Effects of Sugars and Sugar Alcohols on the Gelatinization Temperatures of Wheat, Potato, and Corn Starches

The gelatinization temperature (Tgel) of starch increases in the presence of sweeteners due to sweetener-starch intermolecular interactions in the amorphous regions of starch. Different starch botanical sources contain different starch architectures, which may alter sweetener-starch interactions and the effects of sweeteners on Tgels. To document these effects, the Tgels of wheat, potato, waxy corn, dent corn, and 50% and 70% high amylose corn starches were determined in the presence of eleven different sweeteners and varying sweetener concentrations. Tgels of 2:1 sweetener solution:starch slurries were measured using differential scanning calorimetry. The extent of Tgel elevation was affected by both starch and sweetener type. Tgels of wheat and dent corn starches increased the most, while Tgels of high amylose corn starches were the least affected. Fructose increased Tgels the least, and isomalt and isomaltulose increased Tgels the most. Overall, starch Tgels increased more with increasing sweetener concentration, molar volume, molecular weight, and number of equatorial and exocyclic hydroxyl groups. Starches containing more short amylopectin chains, fewer amylopectin chains that span through multiple clusters, higher number of building blocks per cluster, and shorter inter-block chain lengths exhibited the largest Tgel increases in sweetener solutions, attributed to less stable crystalline regions.


Introduction
Starch granules are botanical energy storage vessels that differ in size (2-100 µm) and molecular architecture between botanical sources, although all native starches share the traits of being semicrystalline, water insoluble, and composed predominantly of two α-glucans: amylopectin and amylose [1][2][3]. Amylose is primarily a linear polysaccharide with a few branch points and a degree of polymerization (DP) ranging from 600 to 6000 [3]. Amylose resides in the amorphous regions of the starch granule, especially near the periphery, although some amylose may also be co-crystallized with amylopectin [2]. Amylopectin is a larger, branched molecule (5% branching) that has a DP of 9600 to >15,900 [3]. The regions near amylopectin branches are amorphous, while the unbranched regions of amylopectin form crystalline, double helical structures [4]. These alternating amorphous and crystalline lamellae have a repeating distance of~9 nm, regardless of the botanical source [5]. On a larger scale, starch granules also have alternating hard crystalline and softer semicrystalline growth rings (also known as shells) that are 100 to 400 nm thick [4]. Within these growth rings are blocklets that range in diameter from 20-500 nm, with smaller blocklets in the semicrystalline growth rings and larger blocklets in the crystalline growth rings. In general, A-type crystalline starches (e.g., cereals) have smaller blocklets than B-type crystalline starches (e.g., tubers) [4,6]. The amorphous lamella and spaces between blocklets result and sorbitol from Alfa Aesar (Ward Hill, MA, USA) ( Table 1). Sodium hydroxide (NaOH) was from Sigma-Aldrich (St. Louis, MO, USA), and hydrochloric acid (37%) (HCl) was from Acros Organics. The water used in this study was processed using reverse osmosis, then filtered by a Barnstead E-Pure Lab Water System (Dubuque, IA, USA) to >17.4 milliohm-cm.

Sweetener Solutions
The 6-carbon (6-C) sweeteners studied consisted of aldoses (glucose, mannose, galactose), a ketose (fructose), and an alditol (sorbitol). Solutions were prepared at 1, 2, 3, and 4 molar (M) concentrations for the 6-C sweeteners except for galactose, for which only 1 and 2 M solutions were used due to solubility limitations. The 12-carbon (12-C) sweeteners consisted of reducing sugars (maltose, isomaltulose), non-reducing sugars (trehalose, sucrose), and sugar alcohols (isomalt, maltitol), and solutions were prepared at 0.5, 1.0, and 1.5 M concentrations, with an additional 2.0 M sucrose solution. The sweetener solutions were prepared in 10 mL volumes in 15 mL centrifuge tubes. Water was added to the sweetener to~80% of the final volume, then the tube was placed in a heating block at 80 • C for several minutes followed by vortexing on a VWR Vortex Mixer (Lebanon, NJ, USA) and/or slower rotational mixing using a Scientific Industries Roto-Shake Genie (Bohemia, NY) until the sweetener fully dissolved. Upon cooling to near ambient temperature, water was added to reach a final 10 mL volume. Solutions were not used if crystals were visible after overnight storage at ambient conditions.

