Metabolism of phenolics in coffee and plant-based foods by canonical pathways: an assessment of the role of fatty acid β-oxidation to generate biologically-active and -inactive intermediates

Abstract

ω-Phenyl-alkenoic acids are abundant in coffee, fruits, and vegetables. Along with ω-phenyl-alkanoic acids, they are produced from numerous dietary (poly)phenols and aromatic amino acids in vivo. This review addresses how phenyl-ring substitution and flux modulates their gut microbiota and endogenous β-oxidation. 3′,5′-Dihydroxy-derivatives (from alkyl-resorcinols, flavanols, proanthocyanidins), and 4′-hydroxy-phenolic acids (from tyrosine, p-coumaric acid, naringenin) are β-oxidation substrates yielding benzoic acids. In contrast, 3′,4′,5′-tri-substituted-derivatives, 3′,4′-dihydroxy-derivatives and 3′-methoxy-4′-hydroxy-derivatives (from coffee, tea, cereals, many fruits and vegetables) are poor β-oxidation substrates with metabolism diverted via gut microbiota dehydroxylation, phenylvalerolactone formation and phase-2 conjugation, possibly a strategy to conserve limited pools of coenzyme A. 4′-Methoxy-derivatives (citrus fruits) or 3′,4′-dimethoxy-derivatives (coffee) are susceptible to hepatic “reverse” hydrogenation suggesting incompatibility with enoyl-CoA-hydratase. Gut microbiota-produced 3′-hydroxy-4′-methoxy-derivatives (citrus fruits) and 3′-hydroxy-derivatives (numerous (poly)phenols) are excreted as the phenyl-hydracrylic acid β-oxidation intermediate suggesting incompatibility with hydroxy-acyl-CoA dehydrogenase, albeit with considerable inter-individual variation. Further investigation is required to explain inter-individual variation, factors determining the amino acid to which C₆–C₃ and C₆–C₁ metabolites are conjugated, the precise role(s) of l-carnitine, whether glycine might be limiting, and whether phenolic acid-modulation of β-oxidation explains how phenolic acids affect key metabolic conditions, such as fatty liver, carbohydrate metabolism and insulin resistance.

Keywords:

• β-Oxidation
• cinnamic acids
• fatty acids
• gut microbiota
• metabolism
• phenolic acids

1. Introduction

There are many studies showing that dietary (poly)phenols administered orally, affect lipid metabolism in multiple ways. Coffee is the largest contributor of phenolic acids in the diet. A systematic review and meta-analysis showed a significantly decreased risk of nonalcoholic fatty liver disease (NAFLD) among coffee drinkers (Wijarnpreecha, Thongprayoon, and Ungprasert 2017). NAFLD involves heavy fat deposits in the liver leading to steatosis, facilitated by high fat diets, and some dietary (poly)phenols give biochemical and histological improvements in NAFLD (Zhang et al. 2011). Further, phenolic acids have attenuated hepatic steatosis in a mouse model by modulating microRNAs and AMP-activated protein kinase (Pang et al. 2017). Elsewhere, phenolic acid degradation products including 3′,4′-dihydroxycinnamic acid 27, which formed during in vitro incubation of quercetin and cyanidin, protected against mitochondrial damage in an in vitro model of hepatocyte steatosis (Rafiei, Omidian, and Bandy 2019). A very plausible mechanism is through effects of phenolic acids, either ingested as such or produced by gut microbiota-catalyzed breakdown of flavonoids, on fatty acid β-oxidation. Modulation of the rate of β-oxidation in mice with nonalcoholic steatohepatitis (NASH), an advanced phase of NAFLD, affected hepatic lipid accumulation, damage and fibrosis (Barbier-Torres et al. 2020).

The seminal studies of fatty acid β-oxidation utilized synthetic ω-phenyl-alkanoic acids 1 (ω-phenyl-fatty acids or aromatic fatty acids) (Figure 1) to investigate the basic reaction sequence (Dakin 1908b, 1909). At the time it had not been anticipated that ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic acids 2 having hydroxy and methoxy substituents on the phenyl ring would later be recognized as important metabolites of dietary phytochemicals. Although in the last 50 years these phytochemical metabolites have been studied extensively, the extent to which they participate in the β-oxidation cycle has rarely been studied explicitly though there is a reasonable possibility they may be involved. This critical review draws together a fragmented literature to investigate:

Figure 1. Generalized Structures. Note that with reference to β-oxidation the double bond in the ω-phenyl-alkenoic acid 2 is always at C2, the ω-phenyl-hydracrylic acid 3 side-chain hydroxyl is always at C3 whatever the overall side chain length, and n may be zero.

1. Are diet-derived ω-phenyl-alkanoic acids 1 and ω-phenyl-alkenoic acids 2 substrates for endogenous β-oxidation, or is β-oxidation of such substrates restricted to the gut microbiota?

2. Does the phenyl-ring substitution determine whether β-oxidation intermediates such as the ω-phenyl-hydracrylic acids 3 form, and if so, do they accumulate? If the phenyl-hydracrylic acids 3 do not progress efficiently to benzoic acids/benzoyl-glycines does that impact phase-2 conjugation and excretion of C6–C3 glucuronides, sulfates, and other amino acid conjugates?

3. Does the metabolism of diet-derived ω-phenyl-alkanoic acids 1, ω-phenyl-alkenoic acids 2 and ω-phenyl-hydracrylic acids 3 impact on metabolic processes such as fatty acid β-oxidation or carbohydrate metabolism?

1.1. The pathway of fatty acid β-oxidation

Fatty acid β-oxidation occurs in peroxisomes and mitochondria and has four distinct stages (Figure 2a). The basic chemistry is identical in both organelles, but the enzymes and encoding genes are unique to each. Each stage in each organelle has multiple enzymes which differ subtly in substrate specificity, particularly fatty-acyl chain length, and extensive reviews and summaries are available (Houten et al. 2016; Agnihotri and Liu 2003; Knottnerus et al. 2018; Wanders and Waterham 2006; Wanders, Waterham, and Ferdinandusse 2016; Adeva-Andany et al. 2019; Palosaari and Hiltunen 1990). Mitochondrial β-oxidation is a vital source of energy, whereas peroxisomal β-oxidation has different metabolic functions, facilitating the degradation of branched-chain fatty acids, very long-chain fatty acids (VLCFA), and the biosynthesis of bile acids by chain shortening of cholesterol, and metabolism of the side chain of some xenobiotics.

Figure 2. a) Generalized fatty acid β-oxidation scheme occurring in peroxisomes and mitochondria. b) Isomerase can inter-convert cis-3- and trans-2-enoyl-CoAs associated with conventional fatty acids, but it cannot accommodate the trans-2-enoyl-CoA geometric isomer of 3-phenyl-alkenoic acids because the side chain is too short to permit the necessary isomerization.

The summary that follows focuses on those aspects of β-oxidation potentially relevant to the metabolism of ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic acids 2. The β-oxidation of fatty acids is a rapid process and the intermediates shown above (Figure 2a) do not accumulate and are not routinely observed, unless one step is blocked or inhibited. Conjugation with coenzyme A is an essential step for β-oxidation to occur, but CoA pools are limited. The CoA pools are affected by metabolic factors and, interestingly, obese rats had lower hepatic CoA pools, implying restricted β-oxidation of fatty acids in obesity (Chohnan et al. 2020). Total CoA pools in rats were also affected by diet. After a high-fat diet for one month, CoA pools were higher in the hypothalamus, cerebellum, and kidney, but lower in liver and skeletal muscle compared with a high-carbohydrate or high-protein diet. Overall body pools were lowered by one fifth (Tokutake et al. 2012).

Stage 1 in the peroxisomes exploits an acyl-CoA-oxidase (EC 1.3.3.6) whereas stage 1 in the mitochondria utilizes an acyl-CoA-dehydrogenase (short chain EC 1.3.8.1; medium chain EC 1.3.8.7; long chain EC 1.3.8.8), and in every case the unsaturated metabolite has trans geometry. The peroxisomal CoA pool is separate from the mitochondria and the rest of the cell (Berge et al. 1983). If ω-phenyl-alkenoic acids 2 can enter the mitochondrion they may enter β-oxidation directly at stage 2 (see Figure 2a). β-Oxidation auxiliary enzymes facilitate the interconversion of the hydroxyacyl-CoA intermediate from 3S 7a to a racemic mixture with a small excess (58 ± 3%) of the 3R isomer 7b through a rapid dehydration–rehydration via the trans-2-enoyl-CoA 6a stage 1 product (Jin, Hoppel, and Tserng 1992). This epimerase (EC 5.1.2.3) is accompanied by an isomerase enzyme (EC 5.3.3.8). As illustrated in Figure 2b, the isomerase can inter-convert cis-3- 6b and trans-2-enoyl-CoAs 6a associated with conventional fatty acids (Kasaragod et al. 2013), but it cannot accommodate the trans-2-enoyl-CoA geometric isomer of 3-phenyl-alkenoic acids 2 because the side chain is too short to permit the necessary isomerization (for example see cinnamoyl-CoA 9, Figure 3).

Figure 3. Outline metabolism of unsubstituted ω-phenyl-alkanoic acids.

The stage 1 hydrogenation is reversible, utilizing an enoyl-CoA reductase (EC 1.3.1.44) which can hydrogenate trans-2-enoyl-CoAs 6a including cinnamoyl-CoA 9 (Cvetanović et al. 1985; Zhao et al. 2019). Stage 2 hydration is reversible and the position of the equilibrium varies with substrate and depends to a considerable extent on the capacity of the enzymes responsible for stages 3 and 4 (Mao, Chu, and Schulz 1994).

Peroxisomes shorten VLCFAs generating medium chain acyl-CoAs, generally described as C₆ to C₁₀ (Yamada et al. 1987), which are then transferred to the mitochondria. This transfer is either as free fatty acids after release by an acyl-CoA-thioesterase enzyme, or after l-carnitine¹ conjugation and transfer by either carnitine O-acetyltransferase (EC 2.3.1.7) or carnitine O-octanoyl transferase (EC 2.3.1.137) which accommodate short and medium chain fatty acids, respectively (Wanders, Waterham, and Ferdinandusse 2016). Fatty acids up to C₈ enter the mitochondrion by diffusion (Schönfeld and Wojtczak 2016). If for some reason the β-oxidation of the stage 2 product, the 3-hydroxy-fatty acyl-CoA 7, cannot proceed, the acid is metabolized by ω-oxidation and is excreted as a 3-hydroxy-dicarboxylic acid (Jin, Hoppel, and Tserng 1992). Importantly, based on structural considerations, this option is not available for ω-phenyl-alkanoic acids 1.

1.2. Nomenclature of phenolic compounds

The shorthand nomenclature commonly used to describe dietary ω-phenyl-alkanoic acids 1 uses C₆ to define the phenyl ring accompanied by a second term to define the number of carbons in the side chain, for example C₆–C₁, C₆–C₂ and C₆–C₃, etc. These are general descriptors and do not define the phenyl-ring or side chain substituents. As appropriate the terms phenyl-propanoic, cinnamic, phenyl-hydracrylic and phenyl-lactic acid will be used as general terms, or where the phenyl-ring substituents do not require further definition. Studies of the cinnamic acids have used the nature-dominant trans-isomer almost exclusively and in this review the geometry will only be specified where there are data for the cis-isomer and a distinction is required. This review also uses the nomenclature proposed by Kay et al. for the gut microbiota-generated metabolites of dietary (poly)phenols that are found in plasma and urine (Kay et al. 2020), and older forms used in cited references have been amended accordingly to facilitate electronic retrieval. Note, because of constraints on space, the metabolic pathways illustrated in Figures 1–26 cannot show every intermediate in every pathway, and as necessary some, particularly the CoA thio-esters have been omitted.

Figure 4. Synthetic ω-phenyl-alkanoic acids used by Dakin.

Figure 5. Outline metabolism of C₆–C₄ ω-phenyl-alkanoic acids.

Figure 6. Structures of C₆–C₃ positional isomers. That on the left is a potential β-oxidation substrate. That on the right is not.

Figure 7. Outline metabolism of [6]-Gingerol by rats. The precise pathways have not been elucidated but appear to be a mixture of ω-, α- and β-oxidation, and phase-2 conjugation.

Figure 8. Outline metabolism of 3′,5′-dihydroxy-ω-phenyl-alkanoic acids.

Figure 9. Outline metabolism of 2′-hydroxy-substituted ω-phenyl-alkanoic acids.

Figure 10. Outline metabolism of 3′-hydroxy-substituted ω-phenyl-alkanoic acids.

Figure 11. Outline metabolism of 4′-hydroxy-substituted ω-phenyl-alkanoic acids.

Figure 12. Outline metabolism of n-nonyl-phenol and n-octyl-phenol by Rainbow trout and Mosquitofish.

Figure 13. Outline metabolism of 2′-methoxy-substituted ω-phenyl-alkanoic acids.

Figure 14. Miscellaneous metabolites.

Figure 15. Outline metabolism of 4′-methoxy-substituted ω-phenyl-alkanoic acids.

Figure 16. Outline metabolism of 3′,4′-dihydroxy-substituted ω-phenyl-alkanoic acids. For subsequent metabolism of 3′-hydroxyphenyl-substituted metabolites see Figure 10.

Figure 17. Outline metabolism of 3′-methoxy-4′-hydroxy-substituted ω-phenyl-alkanoic acids.

Figure 18. Outline metabolism of 3′-hydroxy-4′-methoxy-substituted ω-phenyl-alkanoic acids.

Figure 19. Outline human and rat metabolism of 3′,4′-dimethoxy-ω-phenyl-alkanoic acids.

Figure 20. Outline metabolism of 3′,4′,5′-tri-substituted-ω-phenyl-alkanoic acids associated with flavanols.

Figure 21. Outline metabolism of 3′,4′-methylenedioxy-substituted-ω-phenyl-alkanoic acids.

Figure 22. Outline metabolism of preexisting 3′,4′,5′-tri-substituted-ω-phenyl-alkanoic acids.

Figure 23. Proposed transformation of 4-hydroxy-5-(phenyl)pentanoic acids to 3-hydroxy-5-(phenyl)pentanoic acid enantiomers.

Figure 24. Proposed transformation of 4-hydroxy-5-(phenyl)pentanoic acids to phenylacetic acids.

Figure 25. Metabolism of phenolic acids by cytosolic and mitochondrial enzymes. ω-Phenyl-alkanoic and alkenoic acids can be conjugated in the cytosol or can enter mitochondria. The distribution between reactions depends on the chain length and is defined by the individual rate constants (k).

Figure 26. Structure of cinnamic acids tested in rats for modulation of lipid metabolism.

2. Ω-Phenyl-alkanoic acids 1 and related compounds without ring substituents

This class of compounds include 3-(phenyl)propanoic acid 13 and cinnamic acid 21a, but also related compounds where the aliphatic chain varies in length. Potential reactions and chemical structures are shown in Figure 3. Relatively long-chain ω-phenyl-alkanoic acids 1 occur naturally in the seeds of various aroids (C₆–C₇ to C₆–C₂₃) (Meija and Soukup 2004), in Trichilia claussenii (Pupo et al. 1996), in the skin secretions of the Stinkpot Turtle (C₆–C₁ to C₆–C₇) (Eisner et al. 1977), and in butterfat (C₆–C₃ to C₆–C₁₃) (Schröder et al. 2014). Trichilia claussenii contained odd- and even-numbered ω-phenyl-alkanoic acids 1 from C₆–C₁₀ to C₆–C₁₅. Only odd-numbered side chains were recorded in the turtle secretions, whereas the aroids also produced one even-numbered component (C₆–C₁₂) and butterfat contained components with odd and even-numbered sidechains, these thought to be products of the ruminant microbiota. If β-oxidation goes to completion, normally odd-numbered substrates yield the C₆–C₁ metabolites whereas even-numbered yield the C₆–C₂ metabolites, but streptozotocin-treated “diabetic” or fasted rats can produce benzoic acid 11 slowly from 4-(phenyl)butanoic acid 15 by α-oxidation, probably at the C₆–C₄ stage (Figure 5) (Takahashi et al. 1992).

In order to investigate the oxidation of fatty acids, Dakin adopted the use of synthetic ω-phenyl-alkanoic acids 1 because the overall reaction was slowed making it easier to isolate and identify the intermediates (Dakin 1908b). Dakin employed 3-(phenyl)propanoic 13, 4-(phenyl)butanoic 15 and 5-(phenyl)pentanoic acids 141 (Figure 4) dosed subcutaneously as sodium salts at doses up to 1 g/kg (ca 5–6 mmol/kg) to strict carnivores (cats and dogs that would not normally be exposed to plant derived (poly)phenols) and observed formation of cinnamic acid 21a, cinnamoyl-glycine 23a, 3-hydroxy-3-(phenyl)propanoic acid 59 and hippuric acid from the C₆–C₅ and C₆–C₃ substrates (141 and 13, respectively) and 3-hydroxy-4-(phenyl)butanoic acid 137 and phenylacetic acid 12 from the C₆–C₄ substrate (Figures 3 and 5) (Dakin 1908a, 1909). This seems to be the only record of the C₆–C₅ hydroxy intermediate 162. It is clear from in vitro studies that the sidechain length of ω-phenyl-alkanoic acids 1 is a major determinant of whether β-oxidation occurs in the peroxisome or only in the mitochondrion. Peroxisomes from human skin fibroblasts can shorten the C₉ side chain of the therapeutic drug N-(α-methyl-benzyl)azelaamic acid to C₇ and C₅ but cannot progress to C₃ (Suzuki et al. 1992). Similarly, Mao et al. reported that 3-(phenyl)propanoic acid 13 was a poor substrate for β-oxidation in the peroxisome (Mao, Chu, and Schulz 1994), but rat peroxisomes are able to metabolize ω-phenyl-alkanoic acids 1 with a side chain of 4 to 12 carbons as efficiently as the endogenous fatty acids (Yamada et al. 1987). Mitochondrial uptake and dehydrogenation of ω-phenyl-fatty acids with a chain length above 10, especially with N = 14 or 16, is severely limited because carnitine-palmitoyl transferase and long chain acyl-CoA dehydrogenase both have low activity with these substrates, but after several chain shortening β-oxidation cycles in the peroxisome the shorter ω-phenyl-fatty acids (N = 4–8) are metabolized by mitochondrial medium chain acyl-CoA dehydrogenase (MCAD). These two organelles working in concert can rapidly convert the longer chain substrates stepwise to phenylacetic acid. Although mitochondrial β-oxidation of C₄ to C₈ fatty acids is carnitine-independent, carnitine is required for the β-oxidation of C₄ to C₈ ω-phenyl-alkanoic acids, and the authors comment that “they would hardly permeate the mitochondrial membrane unless they were in the carnitine ester form” (Yamada et al. 1987).

Consistent with these observations of the effects of chain length, Mao et al. concluded that 3-(phenyl)propanoic acid 13 was almost certainly activated in the mitochondrion where it was subjected to near complete but relatively slow β-oxidation because of an unfavorable equilibrium for the hydration of cinnamoyl-CoA 9 (Mao, Chu, and Schulz 1994). This was confirmed by Zhao et al. using rat liver mitochondria—ATP, NAD⁺ and CoA were essential cofactors for β-oxidation but cinnamic acid 21a β-oxidation increased approximately 60% in the absence of l-carnitine, a characteristic shared with short and medium chain fatty acids. Fatty acids significantly inhibited cinnamic acid 21a β-oxidation with decanoic and linoleic being more potent than palmitic, stearic and oleic acid (Zhao et al. 2019). Cinnamoyl-CoA 9 is a substrate for mitochondrial enoyl-CoA reductase regenerating 3-(phenyl)propanoic acid 13, but hydrogenation was also observed in microsomes independent of the mitochondria (Cvetanović et al. 1985; Zhao et al. 2019). Cis- and trans-cinnamoyl-glycine (23a, 23b) have been reported as normal but minor components of human urine, the cis-isomer (23b) not exceeding 0.1 μmol/litre (Wewer et al. 2021; Lagatie et al. 2021).

3-(Phenyl)propanoic acid 13 (25 mg/kg, 167 μmol/kg) given orally to a patient suffering from long chain 3-hydroxy-acyl-CoA dehydrogenase (EC 1.1.1.211) deficiency may be converted to 3-hydroxy-3-(phenyl)propanoic acid 59 and that this occurs in urine as a conjugate, thought to be with glycine (Duran et al. 1991), but this conjugate was not fully characterized. Two studies with healthy volunteers, one using sodium [²H₆]-cinnamate (500 mg 2.94 mmol) (Hoskins, Holliday, and Greenway 1984) and the other 40 mmol unlabeled sodium cinnamate (Snapper and Saltzman 1949), recorded the excretion of hippuric acid 18 (dominant) plus cinnamoyl-glucuronic acid 202 (ca 1% of dose). Labeled cinnamoyl-glycine 23 was not detected. In contrast one patient suffering from pronounced liver cirrhosis given 40 mmol unlabeled sodium cinnamate excreted ca 14% of the dose as benzoyl-glucuronic acid 207 (Snapper and Saltzman 1949) (Figure 3). Benzoyl-glucuronic acid 207 is also excreted by horses (Marsh et al. 1981).

Palir et al. using in vitro procedures established that 4-(phenyl)butanoic acid 15 (Figure 5) also requires MCAD. Both the short chain and long chain hydroxyacyl-CoA dehydrogenases were able to metabolize 3-hydroxy-4-(phenyl)butanoyl-CoA 16, and mitochondrial long chain ketoacyl-CoA thiolase (LCKAT) could metabolize 3-keto-4-(phenyl)butanoyl CoA 17, but surprisingly, the short chain enzyme, SCKAT, could not (Palir et al. 2017), indicating that the proximity of the bulky phenyl residue interferes with SCKAT. Some individuals have an LCKAT deficiency (Bennett and Sherwood 1993). Sodium 4-(phenyl)butanoate has been used therapeutically, at doses up to 500 mg/kg (2.7 mmol/kg) to control hyperammonaemia in urea cycle disorders through its MCAD β-oxidation product, phenylacetic acid 12, which is excreted predominantly as phenylacetylglutamine 110 (Andrade et al. 2019). Volunteers given 360 μmol/kg also excreted 4-(phenyl)butanoyl-glutamine 157 (21.5% of dose), 3S-hydroxy-4-(phenyl)butanoic acid 137 (4.4% of dose), unmetabolized 4-(phenyl)butanoic acid 15 (0.97% of dose), free phenylacetic acid 12 (0.26% of dose) plus 4-(phenyl)butanoyl-glucuronic acid 204 and phenylacetyl-glucuronic acid 179 the ester (acyl) glucuronides of 4-(phenyl)butanoic acid 15 and phenylacetic acid 12, respectively (1.29% and 1.11% of dose) (see Figures 3 and 5). Almost 40% of the dose was unaccounted for (Kasumov et al. 2004). Phenylacetic acid 12 and 4-(phenyl)butanoic acid 15 enhance peroxisomal β-oxidation through interaction with the human peroxisome proliferator-activated receptor (Pineau et al. 1996), in a similar manner to some lipid-reducing drugs, for example clofibrate (Lazarow and De Duve 1976).

Some anaerobes, for example Clostridium sporogenes, convert phenylalanine 112 via 2R-hydroxy-3-(phenyl)propanoic acid 113 to trans-cinnamic acid 21a (Dickert et al. 2000; Buckel 2019; Figure 10). Subtle changes in diet can modulate the biochemical competence of the gut microbiota to process phenylalanine 112 and determine whether rats given commercial rations excrete hippuric acid 18 or 3-(3′-hydroxyphenyl)propanoic acid 14 (Phipps et al. 1998). It is now recognized that for rats there are two phenotypes with regard to gut microbiota phenylalanine 112 metabolism, one producing 3-(phenyl)propanoic acid 13 and hippuric acid 18, and the other less common producing little hippuric acid 18 and substantial 3-(3′-hydroxyphenyl)propanoic acid 14, plus 3′-hydroxycinnamic acid 19 and sometimes 3-(4′-hydroxyphenyl)propanoic acid 20, these phenyl ring hydroxylated products arising from meta-tyrosine 78 or para-tyrosine 91 respectively (Figure 10). It is thought that there are similar human phenotypes, the less common possibly associated with neurological disorders such as schizophrenia and autism (Clayton 2012; Shaw 2010; Xiong et al. 2016), but a thorough examination of abnormal amino acid catabolism by the gut microbiota, and its sequelae, is beyond the scope of this review.

3-(Phenyl)propanoic acid 13, phenylacetic acid 12 and benzoic acid 11 are three of the four dominant phenolic metabolites in human feces (Jenner, Rafter, and Halliwell 2005; Knust et al. 2006). Fedotcheva et al. have reported that these three metabolites, along with cinnamic acid 21a, damage mitochondria at concentrations of 20–100 μM by inhibiting NAD⁺-dependent respiration and decreasing the Ca²⁺ retention capacity (Fedotcheva, Teplova, and Beloborodova 2010). Normal individuals excrete 3-(phenyl)propanoic acid 13 primarily as hippuric acid 18 but, if overloaded and especially in cases of autorecessive MCAD deficiency, 3-(phenyl)propanoyl-carnitine 156 and 3-(phenyl)propanoyl-glycine 22 are excreted because short chain acyl-CoA dehydrogenase is inactive with this small substrate (Glasgow et al. 1992; Rinaldo et al. 1990; Rinaldo et al. 1988; Moore et al. 1990; Carter et al. 1991). Rinaldo et al. reported that MCAD-deficient patients excreted significantly more 3-(phenyl)propanoyl-glycine 22 than healthy controls, 1.1–37 (median 8) compared with <0.1–0.6 μmol/mmol creatinine² (Rinaldo et al. 1993).

In cases of mitochondrial long chain 3-hydroxy-acyl-CoA-dehydrogenase deficiency, elevated concentrations of 3-hydroxy-3-(phenyl)propanoic acid 59 and cinnamic acid 21a were reported (totalling 240 μmol/litre) probably as glycine conjugates (Duran et al. 1991). Infants do not develop the gut microbiota required for production of 3-(phenyl)propanoic acid 13 until aged four months and 33% were still unable to produce this metabolite when aged six months (Bennett et al. 1992), and antibiotic treatment can suppress the anaerobes responsible potentially for producing false negative results for MCAD deficiency (Bhala et al. 1993).

When rats weighing 300 to 400 g were dosed with [4′-²H]-3-hydroxy-3-(phenyl)propanoic acid 59 (180 mg/rat, ca 2.7 to 3.7 mmol/kg), they excreted labeled hippuric acid 18 and a trace of labeled benzoic acid 11 (Kazakoff and Mamer 1978), this low yield suggesting some impediment to its rapid β-oxidation at high doses. When rats were given much smaller doses of [3-¹⁴C/phenyl-²H₅]-cinnamic acid 21a (up to 2.5 mg/kg; 17 µmol/kg), hippuric acid 18 was the dominant metabolite excreted but 3-hydroxy-3-(phenyl)propanoic acid 59 (up to 0.9% of the dose) was also observed (Nutley, Farmer, and Caldwell 1994). Similar behavior was observed with [3-¹⁴C]-cinnamaldehyde 24 (2 mg/kg; 15 µmol/kg i.p. and 250 mg/kg; 1.9 mmol/kg) which resulted in the excretion of 0.4 ± 0.2% and 1.6 to 1.9% of the dose, respectively, as 3-hydroxy-3-(phenyl)propanoic acid 59 (Peters and Caldwell 1994). Sun et al. reported that rats dosed intra-peritoneally (40 μmol) were able to metabolize cis-cinnamic acid 21b to hippuric acid 18 via trans-cinnamic acid 21a. This cis–trans isomerization also occurred in a rat liver cell-free homogenate and in the presence of the β-oxidation inhibitor 3-pentenoic acid suggesting the presence of a constitutive hepatic isomerase (Sun 2003). Both trans- 23a and cis-cinnamoyl-glycine 23b have been reported as normal minor components of human urine (Wewer et al. 2021; Lagatie et al. 2021).

The peroxisomal human thioesterase-2 is considered important in deactivating slowly metabolized xenobiotic acyl-CoAs thus preventing the sequestration of CoA required for other tasks (Hunt et al. 2002). The released xenobiotic must then either be excreted or diffuse to the mitochondrion for further metabolism. Cao et al. concluded that 3-methoxy-4-hydroxybenzoic acid 10 (Figure 17) diffused into the mitochondrion following perfusion through a rat heart of 3,4-dihydroxybenzoic acid 65 (Figure 16) and COMT methylation, suggesting that carnitine-acyl transferases were not involved with this substrate (Cao, Zhang, et al. 2009), but there is a significant lack of definitive information available in the literature on the mitochondrial uptake of the ω-phenyl-alkanoic acids 1, further discussed in part 6.

