The conditional nature of organosulfur auxotrophy in met17Δ yeast

The ability of met17Δ yeast strains to overcome organosulfur auxotrophy at high cell density was first described by Cost and Boeke [19,21]. This phenotype is nonheritable, which ruled out adaptive, secondary mutations concurrent with met17Δ as the cause. We studied the phenotype in two haploid yeast strains in the BY4741 background [24]: (i) a prototrophic strain in which auxotrophies were repaired by genomic integration of the missing genes (prototroph or wild-type (WT): HIS3, LEU2, URA3, MET17); and (ii) an analogous strain in which the met17Δ allele was not repaired (met17Δ: HIS3, LEU2, URA3, met17Δ). As previously observed by Cost and Boeke [19,21], our met17Δ strain was also able to grow in the absence of methionine over the course of 8 days when replicated in thick patches, independently confirming the reproducibility of this phenomenon (S1A Fig).

We started our investigations into the biochemical basis of this phenotype by ruling out amino acid contaminations in commercial yeast nitrogen broth (YNB), a common source of conflicting and unusual growth phenotypes. We validated the concentration of sulfur containing amino acids using liquid chromatography-selected reaction monitoring (LC-SRM) [25]. We detected levels of methionine, homocysteine, and glutathione at a concentration at least 100,000-fold lower than typical levels found in replete yeast media (S1B Fig). Therefore, the growth of met17Δ cells in the absence of methionine was not due to the presence of growth-relevant concentrations of sulfur-containing amino acids in the growth medium. To have a more experimentally tractable system to further our investigations, we next explored the conditions under which the growth phenotype was reproducible in liquid culture in a 96-well plate format. We inoculated the met17Δ strain into methionine-free medium at a range of initial densities (OD_600nm = 0.01 to 0.3) and cultured under four temperatures (22, 26, 30, and 37°C), recording the endpoint OD_600nm at 24 h (S1C Fig) and 48 h (Fig 1B) as a measure of growth. At typical yeast culture temperatures (22, 26, and 30°C), robust growth could be observed where the initial inoculation density was OD_600nm > 0.02, in agreement with previous observations [19]. Growth at the lowest inoculation density (OD_600nm = 0.01) was only observed when the culture temperature was at 26°C, confirming that culture temperatures alter the cell density threshold that permits growth. Similarly, growth was permissive at 37°C only when the initial cell density was high (OD_600nm ≥ 0.08). In contrast, the WT strains could robustly grow at 30°C irrespective of initial inoculation density (Fig 1B). Although we did not observe a strict correlation of temperature with growth as previously reported, growth was greatly enhanced at lower temperatures. Therefore, we established 25°C as the standard culture temperature for all subsequent experiments, with low-density (OD_600nm < 0.01) and high-density (OD_600nm ≥ 0.02) inoculums being nonpermissive and permissive densities for growth in the absence of organosulfur supplementation, respectively. These thresholds were independent of culture size (96-well versus 20 ml flasks; S1D Fig).

Next, to probe whether the cell density limitation is specific to met17Δ or whether it is a more general phenomenon related to methionine or amino acid auxotrophy, we compared the growth of the met17Δ strain to other enzyme knockout strains of sulfur assimilation pathways (met1Δ, met3Δ, met5Δ, met7Δ, or met13Δ) or amino acid metabolism (argΔ, hisΔ, lysΔ, trpΔ, and aroΔ). All auxotrophic strains demonstrated growth deficiency, but only the met17Δ formed biomass close to that of the WT strain (Figs 1C and S1E). This result suggested that the ability to overcome the growth defect is not a common phenotype of methionine or amino acid auxotrophs but rather a specific phenotype that occurs upon the deletion of MET17. We speculated that either another enzyme replaces the function of Met17p in these conditions or else a metabolic shunt exists that allows yeast to circumvent the canonical sulfur utilization pathway to fuel growth.

The ability to overcome organosulfur auxotrophy does not require the exchange of homocysteine or methionine

Cost and Boeke [21] speculated that met17Δ cells could overcome the growth defect through the sharing of organosulfur metabolites between cells to growth-relevant levels. Indeed, previous work from us and others has shown that communal yeast cells can effectively share metabolites to overcome auxotrophies [26,27]. In order to better understand the possible impact of metabolite exchange in the context of the ability of met17Δ cells to grow in the absence of methionine, we employed self-establishing metabolically cooperating (SeMeCo) communities, a synthetic yeast community designed to study metabolite exchange interactions between auxotrophs [27]. SeMeCo communities are composed of cells that express different combinations of auxotrophic markers and are established from a prototrophic founder cell through the stochastic loss of plasmids encoding one or more enzymes that compensate for metabolic deficiencies present in the genome. In the BY4741 background, these are met17Δ, his3Δ, leu2Δ, and ura3Δ, which induce methionine, histidine, leucine, and uracil auxotrophies, respectively. In SeMeCos, the degree of metabolite sharing influences the frequency of a respective auxotroph in the population. Examining data recently acquired by us revealed that met17Δ cells are the most dominant subpopulation within exponentially growing SeMeCo cultures (approximately 60%), and more frequent than cells harboring the three other metabolic deficiencies (S1F Fig) [28]. To understand the role that metabolite sharing plays in overcoming the loss of MET17 in the SeMeCo system, we recorded the metabolic profiles of SeMeCos by LC-SRM and compared these against the metabolic profiles of prototrophic communities. In SeMeCo communities dominated by met17Δ cells, we detected a relative increase in intracellular homocysteine and homoserine concentrations compared to prototrophic communities (Fig 1D, intracellular). However, extracellular concentrations of homocysteine and methionine were not statistically different, although homoserine levels were elevated upon the exclusion of an outlying measurement (Fig 1D, extracellular). This result suggested that indeed, the met17Δ metabolic deficiency might be overcome through the exchange of an upstream precursor rather than directly via methionine sharing.

