OxyR senses reactive sulfane sulfur and activates genes for its removal in Escherichia coli

Reactive sulfane sulfur species such as hydrogen polysulfide and organic persulfide are newly recognized as normal cellular components, involved in signaling and protecting cells from oxidative stress. Their production is extensively studied, but their removal is less characterized. Herein, we showed that reactive sulfane sulfur is toxic at high levels, and it is mainly removed via reduction by thioredoxin and glutaredoxin with the release of H2S in Escherichia coli. OxyR is best known to respond to H2O2, and it also played an important role in responding to reactive sulfane sulfur under both aerobic and anaerobic conditions. It was modified by hydrogen polysulfide to OxyR C199-SSH, which activated the expression of thioredoxin 2 and glutaredoxin 1. This is a new type of OxyR modification. Bioinformatics analysis showed that OxyRs are widely present in bacteria, including strict anaerobic bacteria. Thus, the OxyR sensing of reactive sulfane sulfur may represent a conserved mechanism for bacteria to deal with sulfane sulfur stress.

protecting cells from oxidative stress. Their production is extensively studied, but 23 their removal is less characterized. Herein, we showed that reactive sulfane sulfur is 24 toxic at high levels, and it is mainly removed via reduction by thioredoxin and 25 glutaredoxin with the release of H 2 S in Escherichia coli. OxyR is best known to 26 respond to H 2 O 2 , and it also played an important role in responding to reactive sulfane 27 sulfur under both aerobic and anaerobic conditions. It was modified by hydrogen 28 polysulfide to OxyR C199-SSH, which activated the expression of thioredoxin 2 and 29 glutaredoxin 1. This is a new type of OxyR modification. Bioinformatics analysis 30 showed that OxyRs are widely present in bacteria, including strict anaerobic bacteria. 31 Thus, the OxyR sensing of reactive sulfane sulfur may represent a conserved 32 mechanism for bacteria to deal with sulfane sulfur stress. mechanism is unclear, a recent study suggested that sulfur is transported into the cell 66 in the form of hydrogen polysulfide (Sato et al, 2011), inducing protein persulfidation 67 as a possible toxic mechanism (Islamov et al, 2018). Fungi may use glutathione to 68 reduce polysulfides to H 2 S as a detoxification mechanism (Samrat et  . Both bacteria and fungi show reduced viability being exposed to sulfane 72 sulfur stress (Sato et al, 2011;Xu et al, 2018). Therefore, intracellular sulfane sulfur is 73 likely maintained within a range for microorganisms under normal conditions. 74 Multiple pathways for sulfane sulfur generation have been discovered.  Xin et al, 2016). 80 Catalase can oxidize H 2 S and polysulfides to sulfur oxides (Olson et al, 2017). Most  To confirm the change of intracellular sulfane sulfur in vivo, we constructed a 141 transcription factor (TF)-based reporting plasmid, which contained a sulfane 142 sulfur-sensing TF (CstR) (Luebke et al, 2014), its cognate promoter (Pcst), and a red 143 fluorescent protein (mKate, with a C-terminus degradation tag ssrA) (Fig. 1C). Using 144 the reporting plasmid, the increase of intracellular sulfane sulfur in live cells (Fig. 1A) 145 was reported as the mKate fluorescence (Fig. 1D). When GrxB, GrxC, or GrxD was 146 co-transcribed with mKate under the control of CstR, their expression could partially 147 decrease the sulfane sulfur accumulation as reflected with the mKate fluorescence 148 intensity (Fig. 1D). When TrxA was co-transcribed, it did not affect sulfane sulfur 149 accumulation (Fig. 1D). However, when thioredoxin reductase (TrxB) was 150 co-expressed with TrxA, sulfane sulfur were not increased during the log phase of 151 growth (Fig. 1D). These results indicate that thioredoxin and glutaredoxin reduce 152 sulfane sulfur inside E. coli. 153 The artificial operons contained a negative feedback loop when coupled with an 154 enzyme that reduces sulfane sulfur (Fig. 1C&D). The loop effectively maintained 155 intracellular sulfane sulfur levels within a narrow range, defined by the leaky strength 156 of Pcst and the sensitivity of CstR as well as the controlled enzyme activity. Since 9 KatG under both aerobic and anaerobic conditions 163 We deleted oxyR gene in E. coli and observed that the mutant became more 164 sensitive to exogenously added H 2 S n under both aerobic or anaerobic conditions (Fig.   165 2A). After complementing oxyR into ΔoxyR, the strain reassumed the tolerance to 166 H 2 S n to the same level of the wild type (wt) (Fig. 2A). The results indicated that OxyR  conditions. In wt, all three promoters led to a low mKate expression in the absence of 186 H 2 S n but resulted in obviously higher expression when H 2 S n was added (Fig. 3A). 187 Whereas in the ΔoxyR strain, the three promoters led to constantly low expression of 188 mKate with or without H 2 S n stress (Fig. 3B). After introducing oxyR back to ΔoxyR, 189 the promoters performed the same as that in wt (Fig. 3C). Further, overexpression of 190 trxC, grxA, and katG in E. coli ΔoxyR decreases intracellular sulfane sulfur (Fig. S1).

