An increase in surface hydrophobicity mediates chaperone activity in N-chlorinated proteins

Under physiological conditions, Escherichia coli RidA is an enamine/imine deaminase, which promotes the release of ammonia from reactive enamine/imine intermediates. However, when modified by hypochlorous acid (HOCl), as produced by the host defense, RidAHOCl turns into a potent chaperone-like holdase that can effectively protect the proteome of E. coli during oxidative stress. We previously reported that the activation of RidA’s chaperone-like function coincides with the addition of at least seven and up to ten chlorine atoms. These atoms are reversibly added to basic amino acids in RidAHOCl and removal by reducing agents leads to inactivation. Nevertheless, it remains unclear, which residues in particular need to be chlorinated for activation. Here, we employ a combination of LC-MS/MS analysis, a chemo-proteomic approach, and a mutagenesis study to identify residues responsible for RidA’s chaperone-like function. Through LC-MS/MS of digested RidAHOCl, we obtained direct evidence of the chlorination of one arginine residue (and, coincidentally, two tyrosine residues), while other N- chlorinated residues could not be detected, presumably due to the instability of the modification and its potential interference with a proteolytic digest. Therefore, we established a chemoproteomic approach using 5-(dimethylamino) naphthalene-1-sulfinic acid (DANSO2H) as a probe to label N-chlorinated lysines. Using this probe, we were able to detect the N-chlorination of six additional lysine residues. Moreover, using a mutagenesis study to genetically probe the role of single arginine and lysine residues, we found that the removal of arginines R105 and R128 leads to a substantial reduction of RidAHOCl’s chaperone activity. These results, together with structural analysis, confirm that the chaperone activity of RidA is concomitant with the loss of positive charges on the protein surface, leading to an increased overall protein hydrophobicity. Molecular modelling of RidAHOCl and the rational design of a RidA variant that shows chaperone activity even in the absence of HOCl further supports our hypothesis. Our data provide a molecular mechanism for HOCl-mediated chaperone activity found in RidA and a growing number of other HOCl-activated chaperones.

of RidA is concomitant with the loss of positive charges on the protein surface, leading to an increased 48 overall protein hydrophobicity. Molecular modelling of RidAHOCl and the rational design of a RidA 49 variant that shows chaperone activity even in the absence of HOCl further supports our hypothesis. Our 50 data provide a molecular mechanism for HOCl-mediated chaperone activity found in RidA and a growing 51 number of other HOCl-activated chaperones. 52

INTRODUCTION 53
Phagocytosis is a crucial mechanism of our innate immune system used in the defense against and the 54 elimination of pathogens, such as bacteria and fungi. Within the phagosome, a cellular compartment 55 formed from the cytoplasmic membrane during phagocytosis, pathogens are exposed to a complex 56 mixture of different reactive oxygen and nitrogen species, including superoxide radicals, hydrogen 57 peroxide, peroxynitrite, and hypochlorous acid (for comprehensive reviews see Mortaz et al., 2018;58 Winterbourn et al., 2016). 59 Hypochlorous acid is produced by the heme enzyme myeloperoxidase from H2O2 and chlorine. HOCl is 60 a highly reactive oxidizing and chlorinating agent and one of the most potent oxidants that exist in human 61 cells (Albrich et al., 1981). It is known to cause a variety of modifications in virtually all cellular 62 macromolecules including DNA (Prütz, 1996), lipids (Winterbourn et al., 1992;Carr et al., 1996;63 Deborde and von Gunten, 2008), carbohydrates (McGowan and Thompson, 1989) and proteins. Amino OxyR (Storz et al., 1990;Zheng et al., 1998) or the redox-regulated chaperone Hsp33 (Jakob et al., 1999) 74 in E. coli. Both proteins become activated upon the formation of disulfide bonds and, conversely, 75