met17Δ cells up-regulate enzymes involved in sulfur assimilation, cysteine and methionine metabolism

We have previously recorded proteomes from auxotrophs and prototrophs isolated from SeMeCo cultures using fluorescence activated cell sorting, and a CFP marker expressed from the same plasmid as the auxotrophic marker ([28]; PRIDE project: PXD031160). A differential expression analysis comparing specifically the MET17 and met17Δ subpopulations revealed a clear up-regulation of sulfur assimilation enzymes (Met3p, Met14p, Met5p, and Met10p) and enzymes of the methionine salvage pathway (Meu1p, Spe3p, Spe4p, Bat1p, and Bat2p) in the MET17 strain (Fig 1F). This result is consistent with a typical feedback response to the lack of methionine. Counterintuitively, however, several of the downstream enzymes, including Met6p and Gsh1p, that should carry no flux in the absence of Met17p, had increased protein levels (Fig 1A and 1F). In addition, the abundance of several core enzymes of the methionine cycle (Met6p, Sam1p, Sam2p), serine biosynthetic pathway (Ser1p, Ser33p), and cysteine biosynthetic pathway (Cys3p, Cys4p) were also increased, although this was not observed with all enzymes (Sah1p, Ser3p). This data argued that homocysteine could be further metabolized despite the absence of Met17p, which would suggest the existence of a metabolic bypass to the homocysteine synthase activity of Met17p.

Hydrogen sulfide fixation drives growth in the absence of methionine

We speculated that the intracellular accumulation of homocysteine in met17Δ cells (Fig 1D) could indicate the presence of an enzyme-catalyzed reaction that forms homocysteine independent of Met17p. As sulfide overflow is a defining characteristic of met17Δ cells [19,21], one possible explanation would be the presence of an enzyme that can utilize sulfide or its protonated forms (hydrosulfide (HS⁻) and hydrogen sulfide (H₂S)) to incorporate sulfur and form homocysteine. To test this hypothesis, we exploited the cell density dependency of met17Δ cells in liquid culture that leads to the resolution of organosulfur auxotrophy as a readout (Figs 1B and S1C and S1D). When conditioned medium from met17Δ cells cultured either at high-density (OD_600nm > 0.02, permissive) or low-density (OD_600nm < 0.01, nonpermissive) was filtered and inoculated with fresh cells at low density, only the filtered medium from the high-density culture could support cell growth. Moreover, this property was abolished upon precipitation of H₂S with lead acetate (Fig 2A). To further substantiate whether H₂S could be responsible for the growth, we supplemented two low-density cultures with sodium hydrosulfide (NaHS) that dissociates to form hydrosulfide ions and, subsequently, both aqueous and gaseous forms of H₂S (Fig 2B, reactions (ii-iv)). Taking advantage of the phase equilibria of H₂S, one culture was left unsealed to drive the equilibria towards the formation and dissipation of gaseous H₂S from the system (Fig 2B, reaction (iv)). Growth was only observed in the sealed culture, consistent with the assumption that a critical concentration of H₂S is required to overcome the met17Δ auxotrophy (Fig 2C). In parallel, we manipulated the phase equilibria in the opposite direction, allowing the accumulation of aqueous H₂S in low-density cultures via outgassing from either a high-density culture or a concentrated solution of NaHS by connecting the vessels with a small piece of rubber tubing to create a closed system. In both situations, growth was observed in the low-density culture (Fig 2D). To further substantiate that it is H₂S, which resolves the auxotrophy, we evaluated its presence and uptake across different cultures and growth conditions, exploiting the chemical properties of 7-azido-4-methylcoumarin (AzMC). In the presence of H₂S, AzMC undergoes selective reduction of the azido moiety to form 7-amino-4-methylcoumarin (AMC), which fluoresces when exposed to ultraviolet light [29]. When minimal media was compared with cultures inoculated at low density, no significant difference in AMC fluorescence was observed. Conversely, the fluorescence in NaHS-supplemented cultures decreased in the presence of cells, suggesting utilization of H₂S and its concomitant depletion from the medium (Fig 2E). This fluorescence assay also indicated that high-density cultures had higher levels of aqueous H₂S, relative to low-density cultures, orthogonally confirmed by the precipitation of lead sulfide in high-density cultures (Fig 2A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Hydrogen sulfide is essential for the rescue of organosulfur auxotrophy of met17Δ.

