Unraveling the Complexities of Starch Retrogradation: Insights from Kinetics, Molecular Interactions, and Influences of Food Ingredients

ABSTRACT

Starch retrogradation is important in controlling many properties of starch-rich foods, while many crucial issues remain to be answered in this field. This review for the first time summarized recent understandings on starch retrogradation property, especially from the kinetics and molecular perspectives. Consecutive reaction kinetics (CR) model was recently proposed to help understand nucleation and crystal growth steps, which are critical in determining the overall retrogradation rate. By applying CR model, relations between starch fine molecular structures and retrogradation property have been revealed. Food ingredients, e.g. sugars, salts, proteins, lipids, polyphenols, and various non-starch polysaccharides, all influence starch retrogradation via different mechanisms. These new insights were comprehensively discussed in this review, aiming to help the development of functional foods with both preferable texture and nutrition property, through a better manipulation of starch retrogradation property during food processing.

KEYWORDS:

• Inter- and intramolecular interaction
• nucleation and crystal growth
• short- and long-term retrogradation
• starch molecular structure
• starch retrogradation

Introduction

Retrogradation can largely determine nutrition and textural properties of starch-rich foods.^[1] When preparing starch-rich foods such as rice, noodle, and bakery products, starch undergoes continuing gelatinization and retrogradation process, resulting in food products with desirable eating quality. Retrogradation occurs over the cooling period of starch-rich foods after gelatinization, forming entanglements and double helices between amylose and amylopectin molecules.^[2] With the progression of retrogradation, a sequence of physical alterations occurs, such as the formation of B-type crystallites, increased hardness and exudation of water.^[3] Starch retrogradation (especially amylopectin molecules) should be inhibited in order to avoid the staling phenomena and reduced shelf-life of food products, e.g., breads and Chinese steamed breads. This is because retrogradation could promote the formation of starch double helices and exudation of water.^[4] On the other hand, retrogradation is a nutritionally desirable process, as it can facilitate the development of slowly digestible starch and resistant starch (RS).^[5] RS is able to reach to the colon, where it is fermented into health-promoting metabolites such as short-chain fatty acids (SCFAs).^[6] The definition, production strategies, and effects of RS on gut microbiota have been comprehensively summarized in recent reviews, and the readers are encouraged to refer to these reviews for more information on RS.^[7] Therefore, a better understanding of starch retrogradation process is necessary in order to develop starch-based foods having both preferable texture and nutrition properties.

Many studies have been performed recently in terms of investigating the complex structural basis for starch retrogradation. Although starch retrogradation has been reviewed previously,^[8,9] the following essential aspects are rarely covered. Firstly, a breakthrough in our comprehension of starch retrogradation properties from a reaction kinetics standpoint has been achieved through the development of a consecutive reaction kinetics model.^[10] This model effectively distinguishes the continuous nucleation and crystal growth steps during starch retrogradation, which can help gain insights into the kinetics and mechanisms of retrogradation. Secondly, the importance of starch fine molecular structures, especially amylose molecular structures, e.g., molecular size distribution and chain-length distribution on starch retrogradation property, has been recently explored.^[1] Numerous important findings have been obtained from these investigations. For example, the amount of amylose chains with a degree of polymerization (DP) of ~ 100–1500 was proven to be a significant regulator of rice starch retrogradation rate and extent, as well as the resulted microstructure of retrograded starch hydrogel.^[11] Finally, new aspects with respect to the factors which are important to determine intra- and intermolecular interactions, as well as short- and long-term retrogradation processes were learnt (e.g.,^[12–15]). These factors include common food compositions, e.g., sugars, salts, proteins, lipids, polyphenols, and non-starch polysaccharides.^[16]

This review was thus for the first time aiming to comprehensively summarize all these recent understandings of starch retrogradation property (Fig. 1), with a focus on the kinetics and molecular perspectives. This information is important in advancing our knowledge on the mechanisms of starch retrogradation. Considering the significance of starch retrogradation property in controlling both nutrition and texture property of starch-rich foods, this review supplies critical information that could be generally applied in the future to improve human health as well as the textural attributes of foods.

Figure 1. Retrogradation kinetics modeling and critical factors in determining starch retrogradation, including starch molecular structure, inter- or intramolecular interactions, short- or long-term retrogradation, as well as the addition of sugars, salts, proteins, lipids, polyphenols, and various non-starch polysaccharides. The figure is modified based on our previous publication with permission.^[17]

Characterization of starch retrogradation

Various techniques have been used to comprehensively investigate starch retrogradation process, including rheological, thermal, spectroscopic, X-ray diffraction, microscopic imaging, and mechanical methods.^[8] Differential scanning calorimetry (DSC) could track the endothermic energy alteration of retrograded starch during heating, which has been commonly applied to characterize starch retrogradation property. It could produce parameters including the onset (To), peak (Tp), conclusion (Tc) and enthalpy (ΔH) for the disruption of retrograded amylose and amylopectin double helices.^[8] These parameters are proven to be related to the ordering and amount of recrystallized double helices/entanglements over the starch retrogradation process.^[11] Generally, melting peaks around 50°C, 110°C, and 150°C are believed to be associated with the melting of amylopectin-amylopectin, amylose-lipid, and amylose-amylose helices developed via both intra- and intermolecular interactions, respectively.^[11,18] Gelatinization temperature range, i.e., Tc-To can indicate the heterogeneity (or homogeneity) of starch crystallites, that least stable ones melt with low temperature (To) whereas those remaining crystallites having higher-quality melt with higher temperatures (Tc).^[19,20]

Small amplitude oscillatory shear (SAOS) rheological measurement by rheometer is another common method applied to characterize starch retrogradation property.^[21] Generally, elastic modulus (G′) is determined by the quantities and strength of starch intermolecular bonds.^[21–23] Whereas viscous modulus (G″) is controlled by the energy consumption caused by the friction and movement of starch molecules over the rheological measurement.^[21–23] The advantage of SAOS test over DSC is that it can differentiate starch intra- and intermolecular interactions by, e.g., a temperature sweep test, as G′ and G″ values of retrograded starch hydrogel are related to the amount and ordering of starch intermolecular interactions.^[18] For instance, a temperature sweep (25°C – 85°C) is sufficient in terms of melting amylopectin intermolecular interactions, while amylose intermolecular interactions would be left.^[23] Therefore, the G′ or G″ drops over the temperature sweep could reflect the amount and ordering of amylopectin intermolecular interactions, and the G′ or G″ remained at 85°C are characteristics of the amount of amylose intermolecular interactions. The initial G′ or G″ values prior to the temperature sweep could reflect the overall amount of starch intermolecular interactions.