3. Ω-Phenyl-alkanoic acids 1, Ω-phenyl-alkenoic acids 2 and related compounds with ring substituents

3.1. The origin of the diet-related ω-phenyl-alkanoic 1 and alkenoic acids 2

Preexisting ω-phenyl-alkanoic acids 1 are comparatively rare, especially those with an even-numbered side chain. C₆–C₄ ω-Phenyl-alkandioic acids have been identified in the peel and pulp of the Prickly Pear cactus Opuntia ficus-indica (Mena et al. 2018; De Santiago et al. 2018), accompanied by 2-hydroxy-4-(4′-hydroxyphenyl)butanoic acid 25 (Figure 5) in the cladodes (Petruk et al. 2017). While these C₆–C₄ compounds may survive cooking (De Santiago et al. 2018), there are no definitive data on their absorption and metabolism, but the dicarboxylic acids are unlikely to be β-oxidation substrates and 2-hydroxy-4-(4′-hydroxyphenyl)butanoic acid 25 is probably a candidate for α-oxidation via 2-keto-4-(4′-hydroxyphenyl)butanoic acid 26 (Figure 5). In contrast, the ω-phenyl-alkenoic acids 2 in free and conjugated form are one of the major groups of dietary (poly)phenols, with the best known being the acyl-quinic acids. These may contain one or more residues of the following cinnamic acids—3′,4′-dihydroxycinnamic acid 27 (Figure 16), 3′-hydroxy-4′-methoxycinnamic acid (isoferulic acid) 28 (Figure 18), 4′-hydroxycinnamic acid 29 (Figure 11), 3′,4′-dimethoxycinnamic acid 30 (Figure 19), 3′,5′-dimethoxy-4′-hydroxycinnamic acid 31 (Figure 22), 3′,4′,5′-trimethoxycinnamic acid 32 (Figure 22) and 3′,5′-dihydroxy-4′-methoxycinnamic acid 33 (Figure 22) (Clifford, Kerimi, and Williamson 2020; Clifford et al. 2017; Ludwig et al. 2014; Jaiswal et al. 2010).

Related ω-phenyl-alkanoic acids 1 with hydroxy- and/or methoxy- substituents on the phenyl ring are similarly produced not only from unabsorbed phenylalanine 112 and tyrosine 91 (Figure 10) (Wadman et al. 1973) but also from flavonoids and cinnamic acid conjugates such as the acyl-quinic acids. Following hydroxylation at C8 the gut microbiota catabolism of flavanols and proanthocyanidins yields a C₄ unit (acetoacetic acid) and a C₆–C₅ unit which may potentially undergo two cycles of β-oxidation (Das and Griffiths 1969; Das 1969; Scheline 1970; Groenewoud and Hundt 1986; Meselhy, Nakamura, and Hattori 1997; Le Bourvellec et al. 2019; Stoupi et al. 2010; Bresciani et al. 2021). Gut microbiota incubations with pure monomers and oligomers have demonstrated that dimers and oligomers with A-type linkages are more resistant to metabolism than those with B-type linkages. The yield of C₆–C₅ and C₆–C₃ metabolites over 24 hours declines markedly with increasing degree of polymerization when expressed relative to the moles of monomer incubated (for example 1 mol of pentamer = 5 mol of monomer) (Di Pede et al. 2022). A significant decline in the yield of the C₆–C₅ catabolites in volunteer urine with an increase in the degree of flavanol polymerization has also been reported (Hollands et al. 2020).

Flavanones, dihydrochalcones and flavones yield C₆–C₃ metabolites potentially susceptible to only one cycle of β-oxidation (Hanske et al. 2009; Honohan et al. 1976; Mosele et al. 2014; Braune, Engst, and Blaut 2005; Zeng et al. 2020; Chen et al. 2018; Zhang et al. 2012; Braune and Blaut 2011; Schoefer, Braune, and Blaut 2004). It has been suggested that dihydrogossypetin, an 8-hydroxy-flavanone, might yield a C₆–C₅ fragment (Jeffrey et al. 1972), but hispidulin, a 6-methoxy-flavone was demethylated to 6-hydroxy-flavone and did not yield any C₆–C₅ catabolites (Labib et al. 2006). Although traces of C₆–C₃ metabolites are sometimes reported, flavonols yield predominantly C₆–C₂ metabolites and anthocyanins yield predominantly C₆–C₁ metabolites neither of which are β-oxidation substrates (Baba et al. 1981; de Ferrars et al. 2014; Mansoorian et al. 2019). Isoflavones yield C₆–C₂ metabolites accompanied by distinctive C₆–C₃ metabolites, 2R-(phenyl)propanoic acids (e.g., 56) (Braune et al. 2010; Coldham et al. 2002; Braune and Blaut 2011; Kim and Han 2014), which are not β-oxidation substrates (Figure 6).

Although in theory the C₆–C₂ metabolites of flavonoids could form from C₆–C₄ metabolites by β-oxidation it has not been possible to find references to flavonoids yielding metabolites with longer even-numbered sidechains and these C₆–C₂ metabolites derived from flavonoids will not feature prominently in this review.

Studies on ileostomists have demonstrated that gut microbiota are essential for the degradation of flavanols, proanthocyanidins, flavanones and flavonols, but anthocyanins may simply fragment chemically at the prevailing gut pH value. Although gut microbiota also transforms unabsorbed cinnamic acid conjugates, a range of C₆–C₃ ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic 2 metabolites are produced by ileostomist subjects from the portion of these substrates absorbed in the stomach and upper gastro-intestinal tract. Whether or not gut microbiota metabolites are products of or substrates for β-oxidation they may retain the substitution pattern of their precursor, or the gut microbiota may remove one or more hydroxyls or methoxyls, and after absorption methylation by catechol-O-methyl transferase (COMT, EC 2.1.1.6) may occur, as may conjugation with glucuronide, sulfate, or an amino acid (Williamson and Clifford 2010, 2017; Clifford, Kerimi, and Williamson 2020).

ω-Phenyl-alkanoic acids 1 may also be metabolites of phenyl-alkanes. The environmental contaminants n-nonylphenol 168 and n-octylphenol 169 are susceptible to mammalian ω-oxidation yielding 9-(4′-hydroxyphenyl)nonanoic acid 170 and 8-(4′-hydroxyphenyl)octanoic acid 171 (Thibaut et al. 1998a; Zalko et al. 2003), which can then pass through four or three β-oxidation cycles, respectively, at least in rainbow trout and rats (Figure 12). After ω-oxidation alkyl-resorcinols from cereals (C₆–C₁₇ to C₆–C₂₅) and quinoa (C₆–C₁₇ to C₆–C₂₆) pass through multiple β-oxidation cycles, eventually producing C₆–C₁ metabolites or C₆–C₂ metabolites in the case of the even-numbered sidechains (McKeown et al. 2016; Ross et al. 2017). The allyl- and propenyl-benzenes characteristic of some spices (Clifford 2000), are also sources of C₆–C₃ ω-phenyl-alkanoic acids 1 (Solheim and Scheline 1976, 1973), and potential substrates for one cycle of β-oxidation. Note that other routes to C₆–C₁ metabolites are known (Figures 15 and 19).

The shogaols and gingerols of ginger have 1-(3′-methoxy-4′-hydroxy-phenyl)alkene and 1-(3′-methoxy-4′-hydroxy-phenyl)alkane skeletons respectively (Figure 7). An early study of [6]-gingerol 172 metabolism in rats (50 mg/kg; 210 µmol/kg)³ concluded that ω–1-oxidation and β-oxidation occurred to a limited extent and reported the excretion of 4-hydroxy-6-keto-8-(3′-hydroxy-4′-methoxyphenyl)octanoic acid 173, 4-(3′-methoxy-4′-hydroxyphenyl)butanoic acid 174, 3′-methoxy-4′-hydroxycinnamic acid 79 and 3-methoxy-4-hydroxybenzoic acid 10 as glucuronides in urine. These accounted for ca 12% of the dose and it was concluded that their formation was initiated by the gut microbiota (Nakazawa and Ohsawa 2002). Subsequent studies in which mice were dosed with [6]-gingerol (200 mg/kg; 840 µmol/kg) (Chen et al. 2012), or [6]-shogaol (200 mg/kg; 840 µmol/kg) (Wang et al. 2017) did not report any ω-phenyl-alkanoic acid 1 or ω-phenyl-alkenoic acid 2 derivatives,⁴ and did not refer to the earlier study with rats.

Although ω-phenyl-alkanoic acids 1 with ring substituents have been little studied from a mechanistic standpoint, when some combination of C₆–C₅, C₆–C₃ and C₆–C₁ metabolites are excreted simultaneously it is often assumed that these are linked by β-oxidation, but because the diet contains multiple substrates and there are well established pathways to C₆–C₃ and C₆–C₁ metabolites other than β-oxidation, and because preformed C₆–C₃ and C₆–C₁ phytochemicals may have been present together in the diet, this assumption may not be valid. Even if it is valid, it is often not clear whether the transformations are performed only by the gut microbiota, only by the liver, or by both in combination.

3.2. Mechanistic studies

Pig kidney and human MCAD strongly bind 4′-hydroxycinnamoyl-CoA 34 (Figure 11) and 4′-methoxycinnamoyl-CoA 35 (Figure 15) at their active site in vitro (Rudik et al. 2000). 4′-Methoxycinnamoyl-CoA 35 binds in vitro to the active site of the mitochondrial enoyl-CoA hydratase responsible for stage 2 of β-oxidation, and the affinity is increased by para or meta electron-withdrawing substituents (D’Ordine et al. 1994). The para and meta substituents associated with dietary ω-phenyl-alkanoic acids 1 are electron donating OH and OCH₃ substituents which would thus reduce the affinity for this enzyme, although this effect would be weaker the longer the side chain and might only be significant for the C₆–C₃ ω-phenyl-alkanoic acids 1. The enoyl-CoA hydratase from Pseudomonas has near equal catalytic ability with 4′-hydroxycinnamoyl-CoA 34 (Figure 11), 3′,4′-dihydroxycinnamoyl-CoA 36 (Figure 16) and 3′-methoxy-4′-hydroxycinnamoyl-CoA 37 (Figure 12), but no detectable activity with cinnamoyl-CoA 9 (Figure 3), 2′-hydroxycinnamoyl-CoA 38 (Figure 9), 3′,4′-methylenedioxycinnamoyl-CoA 39 (Figure 14) or 3′,5′-dimethoxy-4′-hydroxycinnamoyl-CoA 40 (Figure 14). From this study the authors concluded that a 4′-hydroxyl was crucial for reactivity, but that this could be abolished by a 5′-substituent (Mitra et al. 1999), but note that the microbial enzyme is likely to have a different and probably broader substrate specificity compared with the human enzyme.

Studies where volunteers consumed cereal products (supplying alkyl-resorcinols) (Landberg et al. 2018; Zhu et al. 2014; Marklund et al. 2014; McKeown et al. 2016) led to the excretion of 5-(3′,5′-dihydroxyphenyl)pentanoic acid 41, 3-(3′,5′-dihydroxyphenyl)propanoic acid 42 and 3,5-dihydroxybenzoic acid 43 (Figure 8) (Landberg et al. 2018; McKeown et al. 2016; Wierzbicka et al. 2017), establishing that, in contrast to the conclusions by Mitra et al., these C₆–C₅ and C₆–C₃ metabolites are substrates for a hydratase, and because of the very rapid absorption and metabolism of these substrates it must be associated with the liver. Taking these data in combination with those of Mitra et al. it is clear that the loss of enzyme activity is not associated with a 5′-substituent per se, but with the 3′,4′,5′-tri-substituted phenyl ring.

3.3. Relative merits and limitations of experimental methods employed to study metabolism through β-oxidation

Very few of the studies reviewed were designed specifically to investigate β-oxidation of ω-phenyl-alkanoic acids 1 and consequently the data generated are not ideal and often fail to provide information that would have been helpful in answering the questions posed. A particular limitation is the inability to discriminate between hepatic metabolism and transformations promulgated by the gut microbiota in studies with animals or healthy volunteers having an entire gastro-intestinal tract unless urine is collected over several relatively short time periods for up to 48 hours. Studies with gnotobiotic animals, studies where animals have been dosed with antibiotics to obliterate the microbiota or where intra-peritoneal dosing was used, studies with healthy ileostomist subjects and studies using isolated tissues or gut microbiota incubations are invaluable in this respect, but some important substrates have not been examined by these methods. With the exception of studies using isolated tissues, use has often been made of whole beverages which contain more than one substrate, and the origin of, for example an excreted benzoic acid, which might be a β-oxidation product, but which might also arise from some other substrate without β-oxidation, or even have been present in the test beverage, also limits the certainty with which conclusions can be drawn, unless the substance of interest has and retains during metabolism a distinctive pattern of phenyl-ring substituents. In this situation studies with pure compounds, especially if labeled, are invaluable, and 30 relevant studies have been published (Baba et al. 1981; Chen et al. 2018; Zeng et al. 2020; Thibaut et al. 1998a; Das and Griffiths 1969; Honohan et al. 1976; Sangster, Caldwell, and Smith 1984; Sangster et al. 1984; Sangster et al. 1983; Sangster et al. 1987; Caldwell and Sutton 1988; Omar et al. 2012; Zeng et al. 2019; Hackett et al. 1983; Ottaviani et al. 2016; Borges, van der Hooft, and Crozier 2016; Fell, Hoskins, and Pollitt 1978; Das and Sothy 1971; Kazakoff and Mamer 1978; Borges et al. 2018; Curtius, Vollmin, and Baerloch 1972; Curtius 1973; Zagalak et al. 1977; De Preter et al. 2004; Hoskins, Holliday, and Greenway 1984; Thibaut et al. 1998b; Kohri, Nanjo, et al. 2001; Kohri, Matsumoto, et al. 2001; Zalko et al. 2003; James and Smith 1973).

A study in which volunteers were given [¹³C]-labeled cyanidin-3-O-glucoside with labels at carbons 6, 8, 10, 3′ and 5′ detected C₆–C₃ and C₆–C₁ metabolites (Czank et al. 2013), but because it is known that anthocyanins can yield C₆–C₁ fragments at the prevailing in vivo pH value without enzymic involvement, it is impossible to judge whether or not the C₆–C₃ metabolites were subject to β-oxidation. There is a lack of human data for the important 8-to-24-hour period after dosing since this is the time of maximum appearance of metabolites from the gut microbiota. Although a particularly important substrate might not have been examined as such, information is sometimes available because the gut microbiota produce it from a precursor, for example 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44 (Figure 18) produced from hesperidin, but this substrate is accompanied by other metabolites and in this respect the incubate in its complexity resembles a beverage. Inevitably, when gut microbiota-metabolism is a prerequisite to generate the substrate of interest, studies with ileostomist subjects are uninformative about the hepatic metabolism of the substrate. Some phase-2 conjugates, particularly those of larger mass, are excreted primarily in bile (Hirom et al. 1972), and have almost certainly been overlooked when only urine and plasma have been analyzed. There have been few relevant studies, but the major biliary metabolite after Mosquitofish were dosed with [¹⁴C]-n-nonylphenol 168 was its glucuronide (M_r = 396) which accounted for 79–84% of biliary radioactivity (3.4.6, Figure 12) (Thibaut, Monod, and Cravedi 2002), whereas in a study with rats administered rosmarinic acid 208, phase-2 conjugates of several C₆–C₃ metabolites (M_r not exceeding 274) were not detectable in bile (Nakazawa and Ohsawa 1998). In contrast, rats receiving 3′,4′-dihydroxycinnamic acid 27 by intestinal perfusion were found to excrete 3′,4′-dihydroxycinnamic 27, 3′-methoxy-4′-hydroxycinnamic acid 79 and 3′-hydroxy-4′-methoxycinnamic acid 28 in bile, all analyzed after enzymic deconjugation (Lafay et al. 2006). In studies using healthy volunteers with an intact colon such conjugates will be deconjugated by the gut microbiota and the aglycone probably further transformed, and although in theory such conjugates might be recovered from ileal effluent that will only be possible if the aglycone was not solely a product of the gut microbiota. Further investigation will require animal studies with bile duct cannulation.

There are also analytical limitations. Some studies focusing on C₆–C₃ metabolites fail to mention any C₆–C₁ metabolites, but this cannot be taken unequivocally as demonstrating that they were not produced, as distinct from them not being mentioned because they were not of primary interest, and authors rarely state explicitly what metabolites were sought but not found. In hot acid phenyl-hydracrylic acids 3 are dehydrated and the C₆–C₃ 3-keto-propanoic acid β-oxidation intermediates are converted to the equivalent acetophenones and in some early studies both may have been overlooked when acid hydrolysis was used for metabolite deconjugation, but stability to such dehydration serves usefully to distinguish the isomeric 2-hydroxy-3-(phenyl)propanoic (phenyl-lactic) acids from the 3-hydroxy-3-(phenyl)propanoic (phenyl-hydracrylic) acids (Armstrong and Shaw 1957; Sangster et al. 1984; Solheim and Scheline 1973). Acyl-glucuronic acids are also extremely labile, susceptible to acyl migration (forming iso-glucuronides), hydrolysis and at least in the case of drugs participating in the trans-acylation of proteins (haptenization). Haptenization is thought to be responsible for idiosyncratic responses to certain drugs and may have other toxic sequelae (Dickinson 2011). There are few reports of the excretion of acyl-glucuronic acid conjugates of the ω-phenyl-alkanoic acids 1 and ω-phenyl-alkenoic acids 2 (see Table 1 and Figures 3, 5, 9, 15 and 17) but such metabolites may have been overlooked.

Table 1. The less common phase-2 conjugations of phenolic acids.

Glycine	Glutamine	Taurine	I-Carnitine	Acyl-glucuronic acid
Data from human studies
Phenylacetyl	Phenylacetyl	Phenylacetyl^e	Benzoyl^b	Benzoyl
4′-Hydroxyphenylacetyl	4′-Hydroxyphenylacetyl		Phenylacetyl^c	Hippuroyl (tentative)
3-(Phenyl)propanoyl^a	4′-Methoxyphenylacetyl		3-(Phenyl)propanoyl^a	Phenylacetyl
3-hydroxy-3-(phenyl)propanoyl^d	4-(Phenyl)butanoyl		3-(3′,5′-dihydroxyphenyl)propanoyl^f	4′-Methoxyphenylacetyl
Cinnamoyl^d				Cinnamoyl
4′-Methoxycinnamoyl				4-(Phenyl)butanoyl
3′-Methoxy-4′-hydroxycinnamoyl
3′-Hydroxy-4′-methoxycinnamoyl
Data from studies of other mammalian species^f
3′-Hydroxycinnamoyl
4′-Hydroxycinnamoyl
2′-methoxycinnamoyl
2′,4′,5′-Trihydroxycinnamoyl
3′,4′-Dimethoxycinnamoyl
3-(3′,4′,5′-Trimethoxyphenyl)propanoyl
3-(2′-Hydroxyphenyl)propanoyl

^aReported only in connection with MCAD deficiency

^bReported only following sodium benzoate therapy for non-ketotic hyperglycinaemia or hyperammonemia

^cReported only in connection with phenylketonuria

^dNow known to be a normal human metabolite at very low concentrations and reported tentatively at elevated concentrations in connection with long chain mitochondrial 3-hydroxy-acyl-CoA-dehydrogenase deficiency

^eAt a very high dose (588 μmol/kg)

^fAnimal metabolism differs from human metabolism and the doses used were very high relative to the likely human exposure and the relevance of these observations for humans is unclear.

To accommodate the low sensitivity and low resolution of the analytical methods then available, such as paper chromatography, the early studies used very high concentrations of test substances given to volunteers and animals, and it is possible that these large doses saturated and overloaded certain enzyme systems, and so the data obtained might not properly represent the routine dietary situation. Also, for reasons of sensitivity, some studies used enzymic deconjugation with β-glucuronidase and sulfatase followed by derivatisation and GC–MS. Incomplete deconjugation, incomplete derivatisation and possible destruction of heat sensitive metabolites during derivatisation, for example ω-phenyl-hydracrylic acids 3, can result in the under-estimation of some metabolites, and completely overlooking them if the surviving concentrations of derivative are below the LOD (Ordonez et al. 2018). LC–MS may be superior in this respect, but the use of acidic modifiers to improve peak shape of acidic analytes can suppress the ionization of analytes lacking a 4′-hydroxyl, for example 3-hydroxybenzoic acid 46, 3-hydroxy-4-methoxybenzoic acid 47, 3′-hydroxy-4′-methoxycinnamic acid 28 and 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44 (Clifford et al. 2007), although their phase-2 conjugates are not suppressed. The modern procedures for work-up of plasma, urine and feces extracts, etc. to purify and concentrate the analytes of interest routinely uses an acid treatment and it is possible that the equilibrium phenyl-γ-valerolactone–4-hydroxy-5-(phenyl)pentanoic acid occurring in vivo has been disturbed (Wiese et al. 2015; Cortes-Martin et al. 2019; Anesi et al. 2019; Castello et al. 2018; Mulek et al. 2015). Frequently LC–MS analyses use selected ion or selected reaction monitoring but inevitably these approaches address only metabolites that have previously been reported or which the investigators have anticipated, and previously unknown or unexpected metabolites may be missed. It is also possible that some metabolites were overlooked because they co-eluted.

Because of clear species differences, human studies remain the gold standard, but for many substrates of interest such data are not available, and in the following section such information as is available from experimental animals is included, subdivided by phenyl-ring substituent pattern.

3.4. The effect of variations in the phenyl-ring substituents on β-oxidation

It is clear from Dakin’s seminal β-oxidation that in the absence of phenyl-ring substituents the process slows sufficiently at high doses (1 g/kg, ca 5–6 mmol/kg) to permit the β-oxidation intermediates to be isolated and identified but benzoic acid 11 is produced from 5-(phenyl)pentanoic acid 141 (Figure 3) (Dakin 1908b, 1909). In this section, we report on the effect of phenyl-ring substituents on the extent of β-oxidation.

3.4.1. 2′-Hydroxyphenyl substrates

This class includes minor dietary components such as 2′-hydroxycinnamic acid 53 reported in nuts and globe artichokes (Gultekin-Ozguven et al. 2015; Dominguez-Fernandez et al. 2022) and 3-(2′-hydroxyphenyl)propanoic acid 48 reported in some apples and cider (Merinas-Amo et al. 2015) (Figure 9). Rats dosed with dihydrocoumarin (100 mg/kg; 676 μmol/kg) excreted 3-(2′-hydroxyphenyl)propanoic acid 48 (melilotic acid) (Jacobi et al. 2016) which undergoes β-oxidation yielding 2-hydroxybenzoic acid 49 (Adams et al. 1998). Rats dosed with coumarin (100 mg/rat; 680 µmol/rat) excreted 3-hydroxy-3-(2′-hydroxyphenyl)propanoic acid 50, 2′-hydroxycinnamoyl-glycine 51 (of unknown geometry) and 3-(2′-hydroxyphenyl)propanoyl-glycine 52. The analogous benzoic and hippuric acids were sought but not detected and clearly β-oxidation did not proceed to completion (Booth et al. 1959). The weight of the rats used in these two studies was not recorded, but the dose of coumarin must have been some 4–5-times larger than the dose of dihydrocoumarin and this might explain the excretion of a significant amount of 3-hydroxy-3-(2′-hydroxyphenyl)propanoic acid 50 and impairment of β-oxidation in the coumarin study.

It is presumed that cis-2′-hydroxycinnamic acid 53a produced by opening of the coumarin lactone ring is the immediate precursor with hydrogenation either by the gut microbiota or the liver. If hydrogenated by the gut microbiota, the resultant 3-(2′-hydroxyphenyl)propanoic acid 48 may have entered β-oxidation yielding trans-2′-hydroxycinnamic acid 53b. Incubation of coumarin 139 with rat microbiota consistently yielded 3-(2′-hydroxyphenyl)propanoic acid 48 but 2′-hydroxycinnamic acid 53 was not found, and dosing rats with coumarin 139 (100 mg/kg; 680 µmol/kg) produced 2′-hydroxyphenylacetic acid 54 as the main urinary metabolite at this much lower dose, accompanied by 3-(2′-hydroxyphenyl)propanoic acid 48 and traces of 3′-hydroxycinnamic acid 19 (Figure 10) (Scheline 2009).

There has been much concern regarding the safety of coumarin 139 and its possible carcinogenicity, and a tolerable daily intake of 0.1 mg/kg b.wt has been set. Human metabolism (Figure 9) is complex and very different from that of the rat which excretes <1% as 7-hydroxycoumarin 140 (umbelliferone), whereas humans excrete ca 80%, plus up to ca 10% 2′-hydroxyphenylacetic acid 54 (Leonart et al. 2017; Abraham et al. 2010; Lewis, Ito, and Lake 2006; Meineke et al. 1998). Volunteers who consumed 200 g of sous vide-cooked artichoke excreted in 24 hours 0.2 μmol 3-(2′-hydroxyphenyl)propanoic acid 48, 0.8 μmol 2-hydroxybenzoic acid 49 and 32 μmol 2′-hydroxyhippuric acid (Dominguez-Fernandez et al. 2022). 2-Hydroxybenzoic acid 49 as a metabolite of aspirin, as distinct from that formed by β-oxidation of a C₆–C₃ precursor, is conjugated to glycine (Figure 10) which in turn may be conjugated with glucuronic acid (Zimmerman et al. 1981; Gallice et al. 1985) and is discussed more fully in sections 4.1.2 and 4.3.

3.4.2. 3′-Hydroxyphenyl substrates

This class includes 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 and 3′-hydroxycinnamic acid 19, and pathways and chemical structures are shown in Figure 10. The major dietary sources of 3′-hydroxyphenyl derivatives are gut microbiota catabolism of unabsorbed phenylalanine 112 and tyrosine 91 (Curtius, Mettler, and Ettlinger 1976), hepatic catabolism of surplus phenylalanine 112 and tyrosine 91 (Guroff et al. 1967; Guroff, Reifsnyder, and Daly 1966), and 3′,4′-dihydroxyphenyl and 3′,4′,5′-trihydroxyphenyl derivatives subjected to gut microbiota dehydroxylation. In vitro incubations of (+)-catechin and a (+)-catechin dimer (procyanidin B3) with rat cecal microbiota for five days generated 5-(3′-hydroxyphenyl)-γ-valerolactone 57, 3-(3′-hydroxyphenyl)propanoic acid 14, 4-hydroxybenzoic acid 45 and eventually 3-hydroxybenzoic acid 46 (Groenewoud and Hundt 1986), demonstrating two cycles of β-oxidation. The unsaturated intermediate 19 and phenyl-hydracrylic acid 58 (Figure 10) intermediate were not mentioned in the report.

The rat studies with 3′-hydroxy-phenyl substrates are inconsistent and may be an early demonstration of the phenotypic variation later demonstrated unequivocally (Phipps et al. 1998). When rats were dosed with 100 mg (552 µmol) m-tyrosine 78, the only metabolites reported were 3-(3′-hydroxyphenyl)propanoic acid 14 and 3′-hydroxyphenylacetic acid 111 (Booth et al. 1960). Further details were promised in a subsequent paper which it has not been possible to locate. Rats fed 3′-hydroxycinnamic acid 19 (50 mg; 305 µmol) excreted the β-glucuronide conjugate with no evidence of β-oxidation, but when the same rats were fed 3-(3′-hydroxyphenyl)propanoic acid 14 (50 mg; 301 µmol) free 3′-hydroxycinnamic acid 19 and 3′-hydroxyhippuric acid 55 were excreted (Booth et al. 1957), suggesting that one cycle of β-oxidation had occurred, but the β-glucuronide conjugate of 3′-hydroxycinnamic acid 19 was not mentioned. A possible explanation for this apparently inconsistent behavior might be that 3-(3′-hydroxyphenyl)propanoic acid 14 can enter the mitochondrion but 3′-hydroxycinnamic acid 19 cannot. Das et al. reported that rats fed (+)-catechin excreted 3-(3′-hydroxyphenyl)-γ-valerolactone 57 and 3-(3′-hydroxyphenyl)propanoic acid 14 but there were no reports of C₆–C₁ metabolites (Das 1969), suggesting that only one β-oxidation cycle had occurred.

Germ-free rats fed a sterile (γ-irradiated) standard rat diet excreted 3-(3′-hydroxyphenyl)propanoic acid 14 (25 ± 22 nmol/24 hours) and after these rats had been fed feces to introduce a normal rat microbiota, this excretion increased to an extremely skewed 3520 ± 5150 nmol/24 hours. In contrast, the excretion of 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 did not increase significantly (116 ± 42 nmol/24 h and 133 ± 70 nmol/24 h), and while the excretion of hippuric acid 18 increased 5-fold there was no mention of either 3-hydroxybenzoic 46 or 3′-hydroxyhippuric acid 55 (Goodwin, Ruthven, and Sandler 1994). There is no obvious simple explanation for this behavior. Collectively these data indicate that some rats can subject 3′-hydroxyphenyl-substituted C₆–C₅ metabolites to two cycles of β-oxidation, but that sometimes the second cycle seems not to occur, and there is a clear tendency for the excretion of an appreciable amount of the 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 intermediate.

3-(3′-Hydroxyphenyl)propanoic acid 14 is one of the four dominant phenolic metabolites in human feces (Jenner, Rafter, and Halliwell 2005; Knust et al. 2006). 3-Hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 is considered a normal urinary metabolite of healthy patients and free-living volunteers. After the consumption of orange juice, it is commonly accompanied by 3′-hydroxyhippuric acid 55 suggesting slow β-oxidation. Unexpectedly, 6 out of 27 volunteers who consumed orange juice did not excrete it, and 3 of the 6 did not excrete 3′-hydroxyhippuric acid 55 (Pereira-Caro et al. 2014; Pereira-Caro et al. unpublished). Aschoff et al also reported 1 out of 12 volunteers who did not excrete 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 (Aschoff et al. 2016). This is further discussed in 5.3 and 6.