(A) met17Δ cultures inoculated at low starting density (OD_600nm < 0.01) and high densities (OD_600nm ≥ 0.02) (upper panel, left and middle). One flask of high-density inoculation was supplemented with Pb(II) acetate to precipitate H₂S, forming lead sulfide (upper panel, right). Conditioned media from these three cultures was subsequently filtered, replenished with YNB and glucose, and reinoculated at low density (lower panel). Data are representative from two biologically independent experiments. (B) Reaction schemes. Hydrolysis of sulfide is in equilibrium with hydrosulfide and hydrogen sulfide (i, iii, iv). NaHS can serve as a hydrosulfide donor (ii) in solution. Hydrogen sulfide preferentially undergoes phase transition to a gaseous state (iv) in unsealed vessels or establishes a dynamic equilibrium with its aqueous counterpart in sealed vessels. (C) Growth of met17Δ cultures supplemented with NaHS and inoculated at low density, sealed vs. unsealed. (D) Cultures inoculated with the met17Δ strain where the vessels are connected via rubber tubing to facilitate gaseous exchange, either between a high-density culture or a high-concentration solution of NaHS (20 mM). Schematic indicates the reactions occurring at each stage to facilitate the growth of low-density cultures. (E) Quantification of hydrogen sulfide production and uptake via conversion of AzMC to AMC [21] between unsupplemented and NaHS supplemented cultures inoculated at low density in the presence (purple) or absence (gray) of met17Δ cells. Blue and red bars indicate a separate experiment where hydrogen sulfide production was quantified met17Δ cultures inoculated at either low (blue) or high (red) density, in a similar experiment as shown in (A). P values calculated via two-tailed Student t test, where n = 3 independent replicates. (F) Isotopic tracing experiment with pathway map indicating the biosynthetic transfer of isotopic S³⁴ from ammonium sulfate to methionine (left group) and glutathione (right group) via methionine and cysteine biosynthetic pathways, respectively. Pie charts indicate the percentage of S³⁴ for cultures supplemented with S³⁴-labeled ammonium sulfate. The data underlying this figure can be found in S1 Data. AMC, 7-amino-4-methylcoumarin; NaHS, sodium hydrosulfide; YNB, yeast nitrogen broth.

https://doi.org/10.1371/journal.pbio.3001912.g002

Finally, to test if sulfur does indeed transition from sulfate through to the biosynthesis of cysteine and methionine in the absence of MET17, we performed isotopic tracing with S³⁴-labeled ammonium sulfate (Figs 2F and S2A). Uptake of this isotope label by sulfur assimilation via the up-regulation of the associated enzymes should lead to S³⁴ dissemination into all subsequent sulfur bearing derivatives via sulfide overflow (Figs 1A and S2B). We quantified the ratios of S³² and S³⁴ sulfur in methionine and reduced glutathione, the downstream products derived from homocysteine and cysteine biosynthesis, respectively, using liquid chromatography mass spectrometry (LC–MS). When native (S³²) ammonium sulfate was supplied, the assay detected S³⁴ at its expected environmental isotopic abundance (S2B Fig; 4% to 6%). Conversely, when we supplied S³⁴-labeled ammonium sulfate, S³⁴-methionine (91.4%) and S³⁴-reduced glutathione (67.9%) accumulated to levels well above natural abundances, indicative of sulfur transfer from inorganic sulfate to organic sulfur-bearing metabolites. Thus, met17Δ assimilated inorganic sulfur from ammonium sulfate, and yeast cells are competent in the use of H₂S as a source for methionine and cysteine biosynthesis.

A targeted genetic screen identifies the bypass enzyme responsible for H₂S-mediated growth in the absence of methionine