Rapid viscosity analyzer (RVA) measures the resistance (i.e., viscosity) of starch samples to a controlled shear rate during both heating and cooling phases, providing valuable information on starch retrogradation.^[24] Key parameters derived from RVA analysis include peak, trough, breakdown, final and setback viscosities. During heating, viscosity increases as starch granules swell, and peak viscosity is reached when granule swelling matches granule collapse. Trough and breakdown viscosities occur when granule collapse surpasses swelling, which could give information on the starch granule integrity. During the short-term cooling period, primarily amylose molecules re-associate, elevating viscosity from trough to the final viscosity. The difference between final and trough viscosity is referred as the setback viscosity. Therefore, the value of setback viscosity is positively correlated with starch short-term retrogradation rates.^[25]

Kinetics modeling

Retrogradation is a kinetic process, consisted of successive nucleation and crystal growth steps, which could be generally represented by the below scheme.^[26]

$\text{Gelatinized\hspace{0.17em}starch }\mathit{nucleation}\mathit{n}\text{Starch\hspace{0.17em}nuclei}\stackrel{\mathit{crystal}\mathit{growth}}{n}\text{Retrograded\hspace{0.17em}starch}$

Understanding the nature of each step is fundamental both theoretically and practically, as they decide the microscopic structure of retrograded starch as well as final physical properties of starch-based foods. Nucleation is generally defined as the first step in the formation of a new thermodynamic phase or structure via self-assembly. During starch retrogradation, it involves the stochastic formation of nuclei through homogeneous or heterogeneous nucleation during cooling from the equilibrium gelatinization temperature to a lower temperature,^[27] possibly due to strong electrostatic attraction force among adjacent starch molecules, as well as high tendency of forming hydrogen bonding among starch hydroxyl groups. Double helices are then formed from these nuclei (crystal growth), which are further aggregated into more ordered crystallites (maturation), reducing the overall free energies of the system.^[27]

Retrogradation kinetics can be described by the evolution of overall DSC melting enthalpy or elastic and viscous modulus (G′ and G″) against retrogradation time.^[2,28] Being able to reduce the retrogradation process into a few kinetics-based parameters is critical, as it enables a direct description of the retrogradation process as well as the structure-property correlation analysis. A consecutive reaction kinetics (CR) model has been recently developed to distinguish nucleation and crystal growth steps from the overall retrogradation kinetics process.^[10] The model can be generally described by equations shown in Table 1. Few important indications have been proposed from this study^[10]: (1) the amount of nuclei was much lower (~10⁸ times) than retrograded starch double helices; (2) a sigmoidal pattern was followed by the accumulation of nuclei during retrogradation, as its amount increased initially then dropped; (3) nucleation is the rate-limiting step, as its rate constant was significantly smaller than that for crystal growth.

Table 1. Kinetics models for understanding starch retrogradation property.

Kinetics model	Equation	Parameters	Advantages	Reference
Consecutive reaction kinetics model	For enthalpy associated with gelatinized starch: $\mathrm{\Delta }{\mathrm{H}}_{\mathrm{GS}}\left(\mathrm{t}\right)=\left(\mathrm{\Delta }{\mathrm{H}}_{\mathrm{RS}}\left({ }^{\mathrm{\infty }}\right)\right)\times \left(e^{-{\mathit{k}}_{\mathrm{n}}\mathit{t}}\right)$	$\mathrm{\Delta }{\mathrm{H}}_{\mathrm{GS}}\left(\mathrm{t}\right)$, $\mathrm{\Delta }{\mathrm{H}}_{\mathrm{SN}}\left(\mathrm{t}\right)$, and $\mathrm{\Delta }{\mathrm{H}}_{\mathrm{RS}}\left(\mathrm{t}\right)$ are DSC melting enthalpy associated with gelatinized starch, starch nuclei, as well as retrograded starch at time t; $\mathrm{\Delta }{\mathrm{H}}_{\mathrm{RS}}\left({ }^{\mathrm{\infty }}\right)$ is available melting enthalpy associated with retrograded starch after infinite retrogradation time; kn and kcg are rate constants for nucleation and crystal growth step.	Being able to differentiate the nucleation and crystal growth step.	^[Citation10]
	For enthalpy related to starch nuclei: $\mathrm{\Delta }{\mathrm{H}}_{\mathrm{SN}}\left(\mathrm{t}\right)=\frac{{\mathit{k}}_{\mathrm{n}}\times \mathrm{\Delta }{\mathrm{H}}_{\mathrm{RS}}\left({ }^{\mathrm{\infty }}\right)}{k_{\mathrm{cg}}-k_{\mathrm{n}}}\times \left(e^{-{\mathit{k}}_{\mathrm{n}}\mathit{t}}-e^{-{\mathit{k}}_{\mathrm{cg}}\mathit{t}}\right)$
	For enthalpy associated with retrograded starch: $\mathrm{\Delta }{\mathrm{H}}_{\mathrm{RS}}\left(\mathrm{t}\right)=\mathrm{\Delta }{\mathrm{H}}_{\mathrm{RS}}\left({ }^{\mathrm{\infty }}\right)\times \left(1-\frac{{\mathit{k}}_{\mathrm{cg}}\times e^{-{\mathit{k}}_{\mathrm{n}}\mathit{t}}-k_{\mathrm{n}}\times e^{-{\mathit{k}}_{\mathrm{cg}}\mathit{t}}}{k_{\mathrm{cg}}-k_{\mathrm{n}}}\right)$
Avrami model	$\mathrm{\Delta }\mathrm{H}\left(\mathrm{t}\right)=\left(\mathrm{\Delta }\mathrm{H}\left({ }^{\mathrm{\infty }}\right)\right)\times \left(1-e^{-{\mathit{k}}_{\mathrm{A}}{t}^n}\right)$	$\mathrm{\Delta }\mathrm{H}\left(\mathrm{t}\right)$ represents DSC melting enthalpy of retrograded starch from time t; $\mathrm{\Delta }\mathrm{H}\left({ }^{\mathrm{\infty }}\right)$ is the max melting enthalpy after infinite retrogradation time; kA is rate constant; n is Avrami exponent.	Being able to reflect crystal growth dimensions.	^[Citation29]
First-order kinetics model	First-order retrogradation kinetics equation: $\mathrm{\Delta }\mathrm{H}\left(\mathrm{t}\right)=\left(\mathrm{\Delta }\mathrm{H}\left({ }^{\mathrm{\infty }}\right)\right)\times \left(1-e^{-{\mathit{k}}_{\mathrm{FOK}}\mathit{t}}\right)$	$\mathrm{\Delta }\mathrm{H}\left(\mathrm{t}\right)$ is the DSC melting enthalpy of retrograded starch from time t; $\mathrm{\Delta }\mathrm{H}\left({ }^{\mathrm{\infty }}\right)$ represents maximum melting enthalpy after infinite retrogradation time; kFOK is rate constant.	Different retrogradation phases with distinct rate constants could be identified.	n.a.
	Transformed LOS equation: $ln\frac{\mathit{d}\mathrm{\Delta }\mathrm{H}\left(\mathrm{t}\right)}{\mathit{dt}}$=$ln\left(\mathrm{\Delta }\mathrm{H}\left({ }^{\mathrm{\infty }}\right)\right)-k_{\mathrm{FOK}}\mathit{t}$

LOS is the abbreviation for logarithm-of-slope. These DSC terms can be replaced by the elastic and viscous modulus (G′ and G″) if rheometer is applied for the investigation of starch retrogradation property.

