Thiosulfate Sulfurtransferase like Domain Containing 1 Protein Interacts with Thioredoxin

Rhodanese domains are structural modules present in the sulfurtransferase superfamily. These domains can exist as single units, in tandem repeats, or fused to domains with other activities. Despite their prevalence across species, the specific physiological roles of most sulfurtransferases are not known. Mammalian rhodanese and mercaptopyruvate sulfurtransferase are perhaps the best-studied members of this protein superfamily and are involved in hydrogen sulfide metabolism. The relatively unstudied human thiosulfate sulfurtransferase-like domain-containing 1 (TSTD1) protein, a single-domain cytoplasmic sulfurtransferase, was also postulated to play a role in the sulfide oxidation pathway using thiosulfate to form glutathione persulfide, for subsequent processing in the mitochondrial matrix. Prior kinetic analysis of TSTD1 was performed at pH 9.2, raising questions about relevance and the proposed model for TSTD1 function. In this study, we report a 1.04 Å resolution crystal structure of human TSTD1, which displays an exposed active site that is distinct from that of rhodanese and mercaptopyruvate sulfurtransferase. Kinetic studies with a combination of sulfur donors and acceptors reveal that TSTD1 exhibits a low Km for thioredoxin as a sulfane sulfur acceptor and that it utilizes thiosulfate inefficiently as a sulfur donor. The active site exposure and its interaction with thioredoxin suggest that TSTD1 might play a role in sulfide-based signaling. The apical localization of TSTD1 in human colonic crypts, which interfaces with sulfide-releasing microbes, and the overexpression of TSTD1 in colon cancer provide potentially intriguing clues as to its role in sulfide metabolism.

Rhodanese domains are structural modules present in the sulfurtransferase superfamily. These domains can exist as single units, in tandem repeats or fused to domains with other activities. Despite their prevalence across species, the specific physiological roles of most sulfurtransferases are not known. Mammalian rhodanese and mercaptopyruvate sulfurtransferase are perhaps the best-studied members of this protein superfamily and are involved in hydrogen sulfide metabolism. The relatively unstudied human thiosulfate sulfurtransferase like domain-containing 1 (TSTD1) protein, a single-domain cytoplasmic sulfurtransferase, was also postulated to play a role in the sulfide oxidation pathway using thiosulfate to form glutathione persulfide, for subsequent processing in the mitochondrial matrix. Prior kinetic analysis of TSTD1 was performed at pH 9.2, raising questions about relevance and the proposed model for TSTD1 function. In this study, we report a 1.04 Å resolution crystal structure of human TSTD1, which displays an exposed active site that is distinct from that of rhodanese and mercaptopyruvate sulfurtransferase. Kinetic studies with a combination of sulfur donors and acceptors reveal that TSTD1 exhibits a low K M for thioredoxin as a sulfane sulfur acceptor and that it utilizes thiosulfate inefficiently as a sulfur donor. The active site exposure and its interaction with thioredoxin, suggest that TSTD1 might play a role in sulfide-based signaling. The apical localization of TSTD1 in human colonic crypts, which interfaces with sulfide-releasing microbes, and the overexpression of TSTD1 in colon cancer, provides potentially intriguing clues as to its role in sulfide metabolism.
Protein persulfidation, a posttranslational modification of cysteine residues, is postulated to be a major mechanism by which H 2 S signals (17). However, the mechanism by which target proteins acquire this modification and the sulfur source(s) used in persulfidation reactions are unclear (18). Sulfurtransferases like rhodanese or the single rhodanese domain containing TSTD1 could potentially play a role in protein persulfidation because of their capacity to form an active site cysteine persulfide during their reaction cycle (18).
The rhodanese superfamily members are distinguished by a structural module with an α/β topology in which α-helices surround a central five-stranded β-sheet core (19). At least three variations of the rhodanese domain are found in family members (20). In some members, only a single rhodanese domain is present such as in the bacterial GlpE (21) and human TSTD1 (22). Alternatively, the rhodanese domain can be present in a tandem repeat, as in rhodanese and MST, where the active sites are located in a cleft between the N-and C-terminal domains (14,15). Finally, rhodanese domains are fused to other protein domains as in Cdc25 phosphatase (20) or in the PRF and CstB proteins (23,24).