Having established homocysteine formation via H₂S fixation as the process by which methionine and cysteine auxotrophy could be overcome, we next used a targeted genetic approach to identify the enzyme responsible. We started by identifying strains in the yeast deletion collection that are associated with sulfur metabolism [20]. We selected 15 strains (S1 Table) that carried a deletion in the sulfur metabolism-associated gene in addition to deletions of four auxotrophic markers in their genome that include MET17 (his3Δleu2Δura3Δmet17Δ) based on two criteria: (i) the deleted gene product was a part of either the sulfur assimilation or organosulfur biosynthetic pathways (S2B Fig); or (ii) the deleted gene product was identified as an ortholog to Met17p according to the eggNOG database [30–32] (Fig 1A and S2 Table). These selected deletion strains were cultured in synthetic drop-in media and tested for met17Δ growth in the presence of NaHS. In a background of MET17 deletion, sul1/2Δ, met3Δ, met14Δ, met5Δ, and met10Δ strains demonstrated robust growth only in the presence of NaHS, while the met6Δ strain lacking the methionine synthase was unable to grow. These growth phenotypes indicated that H₂S utilization and the resolution of organosulfur auxotrophy in the presence of NaHS was independent of sulfur assimilation, but dependent on sulfur utilization (Figs 3A and S2B and S2C). Critically, these experiments also revealed the dependency of the H₂S utilization pathway on homoserine-O-acetyltransferase that is responsible for the conversion of homoserine to OAHS as the met2Δ strain was also unable to grow (Fig 3B). Since OAHS is a substrate shared by both Met17p and Str2p, this implies a similar catalytic mechanism (Fig 1A) [17]. Of the strains deficient in Met17p orthologs, the met17Δcys3Δ strain also could be rescued by NaHS supplementation, albeit with an extended lag phase, which suggests the existence of a direct route towards cysteine formation via H₂S, possibly via cysteine synthase activity (S2C and S3B Figs, gray pathway; [14,33,34]). Notably, the str2Δ and str3Δ strains also had a lengthened lag phase upon NaHS supplementation. This suggested that the degree of transsulfuration from cysteine to homocysteine may act to regulate the reaction efficiency, although neither Str2p nor Str3p appears to directly catalyze the bypass reaction itself. Intriguingly, we observed that YLL058WΔ led to the loss of the ability to utilize H₂S for growth (Fig 3A). This defect was not observed in strains deficient in the other Met17p orthologs (irc7Δ, YML082WΔ, and YHR112CΔ), indicating that YLL058Wp could potentially function as the bypass enzyme. We next asked if the gene products of STR2 and YLL058W could operate as a homocysteine synthase in place of Met17p to rescue organosulfur auxotrophy. Str2p has been shown to be a cystathionine-γ-synthase, catalyzing the conversion of cysteine to cystathionine using OAHS as a substrate (Fig 1A). When operating in tandem with Str3p, this permits the transsulfuration of cysteine to methionine [14,35]. Since Str2p shares the same substrate as Met17p, there was a possibility it could incorporate sulfur via a secondary or promiscuous reaction. The catalytic properties of our other candidate YLL058Wp have not previously been experimentally characterized, although based on its close phylogenetic relationship to Str2p, we hypothesized that it could possess cystathionine-γ-synthase activity or at least interact with similar substrates (S2E Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Identification and characterization of YLL058Wp/Hsu1p as the homocysteine synthase responsible for hydrogen sulfide fixation.

(A) A targeted genetic screen of knockout strains lacking in addition to met17Δ, the genes encoding the homoserine (Met2p), methionine (Met6p) biosynthesis pathways, and selected orthologs of Met17p: Str3p, Str2p, and YLL058Wp. Cultures were inoculated at low density either in the presence or absence of NaHS, which rescues growth in met17Δ strains. Growth curves were recorded over 84 h with 3 biologically independent replicates for each strain (n = 3) as shown. A mutant strain carrying met17Δ alone was used as control. Complete or partial loss of NaHS rescue phenotype in these cultures indicate involvement of the deleted enzyme in the metabolic bypass (purple). Images are representative of the endpoint following growth curve capture. (B) Simplified pathway that summarizes the deletion mutants with a no growth (red) or delayed growth (orange) phenotype during NaHS-mediated growth rescue. Metabolites bearing sulfur are marked in purple. Gray pathway indicates the possible conversion from OAS to cysteine that we observe in vitro with YLL058Wp. (C) Methionine limitation screen comparing met17Δstr2Δ and met17ΔYLL058WΔ. Low amounts of methionine were provided to fuel the first phase of growth, allowing cells to reach a critical density at which sufficient hydrogen sulfide leaked from cells to allow a second phase of growth. In this setup, NaHS supplementation enables the second phase of growth to proceed only if the bypass is functional by allowing cells to freely use H₂S as a sulfur source regardless of limited concentrations of methionine. (D) Schematic indicating the setup of in vitro enzymatic reactions and growth assays used to assess the ability of OAHS reaction supernatants to rescue the indicated deletion strains (met17Δ, met17ΔYLL058WΔ, met17Δstr2Δ). Enzyme was sourced both from GFP immunoprecipitation (Hsu1-GFP) and recombinant (His-Hsu1, His-Str2, His-Met17) sources. (E) Quantification of homocysteine via Ellman’s reagent and spectrometry using postreaction supernatants where immunoprecipitated Hsu1p-GFP was incubated with OAHS as the substrate. RCT_0, 25, and 100 indicate bead volumes of enzyme–resin slurry used in the reaction. RCT(-PLP or -EDTA) indicates reactions whereby pyridoxal 5′-phosphate (PLP) or EDTA has been omitted from the reaction buffer. Data are mean thiol concentrations ± SD where n = 3 biologically independent replicates. Student t test was used to test for significance between indicated samples (***, p < 0.0001, ns = no significance). (F) Enzyme assay as in (E) utilizing recombinant enzymes purified by Ni-NTA affinity chromatography. Data are mean thiol concentrations ± SD where n = 3 biologically independent replicates. Student t test was used to test for significance between indicated samples (*, p < 0.05, ns = no significance). (G) Growth assays whereby each strain was inoculated at low density and supplemented with 50 μl of the indicated in vitro reaction product (n = 3). Final OD_600nm measured following 5 days of outgrowth. For OAHS reaction supplements, data are mean OD_600nm ±SD where n = 3 biologically independent replicates. OD_600nm reached in minimal media and supplementation with homocysteine or methionine (n = 1) are shown for comparison. (H) Growth assay as in (G) using postreaction supernatant generated from recombinant Hsu1 (rHsu1). Growth curves were captured for met17Δ and met17Δstr2Δ over 3 days, and for met17Δhsu1Δ over 5 days with 3 biologically independent replicates for each strain (n = 3). Schemes for Fig 3A, 3C, and 3D were created through BioRender.com. The data underlying this figure can be found in S1 Data. EDTA, ethylenediaminetetraacetic acid; NaHS, sodium hydrosulfide; Ni-NTA, nickel-NTA; OAHS, O-acetylhomoserine; OAS, O-acetylserine.