Gelatinization Temperature (T gel )
The gelatinization temperatures (T gel s) of the starch-sweetener slurries were measured using a method adapted from Allan, Rajwa and Mauer [29]. Approximately 250 mg of 1:2 w/w starch slurries were made by combining 1 part starch with precisely 2 parts solution in 1.5 mL centrifuge tubes, mixing with a stainless steel pin, and vortexing until the slurry appeared to be a homogenous mixture (no visible dry starch). The samples were then capped and stored overnight at ambient temperature (~22 • C). Before analysis, samples were vortexed to re-suspend the starch, and 15 to 20 mg of slurry were transferred into a Perkin Elmer 50 µL pan (BO143017) and hermetically sealed with a lid (BO143003). Due to analysis temperatures exceeding 115 • C and risk of pan failure, the HACS70 and HACS55 samples were loaded into high-pressure Perkin Elmer 50 µL pans (B016-9321) and hermetically sealed with high-pressure lids (B016-9321). Pans were manually transferred into a Perkin Elmer DSC 4000 (Waltham, MA, USA) that was calibrated using water, indium, and zinc. The potato, dent corn, and waxy corn samples were heated from 30 • C up to 100-115 • C at 10 • C/min, while HACS70 and HACS55 samples were heated from 30 • C up to 115-130 • C at 10 • C/min. The T gel of each slurry was measured in triplicate. The T gel was identified as the endothermic event occurring around 60 to 103 • C, and the onset temperature, peak temperature, and enthalpy (∆H) of starch gelatinization were determined using the "peak calculation" function with the "Standard" baseline in Pyris Software (version 10.1.0.0412). T gel peak analysis was performed using data that encompassed~5 to 10 • C before the peak onset to the post peak heat flow maxima (~2 to 5 • C after the peak end) ( Figure S1) while ensuring the peak analysis baseline did not cut through or go under any part of the thermogram. The onset of the T gel peak was calculated by the Pyris software as the intersection of the tangent of the baseline before the peak with the tangent of the inflection of the leading side of the peak ( Figure S1). The ∆H of gelatinization was the measured area of the peak, and the reported J/g was of the 1:2 w/w starch solution slurry (not adjusted to J/g of dry starch). The onset T gel of HACS70 was calculated using the "onset" function due to lack of a clear end of peak and/or only partial gelatinization occurring within the experimental parameters. This onset was determined by the intersection of two tangents, where the 1st tangent was the slope before gelatinization and the 2nd tangent was the slope between the onset and peak. Examples of DSC analysis for starch in water are shown in Figure S1. Since some samples lacked a clear endotherm peak, the onset T gel s were compared further in this study.  The effects of pH on the T gel of potato starch in 1 M glucose solutions were determined using a modified sample preparation method prior to DSC analysis. An initial 150 g sample of a 1:2 w/w starch slurry made with 1 M glucose was mixed in a 250 mL beaker, and then the pH was adjusted to 4, 5, 6, 7, 8, 9, and 10 using 1 M glucose solutions that had been pH-adjusted using 1 M NaOH or 1 M HCl. The pH was measured using an Orion pH probe (ThermoScientific, Waltham, MA, USA) calibrated from pH 4 to 10. The pH of the slurry was adjusted rather than the glucose solution to avoid any unaccounted pH buffering from the phosphates in the potato starch. The final starch-to-slurry w/w ratio was no longer precisely 1:2 w/w due to the addition of the acidified and alkalinized 1 M glucose solutions, but this was not a major concern since the T gel onset in an abundance of solution is not affected by minor starch:solution ratio changes [37]. The slurries were equilibrated overnight, and the T gel of each pH-controlled slurry was then measured in triplicate.

Data Analysis
The effects of sweeteners on the T gel of a starch were compared by grouping the T gel s of a starch in sweetener solutions with equal monosaccharide unit concentrations (e.g., 1 M monosaccharide and 0.5 M disaccharide solutions were grouped together), followed by one-way ANOVA and Tukey post-hoc tests (α = 0.05). The effects of a sweetener solution on the T gel s of different starches were investigated by comparing ∆T gel(i-0) s. The ∆T gel(i-0) was calculated as the temperature difference between the T gel of a starch in a sweetener solution to that in water. Significant differences of ∆T gel(i-0) s between starches in a sweetener solution were identified using one-way ANOVA and Tukey post-hoc tests (α = 0.05). The T gel s of potato starch in 1 M glucose solutions at pHs 4-10 were also compared using one-way ANOVA and Tukey post-hoc tests (α = 0.05). The associations of categorical sweetener solution properties to effects on T gel s were investigated using four-way ANOVA with all possible two-way interactions (α = 0.05). The categorical sweetener solution properties were: monosaccharide unit concentration (e.g., 1 M was 1 M mono-and 0.5 M disaccharide solutions), sweetener size (6-C or 12-C), sweetener type (sugar or sugar alcohol), and if the sweetener was a reducing sugar. The correlation of quantitative starch (Table 1) and sweetener solution properties (Table 2) to the T gel s were investigated using linear correlations and t-statistics. The tested quantitative starch properties were: percent crystallinity, amylose content (apparent and absolute), average amylopectin chain length, percentage of DP 6-12, percentage of DP 13-24, percentage of DP 25-36, percentage of DP ≥ 37, number of building blocks per cluster (NBbl), and inter-block chain lengths (IB-CL). The quantitative sweetener solution properties were: number of hydroxyl groups for intermolecular H-bonding (N OH,s ), number of equatorial and exocyclic hydroxyl groups, dry glass transition (T g ), sweetener molar volume, and capacity factor (K c ) ( Table 2). These starch and sweetener properties were correlated with the ∆T gel(i-0) s in sweetener solutions at 3 M monosaccharide unit concentrations (∆T gel(3M-0) ) and with the slope of the log 10 of T gel s in respect to molar monosaccharide unit concentrations (log T gel s). The significance of the Pearson correlation coefficient (R) was assessed using t-statistics: with n−2 degrees of freedom. The significance of crystalline starch type (A or B-type) was evaluated using a two-tailed t-test. The effective water contents (φ w ,eff ) of sweetener solutions were extracted from van der Sman and Mauer [30]. All ANOVA tests were performed using IBM SPSS Statistics v. 26.0.00 (Armonk, NY, USA), and correlations and t-statistics were calculated using Microsoft Excel 365 (Redmond, WA, USA).

