Cold storage reveals distinct metabolic perturbations in processing and non-processing cultivars of potato

Cold-induced sweetening (CIS) causes a great loss to the potato (Solanum tuberosum L.) processing industry wherein selection of potato genotypes using biochemical information through marker-trait associations has found to be advantageous. In the present study, we have performed nuclear magnetic resonance (NMR) spectroscopy-based metabolite profiling on tubers from five potato cultivars (Atlantic, Frito Lay-1533, Kufri Jyoti, Kufri Pukhraj, and PU1) differing in their CIS ability and processing characteristics at harvest and after one month of cold storage at 4°C. A total of 39 water-soluble metabolites were detected using 1H NMR. Multivariate statistical analysis indicated significant differences in metabolite profiles between processing and non-processing potato cultivars. Further analysis revealed distinct metabolite perturbations as induced by cold storage in both types of cultivars wherein significantly affected metabolites were categorized mainly as sugars, sugar alcohols, amino acids, and organic acids. Significant metabolic perturbations were used to carry out metabolic pathway analysis that in turn tracked 130 genes encoding enzymes (involved directly and/or indirectly) involved in CIS pathway using potato genome sequence survey data. Based on the metabolite perturbations, the possible relevant metabolite biomarkers, significantly affected metabolic pathways, and key candidate genes responsible for the observed metabolite variation were identified. Overall, studies provided new insights in further manipulation of specific metabolites playing a crucial role in determining the cold-induced ability and processing quality of potato cultivars for improved quality traits. Highlight Metabolomic profiling using 1D 1H-NMR and bioinformatics analysis of potato cultivars for the identification of metabolites and genes controlling biochemical pathways in cold-stored potato tubers

Introduction 1 2 Potato (Solanum tuberosum L.)an important staple non-grain vegetable food cropis used 3 globally for both processing and table purposes. Cold storage of potato tubers after harvesting is 4 mandatory to reduce sprouting, prevent diseases, avoid losses due to shrinkage, extend post-5 harvest shelf life, and to ensure year-round supply of quality tubers for consumption (Bianchi et 6 al., 2014;Hou et al., 2017). During cold storage the potato tubers exhibit the phenomenon of cold-7 induced sweetening (CIS), wherein rapid degradation of starch and sucrose hydrolysis has been 8 associated with accumulation of reducing sugars (RS) such as glucose and fructose (Burton, 1969;9 Dale and Bradshaw, 2003). During the frying process, these RS react with free amino acids in a 10 Maillard reaction to generate dark-pigmented products that are bitter and unsightly to consumers. 11 In addition to this, one of the products of the Maillard reaction is acrylamidea potent neurotoxin 12 and carcinogen (Menéndez et al., 2002;Mottram et al., 2002;Hajirezaei et al., 2003). Hence, CIS 13 is considered as one of the critical parameters in potato production as well as in processing; and 14 therefore identification and development of potato tubers resistant to CIS has become a priority in 15 a number of potato breeding programs (Xiong et al., 2002;Hamernik et al., 2009;Colman et al., 16 2017). It is necessary to identify and develop potato cultivars with CIS resistance along with good 17 processing quality attributes to meet the challenges of a frequently-changing market, production 18 circumstances, and improving their economic condition (Kaur and Aggarwal, 2014). In this regard, 19 the metabolic stability of potato tubers during the cold storage period has been identified as one of 20 the prime traits to be investigated for breeding programs worldwide (Sowokinos, 2001;Ali and 21 Jansky, 2015), wherein selection of potato genotypes at early generations using biochemical 22 information through marker-trait associations has been found to be advantageous (Slater et al., 23 2014;Gupta, 2017). 24 25 The potato processing industry is becoming an emerging sector in India and therefore, the demand 26 for processed potato products such as chips, French fries, flakes, etc. is increasing continuously 27 (Rana and Pandey, 2007). Ideally, potato cultivars suitable for processing should possess high 28 specific gravity and dry matter (DM) content along with low RS content ( Kaur et al., 2012;Kaur 29 and Aggarwal, 2014). In this regard, commercially grown processing (Atlantic and Frito Lay 1533) 30 and popular Indian non-processing (Kufri Jyoti and Kufri Pukhraj) potato cultivars (Kaur and 31 1 H total correlation spectroscopy (TOCSY) experiment (using mlevesgppg pulse sequence from 110 Bruker library) was measured with a 6000 Hz of spectral width resolved in 2048  1024 data points 111 with 40 transients per increment. A Hartman-Hahn mixing time of 80 ms was employed for the 112 TOCSY spin-lock using composite blocks of 90-180-90 pulses with 90 pulse width of 25 s 113 at 2.29 W of power. TOCSY data was recorded in States-TPPI mode and Smoothed square shaped 114 (SMSQ10.100) gradients were used with 31% power (after the spin-lock period) and 11% power 115 (before refocusing) for a duration of 1 ms. 116 117

