A glycyl radical enzyme enables hydrogen sulfide production by the human intestinal bacterium Bilophila wadsworthia

Significance This paper describes a pathway for anaerobic bacterial metabolism of taurine (2-aminoethanesulfonate), an abundant substrate in the human intestinal microbiota, by the intestinal bacterium and opportunistic pathogen, Bilophila wadsworthia. This metabolism converts taurine to the toxic metabolite hydrogen sulfide (H2S), an activity associated with inflammatory bowel disease and colorectal cancer. A critical enzyme in this pathway is isethionate sulfite-lyase, a member of the glycyl radical enzyme family. This enzyme catalyzes a novel, radical-based C-S bond-cleavage reaction to convert isethionate (2-hydroxyethanesulfonate) to sulfite and acetaldehyde. This discovery improves our understanding of H2S production in the human body and may also offer new approaches for controlling intestinal H2S production and B. wadsworthia infections.

3 For the D. alaskensis G20 mutational analysis, the previously described lactate/sulfate medium (50) was used with the exceptions that the 8 mM MgSO 4 was replaced by 8 mM MgCl 2 and that 52 mM Na 2 SO 4 was present as the sole electron acceptor; lactate/sulfite medium contained 52 mM Na 2 SO 3 as the sole electron acceptor, and lactate/isethionate medium 60 mM isethionate. For growing D. alaskensis on plates, the previously described MOYLS4 medium was used 22 with 1 mM Na 2 S x 9 H 2 O as the reducing agent and 2 mM K 2 HPO 4 as the phosphorus source.

Total proteomics
Anoxic extracts prepared for the enzyme assays (see below) were used, or extracts prepared under oxic conditions when the cells were disrupted in a bullet blender (Next Advance Inc.) in 50 mM Tris-HCl buffer (pH 8.0) containing Halt protease inhibitor cocktail (Thermo Fisher Scientific). Cell debris was removed by centrifugation (15,000 g, 15 min, 4 °C). The extracts were subjected to peptide fingerprinting-mass spectrometry (PF-MS) at the Proteomics Facility of the University of Konstanz (www.proteomics-facility.uni-konstanz.de) as described previously (25,26,51) with the exception that each sample was analyzed twice on a Orbitrap Fusion with EASY-nLC 1200 (Thermo Fisher Scientific), and that Tandem mass spectra were searched against an appropriate protein database (retrieved from IMG) using Mascot (Matrix Science) and Proteom Discoverer V1.3 (Thermo Fisher Scientific) with "Trypsin" enzyme cleavage, static cysteine alkylation by chloroacetamide, and variable methionine oxidation.

Enzyme assays with cell-free extracts of Bilophila and Desulfovibrio
Cultures were harvested in the late exponential growth phase at an OD 580  Taurine-pyruvate aminotransferase was assayed in 50 mM Tris-HCl buffer (pH 9.0) containing 5 mM taurine, 10 mM sodium pyruvate, and 0.1 mM pyridoxal-5-phosphate. Assays with cell free extract were performed with 200 or 400 µL of extract per mL reaction. Heat inactivation was done by heating the enzymes to 98 °C for ten minutes in a heating block. Samples for HPLC (see below) were mixed with acetonitrile (HPLC grade) in a ratio of 7:3 immediately after sampling in order to stop enzyme reaction, centrifuged to remove precipitate (16,100 g, 10 min, 4 °C), and stored at -20 °C until analysis.
Sulfoacetaldehyde reductase was assayed spectrophotometrically as the oxidation of NADH recorded at 365 nm for 1 min; the standard reaction mixture (1 mL) contained 50 mM Tris-HCl (pH 8.0), 5 mM MgCl 2 , 0.4 mM NADH and 2 mM sulfoacetaldehyde-bisulfite adduct at room temperature. The reactions were started either by addition of substrate or by addition of cell extract. Sulfoacetaldehyde disappearance and isethionate formation during the reactions was confirmed by HPLC (see below) in subsamples taken from the reactions, in which the reaction was stopped by addition of 30% (v/v) acetonitrile; these subsamples were stored at -20 °C until analysis.
