Research Article Distinct properties and structures among B-crystalline starch granules† Varatharajan Vamadevan1,4, Andreas Blennow2, Alain Buléon3, Avi Goldstein1,4, Eric Bertoft1,5 1 Department of Food Science and Nutrition, University of Minnesota, St Paul, MN, U.S.A. 2 Department of Plant and Environmental Sciences, University of Copenhagen, Frederiksberg C, Denmark. 3 UR1268 Biopolymères Interactions Assemblages, INRA, Nantes, France. 4 Present address: Cargill, Minneapolis, MN, U.S.A. 5 Present address: Bertoft Solutions, Turku, Finland. Correspondence: Dr Eric Bertoft, Bertoft Solutions, Gamla Sampasvägen 18, 20960 Turku, Finland. E-mail: email@example.com † This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: [10.1002/star.201700240]. This article is protected by copyright. All rights reserved. Received: September 14, 2017 / Revised: October 16, 2017 / Accepted: October 17, 2017 Abstract Starch granules derived from certain tuber or root crops exhibit a B-type polymorphic pattern and amylopectin with a high content of long B-chains and comparatively long segments between the building blocks. In this investigation, four B-crystalline starches were selected to study their morphology and molecular composition including lintnerization conducted at two different temperatures. The structure of the granules suggested that the B-type crystalline starches could be divided into two distinct groups, with potato and edible canna forming one group having large granules with typical "growth rings" and the molecular structure of the lintners being dependent on the temperature of lintnerization. The other group consisted of shoti and lesser yam starches possessing granules with alternating "granular slices" instead of rings or shells and the molecular composition of the lintners was not dependent on the temperature. We found that the former group of starch possessed lower gelatinization temperatures and swelled at lower temperatures than the latter group, suggesting a more labile granular structure of potato and canna starches compared to shoti and lesser yam. As the properties were not related to the amylose or phosphate content of the granules, the result suggests that B-crystalline starch granules are found as at least two distinct structural types with unique architectures. Keywords: B-type crystalline starch / Granule structure / Thermal properties / Starch composition / Lintnerization 1 Introduction The main components of starch, amylose (essentially linear) and amylopectin (extensively branched), are packed into semi-crystalline granules. In the granules, semi-crystalline shells appear as rings, often called "growth rings", when observed in light microscope and they are embedded in a softer, amorphous matrix . The amylopectin component contributes to the formation of stacks of thin, amorphous and crystalline lamellae in the semi-crystalline rings . The internal chains of amylopectin build up the amorphous lamellae together with most of the branched glucosyl units , which essentially connects to short external chains with a length (CL) of 10–15 glucosyl units through flexible spacer arms . The external chains unite into double-helices that crystallize into either A- or B-type crystalline polymorph patterns [5, 6], of which the latter is found in several starches from tubers, roots and rhizomes. A-type polymorphs are generally found in cereal starches and some starches (e.g. from pulses) exhibit C-polymorphic pattern that is a mixture of Aand B-type unit cells . The internal unit chain profile of amylopectin, as analysed by highperformance anion-exchange chromatography (HPAEC), has been found to have distinct patterns in plants from different genetic backgrounds . The patterns were classified into four types, of which type 4 amylopectin structure appears to confine to most B-crystalline starches. Type 4 is characterized by a high amount of long chains (CL > 38) compared to amylopectins representing type 1, 2, or 3. Clusters, defined as groups of chains with < 9 glucosyl units between branches , are small in type 4 amylopectins [10, 11]. The clusters are essentially composed of groups of very small, densely branched units known as building blocks  and, generally, these units are interconnected by interblock segments with a length (IB-CL) of 5–8 glucose residues. Type 4 amylopectins possess the longest IB-CL (7–8), whereas IB-CL in the other structural types are shorter . In an earlier investigation , we analysed the thermal properties of a range of starches representing all four types of amylopectin structures. Interestingly, we found a strong, positive correlation between the gelatinization onset temperature (To) and IB-CL. This was remarkable as the samples were randomly selected and represented a wide selection of genotypes and tissues. This suggested that starch granules share a common and universal principle of their architechture, in which IBCL of the amylopectin component plays a key role: The longer the inter-block segment, the more flexible it becomes and the better it adapts to stabilise the parallel arrangement of the double-helices in the crystalline lamella , as many of the interblock segments make up the connection of the amorphous and crystalline lamellae . This much resembles the model of starch by Waigh et al. , who described the behaviour of amylopectin as a side-chain liquid crystal. As we found that the correlation of To and IB-CL included all four types of amylopectin internal structures, it also included the different crystalline polymorphs of the granules. However, only two B-type crystalline starches were included in the investigation (edible canna and lesser yam), of which edible canna starch possessed a considerably lower gelatinization temperature than expected based on its molecular structure . From our and others previous works [15-19], we noted that potato starch granules have a low To comparable to that of edible canna. The question arose, therefore, whether potato and canna starch granules, with their apparently "abnormal" gelatinization behaviour, represent an unusual architechture giving rise to properties different from what would be expected based on their molecular structure, which is similar to that of lesser yam [8, 11, 12]. This would be of special interest as potato starch, in particular, has been considered as a model substance for the typical B-crystalline structure. In this work we therefore reinvestigated the thermal properties of all three B-crystalline starches (potato, edible canna and lesser yam) and included one additional B-type, namely shoti starch. This starch, deposited in rhizomes of turmeric, Curcuma zedoaria, is also known under the names of zedoary, white turmeric, and kentjur (Wikipedia, 2017), and has been shown to contain a very high phosphate content – even higher than potato . We also analyzed the crystalline polymorphic pattern and the swelling capacity of the granules. Furthermore, their morphology and molecular composition was analysed prior to and after treatment in dilute hydrochloric acid (lintnerization). We found that these B-type crystalline starches essentially can be divided into two distinct groups based on their structural and thermal properties. 2 Experimental 2.1 Materials Starch was extracted from potato (Solanum tuberosum, variety Dianella) as described previously . Shoti (Curcuma zedoaria) starch was kindly provided by Dr K. Yamamoto, Japan, and edible canna (Canna edulis) and lesser yam (Dioscorea esculenta) starches were generous gifts from Dr K. Sriroth, Thailand. Isoamylase (EC 18.104.22.168) from Pseudomonas sp., 1000 U/mL, and pullulanase (EC 22.214.171.124) from Klebsiella planticola, 700 U/mL, were purchased from Megazyme International (Bray, Wicklow, Ireland). All other chemicals and solvents were of ACS certified grade. 2.2 Microscopy Native and acid hydrolysed starch suspensions were observed under bright field light and cross polarized light using a binocular microscope (Nikon Microscope, Eclipse 80i) equipped with real time viewing (Q-capture Pro™). A QImaging digital camera (QICAM fast 1394, Canada) was used for the image capture. 