Sulfurtransferases are ubiquitously distributed enzymes that function in a range of pathways including sulfur metabolism, iron-sulfur cluster formation, selenium metabolism, and cyanide detoxification (13,14,(25)(26)(27)(28). In humans, there are five sulfurtransferases including mitochondrial rhodanese and MST, which is found in the cytoplasmic and mitochondrial compartments (29,30). Three isoforms of TSTD, 1, 2 and 3, are annotated in the human genome database. Of these, TSTD3 lacks a catalytic cysteine and is not predicted to function as a stand-alone sulfurtransferase. TSTD2 has the conserved active site cysteine, but its function is not known. TSTD1 (previously referred to as KAT) is found in the cytoplasm, in the vicinity of the nuclear membrane (31).
Rhodanese was initially described as a thiosulfate:cyanide sulfurtransferase (32)(33)(34). The catalytic cycle of sulfurtransferases involves two half reactions. In the first, the sulfur is transferred from a donor to an active site cysteine to form a cysteine persulfide (Cys-SSH) intermediate. In the second half reaction, the outer sulfur from the Cys-SSH intermediate is transferred to a thiophilic acceptor to regenerate the resting enzyme (34,35). Human rhodanese preferentially catalyzes sulfur transfer from GSSH to sulfite forming thiosulfate and GSH (eq. 1), and exhibits a >2 ×10 5 -fold preference for this, over the reverse reaction (14,15).
Residues on an active site loop containing the cysteine at which the Cys-SSH intermediate is formed are predicted to confer substrate specificity (Fig. 1A). In rhodanese, the positive charges at the proximal arginine and lysine residues in the CRKGVT motif are predicted to interact with the negative charges on the substrate (e.g. oxygen atoms in the case of thiosulfate). The uncharged serine residue in the corresponding MST motif CGSGVT, participates in a hydrogen-bonding network with two other arginines (at positions 188 and 197) that includes the carboxyl groups of pyruvate (13,36). In fact, the dual replacement of Arg248 and Lys249 in rhodanese with Gly and Ser, respectively to mimic the MST active site, increased the K M value for thiosulfate (~17-fold) and decreased the k cat for rhodanese 6-fold (36).
The structure of the Burkholderia phytofirmans PRF, a persulfide dioxgenaserhodanese fusion protein, displays a unique βhairpin extension in its single rhodanese domain (24). The sequence motif in the active site loop is CRAGGR. PRF preferentially catalyzes a sulfur transfer reaction from thiosulfate to GSH forming sulfite and GSSH, i.e. in the direction opposite to that preferred by rhodanese. (eq. 1). GSSH is a substrate for the fused persulfide dioxygenase domain in PRF. This difference in reaction specificity between rhodanese and the singlerhodanese domain containing PRF raises the question as to whether it is more broadly relevant to other single versus double rhodanese domaincontaining proteins.
While MST and rhodanese have welldescribed roles in sulfide biogenesis and oxidation pathways (13,14), the role of TSTD1 is poorly 3 characterized. Recombinant human TSTD1 utilizes thiosulfate as a sulfur donor, although the kinetics of this reaction were only characterized at the alkaline pH of 9.2 (22). Based on this activity, the involvement of TSTD1 in the sulfide oxidation pathway was proposed, although the remaining enzymes in this pathway exist inside the mitochondrion.
In this study, we report a high-resolution crystal structure and steady-state kinetic characterization of TSTD1 at a physiologically relevant pH. These studies suggest that TSTD1 is more likely to play a role in sulfide signaling than in the sulfide oxidation pathway.

Results
Expression and purification of TSTD1-A synthetic cDNA codon-optimized for E. coli expression of human TSTD1, was obtained from GenScript. The recombinant protein was purified as described previously (22) and obtained in a yield of 20 mg/liter of culture. TSTD1 was judged to be >95% pure by SDS-PAGE analysis (Fig.  1B). TSTD1 eluted from a Superdex 200 column with an apparent molecular mass ~12 kDa, consistent with being a monomer in solution (not shown).
Crystal structure of TSTD1-The crystal structure of TSTD1 was obtained at 1.04 Å resolution by single-wavelength anomalous diffraction phasing with selenomethionine ( Table  1). The final model contains two chains in the asymmetric unit. Each molecule consists of a fivestranded parallel β-sheet core framed by six αhelices ( Fig. 2A). Two of these β-strands are notably short, consisting of only two residues each. Comparison of TSTD1 with the Saccharomyces cerevisiae homolog RDL1 shows that they display similar folds (Fig. 2B). TSTD1 also shares a similar fold with other rhodanese domain-containing proteins. However, some minor differences are observed. TSTD1 displays shorter helices, namely α2, α4, and α6 compared to RDL1 ( Fig. 2A, B) (19). Another point of difference is the presence of an additional α-helix, α3 in TSTD1, which is absent in RLD1.