Starch retrogradation process has also been characterized by Avrami equation (Table 1).^[3,29] As Avrami model has been developed and applied for many years in the literature, it will not be fully discussed in this review. The main difference between Avrami and the CR model is that Avrami model could produce a parameter named Avrami exponent (n), to reflect crystal growth dimensions.^[28] In addition, Avrami rate constant reflects the overall crystallization rate including both nucleation and crystal growth.^[28] Avrami exponent is determined by two factors: (1) types of crystal nucleation and (2) dimensions in which crystal growth occurs. When n = 4, it suggests sporadic nuclei and a spherulitic growth; n = 3 indicates either a disc-like growth from sporadic nuclei or spherulitic growth from instantaneous nuclei; n = 2 suggests either a rod-like growth from sporadic nuclei or disc-like growth from instantaneous nuclei; n = 1 indicates a rod-like growth from instantaneous nuclei.^[28] Starch retrogradation is generally shown having instantaneous nuclei and a rod-like growth, i.e., n = 1.^[30]

As starch retrogradation kinetics curve frequently obeys an exponential growth behavior, it suggests that starch retrogradation could obey the first-order kinetics (FOK) (Table 1).^[31] In this perspective, the transformed logarithm of slope (LOS) plot (Table 1), that is commonly used in fitting starch digestogram to differentiate the number of digestion components/phases (e.g., rapid vs slow),^[32] can also be potentially used to fit the retrogradation kinetics curves. The application of first-order kinetics and LOS plot has advantages over Avrami and CR model in that it could (1) result in rate constants representing the overall retrogradation rate that are comparable among different starch samples; (2) different retrogradation phases i.e., with distinct rate constants, if presented, are able to be identified by using the LOS plot.^[32] For example, although Avrami model was developed based on mechanistic assumptions, its resulted rate constants may not be able to represent the overall retrogradation rate, as the overall reaction rate from Avrami model is also dependent on the Avrami exponent.

However, first-order kinetics and LOS plot are rarely used to fit starch retrogradation kinetics. Therefore, we have applied the first-order kinetics and LOS plot (as described in Table 1) to fit the retrogradation kinetics of ΔH obtained from DSC analyses for 10 different rice starches (Fig. 2), with the original data obtained from the previous publication.^[33] The rice starches shared distinct molecular structures such as the amylose content and the amount of amylopectin trans-lamellar chains, resulting distinct retrogradation kinetics.^[33] Examples of fitting retrogradation kinetics curves of rice starch sample 7 by the LOS, Avrami, and CR model are given in Fig. 2b–d, respectively. The other fittings are given in Figs. S1–S2 in the supplementary information. Only the retrogradation kinetics of rice starches for 7 days were applied here for the modeling testing due to the data availability. Nevertheless, the approach developed here could be readily tested with the retrogradation kinetics of other starches or starch-rich foods with different storage time in the future. For example, a smaller retrogradation rate constant might be applied for the fitting of starch retrogradation kinetics with slower retrogradation rate for longer storage time, i.e., >7 days, compared to the data applied in Fig. 2. Although with different assumptions, these models can all satisfactorily reproduce the retrogradation kinetics curves, with R² > 0.999. Interestingly, all LOS plots (Figs. 2b and S1) showed a continuous linearity, suggesting that there was only one retrogradation phase with a consistent rate constant over the retrogradation process.^[32] However, the slope of LOS plots among different rice starches is significantly different (Fig. S1), indicating that different rice starches had distinct retrogradation rates. All these fitting parameters are given in Fig. 2e,f. As expected, all maximum melting enthalpy of retrograded starches ($\mathrm{\Delta }\mathrm{H}\left(\mathrm{\infty }\right)$) from different model fittings are quite similar (Fig. 2f). In terms of rate constants, CR nucleation rate constants (k_n) was included in Fig. 2e for the comparison with that from first-order kinetics and Avrami model, as nucleation was the retrogradation rate-limiting step, which could dominantly control the overall retrogradation rate. Fig. 2e shows that k_n (nucleation rate constant from CR model) and k_FOK (rate constant from first-order kinetics model) are consistent with each other, suggesting that k_n is useful to represent the overall retrogradation rate like k_FOK. However, the rate constants from Avrami model (k_A) are significantly deviated from both the k_n and k_FOK. It is reasonable as the overall retrogradation rate from the Avarami model is assumed to base on both the rate constant and crystal growth dimensions.^[28] It makes k_A from Avrami model a less useful parameter in order to compare the overall retrogradation rates among different starch samples.

Figure 2. Melting enthalpy kinetics curves for 10 rice starches (A), examples of fitting the rice starch sample 7 by LOS (B), Avrami (C), and CR (D) model, respectively. Comparison of fitting rate constants (E) and maximum melting enthalpy ΔH (F) from FOK, Avrami, and CR kinetics model for these 10 rice starches. Note, CR rate constants compared in graph E are the nucleation rate constants. The experimental data is collected from previous study.^[33]

Taken together, it shows that the first-order kinetics and LOS plot can satisfactorily reproduce the whole retrogradation kinetics curve and identify the number of retrogradation phases (i.e., slow versus fast) in the system. Both the CR model and first-order kinetics model can produce rate constants to represent the overall retrogradation rate. CR model could further distinguish the nucleation and crystal growth step. Although Avrami model can produce meaningful information on the crystal growth dimensions, its rate constant cannot be applied to compare the overall retrogradation rates among different starches. Therefore, these models should be applied together to produce complementary information on starch retrogradation property such as retrogradation phases, nucleation/crystal growth rate and crystal growth dimensions.