2.3 Wide angle X-ray scattering (WAXS) X-ray diffractograms of starch samples were obtained with a Rigaku Ultima IV X-ray diffractometer. The native starches were kept in a desiccator over saturated K2SO4 solutions (aw 0.97) at room temperature for 14 days prior to X-ray diffraction measurements. The hydrated native starches were then packed tightly into a round aluminium holder and WAXS patterns were recorded with operating conditions of target voltage 40 kV, current 44 mA, scanning range 3–30°, scan speed 1.00°/min, step time 0.95, divergence slit width 0.5°, scatter slit width 0.5°, sampling width 0.03°, and receiving slit width 0.3 mm. Jade software (Material Data Inc. CA, USA) was employed for the calculation of the relative crystallinity (RC). 2.4 Differential scanning calorimetry (DSC) TA Instruments, Q2000 differential scanning calorimeter (DSC) equipped with a thermal analysis data station and data recording software (TA Instruments, Universal Analysis 2000) was used to measure the gelatinization parameters of the B-type starches. Starch dispersions in excess water (starch/water ratio = 1:3) were equilibrated for 3 h at room temperature before DSC analysis. The scanning temperature range and the heating rates were 10 °C to 110 °C and 5 °C/min, respectively, and the thermogram was recorded with deionized water as a reference. The transition temperatures reported are the onset (To), peak (Tm), and conclusion (Tc) temperatures. The enthalpy of gelatinization (ΔH) was estimated by integrating the area between the thermogram and a base line under the peak and was expressed as J/g of dry starch. 2.5 Swelling factor The swelling factor of the native B-type crystalline starches was determined according to the method of Tester and Morrison . Starch (50 mg, db) dispersions in deionized water (5 mL) were heated (55–95 °C) in screw cap tubes in a shaking water bath for 30 min. The sample tubes were then cooled rapidly to 20 °C in an ice water bath and 0.5 mL of Blue Dextran (Pharmacia MW 2×106, 5 mg/mL) was added and the contents were mixed gently by inverting tubes. The tubes were centrifuged at 1500 × g for 10 min and the absorbance of the supernatant obtained from samples and the reference tube that contained no starch was measured at 620 nm. The swelling factor was reported as the ratio of the volume of the swollen starch granules to the volume of the dry starch. 2.6 Lintnerization Lintnerization was conducted following the method of Robin et al. . Starch granules were mixed into 2.2 N HCl (25 mg/mL) at 25 or 45 °C and the suspesion was gently stirred daily. Aliquots (15 µL) were diluted, centrifuged (15 min, 1800 × g), and the supernatant was analysed for total carbohydrate content with the phenolsulfuric acid reagent  to follow the hydrolysis of the granules. At 80% hydrolysis, the residual starch granules were centrifuged, washed twice with 0.1 M NaOAc to neutralize the solution and then washed five additional times with water. The final starch residues (lintners) were lyophilized. 2.7 Unit chain profile of amylopectin and lintners Lyophilized amylopectin, previously separated from amylose in butanol-isoamyl alcohol mixture essentially as described by Klucinec and Thompson  but with minor modifications [11, 26], or lyophilized lintners were dissolved in 90% DMSO (50 mg/mL) by stirring on a boiling water bath. The samples were then diluted with 0.01 M NaOAc buffer, pH 5.5, to a final concentration of 5 mg/mL. An aliquote (0.5 mL) was treated with isoamylase and pullulanase (1 µL each) overnight at room temperature (~22 °C). The debranched samples were then analysed by highperformance anion-exchange chromatography coupled to pulsed amperometric detection (HPAEC-PAD) as decribed previously . Untreated lintners were also analysed by HPAEC using the same gradient as for debranched samples. The results are given as average of duplicate analyses. 2.8 Amylose composition The composition of amylose was analysed by debranching whole starch as described for the debranching of amylopectin. The samples were then analysed on a column of Sepharose CL 6B . Long chains of amylose were defined as chains eluting in the void volume and short chains of amylose were chains eluting between the long chains and chains of amylopectin. All results are average values of duplicate analyses. 2.9 Composition of phosphate esters Glucose 3-phosphate and glucose 6-phosphate were analysed with HPAEC following the method of Blennow et al. . 3 Results 3.1 Granular characteristics Microscopy images in polarized light of the four starches analyzed in this investigation showed that all starches possessed birefringence with the well-known "Maltese cross" extending from a distal hilum (Fig. 1). The birefringence pattern is an effect of the directional organisation of the double-helices in the crystalline lamellae towards the granule surface . The granules of canna starch were large with a long axis up to 100 µm, which was about the same size as the potato starch granules in accordance with previous reports [29-31]. Shoti granules were smaller with approximate lengths ≤ 45 µm. Leonel  found a similar size distribution between 15–45 µm, and Jane el al.  showed that the granules of shoti are flat with a thickness of 8–12 µm at the center of the lenticularly shaped granules and only about 1–3 µm at the edges. The sizes of the granules of lesser yam were comparable with shoti granules, but they possessed a different morphology. Many of these granules had shapes resembling three-sided pyramidals. The "Maltese cross" extended from the top of the pyramidals. Some granules had clearly irregular "Maltese crosses", some even with zig-zag patterns (examples are highlighted by arrows in Fig. 1). All four starch granules possessed the typical B-type crystalline pattern [33, 34] when analysed by WAXS (Fig. 2). However, minor differences in the peak patterns were apparent. Thus, the typical doublet at 22–24° 2θ was less pronounced in shoti and lesser yam starch compared to potato and canna starch granules. The same was true for the doublets at 10.0–11.5° and 14.5–15.0° 2θ. Furthermore, the peak at 19–20° 2θ was weaker in shoti and lesser yam starch. The relative crystallinity of the starches demonstrated a crystallinity ranging 26–32% (Table 1), which agreed with previous data on potato, canna and lesser yam [30, 31, 35]. The thermal properties of the starch granules were analysed by DSC (Table 1). Potato and canna starch granules possessed similar thermal characteristics. Thus, the onset gelatinization temperature (To) occurred at 57.5 and 60.4 °C for potato and canna, respectively. The peak gelatinization temperature (Tp) was 61.5 and 65.7 °C, respectively, and the gelatinization range (Tc–To) was 10.5 and 10.8 °C, respectively. Generally, these values were similar to those found elsewhere using similar conditions for the gelatinization [35-37]. Shoti and lesser yam were similar to each other, but possessed considerably higher To (74.5–74.8 °C) and Tp (77.1–77.5 °C) than potato and canna starch. In addition, Tc–To was much narrower (6.5–6.6 °C). In contrast, all four starches possessed rather similar gelatinization enthaphy values (15.1–16.5 J/g). Swelling of the starch granules over the temperature range 55–95 °C illustrated that potato starch granules swelled most extensively of all samples (Fig. 3). Already at 55 °C, the swelling factor (defined as swollen volume/initial volume of dried starch ) approached 20. Maximum swelling was achieved at 85 °C, after which the granules collapsed. A high swelling capacity at comparatively low temperatures compared to other starches is a well known property of potato starch granules [38-40]. Canna granules swelled considerably less than potato, but the temperature range of the swelling was the same, i.e., swelling started at 55 °C and increased up to 85 °C, after which the granules disintegrated. At the maximum temperature the swelling factor of canna granules was 47 as compared to 78 for potato granules. Shoti and lesser yam starch granules were more stable than potato and canna; they only started to swell at a temperature above 65 °C. The maximum temperature was 95 °C, at which the swelling factor of shoti and lesser yam granules was 37 and 44, respectively. 