Aaron
Maurais
and
Eranthie
Weerapana
*
Department of Chemistry, Boston College, Chestnut Hill, MA 02467, USA. E-mail: eranthie@bc.edu
First published on 15th September 2022
The oxidation of methionine to methionine sulfoxide occurs under conditions of cellular oxidative stress, and modulates the function of a diverse array of proteins. Enzymatic systems that install and reverse the methionine sulfoxide modifications have been characterized, however, little is known about potential readers of this oxidative modification. Here, we apply a peptide-crosslinking approach to identify proteins that are able to differentially interact with reduced and oxidized methionine-containing peptides. Specifically, we generated a photo-crosslinking peptide derived from actin, which contains two sites of methionine oxidation, M44 and M47. Our proteomic studies identified heat shock proteins, including HSPA8, as selective for the reduced methionine-containing peptide, whereas the phosphofructokinase isoform, PFKL, preferentially interacts with the oxidized form. We then demonstrate that the favored interaction of PFKL with oxidized methionine is also observed in the full-length actin protein, suggesting a role of methionine oxidation in regulating the actin-PFKL interaction in cells. Our studies demonstrate the potential to identify proteins that can differentiate between reduced and oxidized methionine and thereby mediate downstream protein functions under conditions of oxidative stress. Furthermore, given that numerous sites of methionine oxidation have now been identified, these studies set the stage to identify putative readers of methionine oxidation on other protein targets.
Targeted, reversible Met oxidation can directly regulate protein function.6 For example, Met oxidation has been shown to regulate ROS sensing transcription factors in E. Coli,7 activation of CaMKII8 and the activity of calmodulin.9,10 Of particular interest is the role of Met oxidation in regulating actin dynamics. Actin filaments provide mechanical support, are the driving force for cell movement, and are critical for trafficking intracellular vesicles.11 Eukaryotic actin can exist in either a soluble form (G-actin) or filamentous form (F-actin). After the initial identification of Mical as a cytoskeleton-interacting protein in Drosophila,12 it was shown that Mical can catalyze the stereospecific oxidation of two Met residues in actin, M44 and M47, which are conserved across both vertebrates and invertebrates.13 Mical-catalyzed oxidization of M44 and M47 to Met-R-sulfoxide promotes actin filament disassembly.13,14 MsrB1 reduces both methionine sulfoxide residues and facilitates filament assembly.15
Given the characterized role of Met oxidation in regulating protein function, it is important to fully understand the mechanisms by which this modification is installed, removed and further translated to a downstream functional outcome. Although the role of Mical enzymes in oxidation, and MsrA/B enzymes in reduction of Met sulfoxide, little is known about potential ‘readers’ of the Met sulfoxide modification. We therefore aimed to identify proteins that have the ability to differentially interact with oxidized and reduced Met. One method for identifying ‘readers’ of PTMs is through the chemical synthesis of modified and unmodified peptides16,17 or large scale combinatorial peptide libraries18 to capture proteins which interact with a specific modification state in vitro. In addition, genetic code expansion approaches19 have facilitated the incorporation of PTMs in the context of live cells.20 For labile modifications that are unlikely to persist within the cell, non-hydrolyzable, or more stable analogs have been generated as a proxy for the native modification.21 In the case of Met sulfoxide, the ubiquitous presence of Mical and Msr enzymes will likely either non-specifically install the modification, or reverse the modification from the protein of interest when expressed in cells. Additionally, there is no analog of Met sulfoxide that is resistant to enzymatic reduction. Therefore, we applied a peptide based photo-crosslinking approach17 to identify proteins that differentially interact with Met and Met sulfoxide in vitro. Specifically, we generated a peptide-based photo-crosslinking probe derived from the well characterized Met sulfoxide sites in actin, and demonstrate that HSPA8 and PFKL selectively interact with the Met and Met sulfoxide peptides, respectively.