Wheat Starch
The presence of a sweetener in solution increased the T gel of wheat starch, and increasing sweetener concentrations resulted in further increases in T gel (Figure 1). For example, from a starting point of 60.78 • C in water, the T gel of wheat starch increased by 3. 54, 11.50, 19.20, and 28.86 • C in 0.5, 1.0, 1.5, and 2.0 M sucrose solutions (Figures 1 and 2, Table S1). The type of sweetener also had a significant effect on the wheat starch T gel . In 3 M monosaccharide unit solutions, fructose and mannose increased the T gel the least, whereas isomaltulose followed by sucrose, isomalt, maltitol, and trehalose increased the T gel the most ( Figure 1 and Table S1). The 12-C sweeteners typically increased the wheat starch T gel more than 6-C sweeteners when compared at equal monosaccharide unit concentrations, except for sorbitol which increased the T gel more than maltose (Table 3 and Table S1). The gelatinization endotherm of wheat starch remained as a single unimodal peak in the presence of sweeteners ( Figure 2). Therefore, unlike what is found in low moisture conditions where additional thermal energy is needed to fully melt amylopectin, there was no peak separation between G (solution mediated melting of crystallites that is independent of moisture content) and M1 (melting of remaining crystallites that is dependent on moisture content) [37,44]-increasing sweetener concentrations simply increased the T gel endotherm to a higher temperature.
Sugar alcohols tended to increase the wheat starch T gel more than the sugar counterparts at the same molar concentrations (e.g., Figure 1 and Table S1, sorbitol vs. glucose), except between isomalt and isomaltulose. In Allan, Rajwa and Mauer [29], the wheat starch T gel s in xylitol solutions were also significantly higher than in xylose solutions and sugar alcohols were speculated to increase the wheat starch T gel more than sugars by potentially forming more H-bonds within the amorphous regions of native starch. Sugar alcohols may form more sweetener-starch interactions because they have an open structure [45], greater molar volume, and greater number of exocyclic hydroxyl groups (Table 2) than sugars.
The quantitative sweetener properties in Table 2 were also compared to the sweetener effects on wheat starch T gel in 3 M monosaccharide unit sweetener solutions (∆T gel(3M-0) ) and the slope of the log of T gel s with respect to the sweetener molar monosaccharide unit concentration (log T gel ). Significant correlations (p < 0.10) were found between T gel increases in sweetener solutions and both the number of equatorial and exocyclic hydroxyl groups on the sweetener and the sweetener molar volume, whereas the correlations of T gel with N OH,s , T g , and K c were not significant (Table 4). The correlation between wheat starch T gel and the number of equatorial and exocyclic hydroxyl groups suggests that sweetener stereochemistry influences the T gel increase. Similar findings were found for sweetener-water interactions, for which there was a strong correlation between the number of equatorial hydroxyl groups and dynamic hydration number of a sweetener [46]. The sweetener stereochemistry affects intermolecular H-bonding tendencies because exocyclic and equatorial hydroxyl groups are more reactive than the axial hydroxyl groups [47], and the strength and number of sweetener-starch interactions affect the T gel increase [29]. The sweetener molar volume was also significantly correlated to the effects on wheat starch T gel (Table 4), as were the sweetener solution viscosities (R ≥ 0.96, Table S2). Overall, 12-C sweeteners had larger solute radii and increased the T gel more than 6-C sweeteners, and sugar alcohols had larger radii and increased the T gel more than their sugar counterparts (Tables 2  and 4). Conceptually, sweeteners with larger solute radii can form longer H-bond bridges between chains in the amorphous regions of native starch and therefore have a greater stabilizing effect.
to the starch Tgel ( Figure S2). A low Kc was associated with greater hydrophobicity in a C-18 HPLC column, but the applicability of Kc for determining sweetener effects on starch Tgel was limited as it was not correlated to sweetener hydration numbers, NMR relaxation, terahertz spectroscopy, or viscosity [40]. Poor correlations between sweetener solution aw and Tgel increase have been found [29,30], and there is no chemical need for water during starch gelatinization since starch can gelatinize in pure glycerol [28].