The active site cysteine residue (Cys-79) in TSTD1 is located at the bottom of a shallow pocket. In contrast, in rhodanese and MST, the corresponding cysteines are sequestered in deep clefts (Figs. 3A). The architecture of the TSTD1 active site loop is superficially similar to that of rhodanese and MST (Fig. 3B). Specifically, the orientation of the conserved cysteine and the backbone amide groups are similar in the three proteins. However, a closer inspection reveals that substitutions in the loop residues result in stereoelectronic differences between the three sulfurtransferases and likely reflect differences in their substrate preferences (Fig. 3B). In TSTD1, polar but uncharged residues Gln-80 and Met-81 replace the bulky, and positively charged Arg-248, Lys-249 in bovine rhodanese and Gly-249 and Ser-250 in human MST. The positively charged and bulky residues Lys-83 and Arg-84 in TSTD1 replace valine and threonine residues found in rhodanese and MST in the corresponding positions.
Thiosulfate:cyanide sulfurtransferase activity-The capacity of TSTD1 to detoxify cyanide via sulfur transfer from thiosulfate was assessed. From the dependence of the reaction rate on substrate concentration, K M values of 0.27 ± 0.02 mM for cyanide and 22 ± 3 mM for thiosulfate were obtained (Fig. 4A, B and Table 2). The specific activity of the protein at saturating substrate concentrations was 2.6 ± 0.1 µmol min -1 mg -1 protein at 37 °C corresponding to a k cat of 0.52 s -1 .
Thiosulfate:thiol sulfurtransferase activity-The efficiency of TSTD1 in catalyzing sulfur transfer from thiosulfate to the low molecular weight thiol acceptors, GSH, L-cysteine and Lhomocysteine was assessed (Fig. 5, Table 2). The K M values for the thiols acceptors ranged from 11 ± 1 mM (GSH) to 13.7 ± 1.9 mM (cysteine) while the specific activity ranged from 2.0-3.5 µmol/min/mg of protein at 37 ˚C (Fig. 5A,C,E). Substrate inhibition with thiosulfate was not observed with TSTD1 ( Fig. 5B,D,F) as seen previously with rhodanese (14). The K M for thiosulfate ranged from 14-18 mM in the presence of varying acceptors.
Thiosulfate:thioredoxin sulfurtransferase activity-Unlike rhodanese, TSTD1 can utilize thioredoxin as an acceptor in the presence of thioredoxin reductase and NADPH (Fig. 6, Table  2). While the K M for thiosulfate (K M = 22 ± 3 mM) is high as in the assays described above, the K M for thioredoxin is low (K M = 17 ± 2 µM). The k cat /K M(Trx) is ~175-fold greater than k cat /K M(GSH) ( Table 2).
Persulfide transfer to thioredoxin. To validate transfer of the sulfane sulfur from TSTD1 to thioredoxin, the biotin thiol assay (BTA) was used. This assay distinguishes between persulfides and other oxidative cysteine modifications such as sulfenic acid (37). For this, wild-type thioredoxin and two cysteine mutants, i.e. the catalytic cysteine C32A, and the resolving cysteine C35S, were used (Fig. 7A). Biotin-maleimide alkylation of thiols and persulfides led to immobilization of the proteins on a streptavidin column. DTT treatment reduced the mixed disulfide bonds in the (initially) persulfide-containing proteins and led to their elution from the column (Fig. 7B). Western blot analysis revealed that the presence of wildtype thioredoxin and the C35S mutant but not of the nucleophilic C32A mutant (Fig. 7C). These results are consistent with TSTD1-dependent persulfidation of thioredoxin at the nucleophilic Cys-32 residue. In the absence of DTT, none of the thioredoxins eluted from the column.