3-Hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 has also been detected in the urine of patients recovering from gastro-intestinal surgery receiving only glucose–saline drip and achromycin for at least 4 days (Saini et al. 1974), the urine of volunteers receiving neomycin and [3′,5′-²H₂]-tyrosine 91 (150 mg/kg; 820 μmol/kg), but was not detected when [3′,5′-²H₂]-tyrosine 91 was incubated in vitro with human gut microbiota (Curtius, Mettler, and Ettlinger 1976; Fuchs-Mettler et al. 1980) suggesting that its formation from tyrosine 91 is endogenous. The failure of ileostomists to excrete 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 after consumption of grape juice when it was excreted by volunteers with an intact colon, indicates that the gut microbiota are able to produce an essential precursor from a grape juice constituent (Stalmach et al. 2013).

3-Hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 has been characterized as the (–)-stereo-isomer (Armstrong, Wall, and Parker 1956; Stalmach et al. 2013). Its excretion by healthy individuals with an intact colon is extremely variable (11–494 µmol/mg creatinine),⁵ depending in part on the diet, although it has been clearly demonstrated that its excretion is not inevitably derived from amino acids (Armstrong and Shaw 1957; Rampini et al. 1974; Vollmin et al. 1971), and its excretion, and that of 3-(3′-hydroxyphenyl)propanoic acid 14, 3-hydroxybenzoic acid 46 and 3′-hydroxyhippuric acid 55, are influenced by single nucleotide polymorphisms in the ABCC2 gene encoding multidrug resistance-associated protein-2 in the kidney (Muhrez et al. 2017). 3-Hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 is ostensibly the stage 2 β-oxidation product of 3-(3′-hydroxyphenyl)propanoic acid 14 but might also be the product of 3′-hydroxycinnamic acid 19 which has bypassed stage 1. In at least some urine samples, 3-hydroxy-(3′-hydroxyphenyl)propanoic acid 58 is accompanied by 3′-hydroxyhippuric acid 55 but with free-living volunteers it is not certain that the hippuric acid 55 is derived from the 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58.

There is a single report of 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 excretion after a single volunteer consumed a 1 litre bolus dose of coffee prepared from 64 g of commercial roasted coffee (Shaw and Trevarthen 1958). Borges et al. reported that after consumption of [2-¹⁴C]-(–)-epicatechin (60 mg; 207 μmol), excretion of the [¹⁴C]-label in 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58, 3′-hydroxyhippuric acid 55 and hippuric acid 18 continued for more than 24 h (Borges et al. 2018). 3-Hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 has also been noted as a minor metabolite after healthy volunteers consumed a large dose (10 mg/kg; 16 µmol/kg) of [³H-2′,5′,6′]-rutin (Baba et al. 1981), but there was no reference to any benzoic or hippuric acids being excreted. Similarly, after consuming orange juice supplying 329 µmol hesperetin glycosides, the excretion of 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 increased ca 12-fold to 31 µmol/24 hours for eight out of 12 volunteers, and 3′-hydroxyhippuric acid 55 excretion increased from undetectable to 0.8 µmol/24 hours for nine out of 12 volunteers (Pereira-Caro et al. 2014). A similar orange juice study by Aschoff et al. supplying hesperidin (636 µmol) and narirutin (115 µmol) reported the excretion in 24 hours of ca 115 µmol 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 by 11 out of 12 volunteers (Aschoff et al. 2016). Using LC–HRMS there has been a tentative identification in human urine, but not plasma, of one glucuronide and one sulfate conjugate of 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 (Pereira-Caro, Clifford, et al. 2020). Confirmation is required, but the availability of authentic standards of the conjugates of the isomeric 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 establishes that these known metabolites have not been confused with those assigned tentatively.

After consumption of green tea supplying 634 µmol flavanols there was a ca 7-fold increase in the excretion of 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 to 47 µmol/24 hours although the associated benzoic 46 and hippuric acids 55 were not reported (Roowi et al. 2010). Incubations of rutin, eriodictyol, hesperetin and flavanols with gut microbiota in vitro suggest that the immediate precursor is 3-(3′-hydroxyphenyl)propanoic acid 14 derived from 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (Figure 16) in the case of rutin (Serra et al. 2012), and eriodictyol (Honohan et al. 1976; Miyake et al. 2000), 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44 (Figure 18) in the case of hesperetin (Honohan et al. 1976), and 3-(3′,4′-dihydroxyphenyl)propanoic 60 acid or 3-(3′,4′,5′-trihydroxyphenyl)propanoic acid 61 (Figure 22) in the case of the flavanols (Chen and Sang 2014; Stoupi et al. 2010).

Hackett et al. reported only a small amount of [¹⁴C]-labeled 3-(3′-hydroxyphenyl)propanoic acid 14, 3-hydroxybenzoic acid 46 and 3′-hydroxyhippuric acid 55 were detected in urine after three volunteers consumed 2 g (ca 7 mmol) [U-¹⁴C]-(+)-catechin (Hackett et al. 1983), but there was no mention of the 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 intermediate or the analogous 3′,4′-dihydroxyphenyl metabolites.

Volunteers who consumed [2-¹⁴C]-(–)-epicatechin excreted three phase-2 conjugates of 4-hydroxy-5-(3′,4′-dihydroxyphenyl)pentanoic acids 62 (Figure 16), one phase-2 conjugate of 4-hydroxy-5-(3′-hydroxyphenyl)pentanoic acid 63 (Figures 10 and 20), five phase-2 conjugates of 5-(3′,4′-dihydroxyphenyl)-γ-valerolactone 64 (Figure 16) and one phase-2 conjugate of 5-(3′-hydroxyphenyl)-γ-valerolactone 57 (Figure 10) which collectively accounted for 42 ± 5% of the dose (207 µmol). The only C₆–C₃ metabolite reported was 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 (5.6% of dose) which was accompanied by 3′-hydroxyhippuric acid 55 (8.2% of dose) and hippuric acid 18 (12.8% of dose) (Ottaviani et al. 2016). The excretion of labeled hippuric acid 18 and 3′-hydroxyhippuric acid 55 indicate that some 21% of the substrate underwent β-oxidation but there is no evidence for the formation of either 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (Figure 16) or 3,4-dihydroxybenzoic acid 65, and it appears that β-oxidation occurred only after 4′-dehydroxylation by the gut microbiota. Because two 3′,4′-dihydroxyphenyl-substituted C₆–C₅ metabolites were quantified along with one 3′-hydroxyphenyl-substituted C₆–C₅ metabolite, there is some uncertainty about the flux of 3′-hydroxy-substituted C₆–C₅ metabolites excreted in 24 hours, but this was not less than 33 µmol and not more than ca 50 µmol, although sufficient to overload the system leading to excretion of the 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 intermediate. Madrid Gambin et al. (Madrid-Gambin et al. 2016) state explicitly that 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 is associated with coffee consumption (and use the abbreviation HPHPA to refer to this compound) citing Guertin et al. (Guertin et al. 2015) as the source of this information. However, upon inspection of this reference it appears that Guertin et al. reported 3-(3′-hydroxyphenyl)propanoic acid 14 (Guertin et al. 2015) and Madrid-Gambin et al. have mis-interpreted the structure.

Collectively these data demonstrate that 3′-hydroxyphenyl C₆–C₅ metabolites in humans can undergo two cycles of β-oxidation, but that the second of these may be slow, leading to a noticeable excretion of the 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 intermediate (Figure 10). An important feature, but often overlooked feature of these studies, is that some test materials such as orange juice supply protein-bound/free phenylalanine 112 and tyrosine 91 as well as the phytochemicals of primary interest. Therefore, the yield of 3′-hydroxyphenyl-substituted metabolites is likely to be increased relative to studies with green tea, coffee or pure compounds.

3.4.3. 4′-Hydroxyphenyl substrates

This group includes 4′-hydroxycinnamic acid (p-coumaric acid) 29 and its various esters, tyrosine 91, 2-hydroxy-4-(4′-hydroxyphenyl)butanoic acid 25 and naringenin, plus metabolites formed by 3′-dehydroxylation of 3′,4′-dihydroxyphenyl substrates—see section 3.4.8, Figures 11 and 12. Rat liver mitochondria in vitro interconvert 3-(4′-hydroxyphenyl)propanoic acid 20 and 4′-hydroxycinnamic acid 29 (Ranganathan and Ramasarma 1974), and produce 4-hydroxybenzoic acid 45 from 4′-hydroxycinnamic acid 29. “Reverse” hydrogenation of 4′-hydroxycinnamic acid 29 was inhibited in vitro by octanoic acid (200 μM), and its β-oxidation was inhibited by both benzoic acid 11 and trans-cinnamic acid 21a (Ranganathan and Ramasarma 1971). Das et al. reported that rats fed (+)-catechin excreted 3-(4′-hydroxyphenyl)propanoic acid 20, presumably arising from 3′-dehydroxylation of a C₆–C₅ precursor but there were no reports of the analogous C₆–C₁ metabolites (Das 1969), suggesting that only one β-oxidation cycle had occurred.

Some anaerobes produce 4′-hydroxycinnamic acid 29 from unabsorbed tyrosine 91 via 2-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 66 (Kim et al. 2004), the cinnamic acid being a potential β-oxidation substrate. Booth et al. reported that rats dosed with 3-(4′-hydroxyphenyl)propanoic acid 20 (100 mg/rat; 610 µmol/rat) excreted 4′-hydroxycinnamoyl-glycine 67 plus benzoic acid-4-sulfate 213 (Booth et al. 1960). Griffiths et al. reported that rats dosed with 3-(4′-hydroxyphenyl)propanoic acid 20 (200 mg/rat; 1.2 mmol/rat) excreted unchanged test substance plus 4′-hydroxycinnamic acid 29 and 4-hydroxybenzoic acid 45, but that the associated cecal microbiota did not transform this substrate in vitro—indeed the microflora converted 4′-hydroxycinnamic acid 29 to 3-(4′-hydroxyphenyl)propanoic acid 20 (Griffiths and Smith 1972). When the rats were given oral doses of flavonoids having a 4′-hydroxylated B-ring, e.g., apigenin, apiin, phloridzin and naringin, all excreted 3-(4′-hydroxyphenyl) propanoic acid 20, 4′-hydroxycinnamic acid 29 and 4-hydroxybenzoic acid 45. However, 4-hydroxybenzoic acid 45 was not produced when these flavonoids were incubated with the rat microbiota (Griffiths and Smith 1972). Collectively these observations suggest that 4′-hydroxycinnamic acid 29 is not a β-oxidation substrate for rat microbiota and that hepatic β-oxidation was responsible for at least this last step.

Rats given an oral dose of 100 mg (552 µmol) tyrosine 91 excreted 2-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 66, 3-(4′-hydroxyphenyl)propanoic acid 20, 4′-hydroxycinnamic acid 29, 4-hydroxybenzoic acid 45 and 4′-hydroxycinnamoyl-glycine 67, and this profile was not changed when antibiotics were co-administered (Booth et al. 1960). The excretion of the 4′-hydroxycinnamoyl-glycine 67 suggests that the hepatic β-oxidation pathway was overloaded. The excretion of the 4′-hydroxycinnamoyl-glycine 67 accompanied by benzoic acid-4-sulfate rather than the glycine conjugate 18 (Figure 3) seems to be a characteristic of the rats used by Booth et al. and is probably an early demonstration of the phenotypic variation subsequently recognized (Phipps et al. 1998).

This variable metabolism is confirmed in more recent studies. One study which utilized labeled [²H-2′,3′,5′,6′]-naringin and human gut microbiota in vitro has confirmed the production of 3-(4′-hydroxyphenyl)propanoic acid 20 and 3-(phenyl)propanoic acid 13 (Chen et al. 2018; Chen et al. 2019), but the traces of 4-hydroxybenzoic acid 45 reported when unlabeled naringin was fed to rats (Zeng et al. 2019), were not observed, again suggesting this final step is associated with hepatic metabolism. However, a second study in which [²H-2′,3′,5′,6′]-naringin was fed to rats recorded the excretion of labeled 3-(4′-hydroxyphenyl)propanoic acid 20, 4′-hydroxycinnamic acid 29, 4-hydroxybenzoic acid 45, 4′-hydroxyhippuric acid 68 and hippuric acid 18 in urine but the intermediate 3-hydroxy-3-(4′-hydroxyphenyl) propanoic acid 69, while sought, was not found (Zeng et al. 2020).

Bifidobacterium longum R0175 produced 3-(4′-hydroxyphenyl)propanoic acid 20 (ca 14%) from naringenin over 48 hours in vitro accompanied by 3-(phenyl)propanoic acid 13 (maximally 5% at 36 hours, but only 1.2% at 48 hours indicating further metabolism). Lactobacillus rhamnosus subsp. rhamnosus NCTC 10302 also produced 3-(4′-hydroxyphenyl)propanoic acid 20 (ca 10%) and 3-(phenyl)propanoic acid 13, which increased progressively from 1.3% at 12 hours to 7.3% at 48 hours. It is clear from the disappearance of the substrate (16% and 34%, respectively) that other products must have been formed by both organisms, but neither 3-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 69 nor 3-hydroxy-3-(phenyl)propanoic acid 59 were detected, and there was no evidence for the production of C₆–C₁ metabolites through β-oxidation (Pereira-Caro et al. 2018), although the late decline in 3-(phenyl)propanoic acid 13 suggests that this might have occurred with B. longum.

A study in which rainbow trout (Figure 12) were dosed orally with 25 mg/kg of [³H]-4′-n-nonylphenol 168 produced 9-(4′-hydroxyphenyl)nonanoic acid 170 by ω-oxidation and excreted in bile 7-(4′-hydroxyphenyl)heptanoic acid, 5-(4′-hydroxyphenyl)pentanoic acid, 3-(4′-hydroxyphenyl)propanoic acid 20 and 4′-hydroxycinnamic acid 29 all as glucuronides (Thibaut et al. 1998a, 1998b; Thibaut et al. 2000; Thibaut, Monod, and Cravedi 2002). 4-Hydroxybenzoic acid 45 was not observed in bile but was the main product in urine, accompanied by 3-(4′-hydroxyphenyl)propanoic acid 20 and 4′-hydroxycinnamic acid 29 (Thibaut et al. 1999). Rats dosed with 1 and 10 μg/kg [³H]-4′-n-nonylphenol 168 excreted phase-2 conjugates of 3-(4′-hydroxyphenyl)propanoic acid 20, 4-hydroxycinnamic acid 29 and 4-hydroxybenzoic acid 45 (Zalko et al. 2003),

Human studies with low oral dose (12 mg/kg; 62 μmol/kg) l-[²H₂-3′,5′]-tyrosine 91 have established that the major urinary metabolites are 2-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 66, 4′-hydroxyphenylacetic acid 70 and 2-hydroxy-2-(4′-hydroxyphenyl)acetic acid 142 (Fell, Hoskins, and Pollitt 1978). 2-Hydroxy-2-(4′-hydroxyphenyl)acetic acid 142 excretion was extremely variable across the subjects and may have formed after tyrosine decarboxylation, either endogenously or in the gut. Minor products were not detectable at the low dose used, but from other studies it is known that a small amount proceeds via rapid β-oxidation to 4-hydroxybenzoic acid 45 with little or no accumulation of the β-oxidation intermediates (Booth et al. 1960), this pathway predominantly associated with unabsorbed tyrosine 91 and gut microbiota deamination (Kim et al. 2004; Vanderhe et al. 1971). This is well illustrated by a study where 15 volunteers on two occasions underwent 2-day washouts on low (poly)phenol but nutritionally adequate diets. They excreted predominantly 4′-hydroxyphenylacetic acid 70 (87 ± 6 and 110 ± 15 µmol/24 hours) but relatively little 3-(4′-hydroxyphenyl)propanoic acid 20 (2.2 ± 0.3 and 2.6 ± 0.3 μmol/24 hours) and 4-hydroxybenzoic acid 45 (21 ± 7 and 12 ± 9 μmol/24 hours), and 4′-hydroxyhippuric acid 68 was sought but not detected (Pereira-Caro et al. 2015).

Fecal tyrosine 91 is increased significantly in liver cirrhosis but the actual concentrations were not stated (Huang et al. 2013). A patient with abnormally high colonic tyrosine 91 as a consequence of cystic fibrosis, and receiving medium chain triglyceride administration, excreted 4′-hydroxycinnamic acid 29 (1.59 µmol/mg creatinine),⁶ 3-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 69 (0.92 µmol/mg creatinine), and 4-hydroxybenzoic acid 45 (18.3 µmol/mg creatinine) (Wadman et al. 1973), clearly indicating that β-oxidation of 4′-hydroxyphenyl-substituted ω-phenyl-alkanoic acids 1 can be compromised when the total β-oxidation load is very high.

When volunteers consumed orange juice supplying naringin (165 µmol in orange juice) accompanied by apigenin-6,8-di-C-glucoside (42 µmol), there was no evidence of overloading leading to the excretion of the 3-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 69 at this comparatively low dose. Only 4 out of 12 volunteers showed a significant increase in the excretion of 4-hydroxybenzoic acid 45 relative to placebo, but 11 out of 12 showed a significant increase in the excretion of 4′-hydroxyhippuric acid 68 (21 ± 2 compared with 2.9 ± 0.3 µmol/24 hours). The twelfth volunteer who did not excrete 4′-hydroxyhippuric acid 68 had a significantly increased excretion of 3′-hydroxyhippuric acid 55 (Figure 10) (3.8 µmol/24 hours compared with 0 µmol/24 hours) (Pereira-Caro et al. 2014).

When Nurmi et al. gave volunteers an oregano extract supplying 133 µmol rosmarinic acid 208 (Figures 14 and 16) (out of 223 µmol total phenols), 4-hydroxybenzoic acid 45 (257 ± 83 µmol out of 575 µmol total phenols) was the main metabolite excreted over 48 hours, much higher than the 21 μmol excreted at baseline (Nurmi et al. 2006). Its origin is uncertain but on a purely arithmetic basis it seems as though it must have formed from both the 3′,4′-dihydroxycinnamic acid 27 (Figure 16) and 2R-hydroxy-3-(3′,4′-dihydroxyphenyl)propanoic acid 71 (Figure 16) moieties of rosmarinic acid 208 (Figures 14 and 16). However, it is also possible that some protein was present in the oregano extract, and as this was presented in a gelatin capsule, some phenylalanine 112 and possibly tyrosine 91 would have been consumed and may well have contributed. This increase in the quantity of 4-hydroxybenzoic acid 45 excreted and the apparent absence of 4′-hydroxyhippuric acid 68 is in marked contrast to the results of the study on orange juice (Pereira-Caro et al. 2014), and is further discussed in 4.1.2. When ileostomists consumed Concord grape juice supplying ca 18 µmol of 4′-hydroxycinnamoyl-tartaric acid 209 (Figures 11 and 14), the authors did not report excretion of 3-(4′-hydroxyphenyl)propanoic acid 20.This would suggest that “reverse” hydrogenation did not occur (Stalmach et al. 2011, 2012).

After volunteers consumed sous vide-cooked artichoke providing 12.8 μmol of free and esterified 4′-hydroxycinnamic acid 29, they excreted 16.5 μmol of 4′-hydroxyphenyl-substituted metabolites in the first four hours. These consisted of 0.03 μmol free and 0.43 μmol phase-2 conjugated 4′-hydroxycinnamic acid 29 (Figure 11), plus 0.81 μmol of 2-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 66, 9.12 μmol free 4-hydroxybenzoic acid 45 and 6.1 μmol 4′-hydroxyhippuric acid 68 (Dominguez-Fernandez et al. 2022). There is no evidence that β-oxidation was impeded despite modest phase-2 conjugation, but clearly some 4-hydroxybenzoic acid must have originated from hepatic metabolism of tyrosine 91.

Consistent with the observation that MCAD binds 4′-hydroxycinnamoyl-CoA 34 (Rudik et al. 2000), these data demonstrate that 4′-hydroxyphenyl C₆–C₃ derivatives can undergo in humans one cycle of β-oxidation but that at very high substrate loads the 4′-hydroxycinnamic acid 29 and 3-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 69 intermediates may accumulate. The excretion of 4′-hydroxycinnamic acid 29, 4-hydroxybenzoic 45 and 4′-hydroxyhippuric acid 68 are influenced by single nucleotide polymorphisms in the ABCC2 gene encoding multidrug resistance-associated protein-2 in the kidney (Muhrez et al. 2017).

3.4.4. 2′-Methoxyphenyl substrates

Rats given 2′-methoxycinnamaldehyde 175 (1.3 mmol/kg) intragastrically excreted 2′-methoxycinnamic acid 176 (3.2% of dose), 3-hydroxy-3-(2′-methoxyphenyl)propanoic acid 177 (22.2% of dose), 2′-methoxycinnamoyl-glycine 150 (23.7% of dose) and 3-(2′-methoxyphenyl)propanoyl-glycine 151 (36.9% of dose) clearly demonstrating that 2′-methoxycinnamic acid 176 is not a substrate for complete β-oxidation (Samuelsen et al. 1986) (Figure 13). It has not been possible to locate any human studies.

3.4.5. 3′-Methoxyphenyl substrates

There are very few relevant data. Jenner et al. reported the excretion of 3-(3′-methoxyphenyl)propanoic acid 180 (110–380 μmol/24 hours) by free-living volunteers, but its origin is obscure (Jenner, Rafter, and Halliwell 2005) (Figure 14).

3.4.6. 4′-Methoxyphenyl substrates

Foods contain relatively few 4′-methoxyphenyl-substituted components, and these are always relatively minor components, or the foods are minor components of the diet, with herbs, spices and oranges being the main sources (Vazquez-Fresno et al. 2019; Crozier, Clifford, and Ashihara 2006). Metabolism and structures are shown in Figure 15. Rats dosed intragastrically with 100 mg/kg (562 µmol/kg) 4′-methoxycinnamic acid 72 excreted 3-hydroxy-3-(4′-methoxyphenyl)propanoic acid 73, 4-methoxybenzoic acid 74 and 4′-methoxyhippuric acid 75 possibly suggesting complete β-oxidation. However, the accumulation of the 3-hydroxy-3-(4′-methoxyphenyl)propanoic acid 73 suggests that possibly the liver was overloaded. There was no evidence of “reverse” hydrogenation.

There are data from animal studies that phenylpropanoids such as anethole (3-(4′-methoxyphenyl)prop-2-ene) 114 and estragole (4′-methoxy-allylbenzene; 1-(4′-methoxyphenyl)prop-1-ene) 211, and probably 3-(4′-methoxyphenyl)propane (p-propyl-anisole) 201, are substrates for ω-oxidation (Solheim and Scheline 1973; Sangster et al. 1984; Sangster et al. 1983) yielding 4′-methoxycinnamic acid 72 (Figure 15). However, there is also evidence that the C1 benzylic carbon of 3-(4′-methoxyphenyl)propane 201, and especially the C1 benzylic and allylic carbon of estragole 211, are susceptible to CYP450-hydroxylation and may yield 4′-methoxybenzoic acid 74 independent of β-oxidation (Solheim and Scheline 1973; Cartus et al. 2012). On the evidence currently available it is impossible to judge the relative yields of C₆–C₁ metabolites from these two routes of phenylproanoid metabolism.

Volunteers have been given [methoxy-¹⁴C]-anethole 114 at two doses covering the extremes of normal dietary exposure (1 mg and 50 mg; 6.7 µmol and 338 µmol). After a dose of 1 mg (6.7 µmol), two volunteers excreted 4′-methoxycinnamic acid 72 (<0.2% of dose), 4-methoxybenzoic acid 74 (ca 3.5% of dose) and 4′-methoxyhippuric acid 75 (54–58% of dose) (Sangster et al. 1987) and this pattern was maintained at a very high dose (250 mg; 1.7 mmol) (Caldwell and Sutton 1988). Following a dose of [methoxy-¹⁴C]-3-(4′-methoxyphenyl)propane (100 µg; 0.67 µmol) they excreted 4′-methoxyhippuric acid 75 (ca 11 to 13% of dose) and [¹⁴C]-CO₂ (43% of dose) but the C₆–C₃ metabolites associated with anethole 114 were not detected, strongly suggesting that hydroxylation of the benzylic carbon was the main route of metabolism for this substrate. Similar dosing with estragole 211 yielded 4′-methoxycinnamoyl-glycine 138 (0.8% of dose), 4′-methoxyhippuric acid 75 (12% of dose) plus 2-hydroxy-3-(4′-methoxyphenyl)propanoic acid 77 (4% of dose) and a C₆–C₂ metabolite (0.5% of dose) (Sangster et al. 1987). There is clear evidence of ω-oxidation of these phenylpropanoids at typical human dietary levels, possibly followed by β-oxidation, but it is also possible, except for anethole 114 that a portion of the C₆–C₁ metabolites is produced by CYP450 hydroxylation as discussed above.

This contrasts with a study by Pereira-Caro et al. which used orange juice supplying the 4′-methoxy-substituted flavanone isosakuranetin (19 µmol) and reported no excretion of any isosakuranetin phase-2 conjugates or gut microbiota metabolites with a 4′-methoxyphenyl-substitution. This suggests that demethylation by the gut microbiota was the dominant route of metabolism for this flavanone with subsequent metabolism being identical to naringenin.

These observations are consistent with the observation that MCAD binds 4′-methoxycinnamoyl-CoA 35 (Rudik et al. 2000), although the excretion of 4′-methoxycinnamoyl-glycine 76 suggests that the β-oxidative metabolism of estragole might be susceptible to overloading despite the observation that 4′-methoxycinnamoyl-CoA 35 binds to the active site of the mitochondrial enoyl-CoA hydratase (D’Ordine et al. 1994).

3.4.7. 2′,3′-Dihydroxyphenyl-, 2′,4′-dihydroxyphenyl, 2′,4′,5′-trihydroxyphenyl and 2′,4′,5′-trimethoxyphenyl substrates

For structures see Figure 14. 3-(2′,4′-Dihydroxyphenyl)propanoic acid 193 has been recorded as a gut microbiota metabolite when umbelliferone (617 µmol/kg) or herniarin (568 µmol/kg) are dosed to rats. When dosed per se, no further metabolism was observed (Indahl and Scheline 1971), indicating that it is not a β-oxidation substrate. Similarly, rats dosed with esculin (the 6-glucoside of 6,7-dihydroxy-coumarin), excreted 2′,4′,5′-trihydroxycinnamic acid 194 and 2′,4′,5′-trihydroxycinnamoyl-glycine 153 with the 5′-glucoside intact (Wang et al. 2016). These studies made no reference to any benzoic acids. Volunteers given an extract of Perilla frutescens excreted the ester glucuronide of 2′,4′,5′-trimethoxycinnamic acid 165 (Nakazawa and Ohsawa 2000). 2′,4′,5′-Trimethoxycinnamic acid 165 has been described as the major metabolite of α-asarone (2′,4′,5′-trimethoxy-prop-2-ene) (Antunez-Solis et al. 2009; Serna et al. 2015). 3-(2′,3′-Dihydroxyphenyl)propanoic acid 195 has been reported in the urine of volunteers who consumed artichoke, but its origin is not known (Dominguez-Fernandez et al. 2022).

3.4.8. 3′,4′-Dihydroxyphenyl substrates

This substrate class includes 3′,4′-dihydroxycinnamic acid (caffeic acid) 27 which occurs abundantly in the diet from numerous sources, especially coffee, and for many individuals this will be the dominant ω-phenyl-alkanoic acid 1. Structures and metabolism are shown in Figure 16. The major dietary sources of 3′,4′-dihydroxyphenyl-containing substrates are the conjugates of 3′,4′-dihydroxycinnamic acid 27 and various unabsorbed flavonoids, including procyanidins, that pass to the large bowel where the gut microbiota generate 3′,4′-dihydroxyphenyl-substituted alkanoic acids 1 plus 3′-hydroxyphenyl-substituted alkanoic acids by removing the 4′-hydroxyl (see 3.4.2), or 4′-hydroxyphenyl-substituted alkanoic acids by the less common removal of the 3′-hydroxyl (see 3.4.3).