Waxy, Dent, and High Amylose Corn Starches
As was found for wheat starch Tgel, increasing sweetener concentrations also significantly increased the Tgel of corn starches ( Figure 1), with differences found between sweetener types. For example, at 3 M monosaccharide unit concentration, isomalt increased the Tgel of corn starches the most and fructose increased the Tgel the least (Tables S3-S6). Significant differences were also found between corn starch types (Figures 3 and 4), with dent corn starch generally exhibiting the greatest Tgel increase (ΔTgel(i-0)) in the presence of sweeteners compared to the other corn starches. Exceptions were the ΔTgel(i-0)s of waxy corn in 3 and 4 M mannose solutions and HACS70 in 2 M sucrose solutions, which were greater than the ΔTgel(i-0)s of dent corn (Figures 3 and 4). The HACS70 Tgel in 2 M sucrose Several sweetener properties did not positively correlate with wheat starch T gel , including N OH,s , T g , and capacity factor (K c ) (TableS 2 and 4). In Allan, Rajwa and Mauer [29], the direct effect of sweetener T g on the T gel was doubted due to the presence of extreme outliers and evidence the T g of native starch is well below the T gel [7,48]. Since N OH,s was derived from the T g of the sweetener [49], it was therefore not surprising there was no correlation of T gel with either T g or N OH,s . However, the effective water content (ϕ W , eff ), which is directly related to volumetric density of intermolecular hydrogen bonds of the sweetener solution and calculated using N OH,s [30], was still highly correlated to the starch T gel ( Figure S2). A low K c was associated with greater hydrophobicity in a C-18 HPLC column, but the applicability of K c for determining sweetener effects on starch T gel was limited as it was not correlated to sweetener hydration numbers, NMR relaxation, terahertz spectroscopy, or viscosity [40]. Poor correlations between sweetener solution a w and T gel increase have been found [29,30], and there is no chemical need for water during starch gelatinization since starch can gelatinize in pure glycerol [28]. Table 3. Analyses of sweetener solution properties that influenced the T gel onsets of waxy corn, dent corn, HACS55, HACS70, potato, and wheat starch comparisons of p-values from four-way ANOVA are shown. Sweetener concentrations groupings were by monomeric molar concentrations (e.g., 1 M is 1 M mono-and 0.5 M disaccharide solutions), size groupings were 6-(e.g., glucose, sorbitol) and 12-carbon sweeteners (e.g., sucrose, isomalt); and reducing sugar groupings were reducing (e.g., glucose, isomaltulose) and nonreducing sweeteners (e.g., sucrose, sorbitol). Significant factors and interactions are shown in bold font (α = 0.05).  As was found for wheat starch T gel , increasing sweetener concentrations also significantly increased the T gel of corn starches (Figure 1), with differences found between sweetener types. For example, at 3 M monosaccharide unit concentration, isomalt increased the T gel of corn starches the most and fructose increased the T gel the least (Tables S3-S6). Significant differences were also found between corn starch types (Figures 3 and 4), with dent corn starch generally exhibiting the greatest T gel increase (∆T gel(i-0) ) in the presence of sweeteners compared to the other corn starches. Exceptions were the ∆T gel(i-0) s of waxy corn in 3 and 4 M mannose solutions and HACS70 in 2 M sucrose solutions, which were greater than the ∆T gel(i-0) s of dent corn (Figures 3 and 4). The HACS70 T gel in 2 M sucrose solution was exceptionally high, likely due to the unique shape of the thermogram (Figure 2). The substantial shift in the baseline for this sample may be the onset of amylopectin hydrate dehydration or a T g event. Foods 2020, 9, -0)) is plotted. Error bars are 1 standard deviation, n = 3, and the capital letters indicate significant differences between Tgels.   Similar to what was found for wheat starch T gel s, the categorical sweetener properties that were significantly correlated to the T gel s of all corn starches were sweetener concentration and size. A difference between corn and wheat starch behaviors was that the sweetener type (sugar or sugar alcohol) was significant for the T gel s of all corn starches but not for wheat starch T gel s (Table 3). Corn starch T gel s in 12-C sweetener solutions were generally greater than in 6-C sweetener solutions within the same monosaccharide unit concentration grouping; however, the T gel s in some 6-C sweetener solutions, such as sorbitol, were significantly greater than in some 12-C sweetener solutions, such as maltose (Tables S3-S6). The sweetener type (sugar or sugar alcohol) was significant for all corn starches (Table 3), with the corn starch T gel s in sugar alcohol solutions (sorbitol, isomalt, and maltitol) equal to or greater than in the counterpart sugar solutions (glucose, isomaltulose, and maltose, respectively) (Tables S3-S6). Baek, Yoo and Lim [50] also observed that corn starch T gel s in sugar alcohol solutions were greater than in sugar solutions.

Effects of Starch Properties
The quantitative sweetener properties that were significantly correlated to the T gel elevation of corn starches were the numbers of equatorial and exocyclic hydroxyl groups and molar volumes (p < 0.10), except not the molar volume with log T gel s of HACS55 (Table 4). Similar to what was found for wheat starch T gel , the N OH,s , T g , and K c of sweeteners were not significantly correlated to the sweetener effect on corn starch T gel , while the ϕ W , eff and solution viscosity were highly correlated to the corn starch T gel s ( Figures S3-S6 and Table S2). Sweeteners that formed more H-bonds with neighboring molecules and had larger solute radii increased the T gel s of corn starches to a greater extent.

