Key plant processes involve independent changes in both proteome and phosphoproteome
We queried the data for all measured proteins that change in their abundance and/or phosphorylation status during the 24 h diurnal period. This revealed predominantly proteins involved in translation, metabolism and cell wall biosynthesis, suggesting that these cellular processes are subject to diurnal regulation at the protein level. The translation rates of Arabidopsis enzymes of light-induced metabolic reactions fluctuate diurnally and this correlates with their activity (Seaton et al., 2018). For example, several central metabolic enzymes are synthetized at 50 to 100% higher rates during the light phase of the photoperiod (Pal et al., 2013; Piques et al., 2009). We identified 15 proteins involved in protein translation that have diurnal changes in abundance (Table 3; Supplemental Table 2). Although they belong to several clusters shown in Figure 1A, nine of the proteins are grouped in CL3 that represents a general protein increase at the onset of light. In addition, we found 8 translation-related proteins with changes in their phosphorylation status at L-D and D-L transitions, of which 5/8 are eukaryotic initiation factor (eIF) proteins (Table 3; Supplemental Table 8). Phosphorylation affects eukaryotic translation at the initiation step (Jackson, Hellen, & Pestova, 2010; Le et al., 2000; Muench et al., 2012), and numerous eIFs and ribosomal proteins show differences in phosphorylation levels between light and dark periods (Boex-Fontvieille et al., 2013; Turkina, Klang Arstrand, & Vener, 2011; Uhrig et al., 2019). Our analysis revealed additional diurnally regulated eIFs and suggests that specific translational regulation mechanisms and ribosome composition could be controlled by light changes (e.g. day versus night) and also throughout the 24h photoperiod.
Several enzymes related to fatty acid, biotin, mitochondrial acetyl-CoA and chloroplast metabolism have diurnal changes in abundance (Figure 2; Table 3; Supplemental Table 2). Of particular interest are peroxisomal fatty acid β-oxidation enzymes 3-ketoacyl-CoA thiolase 2 (KAT2/PKT3; AT2G33150) and 3-hydroxyacyl-CoA dehydrogenase (MFP2/AIM1-like; AT3G15290). KAT2 is a central enzyme in peroxisomal fatty-acid degradation for the production of acetyl-CoA that is required for histone acetylation, which in turn affects DNA methylation (Wang et al., 2019), and ABA signaling (Jiang, Zhang, Wang, & Zhang, 2011), which is essential to daily regulation of stomatal conductance. MFP2/AIM1-like is an uncharacterized ortholog of MULTIFUNCTIONAL PROTEIN 2 (MFP2) and ENOYL-COA ISOMERASE (AIM1), which are involved in indole-3-acetic acid and jasmonic acid metabolism (Arent, Christensen, Pye, Norgaard, & Henriksen, 2010; Delker, Zolman, Miersch, & Wasternack, 2007). KAT2 loss-of-function mutants require sucrose to supplement plant acetyl-CoA production, suggesting that diurnal changes in fatty acid degradation through KAT2 and MFP2/AIM1-like are tied to sucrose production and that products downstream of KAT2 and MFP2/AIM1-like (e.g. hormones) are essential to plant growth and development (Pinfield-Wells et al., 2005). Previously, fatty acid and lipid metabolism in leaves and seedlings has been suggested to be diurnally / circadian clock regulated (Gibon et al., 2006; Hsiao et al., 2014; Kim, Nusinow, Sorkin, Pruneda-Paz, & Wang, 2019; Nakamura, 2018; Nakamura et al., 2014). This includes: diurnal changes in fatty acids and lipids (Gibon et al., 2006) in wild-type plants as well as diurnal changes in triacylglycerol (Hsiao et al., 2014) and phosphatidic acid (Kim et al., 2019) in the circadian clock double mutant lhycca1 . Complementing these studies, our findings provide a new protein-level understanding of where fatty acid and lipid regulation may rest which differs from our current transcript / metabolite based knowledge, indicating that further protein-level investigations are required.