When rats were fed 3′,4′-dihydroxycinnamic acid 27 as 1% of their diet⁷, the main urinary metabolite was 3-(3′-hydroxyphenyl)propanoic acid 14 (Figure 10), accompanied by 3′-methoxy-4′-hydroxycinnamic acid 79 (Figure 17), 3-(3′-methoxy-4′-hydroxyphenyl)propanoic acid 80 (Figure 17), 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (Figure 16), and 4′-ethyl-catechol 82 (Figure 16). When gnotobiotic rats were similarly fed, only 3′-methoxy-4′-hydroxycinnamic acid 79 was excreted, and when 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 was fed it produced only 3-(3′-methoxy-4′-hydroxyphenyl)propanoic acid 80 (Figure 17) (Peppercorn and Goldman 1972) without any evidence for hepatic β-oxidation per se. In contrast, rats dosed intragastrically with 100 mg/kg (550 µmol/kg) 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 excreted 3′-hydroxy-4′-methoxycinnamic acid 28 (Figure 18), and 3′-methoxy-4′-hydroxycinnamic acid 79 (Figure 17) in free and sulfate conjugated form, but the cinnamic acids did not proceed any further through the β-oxidation pathway, suggesting that the enoyl-CoA hydratase is not able to handle the 3′,4′-dihydroxycinnamic acid 27 (Figure 16) or its mono-methylated metabolites (Poquet, Clifford, and Williamson 2008a). When an isolated rat liver was perfused with 3′,4′-dihydroxycinnamic acid 27 (30 mg, 167 mmol) 3′-methoxy-4′-hydroxycinnamic acid 79 and 3′-hydroxy-4′-methoxycinnamic acid 28 were detected in the perfusate but the related benzoic acids (65, 10, 47) were not reported (Gumbinger, Vahlensieck, and Winterhoff 1993). Rats dosed with 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (100 μmol/kg) yielded 3-(3′-methoxy-4′-hydroxyphenyl)propanoic acid 80, 3′-methoxy-4′-hydroxycinnamic acid 79, 3′-hydroxy-4′-methoxycinnamic acid 28 plus sulfate and glucuronide conjugates of 60 and sulfate conjugates of 79 and 80. The associated benzoic acids and glycine conjugates were sought but not detected (Poquet, Clifford, and Williamson 2008a).

Rats dosed i.v. with 1,5-dicaffeoylquinic acid (160 mg/kg; 310 µmol/kg) excreted the unchanged substrate plus 14 methylated and glucuronidated metabolites. Traces of low molecular mass metabolites were observed in urine, but these were also present in the urine of animals before dosing, and these metabolites were not characterized (Yang et al. 2006). When rats were dosed with [3-¹⁴C]-3′,4′-dihydroxycinnamic acid there was no report of labeled C₆–C₁ metabolites and it is clear that if β-oxidation of the 3′,4′-dihydroxycinnamic acid moiety did occur, it was to a very limited extent (Omar et al. 2012).

Rats dosed with (+)-catechin excreted 5-(3′,4′-dihydroxyphenyl)pentanoic acid 84 accompanied by 5-(3′-hydroxyphenyl)pentanoic acid 85 (Figure 10), 3-(3′-hydroxyphenyl)propanoic acid 14 (Figure 10) and 3-(4′-hydroxyphenyl)propanoic acid 20 (Figure 11) but not 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 suggesting that β-oxidation only occurs after dehydroxylation by the gut microbiota (Das 1969). Rats dosed with [2-¹⁴C]-(–)-epicatechin excreted over 36 hours labeled C₆–C₅ metabolites equivalent to ca 21% of the dose, but only traces of the 3′-sulfate conjugate of labeled 3-(4′-hydroxyphenyl)propanoic acid 20 (Figure 11) (2.3 × 10⁻³% of the dose), the 3-sulfate conjugate of labeled 4-hydroxybenzoic acid 45 (Figure 11) (3.3 × 10⁻²% of the dose) and labeled hippuric acid 18 (Figure 3) (4.7 × 10⁻²% of the dose) clearly indicating that β-oxidation of 3′,4′-dihydroxyphenyl substrates was extremely limited (Borges, van der Hooft, and Crozier 2016).

In vitro incubations of procyanidin B3, a (+)-catechin dimer, with rat cecal microbiota for five days did not produce 3-(3′,4′-dihydroxyphenyl)propanoic acid 60, and the C₆–C₅ and C₆–C₁ metabolites reported had only 3′-hydroxylation (Groenewoud and Hundt 1986). Rats dosed orally with salvianolic acid B, a rosmarinic acid derivative, excreted 3-(3′-hydroxyphenyl)propanoic acid 14 (Figure 10) and cinnamoyl-glycine 23a (Xu et al. 2007), and rats dosed i.v. with the sodium salt of 2-hydroxy-3-(3′,4′-dihydroxyphenyl)propanoic acid 71 (15, 30 and 60 mg/kg; 68, 136, 272 µmol/kg) excreted only various phase-2 conjugated C₆–C₃ metabolites (Meng et al. 2019), with no evidence of β-oxidation in either study.

In vitro incubation of 3′,4′-dihydroxycinnamic acid 27 using rat hepatocytes (Moridani, Scobie, and O’Brien 2002) and cultured human liver-derived HepG2 cells (Poquet, Clifford, and Williamson 2008a) failed to detect 3,4-dihydroxybenzoic acid 65 and the associated hippuric acid 83 suggesting that β-oxidation of this substrate did not occur in these tissues in vitro. In vitro studies (Cao, Zhang, et al. 2009) with rat heart mitochondria treated with 3,4-dihydroxybenzoic acid 65 failed to detect 3,4-dihydroxybenzoyl-CoA 154, with metabolism proceeding via COMT 3-methylation in the cytosol yielding 3-methoxy-4-hydroxybenzoic acid 10, which diffused into mitochondria where its CoA 152 and glycine conjugates 88 were produced (Figures 16 and 17). Collectively these data indicate that 3′,4′-dihydroxyphenyl-substituted C₆–C₅ and C₆–C₃ metabolites are not subject to direct β-oxidation by rats.

In vitro human gut microbiota fermentations of 5-caffeoylquinic acid and 2-caffeoyl-isocitric acid proceeded via 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 and either 3-(3′-hydroxyphenyl)propanoic 14 (Figure 10) or 3-(4′-hydroxyphenyl)propanoic acid 20 (Figure 11), depending on the donor, but the analogous benzoic acid derivatives (46, 45) were not reported (Vollmer et al. 2017). Phenylacetic acid 12 has also been detected as a significant in vitro gut microbiota metabolite of 3′,4′-dihydroxycinnamic acid 27 (Hasyima Omar et al. 2020) and these authors noted much less 4′-dehydroxylation in one sample. A study using methyl-4,5-dicaffeoylquinate proceeded similarly and again benzoic acids were not reported (Yang et al. 2013). In vitro human gut microbiota convert rosmarinic acid 208 (Figures 14 and 16) (the 3′,4′-dihydroxycinnamoyl ester of 2R-hydroxy-3-(3′,4′-dihydroxyphenyl)propanoic acid) to 3′,4′-dihydroxycinnamic acid 27 (dominant) and 3-(3′-hydroxyphenyl)propanoic acid 14 (Mosele et al. 2014). In vitro human gut microbiota incubation of the flavanone eriodictyol yielded 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (dominant) and 3-(3′-hydroxyphenyl)propanoic acid 14, both potential candidates for β-oxidation (Mosele et al. 2014). In vitro human gut microbiota incubations of (–)-epicatechin yield C₆–C₅ metabolites well in advance of the analogous C₆–C₃ metabolites suggesting that the C₆–C₃ metabolites are produced from the C₆–C₅ metabolites rather than them arising from two independent pathways. The C₆–C₁ metabolites were not detected even at 48 hours, demonstrating that while at least some members of the human gut microbiota have β-oxidation capability with the 5-(3′,4′-dihydroxyphenyl)pentanoic acids, the derived C₆–C₃ metabolites are not direct β-oxidation substrates in vitro (Stoupi et al. 2010), but in a similar study Liu et al. reported that the same C₆–C₅ and C₆–C₃ metabolites appeared together after one hour incubation (Liu et al. 2020). During a study of human gut microbiota metabolism of apple matrix 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 and 3-(4′-hydroxyphenyl)propanoic acid 20 were produced rapidly and had declined substantially or disappeared by 10 hours, whereas 3-(3′-hydroxyphenyl)propanoic acid 14 peaked at 10 hours, and 5-(3′-hydroxyphenyl)pentanoic acid 85 and 3-(phenyl)propanoic acid 13 were still increasing at 50 hours. Only traces of 3-hydroxybenzoic 46 and 4-hydroxybenzoic acid 45 were found (Le Bourvellec et al. 2019). Apple matrix is a complex substrate rich in flavanols, proanthocyanidins and acyl-quinic acids but also containing flavonols and dihydrochalcones plus anthocyanins in some samples, and probably protein, and it is not possible unequivocally to relate all products to precursors, although the flavanols and proanthocyanidins must be the precursors of the C₆–C₅ metabolites. Collectively these in vitro studies suggest that the 3′,4′-dihydroxy-substituted C₆–C₃ metabolites are dehydroxylated rather than subject to direct β-oxidation, and only slowly transformed to 4-hydroxybenzoic 45 or 3-hydroxy-benzoic acids 46, and this conclusion is supported also by studies in volunteers.

As discussed more fully elsewhere (Clifford, Kerimi, and Williamson 2020), 3′,4′-dihydroxycinnamic acid 27 occurs in the human diet primarily as quinic acid conjugates. The associated C₆–C₁ metabolites (3-hydroxybenzoic acid 46, 4-hydroxybenzoic 45, 3,4-dihydroxybenzoic acid 65, 3-methoxy-4-hydroxybenzoic acid 10 (Figure 17), 3-hydroxy-4-methoxybenzoic acid 47 (Figure 18) and/or the corresponding hippuric acids) are rarely reported in human studies after a flavonoid-free source of acyl-quinic acids has been consumed, and in the few exceptions located, only in trace amounts (ca 0.01 to 0.2%) relative to the dose of caffeoylquinic acids (Booth et al. 1957; Gomez-Juaristi et al. 2018b; Dayman and Jepson 1969). In the study by Booth et al. in which a volunteer consumed 1 g (5.5 mmol) of 3′,4′-dihydroxycinnamic acid 27, the 3-hydroxybenzoic acid 46 did not appear in urine before 8-hours post-consumption, suggesting a gut microbiota origin of the 3-hydroxybenzoic acid 46 or its immediate precursor (Booth et al. 1957). 3,4-Dihydroxybenzoic acid 65 was not reported, and its formation and/or excretion may have been prevented by rapid microbial 4′-dehydroxylation.

When Baba et al. gave six volunteers 555 µmol rosmarinic acid 208 (Figures 14 and 16) they excreted free and phase-2 conjugated rosmarinic acid 208, free and phase-2 conjugated 3′-hydroxy-4′-methoxycinnamic acid 28 (Figure 18), and phase-2 conjugated 3′,4′-dihydroxycinnamic acid 27 with no evidence for β-oxidation of C₆–C₃ metabolites (Baba et al. 2005). In contrast, when Nurmi et al. gave volunteers an oregano extract supplying 133 µmol rosmarinic acid 208, 3,4-dihydroxybenzoic acid 65 (31 ± 7 µmol) was excreted in 48 hours (Nurmi et al. 2006), but as discussed for the 4-hydroxybenzoic acid 45 excreted simultaneously, its origin is uncertain.

In view of the strong evidence that C₆–C₃ 3′,4′-dihydroxyphenyl substrates are very poor substrates for β-oxidation, the statistical association between coffee drinking and urinary 3,4-dihydroxybenzoic acid 65 recorded in the EPIC cohort study (Zamora-Ros et al. 2016) is unexpected and difficult to explain unless by chance coffee consumers also consistently consume a commodity directly or indirectly providing the 3,4-dihydroxybenzoic acid 65, perhaps via benzylic carbon hydroxylation of a precursor substrate as discussed previously.

3-Hydroxy-3-(3′,4′-dihydroxyphenyl)propanoic acid 86 is one of several degradation products which form when eriodictyol is heated (Chaaban et al. 2017). The precise yield was not reported but is likely to be negligible during typical commercial orange juice pasteurization (for example 72 °C for 20 s) which is designed to achieve a 5 log cycle reduction in E. coli O157 (Vervoort et al. 2012), because 70 °C for 2 hours only caused a 20% loss of eriodictyol. Whether or not 3-hydroxy-3-(3′,4′-dihydroxyphenyl)propanoic acid 86 can be absorbed from the gut is not known.

When ileostomist subjects consumed coffee providing 331 µmol of 3′,4′-dihydroxycinnamic acid-containing acyl-quinic acids and associated lactones, 14.1 µmol of C₆–C₃ 3′,4′-dihydroxycinnamic acid-derived metabolites were excreted in 24-hour urine, of which ca 23% had been hydrogenated, demonstrating in the absence of any gut microbiota that hepatic “reverse” hydrogenation had occurred. In a dose–response study with ileostomist subjects who received 945, 2005 and 4088 µmol of 3′,4′-dihydroxycinnamic acid-containing acyl-quinic acids and associated lactones, the hydrogenation percentages were in the range 12.7 to 16.5%. No C₆–C₁ metabolites were found in urine or ileal fluid in either study suggesting negligible β-oxidation (Stalmach et al. 2010; Erk et al. 2012).

Hackett et al. investigated the metabolism by volunteers of [U-¹⁴C]-(+)-catechin but did not report any 3′,4′-dihydroxyphenyl metabolites. The excretion of small amounts of [¹⁴C]-labeled 3-(3′-hydroxyphenyl)propanoic acid 14 (Figure 10), 3-hydroxybenzoic acid 46 and 3′-hydroxyhippuric acid 55 suggests that the 4′-dehydroxylation was an early step in the microbial catabolism (Hackett et al. 1983). Volunteers who consumed [2-¹⁴C]-(–)-epicatechin excreted three phase-2 conjugates of 4-hydroxy-5-(3′,4′-dihydroxyphenyl)pentanoic acids 62 (Figure 16), one phase-2 conjugate of 4-hydroxy-5-(3′-hydroxyphenyl)pentanoic acid 63, five phase-2 conjugates of 5-(3′,4′-dihydroxyphenyl)-γ-valerolactones 64 and one phase-2 conjugate of 5-(3′-hydroxyphenyl)-γ-valerolactone 57 (Figure 10) which collectively accounted for 42 ± 5% of the dose (207 µmol). The only C₆–C₃ metabolite reported was 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 (Figure 10) (5.6% of dose) which was accompanied by 3′-hydroxyhippuric acid 55 (8.2% of dose) and hippuric acid 18 (12.8% of dose) (Ottaviani et al. 2016). The excretion of labeled hippuric acid 18 and 3′-hydroxyhippuric acid 55 indicate that some 21% of the substrate underwent two cycles of β-oxidation but there is no evidence for the formation of either 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 or 3,4-dihydroxybenzoic acid 65, and it appears that β-oxidation occurred only after early 4′-dehydroxylation by the gut microbiota. Accordingly, it seems unlikely that the excretion of small amounts of 3,4-dihydroxybenzoic acid 65 and associated phase-2 sulfate conjugates (less than 1% of total phenolic metabolites) observed after the consumption of orange juice is generated by β-oxidation of C₆–C₃ gut microbiota metabolites (Ordonez et al. 2018; Pereira-Caro et al. 2017).

Volunteers who on different occasions consumed pure (–)-epicatechin, procyanidin B1 and a purified procyanidin polymer fraction prepared from cocoa all excreted 5-(3′,4′-dihydroxyphenyl)-γ-valerolactone 64, 4-hydroxy-5-(3′,4′-dihydroxyphenyl)pentanoic acid 62 and 5-(3′,4′-dihydroxyphenyl)pentanoic acid 84 plus 3-(3′-hydroxyphenyl)propanoic acid 14 (Figure 10), 3-(4′-hydroxyphenyl)propanoic acid 20 (Figure 11) and 3-(3′,4′-dihydroxyphenyl)propanoic acid 60, predominantly in conjugated form(s), with yields from (–)-epicatechin considerably greater than from the dimer and polymer (Wiese et al. 2015). The yield of 3-(3′-hydroxyphenyl)propanoic acid 14 dominated the C₆–C₃ metabolites indicating the importance of 4′-dehydroxylation by the gut microbiota. The excretion of benzoic acids barely changed from control values, but hippuric acids excretion was not addressed in this study and it is not possible to judge the extent to which the second cycle of β-oxidation had proceeded. For all metabolites reported there was considerable inter-volunteer variation with the standard deviation often exceeding the mean value. A study in which volunteers consumed sous vide-cooked artichoke providing 8.62 mmol of 3′,4′-dihydroxycinnamic acid equivalents, excreted in the first four hours only 2.9 μmol free and phase-2 conjugated 3,4-dihydroxybenzoic acid 65. This was accompanied by 0.08 μmol free and 23 μmol phase-2 conjugated 3′,4′-dihydroxycinnamic acid 27, plus 0.06 μmol free and 2.5 μmol phase-2 conjugated 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (Dominguez-Fernandez et al. 2022). This indicates that the 3′,4′-dihydroxycinnamic acid 27 absorbed but not methylated by COMT is excreted 80% as phase-2 glucuronide and sulfate conjugates, some 10% diverted to “reverse” hydrogenation, with the balance passing through β-oxidation, assuming that there was no other source of 3,4-dihydroxybenzoic acid 65. The disposal of the 3′,4′-dihydroxycinnamic acid which is methylated is discussed more fully in 3.4.9 and 3.4.11.

Other studies of the consumption of green tea, black tea and cocoa similarly resulted in the excretion of C₆–C₅ metabolites with 3′,4′-dihydroxyphenyl substitution (Roowi et al. 2010; Llorach et al. 2009; Urpi-Sarda et al. 2009; van der Hooft et al. 2012; Llorach et al. 2013; van Duynhoven et al. 2014; Gomez-Juaristi et al. 2019; Urpi-Sarda et al. 2010), but the corresponding C₆–C₃ and C₆–C₁ metabolites are either not mentioned, or did not differ from control values (Urpi-Sarda et al. 2009; Gomez-Juaristi et al. 2019; Urpi-Sarda et al. 2010). 3′,4′-Dihydroxyphenylalanine (DOPA) is a minor dietary component which is also used therapeutically. It is excreted as 3′-hydroxyphenylalanine 78 (m-tyrosine) (Figure 10) (Rekdal et al. 2019; Rekdal et al. 2020) and 3′,4′-dihydroxyphenylacetic acid 94 (Eldrup et al. 1997).

There is evidence that 3′,4′-dihydroxyphenyl-substituted C₆–C₅ metabolites can be subjected to one cycle of β-oxidation at least by the gut microbiota, but the evidence for the next cycle is weak with good evidence for “reverse” hydrogenation in the liver, and it appears that gut microbiota dehydroxylation is the preferred route of metabolism followed by conjugation to sulfate or glucuronide, but not glycine. This reluctance of the 3′,4′-dihydroxyphenyl-substituted C₆–C₃ metabolites to participate in β-oxidation compared with the relative unimpeded participation of the 4′-hydroxyphenyl-substituted C₆–C₃ metabolites is presumably due to the specificity of the human enzyme. The bacterial enoyl-CoA hydratase from Pseudomonas has essentially identical activity with both 3′,4′-dihydroxyphenyl and 4′-hydroxyphenyl-substituted C₆–C₃ metabolites (Mitra et al. 1999) indicating that bacterial enzymes are often poor surrogates for the human enzyme. Note that in humans, the excretion of 3′,4′-dihydroxycinnamic acid 27 is influenced by single nucleotide polymorphisms in the ABCC2 gene encoding multidrug resistance-associated protein-2 in the kidney (Muhrez et al. 2017).

3.4.9. 3’-Methoxy-4’-hydroxyphenyl substrates

This group includes 3′-methoxy-4′-hydroxycinnamic acid (ferulic acid) 79, which is found abundantly in the diet in foods such as cereals and in coffee and many fruits. The metabolism is summarized in Figure 17. The major dietary source of 3′-methoxy-4′-hydroxyphenyl-substituted derivatives is the feruloylquinic acids augmented by the products of endogenous COMT 3′-methylation. Murine gut microbiota transformed 3′-methoxy-4′-hydroxycinnamic acid 79 to 3-(3′-methoxy-4′-hydroxyphenyl)propanoic acid 80 but after intra-peritoneal injection hydrogenation did not occur (Ohue-Kitano et al. 2019). The acyl-glucuronic acid conjugate of 3-methoxy-4-hydroxycinnamic acid 79 has been produced in vitro by mouse microsomes (Piazzon et al. 2012).

When rats were dosed with 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (18 mg,100 µmol per rat), the products included 3′-methoxy-4′-hydroxycinnamic acid 79 but this did not proceed further through the β-oxidation pathway (Poquet, Clifford, and Williamson 2008a). In an early study, rats dosed with 3′-methoxy-4′-hydroxycinnamic acid 79 (100 mg, 515 µmol per rat) were observed to excrete 3′-methoxy-4′-hydroxycinnamoyl-glycine (feruloyl-glycine) 87 and the associated 3-methoxy-4-hydroxybenzoic acid 10 and 3-methoxy-4-hydroxybenzoyl-glycine 88 (Booth et al. 1957), and despite the greater sensitivity of the LC–MS methods compared with paper chromatography, it is possible that this discrepancy can be explained by the dose given, but note that 3-methoxy-4-hydroxybenzoic acid 10 is an endogenous metabolite of adrenaline (epinephrine) and its origin in this study is uncertain.

Small amounts of 3-methoxy-4-hydroxybenzoic acid 10 and associated phase-2 sulfate conjugates (not more than 0.25% of total phenolic metabolites) have been observed in 24-hour urine after the consumption of orange juice, but their origin is uncertain (Ordonez et al. 2018; Pereira-Caro et al. 2017). These C₆–C₁ metabolites were not detected in human urine after a volunteer consumed 5-caffeoylquinic acid (1 g, 284 mmol) or 10 g instant coffee (Booth et al. 1957). The studies by Shaw and Trevarthen and Saini et al. provide the only reports of urinary 3-hydroxy-3-(3′-methoxy-4′-hydroxyphenyl)propanoic acid 120 (Shaw and Trevarthen 1958; Saini et al. 1974), but note that these studies used paper chromatography and did not have access to an authentic standard. It is now known that during coffee brewing caffeoylquinic acids add water across the 3′,4′-dihydroxycinnamic acid 27 double bond yielding (3-(3′,4′-dihydroxyphenyl)hydracryloyl)-quinic acids (Matei et al. 2016; Matei, Jaiswal, and Kuhnert 2012), and the equivalent transformation of feruloylquinic acids may be anticipated. Whether or not either hydracrylic acid moiety could be absorbed, and if so whether the 3′,4′-dihydroxyphenyl metabolite could be 3′-methylated by COMT, is unknown.

When ileostomists consumed coffee providing 48 µmol of 3′-methoxy-4′-hydroxycinnamic acid-containing acyl-quinic acids, 2.9 µmol of C₆–C₃ 3′-methoxy-4′-hydroxycinnamic acid-derived metabolites, including an unknown amount generated by COMT 3′-methylation of 3′,4′-dihydroxycinnamic acid-derived metabolites, were excreted in 24-hour urine, of which ca 28% had been hydrogenated. In a dose–response study with ileostomist subjects who received 109, 214 and 436 µmol, the hydrogenation percentages fell in the range 2.8 to 7.3%. No C₆–C₁ metabolites were found in urine or ileal fluid in either study suggesting negligible hepatic β-oxidation of 3′-methoxy-4′-hydroxycinnamic acid 79 (Stalmach et al. 2010; Erk et al. 2012).

After volunteers consumed sous vide-cooked artichoke providing ca 250 μmol flavonoids and 8.62 mmol 3′,4′-dihydroxycinnamic acid-equivalents but only 3 μmol of 3′-methoxy-4′-hydroxycinnamic acid 79, volunteers excreted in the first four hours 20 μmol of 3′-methoxy-4′-hydroxyphenyl-substituted metabolites demonstrating COMT 3′-methylation of some fraction of the 3′,4′-dihydroxycinnamic acid 27 absorbed. These metabolites included 17.9 μmol phase-2 conjugated 3′-methoxy-4′-hydroxycinnamic acid 79, 0.35 μmol free and 0.37 μmol phase-2 conjugated 3-(3′-methoxy-4′-hydroxyphenyl)propanoic acid 80 and 1.24 μmol phase-2 conjugated 3-methoxy-4-hydroxybenzoic acid 10 (Dominguez-Fernandez et al. 2022). This suggests modest “reverse” hydrogenation (4%), and probably modest β-oxidation (up to 6%) in the first four hours post-consumption, although 3′-methylation of 3,4-dihydroxybenzoic acid 65 cannot be completely excluded (Cao, Zhang, et al. 2009; Figures 16 and 17).

The excretion of 3-(3′-methoxy-4′-hydroxyphenyl)propanoic acid 80, 3′-methoxy-4′-hydroxycinnamic acid 79 and 3-methoxy-4-hydroxybenzoic acid 10 are influenced by single nucleotide polymorphisms in the ABCC2 gene encoding multidrug resistance-associated protein-2 in the kidney (Muhrez et al. 2017).

The evidence for hepatic β-oxidation of 3′-methoxy-4′-hydroxy-substituted C₆–C₃ substrates is weak with a clear preference for conjugation to glycine, sulfate or glucuronide and “reverse” hepatic hydrogenation. This reluctance of the 3′-methoxy-4′-hydroxyphenyl-substituted C₆–C₃ metabolites to participate in β-oxidation compared with the relative unimpeded participation of the 4′-hydroxyphenyl-substituted C₆–C₃ metabolites is, as discussed above, a characteristic of the human enzyme, since the Pseudomonas enoyl-CoA hydratase has essentially identical activity with both (Mitra et al. 1999) as stated previously.

3.4.10. 2′-Hydroxy-4′-methoxyphenyl substrates

For structures see Figure 14. 3-(2′-Hydroxy-4′-methoxyphenyl)propanoic acid 197 is observed as a gut microbiota metabolite of herniarin (4-methoxy-coumarin) when dosed to rats at 100 mg/kg (568 µmol/kg) (Indahl and Scheline 1971). When dosed per se it was excreted unchanged and the demethylated derivative was the only other metabolite observed (Indahl and Scheline 1971). It has not been possible to locate any human data although 2′-hydroxy-4′-methoxycinnamic acid 198 does occur in herbal Chamomile (Villa-Rodriguez et al. 2018).

3.4.11. 3′-Hydroxy-4′-methoxyphenyl substrates

The 3′-hydroxy-4′-methoxyphenyl-substituent pattern is comparatively rare with the citrus flavanone hesperidin being the major dietary source, along with the flavone diosmetin, plus the hepatic COMT metabolite 3′-hydroxy-4′-methoxycinnamic acid 28. Metabolism is summarized in Figure 18. However, when rats produced this metabolite after being dosed with 3-(3′,4′-dihydroxyphenyl)propanoic acid 60, it was diverted to phase-2 conjugation and did not proceed further through the β-oxidation pathway (Poquet, Clifford, and Williamson 2008a). Incubations in vitro with gut microbiota or recombinant Eubacterium ramulus phloretin hydrolase convert the aglycone hesperetin and structurally related neohesperidin dihydrochalcone to 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44, but 3′-hydroxy-4′-methoxycinnamic acid 28 and 3-hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89 were not reported (Honohan et al. 1976; Braune, Engst, and Blaut 2005). Bifidobacterium longum R0175 incubated with 820 nmol hesperetin produced traces of 3-(3′-hydroxy-4′-methoxyphenyl) propanoic acid 44 (4 nmol; ca 0.5%) over 48 hours in vitro accompanied by 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (39 nmol; 4.8%), 3-(3′-hydroxyphenyl)propanoic acid 14 (184 nmol; 22.4%) and 3-(phenyl)propanoic acid 13 (81 nmol; 10%) demonstrating its ability to 4′-demethylate and to dehydroxylate at both 3′ and 4′. Lactobacillus rhamnosus subsp. rhamnosus NCTC 10302 slowly produced 3-(3′-hydroxyphenyl)propanoic acid 14 (92 nmol; 11%) and 3-(phenyl)propanoic acid 13 (74 nmol; 9%) but only traces of 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (1.3 nmol; 0.15%). 3-Hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89 was not detected in this study or other similar studies with gut microbiota (Pereira-Caro et al. 2018; Honohan et al. 1976; Braune, Engst, and Blaut 2005). However, it is clear from the disappearance of the substrate (38% and 36%, respectively), that other products must have been formed by B. longum. The C₆–C₁ metabolites were sought but not found, and if produced through β-oxidation they were below the limit of detection and cannot account for this loss of substrate (Pereira-Caro et al. 2018).

A study in which rats were dosed with [3-¹⁴C]-hesperetin recorded that 40% of the dose was expired as CO₂ providing evidence for β-oxidation of at least some of the C₆–C₃ metabolites, i.e. [2-¹⁴C]-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44, [2-¹⁴C]-3-(3′-hydroxyphenyl) propanoic acid 14 and [2-¹⁴C]-3-(3′,4′-dihydroxyphenyl)propanoic acid 60, produced by the rat gut microbiota in vitro (Honohan et al. 1976).