Potato Starch
Similar to wheat and corn starches, increasing sweetener concentrations increased the T gel of potato starch; however, potato starch T gel responded differently to some sweeteners compared to the other starches ( Figure 1 and Table S7). In 3 M monosaccharide unit concentration solutions, the highest potato starch T gel was in a 1.5 M isomalt solution and the lowest was in a 1.5 M maltose solution (Table S7), which was different from the other starches for which the lowest T gel s were in 3 M fructose solutions (Tables S1-S5). Fewer differences were found in potato starch T gel s, compared to wheat and corn starches, with respect to the different effects of 6-C and 12-C sweeteners on T gel . The 6-C sweeteners tended to elevate the ∆T gel(i-0) of potato starch more than ∆T gel(i-0) s of other starches; however, the 12-C sweeteners did not elevate the ∆T gel(i-0) of potato starch as much as the ∆T gel(i-0) s of the wheat and corn starches. Thus, the T gel of potato starch was one of the most affected by 6-C sweeteners ( Figure 3) and least affected by 12-C sweeteners (Figure 4) compared to the T gel s of wheat and corn starches.
Unlike for other starch types, none of the numerical sweetener factors were significantly correlated to the potato starch T gel (Table 4); however, the number of equatorial and exocyclic hydroxyl groups and molar volumes of 6-C sweeteners were significantly correlated with log T gel s. When correlating log T gel s of potato starch with only 6-C sweetener properties (excluding 12-C sweetener properties), the correlation with the number of equatorial and exocyclic hydroxyl groups increased from R = 0.024 to R = 0.996 (p < 0.05), and the correlation with the molar volume increased from R = 0.340 to R = 0.945 (p < 0.10). The log T gel of potato starch in galactose solutions was excluded from these correlations due to low galactose solubility resulting in a limited number of data points (only 1 and 2 M). The significant correlations of potato starch log T gel with 6-C sweetener molar volumes and the number of equatorial and exocyclic OHs suggest the mechanism by which 6-C sweeteners increased the T gel of potato, wheat, and corn starches was similar. None of the 12-C sweetener properties were correlated with the log T gel s of potato starch, attributed to a potential size limiting effect that altered the sweetener-starch interactions of 12-C sweeteners in the potato starch granule (discussed in detail later).

Effects of Starch Properties
When comparing the effects of sweeteners on the T gels of starches from different botanical sources (Figures 3 and 4), significant differences in the ∆T gel(i-0) between starches were found. Differences were also found in the significant sweetener factors associated with the T gel s (Tables 3 and 4) of these starches. Therefore, sweeteners affected the T gel s of starches from different botanical sources differently, and thus it was of interest to explore how these differences were influenced by starch composition, architecture, and morphology.

Amylose Content
The starches evaluated had varying amylose compositions ranging from ≈0% amylose in waxy corn starch up to 68% apparent amylose in HACS70 (Table 1). Within the corn starch sources, the effects of sweeteners on the ∆T gel(i-0) s were not consistent with the amylose contents. For example, the dent corn ∆T gel(i-0) s were greater than the waxy corn ∆T gel(i-0) s, except in mannose solutions; however, the ∆T gel(i-0) s of HACS55 and HACS70, which contained even more amylose, were also less than dent corn (Figures 3 and 4). When evaluating all starch sources, there were no significant correlations (p < 0.10) between the ∆T gel(3M-0) or log T gel s of a sweetener and the amylose (apparent and absolute) contents of any of the starches (Table S8).
When comparing only waxy and dent corn starches, the presence of amylose increased the ∆T gel(i-0) in the presence of sweeteners. The interspersed amylose in the amorphous regions of native starch may have allowed for a greater stabilization effect from sweetener H-bond bridges. However, this trend did not persist with further increases in amylose content, perhaps because additional starch structural changes were present. The high amylose corn starches have much longer amylopectin branch lengths, greater IB-CLs, and are in the B-type crystalline form; while waxy and dent corn have more comparable amylopectin structures (Table 1).
Other studies have found varying relationships between T gel and amylose content: some studies suggest starches with greater amylose contents have higher T gel s [51,52], others report no effect on T gel [53][54][55][56], and yet others report lower T gel s for starches with higher amylose contents [17]. It has also been proposed that the type of amylose affects the T gel behavior: lipid complexed amylose increases the T gel [19]. Salts have been reported to affect the T gel and ∆H of gelatinization of waxy and normal corn starches in a similar manner [57], thus amylose content made little difference. Therefore, the lack of correlations between the T gel increase in sweetener solutions and the amylose contents of starches (Table S8) suggest amylose contents did not greatly affect the sweetener-starch interactions that increase the T gel .