GSSH:sulfite sulfurtransferase activity-Since the preferred sulfur transfer reaction catalyzed by rhodanese is from GSSH to thiosulfate, the efficacy of TSTD1 in this reaction was assayed. However, unlike human rhodanese, TSTD1catalyzed sulfur transfer from GSSH to sulfite was undetectable under steady-state turnover conditions (Table 2). To investigate whether product inhibition could account for the observed behavior of TSTD1 in this assay, the activity was measured under single turnover conditions (Fig.  8A). Under these conditions, sulfite consumption was stoichiometric within experimental error, with thiosulfate formation. To further investigate product inhibition, the single turnover experiment was repeated in the presence of 2× or 5× thiosulfate. Under these conditions, sulfite consumption was inhibited (Fig. 8B).
Immunohistochemical localization of TSTD1 in colon-Since the colon is routinely exposed to high concentrations (1-2.4 mM) of sulfide produced by gut microbiota (38), we sought to examine the localization of TSTD1 in this tissue. TSTD1 antibody was validated using purified recombinant TSTD1 protein and verified in murine liver and kidney lysates (data not shown).
Staining for TSTD1 was observed in colon epithelium with stronger staining seen at the luminal surface region compared to the crypt base region (Fig. 9A). Increased staining for TSTD1 was observed in seven colon cancer tissues compared to normal colon epithelium (Figs. 9B, C). Overall, the H-scores were significantly higher (p<0.001) in colon cancer versus normal colon epithelium (Fig. S1).

Discussion
The sulfurtransferase superfamily is enigmatic. For instance, eighteen sulfurtransferases are annotated in the Arabidopsis thaliana genome but their specific functions are largely unknown (39). In mammals, the roles of the two sulfurtransferases, MST and rhodanese, involved in sulfide metabolism, are relatively well studied. Although the role of rhodanese was originally described in cyanide detoxification, its high K M for cyanide (29 mM) makes the physiological relevance of this reaction dubious. Instead, it is more likely that rhodanese catalyzes sulfur transfer from GSSH to sulfite forming GSH and thiosulfate and is a component of the mitochondrial sulfide oxidation pathway (14). MST transfers sulfur from mercaptopyruvate to thioredoxin, generating H 2 S (13). The potential for MST and rhodanese to transfer sulfur to other protein acceptors has not been rigorously evaluated.
Like rhodanese and MST, TSTD1 also catalyzes sulfur transfer to cyanide. It exhibits a significantly lower K M for cyanide (0.27 mM) in comparison to rhodanese (29 mM) and MST (6 mM). The catalytic efficiency (k cat /K M ) of TSTD1 for sulfur transfer from thiosulfate to cyanide is ~16-fold lower than for rhodanese but ~5-fold higher than for MST (for sulfur transfer from mercaptopyruvate to cyanide) (13,14). These kinetic data beg the question as to whether TSTD1 could be more proficient in detoxifying cyanide with a different (i.e. preferred) sulfur donor, whose identity remains to be established.
Since the physiologically relevant reaction catalyzed by TSTD1 is unknown, we have examined its activity with a variety of substrates. Kinetic analysis of TSTD1 at the physiologically relevant pH of 7.4, reveals that it can catalyze a variety of sulfur transfer reactions, albeit with low efficiency. Of the sulfur acceptors tested, human TSTD1 exhibits the lowest K M for thioredoxin (18 µM), which is ~650-1000-fold lower than for the physiologically relevant low molecular weight thiol acceptors (Table 2). We note that the K M for GSH while high (11 ± 1 mM), is within a physiologically relevant concentration range for this thiol, which varies from 1-10 mM in tissues (40,41). Thioredoxin is also reported to be an efficient acceptor (K M = 34 µM) for the E. coli single rhodanese domain sulfurtransferase GlpE in the presence of thiosulfate as the sulfur donor (42). Rhodanese from Azotobacter vinelandii reportedly uses thioredoxin as a sulfur acceptor with 2propenyl thiosulfate, an organosulfur compound extracted from garlic (43), as a sulfur donor. However, we were unable to detect activity of human rhodanese with thioredoxin and thiosulfate as donor (data not shown). BTA analysis of thiosulfate-dependent persulfidation catalyzed by TSTD1 demonstrated that persulfidation of thioredoxin requires the catalytic cysteine, Cys-32. Mutation of the resolving Cys-35 in the same CXXC motif, did not affect persulfide transfer to thioredoxin (Fig. 7C).