RESULTS 98
Direct identification of chlorinated sites in RidA after treatment with HOCl revealed chlorination of one 99 arginine and two tyrosines 100 We previously reported that the enamine/imine deaminase RidA from E. coli turns into a potent protein 101 holdase upon N-chlorination of its lysine and arginine residues. RidA, modified in this way (RidAHOCl), 102 can protect other cellular proteins from unfolding due to chlorine stress. Previous mass spectrometric 103 analysis of undigested HOCl-treated RidA showed that the reversible addition of at least 7 and up to 10 104 chlorine atoms to amino acid residues is concomitant with chaperone-like holdase activity in this protein 105 (Müller et al., 2014). However, it remains unknown, which residues are modified and which of those are 106 required for activation. If we consider arginine, histidine, and lysine as potential targets for a reversible 107 N-chlorination, RidA has 15 possible N-chlorination sites: 5 arginines, 1 histidine, 8 lysines, and the 108 protein's N-terminus ( Figure 1). 109

112
To elucidate which amino acids are affected by chlorination, we first performed a mass spectrometric 113 analysis of N-chlorinated RidA after proteolytic digest, specifically searching for peptides having an 114 added mass due to chlorination. Among the above-mentioned 15 possible N-chlorination sites, we were 115 only able to identify R51 to be chlorinated in any of the three replicates after digest with trypsin (Table  116 1, Figure 2). The monoisotopic mass of the chlorinated peptide containing R51 was ~ 33.96 Da heavier 117 than the unmodified peptide, corresponding to the monoisotopic mass of a chlorine atom minus the 118 monoisotopic mass of the hydrogen atom it replaced. The same tell-tale mass difference could be 119 observed in 9 fragment y-ions containing R51 ( Figure 2C). Additionally, we found two chlorinated peptides containing the two tyrosines present in RidA (Y17, Y91) (Table 1). Tyrosine chlorination by 121 HOCl is an irreversible modification that occurs with a much slower rate than the chlorination of lysine, 122 histidine and arginine (Hawkins et al., 2003). But once formed, chlorotyrosine is highly stable 123 (Winterbourn and Kettle, 2000; Hendrikje Buss et al., 2003). Our previous experiments showed that the 124 7 to 10 added chlorine atoms observed in full-length MS are virtually all removable by DTT, inconsistent 125 with chlorinated tyrosine residues. We thus concluded that our finding is, due to the high stability of the 126 modification in question, probably of auxiliary nature with no functional relevance. 127 We suspected that the low number of identified N-chlorinated residues in our experiment could be caused 128 by our use of trypsin. Trypsin, the protease most commonly used for the MS analysis of proteins, cleaves 129 proteins after lysine and arginine, the very residues which we suspected to be modified by HOCl. As 130 chlorination of these residues might interfere with trypsin's ability to recognize them, we additionally 131 prepared a digest with chymotrypsin. Nevertheless, we were still not able to detect more N-chlorinated 132 residues in our chymotryptic digest in any replicate but only re-identified chlorinated tyrosines Y17 and 133 Y91 (Table 1). 134

