The Paraguayan gut microbiome contains high abundance of the phylum Actinobacteriota and reveals the influence of health and lifestyle factors

ABSTRACT

Most gut microbiome studies are focused on populations from developed nations. However, this overlooks the role played by host genetics, lifestyle, and diet, highlighting the need to evaluate under-represented populations. Thus, we performed the first gut microbiome study from a Paraguayan cohort via 16S rRNA sequencing and PICRUSt2 analysis. We evaluated fecal samples from 60 participants from Asunción, while considering categories such as body-mass-index (BMI), sex, age, diet, lifestyle, and clinical history. Firmicutes (76.0 ± 11.6%), Actinobacteriota (10.4 ± 7.9%) and Bacteroidota (9.4 ± 11.4%) were the most abundant phyla. Similarly, the most abundant genera were Blautia (14.1 ± 7.5%), Faecalibacterium (8.07 ± 6.8%), and Bacteroides (6.7 ± 6.8%). Likewise, the most abundant microbial pathways were predicted to be involved with sugar metabolism and fermentation. Interestingly, some categories significantly impacted the gut microbiome composition and function, such as BMI (Moryella, Bifidobacterium), sex (Faecalibacterium), and others. Additionally, dysbiotic indices differed from values previously reported as homeostatic. These observations highlight the need for further studies, considering microbial species and host genetics. Thus, this work expands the knowledge of the gut microbiome from the Collective South, while contrasts found herein reinforce the need for further research with human populations traditionally understudied.

GRAPHICAL ABSTRACT

KEYWORDS:

• Gut
• microbiome
• Paraguay
• microbial
• diversity
• chronic
• illnesses
• diet
• habits
• Actinobacteriota
• collective
• South

Introduction

Trillions of highly diverse microbes colonize the human gut, collectively known as the gut microbiome. The structure and diversity of this microbial community have been linked to host physiology in health and disease. For instance, the gut microbiome has been associated with the onset of cancer, lean or obese physiological states, cardiovascular disease, intestinal inflammation, wound healing, and even mental disorders.^1–5 However, the mechanisms behind many of these associations remain elusive.

The human gut microbiome is composed mainly of bacteria and archaea, belonging to the phyla Bacteroidetes and Firmicutes, and Euryarchaeota, respectively.^6,7 Although the taxonomic composition may vary substantially among subjects, the most abundant microbial metabolic pathways are highly prevalent, pointing toward functional redundancy.⁸ In that sense, the ability of microbes to colonize the gut depends on many metabolic capabilities, including diverse carbon/energy utilization capabilities, and the ability to withstand host-immune responses and physicochemical stressors.^9–14 Moreover, microbial pathways and their metabolites may impact host physiology, such as by generating short-chain fatty acids (SCFA).¹⁵

Most human microbiome studies have been centered in North America, Europe, and Asia, therefore, populations from the Collective South are underrepresented.¹⁶ Importantly, numerous factors drive the microbiota composition, including ethnicity, race, and culture.⁸ Thus, a “normal”, or homeostatic microbiome is not necessarily shared across regions. To our knowledge, there has not been any previous efforts to elucidate the gut microbiome composition from a Paraguayan cohort.

Filling this gap, we characterized the gut microbiome diversity, structure, and functional-potential from 60 fecal donors residing in the Asunción metropolitan area. We also analyzed the participants’ gut microbiome based on their body mass index (BMI), along with other categories such as sex, age, diet, lifestyle, and clinical history. While most observations coincided with previous reports, new correlations were unveiled. Furthermore, dysbiotic index calculations, including the ratios of Firmicutes/Bacteroidota (F/B), and Log Prevotella/Bacteroides (Log P/B), demonstrated that the values for healthy subjects did not align with previous reports. Both observations highlight the need for further studies, considering gut microbial species, local habits, and host genetics.

Overall, this work, the first with a Paraguayan cohort, expands the knowledge of the gut microbiome structure and diversity of the Collective South. The contrasts found herein reinforce the need for further studies with human populations traditionally understudied, especially, given the rise of medical interventions targeting the gut microbiome.