Amylopectin Architecture
The starches in this study had varying amylopectin architectures (Table 1). HACS55 and HACS70 had longer amylopectin chain lengths, smaller percentages of short DP 6-12 amylopectin chains, higher percentages of DP 25-36 and DP ≥37 amylopectin chains, lower NBbls, and higher IB-CLs (Table 1), and the T gel s of these starches were least affected by sweeteners (Figures 3 and 4). The T gel s of wheat and dent corn starches were more affected by sweeteners (Figures 3 and 4), and these starches had shorter amylopectin chain lengths, higher ratios of DP 6-12 chains, lower percentages of DP 25-36 and DP ≥37 amylopectin chains, higher numbers of NBbls, and lower IB-Cls. When comparing all starches in this study, the percentages of DP 25-36 and DP ≥37 chains and the average amylopectin chain lengths of starches were negatively correlated with T gel elevations in different sweeter solutions (Table 5). In contrast, the T gel elevations of starches in sweetener solutions were positively correlated with higher percentages of DP 6-12 amylopectin chains (Table 5). Therefore, the T gel s of starches with higher percentages of short amylopectin chains were more affected by sweeteners than the T gel s of starches with longer amylopectin chains. The T gel increases for starches in several sweetener solutions were also positively correlated with the NBbl of starches and negatively correlated with IB-CLs of starches ( Table 6), suggesting that amylopectin fine structure also influenced the sweetener-starch interactions. The architecture of amylopectin has been associated with the thermal properties of starch, attributed to the stability of the starch crystallites [14,15,19,20,55,[58][59][60]. Starches with higher ratios of short amylopectin chains tend to have lower T gel s because these short chains are too short to crystallize and also act as crystal defects [15,19,20]. Therefore, the starches with higher percentages of DP 6-12 amylopectin chains (e.g., dent corn and wheat starches were 17.9 and 19.0% DP 6-12, respectively) have less stable starch crystallites, and sweeteners presumably formed sweetener-starch interactions that enhanced the crystallite stability and resulted in a greater T gel elevation. Starches with smaller ratios of DP 6-12 amylopectin branches (e.g., HACS55 and HACS70 were 9.7 and 8.5% DP 6-12, respectively) had T gel s that were less affected by sweeteners because there were fewer crystal defects. The negative linear correlations between T gel increases in sweetener solutions and average amylopectin chain lengths, percentages of DP 25-36, and percentages of DP ≥ 37 (Table 5) suggest the T gel s of starches with longer amylopectin branches are less affected by sweetener solutions. The DP 6-12, 13-24, 25-36, and ≥37 amylopectin chains fractions have been defined as A (fa), B1 (fb 1 ), B2 (fb 2 ), and B3 (fb 3 ) chains and span through 1, 1, 2, and 3 crystalline amylopectin clusters, respectively [61,62]. The B2 (DP 25-36) and B3 (DP ≥37) amylopectin branches span through multiple clusters and are long enough to make relatively defect-free crystallites. Crystallites that have fewer crystal defects and are covalently linked to one another are already fairly stable, thus sweeteners in solution have a lower potential for elevating the T gel because there are fewer possibilities for stabilizing sweetener-starch intermolecular interactions. There were also no significant correlations between T gel increases in sweetener solutions and percentages of B1 (DP 13-24) chains, which could be due to the intermediate length between stabilizing and destabilizing amylopectin chain lengths.
The fine structure of amylopectin also affects the T gel , with lower NBbl and higher IB-CL values associated with higher T gel s [11]. There were significant correlations with NBbl and IB-CL of starches and the effects of sweeteners increasing the T gel s ( Table 6). With fewer building blocks in a cluster (NBbl) and greater spacing between building blocks (IB-CL) allowing for more molecular flexibility, the double helices in the crystalline amylopectin clusters are able to form more hydrogen bonds and a stronger crystal structure [11]. Therefore, native starches with more blocks per cluster and shorter spacing between building blocks have less stable crystallites, and thus sweetener-starch interactions have a greater stabilizing effect (Table 5). For example, the T gel s of wheat and dent corn starches increased the most in sweetener solutions (Figures 3 and 4) and these starches had the greatest number of building blocks per cluster (NBbl) and shorter IB-CLs (Table 1). In contrast, the T gel s of high amylose corn starches had the smallest T gel increases in sweetener solutions (Figures 3 and 4) and had the highest IB-CLs and lower NBbl values. Again, the T gel s of starches with more stable crystals were less affected by sweeteners.
Also of note, sweeteners that increased the T gel the most (isomalt, maltitol, and isomaltulose ( Figure 1)) tended to have the strongest correlations between T gel increase and amylopectin lengths, NBb1, and IB-CL (Tables 5 and 6). It was postulated this occurred because these sweeteners had the greatest stabilizing effect due to the larger molar volumes and higher numbers of equatorial and exocyclic hydroxyl groups (Table 2), and starches with the greatest potential for sweetener-starch stabilizing interactions had T gel s that were more affected by these sweeteners.
In summary, amylopectin structure affected the extent to which sweeteners increased the T gel . The T gel s of starches that had less stable crystallites, due to shorter amylopectin chain lengths and sterically hindered double helical clusters, were more affected in sweetener solutions than starches with more stable crystalline regions. Since sweetener-starch interactions drive the T gel increase, starches that have more regions to be stabilized (e.g., wheat and dent corn starches as shown in Tables 5 and 6) had greater T gel increases in the presence of sweetener solutions.

Amylopectin Crystalline Structure
Starches with A-type and B-type amylopectin crystalline structures were used in this study: waxy corn, dent corn, and wheat starch were the A-type polymorphic form while HACS55, HACS70, and potato starches were the B-type polymorphic form ( Table 1). The T gel s of A-type starches increased more in sweetener solutions than the T gel s of B-type starches (Table S9) based on comparing the averages of log T gel s and the ∆T gel(3M-0) s in A and B-type starches. However, the ∆T gel(i-0) s of dent corn (A-type), wheat (A-type), and potato in 6-C solutions (B-type) starches were the most affected by sweeteners, while the ∆T gel(i-0) s of waxy (A-type), potato (B-type) in 12-C solutions, and high amylose corn starches (B-type) were the least affected (Figures 3 and 4). Thus, there did not appear to be a consistent trend in ∆T gel(i-0) response to sweeteners in solution based solely on amylopectin crystal polymorph. A confounding factor was likely that starches with longer amylopectin chains form B-type crystal structures (Table 1) and the T gel s of starches with longer amylopectin branches were less affected by sweeteners (e.g., high amylose corn starches in Figures 3 and 4). In theory, the crystalline form should not affect the influence of sweeteners on T gel , because sweeteners increase the T gel through intermolecular interactions in the amorphous regions of starch [29].