In colon, thiosulfate is the major product of the sulfide oxidation pathway (44). The efficiency of thiosulfate production by rhodanese from GSSH and sulfite (k cat /K M(GSSH) = ~1 ×10 6 M -1 s -1 ) is consistent with its role in the sulfide oxidation pathway (14). Unlike rhodanese, TSTD1 does not catalyze detectable sulfur transfer from GSSH (Table 2), which is due in part, to product inhibition by thiosulfate (Fig. 8B). This product inhibition behavior of TSTD1 is similar to that of the PRF protein, which has a single rhodanese sulfurtransferase domain fused to a persulfide dioxygenase (24). In contrast, the Staphylococcus aureus CstB, which has a persulfide dioxygenase fused to a tandem repeat of rhodanese domains, catalyzes sulfur transfer from coenzyme A persulfide or bacillithiol persulfide to sulfite as an acceptor (23).
Even in the opposite direction, i.e. sulfur transfer from thiosulfate to GSH, TSTD1 is an inefficient catalyst in comparison to rhodanese. Although the specific activity and K M value for GSH are not dramatically different in TSTD1 compared to rhodanese (Table 2), the K M value of TSTD1 for thiosulfate (13 mM) is 38-fold higher than for rhodanese (14). These kinetic data suggest that TSTD1 is unlikely to play a role in the sulfide oxidation pathway using thiosulfate as a sulfur donor as previously postulated (Fig. 10A).
Our conclusion that the sulfur donor for TSTD1 is unlikely to be among the ones that we, and others, have previously tested is supported by its crystal structure. Comparison of the active site loop residues in TSTD1 versus rhodanese and MST (Fig. 3B) reveals significant stereoelectronic differences. Arg-248 and Lys-249, which are postulated to play a role in substrate and Cys-SSH stabilization in rhodanese are replaced by Gln-80 and Met-81 in TSTD1 while similarly charged residues (Lys-83 and Arg-84) in TSTD1 are located at the opposite end of the active site loop within an ~5 Å distance from the active site cysteine. Ser-250 in MST, which is postulated to play a role in substrate binding, is replaced by Met-81 in the TSTD1 active site.
Comparison of the electrostatic potential surfaces of rhodanese, MST and TSTD1 reveals no notable differences (Fig. 3A). Neutral and positively charged residues frame the active site entrance in each protein. However, while the active site of TSTD1 is relatively surface-exposed, it resides in a deep cleft in rhodanese and MST, between the two rhodanese domains (Fig. 3A, B). The differences in the access to and geometry of the active site pockets likely reflect differences in the substrate specificities for these three sulfurtransferases. For instance in rhodanese, which can catalyze many of the same reactions as TSTD1, the deeply set active site pocket likely promotes interactions with low molecular weight substrates like GSSH whereas in TSTD1, the shallow active site pocket might promote interactions with larger substrates, namely proteins. In addition, the sequestered active sites in rhodanese and MST likely contribute to stabilizing the Cys-SSH intermediate, which has been seen in their crystal structures (13,45).
Thioredoxin is postulated to play a role in H 2 S signaling and inhibition of thioredoxin reductase is associated with increased intracellular persulfide levels (46,47). The concentration of thioredoxin is reportedly ~8-10 µM in liver (48) supporting the plausibility of its interaction with TSTD1. In principle, persulfidated thioredoxin can serve as a donor to TSTD1 while we have shown that the thiol form can serve as a sulfur acceptor. We speculate that other single rhodanese domaincontaining proteins like TSTD1 might also have relatively accessible active sites and serve as intermediaries in sulfur transfer reactions involving other proteins (Fig. 10B). The single domain rhodanese proteins GlpE, ThiI and RhdA also exhibit sulfur transfer activity with thiosulfate as a donor (42,49,50). For example, ThiI reportedly carries sulfur between IscS and tRNA. Using [ 35 S]-cysteine, radiolabel transfer from IscS to Tus was detected (51,52). Indeed, IscS transfers sulfane sulfur to TusA, which is then sequentially transferred to TusB/TusC/TusD/TusE and then to tRNA (51)(52)(53). The possible involvement of TSTD1 in a persulfide relay mechanism remains to be elucidated.
TSTD1 is expressed in several human tissues, including kidney, liver, skeletal muscle, heart, colon, thymus, spleen, placenta and lung (31). The TSTD1 protein is reportedly overexpressed in breast, colon and lung carcinoma cell lines (31). The colon is a unique compartment in which host cells are exposed to high concentrations of microbially-derived sulfide, and where thiosulfate rather than sulfate is the predominant sulfide oxidation product (44). As part of a larger focus on colon sulfide homeostasis in our laboratory, we examined the localization of TSTD1 in human colon. Interestingly, TSTD1 expression is higher at the apices of colonic crypts, i.e., at the host-gut microbiota interface (Fig. 9A). Significant increase in protein expression and diffuse localization of TSTD1 was observed in cross sections from cancerous colon tissue (H-score, p<0.001) (Fig. 9B, C, Fig. S1).
