Putative Iron-Sulfur Proteins Are Required for Hydrogen Consumption and Enhance Survival of Mycobacteria

Aerobic soil bacteria persist by scavenging molecular hydrogen (H2) from the atmosphere. This key process is the primary sink in the biogeochemical hydrogen cycle and supports the productivity of oligotrophic ecosystems. In Mycobacterium smegmatis, atmospheric H2 oxidation is catalyzed by two phylogenetically distinct [NiFe]-hydrogenases, Huc (group 2a) and Hhy (group 1h). However, it is currently unresolved how these enzymes transfer electrons derived from H2 oxidation into the aerobic respiratory chain. In this work, we used genetic approaches to confirm that two putative iron-sulfur cluster proteins encoded on the hydrogenase structural operons, HucE and HhyE, are required for H2 consumption in M. smegmatis. Sequence analysis show that these proteins, while homologous, fall into distinct phylogenetic clades and have distinct metal-binding motifs. H2 oxidation was reduced when the genes encoding these proteins were deleted individually and was eliminated when they were deleted in combination. In turn, the growth yield and long-term survival of these deletion strains was modestly but significantly reduced compared to the parent strain. In both biochemical and phenotypic assays, the mutant strains lacking the putative iron-sulfur proteins phenocopied those of hydrogenase structural subunit mutants. We hypothesize that these proteins mediate electron transfer between the catalytic subunits of the hydrogenases and the menaquinone pool of the M. smegmatis respiratory chain; however, other roles (e.g., in maturation) are also plausible and further work is required to resolve their role. The conserved nature of these proteins within most Hhy- or Huc-encoding organisms suggests that these proteins are important determinants of atmospheric H2 oxidation.

The genetic basis and physiological role of atmospheric H 2 oxidation is now largely understood. This process has been most comprehensively studied in the genetically tractable soil actinobacterium Mycobacterium smegmatis (Greening and Cook, 2014). In this organism, atmospheric H 2 oxidation is mediated by two membrane-bound, oxygen-tolerant hydrogenases, Huc (group 2a [NiFe]-hydrogenase, also known as Hyd1 or cyanobacterial-type uptake hydrogenase) and Hhy (group 1h [NiFe]-hydrogenase, also known as Hyd2 or actinobacterialtype uptake hydrogenase) . Additionally, M. smegmatis encodes a third [NiFe]-hydrogenase, Hyh (Hyd3), which mediates fermentative H 2 production during hypoxia . Both H 2 -oxidizing hydrogenases contain a large subunit containing the [NiFe] active site (HucL, HhyL) and a small subunit containing three iron-sulfur clusters (HucS, HhyS), as well as potential additional subunits Cordero et al., 2019b). These two hydrogenases are upregulated in stationary-phase cells, including in response to organic carbon limitation (Berney and Cook, 2010;Berney et al., 2014b). Consistently, when the structural subunits of these hydrogenases are deleted, strains show reduced growth yield and impaired long-term survival during starvation (Berney and Cook, 2010;Berney et al., 2014a;Greening et al., 2014b). Similar findings have been made in Streptomyces avermitilis; the sole hydrogenase of this organism, Hhy, is exclusively expressed in exospores and strains lacking this enzyme exhibit severe survival defects (Constant et al., 2010;Liot and Constant, 2016). Given these findings, it is proposed that bacteria shift from growing on organic compounds to persisting on atmospheric trace gases. Indeed, theoretical calculations indicate that the energy derived from atmospheric H 2 oxidation (0.53 ppmv) can sustain the maintenance of 10 7 to 10 8 cells per gram of soil (Conrad, 1999).
Despite this progress, little is currently known about the biochemical basis of atmospheric H 2 oxidation. One outstanding question is how electrons derived from H 2 oxidation are transferred to the respiratory chain. Most classes of respiratory uptake hydrogenases are predicted to be co-transcribed with a cytochrome b subunit Søndergaard et al., 2016). For example, such subunits interact with the prototypical oxygen-tolerant hydrogenases (group 1d [NiFe]-hydrogenases) of Escherichia coli and Ralstonia eutropha; they anchor the hydrogenase to the membrane and transfer electrons from the hydrogenase small subunit to the quinone pool (Frielingsdorf et al., 2011;Volbeda et al., 2013). However, we did not detect equivalent proteins in the operons encoding the structural subunits of Huc (MSMEG_2261-2270) or Hhy (MSMEG_2722 -2718) in M. smegmatis (Supplementary Figure S1) . Putative iron-sulfur proteins, tentatively annotated as HucE (MSMEG_2268) and HhyE (MSMEG_2718), were encoded downstream of the hydrogenase structural subunits and may potentially fulfill this role instead Greening et al., 2015b). In this work, we characterized the effects of deleting these genes on hydrogenase activity, growth, and survival in M. smegmatis. We also investigated their broader conservation in hydrogenase-encoding bacteria.

