3.6 CRIR1 interacts with MeCSP5 protein
In order to further investigate the biological function of CRIR1 , we performed in vitro RNA pull-down assays and filtered the protein interactome of CRIR1 using mass spectrometry (Figure 6a). A total of 15 proteins, represented by two or more unique peptides, specifically interacted with biotin-labeled sense-CRIR1 but not the control antisense-CRIR1 (Figure 6b). Among them, three proteins, including one member of the Pre-rRNA PROCESSING PROTEIN (RRP5), one 30S ribosomal protein, and one COLD SHOCK DOMAIN CONTAINING PROTEIN (CSP, named as MeCSP5), were found to be associated withCRIR1 RNA function, and the other candidate interactors were mainly secondary metabolism‐related enzymes. CSP has been shown to regulate RNA processing and function by participating in multiple complexes to confer cold tolerance in Arabidopsis (Sasaki & Imai, 2011). The cassava genome contains six CSPs (MeCSP1to 6 ) (Figure S2a-b), two of which (MeCSP1and MeCSP5 ) were regulated by cold stress according to the RNA-seq data. Therefore, we verified possible direct interactions between CRIR1 and MeCSP5. First, we predicted regions ofCRIR1 necessary for interaction with MeCSP5 using the catRAPID database, and found that MeCSP5 was preferentially bound to the N-terminal of CRIR1 (Figure 6c). Further, to check the specificity of the interaction between CRIR1 and the MeCSP5 protein, full-length CRIR1 were separated into three truncated forms 1-3, and were synthesized by in vitro  transcription to evaluate their ability to bind MeCSP5 protein. As presented in Figure 6d, MeCSP5 co-purified with transcripts of segments 1 and 2 but not segment 3, demonstrating a sequence-specific binding activity of MeCSP5. We also checked the interaction of CRIR1 with MeCSP5 using trimolecular fluorescence complementation (TriFC) assay, a modified version of bimolecular fluorescence complementation (BiFC) assay using an MS2 system as described previously (Seo, Sun, Park, Huang, Yeh, Jung & Chua, 2017). We observed that CRIR1 was strongly associated with MeCSP5 in both the nucleus and cytoplasm, but the negative control RNA was not (Figure 6e). Collectively, these results suggest an association between CSP complexes and CRIR1 -mediated cold stress tolerance. Considering that CSP proteins have been confirmed to possess transcription anti-termination activity and have also been thought to enhance translation at low temperatures by eliminating stabilized RNA secondary structures (Sasaki & Imai, 2011). We thus compared the mRNA and protein abundance in WT and transgenic plants. According to the RNA-seq data, transcripts in WT and OE lines have a similar mean abundance under normal conditions. However, low temperature strongly reduced cellular mRNA levels in WT or OE lines (Figure 6f). Besides, we found that the mean protein levels produced from OE lines were significantly higher than that in WT plants under both normal and cold-treated conditions based on the mean protein abundance (Figure 6g). These data indicated that CRIR1 may potentially coordinatilate with MeCSP5 may toincrease translational yield for coping with cold stress.