Isethionate sulfite-lyase was assayed discontinuously at room temperature by monitoring the formation of the two products sulfite and acetaldehyde. The reaction mixture (1 mL) contained 20 mM isethionate, 1 mM S-adenosylmethionine chloride (SAM), 1 mM Ti(III)-NTA, and about 0.2 -0.9 mg total protein in 50 mM Tris-HCl (pH 8.0) containing 5 mM MgCl 2 . First, buffer including isethionate and SAM was degassed under vacuum in glass cuvettes with rubber stoppers, and flushed with nitrogen gas, three times each. Then, Ti(III)-NTA was added, and the reaction was initiated by the addition of anoxic crude extract, each with a syringe and needle through the rubber stoppers. At appropriate time intervals, samples (100 µL) were taken by syringes for routine quantification of sulfite by two independent methods, through a colorimetric assay and by derivatization and HPLC-UV analysis (see below), and for routine quantification of acetaldehyde by derivatization and HPLC-UV analysis (see below).
Analytical chemistry for enzyme assays with Bilophila and Desulfovibrio cell extracts For routine HPLC with ELSD detection, a Shimadzu Prominence HPLC-DAD system coupled with a ZAM3000 ELSD detector (Schambeck SFD GmbH, Germany) was used with a SeQuant ZIC-HILIC hydrophilic interaction liquid chromatography column (5µm, 200 Å, 150 mm x 2.1 mm; Merck). The HPLC conditions were: acetonitrile as eluent A, and 0.1 M ammonium acetate buffer (pH 7.0) containing 10% acetonitrile as eluent B; total flow rate 0.3 mL/min. The elution gradient was: from 90% eluent A to 75% in 20 minutes; from 75% to 65% in 10 minutes; plateau at 65% A for 10 minutes. Under these conditions isethionate eluted at 11.5 min, taurine at 19.5 min, alanine at 20.6 min, NADH at 26.6 min, and NAD + at 28.7 min, while sulfoacetaldehyde eluted only poorly separated in between 5-16 min retention time. Sulfoacetaldehyde (and acetaldehyde) was determined also by HPLC-UV against authentic standards after derivatization with DNPH. Samples were mixed in a 1:1 ratio with the derivatization agent (0.5 mg/mL 2,4-dinitrophenylhydrazine [DNPH] in acetonitrile with 0.1 % H 3 PO 4 ), after which subsamples (5 µL) were analyzed by HPLC-UV. A mixed mode C 18 column (Luna Omega, 5 µm, PS, 5 C18, 100 Å, 140 mm x 3 mm; Phenomenex) was used. The flow rate was 1 mL/min with acetonitrile as eluent A and 0.1 % formic acid in MilliQ water as eluent B. HPLC conditions were as follows: Three minutes at 25 % A, gradient from 25 to 80 % A for ten minutes, reequilibration to 25 % A for seven minutes. Under these conditions, DNPH-derivatized sulfoacetaldehyde eluted at 5.7 min, derivatized acetaldehyde at 6.8 min, and free DNPH at 8.8 min.
Sulfite concentration in enzyme reactions was routinely determined by a colorimetric assay described previously (19). Samples taken from the reactions were immediately added in a ratio of 1:20 to a mixture of 0.56 M H 2 SO 4 , 0.16 ‰ (w/v) fuchsine and 0.16 % formaldehyde. The absorption at 580 nm was measured after 10 minutes against sulfite standards prepared from a fresh sodium sulfite stock solution.