Bacterial Strains and Growth Conditions
All bacterial strains and plasmids used in this study are listed in Supplementary Table S1. Escherichia coli TOP10 was maintained on lysogeny broth (LB) agar plates (10 g L −1 tryptone, 5 g L −1 NaCl, 5 g L −1 yeast extract, 15 g L −1 agar), while Mycobacterium smegmatis mc 2 155 (Snapper et al., 1990) and derived mutants were maintained on LB agar plates supplemented with 0.05% (w/v) Tween 80 (LBT). For broth culture, E. coli was grown in LB. M. smegmatis was grown in either LBT or in Hartmans de Bont (HdB) minimal medium (Hartmans and De Bont, 1992) supplemented with 0.2% (w/v) glycerol. In all cases, liquid cultures were grown in rotary incubators at 37 • C with agitation (200 rpm).

Mutant Strain Construction
Allelic exchange mutagenesis was used to produce markerless deletions of the genes encoding two putative iron-sulfur proteins, hucE (MSMEG_2268) and hhyE (MSMEG_2718) (Supplementary Figure S2). Briefly, a fragment containing fused left and right flanks of the MSMEG_2268 (1800 bp) and MSMEG_2718 (3098 bp) genes were synthesized by GenScript. These fragments were cloned into the SpeI site of the mycobacterial shuttle plasmid pX33 (Gebhard et al., 2006) to yield the constructs pX33-hucE and pX33-hhyE (Supplementary Table S1). These constructs were propagated in E. coli TOP10 and transformed into wild-type M. smegmatis mc 2 155 cells by electroporation. Gentamycin (5 µg mL −1 for M. smegmatis or 20 µg mL −1 for E. coli) was used in selective solid and liquid medium to propagate pX33. Creation of the double iron-sulfur cluster mutant ( hucE hhyE) was achieved by transformation of hhyE electrocompetent M. smegmatis mc 2 155 with the pX33-hucE construct. Briefly, to allow for permissive temperaturesensitive vector replication, transformants were incubated on LBT gentamicin plates at 28 • C until colonies were visible (5-7 days). Resultant catechol-positive colonies were subcultured onto fresh LBT gentamicin plates and incubated at 40 • C for 3-5 days to facilitate integration of the recombinant plasmid flanks into the chromosome. The second recombination event was facilitated by subculturing catechol-reactive and gentamicinresistant colonies onto LBT agar plates supplemented with 10% sucrose (w/v) and incubating at 40 • C for 3-5 days. Catechol-unreactive colonies were subsequently screened by PCR to discern wild-type revertants from hucE, hhyE and hucE hhyE mutants. Primers used for the generation of mutants and for screening are listed in Supplementary Table S2.