DANSO2H, a novel proteomic probe for N-chlorinated lysine residues in RidAHOCl 147
Our previous results showed that at least seven and up to ten residues become N-chlorinated in RidA, 148 when it is active as a chaperone. Thus, we were dissatisfied with our limited ability to detect N-149 chlorinated amino acids using mass spectrometry of proteolytic digests of HOCl-treated RidA. This 150 inability might be due to the inherent instability and high reactivity of N-chloramines. Therefore, we 151 devised a way to stably modify N-chlorinated amino acids. 152 To this end we utilized dansyl sulfinic acid (5-(dimethylamino)naphthalene-l-sulfinic acid, DANSO2H), 153 a derivative of the well-characterized reagent dansyl chloride (5-dimethy1amino)naphthalene-1-sulfonyl 154 chloride, DANSCl), which has been used for a long time to derivatize amines, amino acids, and proteins. 155 It reacts with free amino groups and forms stable, highly fluorescent sulfonamides, that can be separated 156 by HPLC and detected by mass spectrometry. In proteins, DANSCl usually reacts only with the free 157 amino group of lysine and the N-terminus (Hsieh and Matthews, 1985) ( Figure 3A). Unlike DANSCl, 158 DANSO2H is not reactive towards unmodified amino groups. Instead, it reacts with monochloramines, 159 forming the same sulfonamide product as the reaction of DANSCl with corresponding unmodified 160 amines ( Figure 3C). This provides us with a method for the selective labeling of N-chlorinated lysine 161 residues. DANSO2H was synthesized from DANSCl by a reaction with sodium sulfite (Scully et al., 162 1984). The synthesized molecule had a purity of 95 %. The only contamination present was dansyl 163 sulfonic acid (DANSO3H) ( Figure S2), which is chemically inert and does react neither with amino 164 groups nor N-chlorinated residues (Scully et al., 1984). The synthesized DANSO2H was then used to 165 modify RidAHOCl. DANSCl was used as a positive control to modify untreated RidA (RidAUT), and, as a 166 negative control, DANSO2H was incubated with RidAUT. As expected, after treatment with DANSCl, 167 RidAUT showed the typical fluorescence of dansyl-modified proteins ( Figure 3B). The same fluorescence 168 albeit to a lesser extent was observed in DANSO2H-treated RidAHOCl ( Figure 3D). The ~30 % lower 169 fluorescence intensity is consistent with our previous finding that only up to 10 out of 15 potential chlorination sites are chlorinated in fully active RidAHOCl. RidAUT treated with DANSO2H showed 171 virtually no fluorescence, demonstrating the specificity of our probe ( Figure 3D).    Since DANSO2H is most likely not suitable for the detection of N-chlorinated arginine residues, we used 209 a mutagenesis approach to test if individual amino acids have a particular impact on the chaperone 210 function of N-chlorinated RidA. We thus engineered a total of 13 RidA variants with every single lysine 211 or arginine residue exchanged to a serine. 212 These 13 RidA variants were expressed and purified from E. coli BL21(DE3) and their chaperone activity 213 after treatment with HOCl was then determined. Activity was tested in a 4-fold excess over citrate 214 synthase. The activity of the variants was compared against wild-type RidAHOCl under the same 215 conditions. The activity of most of HOCl-treated mutants was not affected in a significant way by the 216 exchange of a single N-containing amino acid to serine, when compared to HOCl-treated wild-type 217 protein ( Figure 5). However, two variants lacking specific arginines (R105S and R128S) showed a 218 significantly reduced chaperone activity when compared to HOCl-treated wild-type RidA (RidAWT).

230
Differences in chaperone activity of RidAWT between the variants were analyzed using a one-way ANOVA with

231
Tukey's comparison test (****, p < 0.0001). The activity of RidAWT was set to 100 %, and all the data are presented 232 in correlation to this control.

233
The concomitant exchange of R105 and R128 does not further decrease chaperone activity 234 Since an exchange of arginine 105 or 128 resulted in decreased chaperone activity, a variant harboring 235 both mutations (R105S_R128S) was constructed to investigate if a synergistic effect can be observed. 236 The variants were then also treated with HOCl and used in an 8-fold excess over citrate synthase. 237 However, the chaperone activity of the variant R105S_R128S remained at approximately the same level 238 as the single exchange variants and a further decrease of chaperone activity was not observed ( Figure 6). 239

246
Differences in chaperone activity between the variants were analyzed using a one-way ANOVA with Tukey's 247 comparison test (***, p < 0.001). The activity of RidAWT was set to 100 %, and all the data are presented in 248 correlation to this control.

249
The activation of RidA's chaperone-like holdase function depends on an overall change of the molecules 250 electrostatic surface. 251 Summed up, our proteomic, chemo-proteomic and mutagenesis studies suggest that no single amino acid 252 acts as a discrete "switch", but rather the modification of multiple N-containing amino acids leads to the 253 activation of RidA's chaperone-like holdase function. Even arginines 105 and 128, whose individual 254 exchanges to the inert amino acid serine had the largest effect on activatability, did not show synergistic 255 effects when both were removed. Chaperones are known for surface patches that enable them to interact 256 with hydrophobic regions of unfolded proteins. Indeed, our previous experiments showed that RidAHOCl 257 did bind a hydrophobic dye much better than RidAUT (Müller et al., 2014). In order to understand how 258 N-chlorination of basic amino acids changes the surface properties of RidA, we predicted the electrostatic 259 surface potential of RidA, using the known X-Ray structure of RidA (Volz, 2008) ( Figure 7A). We then 260 computationally substituted lysine and arginine residues with their chlorinated counterparts within the 261 structure data set. Using a customized force field, we were then able to predict the electrostatic surface 262 potential of an N-chlorinated RidA molecule as well ( Figure 7B). Strikingly, the electrostatic surface 263 potential shifted towards a more negative potential, more or less over the complete surface of the 264 molecule. We thus concluded that the loss of positive electrostatic surface potential is the underlying 265 molecular reason for RidAHOCl's chaperone-like properties. 266 An engineered variant of RidA mimicking RidAHOCl's surface potential is an active chaperone without 267