Results

Demographics and data description

Sixty individuals were selected for the present study, corresponding to 33 overweight (OW, BMI > 25) and 27 normal weight (NW, 18.5 < BMI < 25) individuals, based on their body mass index (BMI) according to the World Health Organization (WHO) (Supplemental Figure S1). Among these, 44 and 16 were women and men, respectively. Regarding age, values ranged from 19 to 58 years old, with a mean of 32.5 years of age. Geographic location: 37 were from Asunción, 3 from Capiata, 5 from Fernando de la Mora, 4 from Lambare, 1 from Limpio, 3 from Luque, 1 from Ñemby, 4 from San Lorenzo, 1 from Mariano Roque Alonso, and 1 from Villa Elisa (Supplemental Figure S2A). Each was given a questionnaire regarding dietary habits and clinical history. Figure 1 shows the questionnaire’s data distribution among participants. Overall, the majority did not declare any base conditions, were omnivores, performed physical activities, and consumed vitamin supplements. However, when segregating the data based on sex o slim/obese states, some notable differences were the higher prevalence of underlying conditions on obese participants, the higher occurrence of constipation among slim and female individuals, and, that fish consumption was more common among female and slim participants. Supplemental Table S1 contains the raw metadata.

Figure 1. Data distribution of participants’ responses on the provided questionary. Data was separated based on participants’ sex, BMI (normal weight [NW] and overweight [OW]), and the combination of both sex and BMI. The percentage of positive responders among each group is shown. The categories include diet, habits, and clinical history.

Microbial diversity and richness

Microbiome analysis was performed via 16S rRNA sequencing. The median read count per sample was 144,030 and operational taxonomic unit (OTU) total count ranged from 50 to 310, with a mean of 190. Shannon index values, reported at a sequencing depth of 5,000 reads across samples (Figure 2A-B), were not significantly different among male (5.61 ± 0.53) and female (5.69 ± 0.52). However, values differed significantly among normal weight (5.81 ± 0.50) and overweight individuals (5.54 ± 0.50), regardless of sex. The mean value across samples was 5.67 ± 0.52. When considering age groups, there was a decrease in diversity in older groups, except with 54–58 group, whose alpha diversity was higher (Supplemental Figure S2C).

Figure 2. Fecal microbiome composition of Paraguayans from the Metropolitan area of Asunción based on 16S rRNA sequencing. (A) Alpha-diversity (Shannon index) distribution across samples, calculated at a depth of 5000 reads. (B) Shannon index values based on sex and BMI values distributed as normal and overweight. (C) Microbial relative abundance at the Phylum level across samples. (D) the twenty most abundant genera across samples. Colors tabulated based on their phyla.

Overview of the microbial relative abundance and prevalence across samples

In terms of phylum prevalence (Supplemental Figure S2B), Firmicutes, Bacteroidota, Proteobacteria, and Actinobacteria were present in 100% of samples. When considering relative abundance, the sixty Paraguayan samples were dominated by Firmicutes with a median of 76.0 ± 11.6%, followed by Actinobacteriota and Bacteroidota at 10.4 ± 7.9 and 9.4 ± 11.4%, respectively (Figure 2C). Regarding gut dysbiotic indices,^17,18 the Firmicutes/Bacteroidetes (F/B) ratio was 8.04 across samples, 9.23 and 7.93 for male and female, respectively, and 12.11 and 6.66 for overweight and normal weight individuals, respectively. At the genus level, the ten most abundant were, in order, Blautia, Faecalibacterium, Bacteroides, Collinsella, Bifidobacterium, Subdoligranulum, Agathobacter, CAG-352, Eubacterium hallii group, and Dorea (Figure 2D). A principal coordinate analysis and heatmap across samples, considering sex and BMI, are shown in Supplemental Figure S2 and S3.

