Gene expression profiles predictive of cold-induced sweetening in potato


Neilson, J., Lagüe, M., Thomson, S., Aurousseau, F., Murphy, A.M., Bizimungu, B., Deveaux, V., Bègue, Y., Jacobs, J.M.E., Tai, H.H. (2017). Gene expression profiles predictive of cold-induced sweetening in potato. Functional and Integrative Genomics, [online] 17(4), 459-476.

Plain language summary

Storage of potatoes after harvest at low temperatures is good for suppressing disease and spouting. But, the cold temperature induces potato tubers to convert starch to sugar if they are stored too long. Sugars in potato tubers, especially reducing sugars, causing browning of fried products like chips and French fries. The study is about developing a tool that can be used at harvest to predict how long tubers can be stored in the cold before starch turns to sugar.


Cold storage (2–4 °C) used in potato production to suppress diseases and sprouting during storage can result in cold-induced sweetening (CIS), where reducing sugars accumulate in tuber tissue leading to undesirable browning, production of bitter flavors, and increased levels of acrylamide with frying. Potato exhibits genetic and environmental variation in resistance to CIS. The current study profiles gene expression in post-harvest tubers before cold storage using transcriptome sequencing and identifies genes whose expression is predictive for CIS. A distance matrix for potato clones based on glucose levels after cold storage was constructed and compared to distance matrices constructed using RNA-seq gene expression data. Congruence between glucose and gene expression distance matrices was tested for each gene. Correlation between glucose and gene expression was also tested. Seventy-three genes were found that had significant p values in the congruence and correlation tests. Twelve genes from the list of 73 genes also had a high correlation between glucose and gene expression as measured by Nanostring nCounter. The gene annotations indicated functions in protein degradation, nematode resistance, auxin transport, and gibberellin response. These 12 genes were used to build models for prediction of CIS using multiple linear regression. Nine linear models were constructed that used different combinations of the 12 genes. An F-box protein, cellulose synthase, and a putative Lax auxin transporter gene were most frequently used. The findings of this study demonstrate the utility of gene expression profiles in predictive diagnostics for severity of CIS.

Publication date