Large-scale identification of ubiquitination sites on membrane-associated proteins in Arabidopsis thaliana seedlings

An analysis of the identification of ubiquitination sites on proteins found at the cell periphery, including over 100 protein kinases.

Dear Editor, Protein phosphorylation and ubiquitination are two of the most frequently observed post-translational modifications in eukaryotes, regulated by thousands of protein kinases, phosphatases, E3 ubiquitin ligases, and ubiquitin proteases. Although previous studies have catalogued several ubiquitinated proteins in plants (Walton et al., 2016), few ubiquitinated membrane-localized proteins have been identified. Receptor kinases (RKs) initiate phosphorylation signal relays that regulate plant growth, development, and stress responses. While the regulatory role of phosphorylation on protein kinase function is well-documented (Couto and Zipfel, 2016), considerably less is known about the significance of ubiquitination on protein kinases, even though their turnover is critical to signaling competence and cellular homeostasis. Here, we describe the large-scale identification of ubiquitination sites on Arabidopsis (Arabidopsis thaliana) proteins associated with or integral to the plasma membrane, including over 100 protein kinases.

Letter
Accepted January 22, 2021. Advance access publication January 28, 2021 V C The Author(s) 2021. Published by Oxford University Press on behalf of American Society of Plant Biologists. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Proteomics and mutagenesis approaches have resulted in the discovery of several phosphorylated residues on BIK1 (Liang and Zhou, 2018). To help us understand the role of ubiquitination on BIK1 function, we set out to identify in vivo ubiquitination sites on BIK1. We enriched for plasma membrane-localized BIK1 by isolating microsomal protein fractions from Col-0/pBIK1:BIK1-HA, cpk28-1/pBIK1:BIK1-HA, and CPK28-OE1/pBIK1:BIK1-HA genotypes, which express 100-fold higher levels of BIK1 and differentially accumulate BIK1 protein compared to wild-type (Monaghan et al., 2014). To increase protein abundance of nonintegral proteins and allow us to potentially capture immune-induced ubiquitination, proteasomal machinery was inhibited with 50 lM MG-132 an hour before treatment with water or 1 lM elf18 (an immunogenic peptide derived from bacterial EF-Tu; Zipfel et al., 2006). Microsomal protein fractions were digested with trypsin, and anti-K-e-GG agarose beads (Udeshi et al., 2013) were used to enrich ubiquitinated peptides by affinity binding. Ubiquitinated lysines were identified based on a shift of $114 Da-the mass of two glycine remnants that remain covalently bound to lysines following trypsin digestion-using liquid chromatography followed by tandem mass spectrometry (Supplementary Methods). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2019) partner repository with the dataset identifier PXD021992 and 10.6019/ PXD021992.
We filtered our data for peptides with the diGly ubiquitin remnant, setting a threshold Mascot ion score of 420 and required multiple spectra for each peptide. This resulted in the identification of a total of 916 ubiquitinated peptides on 450 proteins across several biological replicates with a peptide false discovery rate of 0.025 (Supplemental Table S1), and an additional 526 peptides on 398 proteins observed in single experiments (Supplemental Table S2). Included in these data were seven ubiquitinated lysines on BIK1 (Table 1 and Figure 1; Supplemental Tables S1-S2). Given our particular interest in BIK1, we manually inspected all spectra mapping to BIK1 and found an additional three sites ( Figure 1; Supplemental Figure S1), altogether corroborating five of the ubiquitinated residues reported by (Ma et al., 2020) and revealing five novel ones ( Figure 1). Thus, BIK1 is ubiquitinated on multiple surface-exposed lysines in vivo: three in the N-terminal variable domain (K31, K41, K61), seven in the canonical kinase domain (K95, K106, K155, K170, K186, K286, K337), and five in the C-terminal region (K358, K366, K369, K374, K388; Figure 1). Whether RHA3A/B and PUB25/26 compete for these sites or ubiquitinate distinct lysines remains to be tested experimentally, as does clarifying which E2 conjugating enzymes work with respective E3 ligases to catalyze these events (Turek et al., 2018). Furthermore, as the phospho-status of BIK1 has been shown to affect its regulation by both RHA3A/B and PUB25/26 (Wang et al., 2018;Ma et al., 2020), another challenge will be resolving the biochemical mechanisms underlying this interplay.