3.2 Molecular composition The covalently linked phosphate ester content of the starch granules was analysed by HPAEC and is presented in Table 2. Potato starch granules contained a total of 21.2 nmol phosphate/mg starch, of which 15.2 nmol/mg represented glucose 6-phosphate (G6P) and 6.0 nmol/mg glucose 3-phosphate (G3P), giving a ratio of G6P/G3P of 2.5. Gomand et al.  found that the content of G6P varied between 9.2–20.4 nmol/mg in different potato varieties. Shoti starch granules contained about three times more phosphate than potato and almost all surplus was in the form of G6P. This result was in accordance with a previous report . Edible canna and lesser yam starch contained considerably less phosphate than potato starch, 13.7 and 14.7 nmol/mg, respectively, and these values were also of the same order of magnitude as found in the literature [41-43]. In canna starch, the ratio of G6P/G3P was similar to that in potato, whereas in lesser yam it was only 1.2, mostly due to a relatively large presence of G3P (Table 2). The composition of polysaccharide components in the starch granules was investigated by analysing the unit chain composition of debranched starch by GPC on Sepharose CL 6B. With this technique the long chains typical of amylose elute in front of the shorter chains of amylopectin origin (Fig. 4). The chromatograms suggested fairly similar amylose levels in all samples between 18.6–24.8% (Table 3). Amylose levels in potato starch varies greatly with the cultivar [37, 40] and the concentration obtained here is well within reported values. The amylose content of canna was also in the order of that described by others (23–28%)[30, 41]. However, Ren  recently reported a lower content of amylose (19.23%) in edible canna starch. Amylose content in shoti starch was previously found to be between 28 and 41.8% [44-47], i.e. higher than for the sample analysed here (24.3%). On the other hand, Leonel et al.  reported 20.53% amylose in shoti starch. The amylose content of starches from different varieties of yam (D. esculenta) was found to be 20~24% by Jayakody et al. , whereas Srichuwong et al.  reported the amylose content in lesser yam to be 14.2%, as compared to 18.6% in this report. However, it is doubtful if the samples from different varieties of lesser yam can be compared as the size of the granules was much smaller in the lesser yams analysed previously [30, 48]. In addition, the morphology was different and in the sample analysed by Srichuwong et al.  the crystalline polymorph was of the C-type (mixture of A- and B-types) rather than B-type found in this work. The structure of amylose can be partly characterized from the proportion of long and short chains (LCAM and SCAM, respectively) as obtained by analysing the debranched starch by gel-permeation chromatography (GPC) on Sepharose CL 6B and defined in Fig. 4. Potato amylose contained the largest amount of LCAM (15.4%), but the smallest amount of SCAM (7.1%), thus giving the highest ratio of LCAM:SCAM of 2.2 (Table 3). Canna also possessed much LCAM, but intermediate amounts of SCAM resulting in a ratio of 1.3. Shoti and lesser yam amylose was different. They contained a lower amount of LCAM than SCAM and, thus, the ratio was below one in these starches. The unit chain profile of amylopectin was analysed by HPAEC (Fig. 5). Potato and canna amylopectin possessed very similar chain profiles (Fig. 5a) with only very small differences in the quantity of short (DP 6–38) and long chains (DP ≥ 39). On the other hand, the unit chain profiles of shoti and lesser yam amylopectin were different from potato and canna but were similar between theselves (Fig. 5b). Difference plots between shoti and potato and between lesser yam and potato, respectively (Fig. 5c), demonstrate that shoti amylopectin contained less of the shortest chains with DP between 6–15, and similarly yam had less of chains with DP 6–13. Of these short chains, those with DP 6–8 have been named "fingerprint" Afpchains, because of their characteristic profile in different plants . The rest of the short chains with DP 9–14 is here depicted as "S1". The remaining part of the short chains, depicted "S2", were found in larger amounts than the corresponding chains in potato. Difference plot for potato and canna showed very minor difference (data not shown). The relative number of the chain categories is given in Table 4. Afp-chains were present in about double amount in potato and canna compared to shoti and yam amylopectins. S1-chains represented 31.0 and 32.4% of the chains in canna and potato, respectively, whereas only 25.1 and 27.6% in shoti and lesser yam. As S2chains were more abundant in shoti and yam, the overall content of short chains was rather similar in the four amylopectin samples. The largest difference was found between potato and shoti amylopectin with 88.8 and 85.1% short chains, respectively, and instead, therefore, shoti amylopectin had the largest amount of long chains. The average chain lengths of the short (SCL) and long chains (LCL) are also given in Table 4. As the lengths of both chain types were longer in shoti and lesser yam, the average chain length (CL) for the entire amylopectin in these two starches was longer (24.8 and 23.0, respectively) than in potato and canna (21.4 and 21.9, respectively). 3.3 Lintnerization Lintnerization of the starch granules was performed at two temperatures, 25 and 45 °C (Fig. 6). It is well known that the solubilisation rate is very sensitive to the temperature [49, 50]. Thus, at 25 °C the solubilisation took days rather than hours as at 45 °C. The inital rate of solubilisation at 25 °C up to roughly 30% was similar among the samples, after which the solubilisation rate became diverse for the different samples (Fig. 6a). Potato starch granules reached about 80% solubilisation after 53 days, lesser yam reached this level after more than 80 days. The time to reach the same solubilisation level at 45 °C was only 25 h for potato and canna starch. In contrast, lesser yam reached 80% solubilisation after about 74 hours. At this time shoti still had not reached the same level and the solubilisation continued slowly until 80% solubilisation finally was achieved after 145 h (not shown in Fig. 6b). Srichuwong et al.  linterized starch at 35 °C, at which potato starch was hydrolyzed somewhat faster than edible canna, i.e. in line with what our result suggests. The size-distributions of the dextrins in the residual granules (lintners) at 80% solubilisation are shown in Fig 7. The well established fact that the composition of the dextrins in lintners of potato starch depends on the temperature at which the lintnerization is performed [15, 50, 52] was confirmed in this work. Similarily, the composition of the the lintners of canna starch granules depended on the temperature. At 45 °C the lintner possessed considerably more of large dextrins with DP ≥ 30 than at 25 °C, whereas the amount of smaller dextrins with a peak approximately at DP 15–16 predominated at 25 °C. Interestingly, the dextrin composition in the lintners of shoti and lesser yam was not notably affected by the temperature, as in both starches the small dextrins predominated regardless the temperature (Fig. 7). The average DP of the lintners was approximately 20–21 in all samples, except in potato and canna lintners made at 45 °C, in which DP was 26.6 and 27.0, respectively (Table 5). The lintners were also debranched (not shown) and the apparent chain length (CL) of the dextrins in potato and canna lintners was higher at 45 °C than at 25 °C (Table 5). A slight increase of the average number of chains (NC = DP/CL) at the higher temperature was also found, especially in canna. In contrast, CL and NC were not affected by the temperature in shoti and lesser yam