Correlations between the gut microbiome structure with habits, diet, and clinical history

A multivariate association analysis was performed on the questionnaire previously described, correcting for multiple tests (q < 0.25) and controlling for confounding effects, via the Microbiome Multivariable Association with Linear Models 2 (MaasLin2) R package.¹⁹ The selection of covariates was based on previously published data²⁰ and our data distribution. If a category did not have any effect on the gut microbiome by itself, it was omitted from further analysis as a covariate. These are mentioned in Materials and Methods. Furthermore, chronic intestinal inflammation conditions were pooled into one category called gastrointestinal (GI) symptoms. These included Crohn´s disease, irritable bowel syndrome (IBS), leaky gut, and broadly “intestinal symptoms”, as declared by participants. Finally, fast food or ultra processed foods were considered burgers, fries, soda, pizza, saturated fat, and foods high in cholesterol and low fiber content.²¹ Data was tabulated binarily, as positive, and negative.

At the phylum level (Figure 3A), Actinobacteriota was positively correlated with BMI, while being negatively correlated with the use of salt (NaCl) and vitamins supplementation. Bacteroidota, on the other hand, was positively correlated with vitamins supplementation, as well as with veganism. Cyanobacteria was inversely correlated with intense physical activity, or more than five times a week. Euryarchaeota was correlated with GI symptoms, while being inversely correlated with intense physical activity and with the consumption of probiotics or fermented food. Finally, Verrucomicrobiota was correlated with vitamins supplementation and inversely correlated with intense physical activity. The F/B ratio for participants with GI symptoms was 6.86 and 8.25 for positive and negative responders, respectively. While constipation was not correlated with the gut microbiome structure at the phylum level, its F/B ratio was 6.75 and 8.16 for positive and negative responders, respectively.

Figure 3. Significant associations identified at the phylum (A) and genus (B) levels, via multivariate associations with linear models as implemented by MaAsLin2 R-package¹⁹ (FDR < 0.25), between the gut microbiota structure and the diet, habits, and clinical history of the Paraguayan cohort. Each association analysis was adjusted for fixed effects according to the data distribution and as previously suggested.²⁰

At the genus level (Figure 3B), BMI was correlated with the abundance of Bifidobacterium and Moryella, while being inversely correlated with Eubacterium nodatum group. Constipation was correlated with Allisonella and inversely correlated with Raoultibacter. Fast food consumption was inversely correlated with Dielma. Red meat consumption was inversely correlated with Megasphaera, RF39, Roseburia, and with an uncultured microbe belonging to the order Bacteroidales. Alternatively, white meat consumption was correlated with Candidatus stoquefichus, Megasphaera, and with an uncultured microbe of the Lachnospiraceae family. The GI symptoms category was correlated with DTU014, Merdibacter, Papillibacter, Sanguibacteroides, and with a member of the Tannerellaceae family. Intense physical activity was correlated with Faecalibacterium, Haemophilus, Rikenellaceae RC9 gut group, while being inversely correlated with Akkermansia and Methanobrevibacter. Smoking was correlated with Gemella, while the use of salt (NaCl) was inversely correlated with Frisingicoccus. Among vegans, a significantly higher abundance of RF39 and Eubacterium coprostanoligenes group were detected. Finally, in terms of sex, a significantly lower abundance of DTU089, Defluviitaleaceae UCG.011, Faecalibacterium, and an uncultured genus belonging to order Coriobacteriales among male participants were detected. Contrastingly, female participants contained lower abundance of Enterorhabdus. In terms dysbiotic indices at the genus level,²¹ the log Prevotella/Bacteroides (Log P/B), was −1.53 across samples, while for male and female participants were −1.11 and −1.88, respectively. Similarly, for overweight and normal weight individuals the values were −0.97 and −2.25, respectively. For people with and without GI symptoms, the corresponding values were −2.48 and −1.14, respectively. Finally, for those with and without constipation the values were −2.50 and −1.10, respectively. The full list, including statistical parameters are shown in Supplemental Table S2.