In summary, although the physiological role of TSTD1 is not known, our kinetic studies at a physiologically relevant pH make its postulated role in thiosulfate utilization unlikely. Instead, the crystal structure of the protein and its low K M for thioredoxin suggest a role in persulfide transfer involving protein partners.

Experimental Procedures
Materials Sodium sulfide, sodium sulfite, sodium thiosulfate, GSH, oxidized glutathione, L-homocysteine and L-cysteine were purchased from Sigma.
Expression and purification of TSTD1-A synthetic gene encoding human TSTD1, which was codon-optimized for expression in E. coli, was purchased from GeneScript (Piscataway, NJ). The cDNA was inserted into the pET-28b vector using a 5′-NdeI and a 3′-SacI site. BL21 (DE3) One Shot chemically competent E. coli cells (Invitrogen) were transformed with the TSTD1 expression plasmid. A 200 ml culture in Luria Bertani medium supplemented with 50 µg ml -1 kanamycin was grown overnight at 37 °C and used to inoculate a 6 × 1 L culture in the same medium. Cultures were grown at 28 °C and expression was induced with isopropyl-β-D-thiogalactopyranoside (100 µM) when the optical density at 600 nm reached 0.5. At the time of induction, the temperature was reduced to 16 °C and growth was continued for an additional 12 h. Cells were harvested by centrifugation at 2683 × g for 20 min at 4 °C.
For selenomethionine TSTD1 expression, a one-liter culture in terrific broth medium was grown overnight at 37 o C and used to inoculate a 10 × 1 liter culture in minimal medium supplemented with 50 mg/l selenomethionine to a final absorbance at 600 nm of 0.2. Cultures were grown at 28 °C and expression was induced with isopropyl-β-D-thiogalactopyranoside (200 µM) when the absorbance at 600 nm reached 0.4. Growth was continued for an additional 16 h at 28 °C. Cells were harvested by centrifugation at 10,800 × g for 20 min at 4 °C. TSTD1 was purified as described previously with the following modifications (22). The cell pellet from a 6-liter culture was resuspended in 150 mL of 50 mM Tris buffer, pH 8, containing 0.15 M NaCl (Buffer A), 1 tablet of protease inhibitor cocktail (Roche), DNase (3 mg), soybean trypsin inhibitor (2 mg), aprotinin (3 mg), phenylmethanesulfonyl fluoride (4 mg), tosyllysine chloromethyl ketone (0.45 mg) and MgSO 4 (5 mM final concentration) were added to the cell suspension and stirred at 4 °C for 60 min followed by sonication on ice with the following pulse sequence: 30 sec burst, 1 min rest for a total burst time of 10 min at a power output setting of 6. The sonicate was centrifuged at 38,758 × g for 1h at 4 °C. The resulting supernatant was loaded onto a 20 ml Ni-NTA column equilibrated with Buffer A containing imidazole (final concentration 8 mM). The column was washed with 500 mL of Buffer A containing 8 mM imidazole buffer. TSTD1 was eluted from the column with a linear gradient ranging from 8-250 mM imidazole in Buffer A. Fractions containing the TSTD1 protein were pooled, concentrated, and dialyzed overnight against 4 liter of 50 mM Tris, pH 8. TSTD1 was then loaded on to a 5 mL Mono-Q column equilibrated with 50 mM Tris, pH 8. TSTD1 was eluted from the column with a linear gradient ranging from 0-1M NaCl in 50 mM Tris, pH 8. Fractions containing TSTD1 were pooled, concentrated and store at -80 °C. Protein concentration was determined using the Bradford reagent (BioRad) with BSA as a standard. The SeMet TSTD1 was purified using the same protocol.
Thiosulfate:thiol sulfurtransferase activity assay-The TSTD1 sulfur transfer activities from thiosulfate to various thiol acceptors was monitored as described previously by detecting H 2 S using the lead acetate assay (14). The reactions were performed at 37 °C as described previously and contained 1-50 mM thiosulfate and either 1-50 mM GSH, L-cysteine, or Lhomocysteine. Reactions were initiated by addition of 5-10 µg enzyme. The reactions were monitored by the formation of lead sulfide at 390 nm and H 2 S concentrations were calculated using an extinction coefficient of 5,500 M -1 cm -1 (14,54).