Sulfite in enzyme reactions was also determined by HPLC-UV after derivatization with N-(9-acridinyl)maleimide (see further below) against authentic standards; the column and gradient system for analysis of DNPH-derivatized aldehydes (see above) was used for separation of the sulfite reaction product, which eluted at 3.8 -4.2 min under these conditions. PCR primers and DNA sequencing Primers for cloning were synthesized by Microsynth AG (Balgach, Switzerland), Integrated DNA Technologies (Coralville, IA) or Sigma Aldrich (Table S2) Experiments with mutants of Desulfovibrio alaskensis G20 Transposon insertion mutagenesis was previously used to inactivate Dde_1270, Dde_1272, Dde_1273, Dde_1274, and Dde_1275 (22). The frozen strains were revived on MOYLS4 plates amended with G418 (geneticin, 400 µg/mL) at 37 ˚C, and colonies were inoculated into Hungate tubes containing 5 mL lactate/sulfate medium amended with 800 µg/mL G418. For confirmation of specific transposon insertion by PCR, 20 µL of outgrown lactate/sulfate culture was heated at 95 ˚C for 10 min, and the lysed cells were diluted 1:10 in nuclease-free H 2 O and used as PCR template. The reactions (10 µL) contained 1 µL template, 0.5 µM pRL27_IE_rev1 and 0.5 µM gene-specific primer (see Table S2), and 5 µL 2x Q5 polymerase master-mix. The PCR program was: Denaturation for 12 min at 95 °C, followed by 25 cycles of 30 s denaturation at 95 °C, 30 s annealing at 55 °C, and 90 s elongation at 72 °C. Specific transposon 6 insertion for each mutant was confirmed by the presence of an amplicon at 800-1000 bp on agarose gels.
The growth of the mutants in lactate/sulfite and lactate/isethionate medium was monitored also in 96-well plates with a plate reader in an anoxic chamber. Therefore, precultures in Hungate tubes containing 5 mL lactate/sulfate medium (wildtype) or 5 mL lactate/sulfate medium amended with 800 µg/mL G418 (mutants) were grown at 37 ˚C for 48 h, and 100 µL of each culture was inoculated into fresh 5 mL medium and incubated further for 24 h. Then, 200 µL of each culture was inoculated into a well plate (in quadruplicate) containing 200 µL lactate/sulfite or lactate/isethionate medium, such that the initial OD 600 was 0.02 (pathlength-corrected). For 30 h, the OD 600 was recorded every 1 h with shaking every 20 min.
Heterologous over-production and purification of taurine-pyruvate amintotransferase (Tpa) and sulfoacetaldehyde reductase (SarD) Chromosomal DNA was isolated from B. wadworthia 3.1.6 using the protocol published by the DOE Joint Genome Institute (JGI), "JGI Bacterial DNA Isolation CTAB-2012" (https://jgi.doe.gov/userprogram-info/pmo-overview/protocols-sample-preparation-information/). The constructs for the production of recombinant and His-tagged Tpa and SarD of B. wadworthia were generated when the genes were amplified by PCR using Phusion polymerase (New England BioLabs) and the primers given in Table S2; the primers introduced a NdeI and XhoI site in front of the start and after the stop codons, respectively. The following PCR conditions were used: Denaturation for 90 s at 98 °C, followed by 30 cycles of 30 s denaturation at 98 °C, 30 s annealing at 68 °C, and 60 s elongation at 72 °C; final elongation was for 5 min at 72 °C. The PCR products were purified (QIAquick MiniElute Purification Kit) and blunt-end cloned into a pJet suicide vector (CloneJet PCR cloning kit, ThermoFisher Scientific).
2 µL of the reaction were transformed into chemically competent E. coli NovaBlue cells, by incubating 50 µL competent cells with 10 ng plasmid for 15 minutes on ice, followed by a heat shock at 42 °C for 30 s, and recovery on ice for 1 min. Then, 200 µL of SOC medium (2 % w/v tryptone, 0.5 % w/v yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl 2 , 20 mM glucose) was added and the cells were incubated at 37 °C for 45 minutes before streaking them onto selective LB agar plates (100 µg/mL ampicillin). Positive clones from the agar plates were selected by colony PCR using the insert primers, cultivated overnight in LB medium, and the plasmid was extracted (Zyppy Plasmid Miniprep kit, Zymo Research). Each 0.5 µg of pJet plasmid from a positive clone was digested with 20 U NdeI and 80 U XhoI (New England BioLabs) in CutSmart buffer for 2h at 37 °C. The inserts were then purified (Gel extraction Mini Spin Column kit, Genaxxon biosciences), and ligated (overnight, 16 °C; T4 ligase, NEB) 7 into the expression vector pET 28a(+) (Novagen), which had been linearized by digestion with NdeI and XhoI as described above. 2 µL of the ligation reaction were transformed into chemically competent E. coli NovaBlue and the cells streaked onto selective LB plates (30 µg/mL kanamycin). Positive clones were selected by colony PCR, and the constructs confirmed by DNA sequencing (GATC Biotech).
