Arginine glycosylation enhances methylglyoxal detoxification

Type III secretion system effector proteins have primarily been characterized for their interactions with host cell proteins and their ability to disrupt host signaling pathways. We are testing the hypothesis that some effectors are active within the bacterium, where they modulate bacterial signal transduction and physiology. We previously determined that the Citrobacter rodentium effector NleB possesses an intra-bacterial glycosyltransferase activity that increases glutathione synthetase activity to protect the bacterium from oxidative stress. Here we investigated the potential intra-bacterial activities of NleB orthologs in Salmonella enterica and found that SseK1 and SseK3 mediate resistance to methylglyoxal. SseK1 glycosylates specific arginine residues on four proteins involved in methylglyoxal detoxification, namely GloA (R9), GloB (R190), GloC (R160), and YajL (R149). SseK1-mediated Arg-glycosylation of these four proteins significantly enhances their catalytic activity, thus providing another important example of the intra-bacterial activities of type three secretion system effector proteins. These data are also the first demonstration that a Salmonella T3SS effector is active within the bacterium.


Scientific Reports
| (2021) 11:3834 | https://doi.org/10.1038/s41598-021-83437-0 www.nature.com/scientificreports/ NleB is active and performs important biological functions within the bacterium, prior to its secretion. Here we investigated the potential intra-bacterial activities of the Salmonella SseK enzymes. We found that SseK1 is active within Salmonella enterica and that it glycosylates GloA, GloB, GloC, and YajL on arginine residues to significantly enhance their enzymatic activities and provide increased resistance to MGO.

Results
SseK1 and SseK3 affect methylglyoxal (MGO) resistance. We previously showed that C. rodentium NleB promotes bacterial survival in the presence of hydrogen peroxide by glycosylating the glutathione synthetase enzyme GshB 26 . By contrast, Salmonella is resistant to hydrogen peroxide due to the activity of redundant hydrogen peroxide scavengers 27 . Here we asked whether the NleB orthologs in Salmonella, SseK1, SseK2, and SseK3, might protect Salmonella from other forms of chemical stress. We compared the growth rates of Salmonella strains possessing or lacking SseK orthologs in the presence or absence of 1 mM MGO. Both the sseK1 and sseK3 mutants had a significant growth defect in the presence of MGO, despite having similar growth rates in the absence of MGO, whereas the growth rate of the sseK2 mutant was not significantly different as compared to the wild-type strain (Fig. 1). These data suggest a role for SseK1 and SseK3, but not SseK2 in MGO detoxification.
SseK1 glycosylates MGO detoxification enzymes. SseK1 and SseK3 are arginine-specific glycosyltransferases 18 and Salmonella strains lacking sseK1 or sseK3 are hypersensitive to MGO (Fig. 1). We tested the hypothesis that SseK1 and/or SseK3 glycosylate one or more Salmonella enzymes involved in MGO detoxification to regulate their activity. To do this, we cloned and expressed in Salmonella recombinant forms of the DJ-1 superfamily members YajL, Hsp31, YhbO, and ElbB, the glyoxalases GloA, GloB, and GloC, the l-glutathione lyase Lgl, the methylglyoxal reductase DkgA, the aldo-keto-reductase YdiH, and the putative aldehyde reductase YqhD. We then immunoblotted protein lysates for R-GlcNAc on each protein and observed that the type I glyoxalase GloA, the type II glyoxalase isomers GloB and GloC, and the deglycase YajL were glycosylated on an arginine residue in vivo ( Fig. 2A). We then repeated the glycosylation assays in Salmonella strains possessing or lacking each possible combination of SseK1, SseK2, and SseK3 and concluded that SseK1 glycosylates GloA, GloB, GloC, and YajL, as Argglycosylation was absent in any strain lacking SseK1 but was present independently of SseK2 and SseK3 (Fig. 2B). To further corroborate these data, we performed in vitro glycosylation assays using recombinant enzymes and substrates. All substrates were glycosylated by WT SseK1, but not by an SseK1 (HEN) mutant 18 which lacks glycosyltransferase activity (Fig. 2C).