Thiosulfate:thioredoxin sulfurtransferase activity assay-Steady-state kinetic parameters for TSTD1 were determined in the presence of thioredoxin/thioredoxin reductase and thiosulfate. The reaction mixture contained 300 mM HEPES buffer, pH 7.4, 0.2 mM NADPH, 3.5 µM thioredoxin reductase, (0-150 µM) thioredoxin and thiosulfate (0-50 mM) as described previously (13). The persulfide formed on thioredoxin is spontaneously displaced by the resolving cysteine, releasing H 2 S and forming a disulfide, which is subsequently reduced by NADPH/thioredoxin reductase (Fig. 7B). The initial rate of NADPH oxidation was monitored at 340 nm and an extinction coefficient, of 6,220 M -1 cm -1 was used to determine the concentration of product formed.
Persulfide analysis on thioredoxin using the BTA assay-Persulfide transfer from TSTD1 to thioredoxin was assessed using a modified BTA as described previously (37). Briefly, 120 µg of TSTD1 was pre-incubated with 10 mM thiosulfate for 5 min in 1 mL 100 mM HEPES buffer, pH 7.4, at 37 °C. Then, excess thiosulfate was removed using a 10 K Amicon concentrator (Millipore) and TSTD1 was incubated with 150 µg thioredoxin (wild type, C32A or C35S) in 1 mL 100 mM Tris, pH 8 for 5 min at 37 °C. The reaction mixture was incubated with 50 µM NM-biotin (EZ-Link Maleimide-PEG2-Biotin, Thermo Scientific) for 30 min with gentle mixing at room temperature and subsequently precipitated by addition of 4 volumes of cold acetone. After incubation for 1 h at -20 °C, the samples were centrifuged at 17,000 x g for 10 min, and the precipitated pellets were suspended in 0.1% SDS, 150 mM NaCl, 1 mM EDTA, 50 mM Tris-HCl, pH 8 and gently mixed with 300 µL streptavidin-agarose resin (Thermo Scientific) overnight at 4 °C. 50 µl of each sample was removed for use as equal loading controls. The resin was washed with 10 volumes of buffer (0.1 % SDS, 600 mM NaCl, 1 mM EDTA, 1% Triton X100, 50 mM Tris-HCl, pH 8) and incubated with 500 µl of the elution buffer (150 mM NaCl, 1mM EDTA, 50 mM Tris-HCl, pH 8) with or without DTT (30 mM) for 30 min at room temperature. The eluted protein was concentrated using 10K Amicon concentrators (Millipore) and analyzed by Western blot. The eluted proteins were separated by SDS PAGE, transferred to polyvinylidene fluoride membranes and probed with a rabbit monoclonal anti-thioredoxin antibody (Abcam; ab133524, lot # GR91668-12; 1:5000 dilution).
GSSH:sulfite sulfurtransferase activity assay-The standard reaction was performed at 37 °C with a mixture containing GSSH (0.05-10 mM), sulfite (0.05-10 mM), and enzyme (5-45 µg) in 100 mM HEPES buffer, pH 7.4, 150 mM NaCl (200 µl final volume) as described previously (14,54). The reaction was initiated by the addition of enzyme. After 30 min of incubation, monobromobimane was added to a final concentration of 8 mM and incubated at room temperature for 15 min. Samples containing millimolar concentrations of substrate, were diluted 2-10 fold (final volume 200 µl) before addition of monobromobimane. The derivatization reaction was terminated by addition of 100 µl of 0.2 M sodium citrate, pH 2.0. Production of thiosulfate from GSSH and sulfite was monitored by HPLC. Bimane adducts of GSSH, Na 2 S, GSH, sulfite, and thiosulfate, were prepared in 20 mM sodium phosphate, pH 7.4. Each standard (final concentration 250 µM) was incubated with a final concentration of 3 mM monobromobimane (200 µl final volume) for 10 min at 22 ˚C.