3′-Hydroxy-4′-methoxycinnamoyl-glycine 121 has been detected in urine after volunteers consumed maté supplying 3′,4′-dihydroxycinnamic acid 27 (Gomez-Juaristi et al. 2018a). After volunteers consumed sous vide-cooked artichoke they excreted in the first four hours 5.86 μmol of 3′-hydroxy-4′-methoxyphenyl-substituted metabolites arising from COMT 4′-methylation of 3′,4′-dihydroxycinnamic acid 27. These consisted of 2.29 μmol free and 1.99 μmol phase-2 conjugated 3′-hydroxy-4′-methoxycinnamic acid 28, 1.33 μmol phase-2 conjugated 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44 and 0.25 μmol phase-2 conjugated 3-hydroxy-4-methoxybenzoic acid 47 (Dominguez-Fernandez et al. 2022) suggesting that there might be up to ca 4% β-oxidation, but as discussed in 3.4.9, 4′-methylation may have occurred after β-oxidation of 3′,4′-dihydroxycinnamic acid 27. In contrast with 3′-methylation there is a significant (39%) excretion of free 3′-hydroxy-4′-methoxycinnamic acid 28 suggesting that the 3′-hydroxyl is less easily conjugated with sulfate or glucuronide. Unfortunately, 3-hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89 was not sought in this study and its excretion may have been overlooked (Dominguez-Fernandez et al. 2022).

3-Hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89 was not produced when volunteers were given a jejunal perfusion with hesperetin-7-O-glucoside or hesperetin-7-O-rutinoside (Actis-Goretta et al. 2015), confirming that the gut microbiota generate the 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44 precursor which after absorption enters the β-oxidation pathway but does not necessarilly proceed beyond 3-hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)-propanoic acid 89 (Figure 18). Most volunteers who consumed hesperidin either pure (2 g, 3.3 mmol) or in orange juice (348, 250 or 111 µmol) excreted 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44 and 3-hydroxy-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89 (4.7, 9 or 4.6 µmol, and 72, 20 or 17 µmol, respectively, in the orange juice studies) (Booth, Jones, and DeEds 1958; Pereira-Caro et al. 2014; Pereira-Caro et al. 2017; Pereira-Caro et al. 2015). It is particularly interesting that in the 2014 study three of the 12 volunteers did not excrete detectable 3-hydroxy-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89 (Table 2), and these were also the three volunteers who did not excrete 3′-hydroxyhippuric acid 55 (Pereira-Caro et al. unpublished data). 3-Hydroxy-4-methoxybenzoic acid 47 was not detected in these studies but was subsequently reported but not quantified (Pereira-Caro et al. 2016), although its origin is uncertain and not necessarily as a result of β-oxidation. The accumulation of 3-hydroxy-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89 indicates that, for most but not necessarily all volunteers, hepatic β-oxidation is slow for this pattern of substitution, with glucuronidation of the 3′-hydroxy-4′-methoxycinnamic acid 28 intermediate by UGT1A9 in the liver being the preferred route of metabolism (Wong et al. 2010).

Table 2. Excretion, corrected for placebo, of hesperetin Phase II conjugates and 3′-hydroxy-4′-methoxyphenyl-substituted metabolites after consumption by healthy volunteers of 111 µmol (N = 15) or 348 µmol (N = 12) hesperetin in orange juice hesperetin glycosides in orange juice.

Volunteer number	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27
Hesperetin intake (µmol)	348	348	348	111	111	111	111	111	111	111	111	111	111	111	111	111	348	111	348	348	348	111	348	348	348	348	348
Hesperetin Phase II conjugates excreted (µmol/24 hours)	141.83	100.34	117.89	64.84	19.60	56.86	52.66	51.89	41.50	69.09	21.02	11.93	52.17	15.90	57.10	33.27	25.54	21.11	79.02	41.22	37.22	12.40	51.01	25.92	70.71	58.35	82.60
Nett delivery of hesperetin to colon (µmol/24 hours)	206.17	247.66	230.11	46.16	91.40	54.14	58.34	59.11	69.50	41.91	89.98	99.07	58.83	95.10	53.90	77.73	322.46	89.89	268.98	306.78	310.78	98.60	296.99	322.08	277.29	289.65	265.40
3-(3′-hydroxy-4′-methoxyphenyl) propanoic acid excreted (µmol/24 hours)	n.d.	5.82	6.42	n.d.	6.51	2.57	2.94	2.77	1.54	0.72	2.51	8.31	8.10	3.94	0.09	0.58	4.31	0.95	4.16	6.85	6.30	n.d.	3.00	5.74	7.05	6.72	n.d.
3′-hydroxy-4′-methoxycinnamic acid excreted (µmol/24 hours)				n.d.	n.d.	0.31	0.06	n.d.	n.d.	n.d.	n.d.	0.07	0.04	0.05	0.17	0.10		0.19				0.04
3-hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid excreted (µmol/24 hours)	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	0.85	3.79	7.61	8.28	12.28	16.88	22.46	23.65	51.19	55.55	68.14	80.75	88.81	92.18	94.72	96.54	101.00	101.38	186.00
Notes: n.d. = not detectable. Blank cells = metabolite not sought.

Diosmin (diosmetin-7-rhamnoglucoside), found in some citrus fruit (Zhu et al. 2020) is hydrolyzed and reduced to hesperetin (Spanakis, Kasmas, and Niopas 2009) and when given to volunteers at a dose of 10 mg/kg (ca 1.07 mmol in total for a 65 kg adult) produced the glucuronide of 3-(3′-hydroxyphenyl)propanoic acid 14 (Figure 10) (dominant), 3′-methoxy-4′-hydroxyphenylacetic acid 143, 3-hydroxy-4-methoxybenzoic acid 47 and 3,4-dihydroxybenzoic acid 65 (Figure 16) but no phenyl-hydracrylic acids 58, 89. However, the urine was acidified to pH 1, ether extracted, derivatised and analyzed by GC–MS, and it is possible that the phenyl-hydracrylic acids 58, 89 characteristic of orange juice, were destroyed during sample workup (Cova et al. 1992). The authors do not explicitly record using a standard of 3′-methoxy-4′-hydroxyphenylacetic acid 143 (Figure 14) and its presence as a metabolite of pure diosmetin is difficult to explain. Possibly it should be 3′-hydroxy-4′-methoxyphenylacetic acid 92 (Figure 18).

3.4.12. 3’,4’-Dimethoxyphenyl substrates

3′,4′-Dimethoxyphenyl-substituted derivatives are comparatively rare in the diet, and the major source must be coffee which supplies 3′,4′-dimethoxycinnamic acid 30 both in the free form and as quinic acid conjugates (Clifford et al. 2017; Clifford, Kerimi, and Williamson 2020; Clifford, Knight, et al. 2006; Clifford, Marks, et al. 2006). For structures see Figure 19.

Rats dosed with 3′,4′-dimethoxy-allylbenzene (methyl-eugenol) 116 or 3′,4′-dimethoxy-propenylbenzene (methyl-isoeugenol) 117 at 100 mg/kg (610, 610 µmol/kg) after hydrogenation to 3-(3′,4′-dimethoxyphenyl)propane 119 and ω-oxidation were both excreted as 3-hydroxy-3-(3′,4′-dimethoxyphenyl)propanoic acid 98, 3′,4′-dimethoxycinnamoyl-glycine 99 and 3,4-dimethoxybenzoyl-glycine 100. 3′,4′-Dimethoxycinnamic acid 30 (100 mg/kg 481 μmol/kg) yielded the corresponding phenyl-propanoic acid 115 and glycine conjugates 99 and 100 (Figure 19). 3′,4′-Dimethoxyphenylacetic acid 93 was produced only from methyl-eugenol but 3,4-dimethoxybenzoic acid 101 was observed after consumption of all three test compounds (30, 116, 117) after oral and intraperitoneal dosing demonstrating that production was independent of the gut microbiota. The yield of 3,4-dimethoxybenzoic acid 101 from methyl-eugenol 117 and methyl-isoeugenol 116 did not exceed 3% but the yield from 3′,4′-dimethoxycinnamic acid 30 was not stated. However, the accumulation of 3-hydroxy-3-(3′,4′-dimethoxyphenyl)propanoic acid 98 and 3′,4′-dimethoxycinnamoyl-glycine 99 again suggests that hepatic β-oxidation of 3′,4′-dimethoxycinnamic acid 30 is impeded (Solheim and Scheline 1976). A rat liver perfusion study using 3,4-dimethoxybenzoic acid 101 reported 4-demethylation yielding 3-methoxy-4-hydroxybenzoic acid 10 and 3-demethylation yielding 3-hydroxy-4-methoxybenzoic acid 47 in the ratio 15: 1 (Müller-Enoch et al. 1974).

Following consumption of coffee providing 3′,4′-dimethoxycinnamic acid 30 and 3′,4′-dimethoxycinnamoylquinic acids by eight healthy volunteers, 3′,4′-dimethoxycinnamic acid 30 and 3-(3′,4′-dimethoxyphenyl)propanoic acid 115 were observed in plasma. The authors commented “Interestingly, some volunteers displayed an early appearance [of 3-(3′,4′-dimethoxyphenyl)propanoic acid 115] in plasma within 1 h post-ingestion, accompanied by an increasing concentration at 12 h” (Farrell et al. 2012) suggesting that some but not all volunteers are capable of hydrogenating this 3′,4′-dimethoxycinnamic acid 30 in the liver. Farrell et al. did not report 3,4-dimethoxybenzoic acid 101 or its glycine conjugate 100 in plasma but did not provide any data for urine. There is thus no evidence for β-oxidation, and in the absence of O-demethylation/hydroxylation, phase-2 conjugation would be limited to the carboxyl group.

3.4.13. 3′,5′-Dihydroxyphenyl substrates

Cereal alkyl-resorcinols are the main dietary source of preexisting 3′,5′-dihydroxyphenyl-substituted substrates but substantial amounts may also be derived from the microbial 4′-dehydroxylation of 3′,4′,5′-trihydroxy-flavanols, particularly (–)-epigallocatechin-3-gallate, (–)-epigallocatechin and prodelphinidins (See Figures 8 and 20).

Cultured HepG2 cells converted the C₆–C₁₉ alkyl-resorcinol to 3-(3′,5′-dihydroxyphenyl)propanoic acid 42 (Marklund, McKeown, et al. 2013). Following consumption of 190 mg (ca 1 mmol) of mixed odd-numbered alkyl-resorcinols (C₆–C₁₇ to C₆–C₂₅, predominantly C₆–C₁₉ and C₆–C₂₁) by six healthy volunteers, all ingested alkyl-resorcinols were detected in plasma at 1-hour post-ingestion. Plasma concentrations initially peaked at ca 3 hours, and again at ca 8 hours, thought to reflect absorption in the stomach and duodenum, and lymphatic absorption from the large bowel, respectively (Landberg et al. 2006). Within two hours of consuming the alkyl-resorcinols, 5-(3′,5′-dihydroxyphenyl)pentanoic acid 41, 3-(3′,5′-dihydroxyphenyl)propanoic acid 42, 3′,5′-dihydroxycinnamic acid 96, 3,5-dihydroxybenzoic acid 43 and 3,5-dihydroxybenzoyl-glycine 118 were detected in urine after deconjugation with β-glucuronidase and sulfatase (Zhu et al. 2014; McKeown et al. 2016; Wierzbicka et al. 2017), providing convincing evidence for little if any impediment to hepatic β-oxidation of 3′,5′-dihydroxyphenyl-alkanoic acids. The same metabolites were observed with healthy ileostomist subjects, and when rats were dosed i.v. with C₆–C₁₇ and C₆–C₂₅ alkyl-resorcinols (Marklund, Stromberg, et al. 2013; Marklund et al. 2014). Although a comparatively small amount of 3′,5′-dihydroxycinnamic acid 96 was excreted, neither of the associated β-oxidation stage 2 products, 3-hydroxy-3-(3′,5′-dihydroxyphenyl)propanoic acid and 3-hydroxy-5-(3′,5′-dihydroxyphenyl)pentanoic acid, were reported in these studies. In a study seeking urinary markers of rye consumption Hanhineva et al. reported a potentially suitable but incompletely characterized carnitine metabolite, mass 325.1535 (Hanhineva et al. 2015). It yielded six positive ion fragments, including m/z 85.029 and m/z 60.081 characteristic of carnitine, plus m/z 183.065 characteristic of 3-(3′,5′-dihydroxyphenyl)propanoic acid 42, and we tentatively assign this as 3-(3′,5′-dihydroxyphenyl)propanoyl-carnitine 210 (Figure 8). The significance of this metabolite is discussed in 5.1. A sulfate of 5-(3′,5′-dihydroxyphenyl)pentanoic acid 41 and a glucuronide of 3-(3′,5′-dihydroxyphenyl)propanoic acid 42 have also been reported in human urine (Garcia-Aloy et al. 2015),⁸ and may be accompanied by other phase-2 conjugates (Bondia-Pons et al. 2013; Zhu et al. 2014). Women who participated in these studies excreted more 3,5-dihydroxybenzoyl-glycine 118, 3,5-dihydroxybenzoic acid 43, 3′,5′-dihydroxycinnamic acid 96 and 5-(3′,5′-dihydroxyphenyl)pentanoic acid 41 than men, but the same amount of 3-(3′,5′-dihydroxyphenyl)propanoic acid 42 (Landberg et al. 2018).

3′,5′-Dihydroxy-substituted C₆–C₅ metabolites are excreted after volunteers consumed grape pomace containing prodelphinidins (Castello et al. 2018) and green tea containing (–)-epigallocatechin and (–)-epigallocatechin-3-gallate (van der Hooft et al. 2012; Sang and Yang 2008; Meng et al. 2002), following removal of the 4′-OH by the gut microbiota.

3.4.14. 3′,5′-Dimethoxyphenyl substrates

This pattern of substitution is extremely rare, and structures are presented in Figure 14. Rats given 3′,5′-dimethoxycinnamic acid 167 (200 mg per rat; 961 µmol per rat) excreted the corresponding 3-(3′,5′-dimethoxyphenyl)propanoic acid 192 but the corresponding benzoic and hippuric acids were not found, indicating that β-oxidation did not occur (Griffiths 1969) It has not been possible to locate any human studies.

3.4.15. 3′,4′-Methylenedioxyphenyl substrates

3′,4′-Methylenedioxyphenyl substrates are comparatively rare in foods and have been little studied (Figure 21). Eubacterium, a species recognized as very capable of degrading flavonoids, was unable to metabolize iriflavone which has a 3′,4′-methylenedioxy B-ring (Braune et al. 2010). Rats dosed intragastrically with 1 mmol/kg of 3-(3′,4′-methylenedioxyphenyl)propanoic acid 181 excreted 3′,4′-methylenedioxycinnamic acid 182 (2.3% of dose), 3-hydroxy-3-(3′,4′-methylenedioxyphenyl)propanoic acid 183 (4% of dose) 3-keto-3-(3′,4′-methylenedioxyphenyl)propanoic acid 184 (8% of dose), 3,4-methylenedioxybenzoic acid 185 (1.2% of dose) and 3,4-methylenedioxybenzoyl-glycine 186 (71% of dose) providing convincing evidence for hepatic β-oxidation, albeit the accumulation of the stage 2 and 3 intermediates suggesting that there is significant impediment at the high doses employed. Similar dosing with 3′,4′-methylenedioxycinnamic acid 182 gave the same metabolites (Klungsoyr and Scheline 1981) but there was no evidence of “reverse” hydrogenation. It has not been possible to locate any human studies.

3.4.16. 3′,4′,5′-Tri-substituted substrates

Structures are shown in Figures 20 and 22. The main dietary sources of 3′,4′,5′-trihydroxy derivatives are flavanols, particularly (–)-epigallocatechin-3-gallate and prodelphinidins subjected to microbial catabolism. Apart from 3′,5′-dimethoxy-4′-hydroxycinnamic acid (sinapic acid) 31, preexisting 3′,4′,5′-tri substituted 3-(phenyl)propanoic and cinnamic acids are comparatively rare and minor components of the human diet, and have been little studied. The main sources of 3′,5′-dimethoxy-4′-hydroxycinnamic acid 31 seem to be blood oranges (Rapisarda et al. 1998), other citrus fruit, olives and brassicaceous vegetables such as cauliflower (Niciforovic and Abramovic 2014), and the herbal medicine Gardeniae fructus (Clifford et al. 2010).

Griffiths reported that rats given 200 mg (893 µmol) 3′,5′-dimethoxy-4′-hydroxycinnamic acid 31 excreted 3-(3′,5′-dimethoxy-4′-hydroxyphenyl)propanoic acid 134, 3-(3′-hydroxy-5′-methoxyphenyl)propanoic acid 135 and 3′-hydroxy-5′-methoxycinnamic acid 136 but C₆–C₁ metabolites were not produced (Griffiths 1969). The 4′-glucuronide and 4′-sulfate conjugates of 3′,5′-dimethoxy-4′-hydroxycinnamic acid 31 have also been detected in rat urine (Nakazawa, Yasuda, and Ohsawa 2003). Systematic studies by Scheline and colleagues of a range of tri-substituted cinnamic and 3-(phenyl)propanoic acids, including 3′,5′-dimethoxy-4′-hydroxycinnamic acid 31 dosed to rats (100 mg/kg; 417–505 µmol/kg) with and without antibiotics, or to germ-free rats, plus in vitro incubations with gut microbiota, have demonstrated hepatic dehydrogenation of the 3-(phenyl)propanoic acids and that meta-demethylation occurs only once. The gut microbiota are required for dehydroxylation and/or the second meta-demethylation, and for hydrogenation of the cinnamic acids, although trace excretion of the propanoic acid was observed when antibiotic-treated rats were dosed with 3′,4′,5′-trihydroxycinnamic acid 145. For all tri-substituted compounds studied, a 3′,5′-dihydroxyphenyl product was generated, but in contrast to the human metabolism of this substitution pattern, C₆–C₁ metabolites were not reported (Meyer and Scheline 1972a, 1972b). Glucuronide and/or sulfate conjugates were reported but not fully characterized, including, apparently, for the 3′,4′,5′-trimethoxy-substituted substrates where there would not have been a free hydroxyl available for conjugation. The stalling at stage 1 of the β-oxidation of these tri-substituted substrates is consistent with the observation by Mitra et al. that the Pseudomonas enoyl-CoA hydratase cannot accommodate such substrates (Mitra et al. 1999).

When dosed orally with the 3′,4′,5′-trimethoxy-allylbenzene (elemicin) 187 or 3′,4′,5′-trimethoxy-propenylbenzene (isoelemicin) 188 at 400 mg/kg (1.98 mmol/kg), the main β-oxidation-relevant metabolites were 3′,4′,5′-trimethoxycinnamic acid 32, 3-(3′,4′,5′-trimethoxyphenyl)propanoic acid 146 and 3-(3′,4′,5′-trimethoxyphenyl)propanoyl-glycine 149. Several 3′-demethylated metabolites were detected including 3-(3′-hydroxy-4′,5′-dimethoxyphenyl)propanoic acid 190 and 3′-hydroxy-4′,5′-dimethoxy-cinnamic acid 191, but benzoic acids were not reported. 2-Hydroxy-3-(3′,4′,5′-trimethoxyphenyl)propanoic acid (i.e. 3-(3′,4′,5′-trimethoxyphenyl)lactic acid) 189 was tentatively identified as a metabolite of elemicin for which no reference standard was available (Solheim and Scheline 1980), and which plausibly might in fact have been 3-hydroxy-(3′,4′,5′-trimethoxyphenyl)propanoic acid, the Stage 2 β-oxidation product (Figure 22).

Rats dosed with [³H-4]-(–)-epigallocatechin-3-gallate excreted [³H-3]-5-(3′,5′-dihydroxyphenyl)-γ-valerolactone 130. Incubation of (–)-epigallocatechin-3-gallate with rat cecal microbiota in vitro produced 5-(3′,4′,5′-trihydroxyphenyl)-γ-valerolactone 128 (Figure 20) but there was no evidence of direct β-oxidation, with the main pathway being 4′-dehydroxylation followed by 3′-dehydroxylation leading to 5-(3′,5′-dihydroxyphenyl)pentanoic acid 41 and 5-(3′-hydroxyphenyl)pentanoic acid 85 plus 3-(3′,5′-dihydroxyphenyl)propanoic acid 42 (Figure 8) and 3-(3′-hydroxyphenyl)propanoic acid 14 (Figure 10) (Kohri, Nanjo, et al. 2001; Kohri, Matsumoto, et al. 2001). The authors commented on significant variation in the competence of the gut microbiota from different animals (Takagaki and Nanjo 2010), and this has been demonstrated also by Liu et al. who incubated (–)-epigallocatechin-gallate with fecal samples from 14 volunteers. As for rat gut microbiota incubations 4′-dehydroxylation was more prominent than 3′-dehydroxylation (Liu et al. 2021).

After consumption of 200 g sous vide-cooked artichoke some volunteers excreted 3′,5′-dimethoxy-4′-hydroxycinnamic acid 31 but its origin is uncertain because it was not found in the artichoke (Dominguez-Fernandez et al. 2022). After consumption of an extract of Magnolia officinalis containing 3′,5′-dimethoxy-4′-hydroxycinnamic acid 31, traces of the free acid were found in β-glucuronidase-treated human urine (Nakazawa, Yasuda, and Ohsawa 2003). Jenner et al. reported that free-living volunteers excreted 3′,4′,5′-trimethoxycinnamic acid 32 (30–310 μmol/24 hours) and 3-(3′,4′,5′-trimethoxyphenyl)propanoic acid 146 (20–290 μmol/24 hours) but their origin is obscure (Jenner, Rafter, and Halliwell 2005). The only human data located for 3′,4′,5′-trihydroxy-substituted metabolites are for the gut microbiota metabolism of 3′,4′,5′-trihydroxy-substituted flavanols. 5-(3′,4′,5′-Trihydroxyphenyl)-γ-valerolactone 128 was produced from both (–)-epigallocatechin and (–)-epigallocatechin-3-gallate after incubation in vitro with human gut microbiota (Takagaki and Nanjo 2010, 2015; Liu et al. 2021), and it was excreted by volunteers who consumed green tea beverage (483 μmol 3′,4′5′-trihydroxy-flavanols) (Roowi et al. 2010; Schantz, Erk, and Richling 2010) along with 4-hydroxy-5-(3′,4′,5′-trihydroxyphenyl)pentanoic acid 129 (van der Hooft et al. 2012). Consistent with these observations, Mena et al reported the excretion of an incompletely characterized 5-(methoxy-hydroxyphenyl)-γ-valerolactone-sulfate after volunteers consumed tablets supplying green tea for eight weeks (9.3. mmol 3′,4′,5′-trihydroxy-flavanols) (Mena, Ludwig, et al. 2019) indicative of endogenous COMT methylation at an undefined location. There have been no reports of 3′,4′,5′-trihydroxy C₆–C₃ metabolites and it seems that dehydroxylation, particularly 4′-dehydroxylation, is a major route of gut microbiota catabolism (Takagaki and Nanjo 2015; Liu et al. 2021), yielding 3-(3′,5′-dihydroxyphenyl)propanoic acid 42 (Figure 8) and 3-(3′-hydroxyphenyl)propanoic acid 14 (Figure 10) which have been discussed in 3.4.13 and 3.4.2, respectively.

4. Enzymic transformations ancillary to Β-oxidation

4.1. Amino acid conjugation

In the studies discussed in Parts 2 and 3, a C₆–C₁ or a C₆–C₃ amino acid conjugate is sometimes reported, but this is not inevitable. While in some studies these conjugates might have been overlooked, it is appropriate to look more critically at this conjugation because it could influence how particular ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic acids 2 are handled. Across the animal kingdom many acidic xenobiotics are excreted as amino acid conjugates, with significant inter-species differences. Overall, humans use predominantly glycine, although glutamine dominates for some metabolites, with a minor contribution from taurine. Amino acid conjugation in rats is significantly different from humans and the relevance of rat studies is unclear.

Glycine conjugation was first described in 1845 (Dessaignes 1845) but in a review published March 2021 Rohwer et al. drew attention to a lack of knowledge about the substrate specificity of the enzymes involved (Rohwer, Schutte, and van der Sluis 2021). This conjugation system located within mitochondria has been studied primarily in the context of xenobiotic detoxification but there have been few studies of the conjugation of ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic acids 2. The first step is linking the xenobiotic to adenylate via ATP and the release of inorganic phosphate, followed by conjugation to coenzyme A with the release of AMP, and finally conjugation with glycine and release of coenzyme A. Note that potential substrates for glycine conjugation, i.e., those departing from the β-oxidation system as distinct from preexisting benzoic and cinnamic acids present as such in the diet, are already CoA thioesters, and it is accessibility to the amino acid-conjugating enzyme which determines whether or not a conjugate is formed.

4.1.1. Activation of ω-Phenyl-alkanoic 1 and ω-Phenyl-alkenoic acids 2

There are two human hepatic medium chain (butyryl) acyl-CoA ligases (synthetases), HXM-A (EC 6.2.1.2 also known as ACSM2B) and HXM-B (EC 62.1.1 also known as ACSM1), with HXM-A primarily concerned with activation of xenobiotics and HXM-B with short chain fatty acids. Toxicological studies have focused on a few xenobiotic C₆–C₁ and C₆–C₂ metabolites, but C₆–C₃ substrates and the metabolites of dietary phytochemicals, as distinct from drugs, have received little or no attention. HXM-A has a much greater affinity for benzoic acid 11 than 2-hydroxybenzoic acid 49 and phenylacetic acid 12, and benzoic acid 11 can inhibit the clearance of 2-hydroxybenzoic acid 49 following use of aspirin (van der Sluis and Erasmus 2016). Kasuya et al. reported that a medium chain acyl-CoA synthetase preparation from bovine liver mitochondria, incompletely defined and possibly a mixture, was inhibited in vitro by 2-hydroxybenzoic acid 49 (K_i = 37 µM) (Kasuya et al. 1996). A preparation from rat liver mitochondria described as benzoyl-CoA synthetase was not inhibited in vitro by 3-(3′-hydroxyphenyl)propanoic acid 14 (Figure 10) but it was not a substrate which explains why this metabolite is excreted without conjugation to glycine (Phipps, Stewart, and Wilson 1997; Phipps et al. 1998). Substrates of the bovine liver enzyme have been described as containing “a flat hydrophobic region coplanar to the carboxylate group” and require a negative charge on C-3 of the side chain, and the presence of a C-3 hydroxyl or keto group is inhibitory (Kasuya et al. 1998).

Although it is not known which enzyme is responsible, it is clear from the studies reported in Part 3 that human ligase enzymes can accommodate a much wider range of ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic acids 2 because the associated amino acid conjugates have been reported. These include 3-(phenyl)propanoic acid 13 (Figure 3), 3-hydroxybenzoic acid 46 (Figure 10), 4-hydroxybenzoic acid 45 (Figure 11), 4-methoxybenzoic acid 74 (Figure 15), 3,5-dihydroxybenzoic acid 43 (Figure 8), 3-methoxy-4-hydroxybenzoic acid 10 (Figures 7 and 17), 3′-methoxy-4′-hydroxycinnamic acid 79 (Figures 7 and 17) and trans- and cis-cinnamic acid 21a, 21b (Figure 3), plus some C₆–C₅ metabolites and analogs with even longer sidechains in the case of 3′,5′-dihydroxy-phenyl-ring metabolites. Similarly, and as discussed below in more detail, some C₆–C₄ and C₆–C₂ metabolites are also accommodated. In addition bovine liver mitochondrial medium chain acyl-CoA ligases also activate 3-methoxybenzoic 203 and 3-hydroxy-4-methoxybenzoic acid 47 but not 2-methoxybenzoic acid 147 or 2-hydroxybenzoic acid 49 (Kasuya, Igarashi, and Fukui 1990).

Formation of the acyl-CoA-thioester does not automatically assure conjugation to an amino acid and there is evidence from the study of various xenobiotics to indicate that when glycine conjugation does not occur, the acyl-CoA thioester may disturb mitochondrial function through CoA sequestration. Although CoA pools may be limited (Knights and Vessey 2010), there is no clear evidence of this happening with the dietary ω-phenyl-alkanoic acids 1 and ω-phenyl-alkenoic acids 2, even those that do not pass quickly through β-oxidation.

4.1.2. Glycine and glutamine conjugation

Glycine conjugation utilizes a mitochondrial acyl-CoA:glycine N-acyltransferase (GLYAT) (EC 2.3.1.13 = 2.3.1.71) (Knights, Sykes, and Miners 2007; Knights and Vessey 2010) or the GLYAT-like enzyme GLYATL1 (EC 2.3.1.14 = 22.3.1.68) (Knights and Vessey 2010) which conjugates with glutamine. Some studies appear to have used a mixture of both enzymes. It has long been said that GLYAT and GLYATL1 are restricted to the liver and kidney where activity with phenylacetic acid 12, benzoic acid 11 and 2-hydroxybenzoic acid 49 is higher than in the liver (Kelley and Vessey 1993). More recent studies show that the gene encoding GLYATL1 is expressed to a lower extent in 18 human tissues including prostate, testis, ovary, pancreas and stomach (Zhang et al. 2007).

GLYAT shows kinetic cooperativity rather than the Michaelis–Menten reaction mechanism and published K_m values do not directly reflect affinity (van der Sluis et al. 2017). There are three GLYAT haplotypes and in vitro kinetic analysis of the 156Asn > Ser,199Arg > Cys haplotype suggests that these individuals may have a reduced ability to metabolize benzoic acid 11 compared with the most common haplotype, 156Asn > Ser. Limited GLYAT activity, under some circumstances, could lead to accumulation of acyl-CoAs, inhibition of the medium chain acyl-CoA ligase HXM-A, and energy production in the mitochondria (Rohwer, Schutte, and van der Sluis 2021). GLYAT conjugation of benzoic acid 11, but not 2-hydroxybenzoic acid 49, can be increased by glycine supplementation (Amsel and Levy 1969).