HOCl treatment 268
To test our hypothesis, we decided to engineer a variant of RidA that mimics RidAHOCl's surface 269 potential. To this end, we wanted to mutate all lysine and arginine residues in RidA to more 270 "electroneutral" amino acids. In order to select amino acids that would not lead to a disruption of RidA's 271 structure, we performed a multiple sequence alignment to direct our choice of suitable amino acids. For 272 our mutagenesis, we selected amino acids that occur at the respective position and are mostly uncharged 273 at physiological pH. With the exception of the invariant R105, the arginine in the active deaminase site 274 of RidA, we were able to find a suitable amino acid at all positions. This resulted in a variant, which we 275 termed R0K0_R105 (Figure 8). 276 The predicted electrostatic surface potential of this R0K0_R105 variant showed a high similarity to the 277 predicted surface potential of N-chlorinated RidAHOCl ( Figure 7C). 278   The K0R0_R105 variant was then expressed and purified from E. coli BL21(DE3) and its chaperone 302 activity in comparison to RidAUT and RidAHOCl was determined. As predicted, the RidA variant 303 K0R0_R105 showed already potent chaperone-like activity without HOCl pre-treatment, and treatment 304 with HOCl did not increase its chaperone activity significantly (Figure 9). Overall, the activity of 305 K0R0_R105 was indeed comparable to HOCl-treated RidA, supporting our hypothesis.

315
Taken together, our results suggest that chlorination of basic amino acids by HOCl is not a process that 316 targets one or two crucial amino acids that act as a "switch" that affects the whole protein. Instead, the 317 N-chlorination of multiple residues generates a more negatively charged and hydrophobic molecular 318 surface on RidA, allowing for the interaction with unfolded client proteins. This loss of positive surface 319 charge through N-chlorination is a plausible mechanism for the chaperone-like switch that occurs in a 320 growing number of proteins in response to exposure to HOCl and other reactive chlorine species. 321

DISCUSSION 322
After reacting with HOCl or monochloramine, RidA turns into a potent chaperone-like holdase that can 323 bind client proteins. The underlying mechanism was proposed to be the N-chlorination of lysine and 324 arginine residues (Müller et al., 2014). Since then, several other proteins have been discovered that seem 325 to switch to a chaperone-like function upon N-chlorination of basic amino acids. However, it was still 326 unclear if the chlorination of one or several specific residues is decisive for chaperone activity of RidA 327 or rather a general modification of multiple amino acid side chains that affects the overall surface of the 328

protein. 329
Four proteinogenic amino acids are known to be prone to chlorination (Peskin and Winterbourn, 2001).

Dansyl sulfinic acid can be used to stably modify N-chlorinated lysines 361
Searching for ways to chemically label N-chlorinated lysine, and thus making it accessible for MS-based 362 analysis, dansyl sulfinic acid (DANSO2H) caught our attention. It has been previously used to derivatize 363 low molecular weight monochloramines in water and other fluids, allowing for their detection by HPLC 364 (Scully et al., 1984(Scully et al., , 1986. We synthesized this probe to label the full-length RidA after its chlorination 365 and found that it exclusively reacts with the HOCl-treated protein. We were able to detect a robust 366 fluorescence signal once N-chlorinated RidA was treated with DANSO2H, but DANSO2H did not react 367 with the untreated protein. Therefore, DANSO2H allowed for the specific labeling of N-chlorinated 368 proteins. Using LC-MS/MS analysis of DANSO2H-derivatized RidAHOCl, we identified 6 lysine residues 369 that we were not able to detect using a direct MS-based approach. Interestingly, these 6 lysine residues 370 are also the lysine residues most exposed to the solvent (Table 3), suggesting they are particularly 371 accessible to HOCl. 372 is a chance that the dansyl chloride diffuses from the originally modified amino residue. The second 382 order rate constant of the reaction of dansyl chloride and the ε-aminogroup of lysine is 42 M -1 s -1 (Gray, 383 1967). Therefore, a two-step procedure with blocking of unmodified amino groups might be considered 384 for complex samples prior to labeling with DANSO2H. 385