Protein crystallization-TSTD1 (native and SeMet) was crystallized by the sitting drop vapor diffusion method from a 0.75 µl:0.75 µl mixture of protein stock (10 mg/mL in 50 mM Tris, pH 8, 0.25 M NaCl) and well solution. Initial crystals appeared in 2 days at 20 °C using a well solution of 0.1 M sodium cacodylate, pH 6.5, 0.1 M sodium citrate. Optimal crystals were achieved by microseeding into 0.09 M sodium cacodylate, pH 6.5, 0.77 M sodium citrate. Rhombohedral shaped crystals were cryo-protected by passing harvested crystals through 2 µl of 0.08 M sodium cacodylate, 1.38 M sodium citrate followed by flash cooling in liquid nitrogen.
Data collection and crystal structure determination-Diffraction data were collected at the Advanced Photon Source (APS, Argonne National Laboratory) on the GM/CA beamine 23ID-B (Table 1). For the wild-type and SeMet TSTD1 crystals, data were collected to 1.04 and 1.1 Å, respectively. All data were processed using XDS (55). The TSTD1 structure was determined by single-wavelength anomalous diffraction using Phenix to locate the selenium atoms and to determine the initial phases. Phenix AUTOBUILD was used to build ~50 % of the initial model. Model building and refinement was performed with native data using Coot (56) and Phenix Refine (57). The structure was validated using MolProbity (58). Electron density is complete for all TSTD1 residues excepting two N-terminal residues (Met1 and Ala2) and the N-terminal histidine tag. Figures were prepared using PYMOL (59)(60)(61).
Immunohistochemistry-Immunostaining of paraffin-embedded, formalin-fixed colon tissue sections was performed as described previously using the ABC technique (62). Briefly, seven micro sections of formalin-fixed and paraffinembedded tissues were mounted on slides. The slides were then deparaffinized in xylene, and rehydrated into distilled H 2 O through incubation in graded alcohols. The antigen retrieval process was pursued by microwaving the slides in citrate buffer (pH 6.0; Biogenex, San Ramon, CA) for 10 min. Incubation with hydrogen peroxide (6%) in methanol was used to quench endogenous tissue peroxidase activity. The sections were post-fixed in 10% buffered formalin, washed, and blocked with 1.5% rabbit serum for 1 h. Sections were then incubated with a goat-anti-TSTD1 antibody (Santa Cruz sc-324457, lot #C1411) overnight at 4°C. Slides were washed in PBS and then incubated with a biotinylated anti-goat secondary antibody for 30 min at room temperature. Antigen-antibody complexes were detected with the avidin-biotin peroxidase method, using the chromogenic substrate 3,3′-diaminobenzidine (Vectastain ABC kit -catalog #PK-6105; Vector Laboratories, Burlingame, CA). Immunostained sections were lightly counterstained with hematoxylin and examined by light microscopy. The tumor tissues were staged according to the pTNM pathological classification (63).
The stained tissues were evaluated by a semiquantitative analysis to assign an H-score which was calculated based on the staining intensity and the percentage of stained cells with a specific magnitude of intensity. The H-score was calculated using the following equation: H-score = ƩPi(i), where "i" represents the intensity of staining (0, 1, 2 and 3 are for no, weak, intermediate and strong staining) and "Pi" is the percentage of stained cells with intensities varying from 0 to 100 %. The H-score value ranged from 0 to 300 where 0 is a complete negative staining and 300 is complete positive staining (64,65). The criteria used to evaluate staining intensity are shown in Fig. S1.          A. In one model of the mitochondrial sulfide oxidation pathway denoted by grey arrows, the product of sulfide quinone oxidoreductase (SQR) is thiosulfate (S 2 O 3 2-), which is further processed in the cytoplasm by TSTD1. The product of this reaction is GSSH, which must go across the mitochondrial membranes to be completely oxidized to sulfate. This pathway is considered to be unlikely since the major product of the SQR reaction at physiologically relevant sulfite and glutathione concentrations is predicted to be GSSH. In the mitochondrially contained sulfide oxidation pathway, H 2 S is progressively oxidized by SQR, persulfide dioxygenase (PDO), rhodanese (TST) and sulfite oxidase (SO) to give thiosulfate and sulfate, which are the major products of H 2 S oxidation. The input (H 2 S) and outputs (S 2 O 3 2-, SO 4 2-) of the sulfide oxidation pathway are highlighted in red. B. Based on the structure of TSTD1 and its ability to transfer sulfane sulfur to thioredoxin, we propose that TSTD1 functions in sulfur relay shuttles between donor and acceptor proteins. Alternatively, thioredoxin could potentially function as a persulfide donor and TSTD1 as an intermediary carrier for sulfane sulfur donation to acceptors.