Five GLYAT single nucleotide polymorphisms have been identified in the Japanese population (Yamamoto et al. 2009) and seven in a French Caucasian population (Cardenas et al. 2010). An investigation of human liver and kidney biopsy samples observed significant variation in the ability to conjugate benzoic acid 11, with rates of 254 ± 90.5 nmol min⁻¹ per g liver (N = 110) and 321 ± 99.3 nmol min⁻¹ per g kidney (N = 67) (Temellini et al. 1993).

Human hepatic GLYAT has much lower activity in vitro with 2-hydroxybenzoyl-CoA (22 ± 7%) compared with benzoyl-CoA (Kelley and Vessey 1994). Males are better able to clear 2-hydroxybenzoic acid 49 as a consequence of having a 60% greater capacity for glycine conjugation than females for this substrate, although this advantage is reduced to 20% in females taking oral contraceptive steroids (Miners et al. 1986). This difference in metabolism was confirmed by Navarro et al. although their data, while still statistically significant (p < 0.05), recorded a less pronounced gender difference. The percentage of the dose of aspirin (650 mg; 3.6 mmol) excreted as 2′-hydroxyhippuric acid 102 was 84.7 ± 0.3% for males (N = 264) and 82.4 ± 0.3% for females (N = 264). The molar ratio of 2′-hydroxyhippuric acid 102 excreted relative to 2-hydroxybenzoic acid 49 was 5.06 for males and 4.30 for females (Navarro et al. 2011). In a small study involving 10 males and 10 females, Stanislaus et al. also observed gender differences in excretion of many glycine conjugates, but there was no consistent pattern (Stanislaus, Guo, and Li 2012). Landberg et al. observed that female volunteers excreted more 3,5-dihydroxybenzoyl-glycine, 3,5-dihydroxybenzoic acid, 3′,5′-dihydroxycinnamic acid 96 and 5-(3′,5′-dihydroxyphenyl)pentanoic acid 41 than male volunteers, but the same amount of 3-(3′,5′-dihydroxyphenyl)propanoic acid 42 (Landberg et al. 2018), suggesting that these gender differences are not necessarily restricted to the glycine conjugation step. So far as we are aware this gender difference in amino acid conjugation has never been examined with respect to the major dietary ω-phenyl-alkanoic acids 1 and ω-phenyl-alkenoic acids 2.

GLYAT and GLYATL1 expression are significantly reduced in all hepatocellular carcinomas but not other liver diseases (Matsuo et al. 2012; Guan et al. 2020), and GLYATL1 is over-expressed in prostate cancer (Eich et al. 2019), and phenylacetyl-glutamine excretion 110 is elevated in phenylketonuria (James and Smith 1973). Glutamine conjugation seems to be restricted to C₆–C₂ and C₆–C₄ ω-phenyl-alkanoic acids 1 (Table 1) although such conjugates may have been overlooked. A human study using [¹⁴C]-labeled phenylacetic acid 12 (80 mg/kg, 588 μmol/kg) yielded in 24 hours the glutamine conjugate 110 (Figure 3) 93%, with <0.05% conjugated to glycine 95, and 7% to taurine (James et al. 1972; James and Smith 1973), but phenylacetylglycine 95 (35–170 µmol/mol creatinine)⁹ has been quantified by LC–MS in the urine of healthy free-living adults (Stanislaus, Guo, and Li 2012). Phenylacetic acid 12 may also be conjugated to glucuronic acid and l-carnitine (James et al. 1972; Kasumov et al. 2004), but interestingly free phenylacetic acid 12 has been reported in the urine of female children but not male children (Caterino et al. 2020).

4′-Hydroxyphenylacetic 70 also may be conjugated to glycine 97 or glutamine 90 (Figure 5) (Ju et al. 2016; James et al. 1972). In contrast 4′-methoxyphenylacetic acid 103 and 4-(phenyl)butanoic acid 15 may be conjugated to glutamine 126 or glucuronic acid 127 (Oakley and Seakins 1971; Comte et al. 2002; Kasumov et al. 2004). There is some evidence that glucuronidation occurs when glycine is limiting but glucuronidation occurs in the endoplasmic reticulum and utilizes the free acid rather than the acyl-CoA with conjugation in the mitochondria and the factors controlling which route is followed remain unclear. Several hippuric acids may undergo further phase-2 conjugation discussed in 4.3.

ω-Phenyl-alkanoic acids 1 with substitution at C2 of the side chain, for example 2-hydroxy-3-(phenyl)propanoic acids (3-phenyl-lactic acids) such as 66, 71, 77, 113, 189, 199, and the 2R-(phenyl)propanoic acids (e.g. 56) derived from isoflavones, are not substrates for glycine conjugation (Knights and Vessey 2010). A study in which rats were dosed with 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 (100 μmol/kg) sought but did not find its glycine conjugate (Poquet, Clifford, and Williamson 2008a).

It is clear from the human studies discussed in the preceding sections 3.4.1–3.4.16 that humans can also conjugate 3-(phenyl)propanoic acid 13, trans- and cis-cinnamic acid 21a, 21b, 3-methoxy-4-hydroxycinnamic acid 79, 3′-hydroxy-4′-methoxycinnamic acid 28, 3-hydroxybenzoic acid 46, 4-hydroxybenzoic acid 45, 4-methoxybenzoic acid 74, 3,5-dihydroxybenzoic acid 43 and 3-methoxy-4-hydroxybenzoic acid 10, in vivo. It has not been possible to locate any human studies referring to glycine conjugation of 3,4-dihydroxybenzoic acid 65, 3,4-dimethoxybenzoic acid 101, 3′,4′-dihydroxycinnamic acid 27 or 3′,4′-dimethoxycinnamic acid 30. Lewis-Stanislaus et al. tentatively identified four dihydroxyhippuric acids (M_r = 211) and three hydroxyphenylpropanoyl-glycines (M_r = 223) in human urine for which the hydroxyl locations are unknown (Lewis-Stanislaus and Li 2010). These hippuric acids were not referred to in their subsequent paper (Stanislaus, Guo, and Li 2012) and so much uncertainty remains.

Despite clear evidence that humans can conjugate 4-hydroxybenzoic acid 45 with glycine, when Nurmi et al. gave volunteers an oregano extract, the main metabolite excreted over 48 hours was 4-hydroxybenzoic acid 45 (257 ± 83 µmol out of 575 µmol total phenols) (Nurmi et al. 2006). 4-Hydroxybenzoic acid 45 excretion peaked in the first four hours post-dosing and was thought to be derived primarily from rosmarinic acid 208 (Figures 11 and 14), but 4′-hydroxyhippuric acid 68 was not referred to. It is impossible to judge whether 4′-hydroxyhippuric acid 68 was absent or merely overlooked, but it is abundantly clear that a substantial amount of 4-hydroxybenzoic acid 45 was not conjugated with glycine. This is in marked contrast to the results of an orange juice study where 11 out of 12 volunteers showed a significant increase in the excretion of 4′-hydroxyhippuric acid 68 (21 ± 2 compared with 2.9 ± 0.3 µmol/24 hours) and only four out of 12 volunteers showed a slight increase in the excretion of 4-hydroxybenzoic acid 45 relative to placebo (3.1 ± 0.6 and 3.6 ± 0.4 μmol/24 hours) (Pereira-Caro et al. 2014). There is no certain explanation for this very different behavior, but possibly the generation of nearly two orders of magnitude more 4-hydroxybenzoic acid 45 after oregano compared with orange juice exceeded the glycine available—glycine supplementation has been reported to increase hippuric acid excretion (Amsel and Levy 1969). The twelfth volunteer who after orange juice did not excrete 4′-hydroxyhippuric acid 68 had a significantly increased excretion of 3′-hydroxyhippuric acid 55 (3.8 µmol/24 hours compared with undetectable) but no change in the excretion of hippuric acid 18 relative to placebo (Pereira-Caro et al. 2014). The extreme inter-person variation in 3′-dehydroxylation relative to 4′-dehydroxylation is also well known (Mansoorian et al. 2019).

Unlike humans, rats conjugate phenylacetic acid 12 with glycine and it is uncertain to what extent the rat is a good model for human conjugation with amino acids. 3-(3′-Hydroxyphenyl)propanoic acid 14 did not inhibit GLYAT conjugation of benzoic acid 11 in rat liver mitochondria (Phipps, Stewart, and Wilson 1997; Phipps et al. 1998). Rat liver mitochondria can generate 3-methoxy-4-hydroxybenzoyl-glycine 88 and traces of 3-hydroxy-4-methoxybenzoyl-glycine 81 from preexisting 3,4-dihydroxybenzoic acid 65 via COMT methylation,¹⁰ but 3,4-dihydroxybenzoyl-glycine 83 was not detected (Cao, Zhang, et al. 2009), and this is consistent with volunteers excreting 3-methoxy-4-hydroxy-benzoyl-glycine 88 but not 3,4-dihydroxybenzoyl-glycine 83 after consuming Ginkgo biloba extracts containing 3,4-dihydroxybenzoic acid 65 (Pietta, Gardana, and Mauri 1997). Studies with rats discussed in part 3, albeit with very high doses of test compound, have reported also the excretion of 2′-hydroxycinnamoyl-glycine 51, 3-(2′-hydroxyphenyl)propanoyl-glycine 52, 2′,4′,5′-trihydroxycinnamoyl-glycine 153, 4′-hydroxycinnamoyl-glycine 67, 2′-methoxycinnamoyl-glycine 150, 3-(2′-methoxyphenyl)propanoyl-glycine 151, 3′,4′-dimethoxycinnamoyl-glycine 99, 3,4-dimethoxybenzoyl-glycine 100 and 3-(3′,4′,5′-trimethoxyphenyl)propanoyl-glycine 149. All of these with the possible exception of 4′-hydroxycinnamoyl-glycine 67 are formed from substrates which are very minor components of the human diet and hence may well have been overlooked in human studies, but equally at the much lower levels present in the human diet it is possible that with the exception of 3′,4′-dimethoxycinnamic acid 30 some might have proceeded further along the β-oxidation pathway prior to excretion.

4.2. Hydrolysis of the CoA conjugates

Coenzyme A is recycled during normal metabolism, and any impairment to its recycling can interfere with β-oxidation and other critical biochemical transformations. Human peroxisomal thioesterase-2 is considered important in deactivating slowly metabolized xenobiotic acyl-CoAs, for example acyl-CoAs that cannot be conjugated to glycine, thus preventing the sequestration of CoA required for other tasks (Hunt et al. 2002). Human thioesterase 2 in vitro accepts 3′-hydroxyphenylacetyl-CoA 105, 3′,4′-dihydroxyphenylacetyl-CoA 106, 3′,5′-dihydroxyphenylacetyl-CoA 155, 4′-hydroxyphenylacetyl-CoA 107, phenylacetyl-CoA 104 and 3-hydroxybenzoyl-CoA 108 listed in order of declining k_cat/K_m but had no activity with 4-hydroxybenzoyl-CoA 109 (Cao, Xu, et al. 2009; Cheng et al. 2006). This or a similar enzyme must in vivo be able to hydrolyze the CoA-conjugates of the ω-phenyl-alkanoic acid β-oxidation intermediates, for example 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 and 3-hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89, which are excreted in urine.

4.3. Glucuronide and sulfate conjugation of hippuric acids

The excretion of the glucuronide of 2′-hydroxyhippuric acid 102 (Figure 9), and the sulfates of 3′-hydroxyhippuric acid 55 (Figure 10) and 4′-hydroxyhippuric acid 68 (Figure 11), have been confirmed by NMR and MS or the use of authentic standards (Monti et al. 1985; Correia et al. 2020; Zimmerman et al. 1981), and may have been overlooked. The excretion of the acyl glucuronide of glycine has been reported (Egi 1963) and hence the tentative identification of an acyl glucuronide of hippuric acid 205 (Figure 3) is plausible, but the acyl diglucuronide and mixed anhydride hippuric acid sulfate are less convincing (Agullo et al. 2020; Correia et al. 2020). These structures were assigned by matching the accurate mass of parent ion and associated fragments to records previously entered in a database. As discussed elsewhere any matches must be treated as suggestions, critically evaluated and treated as tentative until confirmed by further studies (Kuhnert and Clifford 2022), particularly for the mixed anhydride hippuric acid sulfate which would be unstable in aqueous solution releasing sulfuric acid.

5. Human metabolism: matters arising

The preceding section summarizes the data relevant to the β-oxidation of ω-phenyl-alkanoic acids 1 and ω-phenyl-alkenoic acids 2 derived from studies in vitro, animal studies and human intervention studies. Using these data this section draws attention to gaps in knowledge and understanding of their β-oxidation in humans.

5.1. Influx to and efflux from the mitochondrion

Although β-oxidation may begin in the peroxisome, the final stages occur in the mitochondrion and the smaller substrates must either be produced in the mitochondrion or enter from the cytosol, after efflux from the peroxisome.

Short and medium chain fatty acids (C₄ up to ca C₁₀) enter the mitochondrion either as free fatty acids by diffusion or as carnitine conjugates after cytosolic activation, this being essential for the transfer of long chain fatty acids across the inner mitochondrial membrane. Yamada et al. reported that compared with C₆–C₆ to C₆–C₁₂ unsubstituted ω-phenyl-alkanoic acids 1 the C₆–C₄ analogues are poorly metabolized in rat liver peroxisomes in vitro and that carnitine is essential for efficient mitochondrial uptake of C₆–C₄ to C₆–C₁₂ unsubstituted ω-phenyl-alkanoic acids 1 by rat liver mitochondria in vitro. In the absence of carnitine, C₆–C₄ uptake is reduced to ca 40%, negligible for C₆–C₆ and C₆–C₈ and undetectable for those with a longer sidechain (Yamada et al. 1987). In contrast, Mao et al. reported that 3-(phenyl)propanoic 13, and 4-(phenyl)butanoic acids 15 entered rat liver mitochondria without carnitine conjugation but paradoxically, in vitro their β-oxidation is suppressed when carnitine is present (Mao, Chu, and Schulz 1994), and the same behavior has been reported for cinnamic acid 21a (Zhao et al. 2019). Carnitine also has a role in the removal of abnormal organic acids associated with acidemias (Longo, Filippo, and Pasquali 2006; Longo, Frigeni, and Pasquali 2016) and possibly carnitine suppresses β-oxidation by facilitating their removal from the mitochondrion. This is consistent with the excretion 3-(phenyl)propanoyl-carnitine 156 by a patient with MCAD deficiency where its β-oxidation is not possible, but not by a healthy control (Moore et al. 1990).

This is not obviously the explanation for the urinary excretion of 3-(3′,5′-dihydroxyphenyl)propanoyl-carnitine 210 by healthy free-living individuals following their normal diet (Hanhineva et al. 2015). We suggest tentatively that this is analogous to the excretion of fatty acyl-carnitines with carnitine conjugation facilitating the transfer of alkyl-resorcinol-derived 3-(3′,5′-dihydroxyphenyl)propanoic acid 42 from the peroxisome to the mitochondrion.

If sufficiently hydrophobic ω-phenyl-alkanoic acids 1 and ω-phenyl-alkenoic acids 2 might enter by diffusion but experimental data are scarce. The excretion of the glycine conjugates of 4′-methoxycinnamic 72, 3′-hydroxy-4′-methoxycinnamic 79 and 3′-hydroxy-4′-methoxycinnamic acid 28 is proof that these cinnamic acids enter the mitochondrion, but the mechanism is not known. The uptake of anionic drugs by rat brain mitochondria is dependent on the experimental pH 7.4 Log D value increasing exponentially from −0.98 to 2 (Durazo et al. 2011). Accordingly, the more hydrophobic longer sidechain acids will diffuse more efficiently than the shorter sidechain acids. It has not been possible to locate experimental pH 7.4 Log D values for molecules relevant to this review and calculated Log D values for anionic molecules are known to be imprecise (Wenlock, Barton, and Luker 2011). It is thus impossible to predict whether C₆–C₅ and/or C₆–C₃ ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic acids 2 enter the mitochondrion by diffusion. However, Cao et al. have stated that 3-methoxy-4-hydroxybenzoic acid 10 does (Cao, Zhang, et al. 2009) and data from rat studies (see 3.4.2) suggest that 3-(3′-hydroxyphenyl) propanoic acid 14 can access the mitochondrion by an undefined mechanism, while 3′-hydroxycinnamic acid 19 might be unable to. The reported mitochondrial influx of relatively hydrophilic C₆–C₁ to C₆–C₄ metabolites without carnitine conjugation is not consistent with a diffusion mechanism.

Durazo et al. have suggested that anionic drugs might enter the mitochondrion via anion transporters (Durazo et al. 2011). Subsarcolemmal mitochondria possess the MCT1 monocarboxylate transporter (Benton et al. 2004), which in other tissues transports benzoic 11, 2-hydroxybenzoic 49, 3-hydroxybenzoic 46, 4-hydroxybenzoic acid 45 (Haughton, Clifford, and Sharp 2007), 3′-methoxy-4′-hydroxycinnamic acid 79 (Ziegler et al. 2016; Poquet, Clifford, and Williamson 2008b) and 4-(phenyl)butanoic acid 15 (Lee and Kang 2016; Gyawali and Kang 2021) suggesting that active transport could be important. Note that some mitochondria, for example intermyofibrillar mitochondria, do not possess MCT1 (Durazo et al. 2011) and clearly further investigation is required to clarify the influx and efflux of these acids.

5.2. Effects of substrate structure on β-oxidation

5.2.1 3’,5’-Di- and 3’,4’,5’-tri-substituted metabolites

There are significant differences in the handling of 3′,5′-dihydroxyphenyl- and 3′,4′,5′-trihydroxyphenyl-substituted ω-phenyl-alkanoic acids 1. As discussed in section 3.4.13, long chain ω-phenyl-alkanoic acids with 3′,5′-dihydroxyphenyl substitution are detected in plasma within 1 hour and within two hours as 3,5-dihydroxybenzoic acid 43 and 3,5-dihydroxybenzoyl-glycine 118, accompanied by small amounts of 3′,5′-dihydroxycinnamic acid 96 (Figure 8) along with phase-2 conjugates of C₆–C₅ and C₆–C₃ metabolites which have not been quantified. Despite some diversion to phase-2 conjugation there is little if any impediment to their hepatic metabolism at intakes above the normal dietary level. Gut microbiota metabolism of alkyl-resorcinols seems not to have been investigated in vitro but there have been no reports of their phenyl ring dehydroxylation in vivo, probably reflecting their limited contact with the gut microbiota compared with the flavanols.

5-(3′,4′,5′-Trihydroxyphenyl)propanoic acid 61 (Figure 22) is a gut microbiota metabolite of flavanols and proanthocyanidins which is predominantly phase-2 conjugated. There is no evidence for the direct β-oxidation of the unconjugated fraction (3.4.16). Its gut microbiota 4′-dehydroxylation is well documented but despite numerous extensive volunteer studies on metabolism of green tea flavanols there have been no reports of 3-(3′,5′-dihydroxyphenyl)propanoic acid 42, 3,5-dihydroxybenzoic acid 43 or 3,5-dihydroxybenzoyl-glycine 118 (Figure 8) in plasma or urine after green tea consumption, in marked contrast to the observations after volunteers have consumed alkyl-resorcinols. Instead, a second gut microbiota dehydroxylation diverts this flavanol metabolite to 3′-hydroxyphenyl-substituted metabolites as discussed in 3.4.3 and 3.4.16. Further support for this divergent behavior is provided by a study in which a volunteer consumed a supplement containing green tea flavanols—there was no increase in fecal 3,5-dihydroxybenzoic acid 43 but a ca six-fold increase in fecal 3-(3′-hydroxyphenyl)propanoic acid 14 (Sanchez-Patan et al. 2011).

5.2.2. Phenylvalerolactones and 4-hydroxy-5-(phenyl)pentanoic acids

Phenylvalerolactones and 4-hydroxy-5-(phenyl)pentanoic acids are major characteristic gut microbiota metabolites of flavanols and proanthocyanidins. Volunteer data indicate that both are predominantly excreted as phase-2 conjugates (Table 3). Even when the phenyl-ring substituents are the same, the flavanol-derived 5-(phenyl)pentanoic acid metabolites differ from the alkyl-resorcinol-derived metabolites by the presence of a C-4 sidechain hydroxyl, and it is this structural feature that allows the formation of the 5-(phenyl)-γ-valerolactones. The formation of a five-membered ring is a highly favored 5-exo-trig reaction according to Baldwin rules (Baldwin 1976), occurring rapidly at acidic or slightly alkaline pH values, and it is generally assumed that the 4-hydroxy-5-(phenyl)pentanoic acids lactonize spontaneously (Mena, Bresciani, et al. 2019).

Table 3. Excretion by rats and humans of each subgroup of C₆–C₅ flavanol metabolites expressed as a percentage of the total C₆–C₅ metabolites.

	Rats			Humans (Castello et al. 2018)
	[2-^14C]-(–)-epicatechin	Red wine extract	Grape seed phenolics	Cocoa	Cocoa enriched in flavanols	Apples	Grape pomace
Conjugated (phenyl)-γ-valerolactones	23.9	38.8	14.8	83.9	79.1	97.7	99.7
Free (phenyl)-γ-valerolactones	21.2			0.94	2.1		0.18
Conjugated 4-hydroxy-5-(phenyl) pentanoic acids	54.9	18.7	25.5	15.2	18.7	1.9	0.11
Free 4-hydroxy-5-(phenyl)pentanoic acids		0.12	0.15	n.d.	n.d.
Conjugated 5-(phenyl)pentanoic acids		41.9	59.3
Free 5-(phenyl) pentanoic acids		0.3	0.26

Under physiological conditions of temperature and pH the rates of equilibration of the 4-hydroxypentanoic acid–γ-valerolactone couple are too slow to be conveniently investigated. However, human serum paraoxonase (PON1), which is secreted by the liver and associates with high density lipoprotein where it hydrolyzes lipid peroxides, is able in vitro to reversibly interconvert γ-valerolactone 160 and 4R/S-hydroxy-pentanoic acid 161, or coumarin 139 and cis-2′-hydroxycinnamic acid 53a at pH 6.0–6.5, establishing that the sidechain can be accommodated and suggesting that the phenyl residue may be tolerated (Figure 14). Lactonization is stimulated by Ca²⁺ and proceeds readily in vitro provided the available substrate acid concentration is greater than the equilibrium value. It is not stereo-specific for R/S orientation of the of the C-4 hydroxyl. At pH values above the hydroxy-acid pK_a there will be more of the ionized form, and this is not available for lactonisation. The pK_a values for the 4-hydroxy-5-(phenyl)pentanoic acids have not been reported but 4-hydroxypentanoic acid has pK_a = 5.7 and therefore the Henderson–Hasselbalch equation predicts that at pH 7.4 it would be 98% ionized. In keeping with this prediction, neither coumarin nor γ-valerolactone hydrolyzed spontaneously but in the presence of PON1, equilibrium was reached quite quickly with 34.5% lactone at pH 6, 14.2% at pH 6.5 and about 1% at pH 7.5 for γ-valerolactone (Teiber, Draganov, and La Du 2003; Shunmoogam, Naidoo, and Chilton 2018). The equilibrium for the 5-(phenyl)-γ-valerolactones has not been reported, but in the absence of a significant electronic effect attributable to the phenyl substituent, it is unlikely to be greatly different. However, many samples of plasma, urine and feces, and samples from in vitro incubations, have been acid treated during work-up to purify and concentrate the analytes of interest and it is possible that the equilibrium occurring in vivo has been disturbed in favor of the γ-lactones, and this possibility must be borne in mind when interpreting the following data

Feeding 4′-hydroxycinnamic acid 29 or 3-(4′-hydroxyphenyl)propanoic acid 20 to rats at 135 µmol/100 g diet¹¹ raises plasma paraoxonase (Lee et al. 2003), but these doses are well in excess of typical dietary intakes. However, Haldari et al. reported that dosing orally with hesperetin (5 mg/kg; 16.5 µmol/kg) also raised plasma paraoxonase in hyperuricemic rats, this dose corresponding to 325 µmol for a 65 kg adult, and thus not greatly different to the doses given as orange juice to volunteers, for example 111, 250 and 348 µmol (Pereira-Caro et al. 2014; Pereira-Caro et al. 2016; Pereira-Caro et al. 2015, 2017).

In all studies on volunteers (Table 3), phase-2 conjugated 5-(phenyl)-γ-valerolactones dominate the C₆–C₅ metabolites excreted in urine with the 4-hydroxy-5-(phenyl)pentanoic acids sometimes undetectable but sometimes accounting for up to 3.8%, and the 5-(phenyl)pentanoic acids for some 0.1 to 0.35%, but also often not reported (Wiese et al. 2015; Cortes-Martin et al. 2019; Castello et al. 2018; Anesi et al. 2019; Gomez-Juaristi et al. 2019; Borges et al. 2018). Based on 24-hour plasma area-under-the-curve data, the 4-hydroxy-5-(phenyl)pentanoic acids accounted for not less than 5.9%, but it is not possible to give a precise figure because some have been quantified with 5-(phenyl)-γ-valerolactones. Neither study reported 5-(phenyl)pentanoic acids in plasma (Ottaviani et al. 2016; Gomez-Juaristi et al. 2019). It is possible that the displacement of the equilibrium in favor of the 5-(phenyl)-γ-valerolactones at plasma pH (assumed pH 7.4) might be a consequence of the more hydrophobic lactone associating with plasma proteins, and displacement in urine would depend on the pH value, this typically being lower with vegetarian diets, although microbial production of ammonia during storage could raise it. The analysis of rat urine has given less consistent results, varying markedly with test substance fed, as summarized in Table 3.

In the absence of a free carboxyl group, 5-(phenyl)-γ-valerolactones could not be substrates for β-oxidation. When 5-(3′,4′-dihydroxyphenyl)-γ-valerolactone 64 (Figure 16) was incubated with human erythrocytes, 4-hydroxy-5-(3′,4′-dihydroxyphenyl)pentanoic acid 62 was a product (Mulek et al. 2015), and when rats were dosed with [U-¹⁴C]-5-(3′-hydroxyphenyl)-γ-valerolactone 57 (Figure 10) they excreted labeled 3-(3′-hydroxyphenyl)propanoic acid 14 which in turn yielded 3′-hydroxyhippuric acid 55 (Das and Griffiths 1969), indicating that lactone hydrolysis and sidechain dehydroxylation occurred in vivo. It is unclear whether the 4-hydroxy-5-(phenyl)pentanoic acids can enter β-oxidation, either directly or after further metabolism. In theory, β-oxidation would be possible if any of the 4-hydroxy-5-(phenyl)pentanoic acids were dehydrated to the cis-geometric isomer of 5-(phenyl)pent-3-enoic acid as outlined in Figure 23, but it has not been possible to locate any reference to a suitable mammalian dehydratase/hydro-lyase enzyme that could facilitate this. Side chain dehydroxylation producing 5-(3′,4′-dihydroxyphenyl)pentanoic acid 84 and 5-(3′-hydroxyphenyl)pentanoic acid 85 (Figure 16) has been observed in in vitro human gut microbiota incubations of (–)-epicatechin (Stoupi et al. 2010; Le Bourvellec et al. 2019), and it is clear from other studies that both are potential sources of 3-hydroxybenzoic acid 46 (Figure 16).

Theoretically, 2-hydroxy-3-(phenyl)propanoic acids (phenyl-lactic acids) would be expected as the product of the first β-oxidation cycle of 4-hydroxy-5-(phenyl)pentanoic acids, as illustrated in Figure 24. They have been associated only rarely with flavanols and proanthocyanidins catabolism (3.4.8 and 3.4.16) but the limited supply of the relevant C₆–C₅ precursor(s) could prevent their accumulation especially if oxidation to 2-keto-3-(phenyl)propanoic acids and α-oxidation to 2-(phenyl)acetic acids were rapid (Figures 3 and 5). Such a pathway could explain why only trace amounts of 5-(phenyl)pentanoic acids are detected in urine, and the excretion by volunteers of the sulfate conjugate of an incompletely characterized 2-(hydroxyphenyl)acetic acid (ca 1.4% of dose) after consumption of [2-¹⁴C]-(–)-epicatechin (Ottaviani et al. 2016), and the excretion of traces of labeled 2-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 66 and 4′-hydroxyphenylacetic acid 70 in the urine of rats dosed intra-peritoneally with [U-¹⁴C]-(+)-catechin (Das and Sothy 1971).

Collectively, these constraints on the entry of the flavanol-derived C₆–C₅ metabolites to β-oxidation (Figure 24) may explain why substantial amounts of the 5-(phenyl)-γ-valerolactones and 4-hydroxy-5-(phenyl)pentanoic acids are excreted as phase-2 conjugates, for example 42 ± 5% of the dose in the human study using [2-¹⁴C]-(–)-epicatechin (Ottaviani et al. 2016), and why little or no C₆–C₃ and C₆–C₁ metabolites are produced in vivo from such precursors, possibly these being derived from only that fraction of the C₆–C₅ metabolites which have undergone microbiota-mediated sidechain dehydroxylation.