Complementation Vector Construction
The genes for the putative iron-sulfur proteins were amplified by PCR and the resulting fragments were cloned into the constitutive expression plasmid pMV261 via PstI/HindIII site for hucE and BamHI/HindIII site for hhyE (Stover et al., 1991) to yield the constructs pMVhucE and pMVhhyE (Supplementary Table S1). Sequence fidelity of the genes was verified through Sanger sequencing and insertion of the genes into the vector was confirmed through restriction-digestion analysis (Supplementary Figure S3). The plasmid constructs were propagated in E. coli DH5α and transformed into M. smegmatis cells by electroporation. Vector pMVhucE was transformed into M. smegmatis wild-type and hucE strains, while pMVhhyE was transformed into wild type and hhyE mutant. In addition, an empty pMV261 was transformed into wild-type, hucE, and hhyE strains. These seven M. smegmatis strains were used for complementation experiments in respirometry and activity staining. Kanamycin (20 µg mL −1 for M. smegmatis or 50 µg mL −1 for E. coli) was used in selective solid and liquid medium to propagate pMV261. Primers used for the generation of the constructs are listed in Supplementary Table S2.

Respirometry Measurements
Cultures of wild-type, derived mutants, and complemented mutant strains of M. smegmatis were grown in 125 mL aerated conical flasks containing 30 mL HdB medium supplemented with 0.2% glycerol. Respirometry measurements were performed with mid-stationary phase cells, i.e., 72 h post OD max (∼3.0). A Unisense H 2 microsensor electrode was polarized at + 800 mV for 1 h using a Unisense multimeter and calibrated against standards of known H 2 concentration. Gas-saturated PBS was prepared by bubbling the solution with 100% (v/v) of either H 2 or O 2 for 5 min. The 1.1 mL microrespiration assay chambers were sequentially amended with stationary-phase cultures (0.9 mL, OD 600 = 3.0), H 2 -saturated PBS (0.1 mL), and O 2 -saturated PBS (0.1 mL). Chambers were stirred at 250 rpm, 37 • C. Changes in H 2 concentration were recorded using Unisense Logger Software, and upon observing a linear change in H 2 concentration, rates of consumption were calculated over a period of 20 s, which corresponds to the most linear uptake of hydrogen by the cells. Oxidation rates were normalized against total protein concentration, which was determined by the bicinchoninic acid method (Smith et al., 1985) with bovine serum albumin standards.

Growth and Survival Assays
Cultures of wild type and derived mutants of M. smegmatis were inoculated into 125 mL conical flasks containing 30 mL LBT medium (initial OD 600 of 0.001), in six biological replicates. Growth was monitored by measuring optical density at 600 nm (1 cm cuvettes; Eppendorf BioSpectrometer Basic); when OD 600 was above 0.5, cultures were diluted ten-fold in LBT before measurement. Specific growth rate during mid-exponential growth was calculated for each replicate using GraphPad Prism (non-linear regression, exponential growth equation, least squares fit). The long-term survival of the cultures was determined by counting colony forming units (CFU mL −1 ) of cultures 21 days post-OD max . Cultures were serially diluted in HdB containing no carbon source and spotted on to agar plates in technical quadruplicates. After incubation at 37 • C for 3 days, the resultant colonies were counted.

Sequence and Phylogenetic Analysis
Sequences homologous to M. smegmatis HucE (MSMEG_2268) and HhyE (MSMEG_2718) were retrieved by protein BLAST (Altschul et al., 1990) using the National Center for Biotechnology Information (NCBI) Reference Sequence (RefSeq) database (Pruitt et al., 2007). The retrieved hits were cross-referenced with the hydrogenase database (HydDB) (Søndergaard et al., 2016) in order to determine which organisms co-encode HucE with HucL and HhyE with HhyL. For downstream phylogenetic and motif analysis, sequences were filtered to remove truncated HucE/HhyE proteins and retain one protein sequence per genus. This resulted in a representative subset of 52 full-length HucE and 26 full-length HhyE sequences. The retrieved sequences were aligned using ClustalW in MEGA7 (Kumar et al., 2016). The phylogenetic relationships of these sequences were visualized on a maximum-likelihood tree based on the Poisson correction method and bootstrapped with 100 replicates. In addition, WebLogo (Crooks et al., 2004) was used to analyze the conserved motifs containing cysteine and histidine residues predicted to bind iron-sulfur clusters. The web-based software Properon (M. Milton 1 ) was used to generate to-scale genetic organization diagrams of the group 1h and group 2a [NiFe]-hydrogenases, with genes labeled according to the nomenclature in HydDB (Søndergaard et al., 2016).