Percent Crystallinity
No correlation was found between the percent crystallinity of starches and the effects of sweeteners on the T gel increase (Table S8). For example, the high amylose corn and wheat starches were all 20% crystalline, yet the T gel of wheat starch was more influenced by sweeteners than the T gels of the high amylose corn starches (Figures 3 and 4). Even though the sweetener-starch interactions that stabilize the granule and increase the T gel occur in the amorphous regions [29], perhaps the amount or distribution of the amorphous regions were not limiting factors within this experimental space, since the native starches in this study were all >50% amorphous.

Potato Starch-The Anomaly
There were similar sweetener effects on ∆T gel(i-0) trends between starches from different botanical sources (Figures 3 and 4) with few exceptions other than for potato starch. Potato starch was also the only starch that had a T gel not affected by the sweetener size (Table 2), since potato starch T gel s in 6-C sweetener solutions were not significantly different from T gel s in 12-C sweetener solutions. It has been reported that 12-C sweeteners (disaccharides) increase the T gel of starch more than 6-C sweeteners (monosaccharides) for other starch botanical sources [22,29,50]. Therefore, the unique aspects of potato starch were explored to elucidate why the sweetener effects on the T gel elevation behavior of potato starch were different than for other starches.
(1) Size Exclusion within the Granule A speculation as to why 12-C sweeteners were not as effective at raising the T gel of potato starch compared to elevating the T gel s of starches from other botanical sources was due to some sort of size exclusion, wherein the diffusion of 12-C sweeteners throughout the potato starch granule was affected differently than diffusion in other starch types. The molecular size exclusion could be due to: (1) the spacing of amylopectin branch points, and/or (2) the spacing between blocklets in the crystalline growth rings. In potato starch, the amylopectin branch length from the backbone to the crystalline double helix is~3 glucosyl units long, much shorter than for waxy rice, tapioca, and wheat starches that have branch lengths of 6, 7, and 7 glucosyl units long, respectively [63]. This short distance between the backbone to the crystalline helix in potato starch creates a smaller region for sweetener-starch interactions in the amorphous lamella, which potentially creates steric hindrance or spatial challenges for a DP 2 sweetener in a space that is~DP 3 in size. This amylopectin branch length between the backbone and double helix is not to be confused with amylopectin lamella spacing, which is consistently~9 nm between starch sources [5]. Another possible size exclusion point is the space between the crystalline blocklets in the crystalline growth rings. The blocklets in potato starch are much larger (200-500 nm) than the blocklets in wheat starch (80-120 nm), which reduces the porosity of potato starch [6,64]. This lower porosity in the crystalline growth ring may restrict the diffusion of 12-C sweeteners and thus affect the localized concentration of sweetener solutions in the amorphous regions of potato starch.
(2) Phosphorous Content Another unique aspect of potato starch is that it contains phosphate monoesters. The phosphate monoesters cause potato starch to be affected by electrolytes [65] and have a lower T gel compared to other starches with similar amylopectin structures, because the crystallites are destabilized when the phosphate groups repel from each other during heating [64,66]. The phosphate groups are predominantly (74-78%) in the amorphous regions [67]; however, phosphate content has been both positively correlated [67] and not correlated [20,56] to the potato starch T gel . It is important to note that the positive correlation of phosphate content and T gel could have been confounded with the effects of longer amylopectin chains [67], which also elevate T gel . Since phosphates are ionizable groups, it was assumed the pK a s of potato starch were similar to the pK a s of phosphoric acid: there would be no pK a3 at 12.67 due to the ester linkage with starch, pK a1 at 2.21, and pK a2 at 7.21 (pK a values from Weast [68]). A similar assumption was made by Marsh and Waight [69], but the pK a s of potato starch have also been assumed to be similar to those of glucose 6-phosphate at pK a1 ≈ 0.94 and pK a2 ≈ 6.11 [70,71]. Regardless, the T gel s of potato starch in 1 M glucose solutions with pHs ranging from 4 to 10 were the same, at~65 • C ( Figure 5). The lack of T gel changes at pHs above and below pK a2 in the presence of a sweetener suggests that the phosphate monoesters do not influence the sweetener-starch interactions to an extent that would affect the T gel . The phosphate monoesters may not affect sweetener-starch interactions because there is a relatively low concentration of phosphate groups: one per~317 glucosyl units [72]. Another explanation could be that the phosphate monoesters are primarily located on long B chains that are greater than DP 20 [72], and DP > 18 amylopectin branches are the stabilizing portions in potato starch since they yield more stable crystals and higher T gel s [20]. The T gel of yam starch, which has about one-seventh of the amount of phosphate monoesters compared to potato starch [73], was also unaffected by changes in pH [74].
Therefore, the most likely reason potato starch T gel s exhibited different trends in the presence of 12-C sweeteners compared to T gel s of starches from other botanical sources is that the sweetener diffusion throughout the amorphous regions of the potato starch granule was restricted more than in other starch types, thereby limiting the amount of 12-C sweetener-starch interactions in native potato starch. Foods 2020, 9, x FOR PEER REVIEW 21 of 25