lintners. The CL of shoti was higher (~18) than in lesser yam (~15). However, it should be noted that the CL values (and therefore also NC) are only apparent, as the debranching of the dextrins with pullulanase and isoamylase is not complete, due to the fact that certain branches in the lintners are resistant to the enzymes [50, 52]. Albeit the phosphate levels generally were lower in the lintners (Table 5) than in the original starch granules (Table 2), suggesting that the phosphate groups to a major part was found in the amorphous layers, in agreement with previous findings [15, 20], details in the phosphate contents were different and suggested rather complex distributions. Potato and canna lintners retained more of the phosphate when lintnerized at 45 °C compared to 25 °C (Table 5). In fact, in the canna lintner the phosphate concentration was even slightly higher (14.7 nmol/mg) than in the original starch (13.7 nmol/mg), due to a high remaining concentration of G6P, which resulted in a high ratio of G6P/G3P. In shoti starch the decrease in phosphate concentration after lintnerization was for the most part due to the removal of G6P, whereas the remaining concentration of G3P even increased, which rendered the ratio of G6P/G3P to decrease from 8.3 (Table 2) to 1.9 and 2.8 at 25 and 45 °C, respectively. In contrast, the decrease of phosphate in the yam lintner was more moderate and did not affect the distribution of G6P and G3P as the ratio largely remained unchanged. The morphology of the residual starch granules was largely unchanged after the lintnerization (Fig. 8), despite the fact that only 20% of the original starch remained in these granules. The granules also possessed the "Maltese cross" in polarised light, which was in agreement with a previous report for potato lintners  and as summarized by Kainuma and French . This showed that the orientation of short chains in the crystalline lamellae remained intact in all starches. The "growth rings" of potato and canna granules were clearly distinguished after lintnerization (Fig. 8a and 8b). However, in shoti and, especially, in lesser yam, these layers appeared to resemble "slices" rather than "rings" (Figs. 8c-f). In fact, lesser yam appeared to contain two distinct types of granules, of which one type possessed the granular "slices" being partly separated from each other at this stage of lintnerization (Figs. 8e and 8f), whereas the other type did not have "slices" but were fragmenting into smaller pieces. The former type of granules were associated with the smooth "Maltese crosses", whereas the latter granules had the irregular crosses and appeared more dense with sharper birefringence in the images (Fig. 8d). Whether shoti possessed granular "slices" or "rings" remained somewhat unclear, but some of the layers appeared to reach the edge of the granules, suggesting more flat structure that possibly were a kind of "slices" (Fig. 8c). All granules were very fragile at this stage of solubilisation and fragmented extensively after freeze-drying. 4 Discussion In this investigation we followed up our earlier finding that potato and canna starch granules are exceptions from the general trend that the gelatinization onset temperature (To) of any kind of normal or waxy starch granules correlates positively with the inter-block chain length (IB-CL) in the amylopectin component . We now extended the analyses of thermal properties with shoti starch granules and found that this starch possessed very similar gelatinization temperature (To, Tm and Tc) as lesser yam (Table 1). The temperature range for the melting (Tc–To) was also similar to yam. This indicated that potato and canna formed one pair and shoti and lesser yam formed another pair with respect to their gelatinization properties, which was unexpected as they all had amylopectin of type 4 molecular structure and, thus, similar IB-CL values between 7–8 [9, 11], suggesting no correlation with the gelatinization temperature. This was interesting because, like potato starch, shoti starch contains very high content of phosphate, whereas both yam and canna has much lower, and between themselves similar, phosphate levels (Table 2). The influence of phosphate on gelatinization of starch granules remains unclear, as contradicting results have been reported [15, 17, 37, 54-56]. Moreover, the distribution of G3P and G6P differed in the lintners in an unpredicted way in the different starches, i.e. was not correlated to any thermal effects. More data is required to discern such variation. Hence, our data suggest that phosphate content had less influence on the thermal properties of starch granules than other molecular structures. However, it should here be noted that once gelatinised, the phosphoester groups clearly affects starch properties due to a higher degree of hydration, which results in more viscous and clearer solutions or gels [47, 57]. The enthalphy of gelatinization (∆H) is a measure of the unwinding of the double-helices in the crystalline lamellae and the melting of the lamellae. This was shown to correlate positively with the length of the external chains (ECL) . ECL was not measured in this investigation, but according to previous works, ECL of potato, canna and yam is fairly similar (13.5–15.0) [8, 58], which agreed with the similar ∆H-values, and also with the rather similar relative crystallinities (Table 1). Using molecular modeling techniques, G3P, but not G6P, has been shown to interfere with the double-helix structure by introducing local amorphization of the crystallites , but the present result did not detect any destabilising effect (i.e. ∆H) by G3P, nor G6P, on the crystallites. Indeed, shoti starch, with the by far highest content of phosphate, also possessed the highest ∆H (Table 1). Possibly, the destabilising effect of G3P is below the detectable limit of DSC. The swelling properties of starch granules has also been considered highly dependent on the phosphorylation of amylopectin [37, 39, 59], based on the well known fact that potato starch granules swell extensively and at a comparatively low temperature. Indeed, this was once again confirmed in this work (Fig. 3). However, we could not confirm that this property of potato granules is defined by its high phosphate content, because shoti starch granules swelled considerably less, and at higher temperature, despite a three times higher content of phosphate. Instead, canna granules behaved similar to potato starch with respect to the temperature interval at which they swelled, but their degree of swelling remained lower. In contrast, lesser yam behaved as shoti starch and, in fact, the swelling factor of yam was slightly higher than for shoti at 95 °C, despite more than four times lower phosphate content. Therefore, one can conclude that also swelling, which is closely related to the gelatinization properties (cf. Table 1 and Fig. 3), was unaffected by the phosphate content in the granules. If phosphate does not exert an influence on the thermal properties of the Bcrystalline starch granules, then some other factor, or factors, must be responsible, such as the packing of the starch components in the granules or the molecular structure of the components, or possibly both, as the structure would affect the packing. As shown in Fig. 2, all starch samples possessed the typical B-crystalline polymorph with signature peaks at 5.6°, 22.0° and 24.0° 2θ . Minor differences were detected, however. For example, the double peak at 22° and 24° 2θ was sharper in potato and canna starch granules. Further, B-type crystalline starches tend to possess a rather poorly resolved doublet peak at 14.5–15° 2θ , but lesser yam possessed a more unusually shaped emerging single peak at this position, which was in common with other D. esculenta varieties . As some Dioscorea starches were reported to have C-type crystalline structures [30, 48], and A-type crystallinity is associated with a single peak at about 15° 2θ , we can not exclude a small proportion of A-type unit cells in lesser yam. Nevertheless, the detected