Structural basis for nucleation, crystal growth and maturation

It is generally believed that initial inter- and intramolecular entanglements between amylose-amylopectin or amylose-amylose molecules could act as nuclei and initiate the retrogradation process. Starch consists of two polymers, amylopectin and amylose. Amylose has a rapid retrogradation rate (e.g., in minutes) possibly due to its flexibility, which develops the backbone of starch hydrogel network. Amylopectin retrogrades with a significantly slower rate (in days) through both intra- and intermolecular interactions within the amylose network.^{[1,11,14,27,28]} Therefore, the short-term retrogradation of amylose molecules could provide crystal seeds for further amylopectin retrogradation.^[34] For example, it was proposed that potato starch had a faster retrogradation rate compared to waxy maize starch, as potato amylose can form an ordered matrix serving as the nuclei seeds for continuous amylopectin retrogradation.^[35] In addition, waxy potato amylopectin has a natural capability to develop a strong retrograded gel, due to its comparably long external and internal chains (although this capability is impeded by the charge repulsion at basic and neutral pH from the ionized phosphate monoesters in potato starches).^[21] Amylose structure was also shown as the critical factor in controlling the kinetics of rice amylopectin long-term retrogradation.^[10]

Starch molecular structures including molecular size distribution and chain-length distribution (CLD) are main driving factors in terms of determining nucleation and crystal growth rate, when starch supramolecular structures including crystalline, lamellar and granule structure are damaged over the gelatinization process. Among starch molecular structural parameters, amylose content (AC) is often a key factor in controlling starch physiochemical properties.^[1,36] By applying the newly developed consecutive reaction kinetics model, a recent study has shown a parabolic relation between AC with the nucleation rate and melting ΔH of long-term retrograded rice starch (Fig. 3a), with the lowest nucleation rate constant observed from rice starches having ~20% AC.^[33] It suggests that either a low or high AC could increase the nucleation rate and long-term amylopectin retrogradation extent. The detailed mechanism for this parabolic relationship between AC and nucleation rate is currently unclear, although two possible mechanisms were proposed. Starch with a higher amylopectin content would have a higher amount of nucleation sites (i.e., glucose molecules around the α-(1, 6) glycosidic branching points) (Fig. 3b,c), where double helices could be rapidly formed between neighboring chains. On the other hand, a faster short-term retrogradation rate is associated with starch having a higher AC, supplying more nuclei and bringing amylopectin molecules closer for a rapid retrogradation rate (Fig. 3b,c). Few other important findings were also obtained from the above study^[10]: (1) consecutive reaction kinetics parameters were mainly related to amylose CLDs, whereas less correlated with amylopectin CLD parameters; (2) a faster nucleation rate was not necessarily associated with a higher starch retrogradation extent; (3) shorter amylose short chains and a lower amount of amylose intermediate-long chains can increase both starch nucleation and crystal growth rates during long-term retrogradation, possibly by affecting the retrograded starch hydrogel structure (Fig. 3d). It has been further indicated that amylose with a larger hydrodynamic radius was able to promote the initial short-term rice starch nucleation rate and form a starch phase with a slower digestion rate, possibly because of its mobility as well as dense nucleation sites.^[37] Amylose molecules with relatively more short to intermediate chains can further promote the propagation rate of crystals based on these initially formed nuclei.^[37] It would eventually form a homogenous hydrogel with a dense matrix after extensive retrogradation period, which shows the digestion pattern having a single first-order kinetics component. On the other hand, a slow initial amylose retrogradation resulted in the development of two different starch fractions with a combination of parallel and sequential digestion pattern.^[11,37] This is consistent with earlier studies, which indicated that amylose chains with a DP about 1100 had the highest inclination to develop intermolecular interactions over retrogradation.^[38] Another study has shown evidence that shortened amylose molecules by sonication were readily leached out during gelatinization, which were then aligned into nuclei and accelerated the initial stage of retrogradation.^[30] Therefore, after retrogradation, sonicated starch had a much higher crystallinity and formed a high amount of resistant starch compared to non-sonicated starch.^[30]

Figure 3. Causal relationships between long-term rice starch retrogradation kinetics with parameters of starch CLDs for 10 various rice starches having a wide range of AC. Graph a shows the parabolic relationship between nucleation rate constant (k_n) and AC. Graphs were modified from previous publications.^[10,11]

The above studies were mainly focusing on homogenous nucleation, where no existing nuclei or second surface is present and starch nuclei formation occurs spontaneously. However, a second phase (e.g., a foreign crystal seed) may be added to interfere the starch retrogradation process (heterogenous nucleation), where nucleation can occur on cavities, surfaces, or cracks of insoluble impurities (crystal seeds) in the system. For example, addition of external crystal seeds can largely increase the starch retrogradation rate, although crystal seeds with different structure may have distinct effects.^[36,39,40] Adding amylopectin crystal seeds with the B-type crystallinity polymorph promoted the crystallinity of retrograded amylopectin as well as overall retrogradation rate, while crystal seeds with A-type crystallinity polymorph had no significant effects on amylopectin retrogradation.^[39] Specific reasons for the phenomenon are unknown, but it is speculated that only nuclei with a B-type crystallinity structure could further induce the crystal growth of gelatinized rice amylopectin molecules.^[39] In addition, addition of rice starch crystal seeds with B-type crystallinity polymorph improved the uniformity of rice starch crystal size and crystal perfection degree when gelatinized rice starch stored at 4°C, while those crystal seeds with A-type crystallinity polymorph had no significant effects on these properties.^[40] Addition of rice amylose crystal seeds improved the short-range order, long-range order, and retrogradation rate of retrograded rice starch.^[36]

Retrogradation rate is also determined by the short-range starch molecular order after gelatinization.^[41] For example, the long-range crystallinity, DSC enthalpy alterations and short-range molecular order for the retrograded starch increased initially and then decreased over a further decrease in the short-range molecular order of the gelatinized starch (determined by the intensity and area of Raman bands at 480 and 1080 cm⁻¹).^[42] It was further shown that the optimal amount of residual short-range molecular order from the gelatinized starch was determined by the water content and heating temperature. The reduced retrogradation rate at a higher loss of the short-range molecular order was possibly due to the reduced potential for nucleation of the well-separated starch molecules.

Short- and long-term retrogradation

There is currently no strict timeframe definition for the short-term and long-term retrogradation. Generally, short-term (happens within hours) and long-term (occurs in days) retrogradation are referred to amylose retrogradation and amylopectin retrogradation, respectively, according to their retrogradation rate.^[3,14] Short-term and long-term retrogradation can differently determine the nutritional and textural properties of starch-based foods. For instance, short-term amylose retrogradation is responsible for the initial reduced digestibility of retrograded starch, while longer storage contributes minimally to the decreased digestibility of retrograded starch possibly due to the slow retrogradation rate of amylopectin molecules.^[43,44] On the other hand, long-term amylopectin retrogradation is mainly responsible for the staling of bread and cake.^[45]