Two RidA variants lacking either R105 or R128 showed a significantly decreased chaperone activity 386
To find out, whether specific amino acids play a key role in the activation of RidA's chaperone-like 387 function, all lysines and arginines were individually changed to serine residues through site-directed 388 mutagenesis. The individual variants were then examined regarding their chaperone activity. The 389 exchange of most lysines and arginines did not affect the chaperone activity of RidA. However, the 390 individual exchange of amino acid residues R105 and R128 resulted in diminished chaperone activity 391 after HOCl treatment. This could indicate that these arginines play a prominent role in chaperone activity. 392 Potentially, these two arginines are particularly accessible to unfolded proteins in N-chlorinated RidA 393 and, therefore, their chlorination might be important for client protein binding. R128 is the C-terminal 394 amino acid of RidA and, based on structural predictions, exposed to the solvent, which makes it 395 especially accessible to unfolded proteins (Table 3). In the bacterial cytoplasm, RidA is present in the 396 form of a trimer (Volz, 2008) and R128, R127, and K3 of every protein subunit form a large positively 397 charged surface patch (visible in the center of the molecule in the upper panel of Fig. 7A), which is 398 predicted to change to a more electroneutral or even negatively-charged patch after N-chlorination ( Fig.  399

B, upper panel) and thus might allow for binding of unfolded client proteins. 400
The other residue, R105, is highly conserved in three of the seven RidA subfamilies that have We also exchanged both arginines R105 and R128 simultaneously through site-directed mutagenesis. If 413 our argumentation that both arginines are of particular importance for client protein binding were true, 414 we would expect the double mutant to be an even less effective chaperone than the individual mutants. 415 Conversely, the chaperone activity of this variant corresponded approximately to the chaperone activity 416 of the individual exchanges. One explanation would be the distant localization of these two residues, 417 excluding a mutual influence on chaperone-like activity, even if both residues are absent. Alternatively, 418 the lower chaperone-like activity in both single and double mutants might not be due to the lack of an N-419 chlorination site but the general structural disturbance that an exchange of these residues (especially in 420 the active site of the protein) induces. 421 The latter possibility is further supported by the fact that the second-order rate constant of the reaction 422 of HOCl with arginines is three orders of magnitude lower than the one for the reaction with lysines (k 423 RidA molecule. Protein hydrophobicity as a driving force for chaperone activity is a known mechanism, 439 and other chaperones are also known to have exposed hydrophobic surfaces that allow for the binding of
The sample was digested at 37 °C for 18 h, and afterward, the peptides were purified using OMIX C18 474 tips. Purified peptides were concentrated to dryness and dissolved in 0.1 % TFA for LC-MS/MS analysis. 475