We identified enzymes in primary metabolism that changed their phosphorylation status at the D-L and L-D transitions (Table 3; Supplemental Table 9). Several of these enzymes were previously identified as phosphorylated proteins (PhosPhat 4.0) (Heazlewood et al., 2008). Our sampling of three closely spaced time-points provides new information about the rate of protein phosphorylation changes at each transition. Moreover, our results demonstrate that in Arabidopsis metabolic enzymes are subject to changes in either protein abundance or phosphorylation, or both, which likely is of regulatory relevance for metabolic pathway flux. This information is useful when deciding which protein isoform would be best for engineering increased pathway flux if two are present simultaneously. For example, in case of the well-characterized NITRATE REDUCTASE 1 (NIA1; AT1G77760) and 2 (NIA2; AT1G37130) proteins their abundance parallels transcript levels but opposed to changes in protein phosphorylation, with NIA2 showing more rapid changes in phosphorylation at the D-L transition compared to NIA1 (Table 3; Supplemental Figure 4). NIA1 and NIA2 are regulated both transcriptionally and post-translationally by phosphorylation (Lillo, 2008; Lillo et al., 2004, Wang, Du, & Song, 2011). Our results define a rate of change in the phosphorylation of these related isozymes at the L-D and D-L transitions and define when peak NIA1 and NIA2 protein levels occur relative to peak transcript levels. These new insights help to better understand the precise functional differences between NIA1 and NIA2. NIA1 and NIA2 have tissue-specific gene expression profiles, with NIA1  expression generally complementing that of NIA2  in the same organ. NIA1 was predominantly found in leaves, while NIA2 was predominantly found in meristematic tissue (Olas & Wahl, 2019). We analyzed whole Arabidopsis rosettes before bolting, of which developing leaves and apical meristematic tissue comprises only a small amount of total tissue. Therefore, it will be interesting to determine if the observed difference in NIA1 and NIA2 phosphorylation rates at known regulatory phosphorylation sites reflect a higher sensitivity of NIA2 to changes in nitrate levels in meristematic and developing tissues (Olas et al., 2019).
Plant genomes often encode multiple forms of enzymes (isozymes) in metabolic pathways. The temporal rate at which related co-expressed protein orthologs are modified by a post-translational modifications such as protein phosphorylation provides more detailed information on cellular regulation. Our analysis of the phosphoproteome at three D-L and L-D time-points shows the dynamics of phosphorylation events at both transitions. Further information about the temporal rate at which different phosphorylation sites in a protein are utilized can then be combined with e.g., enzyme kinetics to reveal how metabolic flux through multiple enzyme reactions may be fine-tuned by PTMs versus changes in protein abundance. NIA1 and NIA2 are a good example that resolving differences in PTM rates helps to better understand the role of PTMs in protein regulation.
We also find diurnal changes in both protein abundance and protein phosphorylation for enzymes involved in carbohydrate metabolism (Table 3; Supplemental Table 2, 4). Starch biosynthesis and degradation is diurnally regulated to manage the primary carbon store in plants (Kotting, Kossmann, Zeeman, & Lloyd, 2010). For example, granule bound starch synthase 1 (GBSS1; AT1G32900) levels increase preceding the D-L transition, likely in anticipation of starch granule formation (Szydlowski et al., 2011). Debranching enzyme 1 (DBE1, AT1G03310) increases in abundance at the end of the light period to facilitate effective starch degradation in the dark (Delatte, Trevisan, Parker, & Zeeman, 2005). Other enzymes such as beta-amylase 1 (BAM1; AT3G23920) were phosphorylated immediately after the onset of light. Although the function of BAM1 phosphorylation is currently unknown, our results provide information to understand its regulation in stomatal starch degradation and sensitivity to osmotic changes in rosettes (Zanella et al., 2016).
Cell wall metabolic enzymes involve both diurnal fluctuations in protein abundance (Figure 2, Table 3; Supplemental Table 2) and changes in phosphorylation status (Figure 3, Table 3; Supplemental Table 4) at the D-L and L-D transitions. Cell wall biosynthesis is a major metabolic activity of growing plants (Barnes & Anderson, 2017; Cosgrove, 2005). We find that cellulose synthase enzymes CESA5 (AT5G09870) and CSLC6 (AT3G07330) were rapidly phosphorylated at the L-D transition. CESA5 has been shown to be phosphorylated and phosphorylation memetic-mutant enzymes increase movement of the cellulose synthase complex (CSC) in dark-grown seedlings, indicating a photoperiod-dependent regulation cell wall biosynthesis (Bischoff et al., 2011). Diurnal cellulose synthesis may also be controlled by the intracellular trafficking of CSC enzymes as a result of changes in metabolism (Ivakov et al., 2017). In dark-grown hypocotyls the ratio of CESA5 to CESA6 phosphorylation in the CSC complex is important for cellulose synthesis (Bischoff et al., 2011). Our phosphoproteome results now provide additional information on the rate of CESA5 phosphorylation at the onset of that dark period. We also find phosphorylation of the plasma membrane H+-ATPase HA1 (AT2G18960) at the L-D transition (Figure 3B). Phosphorylation activates H+-ATPases (Duby & Boutry, 2009; Sondergaard, Schulz, & Palmgren, 2004) and implicates HA1 as a primary candidate H+-ATPase in diurnal cell wall acidification to facilitate cell expansion during the night (Ivakov et al., 2017).