Microbial metabolic potential across samples and categories

Based on PICRUSt2²² metabolic potential prediction, there were 304 microbial pathways across samples, of which 190 were 100% prevalent (Supplemental Table S3). The ten most abundant microbial pathways were pentose phosphate pathway (non-oxidative branch), pyruvate fermentation to isobutanol pathway (engineered), starch degradation V, L-isoleucine biosynthesis II, glycogen biosynthesis I (from ADP-D-Glucose), sucrose degradation III (sucrose invertase), glycogen degradation I (bacterial), Calvin-Benson-Bassham cycle, L-isoleucine biosynthesis I (from threonine), and L-valine biosynthesis (Figure 4). To identify correlations between microbial pathways abundances and the metadata, a multivariate association analysis was performed following the previously mentioned tool and criteria. After controlling for multiple testing and potential confounding effects, six categories had a significant effect on the gut microbial metabolic potential: age, BMI, sex, the consumption of fast food, vitamins supplementation, and performing intense physical activity (Figure 5). Regarding age, PWY-7332 superpathway of UDP-N-acetylglucosamine-derived O-antigen building blocks biosynthesis was inversely correlated with this category. BMI was correlated with PWY-6471 peptidoglycan biosynthesis IV (Enterococcus faecium) pathway. Concerning sex, among male participants there was a lower relative abundance of PWY490-3 nitrate reduction VI (assimilatory), and PWY-6749 CMP-legionaminate biosynthesis I pathways. Fast food consumption was inversely correlated with P562-PWY myo-inositol degradation I, and PWY-7237 myo-, chiro- and scillo-inositol degradation pathways. Finally, vitamins supplements consumption and intense physical activity were involved with 108 and 63 microbial pathways, respectively. The full list is shown in Figure 5A-B and in the Supplemental Table S4 and S5. However, the five with the largest effect size were: for vitamins supplementation, FASYN-ELONG-PWY fatty acid elongation – saturated, PWYG-321 mycolate biosynthesis, PWY-7664 oleate biosynthesis IV (anaerobic), PWY-5989 stearate biosynthesis II (bacteria and plants), and PWY-6282 palmitoleate biosynthesis I (from (5Z)-dodec-5-enoate). And for intense physical activity, the five with the largest effect size were PWY-2941 L-lysine biosynthesis II, PWY0-1296 purine ribonucleosides degradation, PWY-6891, thiazole biosynthesis II (Bacillus), PWY0-1297 superpathway of purine deoxyribonucleosides degradation, and PWY0-1298 superpathway of pyrimidine deoxyribonucleosides degradation. Interestingly, 22 pathways were common and displayed inverse correlation along both categories (Supplemental table S6). Additionally, BMI and the consumption of fast food also shared one pathway each with vitamins supplement intake. Specifically, PWY-6471 peptidoglycan biosynthesis IV (Enterococcus faecium) was inversely correlated with vitamins supplementation and positively correlated with BMI. Alternatively, PWY-7237 myo-, chiro- and scillo-inositol degradation was correlated with vitamins supplementation, while being inversely correlated with fast food consumption.

Figure 4. Functional characterization of the gut microbiome of Paraguayans based on PICRUSt2²² analysis of 16S rRNA data. The 27% most abundant across samples are shown.

Figure 5. Significant associations identified via multivariate associations with linear models as implemented by MaAsLin2 R-package¹⁹ (FDR < 0.25), between the gut microbiota functional potential (i.e., PICRUSt2 data) and the diet, habits, and clinical history of the Paraguayan cohort. Top 20 pathways associated with physical activities (A), and intake of vitamins and supplements (B), respectively. (C) pathways associated with habits and clinical history. Each association analysis was adjusted for fixed effects according to the data distribution and as previously suggested.²⁰

Discussion

The human microbiome has received much attention due to its role in health and disease. However, there is a lack of representation from the Collective South. To our knowledge, the Paraguayan gut microbiome has not been previously evaluated. Thus, we performed its first characterization, evaluating the fecal microbial structure, diversity, and functional potential of sixty adult individuals from the urban cluster of Gran Asunción.

Based on this, we uncovered that most participants did not have any base conditions, were born by vaginal delivery, had an active lifestyle, and their diet was primarily based on meat, grains, tubercules, and fruits/vegetables. A previous study reported that while urban and rural households had similar access to different food groups, socioeconomic factors had a significant effect on food access.²³ In our study, this element was not considered.