Peptide cleanup using OMIX C18 tips 476
The samples after tryptic or chymotryptic digest were first adjusted to 0.1 % (v/v) TFA using 10 % TFA 477 solution. Then the tips were first equilibrated 2 times with 100 µL 100 % ACN, followed by 2 times The availability of modification was monitored using "dansylation-sites.txt", "chlorination-sites.txt" and 505 "modificationSpecificPeptides.txt" output files. Data was imported and analyzed using Microsoft Excel 506 gene were exchanged using PCR-based mutagenesis, known as QuickChange PCR. PCR was performed 515 using 150 ng of pUC19_ridA or pEX_ridA-K0R0 as a template and 125 ng of each specific primer (Table  516 5) for every single exchange. 20 µL of the PCR product were digested with 20 U of DpnI at 37 °C for 517 1 h to eliminate the template plasmid. Subsequently, E. coli XL-1 blue cells were transformed with the 518 sample using a standard heat-shock method and plated on LB agar plates supplemented with 100 mg/ml 519 ampicillin. Plasmid DNA was isolated from single colonies, and successful mutagenesis was verified by 520 sequencing. Afterwards, ridA gene variants were subcloned into pET22b(+) expression vector via the 521 restriction sites NdeI and XhoI. The resulting pET22b(+)-based constructs were transformed into E. coli 522 BL21(DE3) cells for the subsequent overexpression of RidA variants. 523  expression was induced by the addition of 1 mM isopropyl 1-thio-ß-D-galactopyranoside (IPTG) to the 532 culture. After 3 h of incubation at 37 °C, 120 rpm, cells were harvested by centrifugation at 7800 x g and 533 4 °C for 45 min and either stored at -80 °C or directly used for the purification procedure. 534 The resulting cell pellet was washed once with lysis buffer (50 mM sodium phosphate, 300 mM NaCl, 535 10 mM imidazole, pH 8.0) containing 1 ml of EDTA-free protease inhibitor mixture (Roche Applied 536 Science). Cells were disrupted by passing the cell suspension three times through a Constant systems cell 537 disruption system TS benchtop device (Score Group plc, Aberdeenshire, UK) at 1.9 kbar and 4 °C, 538 followed by the addition of PMSF to a final concentration of 1 mM. 539 The cell lysate was centrifuged at 6700 x g, 4 °C for 1 h, and the supernatant was vacuum filtered through 540 a 0.45 μm filter. The filtrate was loaded onto a Ni-NTA affinity column. The column was washed with 541 10 mL washing buffer (50 mM sodium phosphate, 300 mM NaCl, 20 mM imidazole, pH 8.0). For 542 purification of the R0K0_R105 variant, washing buffer was additionally supplemented with 1 M NaCl 543 and 0.1 % SDS. Purified proteins were stored at -80 °C and K0R0_R105 was stored at room temperature. 544 Protein concentrations were determined using a JASCO V-650 spectrophotometer using the extinction 545 coefficient ε280=2,980 M -1 cm -1 . 546 Protein aggregation assay with citrate synthase 547 Citrate synthase was chemically denatured in 4.5 M GdnHCl, 40 mM HEPES, pH 7.5 at room 548 temperature overnight. The final concentration of denatured citrate synthase was 12 mM. 549 To monitor initial aggregation of citrate synthase, 20 µl of denatured protein were added to 1580 µl of 550 40 mM HEPES, pH 7.5 to a final concentration of 0.15 µM after 20 s of measurement. To test the 551 chaperone activity, RidA or RidA variants were added prior to the addition of citrate synthase to the 552 buffer at different molar access (0.5-16-fold) over dimeric citrate synthase. The increase in light 553 scattering was monitored for 240 s using a JASCO FP-8500 fluorescence spectrometer equipped with an 554 EHC-813 temperature-controlled sample holder (JASCO, Tokyo, Japan). Measurement parameters were 555 set to 360 nm (Em/Ex), 30 °C, medium sensitivity, slid width 2.5 nm (Em/Ex). Relative chaperone 556 activity of different RidA variants was calculated based on the difference between initial and final light 557 scattering. The chaperone activity of wild-type chlorinated RidA was set to 100 %. 558