Starch fine molecular structures play critical roles in determining the short-term and long-term retrogradation. For example, longer amylopectin external and internal chains could increase the long-term starch retrogradation extent, with longer internal chain segments possibly allowing a more flexible amylopectin structure.^[18,46] However, a recent study has shown that intermediate amylopectin chains possibly participated in the short-term retrogradation, as relatively shorter intermediate amylopectin chains were positively correlated with a higher final viscosity during RVA test.^[47] Furthermore, more amylopectin short and amylose long chains (e.g., DP > 1500) are associated with a higher melting enthalpy at the temperature around 50°C for long-term retrograded starches.^[47] It suggests that both amylose and amylopectin molecules participated during the long-term retrogradation. Therefore, it seems that there is no strict separation of amylopectin and amylose molecules in terms of forming double helices or entanglements during the short- and long-term retrogradation. These terms (short- versus long-term) could only refer to the retrogradation process occurring at different time. Furthermore, evidence has shown that the heterogeneity (or homogeneity) of amylopectin long-term retrograded double helices was determined in a very short period (e.g., less than a day) and a further prolonged storage time only duplicated these double helices with the same heterogeneity.^[33]

Intermolecular or intramolecular interactions

Entanglements and double helices can form through both inter- and intramolecular interactions over starch retrogradation process.^[18,48] Different interaction patterns can frequently result in retrograded crystallites with distinct structures, which eventually determine physical properties of foods.^[21] For instance, more intermolecular interactions would result in a firmer starch hydrogel, while intramolecular interactions contribute minimally to the hardness of retrograded starch hydrogel.^[21] In addition, double helices formed through intramolecular interactions tend to inhibit activity of digestive enzymes by binding to these enzymes, while starch gel matrix formed through intermolecular interactions could form a physical barrier inhibiting its accessibility towards starch digestive enzymes.^[11,49] Therefore, manipulation of these interaction patterns among starch molecules over retrogradation might offer ways to develop novel foods with both preferable texture and nutrition properties.

As mentioned in the above section, starch molecular structures could largely determine starch retrogradation property, including the interaction pattern.^[11,20] Amylopectin molecules frequently form short-range double helices through intramolecular interactions, due to their short chain length and high steric hindrance of α-(1, 6) glycosidic linkages, which can be characterized as the melting peak around 50°C in DSC endo-thermogram.^[18,19] On the other hand, amylose molecules can generally develop long-range double helices via both intra- and intermolecular interactions, characterized by the DSC endothermic transition peak about 150°C.^[50] Lately, there has been evidence showing that longer amylopectin external chains or smaller amylopectin molecules also tend to develop both intra- and intermolecular interactions over the long-term retrogradation possibly due to their high mobility.^[18] However, in this study,^[18] it was not mentioned if these intermolecular interactions were between amylose and amylopectin, or between amylopectin and amylopectin. Similarly, waxy potato starch with elongated chains can form a significantly strong viscoelastic gel via intermolecular interactions, with an acidic condition during the retrogradation process, which can reduce the intermolecular electrostatic repulsion among phosphate monoesters.^[21] The degree of retrogradation of waxy rice and waxy corn starches increased with the reduced molecular weight.^[51] The internal chain segments of amylopectin molecules seem participated in the intermolecular interactions, because even small variations in the length of these segments from waxy starches resulted in distinct changes of the gel morphology after retrogradation.^[52] Typically, starches with longer internal chain length formed a harder gel with denser microstructure. By applying 17 starches from different plant sources, a strong correlation was found between the external chain length and the melting temperatures of retrograded starches.^[46] More importantly, the inter-block chain length (i.e., distance between tightly branched units) was positively correlated with the melting temperature and enthalpy of retrograded starches, possibly due to their flexible amylopectin structure.^[46] A recent study has further shown the significance of amylose in the development of intermolecular short-range double helices with amylopectin molecules over the retrogradation process.^[53] For example, more amylose chains could promote the development of more intermolecular short-range interactions (although less thermally stable), while inhibit the development of short-range intramolecular interactions.^[53]

Storage temperature is also an important factor in controlling starch retrogradation patterns. For example, storage temperatures near the glass transition temperature of gelatinized starch could promote starch nucleation rate, whereas temperatures around starch gelatinization temperature favor the starch crystal growth rate.^[50,54] Therefore, a temperature cycling between optimum temperatures for nucleation and crystal growth for set periods of time could often facilitate the development of more retrograded starch crystallites than isothermal conditions.^[50,54,55] A recent study has further shown that more homogenous intra- and intermolecular double helices (although less thermally stable) were formed when sago starch retrograded at −20°C compared to that stored under room temperature (RT) and 4°C, which finally resulted in a fracture-like microstructure (with compact and smooth surfaces) instead of sponge-like microstructure (composed of many thin slices of aggregates) (Fig. 4).^[56] In addition, gelatinized sago starch stored under −20°C and 4°C showed a significantly higher proportion of intermolecular double helices compared to that stored under RT. Different mechanisms could be involved for this finding, including (1) more nuclei were developed under −20°C compared to RT and 4°C,^[57] and (2) ice crystals formed under −20°C have sheared sago starch molecules into smaller ones, which can retrograde faster due to their flexibility.^[18]

Figure 4. Microstructure of sago starch after the retrogradation treatment at RT, −20°C and 4°C for 5 days under scanning electron microscopy (SEM). Bottom graphs have a higher magnification compared to upper graphs. These graphs are collected from the literature.^[56]

Influence of food ingredients on starch retrogradation

In real foods, starch is commonly present with other food compositions, including sugars, salts, proteins, lipids, polyphenols, and various non-starch polysaccharides. Interaction between starch and these food compositions during food processing could significantly affect starch retrogradation and final food properties,^[9] which is thus summarized in the following sections.

Water content

Water content is a critical factor in determining starch retrogradation, although its effects on starch retrogradation is complex and depends on many other factors, such as starch varieties.^[16] For example, a parabolic relationship was observed between the retrogradation enthalpy of wheat starch gels and water content from 20% to 80%.^[58] It suggests that an optimal water content is needed to reach the maximum starch retrogradation rate. Two possible mechanisms could be involved to explain the above-mentioned parabolic relationship.^[16] Firstly, excessive amount of water could cause a large distance among starch chains, limiting the possibility of forming intermolecular interactions during retrogradation. On the other hand, a limited water content may restrict the amylose leaching during gelatinization,^[59] subsequently inhibiting the formation of short-term crystal nucleus for crystal propagation and maturation. As mentioned above, starch molecular structure is another important factor, which could either synergistically or antagonistically determine starch retrogradation together with water content. For example, retrogradation occurred for normal corn starch at the water content of 80%, while there was no clear retrogradation for waxy starch at this water content.^[60] On the other hand, the retrogradation of waxy corn starch occurred when the water content dropped from 80% to 70%, although its enthalpy was lower compared to the normal corn starch retrograded at such water content.^[60] It suggests that the retrogradation of corn starch is promoted by the presence of amylose molecules.

Sugars

Sugars including both natural sugars and sweeteners are a common ingredient in many food products, which could largely alter starch retrogradation.^[9] The current widespread interests in retrogradation as a critical starch functionality in starch-rich foods, and hindrance for the reformulation in order to decrease added sugars, have inspired numerous efforts in investigating effects of sugars on starch retrogradation.