small differences in WAXS patterns were largely confined to the two apparent groups of Bcrystalline samples (potato and canna versus shoti and yam), but the significance with regards to the crystalline structure is more difficult to unravel. Overall, the relative crystallinity (RC) based on the WAXS results did not obey any special pattern among the samples (Table 1). Muhr et al.  concluded that ∆H of gelatinization as measured by DSC is not related to the degree of crystallinity in the granules and Cook and Gidley  found that ∆H primarily measures the melting of helical structures, but many other physical events induced by heating play also important roles. This was confirmed here, as the measured ∆H did not reflect the RC based on WAXS. However, the gelatinization temperature can be considered as a measure of the stability of the crystallites, and the temperature range (Tc–To) is an indication of the homogeneity of the crystallites [19, 38, 62]. For shoti and yam starch, Tp was high and Tc–To was narrow showing that the crystallites in these granules were more stable and homogenous in structure in comparison to potato and canna. Even though the crystallites generally are formed by the major group of short chains of amylopectin, the properties of the crystallites are affected by the chains in the amorphous lamella  and especially the length of the inter-block segments (IBCL) have been shown to be important [13, 14]. In addition to amylopectin, amylose is assumed to be found to a large part in the amorphous lamellae and, if interacting with inter-block segments – or even more likely, with the longer inter-cluster segments  – it might possess an indirect effect on the melting property of the amylopectin crystallites. Indeed, amylose was suggested to be mixed with amylopectin in potato starch, as opposed to maize starch where it is separated from amylopectin and no or only weak interactions between the two components exist [64, 65]. It is also possible that amylose segments can form double helical segments with the short amylopectin chains if amylose transverses the stacks of lamellae [66, 67]. The amylose content in the B-crystalline starch granules in this work was fairly similar (Table 3) and did not suggest any correlation with the thermal properties or the swelling of the granules. On the other hand, the structure of the amylose component was different in the starches and appeared to correlate with the gelatinization properties. Thus, potato and canna starch contained considerably more of long amylose chains, so that the ratio of LCAM:SCAM was about two to three times higher than in shoti and yam (Table 3, Fig. 4). It is well known that amylose exist both as linear and branched molecules and the proportion varies between types of starches . Branched amylose has generally larger molecular weight than linear amylose, but the chains are comparatively short [68-71]. Thus, a high amount of short chains (low ratio of LCAM:SCAM) implies a higher proportion of branched amylose and vice versa. Along with this line, shoti and yam starch had more branched amylose, whereas potato and canna were enriched in linear amylose. At this time, however, a discussion of how and to what extent these differences will affect the properties of the granules rests solely on speculation. Details in the unit chain distribution of the amylopectin component revealed that the amount of Afp-chains (DP 6–8) was considerably larger in potato and canna compared to shoti and yam (Fig. 5, Table 4). As these chains are too short to effectively participate in the double-helical structures , they have been considered to introduce defects into the crystallites rather than contributing to their structure [73, 74]. Therefore, this could explain at least partially the weaker crystal structure in potato and canna starch. In addition, also other short chains with DP up to ~14 (here designated "S1") were more common in potato and canna and appeared therefore to have a negative effect on the crystal stability. Many investigators have found the same effect, i.e. amylopectin chains with DP ≤ 12 tend to be associated with low gelatinization temperature [30, 35, 75, 76] and high swelling power  and a possible explanation is that these chains are shorter than the optimal length to fill up the crystalline lamella . This explanation is not immediately clear, however, especially as chains with DP around 12 are commonly found in lintnerized starches [19, 51, 52, 79] and corresponds fairly well to the thickness of the crystalline lamellae, which is 4–6 nm (or DP 12~17) [66, 74]. A considerable number of chains with DP ≤ 12 are probably A-chains, which – if correctly defined – are chains not carrying other chains at the C6-positions , but others are B-chains (which carry other chains) and some A-chains are, in fact, longer than DP 12 . It is possible that some A-chains form double helices with other A-chains along the backbone of amylopectin . In contrast to A-B double-helices, such A-A double-helices have been very little considered hitherto, if at all. In the A-B chain union the double-helix is stabilised by the branching point , but this is not the case in the A-A doublehelix and, especially if in addition being comparatively short, it would be less stable than the majority of the double-helices, contributing to a lower gelatinization temperature in potato and canna. Shoti and yam starches had more S2-chains (DP 15–38) than potato and canna (Fig. 5, Table 4). These chains correspond mostly to the major part of the short, internal chains (so called BSmajor-chains)  and they were suggested to be involved in the interconnection of building blocks along the backbone, but probably they are also to a large extent forming short branches to the backbone . I.e., these chains possess the comparatively long inter-block segments with DP 7–8 (typically found in type 4 starches) and more of these chains will therefore stabilise the crystals in line with the finding that IB-CL correlates positively with To . The starch granules reacted very different when treated in dilute hydrochloric acid and, again, they could be divided into two groups. As expected, the rate of hydrolysis was much slower at 25 °C compared to 45 °C (Fig. 6). Initially, all starches were hydrolysed by approximately the same rate at the respective temperature. During this period, most of the amorphous regions are removed by the acid  and it appeared that the constitution of these parts, which are formed mostly by the internal chains of the amylopectin and a large part of amylose, were similar in the starches. The second, slower stage was different, however. Potato and canna were hydrolyzed with a higher rate at 45 °C and the difference to shoti and yam was large. It has been suggested that during this stage both amorphous and crystalline structures are simultaneously degraded, as the relative crystallinity in the granules do not increase anymore . As the molecular distribution of the single chain fraction found around the peak at DP 15~17 [23, 84] do not gradually change into shorter chains by time, it was suggested that the double-helices slowly dissolves into the solution and finally degrades further in the solution rather than as being part of the residual granules . As this step was clearly different in the two groups of starches, it suggested that the crystallites – and the residual amorphous areas – possessed different resistance towards the solubilisation of the helices at the higher temperature. As all starches were B-crystalline, the difference could not depend on the organisation of the doublehelices into the crystalline pattern per se, but other factors must be involved, such as the degree of defects caused by short chains, amylose tie chains, etc. Another striking result was the fact that the molecular composition of potato and canna lintners was strongly dependent on the temperature, whereas shoti and yam lintners were not (Fig. 7). The apparent chain length (CL) also increased at the higher temperature in potato and canna, and the number