https://doi.org/10.1371/journal.pbio.3001912.g003

To characterize how YLL058Wp and Str2p contributed to the bypass reaction, we devised an experiment in which we supplemented only a limited concentration of methionine into minimal media that was rapidly depleted (Fig 3C). In essence, the assay separates growth into two phases; the first phase is driven by methionine until the cells have reached the critical cell density that is required by the cells to overcome the MET17 deficiency using the bypass enzyme. The met17Δstr2Δ strain maintained robust growth, both in the presence and absence of supplemented NaHS, indicating that the bypass reaction does not require Str2p. In contrast, growth of the met17ΔYLL058WΔ strain slowed after the initial biomass accumulation phase that is driven by the presence of methionine and did not achieve robust growth even in the presence of ample NaHS (Fig 3C). Thus, YLL058Wp is necessary for the cell to assimilate sulfide and sufficient to catalyze homocysteine biosynthesis from OAHS and sulfide, while Str2p appears to indirectly affect the bypass reaction. Considering the H₂S dependency of YLL058Wp, we henceforth refer to it as Hydrogen Sulfide Utilizing-1 (Hsu1p).

Hsu1p is a homocysteine and cysteine synthase that fixes sulfur from hydrogen sulfide

Having identified an enzyme required for the bypass reaction, we next sought to further characterize the levels and activity of Hsu1p, using a yeast strain that expresses a Hsu1-GFP fusion protein in an auxotrophic background lacking Met17p (leu2Δura3Δmet17Δ) [36]. We first verified by PCR the identity of the strain isolated from this library (S3A Fig). Monitoring GFP expression via flow cytometry, we observed that the GFP fluorescence increased 1.5-fold in minimal media supplemented with NaHS, indicating that HSU1 expression is responsive to sulfide levels (S3B Fig). We next immunoprecipitated Hsu1-GFP from yeast cells grown under these conditions. Then, we tested in vitro if the immunopurified protein was capable of transferring the sulfur from H₂S onto OAHS to form homocysteine (Figs 3D and S3C). We also tested the ability of the enzyme to function as a cysteine synthase using O-acetylserine (OAS) as an alternative substrate, since this could potentially explain the growth of the met17Δcys3Δ strain in the presence of NaHS (Figs 3B and S3D). We quantified the formation of homocysteine or cysteine from OAHS or OAS, respectively, upon covalent linkage of inorganic sulfide to organic backbones using Ellman’s reagent and spectrophotometry [37]. While the detection of homocysteine was hampered by the presence of a high background, the reaction of OAS to cysteine could be followed without strong interference. Cysteine synthase activity was observed at high enzyme concentrations (RCT_100; S3D Fig). To further validate homocysteine synthase activity in Hsu1p and to overcome the issue of high background, we generated recombinant versions of Hsu1p and Met17p by expressing the proteins in E. coli and purifying them via affinity chromatography (nickel-NTA (Ni-NTA)) and molecular weight cutoff (MWCO) filtration. We again tested for homocysteine synthase activity, this time using the purified recombinant enzymes in the presence or absence of OAHS substrate and reduced levels of PLP and ethylenediaminetetraacetic acid (EDTA) (Figs 3F and S3F). We observed an increase in homocysteine in the presence of OAHS for both Met17p and Hsu1p, again indicating that Hsu1p possesses homocysteine synthase activity. However, high enzyme levels were also required, indicating that this might be a secondary catalytic activity. We therefore tested whether the increase in homocysteine levels that we observed was relevant to our phenotype and linked the enzyme assay to a growth assay. Specifically, we asked whether supplementation with postreaction supernatant could rescue the growth of strains that previously had a growth defect in the absence of methionine or NaHS supplementation (met17Δ, met17Δhsu1Δ, met17Δstr2Δ, and met17Δcys3Δ). Supernatants obtained from the in vitro OAHS and OAS reactions (RCT_0, RCT_25, and RCT_100) were supplemented into minimal media without methionine and inoculated with the above strains. At 5 days postinoculation, the endpoint OD_600nm of the OAHS supplemented cultures were compared against cultures directly supplemented with 0.15 mM methionine and homocysteine. Critically, we observed that supplementation with postreaction supernatant generated in the presence of immunoprecipitated Hsu1-GFP restores growth in the mutant backgrounds, akin to direct supplementation with an organosulfur compound (Fig 3G). Conversely, the supernatants taken from both OAHS and OAS reactions did not support growth of any tested strain in the absence of the immunoprecipitated enzyme (RCT_0; Figs 3G and S3E). Similarly, growth profiles captured from cultures supplemented with the OAHS postreaction supernatant generated in the presence of recombinant Hsu1p could also rescue the growth of the deletion strains, converse to supernatants that did not contain OAHS (Fig 3H). Hence, the reaction products formed in the presence of Hsu1p were sufficient to rescue the growth defects of the organosulfur auxotrophs.