At the phylum level, the Paraguayan gut microbiome was dominated by Firmicutes, followed by Actinobacteriota and Bacteroidetes. Compared to regional cohorts, it differed from Argentina, Chile, Colombia, and Brazil.^24,25 Interestingly, samples from Paraguay looked more like those of Italy and the Netherlands, that also displayed a high relative abundance of Actinobacteriota.^24,26 Regarding dysbiotic indices, the F/B ratio also differed from what is typically considered homeostatic (i.e., <1.5).^17,18,21 In our study, the absence of GI symptoms was > 7.9 for F/B. Noteworthy, the F/B values for normal weight individuals coincided with intestinal symptoms (i.e., <7), suggesting that it was not a predictor for a healthy state. The F/B ratio is usually associated with normal intestinal homeostasis where changes in this ratio is related as dysbiosis, potentially leading to several pathologies.²⁷ Moreover, at the genus level, while the most abundant found herein are considered normal gut microbes, Collinsella, the fourth most abundant genus, has previously been associated with several health issues. These include obesity,²⁸ type 2 diabetes (T2D),²⁹ systemic inflammation,³⁰ hypercholesterolemia,³¹ while also being correlated with the western diet.³² While dysbiotic indices intent to classify intestinal homeostasis,²⁷ our findings underscore the challenges of defining a normal or homeostatic gut microbiome, and of using the same rule across populations.

Among the identified correlations, sex is a known predictor of the gut microbiome structure, whereby females contain higher diversity²⁰ and relative abundance of SCFA-producing microbes.²⁶ Similarly, Faecalibacterium was more abundant among females in our study, a microbe that shares this metabolic trait, and as previously observed.³³ In terms of BMI, the phyla Firmicutes, Proteobacteria and Fusobacteria were previously correlated with this category, but not always. Conversely, Actinobacteriota was only inversely correlated.³⁴ Contrastingly, our data reveals that Actinobacteriota is positively correlated with BMI. At the genus level, several were previously found inversely correlated with this category, however, none of them coincided with our results.^35–37 Furthermore, the positive association between Bifidobacterium and BMI found herein is noteworthy. This genus was previously inversely correlated with BMI, contrasting with our data.^38–40 However, a meta-analysis identified Bifidobacterium species associated with different diets,⁴¹ pointing toward its metabolic diversity. Since host genetics also plays a key role in the gut microbiome structure,⁴² identifying the Bifidobacterium species more abundant among Paraguayans will expand our knowledge about this microbe and its role in host physiology. In terms of alpha diversity, the higher values among normal weight individuals agrees with previous studies.^1,43 Overall, the mechanisms driving the gut microbiome – BMI association are still debated; however, they might relate to energy generation via SCFA by gut microbes.⁴⁴

When considering GI symptoms, it was previously shown a higher relative abundance of Firmicutes, Proteobacteria, and Actinobacteriota among Crohn’s disease patients, and higher relative abundance of Firmicutes, Bacteroidetes and Actinobacteriota in ulcerative colitis.⁴⁵ This contrasts with our data since only Euryarchaeota was correlated with this category. Here, however, we have pooled several gut inflammation conditions together, which does not allow to identify correlations with specific clinical diagnostics. At the genus level, however, previous studies agree with our results.^46–49 Moreover, we have identified additional correlations: DTU014 and the family Peptococcaceae. DTU014 contains genes associated with pathogenesis⁵⁰ and can utilize SMFA as carbon/energy source.^51,52 Alternatively, to our knowledge, the family Peptococcaceae was not previously correlated with this category.

In terms of constipation, the families Lachnospiraceae and Ruminococcaceae, and the genera Agathobacter and Dorea were previously correlated with this category.⁵³ While these groups belong to the phylum Firmicutes, another study found its inverse correlation with constipation.⁵⁴ Here, we identified additional genera correlated with this category, Allisonella, a Firmicute, and Raoultibacter, an Actinobacteriota.

Regarding the Log P/B ratio, values above −0.15 were previously associated with the ease of body weight loss via calorie restriction. Here, however, values for normal weight individuals and those with GI symptoms were <−2.25. The absence of the latter corresponded to >−1.1, while overweight participants were −0.97. Thus, here, the Log P/B ratio for normal weight individuals without intestinal symptoms was −1.1 < Log P/B < −0.97, supporting the notion for the challenges of a universal dysbiotic index.