HucE and HhyE Are Predicted to Be Iron-Sulfur Proteins Associated With Group 2a and Group 1h [NiFe]-Hydrogenases
We investigated the diversity of putative iron-sulfur cluster proteins associated with [NiFe]-hydrogenases by conducting a homology-based search using the amino acid sequences of HucE (MSMEG_2268) and HhyE (MSMEG_2718) from M. smegmatis (Supplementary Figure S1). Homologous sequences were retrieved from 14 phyla and 104 genera of bacteria (Supplementary Figure S4 and Supplementary Table S3).
Multiple sequence alignments show that HucE and HhyE proteins contain highly conserved motifs potentially involved in binding iron-sulfur clusters (Supplementary Figures S5,  S6). Both HucE and HhyE contain a CxxC motif within a domain homologous to NifU proteins (Yuvaniyama et al., 2000). The C-terminus of HhyE proteins contains two CxxC motifs typical of iron-sulfur proteins (e.g., rubredoxins). In contrast, the HucE proteins contain an C-terminal motif CxH(x 15−18 )CxxC that matches the signature motif of Rieske iron-sulfur clusters (Schmidt and Shaw, 2001) (Supplementary Figure S6). A subset of the species surveyed contain truncated HucE and HhyE proteins that contain the NifU-like domain, but lack the C-terminal domains (Supplementary Figures S4, S5).

HucE and HhyE Are Essential for H 2 Oxidation in Mycobacterium smegmatis
We used allelic exchange mutagenesis to generate markerless single and double mutants of the hucE and hhyE genes in M. smegmatis, i.e., hucE, hhyE, and hucE hhyE. Gene deletion was confirmed by PCR targeting chromosomal sequences adjacent to the flanking regions used for homologous recombination (Supplementary Figure S2). Assays were used to compare H 2 oxidation of these strains with the wild-type strain and strains containing previously generated deletions of the hydrogenase structural subunits, i.e., hucS, hhyL, and hucS hhyL, that lack hydrogenase activity (Berney and Cook, 2010;Berney et al., 2014b;Greening et al., 2014a).
We first used a H 2 electrode to measure rates of aerobic H 2 respiration mediated by whole cells of each strain. There were significant differences in the rate of H 2 oxidation for all deletion strains compared to the wild type (Figure 2A). Loss of hucE and hhyE resulted in reductions of 1.8-fold and 8.4-fold, respectively; such reductions were statistically indistinguishable from those observed in the mutants of the hydrogenase structural subunits hucS and hhyL. Deletion of both iron-sulfur proteins ( hucE hhyE) or both hydrogenase structural subunits ( hucS hhyL) caused complete cessation of H 2 oxidation, highlighting that these two hydrogenases are solely responsible for H 2 oxidation and that the putative iron-sulfur proteins are indispensable for this process. The low-level negative rates in hucE hhyE and hucS hhyL strains most likely reflect drift of the electrode rather than actual H 2 production by Hyh (Hyd3), since this hydrogenase is only upregulated during hypoxia . We successfully complemented the hucE and hhyE strains by reintroducing the hucE and hhyE genes on the episomal plasmid pMV261 (Stover et al., 1991) ( Figure 2B); in contrast, introducing the empty vector caused no effect and neither did introducing the complementation vectors in a wild-type background. This restoration of Huc and Hhy activities in complemented iron-sulfur protein deletion mutants strongly indicate that HucE and HhyE are essential for Frontiers in Microbiology | www.frontiersin.org The tree was constructed using the maximum-likelihood method (gaps treated with partial deletion), bootstrapped with 100 replicates, and rooted at the mid-point. The sequences used to create this tree are provided in Supplementary Table S3. H 2 oxidation. Moreover, the similarity in H 2 oxidation rates between the strains containing deletions of the catalytic subunits, compared to the putative iron-sulfur proteins, is consistent with HucE and HhyE being functionally linked with the Huc and Hhy hydrogenases, respectively.
In an interrelated assay, we performed activity staining of the Huc and Hhy hydrogenases using whole-cell lysates of wild-type and deletion mutant strains, with and without the complementation vectors, in the presence of the artificial electron acceptor nitroblue tetrazolium chloride. In the Huc activity staining gel (Figure 3A), three bands were observed in the wholecell lysates of wild-type strains, with or without complementation vectors: the top high-MW band, middle mid-MW band, and bottom low-MW band. Both the high-MW and low-MW bands correspond to Huc activity (Cordero et al., 2019b) and these bands were not observed in strains lacking either hucS or hucE. However, Huc activity was restored when the hucE strain was complemented by episomal expression of hucE. For Hhy activity  staining (Figure 3B), a mid-sized MW band was be observed in all wild-type strains. This band, which is the same middle band observed in the Huc activity stain (Figure 3A), corresponds to Hhy activity Cordero et al., 2019b). No Hhy staining was detected with the loss of either hhyL or hhyE, but complementation of the hhyE strain with hhyE restored Hhy activity. The similarity in the staining bands observed between hucS and hucE strains or between hhyL and hhyE indicate that the putative iron-sulfur proteins HucE and HhyE, like their respective hydrogenase core subunits HucS and HhyL, are important for hydrogenase activity. The artificial electron acceptor cannot compensate for the loss of HucE/HhyE and neither can HucE for HhyE nor HhyE for HucE. This further supports the model that HucE and HhyE form a functional association with Huc and Hhy, respectively.