SseK1-mediated glycosylation increases the activity of GloA, GloB, GloC, and YajL. We hypothesized that SseK1 glycosylation of GloA, GloB, GloC, and YajL might affect the respective activity of www.nature.com/scientificreports/ these enzymes, consistent with the sseK1 growth phenotype in MGO. We purified GloA, GloB, GloC, and YajL in their glycosylated or unglycosylated forms by co-expressing them (or not) with SseK1 (Fig. 3A, top), verified their glycosylation state using immunoblotting (Fig. 3A, bottom), and then used these recombinant proteins in enzyme assays. GloA is a type I glyoxalase that uses glutathione as a cofactor to convert hemithioacetal, which absorbs at OD 288 , into S-lactoylglutathione, which absorbs at OD 240 4 . We assayed the activity of GloA as a function of its glycosylation state and observed a significantly enhanced activity in comparison to the native enzyme ( Fig. 3B), as measured by a significant increase in the rate of hemithioacetal consumption. To corroborate these data, we also quantified the production of S-lactoylglutathione and observed a corresponding increase in S-lactoylglutathione production due to GloA glycosylation (Fig. 3C). We used GloB as a control in these assays and, as expected, observed no significant activity.
GloB and GloC are glyoxalase II isomers that catalyze the conversion of S-lactoylglutathione into d-lactate 28 . GloB is described as the major glyoxalase II, while GloC has a minor glyoxalase II activity, with maximal activity is reached by a combination of GloB and GloC 7 . We performed a glyoxalase II assay to compare the activity of individual or a combination of glycosylated and unmodified forms of GloB and GloC (Fig. 3D). As reported previously 7 , GloB was more active than GloC (Fig. 3D). SseK1-mediated glycosylation of GloB and GloC significantly enhanced their activity (Fig. 3D), as measured by a significant increase in the rate of S-lactoylglutathione consumption, as well a significant increase in the rate of d-lactate production (Fig. 3E).
YajL is a deglycase that repairs MGO-damaged proteins 8 . We measured the capacity of YajL to repair MGOmediated glycated GAPDH, by performing GAPDH activity assays as a function of YajL glycosylation state. The glycosylated form of YajL was significantly more active than the native form of YajL, as measured by an increase in the activity of repaired GAPDH as a function of YajL glycosylation (Fig. 3F). Taken together, these data show that SseK1-mediated glycosylation of the glyoxalases GloA, GloB, and GloC, as well as the deglycase YajL increases their enzymatic activities, which explains, at least in part, the reduced growth of the sseK1 mutant in comparison to the parental strain in the presence of MGO.
Glycosylation site mapping. We used mass spectrometry and site-directed mutagenesis to identify the glycosylation sites on each protein substrate. We determined by using mass spectrometry that SseK1 glycosylates GloB on R190, GloC on R160 or R165, and YajL on R149 (Fig. 4A). We corroborated the mass spectrometry data by performing R-GlcNAc specific immunoblots with wild-type and mutant proteins, confirming these sites and the modification of R160 on GloC (Fig. 4B,C). Although GloA was examined using both Lys-C and trypsin digestion, we were unable to achieve complete coverage of all arginine residues within this protein. Within GloA, multiple arginines are flanked by lysine residues, which, when cleaved by Lys-C or trypsin, result in short (< 6 AAs) peptides which are poorly amenable to LC-MS analysis. Because a GloA peptide could not be generated for MS/MS analysis, we instead used site-directed mutagenesis to localize the glycosylation site on this protein and found that SseK1 glycosylates GloA on R9 (Fig. 4D).

Discussion
This study was undertaken to investigate the potential role of Salmonella T3SS effectors in providing resistance to methylglyoxal and to determine whether the NleB orthologs SseK1, SseK2, and SseK3 are active within Salmonella. Our data show that GloA, GloB, GloC, and YajL are glycosylated on specific arginine residues by SseK. Such glycosylation enhances resistance to MGO and also enhances the repair of MGO-damaged proteins (Fig. 5).