In terms of diet, Bacteroidota was previously correlated to a high-salt diet,⁵⁵ contrasting with our results. Here, however, we considered salt intake as a binary category, regardless of its amount. Interestingly, despite the WHO guidelines of 5 g/day of salt intake, Paraguayan average is 13 g/day.⁵⁶ Meat consumption, on the other hand, was previously inversely correlated with Anaerostipes and Faecalibacterium and positively correlated with Roseburia.⁵⁷ However, the authors of this metanalysis highlight inconsistencies in directionalities among studies or the lack of data about meat subtypes. Contrasting with our results, Roseburia displayed a lower relative abundance among red meat consumers. Additional inverse correlations found herein included RF39 and Megasphaera. The latter was previously correlated with a healthy gut.⁵⁸ Alternatively, a high fish and seafood, and low red meat consumptions, were correlated with the relative abundances of Coprococcus and Roseburia.^59,60 On the other hand, ultra processed food intake was previously correlated with bile-tolerant microbes, such as Bilophila, Alistipes and Bacteroides,⁶¹ contrasting with our finding, that showed a negative correlation with Dielma. Interestingly, this microbe was inversely correlated with BMI in children.⁶² Among vegans, Bacteroides was previously shown to be enriched among this group, coinciding with our data.^63,64 Additionally, genus RF39, a Firmicute, was also identified to be correlated with veganism in this study. This group was previously correlated with a healthy lifestyle and with fruits- and vegetables-rich diet.⁶⁵ Fermented food and/or probiotics consumption was previously associated with Bacteroides, Dorea, Prevotella, Faecalibacterium, among others.⁶⁶ However, we have only observed a reduction in the relative abundance of Euryarchaeota.

In terms of habits, people that perform intense physical activity contain a higher relative abundance of SCFA-producing microbes such as Faecalibacterium and Akkermansia, while a lower abundance of Methanobrevibacter.⁶⁷ This agrees with our results, except for the lower relative abundance of Akkermansia, the additional genera correlated with this category identified, and the negative correlation with Cyanobacteria and Verrucomicrobiota. In terms of vitamin supplementation, previous studies have shown that vitamins C and E increase the relative abundance of Bifidobacterium.⁶⁸ Similar effect was observed with Alistipes and Clostridium by vitamin B2 intake.⁶⁹ Additionally, vitamin C supplementation has been shown to increase fecal SCFAs levels.⁷⁰ Contrastingly, our data shows higher and lower relative abundance of Bacteroidota and Actinobacteriota, respectively, within this category. Finally, smoking was previously inversely correlated with the genus Gemella, contrasting with our results.⁷¹ Taken together, while we could argue that certain habits could lower the abundance of disease-associated taxa, such as increasing physical activity to lower the abundance of GI-symptoms-associated Euryarchaeota, more data and confirming microbial identities are key for asserting strong recommendations. Furthermore, although we used previously known covariates for analysis, we did not employ the LASSO approach, that while being a powerful tool, it is still relatively underused in microbiome studies.⁷²

