Analysis of light-dark transitions in a diurnal cycle reveal dynamic fluctuation in the Arabidopsis phosphoproteome
Protein phosphorylation is often associated with changing environmental conditions (Li et al., 2017; S. Zhang et al., 2019; Zhao et al., 2017). Therefore, we examined time-points before (30 min) and after (10 min, 30 min) the D-L and L-D transitions for changes in the phosphoproteome (Supplemental Figure 1). We identified 1776 phosphopeptides from 1091 proteins (phosphorylation site probability score ≥ 0.8) and quantified 1056 of these phosphopeptides from 725 proteins at the two light transitions (Table 1, Supplemental Table 3). We found that 176 phosphopeptides from 153 proteins at the D-L transition and 164 phosphopeptides from 144 proteins at the L-D transition had significant changes in abundance (Supplemental Figure 2 and 3; Supplemental Table 4). We then benchmarked the quality of our dataset by querying it for proteins known to be diurnally regulated by protein phosphorylation (Supplemental Table 5). This revealed phototropin 1, nitrate reductase (NIA1 and NIA2) and CF1 ATP synthase. Phototropin 1 is phosphorylated in the light (Sullivan, Thomson, Kaiserli, & Christie, 2009; Sullivan, Thomson, Lamont, Jones, & Christie, 2008), while the NIA1, NIA2 and the CF1 ATP synthase beta-subunit are phosphorylated in the dark (Kanekatsu, Saito, Motohashi, & Hisabori, 1998; Lillo, Meyer, Lea, Provan, & Oltedal, 2004; G. Moorhead et al., 1999; Reiland et al., 2009). Our quantitation of NIA1 and 2 protein phosphorylation changes across time-points revealed that NIA2 was more rapidly dephosphorylated on Ser534 at the D-L transition than NIA1, potentially relating to regulatory differences between NIA1 and 2. Additionally, we found a new NIA2 phosphorylation site at Ser63 with opposing diurnal changes in phosphorylation at the same transition (Supplemental Figure 4).
We next performed a GSEA of all significantly changing phosphoproteins (P value ≤ 0.01, FDR ≤ 0.05, gene set size ≥ 2) at each transition. Enriched biological processes at the D-L transition include phosphoproteins involved in light detection, nitrogen metabolism, cell wall-related processes and phosphorylation signaling, while phosphoproteins identified at the L-D transition are involved in light detection, vesicle-mediated transport, auxin signaling and nucleus organization (Table 2). We then generated a hierarchical heat map of the phosphopeptides to identify clusters of proteins at each light transition with similar phosphorylation dynamics (Supplemental Figure 2 and 3). When compared to datasets of phosphorylated proteins previously identified in Arabidopsis growing under free-running cycle conditions (Choudhary et al., 2015; Krahmer et al., 2019), or at the ED and EN time-points of a 12-hour photoperiod (Reiland et al., 2009; Uhrig et al., 2019), our data reveals proteins that have diurnal changes in their phosphorylation status and also novel rate-of-change information for these phosphorylation events (Supplemental Figure 2 and 3). For example, the L-D cluster I has phosphoproteins involved in nitrogen metabolism and the cell cycle (AD10 and AD30) and the L-D cluster III (BD30) has phosphoproteins involved in plastid organization (Supplemental Figure 2). In contrast, the D-L cluster II (AL10 and AL30) has phosphoproteins involved in central and carbohydrate metabolism (Supplemental Figure 3). Interestingly, parallel phosphorylation changes in L-D cluster I occur on proteins involved in nitrogen metabolism and the cell cycle. Nitrogen is acquired by plants primarily in the form of nitrate or ammonium, and is an essential macronutrient for plant growth. Nitrate signaling is linked to cell cycle progression through the TEOSINTE BRANCHED 1/ CYCLOIDEA/PROLIFERATING CELL FACTOR 20 (TCP20) – NIN-LIKE PROTEIN 6/7 (NLP6/7) regulatory network. TCP20 positively regulates genes encoding proteins involved in nitrate assimilation and signaling and downregulates the expression of CYCB1;1 , which encodes a key cell-cycle protein involved in the G2/M transition (Guan, 2017). Our data suggests that in addition to TCP20 transcriptional regulation, reversible protein phosphorylation may also play a role in this regulatory intersection between nitrate signaling and the cell cycle.
Similar to our analysis of protein abundance changes, we built association networks using STRING-DB to complement the GSEA analysis of the phosphoproteome (Figure 4). Association networks were generated based on phosphopeptide quantification data and in silicosubcellular localization information to examine relationships between the significantly changing phosphoproteins at both the D-L and L-D transitions. Most of the node clusters overlap between both the D-L (Figure 4A) and L-D (Figure 4B) networks, with larger clusters consisting of proteins involved in light detection and signaling, carbon and nitrogen metabolism, protein translation, hormone signaling, ion transport, cell wall related processes and protein phosphorylation. L-D transition-specific node clusters include RNA processing, transcription and secretion, and protein transport (Figure 4B). Similar to our proteome analyses, network association and GSEA analyses showed a high degree of overlap, indicating that the two approaches revealed the same cell processes in which proteins show differences in phosphorylation.
The dynamics of the measured Arabidopsis proteome during the diurnal period suggests that the proteins affected by abundance changes have significant functions in cellular processes. But as discussed above, protein abundance changes are generally not as widespread and extensive as transcriptome-level changes during the 24 h period. Comparison of our diurnal proteome time-series to reported transcriptome time-series substantiates data from earlier reports of ED vs EN studies examining Arabidopsis rosettes (Baerenfaller et al., 2012; Seaton et al., 2018; Uhrig et al., 2019) and circadian clock mutants (Graf et al., 2017) that transcript changes often do not coincide with protein abundance fluctuations (Baerenfaller et al., 2012; Graf et al., 2017). Changes in protein phosphorylation could be dependent or independent of protein abundance fluctuations. Our results show that the majority of significantly changing diurnal phosphorylation events occur independently of protein abundance changes, indicating that they most likely regulate protein function (Duby & Boutry, 2009; Le, Browning, & Gallie, 2000; Lillo et al., 2004; Muench, Zhang, & Dahodwala, 2012). Further research is required to elucidate the roles of the phosphorylation events on the Arabidopsis proteins we have identified. Based on our results it will be interesting to investigate which of the seemingly stable proteins / phosphoproteins and significantly changing phosphoproteins are in fact undergoing changes in their translation and turnover, but maintain their overall abundance (Li et al., 2017).