5.2.3. 3-Hydroxy-5-(phenyl)pentanoic acids

The only reports of a β-oxidation intermediate associated with the first cycle of β-oxidation of a C₆–C₅ metabolite are the excretion of 3-hydroxy-5-(phenyl)pentanoic acid 162 in the urine of cats and dogs after subcutaneous injection of 5-(phenyl)pentanoic acid 141 (Dakin 1908a) and in the urine of non-human primates in a study investigating the effects of their exposure to γ-irradiation (Cheema et al. 2019) (Figure 3). The non-human primates were fed a commercial cereal-based diet, but its precise composition was not recorded, and the origin of this metabolite is uncertain. Its identification should be treated as tentative because the reported fragmentation is difficult to reconcile with the suggested structure.

Such intermediates have never been reported in volunteer studies even for those metabolites with a 3′-hydroxy substituent on the phenyl ring where a phenyl-hydracrylic acid 3 sometimes accumulates in the second cycle. The possibility that, for example, 4-hydroxy-5-(3′-hydroxyphenyl)pentanoic acid 63 (Figure 10) (a well-known (–)-epicatechin gut microbiota metabolite) has been confused with 3-hydroxy-5-(3′-hydroxyphenyl)propanoic acid 163 (Figure 10) (the β-oxidation intermediate) can be excluded because in recent studies authentic standards of the C-4 hydroxylated isomers have been available. Accordingly, one might conclude that this C₆–C₅ β-oxidation intermediate does not accumulate even though the analogous C₆–C₃ β-oxidation intermediate may, and this suggests that in the longer side chain metabolite any adverse steric or electronic effects are too far removed from the enzyme’s active site to cause any impediment. Note, however, that the flux of 3′-hydroxyphenyl-substituted C₆–C₅ metabolites from a comparatively restricted number of dietary sources (flavanols, proanthocyanidins) is less than the flux of 3′-hydroxyphenyl-substituted C₆–C₃ metabolites (to which some other flavonoids, acyl-quinic acids, phenylalanine 112, tyrosine 91, etc. can also contribute) (Figure 10), and it is possible that under normal dietary conditions the flux from the C₆–C₅ precursor alone is insufficient to overload the first cycle.

5.2.4. 3-Hydroxy-3-(phenyl)propanoic acids

3-Hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89 (Figure 18) is the most frequently reported phenyl-hydracrylic acid of purely phytochemical origin. Oranges and related products are the dominant dietary sources of 3′-hydroxy-4′-methoxyphenyl-substituted substrates hesperetin and diosmetin glycosides, and this phenyl-hydracrylic acid is considered a marker for their consumption. However, as discussed in 3.4.11 some volunteers do not excrete a detectable amount and this person-to-person variation will be explored more fully in 5.3.

The only other significant sources of 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44 and/or 3′-hydroxy-4′-methoxycinnamic acid 28 (isoferulic acid) are indirect, that is, those sources rich in the COMT-substrate 3′,4′-dihydroxycinnamic acid 27, predominantly coffee, maté, artichoke, apples, blueberries and cranberries. However, there have been no reports of volunteers who have consumed substantial amounts of these commodities excreting 3-hydroxy-3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 89, even though early excretion of phase-2 conjugated 3-hydroxy-4-methoxybenzoic acid 47 (see 3.4.11) is suggestive of limited β-oxidation. Note, however, the point at which 4-methylation occurred is not certainly known and 3,4-dihydroxybenzoic acid 65 (Figure 16) might be the immediate precursor.

A possible explanation could be that the flux of 3′-hydroxy-4′-methoxyphenyl-substituted metabolites entering β-oxidation sometimes does not exceed the capacity of the enzymes involved and the phenyl-hydracrylic 89 intermediate does not accumulate. It is not possible to make precise comparisons of flux across studies because the studies differ regarding the methods of analysis and which metabolites were reported, but such data as are available are presented in Table 4.

Table 4. 24-Hour excretion of 3′-hydroxy-4′-methoxyphenyl-substituted C₆–C₃ metabolites by volunteers.

Commodity	Intake	Excretion of 3′-hydroxy-4′methoxyphenyl-substituted C6–C3 metabolites			Reference
		Number of metabolites quantified	Mean Flux (μmol/24 hours)	Maximum flux (μmol/24 hours)^b
Artichoke	8.62 mmol 3′,4′-dihydroxycinnamic acid equivalents	4 phase-2 conjugates 82% + the cinnamic acid 18%	54.8		Dominguez-Fernandez et al. 2022
Coffee	4.4 mmol 3′,4′-dihydroxycinnamic acid equivalents	4 phase-2 conjugates	49.5		Erk et al. 2012
Coffee	2.15 mmol 3′,4′-dihydroxycinnamic acid equivalents	4 phase-2 conjugates	32.7		Erk et al. 2012
Coffee	1.01 mmol 3′,4′-dihydroxycinnamic acid equivalents	4 phase-2 conjugates	19.9		Erk et al. 2012
Coffee	0.75 mmol 3′,4′-dihydroxycinnamic acid equivalents	2 phase-2 conjugates	8.1		Stalmach, Williamson, and Crozier 2014
Coffee	0.54 mmol 3′,4′-dihydroxycinnamic acid equivalents	2 phase-2 conjugates	7.5		Stalmach, Williamson, and Crozier 2014
Coffee	0.35 mmol 3′,4′-dihydroxycinnamic acid equivalents	2 phase-2 conjugates	5.3		Stalmach, Williamson, and Crozier 2014
Coffee	0.19 mmol 3′,4′-dihydroxycinnamic acid equivalents	5 phase-2 conjugates	15.25		Gomez-Juaristi et al. 2018b
Maté	0.23 mmol 3′,4′-dihydroxycinnamic acid equivalents	4 phase-2 conjugates 88% + the cinnamic acid 28 12%	10.58		Gomez-Juaristi et al. 2018a
Apple (white fleshed)	0.076 mmol 3′,4′-dihydroxycinnamic acid equivalents	5 phase-2 conjugates 75.5% + the phenylpropanoic 44 1.3% and cinnamic acids 28 23.2%	6.30	18.8	Rubió et al. 2021
Apple (red fleshed)	0.195 mmol 3′,4′-dihydroxycinnamic acid equivalents	5 phase-2 conjugates 88% + the phenylpropanoic 44 6.3% and cinnamic acids 28 5.7%	9.25	25.4	Rubió et al. 2021
Orange juice	0.287 mmol hesperetin (nett)^a	Phenyl-hydracrylic acid 89 (94%) and phenyl-propanoic acid 44 (6%)^c	76.8 (72 phenyl-hydracrylic acid 89)^c	186 (186 phenyl-hydracrylic acid 89)	Pereira-Caro et al. 2014
Orange juice	0.079 mmol hesperetin (nett)^a	Phenyl-hydracrylic acid 89 (57%), phenyl-propanoic acid 44 (21%) and cinnamic acid 28 (21%)^c	19.2 (11.0 phenyl-hydracrylic acid 89)^c	92.2 (92.1 phenyl-hydracrylic acid 89)	Pereira-Caro et al. 2015
Orange juice	0.079 mmol hesperetin (nett)^a	Phenyl-hydracrylic acid 89 (65%), phenyl-propanoic acid 44 (18%) and cinnamic acid 28 (17%)^c	26.0 (17.0 phenyl-hydracrylic acid 89)^c	–	Pereira-Caro et al. 2015
Orange juice	0.24 mmol hesperetin (nett)^a	3 phase-2 conjugates 37.5% + the phenyl-hydracrylic acid 89 62.5%	26.0 (17.0 phenyl-hydracrylic acid 89)		Pereira-Caro et al. 2017
Orange juice	0.24 mmol hesperetin (nett)^a	3 phase-2 conjugates 32.9% + the phenyl-hydracrylic acid 89 67.1%	29.9 (18.5 phenyl-hydracrylic acid 89)		Pereira-Caro et al. 2017

a The intake of hesperetin glycosides has been corrected for mean excretion of hesperetin phase-2 conjugates.

b Value for the volunteer with the greatest excretion.

c Quantified by GC–MS after deconjugation with β-glucuronidase and sulfatase.

Volunteer studies in which caffeoylquinic acids were consumed indicate that the highest mean fluxes for total 3′-hydroxy-4′-methoxyphenyl-substituted metabolites (46–55 μmol/24 hours) have been recorded without detection of the phenyl-hydracrylic acid 89 and that conjugated phase-2 metabolites always dominated (>82%) (Table 4). In contrast, after consumption of hesperetin, excretion of the phenyl-hydracrylic acid has been recorded for a mean flux as low as 20 μmol/24 hours. Unfortunately, most studies on orange juice used enzymic deconjugation of biological fluids, but for the two studies where LC–MS analysis was used without prior hydrolysis, the phase-2 conjugates are a relatively small proportion (<40%) with the phenyl-hydracrylic acid clearly dominating, although with considerable person-to-person variation as discussed more fully in 5.3.

Fortuitously, some studies have recorded the excretion data for a series of shorter intervals. For example, Stalmach et al. reported that the ileostomists who consumed 345 μmol 3′,4′-dihydroxycinnamic acid equivalents excreted ca 60% as two 3′-hydroxy-4′-methoxyphenyl-substituted metabolites in the first five hours at an average flux of ca 0.5 μmol/hour (Stalmach et al. 2010). Assuming that this applies to the study by Erk et al. with ileostomists, a flux averaging ca 2.9 μmol/hour (two metabolites) for the first five hours can be expected at the highest dose of 4.4 mmol 3′,4′-dihydroxycinnamic acid equivalents, and a flux (two metabolites) of ca 1 μmol/hour for volunteers with an intact colon after 745 μmol 3′,4′-dihydroxycinnamic acid equivalents (Stalmach, Williamson, and Crozier 2014). Despite a larger dose (8.62 mmol 3′,4′-dihydroxycinnamic acid equivalents), volunteers who consumed artichoke recorded a flux of five metabolites averaging only 1.9 μmol/hour over the first eight hours and only 1.47 μmol/hour over the first 4 hours (Dominguez-Fernandez et al. 2022).

In contrast, Pereira-Caro et al. recorded that after volunteers consumed orange juice ca 60% of the 3′-hydroxy-4′-methoxyphenyl-substituted metabolites were excreted in the five to ten hour-period (Pereira-Caro et al. 2014), with a flux of ca 9 μmol/hour (two metabolites quantified after enzymic deconjugation) following a nett consumption of 0.29 mmol hesperetin glycosides. The rate for the phenyl-hydracrylic acid 89 alone was 8.6 μmol/hour, and 0.4 μmol/hour for the 3-(phenyl)propanoic acid 44. The equivalent data following consumption of a 3′,4′-dihydroxycinnamic acid-rich source are not available for volunteers with an intact colon. There are some less precise data for coffee and maté consumption where four and seven metabolites were quantified, but without the use of authentic standards, which recorded average fluxes of 0.8 μmol/hour in the five to 12 hour-period and 0.9 μmol/hour in the four to 12 hour-period, respectively (Gomez-Juaristi et al. 2018a, 2018b). As stated above, Stalmach et al., reported 8.1 μmol excreted in 24 hours after 745 μmol 3′,4′-dihydroxycinnamic acid equivalents intake (Stalmach, Williamson, and Crozier 2014), and even if this had all been excreted in the five to 10 hour-period it would only correspond to 1.6 μmol/hour, and it is clear that in this time period, coffee consumption is unlikely ever to generate fluxes of 3′-hydroxy-4′-methoxyphenyl-substituted metabolites as high as orange juice—indeed they may well be an order of magnitude lower. Even artichoke providing 8.62 mmol 3′,4′-dihydroxycinnamic acid equivalents only produced a 2.46 μmol/hour flux of five metabolites in the four to eight hour-period. This seems surprisingly low and must in part reflect the influence of the artichoke matrix (Dominguez-Fernandez et al. 2022), largely absent in coffee, maté and orange juice, but almost certainly other factors are operating. Nevertheless, the different handling of orange juice is striking, and we offer the following explanation based on in vitro studies.

With 3′,4′-dihydroxycinnamic acid as the substrate, 4′-methylation exceeded 3′-methylation several-fold in the gastric epithelium in vitro, and there was no further phase-2 conjugation (Farrell, Dew, et al. 2011). According to Kern et al. (Kern et al. 2003), differentiated Caco-2 cells slowly transform 3′,4′-dihydroxycinnamic acid to approximately equal amounts of its 3′- and 4′-methylated conjugates, but other investigators have reported only 3′-methylation (Farrell, Poquet, et al. 2011; Martini, Conte, and Tagliazucchi 2019). Similarly, 3-(3′,4′-dihydroxyphenyl)propanoic acid arising from gut microbiota hydrogenation of the cinnamic acid, is only 3′-methylated by Caco2 cells (Poquet, Clifford, and Williamson 2008a), and hence the bulk of the 4′-methylated metabolites produced from 3′,4′-dihydroxycinnamic acid 27 are associated with upper GIT metabolism. Accordingly, colon metabolism is associated primarily with 3′-methylation, and because comparatively little dietary caffeoylquinic acids/3′,4′-dihydroxycinnamic acid 27 are absorbed in the upper GIT, gut microbiota metabolism will dominate for these substrates. With hesperetin as substrate, there is no requirement for methylation, and thus it appears that the relatively high fluxes of 3′-hydroxy-4′-methoxyphenyl-substituted metabolites overload the pathways for disposing of the propanoic and cinnamic acid (sulfation, glucuronidation and glycine conjugation, plus “reverse” hydrogenation for the cinnamic acid 28) leading to an accumulation of the phenyl-hydracrylic acid 89 intermediate which cannot be cleared by the mitochondrial 3-hydroxy-acyl-CoA-dehydrogenase, essentially stalling β-oxidation of 3′-hydroxy-4′-methoxyphenyl-substituted substrates, for some but not all volunteers.

5.2.5. Hepatic “reverse” hydrogenation

There is evidence of human hepatic “reverse” hydrogenation of 3′,4′-dihydroxycinnamic acid 27, 3′-methoxy-4′-hydroxycinnamic acid 79 and 3′,4′-dimethoxycinnamic acid 30 in vivo and 4′-hydroxycinnamic acid 29 in vitro, but the specific enzyme(s) responsible are not known. There are cytosolic and mitochondrial 2-enoyl-CoA reductases, both of which accept cinnamoyl-CoA 9 (Kim et al. 2014; Cvetanović et al. 1985; Zhao et al. 2019) and which might accept other ω-phenyl-alkenoic acids 2. Only the cytosolic enzyme could hydrogenate a cinnamic acid unable to enter the mitochondrion.

Phenylacetyl-CoA 104 is a substrate for the cytosolic fatty acid synthase (FAS) yielding a series of even-numbered ω-phenyl-fatty acids up to C₆–C₁₆, predominantly C₆–C₁₂, but chain elongation proceeded more slowly than with acetyl-CoA. Although it has been reported that odd-numbered ω-phenyl fatty acids are not produced because chain elongation does not occur with benzoyl-CoA (Smith and Stern 1983), the excretion by horses of 3-hydroxy-3-(phenyl)propanoic acid (3-(phenyl)hydracrylic acid) 59, and 3-keto-3-(phenyl)propanoic acid 164 after feeding [²H₅]-benzoic acid 11 (see Figure 3) were thought to be due to FAS (Marsh et al. 1982). Although there are no data for ω-phenyl-alkenoic acids 2 with phenyl-ring substituents possibly the cytosolic FAS is responsible for cinnamic acid hydrogenation.

The factors determining the nett direction of this reaction in mitochondria seem not to have been investigated.

5.2.6. Amino acid conjugation

As discussed in part 4 the amino acid conjugations are not well characterized. For conjugation to occur the ligand must either enter the mitochondrion or be produced therein. Glycine conjugates of ω-phenyl-alkanoic acids 1 and ω-phenyl-alkenoic acids 2 predominate in human urine and have been observed for ω-phenyl-alkanoic acids 1 and ω-phenyl-alkenoic acids 2 acids with side chains of 1 to 3 carbons, whereas glutamine and taurine conjugation has been observed only for metabolites with 2 and 4-carbon sidechains (Figures 3 and 5). Phenylacetic acid 12 is conspicuously promiscuous with reference to the amino acid utilized (Table 1). The precise factor(s) determining which route(s) are utilized are unclear, but inherited metabolic diseases, such as MCAD deficiency, phenylketonuria, and non-ketotic hyperglycinaemia or hyperammonaemia, are certainly one factor (James and Smith 1973; Moore et al. 1990; Fischer et al. 2000; Sakuma, Sugiyama, and Wada 1992; Van Hove et al. 1995), and other genetic variations and substrate loading will also contribute, as will the relative rates of the competing routes of metabolism (see Figure 25). The lack of experimental data for mitochondrial uptake is a serious constraint to unraveling the metabolism of these metabolites of phytochemical origin which overlap with endogenous metabolism.

5.3. Person-to-person variation

Individual variation in phytochemical metabolism is well illustrated by data from plasma C_max and 24-hour urinary excretion with standard deviations or standard errors near to or exceeding the mean value, and quotients for the (largest value/smallest value) in a single data set occasionally exceeding 200 (Dominguez-Fernandez et al. 2022; Rubió et al. 2021). Such variation is rarely discussed, and without access to the raw data post-hoc, further investigation is not possible. Our examination of unpublished data for individual’s excretion of 3′-hydroxy-4′-methoxyphenyl-substituted metabolites after orange juice consumption follows.

5.3.1. Person-to-person variation in the excretion 3’-hydroxy-4’-methoxyphenyl-substituted metabolites after consumption of orange juice

The metabolic pathways are presented in Figure 18. Inspection of raw data (rather than published mean ± s.d.) for samples hydrolyzed and analyzed by GC–MS, identifies four volunteers out of 27 who consumed orange juice but did not excrete detectable 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44 (Table 2).

One of the four (22) excreted the cinnamic acid 28 (0.04 μmol/24 hours) and the phenyl-hydracrylic acid 89 (92.18 μmol/24 hours), one (4) did not excrete either, but unfortunately the equivalent data for the cinnamic acid are not available for the other two participants (1, 27). However, 27 excreted the phenyl-hydracrylic acid 89 (186 μmol/24 hours) which presumably was produced from the phenyl-propanoic acid 44 via the cinnamic acid 28 following delivery to the colon of a nett intake of 265 μmol hesperetin glycosides, i.e., the gross intake of 348 μmol corrected for excretion of hesperetin phase-2 conjugates.

In contrast volunteers 5 and 8 both excreted the phenyl-propanoic acid 44 (6.51 and 2.77 μmol/24 hours, respectively for nett intakes of 91 and 59 μmol, respectively), but did not excrete the cinnamic acid 28 or the phenyl-hydracrylic acid 89, suggesting that either β-oxidation did not occur at all, or may have proceeded to completion without accumulation of C₆–C₃ intermediates.

Volunteers 9–11 with nett hesperetin intakes of 70, 42 and 90 μmol, respectively, excreted the phenyl-propanoic acid 44 (1.54, 0.72 and 2.51 μmol/24 hours, respectively) and the phenyl-hydracrylic acid 89 (0.85, 3.79 and 7.61 μmol/24 hours, respectively) but the cinnamic acid 28 was not detectable in urine.

These data clearly demonstrate the substantial person-to-person variation largely masked when mean values are used. Some of the variation will almost certainly reside with the gut microbiota production of 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44, but there is also considerable variation in endogenous handling of this substrate. Excretion of substantial amounts of the phenyl-hydracrylic acid 89 by some individuals suggests that the hydroxy-acyl-CoA dehydrogenase might be a limiting factor.

However, eight volunteers (1–8) did not excrete the phenyl-hydracrylic acid 89 suggesting that for these volunteers either (a) the hydroxy-acyl-CoA dehydrogenase was not a limiting factor, or (b) the phenyl-hydracrylic acid 89 was not produced, possibly because the phenyl-propanoic acid 44 was not produced by the gut microbiota. However, of those eight, six did excrete the 3-(phenyl)propanoic acid 44 (2.57–6.51 μmol/24 hours), and two (6, 7) of the six did excrete the cinnamic acid 28 (0.31 and 0.06 μmol/24 hours). Possibly for these eight volunteers the 3-(phenyl)propanoic acid 44 and the cinnamic acid 28 were diverted from β-oxidation to Phase-2 conjugation pathways but because the urines had been enzyme treated prior to derivatisation and GC–MS, this hypothesis cannot be tested with these data. Clearly the situation is complex and carefully targeted experiments will be required to obtain clarification.

5.4. Studies with substrates for which there are no human data

Data for metabolism of ω-phenyl-alkanoic acids 1 having other patterns of phenyl substitution are either not available, or available only from animal studies, predominantly with rats. There are significant inter-species differences in metabolism (Scheline 1978), and with many substrates the data obtained from such studies are not good models for human metabolism. For example, rats and humans differ significantly in their metabolism of coumarin 139 (see 3.4.1 and Figure 9), 3-(4′-hydroxyphenyl)propanoic acid 20 (see 3.4.3 and Figure 11) and hesperetin (see 3.4.11 and Figure 18).

Studies with rats given very large doses (100 mg/kg and 200 mg or 400 mg per rat = ca 6 to 100 g per adult human) of pure test substances, with and without antibiotic to suppress the gut microbiota, suggest that C₆–C₃ substrates with 2′-hydroxy-, 2′-methoxy-, 2′,4′,5′-trihydroxy-, 3′,4′-dimethoxy- and 3′,4′,5′-trimethoxyphenyl-substitution enter the mitochondrion but do not undergo complete β-oxidation as judged by the excretion of C₆–C₃ glycine conjugates (Table 1) and the failure to excrete the corresponding benzoic or hippuric acid. However, it is not clear whether the cinnamic acids can access the mitochondrion as such, or only after cytosolic hydrogenation. Those substrates with 2′,4′-dihydroxy-, 2′-hydroxy-4′-methoxy-, 3′,4′-methylenedioxy-, 3′,4′,5′-trihydroxy- and 3′,5′-dimethoxy-4′-hydroxyphenyl-substitution are also excreted as C₆–C₃ metabolites but without glycine conjugation (Booth et al. 1959; Samuelsen et al. 1986; Meyer and Scheline 1972b, Griffiths 1969; Diao et al. 2018).

The metabolism of the 2′-hydroxy-, 2′-methoxy-, and 3′,4′-dimethoxy-substituted test substances all stalled at the phenyl-hydracrylic acid 3 intermediate and when 3-(3′,4′-methylendioxyphenyl)propanoic 181 and 3′,4′-methylendioxycinnamic acid 182 were dosed to rats at 1 mmol/kg they showed unique behavior with significant accumulation not only of the corresponding stage 2 intermediate, 3-hydroxy-3-(3′,4′-methylenedioxyphenyl)propanoic acid 183, but also the stage 3 intermediate, 3-keto-3-(3′,4′-methylenedioxyphenyl)propanoic acid 184, despite ultimately yielding the associated hippuric acid (Figure 21) (Klungsoyr and Scheline 1981).

Rats transformed all tested 3′,4′,5′-tri-substituted substrates to 3-(3′,5′-dihydroxyphenyl)propanoic acid 42, a human β-oxidation substrate, suggesting that possibly the original 3′,4′,5′-tri-substituted substrates could be handled by humans without accumulation of the intermediates at the low doses that might occur in their diet.

5.5. Relative contributions of the gut microbiota and hepatic β-oxidation to phenolic catabolism

The human gut microbiota is very diverse, typically containing well over 1,000 species (Bäckhed et al. 2005), collectively capable of numerous transformations of a vast range of substrates, and even for a single substrate there is likely to be more than one route to any particular product (Williamson and Clifford 2017, 2010). It is well-established that many anaerobes, including some members of the gut microbiota such as some strains of Escherichia coli, can rupture the phenyl ring and degrade the resultant aliphatic metabolites by β-oxidation (Dı́az et al. 2001) but in this review the question posed is can the gut microbiota shorten the side chain of ω-phenyl-alkanoic acids 1 by β-oxidation while keeping the phenyl ring intact? Eubacterium ramulus, Clostridium butyricum and Flavonifractor plautii (Clostridium orbiscendens) and some lactic acid bacteria are able to produce C₆–C₃ metabolites directly from susceptible flavonoids or indirectly from an intermediate flavonoid degradation product (Braune and Blaut 2016; Schoefer et al. 2003; Pereira-Caro et al. 2018; Guo, Guo, and Li 2021; Braune, Engst, and Blaut 2005; Honohan et al. 1976), but C₆–C₁ metabolites were not reported in these pure culture incubations.

In some studies, for example incubation of (–)-epicatechin with human fecal flora, the C₆–C₃ metabolites did not appear until the C₆–C₅ metabolites were declining (Stoupi et al. 2010), suggesting a sequential production consistent with β-oxidation. Flavonifractor plautii can also produce the C₆–C₅ metabolites from flavonoids and might therefore be a candidate for such β-oxidation. In contrast, as discussed in 3.4.11, incubation of [3-¹⁴C]-hesperetin with rat cecal microbiota produced labeled 3-(3′-hydroxy-4′-methoxyphenyl)propanoic acid 44, 3-(3′,4′-dihydroxyphenyl)propanoic acid 60 and 3-(3′-hydroxyphenyl)propanoic acid 14 but the associated benzoic acids were not detected until the labeled hesperetin was fed to rats (Honohan et al. 1976), clearly indicating that mammalian metabolism was essential for the β-oxidation. Similar results were observed with several other flavonoids and 4′-hydroxycinnamic acid 29 was hydrogenated (Griffiths and Smith 1972). Similarly, as discussed in 3.4.3 when [²H-2′,3′,5′,6′]-naringin was incubated with human gut microbiota in vitro the production of 3-(4′-hydroxyphenyl)propanoic acid 20 and 3-(phenyl)propanoic acid 13 was confirmed, but the benzoic acids were not reported (Chen et al. 2018; Chen et al. 2019). In contrast, loss of unconjugated 4-hydroxybenzoic acid 45 and 3-(4′-hydroxyphenyl)propanoic acid 20 in feces by a patient suffering from cystic fibrosis has been attributed to gut microbiota β-oxidation of 3-(4′-hydroxyphenyl)propanoic acid 20 formed from unabsorbed tyrosine 91 (Vanderhe.C et al. 1971). Accordingly, there seems little doubt that some gut microbiota can subject at least some ω-phenyl-alkanoic acids 1 to β-oxidation, but it is equally clear that in some situations C₆–C₅, C₆–C₃ and C₆–C₁ metabolites produced in the colon may arise independently.

5.6. Possible interactions between ω-phenyl-alkanoic acids 1 and lipid and carbohydrate metabolism

5.6.1. Carbohydrate metabolism

Studies with isolated perfused rat livers have demonstrated that ω-phenyl-alkanoic acids 12, 13, 15, 141 (4 mm) inhibit gluconeogenesis from l-lactate, pyruvate, and l-alanine by 25 to 35% with inhibition increasing with the side-chain length in the range C₆–C₂ to C₆–C₆, but there was no chain length effect with d-fructose (10% inhibition) or glycerol (20% inhibition). During the perfusion the ω-phenyl-alkanoic acids 12, 13, 15, 141 were metabolized to benzoic 11 and hippuric acid 18 or, phenylacetic acid 12 and phenylacetyl-glycine 95 (phenylaceturic acid) as appropriate (Gonzalez, Bressler, and Brendel 1973). Phenylacetyl-CoA 104, but not phenylacetic acid 12, inhibits pyruvate carboxylase and decreased pyruvate carboxylation in a concentration dependent manner. Infusion of 3-(phenyl)propanoic acid 13 (7 µmol/min) decreased plasma glucose in normal rats from 110 ± 12 to 66 ± 11 mg/dL and from 295 ± 14 to 225 ± 12 mg/dL in streptozotocin diabetic rats (Bahl et al. 1997), but there was no effect in diabetics after a 30.4 mmol bolus dose of phenylacetic acid 12 (Wajngot et al. 2000). Intravenous administration of 3′-hydroxy-4′-methoxycinnamic acid 28 (26 µmol/kg) to streptozotocin diabetic rats for one day lowered plasma glucose and increased GLUT4 mRNA and muscle glycogen synthesis, suggesting inhibition of liver gluconeogenesis (Liu et al. 2000). It has also been reported that 3′-hydroxy-4′-methoxycinnamic acid 28 stimulates the α_1a-adrenoceptor of murine myoblast C₂C₁₂ cells increasing glucose uptake even at sub-nanomolar concentrations (Liu et al. 2001). 3′-Hydroxy-4′-methoxycinnamic acid 28 occurs in plasma primarily as phase-2 conjugates (Clifford, Kerimi, and Williamson 2020; Pereira-Caro, Clifford, et al. 2020), but maximal concentrations of the free acid at ca 20 nmol/L have been reported (Mills et al. 2017; Scherbl et al. 2017).

A study in which volunteers consumed 300 ml orange juice per day for 60 days delivering 120 µmol hesperidin and 24.5 µmol naringin produced a reduction in insulin resistance (–44%), insulin (–33%) and plasma glucose (–6.5%) which returned to their initial values after washout. There were no changes in body weight or fat during the volunteer study (Fidélix et al. 2020). The gut microbiota metabolites were not investigated but the studies of Pereira-Caro et al. (see 3.4.11, Figure 18 and Table 2) indicates that significant amounts of C₆–C₃ metabolites would enter the plasma.