191
The induction by H 2 S n was further conformed by in vitro transcription-translation 192 experiments. DTT or H 2 S n treated-OxyR and a DNA fragment containing the trxC 193 promoter and mKate (P trxC -mKate) were added into the cell-free 194 transcription-translation system. When DTT-treated OxyR (the reduced form) was 195 used, mKate expression was low. Whereas, when H 2 S n -treated OxyR was used, mKate 196 expression was significantly increased (Fig. 4). These results indicated that H 2 S n 197 induces the trxC promoter via directly modifying OxyR. We also test the induction by H 2 S n under anoxic conditions using qPCR as  To find out the molecular mechanism on how OxyR senses H 2 S n , mass 241 spectrometry analysis was performed to analyze the H 2 S n -treated OxyR. A short

247
Cys 208 was not modified by IAM indicating that it is not accessible to IAM, consistent 248 with a previous report that Cys 208 is buried in the protein (Kim et al, 2002). No 249 peptide containing both Cys 199 and Cys 208 was detected. These data collectively 250 indicated that OxyR senses H 2 S n via persulfidation on Cys 199 , other than forming 251 disulfide or -S n -(n≥3) bond between Cys 199 and Cys 208 .   and SodC, but did not affect the expression of CARS (encoded by cysS) (Fig. 8B).  (Fukuto et al, 2018). 304 GrxA, TrxC, and KatG were similarly upregulated by H 2 S n and H 2 O 2 stress; Trx A 305 and TrxB were not obviously affected by H 2 S n stress, but H 2 O 2 stress upregulated 306 them ( Fig. 8A and B). Overall, the transcriptomic data indicated that H 2 S n and H 2 O 2 307 stresses lead to largely different responses in E. coli. However, some proteins like 308 TrxC, GrxA, and KatG, are likely involved in alleviating both stresses.  (Table S1). Thus, OxyR is widely 321 distributed in bacteria. It is worth noting that OxyR is also present in many obligate 322 anaerobic bacteria, such as Bacteroides spp., Prevotella spp., and Porphromonas spp.    proteins are also analogous, i.e., protein-SSH vs protein-SOH (Mishanina et al, 2015). 360 From an evolutionary perspective, the former's history can be traced back before the we also observed that E. coli has obviously responding-discrepancies when 368 confronting H 2 O 2 or HSSH (Fig. 7). These discrepancies are in agreement with the 369 multi-level transcriptional responses of OxyR when activated by different reagents 370 (Haridas et al, 2005;Kim et al, 2002;Seth & Stamler, 2012). 371 In conclusion, we discovered that E. coli uses thioredoxins, glutaredoxins, and 372 catalase to control homeostasis of intracellular sulfane sulfur. OxyR functions as a 373 reactive sulfane sulfur sensor via persulfidation of its Cys 199 both under aerobic or 374 anoxic conditions. This is the fifth type modification observed for OxyR activation.

375
Since OxyR is widely distributed in both aerobic and anaerobic bacteria, the 376 OxyR-regulated network may represent a conserved mechanism that bacteria can     The H 2 S n -reacted OxyR (0.5 mg/ml) was mixed with iodoacetamide (IAM), and 469 then digested with trypsin by following a previously reported protocol (H et al, 2017). 470 The Prominence nano-LC system (Shimadzu) equipped with a custom-made silica   first-strand DNA fragments with ligated adaptors on both ends were selectively 508 enriched in a 10-cycle PCR reaction, purified (AMPure XP), and the library was 509 28 quantified using the Agilent High Sensitivity DNA assay on the Agilent Bioanalyzer 510 2100 system. The library was sequencing on Illumina Hiseq 2500 platform.

511
Sequencing was performed at Beijing Novogene Bioinformatics Technology Co., Ltd.

512
The clean data were obtained from raw data by removing reads containing adapter,