For heterologous overproduction of Tpa and SarD, the pET28 constructs were transformed into chemically competent E. coli Rosetta 2 DE3 cells as described above, and the cells grown in 5 mL Heterologous over-production and purification of glycyl radical enzymes (GREs), GRE activating enzymes, acetaldehyde dehydrogenase (AdhE) and substrate binding protein The GRE genes were assembled into pET28a plasmids, and the GRE activating enzyme, AdhE, and DctP genes were assembled into pET29b plasmids, each via Gibson assembly. Inserts were prepared when 20 ng of genomic DNA of B. wadsworthia or D. desulfuricans (prepared as described above) was mixed with primers (0.5 µM of each, see Table S2) and 1x Phusion master-mix (New England BioLabs) on a 50 µL scale. The inserts were amplified using the following protocol: Denaturation for 30 s at The expression host E. coli BL21(DE3) ΔiscR (52) for expression of the GRE activating enzyme (see below) was constructed as follows. The strain E. coli BL21(DE3) ΔiscR::kan was generated using P1 transduction with E. coli strain JW2515-3 (Coli Genetic Stock Center), which contains the ΔiscR777::kan mutation, as the donor strain and E. coli BL21(DE3) as the recipient strain as previously described (53).
Recombinant enzymes that required anoxic conditions (see below) were handled in an anoxic cabinet (MBraun) (100% N 2 atmosphere) or an anoxic soft vinyl chamber (Coy Laboratories) (97% N 2 /3% H 2 atmosphere). Samples were routinely rendered anoxic by one of the following methods. Consumable goods were brought into the glovebox the day before being used. Solid chemicals were brought into the anoxic chamber in Eppendorf tubes that had been perforated. Protein solutions were rendered anoxic as described in the specific section related to their purification. Media components were routinely rendered anoxic by sparging them with argon or N 2 prior to sterilization.
For heterologous overexpression, 50 ng of plasmid was transformed into 50 µL chemically competent E. coli BL21, or chemically competent E. coli BL21 ∆iscR (GRE activating enzymes). The cells were incubated on ice with DNA (5 min), incubated at 42 ˚C (30 s), and recovered on ice (2 min); SOC (200 9 µL) was added, and the cells were shaken at 37 ˚C for 1 h. Cells were then plated on LB-Kan50 and grown overnight. Single colonies were inoculated into 5 mL LB-Kan50, and these cultures transferred further into 100 mL LB-Kan50, if appropriate. For expression of GREs and the GRE activating enzymes, an entire 100 mL starter culture was inoculated into 4 L LB-Kan50, which was split equally between two 4 L shake-flasks, and for AdhE and DctP, a 5 mL starter culture was inoculated into 1 L LB-Kan50. For overexpression of the activase, the LB medium was supplemented with glucose (1% w/v) and Fe(III)ammonium-citrate (2 mM). The cultures were grown at 37 ˚C until they reached an OD 600 of 0.6, at which point IPTG (0.3 mM) was added. Then, the temperature was lowered to 15 ˚C and the cultures incubated overnight. At the point of induction, the cultures expressing the GRE activating enzymes were additionally sparged with argon for 20 min, and cysteine (2 mM) and sodium fumarate (20 mM) were added, before the cultures were sealed with screw-cap tops. These cultures were then transferred into the anoxic chamber that had been chilled to 15 ˚C and incubated overnight without shaking.