Metabolite Identification and Quantification 118
All of the NMR data were processed using Topspin (v3.5) software 119 (www.bruker.com/bruker/topspin). 1 H NMR raw data was multiplied with exponential function 120 and zero-filled to 64K data points prior to Fourier transformation. All the spectra are manually 121 phased and the baseline is corrected before subjecting to further analysis. 1  After identification of metabolites, respective peaks were manually picked, integrated using 135 Topspin v3.5, and converted to absolute concentrations of individual metabolites using Chenomx 136 NMR suite 8.1 by comparing with the peak integrals from an external reference compound DSS 137 of known concentration (400 µM). The absolute concentrations obtained above were then 138 normalized using the dry weight obtained from the tuber mass used for metabolite extraction. The 139 data matrix file was created using concentrations of metabolites as obtained above from 30 distinct 140 samples. The lower limit of quantification achieved using above-mentioned NMR parameters was 141 0.25 M for the methyl peak of DSS at a s/n ratio of 10. 142 143 Metabolic pathway analysis, Blast similarity searching, gene identification, notation and 144 location on potato chromosomes 145 Metabolic pathway analysis depicting significantly affected metabolites in cold-stored potato 146 tubers was performed by comparing the primary metabolites based on KEGG and the reference 147 pathway (Sowokinos, 2001;Malone et al., 2006)  analysis. Next, pair-wise analysis of all five cultivars in FH and CS treatments was achieved using 169 Volcano plot analysis, where metabolites were selected based on dual criteria, 1) the significance 170 (false discovery rate (FDR) corrected p-value < 0.05), and fold-change in concentration (cut-off 171 for fold change was set to 1.5 fold increase or decrease). In addition to this, the VIP score plot 172 obtained by PCA identified the key metabolites responsible for the clustering of various groups. 173 Metabolites with a VIP score of ≥1.0 are generally considered to be statistically significant (Ma et 174 al., 2016;Wu et al., 2018). A union set of significant metabolites (those identified from volcano 175 plot analysis, and from VIP score following the above-mentioned criteria) were taken for 176 generating Box and Whisker plots to highlight the variation of a particular metabolite across 177 replicates, different cultivars, and in different treatment conditions. Metabolites, e.g. ascorbate, 178 having low signal-to-noise (s/n < 15) in NMR spectra, although identified with confidence, were 179 not included in box and whisker plot analysis as they might be prone to over-or under-estimation 180 of concentrations. Further, correlation plots were drawn to identify all the correlated metabolites 181 in FH and CS treatments for all five cultivars. The significantly affected pathways were then 182 identified using significantly perturbed metabolites as input in MetaboAnalyst tool and KEGG 183 pathway database (www.genome.jp/kegg/pathway.html). were also found to be different in processing, non-processing, and locally grown cultivars ( Fig. 1). 207 These differences in the metabolite content in the different cultivars at the two time-points could 208 be attributed to the genetic make-up of each cultivar used in the present study. In fact, previous 209 studies have also reported such variability in the metabolite content from different potato cultivars 210 differing in their genetic background (Defernez et al., 2004;Uri et al., 2014). 211 212 Pair-wise analysis of metabolic changes upon cold storage in processing, non-processing, and 213