Fig. 1 Photo-crosslinking peptide probes to identify readers of methionine oxidation. (A) Structures of actin probes 1 and 2. (B) Probes 1 and 2 label MCF7 lysates in a concentration-dependent manner and show distinct labeling patterns. Uncropped gel image provided in Fig. S5 (ESI†). (C) Workflow used to identify proteins that interact selectively with oxidized or reduced methionine. 50 μM probes 1 or 2 are crosslinked in isotopically heavy or light MCF7 lysates. After UV crosslinking, equal amounts of heavy and light lysate are combined. Proteins that are bound to the probe are enriched on streptavidin agarose beads, trypsinized, and subjected to LC–MS/MS analysis. (D) 2D plot showing SILAC RL/H for each protein in the “forward” (x-axis) and “reversed” (y-axis) experiments. Proteins that interact preferentially with probe 1 are in the upper right portion of the plot and preferential interactors of probe 2 are in the bottom left portion. |
A significant limitation of using a peptide-based crosslinking approach is that the peptide can only duplicate the primary amino acid sequence in actin between H42 and K52. Secondary and tertiary structural elements will not be preserved. Therefore, protein interactions with the actin peptide probes may not necessarily translate to the full-length protein. To determine if these peptide-based probes can interact with proteins known to interact with this specific region of actin, we tested the oxidase activity of Mical1 against probe 1. Mical redox enzymes have been shown to induce actin disassembly through targeted oxidation of M46 and M49.13–15,23 Although MICALs have some basal level of NADPH consumption even in the absence of substrate,23,24 they are known to show an increased rate of NADPH consumption in the presence of F-actin.13,14,23,24 We treated the reduced peptide probe 1 with the purified catalytic domain of human Mical1, and monitored the appearance of the oxidized form of this peptide. We observed the formation of oxidized product by LC/MS (Fig. S2, ESI†), suggesting that although our peptide probes are not perfect representations of full-length actin, they do have the ability to interact with interacting proteins known to recognize that specific region of actin.
To test the cross-linking ability of probes 1 and 2, the probes were added to MCF7 cell lysates at varying concentrations. Upon irradiation of the probe-treated cell lysates, cross-linked proteins were visualized by biotin blot. Both probes displayed concentration-dependent covalent labeling of proteins within the MCF7 lysates. The majority of labeled protein bands were shared across the two probes, suggesting that the most interacting proteins were ambivalent as to the oxidation state of the Met residues within the peptide. However, a high intensity protein band was observed at ∼75 kDa in the probe 1 samples (Fig. 1B), suggesting the presence of proteins that could potentially interact selectively with either the Met or Met sulfoxide probes. We next sought to identify these protein interactors using a mass-spectrometry (MS) proteomics workflow.
HSPA8 and HSPA1B are members of the heat shock protein (HSP) family that serve a wide array of functions related to maintaining cellular protein homeostasis.25 These HSPs contain a central ATP/ADP binding site, and undergo a conformational change upon the hydrolysis of ATP. Although smaller HSPs are known to interact with actin,26 a direct interaction between HSPA8 and HSPA1B with actin has not been documented. Given that the molecular weights of HSPA8 and HSPA1B are within the ∼70 kDa range, the protein band visualized in the gel analysis (Fig. 1B) to be enriched in the probe 1 treated lysates is likely attributed HSPA8 and HSPA1B. HNRNPM is a heterogeneous nuclear riboprotein that regulates splicing thorough pre-mRNA binding, and PITRM1 is a mitochondrial matrix-localized metalloendopeptidase. Give that the cellular localization of these two proteins precludes interaction with actin and the cytoskeleton, it is unlikely that HNRNPM and PITRM1 are direct interactors with actin.