of chains per molecule (NC) increased especially in canna, showing that some branches escaped the acid hydrolysis (Table 5). This phenomenon has been described earlier for potato starches with different amylose contents as well as phosphate contents and was suggested to depend on a reorganisation of the double-helices in a fashion similar to what happens during annealing [15, 50, 52]. Apparently, the double-helices in shoti and yam starch granules were more fixed into their positions, which prevented similar movements in these starches. Indeed, the gelatinisation temperature interval was considerably narrower for shoti and lesser yam (Table 1), suggesting that these granules already possessed a well optimised crystalline structure, which would make them more stiff in comparison to the crystallites in potato and canna. Interestingly, an optimised crystalline structure did not necessarily lead to an optimal organisation of the crystallites in a radial fashion inside starch granules. Fig. 8 shows that the granules generally retained the polarisation cross after lintnerization. In lesser yam, many granules possessed a very irregular cross (Fig. 8d), suggesting poorer orientation of the crystallites. Indeed, in these granules there were no sign of the "granular rings" even after lintnerisation, suggesting a lack of higher-range order as well. Yet, the size-distribution of the dextrins in yam linters was not different from shoti (or the other starches at 25 °C) (Fig. 7). Another remarkable finding was that the other part of the yam granules, i.e. those with a regular polarization cross, possessed "granular slices" instead of the common rings (Figs. 8e-f). This suggested a very well developed organisation inside these granules with the crystallites found in one direction from the distal hilum at the top of the pyramide-like granules to the base of the pyramidals, instead of a radial organisation. A quite similar organisation of the helices was described for the giant starch granules from the pseudo-bulbs of the orchid Phajus grandifolius . Some of these granules, which are B-crystalline, have protruberances extending from an ogival body. The protruberances have almost planar "growth rings", much resembling slice-like structures, that at the edges turn and appear to fuse together . Possibly the "slices" in lesser yam granules also fuse together at the ends, but these parts were eventually removed by the acid. The internal structure of shoti starch granules was not as clearly observed as in yam, but several granules appeared to possess layers that ended at the edges (Fig. 8c), thus resembling the structures in lesser yam or Phajus grandifolius granules. 5 Conclusions The results in this investigation, which included four starch samples, showed that the B-crystalline starch granules can be divided into two types based on their structure and granular properties. However, as more samples are investigated a broader diversity might be found. Nevertheless, one structural type is represented by potato and edible canna starch granules with growth rings, or shells, and radially organised double-helices. The amylose component in these starches is comparatively little branched, whereas the amylopectin component has much short chains, especially Afp-chains, which possibly contribute to defected, less organised crystallites characterised by low melting temperatures as well as low swelling temperatures. The other structural type is represented by shoti and lesser yam granules, which appear to possess more planar internal structures with double-helices oriented in one direction from a proximal hilum to a distal, more flattened end. In this group amylose is more branched and amylopectin contains less short chains. This results in more organised and rigid crystallites with high melting and swelling temperatures, as well as a higher resistance to acid hydrolysis. It is noted that all starches in this investigation have amylopectin with an internal structure classified as type 4, for which one of the structural signatures is long inter-block segments, resulting in expected high gelatinization temperatures. In this aspect, potato and canna starch are exceptions possessing unexpectedly low gelatinization temperatures. While this work have highlighted a correlation between the molecular and granular structure with the thermal properties of B-crystalline starch granules, the actual relationship between the crystalline polymorph, thermal properties and branch configuration and the proposed higher “blocklet” structure remains to be understood. Acknowledgement Ms Jeanette Wikman is acknowledged for skillful laboratory assistance. Authors declare no conflict of interest. 6 References  Pérez, S., Bertoft, E., The molecular structures of starch components and their contribution to the architecture of starch granules: A comprehensive review. Starch/Stärke 2010, 62, 389-420.  Jenkins, P. J., Cameron, R. E., Donald, A. M., A universal feature in the structure of starch granules from different botanical sources. Starch/Stärke 1993, 45, 417-420.  Bertoft, E., On the building block and backbone concepts of amylopectin structure. Cereal Chem. 2013, 90, 294-311.  Waigh, T. A., Kato, K. L., Donald, A. M., Gidley, M. J., Clarke, C. J., Riekel, C. S., Side-chain liquid-crystalline model for starch. Starch/Stärke 2000, 52, 450460.  Buléon, A., Colonna, P., Planchot, V., Ball, S., Starch granules: structure and biosynthesis. Int. J. Biol. Macromol. 1998, 23, 85-112.  Popov, D., Buléon, A., Burghammer, M., Chanzy, H., Montesanti, N., Putaux, J.L., Potocki-Véronèse, G. Riekel, C., Crystal structure of A-amylose: A revisit from synchrotron microdiffraction analysis of single crystals. Macromolecules 2009, 42, 1167-1174.  Buléon, A., Gérard, C., Riekel, C., Vuong, R., Chanzy, H., Details of the crystalline ultrastructure of C-starch granules revealed by synchrotron microfocus mapping. Macromolecules 1998, 31, 6605-6610.  Bertoft, E., Piyachomkwan, K., Chatakanonda, P., Sriroth, K., Internal unit chain composition in amylopectins. Carbohydr. Polym. 2008, 74, 527-543.  Bertoft, E., Composition of building blocks in clusters from potato amylopectin. Carbohydr. Polym. 2007, 70, 123-136.  Bertoft, E., Composition of clusters and their arrangement in potato amylopectin. Carbohydr. Polym. 2007, 68, 433-446.  Bertoft, E., Koch, K., Åman, P., Building block organisation of clusters in amylopectin of different structural types. Int. J. Biol. Macromol. 2012, 50, 1212-1223.  Bertoft, E., Koch, K., Åman, P., Structure of building blocks in amylopectins. Carbohydr. Res. 2012, 361, 105-113.  Vamadevan, V., Bertoft, E., Seetharaman, K., On the importance of organization of glucan chains on thermal properties of starch. Carbohydr. Polym. 2013, 92, 1653-1659.  Vamadevan, V., Bertoft, E., Soldatov, D. V., Seetharaman, K., Impact on molecular organization of amylopectin in starch granules upon annealing. Carbohydr. Polym. 2013, 98, 1045-1055.  Wikman, J., Blennow, A., Buléon, A., Putaux, J.-L., Pérez, S., Seetharaman, K., Bertoft, E., Influence of amylopectin structure and degree of phosphorylation on the molecular composition of potato starch lintners. Biopolymers 2014, 101, 257-271.  Fredriksson, H., Silverio, J., Andersson, R., Eliasson, A.-C., Åman, P., The influence of amylose and amylopectin characteristics on gelatinization and retrogradation properties of different starches. Carbohydr. Polym. 1998, 35, 119-134.  Wischmann, B., Adler-Nissen, J., Melting properties and lintnerisation of potato starch with different degrees of phosphorylation. Starch/Stärke 2002, 54, 499507.  Varatharajan, V., Hoover, R., Li, J., Vasanthan, T., Nantanga, K. K. M., Seetharaman, K., Liu, Q., Donner, E., Jaiswal, S., Chibbar, R. N., Impact of structural changes due to heat-moisture treatment at different temperatures on the susceptibility of normal and waxy potato starches towards hydrolysis by porcine pancreatic alpha amylase. Food Res. Int. 2011, 44, 2594-2606.  