For the preparation of the GRE activating enzymes, all subsequent steps took place at 4 ˚C in an anoxic chamber unless otherwise specified. For the steps taking place outside of the anoxic chamber (centrifugation and incubation on a nutator), the containers were sealed with electrical tape before removing them from the chamber. For the GREs, AdhE and DctP, the cells were lysed by two passages through an Avestin EmulsiFlex-C3 cell disruptor . The lysates were clarified by centrifugation (30 min, 20,000 g). For the GRE activating enzymes, the cells were incubated with the lysozyme at 4 ˚C for 1 h. The cells were then lysed by sonication with a ½" horn (7 min total sonication, 10 s on, 30 s off, 25% amplitude) in a cold water bath. The dark gray solution was transferred into a fresh 50 mL plug-seal conical vial that was sealed with electrical tape, and the lysates were clarified by centrifugation (20,000 g, 30 min).
For protein purification, the supernatant was incubated with Ni-NTA resin (Qiagen) (6 mL for the GREs, 3 mL for the activating enzymes, 1 mL for AdhE and DctP) that had been equilibrated with 10 column volumes of the respective lysis buffer for 1 h. The resin was pelleted (500 g, 5 min), and the supernatant was decanted. The resin was resuspended in a minimal volume of lysis buffer and then transferred into a column, except for purification of the GREs, where the mixture was directly decanted into a column.
After the flowthrough was collected, the resin was washed with 50 mL lysis buffer (10 mL for AdhE and DctP). The proteins were eluted by sequential washes with elution buffer (50 mM HEPES pH 7.5, 200 mM NaCl, 250 mM imidazole, pH 8.0 for the acetaldehyde dehydrogenase); for the GREs, this was four steps with 6 mL each, for the activating enzymes three steps of 4 mL, and for AdhE four steps of 1 mL.
SDS-PAGE was used to identify the fractions containing the proteins and their purity.
Purified proteins were loaded into a dialysis cassette of an appropriate size; for GREs, the dialysis cutoff was 20 kDa MWCO, and 10 kDa for activating enzymes, AdhE and DctP. The proteins were dialyzed three times against 1.3 L dialysis buffer (50 mM HEPES pH 7.5, 50 mM NaCl, 10% (v/v) glycerol) for two 2 h steps and one overnight step; for AdhE and DctP, one of the 2 h steps was omitted.
In order to render the GREs anoxical, the dialyzed protein solution was concentrated via repeated spins in a 20 mL 30 kDa centrifuge filter (3,220 g, 10-20 min spins) until a volume of 5 mL was reached.
Thereafter, degassing was accomplished by 12 cycles of a short vacuum pull followed by refilling with argon, then a 5 min pull of vacuum and a 5 min refill with argon; the entire process was repeated two times. Then the flasks were sealed and transferred into the anoxic chamber (97% N 2 /3% H 2 ). Finally, all proteins were aliquoted into 200 µL portions in cryovials fitted with an O-ring, flash frozen in liquid N 2 , and stored at -80 ˚C. The cryovials with GREs and activating enzymes were sealed in anoxic Hungate tubes before freezing.

In vitro assays for recombinant Tpa and SarD
Enzyme assays were conducted in 50 mM Tris-HCl buffer at a pH of 9.0 containing 5 mM taurine, 5 mM sodium pyruvate, and 0.1 mM pyridoxal-5-phosphate. Typically, 50 µg/mL Tpa and 25 µg/mL SarD were used in an assay, and SarD and 2 mM NADH were added to the reactions after 90 minutes. Samples taken for HPLC were mixed with acetonitrile (HPLC grade) in a ratio of 7:3 immediately after sampling and stored at -20 °C until analysis.
For separation of taurine, isethionate, sulfoacetaldehyde, alanine, pyruvate, NAD + , NADH, sulfate and sulfite, the HPLC DAD ELSD system (see above) fitted with the hydrophilic interaction liquid chromatography column (ZIC-HILIC, see above) was used as follows: acetonitrile as the eluent A, and 0.1 M ammonium acetate buffer (pH 7.0) containing 10% acetonitrile as the eluent B; total flow rate 0.3 mL/min. The elution gradient was: from 90% eluent A to 75% in 20 minutes; from 75% to 65% in 10 11 minutes; plateau at 65% A for 10 minutes. Under these conditions, pyruvate eluted at 6.6 min, isethionate at 11.5 min, alanine at 20.6 min, NADH at 26.6 min, and NAD + at 28.7 min. Sulfoacetaldehyde eluted only poorly separated in between 5-16 min retention time, and taurine eluted at 19.5 min but co-eluted with the peaks for Tris and sodium.