PFKL is the liver isoform of phosphofructokinase (PFK) and was found to selectively interact with the oxidized Met probe. We identified peptides that are common to all 3 PFK isoforms, however PFKL is the only family member that contained unique peptide sequences in our datasets. Glycolytic enzymes, including PFK, have long been known to directly interact with F-actin filaments, whereupon actin binding serves to increase the rate of glycolytic activity.27 Enrichment studies with antibodies against specific peptide sequences within actin, indicated that PFK can bind to actin at multiple distinct sites, including between residues spanning the peptide sequences used for our crosslinking studies.28 However, preferential interaction of PFK with oxidized actin has not been previously reported.
We focused our attention on HSPA8 and PFKL as putative proteins that selectively interact with reduced and oxidized Met-containing actin, respectively. To confirm our MS data and demonstrate selective interaction of HSPA8 and PFKL with reduced and oxidized Met, we treated cell lysates with each of the probes, performed the UV crosslinking and avidin enrichment. The enriched proteins were then eluted from the beads and assessed by western blotting to monitor the presence of HSPA8 and PFKL. In agreement with the SILAC-MS data, western blotting confirmed that HSPA8 selectively binds to probe 1 containing reduced Met residues. In contrast, PFKL preferentially bound to probe 2 containing oxidized Met residues (Fig. 2A). PFKL did show binding to the reduced Met probe as well, but displayed a 2.5-fold preference for the oxidized form.
Fig. 2 Validation of HSPA8 and PFKP as selective binders for Met and Met sulfoxide-containing peptide probes. (A) Detection of endogenous PFKL and HSPA8 in MCF7 lysates enriched with probe 1 or 2. MCF7 lysates were crosslinked with 75 μM of the indicated probe. An aliquot of the labeled lysate was taken and run as the “input” fraction. The labeled proteins were enriched on streptavidin-agarose beads. The bound proteins were eluted by heating and treatment with 50 μM biotin to obtain the “enriched” fraction. Each fraction was analyzed by western blot using antibodies for HSPA8 and PFKL. Streptavidin-HRP was used to detect probe-labeled proteins. Uncropped gel images provided in Fig. S6 (ESI†). (B) 12 μM of the indicated probe was combined with 1 μM purified HSPA8 and (C) PFKL in vitro. Each lane is a biological replicate performed with a new preparation of protein. Uncropped gel images provided in Fig. S7 (ESI†). (D) Mean densitometry measurements of 3 biological replicates shown in B and C. Band intensities from the streptavidin blot were normalized to the intensity of the HSPA8 or PFKL band and then normalized intensity of the most intense replicate, such that the intensity of the most intense replicate is set to 1. Replicates were statistically analyzed by Student's t test (*p < 0.005). Error bars represent the standard deviation of the 3 measurements. |
Upon confirming the selective interaction of probe 1 and 2 with HSPA8 and PFKP in cell lysates, we then proceeded to monitor the ability of probes 1 and 2 to bind to recombinant purified HSPA8 and PFKL. HSPA8 with an N-terminal HA tag was expressed and purified from E. coli. PFKL was expressed and purified from sf9 insect cells via baculovirus transduction.29 To verify that PFKL produced from sf9 cells was active, we performed a spectrophotometric coupled enzyme assay,30 which demonstrated that the sf9-produced PFKL was active with a Vmax of 2.45 × 10−3 s−1, and Km,F6P of 2.3 μM (Fig. S4, ESI†).
Upon obtaining purified HSPA8 and PFKL protein, crosslinking reactions with recombinant protein were performed with 1 μM HSPA8 or PFKL and a 12 molar excess of probes 1 or 2. We also included 0.2 mg mL−1 casein in the reaction mixture as a non-specific protein interactor. The purified protein and casein mixtures were treated with the probes and irradiated to induce the covalent crosslink. Covalent protein labeling of HSPA8 and PFKL by probes 1 and 2 was monitored by biotin blot. Western blotting for recombinant HSPA8 (anti-HA) and PFKL (anti-PFKL) confirmed equal protein loading in each lane. The experiment was performed in 3 biological replicates, and both probes showed a statistically significant preference for the expected protein target. Probe 1 showed a ∼6 fold selectivity for recombinant HSPA8 and probe 2 showed a ∼3 fold selectivity for recombinant PFKL (Fig. 2B–D).