The influence of sugars on starch retrogradation varies depending on types of starch and sugars, amount of sugars as well as storage conditions (Table 2).^[61–64] For example, addition of fructose, glucose and maltose can all restrict starch retrogradation process when stored under 4°C, while starch retrogradation was increased after adding fructose and glucose when stored at −22°C.^[63,65] The mechanisms underneath the above-mentioned phenomena are not well understood yet, but are likely related to the changed water activity and steric hindrance effects brought by sugar addition.^[65] Disaccharides with distinct molecular structures can all retard starch retrogradation following the order of maltose > lactose > sucrose.^[64,66] In comparison with sucrose, maltose is more efficient in terms of preventing staling behaviors of glutinous rice products.^[15] Two main mechanisms were raised in literature for justifying effects of sugar addition on starch properties such as retrogradation, including (1) hydrogen bonding between starch and sugars^[63,67,68] and (2) change of water activity and solution viscosity due to the hydrophilicity of sugars.^[67] For example, polyols including mannitol, xylitol and isomalt could all inhibit the retrogradation of wheat starch, possibly via forming hydrogen bonding with wheat starch molecules and inhibit the mobility of starch molecules to form double helices during retrogradation.^[69] On the other hand, the enthalpy of retrograded wheat starch was negatively correlated with the solution viscosity of sucrose, allulose and oligosaccharides (such as fructo-oligosaccharide, galacto-oligosaccharide, isomalto-oligosaccharide, and xylo-oligosaccharide).^[70]

Table 2. Effects of food ingredients such as sugars, salts, proteins, lipids, polyphenols, and various non-starch polysaccharides on starch retrogradation as well as corresponding mechanisms.

Compositions	Influence on starch retrogradation	Mechanisms
Sugars	Influence varies according to the types of starch and sugars, amount of sugars and storage conditions	(1) Hydrogen bonding interactions between starch and sugars; (2) Change of water activity and solution viscosity
Salts	Influence varies with the type of starch and salt ions, salt concentrations and storage conditions	(1) Electrostatic interactions among hydroxyl groups of salts and starch; (2) Structure-breaking and structure-making effects of salts on water and starch molecules
Protein	Effects are related to protein compositions as well as storage conditions	(1) Changing the availability of water; (2) Affecting amylose leaching during gelatinization
Lipids	Addition of lipids can generally restrict the degree of starch intra- and intermolecular interactions	(1) Formation of amylose-lipid complex; (2) Affecting water mobility and starch gelatinization property
Polyphenols	Phenolic compounds generally retard starch retrogradation	(1) Hydrophobic interactions; (2) Electrostatic interactions; (3) Hydrogen bonds
Non-starch polysaccharides	The effects are frequently dependent on the chemical compositions of non-starch polysaccharides, gel preparation method, storage temperature and measurement methods	(1) Competing with starch for the free water; (2) Forming physical barriers and inhibiting starch gelatinization

Depending on the above-mentioned mechanisms, water, sugars and polyols might be viewed as a single effective solvent by their available sites for hydrogen bonding interactions with starch molecules.^[71] In a recent study, it has been proposed that starch gelatinization temperatures could be predicted by the volume density of available hydrogen bonding sites for interacting with starch molecules in solutions with polyols, amino acids, oligofructoses and sugars.^[72] This theory could also be applicable for the investigation of influence of adding sugars on starch retrogradation in the future.

Salts

The effects of adding salts on starch retrogradation could be very complex, as it varies with the type of starch and salt ions, salt concentrations and storage conditions (Table 2).^[16,73–75] Various mechanisms were raised in literature for explaining how salts affect starch retrogradation. Among these mechanisms, the most popular ones include (1) electrostatic interactions among hydroxyl groups of salts as well as starch; and (2) structure-breaking and structure-making influence of salts on water as well as starch molecules.^[73,74,76] In these theories, structure-breaking ions e.g., I⁻, SCN⁻ and Li⁺ can break hydrogen bonds among starch molecules, thereby inhibiting starch retrogradation. Whereas, structure-making ions, e.g., F, SO₄²⁻ and K⁺, could promote and protect hydrogen bonds among starch molecules, thereby facilitating the starch retrogradation.^[74,77] Furthermore, starch molecules have a electronegative nature.^[78] Cations tend to generally attract starch hydroxyl groups and can thus inhibit starch retrogradation. Whereas anions could repel starch groups and thus promote and stabilize starch retrogradation.^[73] Therefore, the influence of salts on starch retrogradation depend on a coordination of both electrostatic interactions and structure-making/breaking effects. For example, increasing the charge density of structure-breaking cations might promote their attraction to hydroxyl groups of starch molecules and retard starch retrogradation. On the other hand, increasing the charge density of structure-making anions could promote their repulsion to hydroxyl groups of starch molecules and increase starch retrogradation. The effects of structure-making cations or structure-breaking anions on starch retrogradation depend on both effects (conflicting with each other) mentioned above, which is thus more complex.

The influence of salts on starch retrogradation are also affected by the storage conditions. For example, NaCl is normally included in dough formulations, in order to improve the sensory properties, microstructure, and physical characteristics of finished cereal-based products.^[79] The addition of 5% NaCl inhibited starch retrogradation at 4°C possibly due to the repulsion between starch and Cl⁻, whereas it promoted starch retrogradation under −22°C possibly due to the losing of binding ability between water and starch molecules.^[65] The application and relevance of the above-mentioned finding to the food industry should be further investigated, as 5% NaCl is quite high to be applied in any starchy foods.

Protein, lipids, and polyphenols

Protein, lipids, and polyphenols are common ingredients in our daily diet. A better understanding of interactions among starch, protein, lipid, and polyphenols, as well as their influence on starch retrogradation have realistic applications to help develop foods with preferable properties.

Generally, the influence of protein on starch retrogradation are related to the protein compositions as well as storage conditions (Table 2).^[80] For instance, isolated wheat glutenin could retard wheat starch retrogradation because of its long chain, which can form a large amount of hydrogen bonds with water molecules and thus decrease the water activity.^[81] On the opposite, isolated wheat globulin, albumin, and gliadin can all promote wheat starch retrogradation.^[81] Corn starch retrogradation is retarded by soybean 7S globulin (which may have a distinct structure compared with the isolated wheat globulin), while promoted by soybean 11S globulin.^[82] However, soybean protein isolate had negligible effects on corn starch retrogradation.^[82] Additional investigations are needed to examine whether the structural differences between soybean 7S globulin, 11S globulin, and protein isolate contribute to their disparate effects on corn starch retrogradation. Moreover, further exploration is required to understand their contradictory effects in comparison to isolated wheat globulin on wheat starch retrogradation. The crystallinity and hardness of retrograded rice starch hydrogel was decreased over the increase of rice protein from 0% to 8%, indicating that rice protein can retard rice starch long-term retrogradation.^[83] This is possibly because of the high water-holding capacity of rice protein.^[83] The starch setback viscosity during RVA testing was decreased after the removal of total starch granule-associated proteins, indicating that these proteins could affect the short-term rice starch retrogradation process, possibly because of the ability of these granule-associated proteins in preserving the structural stability of rice starch.^[84] Addition of whey protein isolate accelerated the corn starch retrogradation, possibly due to its restriction on starch gelatinization process by interacting with starch through hydrogen bonds or limiting the water mobility.^[85] The short-term retrogradation of corn starch was significantly retarded by the addition of porcine plasma protein hydrolysates, which could competitively absorb water and decrease the starch mobility during retrogradation.^[86] Furthermore, these porcine plasma protein hydrolysates might interact with starch granules and inhibit amylose leaching during gelatinization, which can ultimately restrict the effectiveness of leached amylose molecules to develop a hydrogel network during retrogradation.^[87] Therefore, it seems the main mechanism how protein affects starch retrogradation is by changing the availability of water and affecting amylose leaching during the gelatinization and retrogradation process.