Although both the sseK1 and sseK3 mutants had a growth phenotype in the presence of MGO (Fig. 1), we did not detect any Salmonella glycosylation targets for SseK3. Possible explanations include either the relatively small number of potential substrates we characterized in this study or the possibility that SseK3 might bind a bacterial substrate without glycosylating it, as seen for the host protein TRIM32 29 . We did not observe a role for the SseK enzymes in mediating Salmonella resistance to either hydrogen peroxide or glyoxal (data not shown). The importance of MGO detoxification to bacterial virulence has been reported in several other systems, including Burkholderia pseudomallei and Group A Streptococcus (GAS) 30   www.nature.com/scientificreports/ reductase activity 30 . Deleting GloA from GAS sensitizes GAS to MGO and increases its susceptibility to killing by human neutrophils 31 . In Salmonella, the lactoylglutathione lyase Lgl is known to also be important for MGO detoxification 11 , but we did not detect Lgl as an SseK1 substrate. Similar to our previous study with NleB-mediated enhancement of glutathione synthetase activity in C. rodentium 26 , it remains to be determined why Arg-glycosylation affects the activity of the GloABC and YajL enzymes in Salmonella. The glycosylated arginines are distant from the enzyme active sites 10,32,33 ; thus, activation might be the result of the post-translational modification mimicking an allosteric modification of the enzymes. Overall, these data reinforce the notion that T3SS effectors are active within bacterial cells and their activity appears to be tightly integrated with bacterial physiology, in this case, the ability to resist exogenous stressors. These data are also the first demonstration that a Salmonella T3SS effector is active within the bacterium.
Our work prompts several important questions that remain to be answered regarding the apparent dual activity of the NleB/SseK T3SS effectors within both the bacterium and the host cell. T3SS effectors are typically bound by a cognate chaperone to be guided to the injectisome complex 34 . This allows effectors to remain partially unfolded and competent for secretion through the secretion channel 35 . Some chaperones also prevent mis-targeting of effectors to bacterial membranes 36 and inhibit their degradation by proteases 37 . No chaperone for SseK1 or NleB has been identified and it is currently unclear to what extent chaperone binding, or the lack thereof, may explain the intra-bacterial activity of the SseK1 we observe here and in earlier studies 26 .
We previously evaluated whether the multi-cargo chaperone CesT affected C. rodentium NleB activity but found that NleB activity was independent of CesT 26 . Thus, it is unclear what role, if any, chaperones may play in regulating the folding state of SseK1 to allow the effector to function within Salmonella and then be secreted through the T3SS. If there is an NleB/SseK cognate chaperone, the effector may shuttle between folded (active) and unfolded (inactive) states as a function of its binding to the chaperone. It is conceivable that such an equilibrium might be affected by exogenous stressors (e.g. high MGO concentrations) requiring intra-bacterial effector activity. Such a scenario would likely require post-translational, rather than co-translational secretion of SseK1.
The percentage of NleB/SseK that is secreted and the percentage of NleB/SseK that remains within the bacterium is also unclear. While such concepts could theoretically be addressed by performing pulse-chase and single-cell imaging experiments, this issue remains beyond the scope of the experiments described here. Such host/bacterial effector ratios would conceivably be influenced by the environmental conditions the bacterium is facing, i.e. whether the bacterium is in a secretion-competent state. NleB expression is induced and secreted in conditions that permit the expression of the EspADB translocon 38 . In Salmonella, SseK1 is expressed in both SPI1-and SPI2-inducing conditions 39 , but is higher in SPI2-inducing conditions 40 . SseK2 and SseK3 are expressed in SPI2-inducing conditions and are dependent upon the SsrA/SsrB two-component regulatory system 20 .