Synthesis of N,N-Dimethyl-1-amino-5-naphtalenesulfinic acid (Dansyl sulfinic acid, DANSO2H) 559
Dansyl sulfinic acid (DANSO2H) was synthesized from dansyl chloride, as described in Scully et al.,560 1984 with minor modifications. Dansyl chloride (5 g) was added to a continuously stirred aqueous 561 solution of sodium sulfite (10.7 g in 50 mL) warmed to 70 °C. The reaction temperature was kept at 562 80 °C for 5 h. After the solution was cooled, DANSO2H was precipitated from the product mixture by 563 acidifying the solution to pH 4 with concentrated sulfuric acid. The precipitate was then filtered. The 564 precipitate was then dried in a vacuum desiccator over silica gel. The powder was re-dissolved in a cold, 565 aqueous solution of NaOH (2.8 M). The resulting solution was filtered and titrated to pH 4.0 using 566 sulfuric acid. The resulting DANSO2H was dried again in a vacuum desiccator and stored in the light-567 protected vial at 4 °C. The purity of the resulting dansyl sulfinic acid (retention time 6.48 min) was 568 confirmed by HPLC (Fig. S1) and was 95 %. The 5 % contamination by a corresponding sulfonic acid 569 RidA was chlorinated with HOCl as described above. Using an NAP-5 gel filtration column, residual 576 HOCl was removed, and the buffer was exchanged to 200 mM NaHCO3 buffer, pH 9.0. 250 µM 577 RidAHOCl or RidAUT were incubated with a 50-fold molar excess of DANSCl or DANSO2H for 1h, 37 578 °C, 1300 rpm. The derivatizing agent was removed using an NAP-5 gel filtration column. Successful 579 derivatization of RidA by monitoring the fluorescence of the resulting sulfonamide was determined by a 580 fluorescence emission scan from 360-600 nm in a JASCO FP-8500 fluorescence spectrometer with the 581 following parameters: 340 nm excitation, 2.5 nm slit width (Ex/Em) and medium sensitivity. 582 Preparation and chymotryptic digest of dansylated proteins for mass spectrometry analysis 583 5 µL (approx. 8 µg of protein) of each dansylated sample prepared above were mixed together with 25 µl 584 ultrapure water and 10 µl 5 x chymotrypsin digestion buffer (500 mM Tris-HCl, 10 mM CaCl2, pH 8.0). 585 pH of the resulting solution was monitored to be around 8.0. Then, DTT was added to the solution to a 586 final concentration of 10 mM and incubated at 60 °C for 45 min. After the sample was cooled to room 587 temperature, 3.5 µL of freshly prepared 500 mM iodoacetamide solution in pure water were added to a 588 final concentration of 20 mM. The sample was incubated at room temperature for 30 min, protected from 589 light. To quench the alkylation reaction, 1 ul of 500 mM DTT was added, and the volume of the solution 590 was adjusted to 50 µl using pure water. Chymotrypsin, reconstituted to a concentration of 1 mg/mL in 1 591 mM HCl, was then added to the sample for a final 1:20 enzyme-to-protein ratio. The reaction mixture 592 was incubated at 37 °C for 18 h. The sample was then desalted using OMIX C18 tips (Agilent 593 Technologies, USA) according to the manufacturer's instructions. Purified peptides were concentrated 594 to dryness and dissolved in 0.1 % TFA for LC-MS/MS analysis. 595 Protein structure modification and electrostatic surface modelling of RidA, RidAHOCl, and K0R0_R105 596 E. coli RidA crystal structure was accessed using PDB entry number 1QU9 and visualized using PyMOL 597 2.3.4 (Schrödinger, New York, NY, USA). This structure was converted to a PQR file using the 598 PDB2PQR webserver (Jurrus et al., 2018) under an assumed pH of 7.0, using PROPKA to assign 599 protonation and the AMBER forcefield to assign partial charges and atomic volumes to the structure's 600 atoms. Using the resulting PQR file as input, PyMol's APBS plugin was used to calculate the electrostatic 601 surface potential. 602 To model the N-chlorinated RidAHOCl, the PQR file of RidA was manipulated using a custom python 603 script (Supplemental Material S2). This script removes charge bearing protons from specified lysine and 604 arginine residues, rectifies binding angles (in case of deprotonated arginine) and replaces one nitrogen-605 attached hydrogen atom in these residues with a chlorine atom using the following bond geometries and 606 partial charges: Charge of the chlorine atom: 0.07, charge of the nitrogen atom: -0.61 (based on values 607 for CH3NHCl (Heeb et al., 2017)), radius of the chlorine atom 1.75 Å (Bondi, 1964), length of the N-Cl 608 bond: 1.784 Å (based on values for NH2Cl (Harmony et al., 1979)). The resulting manipulated PQR file 609 was then used to calculate the electrostatic surface potential of RidAHOCl using the APBS plugin of 610

PyMol. 611
The structure of RidA variant R0K0_R105 was modeled using the mutagenesis function of PyMOL. The 612 specified amino acids in wild-type RidA were exchanged as follows: 613 The electrostatic surface potential for this variant was then calculated as outlined for the wildtype 626 above (conversion of the pdb-file to a PQR file using the PDB2PQR webserver under an assumed pH 627 of 7.0, using PROPKA to assign protonation and the AMBER forcefield and subsequent use of the 628 ABPS plugin in PyMol). 629