Lipids could interact with amylose into amylose-lipid V-type complex during retrogradation via the hydrophobic force (Table 2).^[8] It has also been shown that lipid can form complex with amylopectin outer chains.^[88,89] Therefore, it is generally believed that addition of lipids can restrict the degree of intra- and intermolecular starch interactions over the retrogradation process (Fig. 5a).^[90] For example, removal of endogenous lipids from adlay seed starch has increased the RVA setback viscosity (i.e. increased short-term retrogradation),^[80] while adding saturated fatty acid to rice starch has decreased the retrogradation endotherm.^[91] However, it is worth noting that only monoglycerides and free fatty acids could form amylose-lipid complexes, with the complexation ability depending on their molecular structures such as carbon chain length and saturation degree.^[92] For example, fatty acids with shorter carbon chains had higher complexation ability with amylose molecules, while longer chain fatty acids showed greater inhibition on starch retrogradation presumably because of the formation of more stable complexes with amylose molecules.^[8,89] Triglycerides are typically incapable of forming such complexes because of their large steric hindrance. A recent published review has discussed all these perspectives,^[88] so it is not further discussed in this review. In addition, lipids can affect starch retrogradation indirectly by affecting water mobility and starch gelatinization property (Fig. 5b),^[93] which might develop a partially gelatinized starch state in foods and distinct scenarios for continuing starch retrogradation process.^[94] This principle might also be applicable to explain the effects of other food compositions, e.g., protein, non-starch polysaccharides, on starch retrogradation.^[16]

Figure 5. Influence of fatty acids on starch retrogradation property through forming amylose-lipid complex during cooling (A) and indirectly limiting starch gelatinization (B). AM is amylose. AP is amylopectin. FA is fatty acids.

The effects of polyphenolic compounds including rutin, tea polyphenols, and polyphenols from pomegranate peel, mango peel, Chinese gall, and Chinese hawthorn on starch retrogradation have been investigated.^[95–98] It is generally concluded that the addition of phenolic compounds can retard starch retrogradation (Table 2). However, the effects of polyphenolic compounds on starch retrogradation depend on the differences of types and structures of these phenolic compounds.^[99–102] For example, the presence of isoferulic acid, ferulic acid, caffeic acid, and 4-hydroxybenzoic acid inhibited the long-term retrogradation of purple sweet potato starch, and the inhibition effects followed an order of 4-hydroxybenzoic acid > ferulic acid > caffeic acid > isoferulic acid.^[103] Compared to epigallocatechin and epicatechin, epigallocatechin gallate was more effective in restricting the retrogradation process of rice starch.^[104] The distinct effects of phenolic compounds on starch retrogradation might be driven by the various interaction types between starch and phenolic compounds, such as hydrophobic interactions, electrostatic interactions, and hydrogen bonds.^[105] The number of hydroxyl groups in these polyphenols was positively related to their effectiveness in retarding the potato starch retrogradation.^[61] Starch structure is another factor that could determine the effects of polyphenolic compounds on starch retrogradation. For instance, the inhibitory effects of oligomeric procyanidins on high-amylose maize starch was different from that for normal and waxy maize starches due to the variation on the molecular interactions between different types of starches and oligomeric procyanidins.^[100]

Non-starch polysaccharides

In order to improve starch physicochemical properties such as retrogradation, non-starch polysaccharides and their derivatives have been quite commonly incorporated in starch-rich foods.^[106,107] The effects on starch retrogradation frequently depend on chemical compositions of these non-starch polysaccharides, gel preparation method, storage temperature, and measurement methods, although in most cases the addition of non-starch polysaccharides can retard retrogradation (Table 2).^[8] For example, water-extractable arabinoxylan with a high molecular weight (M_w) can largely inhibit the long-term wheat retrogradation, while those with low M_w primarily retard short-term wheat retrogradation.^[108] RVA setback viscosity decreased first and then increased slightly with the increase of inulin concentration.^[109] The percentage of long-term wheat retrogradation decreased firstly while increased over the increasing of inulin concentration.^[109] Inulin types with DP ≤ 10 and 2–60 had stronger influence on starch retrogradation compared to that from inulin type with DP ≥ 23.^[109] Effects of chitosan, carrageenan, arabic gum, xanthan, and guar gum on both short-term and long-term retrogradation of lotus seed starch have been investigated.^[110,111] It was concluded that arabic gum can promote the short-term retrogradation possibly due to the release of free water, while carrageenan, guar gum, and xanthan could all retard short-term retrogradation by diminishing the release of free water.^[110,111] During long-term retrogradation, chitosan can induce water migration from starch and diminish starch-water interactions, while guar gum and xanthan can suppress water mobility and decrease the release of free water.^[110,111] Consequently, guar gum and xanthan inhibited the development of highly ordered starch structure during retrogradation.^[110,111] Although no general conclusion on the influence of non-starch polysaccharides on starch retrogradation has been reached, competing with starch for the free water is one of the main mechanisms.

The influence of intact cell wall on starch gelatinization has been investigated, which showed that starch gelatinization is significantly retarded by the presence of intact cell wall, possibly via limiting water ingress, restricting starch granule swelling, and impeding heat transfer.^[112] It could thus be implemented that starch retrogradation could be indirectly controlled by the intact cell wall through affecting starch gelatinization degree in foods. Because starch retrogradation rate could be largely different when starch is partially gelatinized compared to fully gelatinized.^[94] For example, starch molecules might be juxtaposed with each other and readily retrograde into double helices during cooling from its partially gelatinized state. Furthermore, rest amylopectin double helices in partially gelatinized starch can act as starting nuclei and promote the starch retrogradation rate.^[36,40]

Effects of retrogradation on nutritional and sensory properties of foods

As summarized above, starch retrogradation is a complex process and determined by many individual factors. A better understanding of the nature of starch retrogradation is crucial, as it can largely determine food quality such as nutritional and sensory properties (Table 3).