Conclusions
The effects of 6-C and 12-C sweeteners on the Tgels of six different starches were investigated. The Tgels of all starches increased in sweetener solutions compared to Tgels in water, with greater Tgel increases as sweetener concentration increased; however, the magnitude of the increase in Tgel varied between different types of sweeteners and starches. The starch Tgel was increased more by sweeteners that had larger molar volumes and higher numbers of equatorial and exocyclic hydroxyl groups (e.g., sugar alcohols and 12-C sweeteners). Sweeteners with these properties formed more sweetenerstarch interactions, stabilizing the amorphous starch regions, and thereby increased the Tgel to a greater extent. Sweetener traits that were not associated with increases in starch Tgel were: dry Tg, Kc, NOH,s, and whether or not the sweetener was a reducing sugar. The amylopectin structure of starches also affected the Tgel increase in sweetener solutions. The Tgels of starches with higher percentages of A-chains (DP 6-12) increased the most in sweetener solutions, whereas the Tgels of starches with higher percentages of B2 (DP 25-36) and B3 (DP ≥37) amylopectin chains increased less in sweetener solutions. The amylopectin fine structure also influenced the effects of sweetener solutions on the Tgel, where starches with fewer building blocks per cluster and longer inter-block chain lengths formed more stable crystals and the Tgels were less affected. A-type starches were more affected by sweeteners than B-type starches, but this may be confounded with B-type starches having longer amylopectin chains. Starch traits that were not associated with Tgel increases in the presence of sweeteners included: amylose content and percent crystallinity. The effects of sweeteners on the Tgel elevation of potato starch were unique, because the potato starch Tgel was one of the most affected by 6-C sweeteners and the least affected by 12-C sweeteners. This was attributed to a size exclusion phenomenon that altered 12-C sweetener diffusion into the potato starch granules and was not due to ionic effects of the phosphate monoesters. In summary, starches with less stable crystalline regions are more susceptible to sweetener-starch stabilizing effects and exhibited greater increases in Tgel in the presence of sweeteners. Native starches with structures that limit sweetener-starch interactions, such as the more stable crystalline regions in high-amylose corn starches or the unique structures in potato starch that limit diffusion of 12-C sweeteners, exhibit smaller increases in Tgel. However, sweeteners with molecular conformations that were favorable for intermolecular interactions increased the Tgel the most, regardless of starch architecture. This study has shown the different Tgel elevation effects of sweeteners on multiple starch botanical sources and provided insights into structural reasons for these differences, information that is useful for improving the understanding

Conclusions
The effects of 6-C and 12-C sweeteners on the T gel s of six different starches were investigated. The T gel s of all starches increased in sweetener solutions compared to T gel s in water, with greater T gel increases as sweetener concentration increased; however, the magnitude of the increase in T gel varied between different types of sweeteners and starches. The starch T gel was increased more by sweeteners that had larger molar volumes and higher numbers of equatorial and exocyclic hydroxyl groups (e.g., sugar alcohols and 12-C sweeteners). Sweeteners with these properties formed more sweetener-starch interactions, stabilizing the amorphous starch regions, and thereby increased the T gel to a greater extent. Sweetener traits that were not associated with increases in starch T gel were: dry T g , K c , N OH,s , and whether or not the sweetener was a reducing sugar. The amylopectin structure of starches also affected the T gel increase in sweetener solutions. The T gel s of starches with higher percentages of A-chains (DP 6-12) increased the most in sweetener solutions, whereas the T gel s of starches with higher percentages of B2 (DP 25-36) and B3 (DP ≥37) amylopectin chains increased less in sweetener solutions. The amylopectin fine structure also influenced the effects of sweetener solutions on the T gel , where starches with fewer building blocks per cluster and longer inter-block chain lengths formed more stable crystals and the T gel s were less affected. A-type starches were more affected by sweeteners than B-type starches, but this may be confounded with B-type starches having longer amylopectin chains. Starch traits that were not associated with T gel increases in the presence of sweeteners included: amylose content and percent crystallinity. The effects of sweeteners on the T gel elevation of potato starch were unique, because the potato starch T gel was one of the most affected by 6-C sweeteners and the least affected by 12-C sweeteners. This was attributed to a size exclusion phenomenon that altered 12-C sweetener diffusion into the potato starch granules and was not due to ionic effects of the phosphate monoesters. In summary, starches with less stable crystalline regions are more susceptible to sweetener-starch stabilizing effects and exhibited greater increases in T gel in the presence of sweeteners. Native starches with structures that limit sweetener-starch interactions, such as the more stable crystalline regions in high-amylose corn starches or the unique structures in potato starch that limit diffusion of 12-C sweeteners, exhibit smaller increases in T gel . However, sweeteners with molecular conformations that were favorable for intermolecular interactions increased the T gel the most, regardless of starch architecture. This study has shown the different T gel elevation effects of sweeteners on multiple starch botanical sources and provided insights into structural reasons for these differences, information that is useful for improving the understanding of structure-function relationships and behaviors of starches in formulations containing different sweeteners.