local cultivars 214
The variations in metabolite profiles of potato cultivars differing in their genetic constitution offer 215 a potential tool to develop CIS resistant potatoes with genotypes encoding improved processing 216 characteristics. However, studies investigating the metabolic diversity from cold-stored potato 217 11 tubers differing in their processing and non-processing characteristics have been limiting. In order 218 to highlight the differences in the FH and the CS condition from the processing, the non-219 processing, and the local potato cultivars used in the study, pair-wise analysis was done (Fig. 2). 220 In addition to this, volcano plot analysis (Fig. 3) and VIP score plot analysis (Fig. 4)  in PC1 and 6.1% variation in PC2 (Fig. 2D), respectively upon CS. The levels of glucose, fructose, 240 mannose, galactose, aspartate, malate, fumarate, leucine, proline, trigonelline, asparagine, and 241 serine were increased in the Kufri Jyoti cultivar upon CS, while the levels of sucrose and alanine 242 were reduced ( Fig. 3C and Fig. 4C). Similarly, CS treatment of the Kufri Pukhraj cultivar was 243 associated with significant increase in the levels of fructose, glucose, 3-hydroxyisobutyrate, 244 mannose, malate, leucine, aspartate, serine, proline, isoleucine, adenosine, arginine, asparagine, 245 and methanol on one hand; it significantly decreased the levels of chlorogenate and formate upon 246 CS ( Fig. 3D and Fig. 4D). In the local PU1 cultivar, levels of formate, tryptophan, and sucrose 247 were significantly decreased, while 3-hydroxyisobutyrate, methanol, fructose, glucose, proline, 4-12 aminobutyrate, trigonelline, myo-inositol, arginine, aspartate, uridine, and sn-glycero-3-249 phosphocholine showed significant increase upon cold storage treatment ( Fig. 3E and Fig. 4E). 250 251 Metabolomics approach has been previously used to assess the effect of storage conditions on a 252 variety of potato cultivars. For example, metabolic profiles in different life cycle stages of potato 253 tubers were characterized to link temporal changes in metabolites related to trait development 254 . In a recent study, comprehensive metabolomics and ionomics analysis on 255 raw and cooked potato tubers of 60 unique genotypes were performed to understand the chemical 256 variation and nutritional values in different varieties (Chaparro et al., 2018). In another study, 257 storage of commercial cultivars at 20-22 C in the dark suggested a significant decrease in sucrose 258 and fructose (Uri et al., 2014). Here, we have reported that the storage of potato tubers at 4°C 259 significantly increased the levels of sucrose, particularly in Atlantic and Frito Lay 1533, while it 260 was significantly decreased in Kufri Jyoti and PU1, and remained invariant in Kufri Pukhraj (Fig. 261 5). On the other hand, we found that the increase in RS was more pronounced in the non-processing 262 cultivars Kufri Pukhraj, Kufri Jyoti, and PU1 as compared to the processing cultivars, Atlantic and 263 Frito Lay 1533 (Fig. 3, Fig. 4, and Fig. 5). These results are in agreement with other studies that 264 observed an increase in RS after cold storage of potato tubers (Kaur and Aggarwal, 2014;265 Aggarwal et al., 2017;Datir et al., 2019), which has been attributed to the enhanced activity of the 266 vacuolar invertase (Lin et al., 2013). The effect of silencing of vacuolar invertase, which converts 267 sucrose into glucose and fructose, on sugar metabolism pathways has previously been studied to 268 find suitable targets for further genetic manipulation to improve the tuber quality (Wiberley-269 Bradford et al., 2014). Brummell et al., (2011)  13 associated with the variation in RS levels in these cultivars (Datir et al., 2019). However, these 280 results need to be further validated using a qRT-PCR expression of vacuolar invertase inhibitor 281 gene before and after cold storage in these cultivars. 282