Evidence that Hsu1p is part of a sulfur limitation response in yeast

One prominent feature of our proteomic analysis was that met17Δ cells up-regulate enzymes involved in sulfur assimilation and methionine salvage (Fig 1E and 1F). This suggested that met17Δ cells experience sulfur limitation, owing to the reduced methionine and cysteine concentrations that would stem from the loss of its homocysteine synthase activity. Furthermore, an increase in the levels of Hsu1p is also triggered by increased sulfide concentrations (S3B Fig), which suggests that Hsu1p is regulated by the extracellular sulfide concentration. We hence sought to understand the conditions under which Hsu1p levels are increased in the presence of Met17p, which may help identify ecological conditions in which Hsu1p is essential, helping to explain its conservation. Previous studies have also shown that HSU1 is localized within a regulon-like gene cluster that is believed to be up-regulated for the utilization of alternative sulfur sources [38,39]. We therefore hypothesized that Hsu1p could function under sulfur limitation conditions in conjunction with other enzymes that enable the usage of alternative sulfur sources, such as H₂S. To explore this possibility, we analyzed the transcriptomic data acquired from a prototrophic yeast strain grown under sulfur limitation [40] together with genes identified as part of alternative sulfur usage regulons [39]. Comparing the genes within the datasets and the overlap between them revealed the up-regulation of sulfur assimilation enzymes and alternative organosulfur transporters/permeases upon sulfur limitation (Fig 4A). Hsu1p was found up-regulated in both datasets, together with Mmp1p (high-affinity S-methylmethionine permease), Jlp1p (Fe(II)-dependent sulfonate/alpha-ketoglutarate dioxygenase), Yct1p (high-affinity cysteine-specific transporter), Soa1p (sulfonate and inorganic sulfur transporter), and Bds1 (bacterially derived sulfatase). Therefore, the induction of HSU1 in parallel with other genes involved in the uptake or utilization of alternative sulfur sources suggested a functional association of these genes to a sulfur limitation response. Furthermore, the up-regulation of sulfur assimilation enzymes (Met1p, Met3p, Met8p, Met10p, Met16p, and Sul2p) is similarly observed in the met17Δ proteomes (Fig 1G), which is strongly suggestive of a common sulfur limitation signature.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Hsu1p confers a growth advantage under prolonged sulfur limitation in the presence of Met17p.

(A) Comparison of genes identified by transcriptomics that are up-regulated under sulfur limitation against genes identified as part of the sulfur limitation regulon regulated by Met4p [39,40]. Of the 68 genes up-regulated in response to sulfur limitation, 24 (35%) have functional annotations related to sulfur metabolism or transport. Gene functions are classified as indicated. * indicates gene products that have a defined enzymatic function related to the usage of alternative sulfur sources. (B) Profiling of Hsu1-GFP by flow cytometry in a prototrophic strain exposed to sulfur limitation conditions via titration of MgSO₄ (mg/L). SM indicates minimal media containing standard concentrations of (NH₄)₂SO₄ (5 g/L) and MgSO₄ (500 mg/L) giving a total sulfate concentration of 5.5 g/L. Histograms indicate population of GFP+ cells measured (>20,000); red line indicates relative median fluorescence intensity of the captured population. (C) Growth profiles and endpoint OD_600nm of control, HSU1, and hsu1Δ strains under sulfur-replete (SM) and limited (SL_5) conditions. All strains express MET17 due to the repair of auxotrophies via plasmid complementation with pHLUM. Opposed to the HSU1 (his3Δ) and hsu1Δ strains, the control strain does not harbor any gene knockout and is Kan−. The HSU1 strain is his3Δ, which serves to control for the position effect of the Kan4MX cassette, thereby making it comparable to hsu1Δ. (D) Setup of a growth competition assay competing the control strain with either HSU1 or hsu1Δ strains grown in SM vs. SL_5. Since both knockout strains harbor a Kan4MX cassette their frequency can be assessed by replica plating onto SM±G418. The number of positive colonies can be counted via image analysis and presented as a ratio between the two replica plates per condition and per time point. (E) Frequency of Kan+ and Kan− populations across all conditions and time points tested (pie charts). Adjacent barplot summarizes the fold change in Kan+ populations over 48 h between SM and SL_5 conditions. The data underlying this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3001912.g004