The functional profiling via PICRUSt2 prediction revealed the most abundant microbial pathways across the Paraguayan cohort, largely agreeing with a previous metanalysis.⁷³ Significant associations with some of the categories evaluated were also identified. In terms of BMI, coinciding with our results, peptidoglycan biosynthesis IV was previously correlated with BMI among people with insulin-related pathologies.^74,75 Considering age, here inversely correlated with superpathway of UDP-N-acetylglucosamine-derived O-antigen building block synthesis, was previously correlated with child growth.⁷⁶ Fast food consumption was inversely correlated with inositol degradation pathways, potentially indicating a decrease in its availability for gut microbiomes use. Myo-inositol supplementation improves bone stability and general performance in animal models.^77–79 Furthermore, myo-inositol and chiro-inositol supplementation facilitates insulin signaling, reducing glycemia in disorders linked to insulin resistance.^80–82 Accordingly, inositol deficiency was previously correlated to metabolic syndrome and diabetes.^83,84 Intense physical activity was correlated with numerous microbial pathways. Some noteworthy are N10-formyl-tetrahydrofolate biosynthesis, which was more abundant, and the superpathway of L-methionine biosynthesis, less abundant. The expression of genes responsible for folate/methionine biotransformations is correlated with physical activity and body weight change.⁸⁵ Additionally, homolactic fermentation pathway decreased among this group, agreeing with previous observations.⁸⁶ Bradly, athletes contain significantly higher abundance of microbial pathways related to carbohydrate and secondary metabolite metabolism, and biosynthesis of organic cofactors.⁸⁷ In terms of sex, differences in nitrate and nitrite metabolism were previously observed,⁸⁸ including increased nitrate reductase activity in females.⁸⁹ Finally, vitamins supplementation was correlated with several microbial pathways. Previous studies have focused on specific vitamins supplementation, such as D and B, in the context of obesity or diabetes. Some pathways included folate biosynthesis, amino acids metabolism, lipid and fatty acids biosynthesis, and cofactors and vitamins metabolism.⁹⁰ These microbial functions have been suggested to impair a positive health effect related to lowering blood glycemic levels,⁹¹ and vitamins absorption.^92,93

Overall, this study represents the first characterization of the gut microbiome of a Paraguayan cohort, while considering their diet, habits, and clinical history. While most observations were in line with previous studies, our results also identified new associations. Likewise, dysbiotic indices among healthy subjects differed from other studies. This warrants further studies to verify these associations. Host genetics, lifestyle, and diet may all contribute to these observations. In summary, this underscores the need for further fundamental studies across Collective South populations, especially considering the rise of personalized medicine in the context of the human microbiome.

Materials and methods

Subjects

Sixty urban volunteers from Greater Asunción, Central, participated in this study. Each was provided with a questionnaire about medical history, lifestyle, and diet. Exact categories are presented in the Results section. This data was analyzed to evaluate correlations with the gut microbiome. Exclusion criteria: people that have consumed antibiotics within the last 6 months, that have undergone chemotherapy or used immunosuppressors within the last year, and people whose age was < 18 and >60 years.

Ethics

The volunteers were informed about the nature and purpose of this research, in compliance with the Helsinki Declaration. All participants gave their consent.

Sample collection and prokaryotic (bacterial and archaeal) DNA extraction

Fecal samples were collected in 50 mL sterile Falcon tubes, and participants were asked to freeze them soon after collection. Samples were then stored at −80°C until DNA extraction. DNA extraction was performed using the DNeasy Power Soil Pro Kit (Qiagen, Germany), following manufacturer’s instructions.

16S rRNA sequencing and analysis

Sequencing

Barcoding and library preparation were carried out as described.^94–97 Briefly: 12.5 ng of DNA was amplified using universal primers targeting the V4 region of the bacterial 16S rRNA gene, containing overhang adapters appended to the 5’ end of each primer for compatibility with the Illumina sequencing platform.⁹⁸ The complete sequences of the primers were:

515F–5’ TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGTGCCAGCMGCCGCGGTAA 3’

806 R–5’GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGGACTACHVGGGTWTCTAAT 3’.

Master mixes contained 12.5 ng of total DNA, 0.5 µM of each primer, and 2× KAPA HiFi HotStart ReadyMix (KAPA Biosystems, Wilmington, MA). Thermal profile as recommended by the manufacturer. Each 16S amplicon was purified using the AMPure XP reagent (Beckman Coulter, Indianapolis, IN). Then, each sample was amplified using a limited cycle PCR program, adding Illumina sequencing adapters and dual‐index barcodes (index 1(i7) and index 2(i5)) (Illumina, San Diego, CA) to the amplicon target. The final libraries were again purified using the AMPure XP reagent (Beckman Coulter), quantified, and normalized prior to pooling. The DNA library pool was then loaded on the MiSeq reagent cartridge (Illumina) and the MiSeq instrument (Illumina). Automated cluster generation and paired – end sequencing with dual reads were performed according to the manufacturer’s instructions.