283
Although CS resulted in several significant metabolic perturbations, it is important to highlight the 284 significance of some metabolites that can be related to the CIS status of the potato cultivars. For 285 example, it is noteworthy that FL-1533 exhibited significantly higher citrate levels as compared to 286 rest of the cultivars after CS (Fig. 5), which might be associated with CIS resistance along with 287 chips with an acceptable colour. This is particularly important as citric acid is known as a popular 288 anti-browning agent, mainly because it not only inhibits the polyphenol oxidase by reducing pH 289 but also chelates copper at the enzyme-active site (McCord and Kilara 1983). Likewise, the 290 changes in the levels of total amino acids, specifically the levels of asparagine, and the ratio of free 291 asparagine to RS during cold storage were found to be significantly varied among different potato 292 cultivars upon CS (Fig. 5). These factors, therefore, can further influence the processing quality of 293 potato tubers. It is interesting to note that, among all the cultivars, PU1 cultivar in particular 294 showed significantly higher levels of methanol after CS (Fig. 5). The amount of methanol released 295 on saponification is the measure of the degree of pectin methylation and was found to be associated 296 indirectly with the potato tuber texture properties (Ross et al., 2010b). It can also be presumed that 297 some of these significantly affected metabolites might have acted as osmolytes such as proline, 298 trigonelline, 4-aminobutyrate (GABA), etc. (Evers et al., 2010) (Fig. 5, Fig. 6) as an acclimation 299 response under CS treatment. However, not much research has been focused on understanding the 300 function of various metabolites in the CIS process of potato tubers. Therefore, these uniquely 301 been previously reported (Dobson et al., 2010;Chaparro et al., 2018). Significant metabolite 337 variation and metabolite-metabolite correlations were detected from a collection of 60 unique 338 potato genotypes that span 5 different market classes such as russet, red, yellow, chip, and 339 speciality (Chaparro et al., 2018), where authors concluded that metabolite diversity and 340 correlations data can support the potential to breed new cultivars for improved health traits. 341

Metabolite biomarkers for the identification of CIS resistant and susceptible genotypes 343 CIS is a multigenic complex trait involving multiple intricate metabolic pathways which clearly 344
indicates that it is unlikely to be controlled by a single metabolite; thus, multiple metabolites would 345 come-up as plausible biomarkers for CIS in potatoes. Previous studies have suggested that various 346 primary metabolites in potato tubers can be utilized as biomarkers in breeding programs for 347 predicting agronomically important traits such as black spot bruising and chip quality (Steinfath 348 et al., 2010;Instroza-Blancheteau et al., 2018). We would like to point out that in addition to the 349 amount of RS, the total and individual amino acid content, the asparagine content, levels of organic 350 acids, and other metabolites could be considered as important processing parameters. Breeders aim 351 for the identification and development of processing potato cultivars with low free-asparagine and 352 RS as desirable characteristics for processing purpose. In the current study, a unique metabolite 353 combination was observed for the processing cultivar, FL-1533, which was represented by the 354 lowest amount of RS and asparagine compared to rest of the cultivars CS (Fig. 3, Fig. 4, and Fig.  355 5). The levels of RS and asparagine have been used as markers for potato trait development 356 . 357

358
Amongst the different TCA cycle metabolites, the levels of citrate were found to be significantly 359 higher in the processing cultivar, FL-1533, whereas Kufri Jyoti and Kufri Pukhraj showed 360 significantly higher levels of malate after CS (Fig. 6). Citrate and malate are critical in 361 determination of non-enzymic browning reactions, after cooking darkening, physiological age/ 362 stages of development in the storage of potato tubers (Wichrowska et al., 2009;Reust and Aerny, 363 1985). In addition, they indirectly influence the texture of cooked and fried potato products 364 (Heisler et al., 1964;Thomas et al., 1979;Lynch and Kaldy, 1985). Hence, it is necessary to 365 develop the indicators of tuber browning and physiological age mainly because both the growers 366 and seed companies can optimize the storage conditions for individual cultivars. Moreover, such 367 indicators will be extremely important in the determination of the suitability of potato tubers for 368 culinary use and industrial processing (Reust and Aerny, 1985). 369

370
The texture of potato tubers is a key determinant of the quality of processing as well as cooked 371 potato as has been shown to greatly influence the consumer's preference (Shomer and Kaaber,372 wall middle lamella during cooking, and the correlation between pectin methylesterase activity 374 and the degree of methylation of cell wall pectin (reviewed in Taylor et al., 2007;Ross et al., 375 2010b). The amount of methanol released on saponification is the measure of the degree of pectin 376 methylation and is indirectly associated with the potato tuber texture properties (Ross et al., 377 2010b). Significantly highest levels of methanol were exclusively recorded in the PU1 cultivar 378 after CS (Fig. 5). Therefore, the amount of methanol present in potato tubers can be used as a 379 potential marker for screening of potato cultivars for texture properties. 380 381 Importantly, several other metabolites such as fumarate, adenosine, sn-glycero-3-phosphocholine, 382 4-aminobutyrate, 3-hydroxyisobutyrate, trigonelline, and chlorogenate were significantly varied 383 upon CS (Fig. 3, Fig. 4, and Fig. 6) indicating that these metabolites might have a role in the CIS 384 process, as well as the determination of processing quality of these cultivars. In order to improve 385 potato (Solanum tuberosum L.) genotypes through selection or breeding, it is helpful to determine 386 the chemical composition of tubers (Pal et al., 2008). Maintaining the quality of potato tubers 387 during storage is a major challenge. Therefore, the information on the response of potato cultivars 388 to cold storage and metabolite accumulation can be useful for the development of biomarkers 389 predicting severity of CIS of different potato genotypes. Such biomarkers (supplementary table  390 S4) can then be tested on a wide range of potato genotypes differing in CIS response and easily 391 integrated into the existing potato storage management and breeding methods. Moreover, such 392 predictive biomarkers can be used in selection for potato breeding and for tailoring storage 393 conditions for each lot of harvested tubers (Neilson et al., 2017). Furthermore, biomarkers can be 394 utilized for the manipulation of a specific metabolite pathway for developing potato genotypes 395 with improved processing characteristics. 396 397