Jacobs, H., Eerlingen, R. C., Rouseu, N., Colonna, P., Delcour, J. A., Acid hydrolysis of native and annealed wheat, potato and pea starches - DSC melting features and chain length distributions of lintnerised starches. Carbohydr. Res. 1998, 308, 359-371.  Blennow, A., Bay-Smidt, A. M., Olsen, C. E., Møller, B. L., The distribution of covalently bound phosphate in the starch granule in relation to starch crystallinity. Int. J. Biol. Macromol. 2000, 27, 211-218.  Blennow, A., Bay-Smidt, A. M., Wischmann, B., Olsen, C. E., Lindberg-Møller, B., The degree of starch phosphorylation is related to the chain length distribution of the neutral and the phosphorylated chains of amylopectin. Carbohydr. Res. 1998, 307, 45-54.  Tester, R. F., Morrison, W. R., Swelling and gelatinization of cereal starches. I. Effects of amylopectin, amylose, and lipids. Cereal Chem. 1990, 67, 551-557.  Robin, J. P., Mercier, C., Charbonnière, R., Guilbot, A., Lintnerized starches. Gel filtration and enzymatic studies of insoluble residues from prolonged acid treatment of potato starch. Cereal Chem. 1974, 51, 389-406.  Dubois, M., Gilles, K. A., Hamilton, J. K., Rebers, P. A., Smith, F., Colorimetric method for determination of sugars and related substances. Anal. Chem. 1956, 28, 350-356.  Klucinec, J. D., Thompson, D. B., Fractionation of high-amylose maize starches by differential alcohol precipitation and chromatography of the fractions. Cereal Chem. 1998, 75, 887-896.  Bertoft, E., Källman, A., Koch, K., Andersson, R., Åman, P., The cluster structure of barley amylopectins of different genetic backgrounds. Int. J. Biol. Macromol. 2011, 49, 441-453.  Blennow, A., Bay-Smidt, A. M., Olsen, C. E., Lindberg-Møller, B., Analysis of starch-bound glucose 3-phosphate and glucose 6-phosphate using controlled acid treatment combined with high-performance anion-exchange chromatography. J. Chromatogr. A 1998, 829, 385-391.  French, D., Fine structure of starch and its relationship to the organization of starch granules. J. Jpn. Soc. Starch Sci. 1972, 19, 8-25.  Jane, J.-l., Kasemsuwan, T., Leas, S., Zobel, H., Robyt, J. F., Anthology of starch granule morphology by scanning electron microscopy. Starch/Stärke 1994, 46, 121-129.  Srichuwong, S., Sunarti, T. C., Mishima, T., Isono, N., Hisamatsu, M., Starches from different botanical sources I: Contribution of amylopectin fine structure to thermal properties and enzyme digestibility. Carbohydr. Polym. 2005, 60, 529538.  Ren, S., Comparative analysis of some physicochemical properties of 19 kinds of native starches. Starch/Stärke 2017, 68, 1600367.  Leonel, M., Sarmento, S. B. S., Cereda, M. P., New starches for the food industry: Curcuma longa and Curcuma zedoaria. Carbohydr. Polym. 2003, 54, 385-388.  Katz, J. R., Itallie, T. B. v., Abhandlungen zur physikalischen Chemie der Stärke und der Brotbereitung. V. Alle Stärkearten haben das gleiche Retrogradtionsspektrum. Z. physikal. Ch. (A) 1930, 150, 90-99.  Buléon, A., Bizot, H., Delage, M. M., Pontoire, B., Comparison of X-ray diffraction patterns and sorption properties of the hydrolyzed starches of potato, wrinkled and smooth pea, broad bean and wheat. Carbohydr. Polym. 1987, 7, 461-482.  Gomand, S. V., Lamberts, L., L, J., Derde, Goesaert, H., Vandeputte, G. E., Goderis, B., Visser, R. G. F., Delcour, J., Structural properties and gelatinisation characteristics of potato and cassava starches and mutants thereof. Food Hydrocoll. 2010, 24, 307-317.  Hoover, R., Vasanthan, T., The effect of annealing on the physiochemical properties of wheat, oat, potato and lentil starches. J. Food Biochem. 1993, 17, 303-325.  Karim, A. A., Toon, L. C., Lee, V. P. L., Ong, W. Y., Fazilah, A., Noda, T., Effects of phosphorus contents on the gelatinization and retrogradation of potato starch. J. Food Sci. 2007, 72, C132-C138.  Vamadevan, V., Bertoft, E., Structure-function relationships of starch components. Starch/Stärke 2015, 67, 55-68.  Hoover, R., Composition, molecular structure, and physicochemical properties of tuber and root starches: a review. Carbohydr. Polym. 2001, 45, 253-267.  Yusuph, M., Tester, R. F., Ansell, R., Snape, C. E., Composition and properties of starches extracted from tubers of different potato varieties grown under the same environmental conditions. Food Chem. 2003, 82, 283-289.  Thitipraphunkul, K., Uttapap, D., Piyachomkwan, K., Takeda, Y., A comparative study of edible canna (Canna edulis) starch from different cultivars. Part I. Chemical composition and physicochemical properties. Carbohydr. Polym. 2003, 53, 317-324.  Srichuwong, S., Sunarti, T. C., Mishima, T., Isono, N., Hisamatsu, M., Starches from different botanical sources II: Contribution of starch structure to swelling and pasting properties Carbohydr. Polym. 2005, 62, 25-34.  Zhu, F., Isolation, composition, structure, properties, modifications, and uses of yam starch. Compr. Rev. Food Sci. Food Safety 2015, 14, 357-386.  Glaring, M. A., Koch, C. B., Blennow, A., Genotype-specific spatial distribution of starch molecules in the starch granule: A combined CLSM and SEM approach. Biomacromol. 2006, 7, 2310-2320.  Jyothi, A. N., Moorthy, S. N., Vimala, B., Physicochemical and functional properties of starch from two species of Curcuma. Int. J. Food Prop. 2003, 6, 135-145.  Mukerjea, R., Mukerjea, R., Robyt, J. F., Controlled peeling of the surface of starch granules by gelatinization in aqueous dimethyl sulfoxide at selected temperatures. Carbohydr. Res. 2006, 341, 757-765.  Thygesen, L. G., Blennow, A., Engelsen, S. B., The effects of amylose and starch phosphate on starch gel retrogradation studied by low-field 1H-NMR relaxometry. Starch/Stärke 2003, 55, 241-249.  Jayakody, L., Hoover, R., Liu, Q., Donner, E., Studies on tuber starches. II. Molecular structure, composition and physicochemical properties of yam (Dioscorea sp.) starches grown in Sri Lanka. Carbohydr. Polym. 2007, 69, 148163.  Jane, J.-l., Wong, K.-s., McPherson, A. E., Branch-structure difference in starches of A- and B-type X-ray patterns revealed by their Naegeli dextrins. Carbohydr. Res. 1997, 300, 219-227.  Bertoft, E., Lintnerisation of two amylose-free starches of A- and B-crystalline types, respectively. Starch/Stärke 2004, 56, 167-180.  Srichuwong, S., Isono, N., Mishima, T., Hisamatsu, M., Structure of linterized starch is related to X-ray diffraction pattern and susceptibility to acid and enzyme hydrolysis of starch granules. Int. J. Biol. Macromol. 2005, 37, 115121.  Wikman, J., Blennow, A., Bertoft, E., Effect of amylose deposition on potato tuber starch granule architecture and dynamics as studied by lintnerization. Biopolymers 2013, 99, 73-83.  Kainuma, K., French, D., Nägeli amylodextrin and its relationship to starch granules structure. I. Preparation and properties of amylodextrins from various starch types. Biopolymers 1971, 10, 1673-1680.  Muhrbeck, P., Eliasson, A.-C., Influence of the naturally occurring phosphate esters on the crystallinity of potato starch. J. Sci. Food Agric. 1991, 55, 13-18.  Muhrbeck, P., Svensson, E., Annealing properties of potato starches with different degrees of phosphorylation. Carbohydr. Polym. 1996, 31, 263-267.  Noda, T., Takigawa, S., Matsuura-Endo, C., Kim, S.-J., Hashimoto, N., Yamauchi, H., Hanashiro, I., Takeda, Y., Physicochemical properties and amylopectin structures of large, small, and extremely small potato starch granules. Carbohydr. Polym. 2005, 60, 245-251.  Blennow, A., Engelsen, S. B., Helix-breaking news: fighting crystalline starch energy deposits in the cell. Trends Plant Sci. 2010, 15, 236-240.  Wikman, J., Larsen, F. H., Motawia, M. S., Blennow, A., Bertoft, E., Phosphate esters in amylopectin clusters of potato tuber starch. Int. J. Biol. Macromol. 2011, 48, 639-649.  Gomand, S. V., Lamberts, L., Visser, R. G. F., Delcour, J. A., Physicochemical properties of potato and cassava starches and their mutants in relation to their structural properties. Food Hydrocoll. 2010, 24, 424-433.  Muhr, A. H., Blanshard, J. M. V., Bates, D. R., The effect of lintnerisation on wheat and potato starch granules. Carbohydr. Polym. 1984, 4, 399-425.  