To experimentally test this hypothesis, we modified our minimal medium, replacing ammonium sulfate with ammonium chloride, and titrating magnesium sulfate (MgSO₄, 0 to 500 mg/L) together with magnesium chloride (MgCl₂, 500 to 0 mg/L)) to maintain the magnesium concentration while modulating the sulfate concentration (see Materials and methods). In a background of MET17 prototrophy, the Hsu1-GFP strain demonstrated an increase in GFP fluorescence during conditions of sulfur limitation (Fig 4B; [MgSO₄] = 5 mg/L). This phenotype is abrogated when sulfate concentrations are increased (SM, or [MgSO₄] = 50 or 500 mg/L), suggesting that the increase in Hsu1p levels is conditional upon and regulated by either sulfur or organosulfur limitation. Growth profiling of both the HSU1 and hsu1Δ strains during sulfur limitation reveals that the presence of Hsu1p confers a moderate growth advantage under these conditions (Fig 4C). To further investigate this observation, we subjected both strains to a competition assay against a control strain that expresses HSU1. In these experiments, our experimental HSU1 strain carried a Kan4MX marker (Kan4MX::his3Δ) and a pHLUM plasmid expressing HIS3, LEU2, URA3, and MET17. The control strain carried HSU1 and had no Kan4MX marker, allowing us to use kanamycin as a marker to distinguish the experimental HSU1 and hsu1Δ strains from the control HSU1 strain. Following 48 h of competition in both sulfur-replete (SM) and sulfur-limited (SL_5) conditions, we plated a sample of the cultures onto minimal media, followed by replica plating onto SM or SM+G418 to assess the proportion of Kan+ cells in the original culture (Fig 4D). In agreement with the growth curve analysis (Fig 4C), we observed that HSU1 expression confers a moderate but significant 1.2-fold growth advantage under sulfur-limited conditions, which was not observed when sulfur was replete (Fig 4E). Conversely, HSU1 expression appears to be detrimental in sulfur-replete conditions, as the hsu1Δ strain had a similar growth advantage. This suggests that cells under rich conditions suppress HSU1 expression to alleviate this growth defect. We suspect this growth defect may be a result of substrate competition, most likely for OAHS, which is shared among Met17p, Str2p, and Hsu1p. Therefore, it appears that HSU1 expression is tightly regulated, is strongly conditional on changes in sulfide concentration, and its presence may provide a competitive advantage to cells growing in low-sulfur conditions.

Yeast culture

Yeast strains (S1 Table) were streaked from cryostock onto YPD, synthetic minimal medium (SM, 6.8 g/L YNB without amino acids (Sigma, Y0626; Lot: MKCF2863), 2% w/v glucose (Sigma), 2% w/v agarose (Life Technologies)) or supplemented synthetic minimal medium (SM+histidine (H, 20 mg/L), leucine (L, 100 mg/L), uracil (U, 20 mg/L), and methionine (M, 20 mg/L) and incubated for 3 days at 30°C before a single colony was picked to inoculate 20 ml of media in a conical flask and subsequently incubated at 30°C for growth under various conditions as outlined in the experiments above. Prototrophic (WT) and SeMeCo strains were revived onto SM medium as described in ref [28], while strains harboring only met17Δ were streaked onto SM+methionine. Strains where auxotrophy complementation was introduced were cultured in selective media (variations of SM±HLUM) at all times unless otherwise stated. Cultures that required activation of the bypass reaction (high-density inoculations or supplementation with NaHS) were cultured at 25°C to maximize growth (see Fig 1B).

Growth curve profiling

For the growth analysis of amino acid auxotrophs (Fig 1C), deletion mutants from the BY4741 prototrophic collection and the met17Δ strain were revived on SM or SM+Methionine for 36 h. Single colonies were picked for each strain, resuspended in water (Millipore, autoclaved) in a short 96-well plate, and then pinned onto synthetic complete (SC) agar medium using the Singer RotorHD. After 72 h, colonies were used to inoculate a 96-well plate containing SM liquid medium and repinned onto SM solid medium. Growth curves were recorded by placing the plate containing liquid medium in a Tecan Infinite M200 PRO spectrophotometer to determine optical density at 600 nm (10 flashes/read) for 72 h. The solid media plate was incubated at 30°C for 72 h. Growth curve analysis was conducted using the growfit package in R(v4.0.2) and ΔOD_600nm was calculated, extracted, and plotted using custom scripts. In NaHS supplementation and the other growth curves, growth curve profiling was extended to 72 to 100 h to capture a complete profile and plotted directly using custom scripts.

Systematic growth analysis of met17Δ in minimal media

Single colonies of the met17Δ and the prototrophic WT strain (n = 3 replicates per genotype) were inoculated in SM+Methionine, grown overnight at 30°C in Erlenmeyer flasks, diluted 1:20 in fresh SM+Methionine or SM media, and incubated for 4 h. Cells were collected, washed in autoclaved H₂O (Millipore), resuspended in SM, and the OD_600nm of each strain was adjusted to 0.6. Each set of replicates (n = 3 for the deletion strain, and n = 3 for the control strain) were then transferred in a deep 96-well plate where a sequential 1:1 dilution of 100 μl culture was performed with 100 μl SM media to obtain an OD_600nm range of 0.3 to approximately 0.0003. 4 × 200 μl of each well was transferred to 4 × 96-well microtiter plates (Costar). The plates were sealed using a nonpermeable, clear foil (Roth) and were incubated at 22°C, 26°C, 30°C, and 37°C without shaking. Optical density was recorded by placing the plates in an Infinite M200 Pro plate reader (Tecan), and OD_600nm was obtained from the median of 9 multi-well reads per well. Subsequently, for each plate, the median OD_600nm of all blank (n = 8 per plate) values was subtracted from obtained OD_600nm values.