Metabolic Pathway Analysis 398
We performed the pathway analysis depicting significantly affected metabolites in cold-stored 399 potato tubers by comparing the primary metabolites based on KEGG and the reference pathway 400 ( Fig. 6) (Sowokinos, 2001;Malone et al., 2006). Cold temperature induces starch degradation in 401 potato tubers to principal sugars including sucrose, glucose, and fructose, thereby leading to an 402 imbalance between starch degradation and sucrose metabolism in tubers. So far, CIS studies have 403 mainly concentrated on the activity of enzymes involved in the conversion of starch and sugars 404 (Jansky and Fajardo, 2014). However, potato tubers displayed diverse biochemical mechanisms 405 during CIS and the amount of sugar in potato tubers is influenced by several candidate genes 406 operating in glycolysis, hexogenesis, and mitochondrial respiration (Sowokinos, 2001). The 407 metabolic pathway analysis presented in this study suggests that several metabolites were affected 408 during cold storage and mainly resulted from the alanine, aspartate, and glutamate metabolism; 409 valine, leucine, and isoleucine biosynthesis; arginine and proline metabolism; glycine, serine, 410 and threonine metabolism; the TCA cycle, fructose and mannose metabolism, galactose 411 metabolism, nicotinate and nicotinamide metabolism, glycolysis; and sucrose metabolism along 412 with several other metabolites (Fig. 6). Also, the levels of metabolites were found to be 413 specifically different depending on potato cultivars (Fig. 3, Fig. 4, and Fig. 5) indicating that the 414 specific metabolites might play a crucial role in determining the cold-induced ability of potato 415 cultivars. Also, the molecular events controlling such metabolic perturbations in potato tubers after 416 cold storage are still puzzling. Among various metabolic processes, carbohydrates, amino acids 417 and organic acids were identified as the main players in the CIS process and were either decreased 418 or increased under cold storage condition. In the amino acid metabolism pathways, the distinct 419 significantly affected pathways include metabolism of 11 amino acids: isoleucine, glutamate, 420 glutamine, leucine, alanine, arginine, proline, tryptophan, aspartate, asparagine and serine 421 metabolism. In the TCA cycle, the levels of citrate, malate, and fumarate were significantly 422 affected by CS. Particularly, citrate and fumarate synthesis was up-regulated in FL-1533 423 cultivar (Fig. 6). Several other metabolites such as 3-hydroxyisobutyrate, trigonelline, 424 galactose, mannose, etc. were either up-regulated or down-regulated in response to cold storage 425 (Fig. 6). On the other hand, methanol production was significantly enhanced in Atlantic, Kufri 426 Pukhraj, and PU1 cultivars although the extent of this increase was significantly higher in PU1 427 (Fig. 6). The GABA shunt pathway was significantly enhanced as seen by the increased levels of 428 4-aminobutyarte in PU1 upon cold storage, whereas it was not significantly affected in any other 429 potato cultivar. The convergence and divergence of various pathways involved in CIS revealed a 430 complex metabolic network. However, the roles of these metabolites and their accumulation 431 pattern in response to cold storage in different potato cultivars remains to be further investigated. 432 A possible approach to achieve this goal is to identify the genes which are putatively involved in 433 the formation of enzymes involved in the biosynthesis of these metabolites. 434