HucE and HhyE Mutant Strains Have Significant Growth and Survival Defects
Previous genetic studies have shown that the hydrogenases modestly increase growth yield and long-term survival of M. smegmatis (Berney and Cook, 2010;Greening et al., 2014b). We therefore tested whether these findings extended to the putative iron-sulfur proteins by analyzing the growth rate, growth yield, and long-term survival of the seven aforementioned strains when cultured aerobically on rich media (LBT). In line with previous findings Greening et al., 2014b), no significant differences in specific growth rate were observed between the strains ( Figure 4A). However, there was a 10% reduction in the specific growth yield of the HhyE mutant compared to the wild-type strain (OD maxwt = 4.19 ± 0.21; OD max hhyE = 3.81 ± 0.09; p = 0.008) ( Figure 4B). This phenotype extended to the double mutant strain ( hucE hhyE) and again phenocopied single and double mutants lacking the hhyL gene.
We also tested whether the strains were defective in longterm survival by counting colonies of aerobic cultures 21 days following OD max . There were significant reductions in the survival of most strains compared to the wild-type ( Figure 4C). Cell counts were approximately two-fold lower for the hhyE and hhyL strains (p < 0.02), and four-fold lower for the double mutant strains (p < 0.002), relative to the wild-type. These findings agree with previous reports that atmospheric H 2 oxidation by the hydrogenases enables M. smegmatis to survive energy starvation  and further supports that the putative iron-sulfur proteins contribute to this function. For reasons currently unclear, no phenotypes were observed for the hucE strain.