Table 3. Possible effects of retrogradation on starch digestibility and texture of foods.

Property	Results	Reference
Digestion	Increased retrogradation decreased both catalytic efficiency of amylase and total digestible starch	^[Citation49]
	Digestion degree of retrograded sago starch gel decreased rapidly after storage for 1 h, while little further drop observed afterwards	^[Citation27]
	No change of RS content over retrogradation	^[Citation44]
	RS content increased over retrogradation	^[Citation113]
	RS content increased with a temperature-cycling storage	^[Citation54,Citation55]
Sensory	An increased hardness associated with retrogradation	^[Citation3]

For example, starch digestibility is an important nutritional property of starch-based foods. A slower starch digestibility is frequently preferred to maintain a stable postprandial glycemic response.^[114] In this perspective, starch retrogradation is a desirable process, as retrogradation can generally decrease both the catalytic efficiency of amylase and total digestible starch.^[27,49] Many factors could individually or combinedly determine the effects of retrogradation on starch digestibility, among which storage time and temperature are crucial ones. During short-term storage (i.e., few minutes to few hours), amylose molecules are mainly involved in retrogradation and responsible for the initial drop of starch digestibility.^[27] Starch digestibility alters less significantly during the long-term storage (i.e., days), due to the slow rate of amylopectin retrogradation.^[33] However, the effects of storage time on the RS content remain less clear. For example, little effect was observed for RS content after retrogradation for different times,^[44] while another study found an increased RS content over the retrogradation period.^[113] Storage temperature is another critical factor in determining the effects of retrogradation on the starch digestibility. For example, a temperature-cycling between optimum temperatures for nucleation and crystal growth is often applied to augment the degree of retrogradation and reduce starch digestibility (Fig. 6).^[54,55,115]

Figure 6. The hydrolysis rate of waxy starch gels at a starch to water ratio of 1:2 (w/v), after retrogradation at an isothermal temperature of 4°C for 0 day and 14 days, as well as with the temperature-cycling of 4/25°C at 24 h time interval for 14 days. The figure is adopted from the literature with permission.^[115]

On the other hand, texture is one of the most important sensory attributes for starch-based foods, which as mentioned in the introduction section is also significantly affected by the retrogradation process. However, it is currently challenging for food industry to develop starch-based foods with both slow starch digestibility and desirable texture, as an increased hardness is frequently associated with the progression of retrogradation.^[3] A recent review has proposed some desirable starch fine molecular structures contributing to both preferable texture and slow starch digestibility in cooked rice, based on extensive correlation analysis among starch molecular structures, starch digestibility, and texture parameters.^[116] This strategy is worth further exploration.

Conclusions and future remarks

A significant progress has been recently achieved in terms of understanding the starch retrogradation process, which is an important factor in determining food properties. Consecutive reaction kinetics model has been developed to distinguish nucleation and crystal growth steps from the overall starch retrogradation process, which offers a powerful tool to better understand the structural basis for starch retrogradation. For example, an AC of ~20% could be selected to inhibit the nucleation rate, while larger amylose molecules to promote the nucleation rate of rice starch. In addition, external crystal seeds could be added to promote the overall retrogradation rate. Amylose and amylopectin with different chain lengths were involved in inter- versus intramolecular interactions, as well as over the short- versus long-term retrogradation time. For instance, smaller amylopectin molecules and longer amylopectin external chains prefer to form both inter- and intramolecular interactions during the long-term retrogradation process. Amylose is involved in the development of short-range intermolecular interactions. More homogenous double helices and a fracture-like microstructure are formed when sago starch stored at −20°C compared to those stored under RT and 4°C, which form a sponge-like microstructure. The heterogeneity of amylopectin long-term retrograded double helices is determined in a short period (e.g. < a day) and further prolonged storage only duplicates these double helices with the same heterogeneity. Different food ingredients have been explored in terms of their influence on starch retrogradation. Although there is no general conclusion, possible mechanisms were proposed for each case to explain the influence of various food compositions on starch retrogradation.

Despite a significant progress has been achieved in understanding starch retrogradation, there are still essential questions waiting to be further investigated. It is evident that starch molecular structure, sugars, salts, proteins, lipids, polyphenols, and various non-starch polysaccharides could all alter starch retrogradation. However, it remains elusive how these distinct factors together change starch retrogradation. Past investigations on starch retrogradation have mostly focused on one or two of these factors. It is thus highly required to comprehensively study the influence of all these factors on starch retrogradation together. Conceivably, consideration of all these different factors in determining the starch retrogradation is challenging. A possible way to solve this issue is to develop a machine-learning algorithm to integrate all these factors in order to determine their influence on starch retrogradation. For instance, a gradient boosting regression model depending on hundreds of decision trees can be developed to integrate starch molecular structure, sugars, salts, proteins, lipids, polyphenols, and various non-starch polysaccharides to predict their individual effects on the starch retrogradation property. The model could be trained firstly depending on a large variety of starch, sugars, salts, and so on in the discovery phase, followed by another combination of these factors to validate the prediction model. Achieving this is important for a comprehensive understanding of the starch retrogradation property. Furthermore, starch is frequently present in a complex food matrix with other ingredients such as the gluten network in bakery products. Although the importance of these physical structures in terms of determining starch retrogradation property has been investigated in the literature to some extent,^[12] there is still much more remaining to be further investigated considering the complex nature of food matrix. For instance, what is the optimal protein composition in wheat flour in terms of developing a bakery product with desirable starch retrogradation property? Is this optimal protein composition consistent among bakery products with different moisture contents (a well-known critical factor that can largely change the starch retrogradation)? What is currently the best available technique in terms of characterizing the effects of food physical structures on starch retrogradation? All these questions remain further investigations to better understand the starch retrogradation. Finally, novel food processing techniques such as extrusion-based 3D printing, supercritical fluid extrusion technology, two-stage or multi-stage extrusion, and electrospinning have emerged in the food industry. It would be desirable to investigate their influence on starch retrogradation to help develop food products with preferable starch retrogradation with these new techniques.

Acknowledgments

CL would like to thank the financial support from the Direct Grant for Research 2023/2024 from the Chinese University of Hong Kong (Project code: 4053622), CUHK-Improvement on Competitiveness in Hiring New Faculties Funding Scheme (PI: Ref. 310) and CUHK-School of Life Sciences Start-up funding.

Disclosure statement

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

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/87559129.2024.2347467

Additional information

Funding

This work was supported by the Direct Grant for Research 2023/2024 from the Chinese University of Hong Kong (Project code: 4053622), CUHK-Improvement on Competitiveness in Hiring New Faculties Funding Scheme [PI: Ref. 310] and CUHK-School of Life Sciences Start-up funding.