5.6.2. Lipid metabolism

Although possible interactions between the β-oxidation of ω-phenyl-alkanoic 1 or ω-phenyl-alkenoic acids 2 and fatty acid metabolism have rarely been the explicit objective of mechanistic studies, there are published data that suggest that such interactions may occur. As presented in 3.4.3, the clearest evidence for such interaction is provided by a study where a patient on a high protein diet and receiving medium chain triglyceride administration while suffering from cystic fibrosis (causing severely impaired intestinal amino acid resorption) excreted 4′-hydroxycinnamic acid 29 (1.59 µmol/mg creatinine),¹² 3-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 69 (0.92 µmol/mg creatinine), and 4-hydroxybenzoic acid 45 (18.3 µmol/mg creatinine), attributed to an abnormal hepatic β-oxidation load (Wadman et al. 1973) arising from the fatty acids and C₆–C₃ metabolites of unabsorbed tyrosine 91 (Figure 11). This is the only report of urinary 3-hydroxy-3-(4′-hydroxyphenyl)propanoic acid 69—it has never been reported in volunteer studies where ca 200 μmol 4′-hydroxyphenyl-substituted substrates have been consumed. In contrast to the cinnamic 29 and benzoic acids 45 excreted by this patient, the phenyl-hydracrylic acid 69 was not detected in feces, strongly suggesting that it was of endogenous rather than bacterial origin. Unfortunately, the authors did not report medium chain fatty acyl-carnitines or 3-hydroxy-dicarboxylic acids and it is uncertain whether fatty acid β-oxidation was also impeded, but some support is provided by the observation that perfusing a rat heart with either 3,4-dihydroxybenzoic acid 65 or 3-methoxy-4-hydroxybenzoic acid 10 (40 μM) suppressed β-oxidation of perfused palmitic acid and significantly reduced the acyl-CoA: free CoA ratio. The use of the COMT inhibitor tolcapone demonstrated that 3-methoxy-4-hydroxybenzoic acid 10 was responsible for the suppression (Cao, Zhang, et al. 2009). As discussed above (3.4.8 and 3.4.9) these acids (10, 65) are at most only minor products of the β-oxidation of C₆–C₃ ω-phenyl-alkanoic acids 1. Free 3,4-dihydroxybenzoic acid 65 has been observed in plasma with a C_max values of 3.3 ± 0.7 µmol/litre at 1 hour after consuming 150 g of chicory (Zheng et al. 2019), but the free 3-methoxy-4-hydroxybenzoic acid 10 concentration is much lower (Koli et al. 2010), and inhibition in vivo will be limited.

Smith and Stern reported that benzoyl-CoA inhibits FAS¹³ with a K_i value of ca 40 μM by competing with acetyl-CoA and malonyl-CoA for substrate-binding sites. The K_m for malonyl-CoA is 0.2 μM and for acetyl-CoA even lower. Clearly benzoyl-CoA binds much more weakly than the endogenous substrates. but its local intracellular concentration could be very high because (poly)phenol-rich diets generate a significant hepatic flux. Volunteers consuming black tea have recorded hippuric acid urinary fluxes of 82–186 μmol/hour (Clifford et al. 2000) and other studies have produced similar values (Mulder, Rietveld, and van Amelsvoort 2005; Krupp et al. 2012), but whether the local intracellular concentration of benzoyl-CoA could be sufficient to inhibit FAS II in vivo is uncertain.

Phenylacetyl-CoA 104, but not benzoyl-CoA, is a substrate for FAS type II yielding in vitro a series of even-numbered ω-phenyl-fatty acids up to C₆–C₁₆, predominantly C₆–C₁₂, but chain elongation proceeded more slowly than with acetyl-CoA (Smith and Stern 1983). However, the excretion by horses of 3-hydroxy-3-(phenyl)propanoic acid (3-(phenyl)hydracrylic acid) 59, and 3-keto-3-(phenyl)propanoic acid 164 after feeding [²H₅]-benzoic acid 11 (see Figure 3) were thought to arise through the chain elongating action of FAS (Marsh et al. 1982).

It has not been possible to locate any data relating to chain elongation of ω-phenyl-alkanoic acids 1 or ω-phenyl-alkenoic acids 2 having phenyl-ring substituents. However, the lack of references to longer chain even-numbered ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic acids 2 having phenyl-ring substituents, despite the regular occurrence of the C₆–C₂ analogues, suggests that there is little or no chain elongation with these potential substrates.

There have been many animal and in vitro studies utilizing phenyl-propanoic and cinnamic acids, predominantly compounds not found in the diet, but some of dietary origin at non-dietary doses. 2-Hydroxy-3-(3′,4′,5′-trihydroxyphenyl)prop-2-enoic acid (2,3′,4′,5′-tetrahydroxycinnamic acid) 199 inhibited FAS in vitro (Zhang et al. 2009), but the IC₅₀ was not reported and this is not a known dietary metabolite. It is the enol tautomer of 2-keto-3-(3′,4′,5′-trihydroxyphenyl)propanoic acid 200 (Figure 14). 4′-Hydroxycinnamic acid 29 in vitro promoted fatty acid β-oxidation (Yoon et al. 2013), but only at µm concentrations unlikely to be seen in vivo.

3-(3′,4′-Dihydroxyphenyl)propanoic acid 60, 3′-hydroxycinnamic acid 19 and 3′-methoxy-4′-hydroxycinnamic acid¹⁴ 79 and hesperetin (660 µmol/100 g diet)¹⁵ lowered rat plasma total cholesterol and triglyceride, apparently through inhibition of hepatic hydroxy-methyl-glutaroyl-CoA reductase (HMG-CoA-reductase) (ca 40–50%) and acyl-CoA: cholesterol acyl-transferase (ACAT) by ca 35% (Kim et al. 2003). A similar study in which rats were fed naringin, 3-(4′-hydroxyphenyl)propanoic acid 20 or 4-hydroxybenzoic acid 45 (34, 72 and 87 μmol/100 g of diet, respectively) was less effective, suppressing these enzymes ca 15–25% and ca 20–30%, respectively (Jeon et al. 2007).

Sub-cutaneous treatment of hypercholesterolemic rats with 2′,4′,5′-trimethoxycinnamic acid 165 (80 mg/kg; 336 µmol/kg) produced a ca 25% reduction in serum cholesterol (Antunez-Solis et al. 2009). Humans excrete 2′,4′,5′-trimethoxycinnamic acid 165 as the ester glucuronide (Nakazawa and Ohsawa 2000). 3′,4′-Dimethoxycinnamic acid 30 had similar potency but 2′,4′-dimethoxycinnamic 166 and 3′,5′-dimethoxycinnamic acid 167 were much less potent (Serna et al. 2015). There are no metabolic data for 2′,4′-dimethoxycinnamic acid 166 but in rats 3′,4′-dimethoxycinnamic acid 30 yields the phenyl-hydracrylic acid 98 and in humans is subject to “reverse” hydrogenation (see 3.4.12 and Figure 19). 3′,5′-Dimethoxycinnamic acid 167 in rats is hydrogenated (see 3.4.14 and Figures 14 and 26) but there are no human data.

In addition to reductions in insulin resistance and plasma glucose after volunteers consumed orange juice for 60 days—see 5.6.1—Fidelix et al. also reported lower LDL-cholesterol (–16%) and triglycerides (–30%) which returned to their initial values after washout (Fidélix et al. 2020). The gut microbiota metabolites were not investigated but the studies of Pereira-Caro et al. (see 3.4.11, Figure 18 and Table 2) indicates that significant amounts of C₆–C₃ metabolites would enter the plasma.

6. Concluding remarks

It has been known for over 100 years that ω-phenyl-alkanoic acids 1 lacking phenyl-ring substituents are β-oxidation substrates that are metabolized more slowly than fatty acids (Dakin 1908a, 1909) producing cinnamic, phenyl-hydracrylic and 3-keto-3-(phenyl)propanoic acid intermediates. Rat peroxisomes handle C₆–C₄ to C₆–C₁₂ substrates while rat mitochondria handle C₆–C₄ to C₆–C₈ substrates. Some details of their metabolism have not been investigated, but there are bigger gaps in our knowledge and understanding of the β-oxidation of ω-phenyl-alkanoic acids 1 with phenyl-ring substituents. The necessary experiments have simply not been performed. Table 5 and Figure 25 summarize the human metabolism of the dietary ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic acids 2. As discussed in 5.1, 5.2.5 and 5.2.6 the mechanisms of mitochondrial uptake, the factors determining whether a cinnamic acid is hydrogenated and/or passed directly to β-oxidation, and those determining the ligand(s) to which the C₆–C₁ metabolites are conjugated, remain obscure. Some ω-phenyl-alkanoic acids 1 are gut microbiota β-oxidation substrates (3.4.8 and Figure 16).

Table 5. A summary of the human metabolism of the major classes of ω-phenyl-alkanoic acids 1 and phenyl-alkenoic acids 2.

Origin of C6–C3 ω-phenyl-alkanoic 1 and / or ω-phenyl-alkenoic acids 2				Labeled study (volunteers or human gut microbiota)	Fate of the C6–C3 metabolite
Phenyl-ring substituents	Present preformed in the diet	Produced by gut microbiota metabolism	Produced by endogenous ω-oxidation	References	Phase-2 methylation (M), sulfate (S), or glucuronide (G) conjugation	C6–C3-glycine conjugate	Hepatic “reverse” hydrogenation	3-(Phenyl)-hydracrylic acid 3 excreted	Benzoic acid (B) and / or benzoyl-glycine (BG) excreted
None	No	Yes		Hoskins, Holliday, and Greenway 1984	Ester glucuronide	Yes	Yes ^c	Yes	Yes (B, BG)
2′-hydroxy	Yes, minor	No			No	Yes			Yes (B, BG)
3′-Hydroxy ^a	No	Yes		Borges et al. 2018; Baba et al. 1981; Hackett et al. 1983; Ottaviani et al. 2016.	Tentative report of S and G of the hydracrylic acid 58	No		Yes	Yes (B, BG)
4′-Hydroxy ^a	Yes	Yes	Yes, minor	Chen et al. 2018; Chen et al. 2019	Yes (S, G)	No	Yes ^d	Yes	Yes (B, BG)
4′-Methoxy	No	Yes	Yes, minor	Sangster et al. 1987	No	Yes			Minor (B, BG)
3′,4′-Dihydroxy ^a,b	Yes, major	Yes	Extremely limited	Hackett et al. 1983; Ottaviani et al. 2016	Major (M, S, G)	No	Yes ^e	No	Minor (B)
3′-Methoxy-4′-hydroxy	Yes	Yes	Minor		Major (S, G)	Yes	Yes ^e	Requires confirmation	Minor (B)
3′-Hydroxy-4′-methoxy	No	Yes			Major (S, G)	Yes		Yes, major	Minor (B)
3′,4′-Dimethoxy	Yes				No	Yes	Yes ^f		No
3′,4′,5′-Trihydroxy ^a^,b	No				Only C6–C5 (M, S)	No			No
3′,5′-Dihydroxy^a^,b	No	Yes	Yes		Yes (S, G)	No		No	Yes (B, G)

a C₆–C₅ precursors yield significant amounts of 5-(phenyl)-γ-valerolactones which are not β-oxidation substrates.

b C₆–C₃ and C₆–C₅ precursors, with the exception of the alkyl-resorcinols, are susceptible to gut microbiota dehydroxylation.

c In mitochondria (enoyl-CoA reductase) and microsomes (Zhao et al. 2019).

d In mitochondria (Ranganathan and Ramasarma 1974).

e Ileostomists (Stalmach et al. 2010).

f Volunteers within 1 hour of consuming 3′,4′-dimethoxycinnamic acid (Farrell et al. 2012).

The best evidence for the β-oxidation of ring-substituted ω-phenyl-alkanoic acids 1 is provided by studies with 4′-hydroxy and 3′,5′-dihydroxy long-chain ω-phenyl-alkanoic acids 1 which yield C₆–C₁ metabolites after multiple cycles (3.4.3, 3.4.13, Figures 8 and 12). The excretion of C₆–C₃ cinnamic and phenyl-hydracrylic acid intermediates plus phase-2 conjugates suggests that the enoyl-CoA hydratase and hydroxy-enoyl-CoA dehydrogenase can be overloaded in the final cycle. These intermediates have not been observed in earlier cycles but may have been overlooked. Metabolism begins in the peroxisomes and continues in the mitochondria. The overlap and point of transfer are uncertain, but possibly at the C₆–C₃ stage after conjugation to carnitine for 3′,5′-dihydroxyphenyl-substituted metabolites (3.4.13, 5.1 and Figure 8). The active enzymes have not been precisely identified but there is evidence that, because of their greater bulk, short sidechain ω-phenyl-alkanoic 1 and ω-phenyl-alkenoic acids 2 require an enzyme able to accept medium chain length fatty acids, for example LCKAT rather than SCKAT—see Part 2.

Medium chain C₆–C₅ ω-phenyl-alkanoic acids 1 with 3′-hydroxy substituents produced from dietary (poly)phenols by the gut microbiota pass through two β-oxidation cycles, although the excretion of 3-hydroxy-3-(3′-hydroxyphenyl)propanoic acid 58 suggests that the second cycle can be overloaded, especially when influx from amino acid catabolism is high (3.4.3 and Figure 10). Medium chain C₆–C₅ ω-phenyl-alkanoic acids 1 with 3′,4′-dihydroxy substituents certainly pass through one cycle (3.4.8 and Figure 16), but those with 3′,4′,5′-trihydroxy substituents (3.4.16 and Figure 20) are disposed of only after one or two microbial dehydroxylations. Excretion within four hours of ca 3 μmol 3,4-dihydroxy C₆–C₁ metabolites after consuming only 1.7 μmol of 3,4-dihydroxybenzoic acid 65 suggests limited β-oxidation of the C₆–C₃ precursor (3.4.8 and Figure 16).

For all other patterns of substituents, data are available only for C₆–C₃ substrates entering β-oxidation, and possibly metabolism will be restricted to the mitochondrion. Because the associated C₆–C₁ compounds may be present in the diet and may form by routes other than β-oxidation (CYP450—see 3.4.6, 3.4.12, 3.4.16, Figures 15 and 19), it is difficult to judge the extent of C₆–C₃ β-oxidation from the yield of C₆–C₁ metabolites. The excretion of small amounts of 2′-hydroxycinnamic acid 53b accompanied by substantial amounts of 2′-hydroxyhippuric acid 102 suggest that β-oxidation has occurred (3.4.1 and Figure 9). There is evidence of β-oxidation of 3′,4′-dimethoxycinnamic acid 30 in rats (3.4.12 and Figure 19), but human metabolism has not been extensively investigated and only “reverse” hydrogenation has been reported.

There is evidence for modest β-oxidation of 4′-methoxycinnamic 72 (3.4.6 and Figure 15), 3′-hydroxy-4′-methoxycinnamic 28 (3.4.11 and Figure 18) and 3′-methoxy-4′-hydroxycinnamic acid 79 (3.4.9 and Figure 17) but the excretion the C₆–C₃-glycine conjugates, phase-2 conjugates of 28 and 79 plus the two 4′-methoxyphenyl-substituted phenyl-hydracrylic acids (73, 89) in substantial quantities indicates that other routes are more important.

The magnitude of person-to-person variation is clearly indicated by data for individual volunteers consuming orange juice with excretion of 3′-hydroxy-4′-methoxy substituted C₆–C₃ metabolites ranging from undetectable to 186 μmol/24 hours and undetectable to 92 μmol/24 hours after consumption of 348 and 111 μmol hesperetin, respectively. Failure to detect any of these metabolites suggests that the gut microbiota did not produce the phenyl-propanoic acid 44, but for producers, variation in the relative amounts of the phenyl-propanoic 44, cinnamic 28 and phenyl-hydracrylic acid 89 excreted indicate variation in endogenous metabolism (3.4.11, 5.2.4, Figure 18, Table 2). Such variation makes it difficult to compare metabolism across different phenyl-ring substituent patterns because in most cases different groups of volunteers have been involved. In studies where one group of volunteers have consumed simultaneously several substrates with more than one phenyl-ring substituent pattern the amounts of each consumed have differed by an order of magnitude.

The failure to excrete a β-oxidation intermediate might indicate that it is not formed, but equally it might indicate it does not accumulate. The tendency to excrete the cinnamic and/or the phenyl-hydracrylic acid suggests that the enoyl-CoA hydratase and/or the hydroxy-acyl-CoA dehydrogenase are susceptible to inhibition by the phenyl moiety when some undefined but critical flux is exceeded. It is impossible to judge for any individual whether this limiting value is the same for all phenyl-ring substituent patterns, and whether the critical factor is total flux, i.e., the sum of the fluxes of every phenyl-ring substituent pattern being metabolized simultaneously, or that value plus the simultaneous flux of fatty acids. Whether fatty acid β-oxidation can be impeded by the ω-phenyl-alkenoic acids 2, or vice versa, remain open questions.

Because of constraints on analytical sensitivity, many human studies have used doses that greatly exceed typical dietary intakes, possibly generating excessive fluxes, and therefore the results obtained are unlikely to represent the typical dietary situation. This limitation applies also to animal studies, even if the human metabolic pathways are the same (see Figures 9, 19, 20 and 22).

Dietary intakes of a given commodity vary extensively across and within populations, and the flux of substrates varies in parallel. The flux also varies with the phenyl-ring substituent pattern because certain substrates are present in foods and beverages at higher concentrations, for example concentrations of 3′,4′-dihydroxyphenyl-substituted are greater than 4′-hydroxyphenyl-substituted, in turn greater than 3′,5′-dimethoxy-4′-hydroxyphenyl-substituted substrates, and coffee, a major source of 3′,4′-dihydroxycinnamic acid 27 may be consumed at regular short intervals whereas sources of 4′-hydroxycinnamic acid 29 (such as apples and artichokes) and 3′,5′-dimethoxy-4′-hydroxycinnamic acid 31 (such as citrus fruit, brassicaceous vegetables and olives) tend to be consumed less frequently. Ideally, in assessing the ease with which substrates having different substitution patterns pass through β-oxidation, a controlled dietary load of both phenolic substrate and fatty acids should be used, but such studies have never been performed.

It is intriguing that substrates having the two phenyl-substituent patterns which frequently will dominate the diet—3′,4′,5′-trihydroxyphenyl (3.4.16, Figure 20) and 3′,4′-dihydroxyphenyl (3.4.8, Figure 16)—are substantially diverted to the 3′-hydroxyphenyl-pathway (3.4.2, Figure 10) by gut microbiota dehydroxylation followed by extensive phase-2 conjugation. This suggests that there is some advantage associated with the strategy, either to the gut microbiota, or symbiotically to the host—two possibilities are the removal of metabolites potentially susceptible to redox cycling and formation of 1,2-quinones and/or 1,4-quinone-methides, plus reducing the demand for limited pools of coenzyme A.

There is insufficient evidence to determine by how much β-oxidation of ω-phenyl-alkanoic acids 1 can impact on fatty acid β-oxidation or carbohydrate metabolism, but modest effects are plausible. The existence of multiple pathways for the clearance of C₆–C₃ ω-phenyl-alkanoic acids 1 implies that the effect will be mechanistically complex—for summary see Table 5 and Figure 25.

7. Suggestions for future studies

1. Some C₆–C₅ and C₆–C₃ β-oxidation intermediates, particularly the ω-phenyl-hydracrylic acids 3, have not been sought in urine and re-interrogation of stored LC–MS TIC data would be useful. In future such intermediates plus the less common amino acid conjugates and acyl-glucuronide conjugates should be sought routinely.

2. Comparison of cell fractions, peroxisomes and mitochondria for uptake mechanisms and ability to metabolize C₆–C₅ and C₆–C₃ ω-phenyl-alkanoic acids 1 and ω-phenyl-alkenoic acids 2 with a range of phenyl-ring substituents would be helpful to elucidate the contribution of the different CoA pools. In addition, cells lacking one of the metabolic enzymes, either naturally or by knockdown, would be useful for indicating the individual contribution of different pathways. Interactions with mitochondrial FAS and the metabolism of cis-isomers of the ω-phenyl-alkenoic acids 2 should be included.

3. MCAD deficiency is a rare metabolic condition, and it may be impractical to mount a study using such individuals to assess the contribution of this enzyme to ω-phenyl-alkanoic acids 1 metabolism. An alternative would be to have healthy volunteers consume foods/beverages providing substantial amounts of the ω-phenyl-alkanoic acids 1 with and without a particular high fat meal, with fatty acids targeted at the medium chain enzyme, to see if this affected the metabolic fingerprint.

Enzymes

Acetyl-CoA carboxylase (EC:6.4.1.2)

Acyl-CoA: cholesterol acyl transferase (EC 2.3.1.26)

Acyl-CoA-dehydrogenase long chain (EC 1.3.8.8)

Acyl-CoA-dehydrogenase medium chain (EC 1.3.8.7)

Acyl-CoA-dehydrogenase short chain (EC 1.3.8.1)

Acyl-CoA-oxidase (EC 1.3.3.6)

Acyl-CoA-thioesterase (EC 3.1.2.2)

Δ3,Δ2-Enoyl-CoA Isomerase (EC 5.3.3.8)

Carnitine O-acetyltransferase (EC 2.3.1.7)

Carnitine-O-palmitoyl transferase (EC 2.3.1.21)

Carnitine O-octanoyl transferase (EC 2.3.1.137)

Catechol-O-methyl transferase (EC 2.1.1.6)

Enoyl-CoA reductase (EC 1.3.1.44)

Fatty acid synthase (EC 2.3.1.85)

Hepatic medium chain (butyryl) acyl-CoA ligases (synthetases), HXM-A (also known as ACSM2B) (EC 6.2.1.2) and HXM-B (also known as ACSM1) (EC 6.2.1.1)

HMG CoA reductase (EC 1.1.1.88)

Hydroxy-acyl-CoA dehydrogenase (EC 1.1.1.35)

3-Hydroxy-acyl-CoA epimerase (EC 5.1.2.3)

Long chain 3-hydroxy-acyl-CoA dehydrogenase (EC 1.1.1.211)

Long chain keto-acyl-CoA thiolase (EC 2.3.1.16)

Mitochondrial acyl-CoA:glycine N-acyltransferase (GLYAT) (EC 2.3.1.13 = 2.3.1.71) Mitochondrial acyl-CoA:glycine N-acyltransferase GLYATL1 (EC 2.3.1.14 = 22.3.1.68)

Short chain keto-acyl-CoA thiolase (EC? ??)

Trans-2-enoyl-CoA reductase (NADPH) (EC 1.3.1.38)

Abbreviations

ACAT	=	Acyl-CoA: cholesterol acyl-transferase
ACC	=	Acetyl-CoA carboxylase
CoA	=	Coenzyme A
COMT	=	Catechol-O-methyl transferase
CPT1	=	Carnitine-O-palmitoyl transferase
DOPA	=	3′,4′-Dihydroxyphenylalanine
FAS	=	Fatty acid synthase
GIT	=	Gastro-intestinal tract
GLYAT	=	Mitochondrial acyl-CoA:glycine N-acyltransferase
GLYATL1	=	Mitochondrial acyl-CoA:glycine N-acyltransferase
HMG-CoA-reductase	=	Hydroxy-methyl-glutaroyl-CoA reductase
HXM-A (also known as ACSM2B)	=	Hepatic medium chain (butyryl) acyl-CoA ligase (synthetase)
HXM-B (also known as ACSM1)	=	Hepatic medium chain (butyryl) acyl-CoA ligase (synthetase)
l-BP	=	l-bifunctional protein
d-BP	=	d-bifunctional protein
LC–HRMS	=	Liquid chromatography–high resolution mass spectroscopy
LOD	=	Limit of detection
LCKAT	=	Mitochondrial long chain keto-acyl-CoA thiolase
MCAD	=	Mitochondrial medium chain acyl-CoA dehydrogenase
MS	=	Mass spectroscopy
NAFLD	=	Nonalcoholic fatty liver disease
NASH	=	Nonalcoholic steatohepatitis
n.d	=	not detected
NMR	=	Nuclear magnetic resonance
SCKAT	=	Mitochondrial short chain keto-acyl-CoA thiolase
VLCFA	=	Very long-chain fatty acids

ACAT	=	Acyl-CoA: cholesterol acyl-transferase
ACC	=	Acetyl-CoA carboxylase
CoA	=	Coenzyme A
COMT	=	Catechol-O-methyl transferase
CPT1	=	Carnitine-O-palmitoyl transferase
DOPA	=	3′,4′-Dihydroxyphenylalanine
FAS	=	Fatty acid synthase
GIT	=	Gastro-intestinal tract
GLYAT	=	Mitochondrial acyl-CoA:glycine N-acyltransferase
GLYATL1	=	Mitochondrial acyl-CoA:glycine N-acyltransferase
HMG-CoA-reductase	=	Hydroxy-methyl-glutaroyl-CoA reductase
HXM-A (also known as ACSM2B)	=	Hepatic medium chain (butyryl) acyl-CoA ligase (synthetase)
HXM-B (also known as ACSM1)	=	Hepatic medium chain (butyryl) acyl-CoA ligase (synthetase)
l-BP	=	l-bifunctional protein
d-BP	=	d-bifunctional protein
LC–HRMS	=	Liquid chromatography–high resolution mass spectroscopy
LOD	=	Limit of detection
LCKAT	=	Mitochondrial long chain keto-acyl-CoA thiolase
MCAD	=	Mitochondrial medium chain acyl-CoA dehydrogenase
MS	=	Mass spectroscopy
NAFLD	=	Nonalcoholic fatty liver disease
NASH	=	Nonalcoholic steatohepatitis
n.d	=	not detected
NMR	=	Nuclear magnetic resonance
SCKAT	=	Mitochondrial short chain keto-acyl-CoA thiolase
VLCFA	=	Very long-chain fatty acids

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

The author(s) reported there is no funding associated with the work featured in this article.

Notes

1 l-Carnitine is also referred to as (–)-carnitine. It is 3R-hydroxy-4-(trimethylammonio)butanoate — see 156, 158 and 159, Figure 3.

2 According to Feher (2017, page 723) a typical 70 kg adult male excretes about 2 g (ca 0.18 mol) creatinine per day suggesting that MCAD patients excrete ca 0.2–0.7 μmol 3-(phenyl)propanoyl-glycine 22 per day compared with not more than ca 100 nmol per day for healthy controls (Feher 2017).

3 This is a substantial dose corresponding to 3.25 g for a 65 kg adult.

4 These are massive doses corresponding to 13 g for a 65 kg adult.

5 According to Feher (2017, page 723) a typical 70 kg adult male excretes about 2 g creatinine per day suggesting excretion of 3-hydroxy3-(3′-hydroxyphenyl)propanoic acid might be of the order of 20 to 1,000 µmol per day (Feher 2017).

6 According to Feher (2017, page 723) a typical 70 kg adult male excretes about 2 g creatinine per day (Feher 2017).

7 1% of diet is a substantial dose. If a 300 g rat eats 30 g/day it equates to 300 mg or 1.67 mmol

8 Unless one of these hydroxyls or conjugates is on the side chain, these must be 3-(3′-hydroxyphenyl)propanoic acid-5′-sulfate, and 5-(3′-hydroxyphenyl)pentanoic acid-5′-glucuronide because the phenyl ring is symmetrically substituted.

9 According to Feher (2017, page 723) a typical 70 kg adult male excretes about 2 g (ca 0.18 mol) creatinine per day. suggesting ca 6–30 μmol phenylacetylglycine (Feher 2017).

10 Note that rat liver mitochondria in vitro and rats in vivo can glycine conjugate pre-existing 3-hydroxy-4-methoxybenzoic acid 47 efficiently (Kasuya, Igarashi, and Fukui 1990).

11 This is a substantial dose. Assuming a 250 g rat consumes 25 g of food per day this corresponds to 135 µmol/kg or ca 1.5 g for a 65 kg adult.

12 According to Feher (2017, page 723) a typical 70 kg adult male excretes about 2 g creatinine per day (Feher 2017).

13 Smith and Stern commented in 1983 ‘For aromatic carboxylic acids to influence directly the de novo fatty acid synthesis pathway would necessitate access of the CoA thioesters of these compounds to the cytosolic compartment’ but more recently a mitochondrial FAS has been reported (Zhang, Joshi, and Smith 2003) now known to coordinate oxidative metabolism in mammalian mitochondria (Nowinski et al. 2018). Interactions between mitochondrial FAS and ω-phenyl-alkanoic acids 1 have not been investigated.

14 Kim et al. claim this is a metabolite of hesperetin but appear to have confused it with 3′-hydroxy-4′-methoxycinnamic acid 44 — see 3.4.11, Figure 18 and Table 2.

15 This is a substantial dose. If a 300 g rat consumes 30 g feed per day it corresponds to 198 µmol, substantially more on a body weight basis than used in the orange juice volunteer studies — see 3.4.11, Figure 18 and Table 2.