Fig. 3 PFKL preferentially interacts with oxidized full-length actin. (A) Immunoprecipitation of ACTB from MCF7 lysates was followed by western blotting of the input (I) and bead-bound (E) fractions with anti-HSPA8, PFKL and GAPDH antibodies. Uncropped gel images provided in Fig. S8 (ESI†). (B) 25 μg biotin-phalloidin was added to MCF7 lysates and then incubated with streptavidin-agarose beads. The input (I) and bead bound (E) fractions were analyzed by western blot with anti-HSPA8, PFKL and GAPDH antibodies. Uncropped gel images are provided in Fig. S9 (ESI†). (C) 50 μg reduced or Mical oxidized G-actin was labeled with biotin-NHS then bound to streptavidin-agarose beads. 6 μM of recombinant HSPA8 and PFKL were combined and incubated with the immobilized actin. The bead bound fractions for reduced (M), oxidized (M(O)), and no actin (−) was analyzed by western blot using anti-PFKL and HA (HSPA8) antibodies. A representative of 3 biological replicates is shown. Uncropped gel images shown in Fig. S10 (ESI†). (D) Mean densitometry measurements of 3 biological replicates of the experiment shown in replicates were statistically analyzed by Student's t test (*p < 0.05, **p < 0.005). Error bars represent the standard deviation of the 3 measurements. |
To better understand how the oxidation state of ACTB affects HSPA8 and PFKL binding, reduced and oxidized ACTB was treated with recombinant HSPA8 and PFKL, and the levels of HSPA8 and PFKL that interact with ACTB were monitored upon immunoprecipitation of ACTB. Similar to the interaction we observed with the ACTB-derived peptide probes, we show that PFKL interacts with both reduced and oxidized ACTB, but interacts to a greater extent with the oxidized form (Fig. 3C and D). As before, we could not detect any interaction of HSPA8 with either reduced or oxidized ACTB, suggesting that HSPA8 only stably interacted wtih the truncated ACTB-based peptide probes and not the full-length protein.
In contrast to PFKL, although HSPA8 showed robust and selective labeling by probe 1, this interaction could not be recapitulated at the protein level. There are two possible reasons for this; (1) the interaction between HSPA8 and actin is transient and not strong enough to withstand immunoprecipitation; or, (2) HSPA8 only interacts with the actin-derived peptide probe, but not the full-length protein. Additional experiments are warranted to further characterize the biological consequences of methionine oxidation on the dynamics between HSPA8 and actin.
Together, our studies apply photo-crosslinking peptide-based probes to identify proteins that preferentially interact with reduced and oxidized methionine. Although similar strategies have been widely applied to numerous PTMs, methionine oxidation has not been previously investigated using this approach. We note several limitations of this approach to identify bona fide Met/Met sulfoxide interactors. First, as demonstrated for HSPA8, interaction with the peptide probes do not necessarily translate to interactions with the full-length protein. Additionally, Met sulfoxide exists as Met-S-sulfoxide and Met-R-sulfoxide, and it is unclear if reader proteins exist that can differentiate between both these isomers. In our studies we used a mixture of Met-S-sulfoxide and Met-R-sulfoxide, but resolution into the individual isomers may lead to other protein targets that show stereochemical preference for this oxidative modification. Lastly, it is likely that no general Met sulfoxide reader protein exists, and instead each unique Met sulfoxide site may have a dedicated reader protein associated with it. Therefore, a next step is to repeat these studies with peptide probes derived from other known functionally relevant sites of Met oxidation. Future studies will aim to address these limitations and broaden our understanding of how Met oxidation can regulate protein function through selective reader protein-mediated interactions.
Footnote |
† Electronic supplementary information (ESI) available: Experimental procedures, supplemental figures, and supplemental tables. See DOI: https://doi.org/10.1039/d2cb00183g |
This journal is © The Royal Society of Chemistry 2022 |