For HPLC with ESI-MS-MS detection, an Agilent 1100 HPLC system fitted with the same chromatography system was connected to a Thermofisher LCQ ion trap mass spectrometer. The fragmentation spectra for each analyte were recorded from the total ions (TIC) with the following ranges The entire activation mixture was then loaded into a J. Young EPR tube (4 mm outer diameter and 8" length, from Wilmad Lab-Glass), sealed, removed from the anoxic chamber, and slowly frozen in liquid N 2 . Perpendicular mode X-band EPR spectra were recorded on a Bruker ElexSysE500 EPR instrument equipped with a quartz finger dewar (Wilmad Lab-Glass) for acquiring spectra at 77 K with liquid N 2 .
The samples were acquired with the following parameters: microwave frequency: 9. ms; modulation amplitude: 4 G; modulation frequency: 100 kHz. Normalization due to differences in modulation gain were automatically performed by the spectrometer. Typically, only a single scan was recorded to minimize any disruption due to bubbling from the liquid N 2 . The field was calibrated by using an external standard of bisdiphenylene--phenylallyl (BDPA) with g = 2.0026. An external standard of Frémy salt was prepared by dissolving K 2 (SO 3 ) 2 NO in anoxic 0.5 M KHCO 3 and diluting the standard to ~0.5 mM. The concentration of the standard was more precisely ascertained by measuring its absorbance at 248 nm (ε = 1,690 M -1 cm -1 ) using a NanoDrop 2000 UV-Vis Spectrophotometer. The double integral of the Frémy salt standard was calculated on the EPR spectrometer and then used to determine the concentrations of each of the protein samples. The EPR spectra from the activation mixtures were simulated using EasySpin in MatLab using the genetic algorithm followed by the Levenberg/Marquardt algorithm after sufficient optimization. These simulations yielded the g-value, the hyperfine coupling constant, and the linewidth associated with the GRE signal.

Enzyme assays with activated GREs
The GRE was first activated as described above for EPR spectroscopy. Activated GRE (1 µM) was then added to reaction buffer (50 mM HEPES pH 7.5, 50 mM NaCl) in a 50 µl, 250 µl or 500 µL scale, as specified in the following sections, and the reaction initiated by addition of isethionate. If appropriate, reaction buffer containing yeast alcohol dehydrogenase (8 µM) and NADH (3 mM) was used, and then activated GRE and isethionate was added. In the coupled assay with yeast alcohol dehydrogenase, acetaldehyde is reduced to ethanol, which allows for monitoring the GRE reactions spectrophotometrically via the NADH conversion (see below). Further, without including these components, sulfite and acetaldehyde condense to form 1-hydroxyethanesulfonate, which can be indistinguishable from isethionate by LC-MS. Aerobic control reactions were first removed from the anoxic chamber and gently aerated by pipetting, prior to mixing with the isethionate. Acetaldehyde standards in reaction buffer were derivatized at the same time. The samples were incubated in the dark for 1 h at room temperature, centrifuged (16,100 g, 10 min), and then held at 10 ˚C until analysis was carried out (<24 h). HPLC analysis was used to detect the DNPH-derivatized acetaldehyde, when separated using a Dikma C 18 Inspire (50 x 4.6 mm, 5 µm) column. Eluent A was 0.1% formic acid in H 2 O and eluent B 100% acetonitrile, and the flow rate was 1.0 mL/min. The gradient was started at 20% B, was held at 20% B for 1 min, ramped to 80% B over 1.5 min, held at 80% B for 0.5 min, decreased to 20% B over 1 min, and reequilibrated at 20% B for 2 min. The absorption at 360 nm was monitored. 5 µL of sample was injected. For detection of CoASH and acetyl-CoA, 5 µL of each sample was injected onto the Sequant ZIC-HILIC column (see above). Eluent A was 20 mM ammonium acetate pH 8.0, and eluent B was acetonitrile. The flow rate for the column was 0.4 mL/min, and the column was started at 100% B which was decreased linearly to 0% B over 10 min, held at 0% B for 2 min, increased to 100% B over 2 min, and held at 100%

Bioinformatic identification of putative sulfoacetaldehyde reductases
The distribution of the sulfoacetaldehyde reductase was examined by using the B. wadsworthia sequence as the query for a BLAST search in the NCBI non-redundant protein database (September 7, 2017).