Analysis of gene ontology (GO) terms associated with proteins identified in the high-confidence dataset (Supplemental Table S1) indicated an enrichment of proteins localized to the "plasma membrane" (p = 1.53 Â 10 -114 ; Supplemental Table S3). Because we analyzed the samples in the mass spectrometer in data-dependent mode, without quantification, we are unable to comment on differences between genotypes or immune treatments. Therefore, any immune-triggered events must be corroborated experimentally. Multiple sequence alignments of peptides spanning -10 to + 10 amino-and carboxyl-terminal to the modified lysines indicated very little consensus and no significant motifs (Supplemental Figure S2). Unlike other post-translational modifications, the ubiquitination reaction requires coordination between E1 activating, E2 conjugating, and E3 ligase enzymes (Vierstra, 2012). While it may be possible for individual E2-E3 pairs to exhibit residuelevel specificity on their target proteins, data from multiple species suggest that surface-availability may be the only unifying feature of ubiquitinated residues (Danielsen et al., 2011).
We identified ubiquitinated peptides mapping to proteins from diverse families, including aquaporins, H + and Ca 2 + ATPases, remorins, several classes of transporters, cellulose synthases, and others (Supplemental Tables S1-S2). Comparison between our dataset and eight published Arabidopsis ubiquitome datasets, as well as manual inspection of the literature, revealed 268 novel ubiquitin targets (Supplemental Table S4). We noted that molecular function GO terms "protein modification" (p = 1.79 Â 10 -12 ), "phosphorylation" (p = 2.15 Â 10 -26 ), and "response to stimulus" (p = 6.44 Â 10 -21 ) were particularly enriched in our dataset (Supplemental Table S3). Interestingly, we identified multiple ubiquitinated lysines on over 70 RKs representing diverse subgroups, including FLS2, EFR, CERK1, LORE, RLK7, SOBIR1/EVR, LIK1, RKL1, WAK1, WAK2, FER, ER, BAM1, BAM2, and others (Table 1). We also identified ubiquitination sites on more than 20 plasma membrane-associated cytoplasmic protein kinases from several subgroups (Table 1). Because analysis of tryptic peptides with ubiquitinated lysine residues enriched by anti-K-e-GG does not allow for discrimination between mono-or poly-ubiquitination, it is likely that we have captured both degradative and nondegradative ubiquitination on these protein kinases. Given the broad interest in phosphorylation-based signal transduction and protein homeostasis, we expect this information will be valuable to the plant research community and look forward to future studies that explore the function of these ubiquitination events. Table 1. Ubiquitinated protein kinases identified in this study. Proteins matching the gene ontology term "kinase activity" were filtered from Supplementary Tables S1 and S2 and classified based on phylogenies presented by Bleecker (2001, 2003). Residues that are only supported by a single observation (Supplementary Table S2) are indicated by an asterisk and should be interpreted with caution. Residues that were observed only after manual inspection of mass spectra matching BIK1 are indicated with two asterisks and shown in Supplementary Figure S1 Receptor-like protein kinases    Figure 1. BIK1 is ubiquitinated on multiple lysines in vivo. A, Comparison between this study and Ma et al. (2020) indicates that BIK1 is ubiquitinated on three lysines at its amino (N) terminus, seven in its kinase domain, and five at its carboxyl (C) terminus. Ubiquitinated lysines identified in Ma et al. (2020) are shown in green, those identified in this study are shown in blue, and residues identified in both studies are in magenta. The ATP-binding site (ABS), catalytic loop (CL), activation loop (AL), and P + 1 loop (PL) are indicated; the ABS is not surface-exposed, but the CL is shown in dark gray, the AL in white, and the PL in black. Although the structure of the BIK1 canonical kinase domain was recently solved , we modeled BIK1 in Phyre2-intensive mode (Kelley et al., 2015) in order to include the disordered N-and C-terminal ends in this surface representation in PyMol (The PyMol Molecular Graphics System, Version 2.0 Schrodinger, LLC). Phyre2-intensive modeling maximises sequence coverage and confidence to model regions for which there is no template information by an ab initio simplified-folding physics simulation; while 354/395 (90%) of the residues were modeled at 490% accuracy, it is likely that the model does not completely reflect the protein structure.

Supplemental data
The following materials are available in the online version of this article. Supplemental Methods. Methods used in this study. Supplemental Figure S1. Ubiquitinated residues identified on BIK1.
Supplemental Table S1. High-confidence peptides identified in multiple experiments.
Supplemental Table S2. Peptides identified in single experiments.
Supplemental Table S3. Gene ontology terms associated with proteins identified in this study.
Supplemental Table S4. Comparative analysis reveals 268 unique ubiquitin targets identified in this study.