Bioinformatics

Sequencing output from the Illumina MiSeq platform was converted to fastq format and demultiplexed using Illumina Bcl2Fastq 2.18.0.12. The resulting paired-end reads were processed using QIIME 2 2018.11.⁹⁹ Index and linker primer sequences were trimmed using the QIIME 2 invocation of cutadapt. The resulting paired-end reads were processed with DADA2 through QIIME 2, including merging paired ends, quality filtering, error correction, and chimera detection.¹⁰⁰ Amplicon sequencing units from DADA2 were assigned taxonomic identifiers with respect to Green Genes release 13_08 using the QIIME 2 q2-featureclassifier.¹⁰¹ Alpha diversity with respect to: Faith PD whole tree, Evenness (Shannon) index, and observed species number metrics was estimated using QIIME 2 at a rarefaction depth of 5,000 sequences per subsample. Beta diversity estimates were calculated within QIIME 2 using weighted and unweighted Unifrac distances as well as Bray-Curtis dissimilarity between samples at a subsampling depth of 5,000. Results were summarized and visualized through principal coordinate analysis, and significance was estimated as implemented in QIIME 2.

Data analysis

Analysis was performed using the Microbiome Multivariable Associations with Linear Models tool (MaAsLin2)¹⁹ as implemented in R: minimum prevalence of 10%, using total-sum scaling (TSS) normalization to obtain data as relative abundance, and arcsine squared transformation (AST). Benjamini-Hochberg (BH) correction method for multiple testing was used, for a target q-value of q < 0.25, to be considered significant.²⁴ The output, or model coefficient, indicates the difference between the specified category against the reference, which is set by the algorithm, taking the first category in alphabetical order. To control for confounding effects, for each category (predictor), covariates were added, based on known confounders²⁰ and our data distribution. In this study, we did not employ the still underused⁷² LASSO approach to uncover potentially additional covariates. Functional analysis was performed via PICRUSt2 analysis, as previously described.²² STAMP (Statistical Analysis of Metagenomics Profiles)¹⁰² software was used to analyze taxa abundance data between normal and overweight participants (Supplemental Figures S3 and S4). For this test, two-sided Withe’s non-parametric t-test was performed.¹⁰³

Table

Table of predictors vs covariates
Predictor	Covariates
Age	Sex, BMI
Alcohol intake	Sex, BMI
Red meat consumption	None.
Dairy milk consumption	GI symptoms
Sex	GI symptoms, hyper/hypothyroidism
Constipation	BMI, GI symptoms, Sex
BMI	Constipation, GI symptoms, Sex
Fish consumption	BMI, Sex
Legumes consumption	GI symptoms
No physical activity	BMI, Sex
Olive oil consumption	BMI, Sex
Proteins or amino acids consumption	None.
Salt (NaCl) consumption	BMI, Age, No physical activity, GI symptoms
Sugar consumption	Age, Sex, BMI
GI symptoms	Sex, Constipation, dairy milk consumption
Fast food consumption	BMI, Sex
Intense physical activity	BMI, Sex
Probiotic/fermented food consumption	BMI, Sex
Smoke	BMI
Vegan	Sex, dairy milk consumption, Age
Vitamins supplementation	Sex, Age
White meat consumption only	BMI, Sex

Acknowledgments

This project was funded by the Chancellor’s Office at the National University of Asuncion (UNA), project number FACEN/02/21, and by the Paraguayan Society for Microbiota, Probiotics and Prebiotics (SPMpyP). The authors would like to thank Jonas Fernandez, Kamila Pintos, Martin Nuñez, and Lourdes Cardozo for their assistance in the lab, and Nelson Guzman for sample collection. We would also like to thank Danilo Fernandez for his very kind and generous support by providing us with access to hardware for bioinformatic analysis.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data presented in this study are available online at https://dataview.ncbi.nlm.nih.gov/object/PRJNA992776, accession number PRJNA992776.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/29933935.2024.2332988.

Additional information

Funding

This study was funded by the Chancellor’s Office at the National University of Asunción (UNA) with project code FACEN/02/21, and by the Paraguayan Society for Microbiota, Probiotics and Prebiotics (SPMpyP).