Authentic acetaldehyde dehydrogenases, such as CutO from the choline utilization cluster, exhibit only modest sequence identity to SarD (33% ID/E-55). To identify putative SarD homologs, we gathered only the sequences that exhibited substantially more homology than that relationship (27 sequences in total; the least similar sequence was 39% ID/5E-90 to the query sequence). The primary sequences of these proteins and several experimentally verified alcohol dehydrogenases were aligned as outlined above.
The putative metal-binding residues within all proteins (36 proteins (September 7, 2017). The 250 highestscoring sequences were exported (down to ~35% ID/E-145 score). As many of these sequences are unlikely to be authentic isethionate sulfite-lyases, these sequences were then clustered into a sequence similarity network using Option C of the Enzyme Function Initiative's Enzyme Similarity Tool (EFI-EST; 59). The network was generated with a sequence length requirement of >750 amino acids and with an initial alignment score of E-200. The full network was downloaded, and the edge value was subsequently refined in Cytoscape 3.2 to 62% ID, as this edge value had previously been determined to be likely sufficient to separate proteins that catalyze distinct reactions into distinct sequence similarity clusters (32). The three confirmed isethionate sulfite-lyases (from B. wadsworthia, D. desulfuricans, and D. alaskensis) co-occurred in the same cluster with 115 GRE sequences, suggesting that this cluster encodes isethionate sulfite-lyases or other C-S bond cleaving GREs.
Multiple tactics were taken to validate this hypothesis. First, a multiple sequence alignment of these sequences was constructed as described above. A homology model was also constructed of the B.
wadsworthia IslA by using the MPI Bioinformatics Toolkit. The protein was used as the query sequence for an HHPred search of the most recent PDB release (as of January 3, 2017), yielding structurally characterized proteins with significant sequence similarity. The top 8 returned hits had similarity scores more stringent than E-138 and represented all structurally-characterized GREs other than ribonucleotide reductase. These proteins were used as the templates for homology model construction with Modeller.
The homology model, 1,2-propanediol dehydratase from Roseburia inulinivorans (PDB ID 1R9D), and choline trimethylamine-lyase from D. alaskensis (PDB ID 1FAU) were visualized in Pymol. The multiple sequence alignment and a structural alignment in Pymol indicated that residues Q193, C468, E470, and G805 in the B. wadsworthia IslA align with the residues critical for dehydration in 1,2-propanediol dehydratase or trimethylamine elimination in choline trimethylamine-lyase. These residues may thus dictate specificity for bisulfite elimination; notably, these four residues are universally conserved in the 115 putative isethionate sulfite-lyases. Other residues in the active site may be less than 100% conserved, and hence it will take more (structural) work to discern whether the proteins are likely isethionate sulfitelyases or operate on other organosulfonates.
The organisms encoding these putative IslAs were tabulated. The number of sequenced strains within the NCBI non-redundant protein database in a given taxonomic class was estimated by searching the NCBI RefSeq assemblies using "Desulfovibrio[orgn] AND latest refseq[filter]", where the "Desulfovibrio" portion was replaced with the taxon of interest. Thus, we could roughly calculate the proportion of strains within a given taxonomic order that encode an isethionate sulfite-lyase.       Cultures were grown in either lactate/sulfite medium (blue) or lactate/isethionate medium (red).

Construction of a phylogenetic tree for the glycyl radical enzymes
The data shown is the mean of four replicate growth experiments ± standard error of the mean after baseline correction.