It is clear from our studies here, as well as our previous study of C. rodentium NleB 26 , that the NleB/SseK orthologs function within both the bacterium and in the host cell after secretion through the T3SS. However, one potentially significant issue that may limit the broad impact of our data is that is generally unclear whether the activities of NleB/SseK are representative of the many other T3SS effectors. It is conceivable that NleB and other non-LEE encoded effectors might be less tightly transcriptionally regulated as compared to other effectors and might thus have a more constitutive expression profile. This class of effector may function normally inside the bacterium and their functions within host cells may be of secondary importance. In contrast, T3SS effectors whose expression is tightly coordinated with T3SS activation signals and whose secretion is strongly dependent upon cognate chaperones may indeed not be found to have intra-bacterial activity. To address such questions, we await the development of assays suitable for monitoring the potential intra-bacterial activities of other T3SS effectors with defined enzymatic activities.

Materials and methods
Plasmids, strains, and cloning. The plasmids and strains used in this study are listed in Tables 1 and 2, respectively. Wild-type sseK1 (Salmonella enterica) and its derivative H244A E255A N256A, were cloned into pET42a. Wild-type gloA, gloB, gloC, and yajL, as well as their arginine/alanine derivatives were cloned into a modified pET28a vector (in which the T7 promoter was replaced with the tac promoter) by using the ABC cloning method 41 . Gene deletions were constructed using lambda red recombination with the pKD3 and pKD119 plasmids 42 . Protein purification was performed as described previously 26 . Bacterial growth assays. One half of one percent of an overnight culture of Salmonella strains was used to inoculate 50 ml of LB medium in the presence or absence of 1 mM methylglyoxal. Bacterial growth was monitored for 16 h at 37 °C using an automated plate reader.
GAPDH activity assay. Bacterial GAPDH was incubated overnight at room temperature in a buffer containing 100 mM NaH 2 PO 4 pH 7.0, in the presence of 5 mM of MGO. Glycated GAPDH (200 nM) was then incubated with either 100 nM YajL, YajL-GlcNAc, or no enzyme. GAPDH activity was monitored using the GAPDH Activity Assay Kit (# MAK277, Sigma).
Glyoxalase I activity assay. Glutathione (5 mM) was incubated with 5 mM MGO for 1 h at room temperature in 100 mM NaH 2 PO 4 pH 7.0 to generate hemithioacetal. Hemithioacetal was then incubated with either 50 nM GloA or GloA-GlcNAc and both hemithioacetal consumption and S-lactoylglutathione production was measured as a function of time by monitoring OD 288 and OD 240 , respectively, in an automated plate reader.
Glyoxalase II activity assay. S-lactoylglutathione (1 mM) was incubated with 50 nM GloB, GloB-GlcNAc, GloC GloC-GlcNAc or combinations of either form of GloB and GloC in 100 mM NaH 2 PO 4 pH 7.0. Glyoxalase II activity was monitored by quantifying the amount of S-lactoylglutathione remaining as a function of time by monitoring OD 240 in an automated plate reader.
Digest of gel-separated proteins. Affinity Mass spectrometry data analysis. Identification of Arg-glycosylation events was accomplished using MaxQuant (v1.6.3.4) 44 . The predicted amino acid sequences for GloA, GloB, GloC, and YajL were combined into a database with the Salmonella typhimurium SL1344 proteome (Uniprot accession: UP000008962) and searched, allowing carbamidomethylation of cysteine set as a fixed modification and the variable modifications of oxidation of methionine and Arg-GlcNAcylation (H 13 C 8 NO 5 ; 203.0793 Da to Arginine). Searches were performed with either Trypsin or Lys-C cleavage specificity towards each protein sample, allowing 2 miscleavage events with a maximum false discovery rate (FDR) of 1.0% set for protein and peptide identifications. The resulting modified peptide output was processed within the Perseus (v1.4.0.6) 45 analysis environment to remove reverse matches and common protein contaminants. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 46 partner repository with the dataset identifier PXD021878.
Statistical analyses. Bacterial growth assays were analyzed using non-linear regression followed by Dunn's multiple comparison testing. Enzyme assays were analyzed using linear regression. P values < 0.05 were considered significant.