DISCUSSION
In summary, this study shows that HucE and HhyE are required for the enzymatic activity and physiological function of the mycobacterial uptake hydrogenases. Strains lacking these proteins showed no hydrogenase activity in either amperometric or zymographic assays. Furthermore, they exhibited growth and survival phenotypes similar to those of knockouts of hydrogenase structural subunits (Berney and Cook, 2010;Greening et al., 2014b); as with the structural subunit mutants, these phenotypes are relatively minor, likely reflecting the numerous survival mechanisms present in M. smegmatis such as the ability to persist on carbon monoxide (Cordero et al., 2019a). Despite some sequence similarity between the two proteins, they are non-redundant, as there was no compensation in hydrogenase activity in the single mutant strains. The genomic survey and phylogenetic analysis indicate that hucE and hhyE genes co-evolved with the genes encoding the structural subunits of the group 2a and group 1h [NiFe]-hydrogenases. Their detection in the genomes of most but not all characterized high-affinity H 2 oxidizers indicate they are important but overlooked mediators of atmospheric H 2 oxidation. They are also associated with the group 2a [NiFe]hydrogenases of H 2 -recycling Cyanobacteria and various aerobic hydrogenotrophic bacteria that are not currently known to oxidize atmospheric H 2 .
This study lends some support to the hypothesis that these proteins serve as the immediate electron acceptors for the group 2a and group 1h [NiFe]-hydrogenases. There are broadly five lines of evidence that support this hypothesis: (i) the presence of highly conserved motifs for binding iron-sulfur clusters, (ii) the essentiality of these proteins for the function of these hydrogenases, (iii) their association with the structural rather than maturation operons of the hydrogenases , (iv) co-localization of HhyL, HhyS, and HhyE subunits on native polyacrylamide gels (Cordero et al., 2019b), and (v) their genomic association with hydrogenases that lack known electron transfer subunits (e.g., cytochrome b subunits). With the respect to the latter point, it is interesting that these proteins are conserved in Cyanobacteria, given the immediate electron acceptors of their uptake hydrogenases have long remained enigmatic (Tamagnini et al., 2002). It is also notable that HucE proteins encode the signature motifs of a Rieske iron-sulfur cluster. Given their unusual ligands, these clusters have a higher standard redox potential (E o ' > −150 mV) than most ironsulfur clusters (e.g., ferredoxins) (Brown et al., 2008). They would therefore be well-poised to accept the relatively high-potential electrons derived from atmospheric H 2 and transfer them to menaquinone. Consistently, zymographic studies suggest that the high-affinity hydrogenases operate at higher redox potential than prototypical hydrogenases, given they are reactive with the nitroblue tetrazolium (E o ' = −80 mV) but not viologen compounds (E o ' = −360 mV) (Pinske et al., 2012;Greening et al., 2014a).
While this study demonstrates HucE and HhyE are important for mycobacterial hydrogenase activity, further work is ultimately needed to resolve their respective function. While a role in electron transfer is most plausible, we have not demonstrated that these proteins interact with the hydrogenases and it is notable that the artificial electron acceptor nitroblue tetrazolium chloride cannot compensate for their absence. In this regard, other roles are also possible and compatible with the available evidence, for example as specific assembly factors and/or structural scaffolds for the hydrogenases. For example, it has been demonstrated that a rubredoxin-related protein is important for aerobic maturation of the group 1d [NiFe]-hydrogenase in R. eutropha (Fritsch et al., 2014). Furthermore, it is possible that other hypothetical proteins downstream of HucE and HhyE may also serve as electron acceptor candidates, in particular MSMEG_2717 that shares homology to PHG067, the proposed electron acceptor of R. eutropha (Schäfer et al., 2013). Biochemical studies, including studying the redox chemistry of these proteins and their interactions with the as-yet-unpurified hydrogenases, are now required to distinguish these possibilities and develop a sophisticated understanding of their function.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
CG conceived, designed, and supervised the study. CG and ZI were responsible for the phylogenetic analysis and analyzed the data. ZI and PC were responsible for the knockout generation, hydrogen electrode measurements, and activity staining. PC was responsible for the complementation experiments. ZI was responsible for the phenotypic assays. ZI, PC, and CG wrote the manuscript.

FUNDING
This work was supported by an ARC DECRA Fellowship (DE170100310; awarded to CG), an Australian Government Research Training Program Stipend (awarded to ZI), and a Monash University Doctoral Scholarship (awarded to PC).