Cooke, D., Gidley, M. J., Loss of crystalline and molecular order during starch gelatinisation: origin of the enthalpic transition. Carbohydr. Res. 1992, 227, 103-112.  Tester, R. F., Debon, S. J. J., Annealing of starch – a review. Int. J. Biol. Macromol. 2000, 27, 1-12.  Perry, P. A., Donald, A. M., The role of plasticization in starch granules assembly. Biomacromol. 2000, 1, 424-432.  Zobel, H. F., Molecules to granules: a comprehensive starch review. Starch/Stärke 1988, 40, 44-50.  Saibene, D., Seetharaman, K., Amylose involvement in the amylopectin clusters from potato starch granules. Carbohydr. Polym. 2010, 82, 376-383.  Kozlov, S. S., Blennow, A., Krivandin, A. V., Yuryev, V. P., Structural and thermodynamic properties of starches extracted from GBSS and GWD suppressed potato lines. Int. J. Biol. Macromol. 2007, 40, 449-460.  Vamadevan, V., Hoover, R., Bertoft, E., Seetharaman, K., Hydrothermal treatment and iodine binding provide insights into the organization of glucan chains within the semi-crystalline lamellae of corn starch granules. Biopolymers 2014, 101, 871-885.  Takeda, Y., Hizukuri, S., Takeda, C., Suzuki, A., Structures of branched molecules of amyloses of various origins, and molecular fractions of branched and unbranched molecules. Carbohydr. Res. 1987, 165, 139-145.  Takeda, Y., Maruta, N., Hizukuri, S., Juliano, B. O., Structures of indica rice starches (IR48 and IR64) having intermediate affinities for iodine. Carbohydr. Res. 1989, 187, 287-294.  Hizukuri, S., Takeda, Y., Maruta, N., Juliano, B. O., Molecular structure of rice starch. Carbohydr. Res. 1989, 189, 227-235.  Takeda, Y., Maruta, N., Hizukuri, S., Structures of amylose subfractions with different molecular sizes. Carbohydr. Res. 1992, 226, 279-285.  Gidley, M. J., Bulpin, P. V., Crystallisation of malto-oligosaccharides as models of the crystalline forms of starch: Minimum chain-length requirement for the formation of double helices. Carbohydr. Res. 1987, 161, 291-300.  Vermeylen, R., Goderis, B., Reynaers, H., Delcour, J. A., Gelatinisation related structural aspects of small and large wheat starch granules. Carbohydr. Polym. 2005, 62, 170-181.  Genkina, N. K., Wikman, J., Bertoft, E., Yuryev, V. P., Effects of structural imperfection on gelatinization characteristics of amylopectin starches with Aand B-type crystallinity. Biomacromol. 2007, 8, 2329-2335.  Nakamura, Y., Sakurai, A., Inaba, Y., Kimura, K., Iwasawa, N., Nagamine, T., The fine structure of amylopectin in endosperm from Asian cultivated rice can be largely classified into two classes. Starch/Stärke 2002, 54, 117-131.  Vandeputte, G. E., Vermeylen, R., Geeroms, J., Delcour, J. A., Rice starches. I. Structural aspects provide insight into crystallinity characteristics and galatinisation behaviour of granular starch. J. Cereal Sci. 2003, 38, 43-52.  Vandeputte, G. E., Vermeylen, R., Geeroms, J., Delcour, J. A., Rice starches. II. Structural aspects provide insight into swelling and pasting properties. J. Cereal Sci. 2003, 38, 53-59.  Vandeputte, G. E., Delcour, J. A., From sucrose to starch granule to starch physical behaviour: a focus on rice starch. Carbohydr. Polym. 2004, 58, 245266.  Wang, S., Copeland, L., Effect of acid hydrolysis on starch structure and functionality: A review. Crit. Rev. Food Sci. Nutr. 2015, 55, 1081-1097.  Peat, S., Whelan, W. J., Thomas, G. J., Evidence of multiple branching in waxy maize starch. J. Chem. Soc. 1952, 4546-4548.  Imberty, A., Pérez, S., Conformational analysis and molecular modelling of the branching point of amylopectin. Int. J. Biol. Macromol. 1989, 11, 177-185.  Bertoft, E., Laohaphatanaleart, K., Piyachomkwan, K., Sriroth, K., The fine structure of cassava amylopectin. Part 2. Building block structure of clusters. Int. J. Biol. Macromol. 2010, 47, 325-335.  Wang, S., Blazek, J., Gilbert, E., Copeland, L., New insights on the mechanism of acid degradation of pea starch. Carbohydr. Polym. 2012, 87, 1941-1949.  Watanabe, T., French, D., Structural features of naegeli amylodextrin as indicated by enzymic degradation. Carbohydr. Res. 1980, 84, 115-123.  Chanzy, H., Putaux, J.-L., Dupeyre, D., Davies, R., Burghammer, M., Montanari, S., Riekel, C., Morphological and structural aspects of the giant starch granules from Phajus grandifolius. J. Struct. Biol. 2006, 154, 100-110. Legends to figures Figure 1. Microscopy in polarized light of starch granules from potato, edible canna, shoti and lesser yam. Irregular birefringence patterns in the latter are highlighted by white arrows. Bar = 50 µm. Figure 2. Wide-angle X-ray scattering pattern of B-crystalline starch granules. Figure 3. Swelling factor of B-crystalline starch granules at 55–95 °C. Figure 4. Gel permeation chromatography on Sepharose CL 6B of debranched starches of potato () and canna () (upper figure), and of shoti () and lesser yam () (lower figure). Long chains of amylose (LCAM), short chains of amylose (SCAM), and chains of amylopectin are indicated. Figure 5. (a) Unit chain profiles obtained by HPAEC of amylopectin from potato () and canna (), and (b) from shoti () and lesser yam (). (c) Difference plots of shotii () and lesser yam () versus potato amylopectin. Chain categories are indicated. Figure 6. Lintnerization of B-crystalline starch granules at 25 °C (a) and 45 °C (b): Potato (); canna (); shoti (); lesser yam (). Note the different time scales in (a) and (b). Note also that 80% solubilisation with shoti starch, in fact, was obtained after 145 h (not shown in this figure). Figure 7. HPAEC analysis of the composition of dextrins in lintners after about 80% solubilisation at 25 °C (----) and 45 °C (–––––). Figure 8. Microscopy in polarized light of lintnerized starch granules from potato, edible canna, shoti and lesser yam. (f) shows a magnification of granules in (e). Arrows in (c), (e) and (f) highlight granules that appears to have "slices" rather than granular "rings". Arrows in (d) highlight granules with irregular "Maltese cross" and no "slices". Bar = 50 µm. Table 1. Relative crystallinity and gelatinization parameters of starch granulesa RC To Tp Tc Tc–To ∆H (%) (°C) (°C) (°C) (°C) (J/g) Potato 32 57.3 61.5 67.8 10.5 15.1 Canna 26 60.4 65.7 71.2 10.8 16.1 Shoti 30 74.5 77.1 81.1 6.6 16.5 Lesser yam 30 74.8 77.5 81.3 6.5 15.7 Sample a RC = Relative crystallinity; To, Tp, and Tc = Onset, peak, and conclusion temperature of gelatinization, respectively; ∆H = Enthalpy of gelatinization. Values shown are average of duplicates. Table 2. Phosphate content in starcha Sample Total P G6P G3P (nmol/mg) (nmol/mg) (nmol/mg) Potatob 21.2 15.2 6.0 2.5 Canna 13.7 9.6 4.1 2.3 Shoti 63.2 56.4 6.8 8.3 Lesser yam 14.7 8.0 6.7 1.2 a G6P/G3P P = Phosphate; G6P = Glucose 6-phosphate; G3P = Glucose 3-phosphate. From . Values are average of duplicates. b Table 3. Amylose content and structure based on debranched starches analysed by GPC on Sepharose CL 6Ba Sample Amylose LCAM SCAM (%) (%) (%) Potato 22.5 15.4 7.1 2.2 Canna 24.8 14.0 10.8 1.3 Shoti 24.3 8.9 15.4 0.6 Lesser yam 18.6 8.4 10.2 0.8 a LCAM:SCAM LCAM = Long chains of amylose; SCAM = Short chains of amylose. Values are average of duplicates. Table 4. Relative amount of chains (mol%) and chain lengths in amylopectin Sample Afp S1 S2 S L CL SCL LCL Potato 7.0 32.4 49.4 88.8 11.2 21.4 16.7 58.5 Canna 6.5 31.0 50.0 87.5 12.5 21.9 16.7 58.6 Shoti 3.9 25.1 56.1 85.1 14.9 24.8 18.4 61.9 Lesser yam 3.5 27.6 56.2 87.3 12.7 23.0 18.0 59.4 DP-division of chain segments based on Fig. 5: Afp = 6–8, S1 = 9–14, S2 = 15–38, S = ≤ 38, L = ≥ 39, CL = chain length, SCL = CL of S-chains, LCL = CL of L-chains. Values are average of duplicates. Table 5. Molecular structure of lintnerized starches Lintner Temp. DP CLa NCb (°C) Potato Canna Shoti Lesser yam a b Total P G6P/G3P (nmol/mg) 25 21.0 14.9 1.4 11.3 1.9 45 26.6 17.7 1.5 15.9 1.9 25 20.3 14.8 1.4 10.4 2.1 45 27.0 16.1 1.7 14.7 3.7 25 21.3 17.9 1.2 36.3 1.9 45 20.2 18.3 1.1 27.0 2.8 25 19.9 15.0 1.3 9.2 1.1 45 20.9 15.1 1.4 7.5 1.0 Apparent average chain length after debranching with isoamylase and pullulanase. Average number of chains per molecule = DP/CL.