Hydrogen sulfide production quantification

Qualitative assessment of hydrogen sulfide production was indicated by the precipitation of lead sulfate in growth media containing 0.1 g/L lead(II) acetate. Quantification of hydrogen sulfide was conducted using AzMC, which undergoes selective reduction of the azide moiety in the presence of hydrogen sulfide to form AMC. In brief, 10 μl of filtered yeast media was mixed with 200 μl of 10 μM AzMC and incubated at RT for 30 min before the fluorescence was recorded on a Tecan spectrophotometer at _emλ = 450 nm (_exλ = 365 nm). For the generation of hydrogen sulfide, we utilized NaHS as the donor, reconstituted with water to a stock concentration of 20 mM and a working concentration in the medium of 0.2 or 0.4 mM.

Metabolite quantification and isotopic tracing by LC–MS/MS

Ortholog mapping and phylogenetic analyses

Orthologs of Met17p were identified in the Eggnog Database (http://eggnog5.embl.de/; [32]) using Met17p as a query protein and considering orthologs from the Eukaryota taxa. This yielded 753 proteins from the KOG0053 orthogroup, from which we only considered orthologous proteins belonging to S. cerevisiae (7 proteins; S2 Table). All these genes were considered for our screen of H₂S utilization in a met17Δ background.

To understand the phylogenetic relationship between the S. cerevisiae orthologs of Met17p (S2E Fig), each of these sequences was then used as a query in NCBI pBLAST (https://blast.ncbi.nlm.nih.gov/) focusing on the genomes of 26 model organism species including 17 fungal species, 2 plant species, 5 animal species, a choanoflagellate (Salpingoeca rosetta), and E. coli (S4 Table). Duplicates and highly similar proteins were removed (Similarity score over 60 using the globalms algorithm from biopython’s pairwise2 package with match_points = 1, mismatch_points = −1, gap_open = −0.5, gap_extension = −0.1 and penalize_end_gaps = False) [47]. Multiple Sequence Alignments (MSAs) were then generated using mafft v7 with–genafpair and–maxiterate 1000 options. The MSAs were trimmed using TrimAl 1.2 with -gappyout [67] and trees were generated from the trimmed alignment using IQTREE v1.6.12 and the model LG+R5 with E. coli protein EEV6508859.1 as the outgroup [68]. The full tree is included in the Supporting information, and the topology of this tree (but with branch lengths not drawn to scale) was used to construct the summary tree manually (S2E Fig).

Immunoprecipitation of Hsu1-GFP

The strain was streaked from the GFP library and the coding region was verified via specific primers [20] (S3 Table). To maximize the recovery of Hsu1-GFP protein, the strain was cultured for 5 days in 150 ml minimal media supplemented with leucine, uracil, and 0.4 mM NaHS at 25°C. Following this, cells were pelleted by centrifugation at 3,220g for 5 min at 4°C and resuspended in an appropriate volume of Y-PER Yeast Protein Extraction Reagent (Thermo Fisher) containing 1× Protease Inhibitor cocktail (P8215, Sigma). Purified protein extract was obtained following the manufacturer’s protocol, to which 250 μl of GFP-clamp resin (in-house, Crick-Structural Biology Scientific Technology Platform) was added to the lysis solution and incubated at 4°C for 2 h with rotation.

Generation of recombinant Hsu1p and Met17p

Genomic DNA was isolated from a prototrophic BY4741 strain [28] following an established protocol [48]. The coding sequence from each protein was cloned via two rounds of PCR, firstly to amplify the coding sequence and secondly to attach the leader sequences required for subcloning into a bacterial expression vector via ligation independent cloning (LIC; primers are available in S3 Table). The His-tagging, thioredoxin-fusion expression vector, pN-TrxT, was obtained from Addgene (#26101; [49]) and digested with BsaI to release the sacB stuffer fragment and linearize the vector. Following gel purification of both linearized vector and PCR fragments, LIC was initiated following treatment with T4 DNA polymerase in the presence of either dGTP (vector) or dCTP (inserts). Vector and inserts were mixed at a 1:3 ratio and incubated at RT for 5 min to facilitate annealing. Nicks were sealed upon transformation into competent DH5α E. coli (NEB) and plated onto selective media (LB+10% sucrose), which eliminates clones carrying undigested vector. Clones were then picked and insertion was verified by PCR and sequencing before amplification and extraction of the plasmids via miniprep (Qiagen). Plasmids were then retransformed into competent BL21 E. coli (NEB) and selected on LB+Kan plates. For recombinant protein production, a single colony was picked for each recombinant protein and used to inoculate the initial 1L culture, to which IPTG was added at a final concentration of 400 μM when OD_600nm = 0.8. Cultures were then grown overnight at 22°C with shaking and harvested the following morning via centrifugation at 4,000 rpm for 10 min. Bacterial pellets were lysed in B-PER Bacterial Protein Extraction Reagent (Thermo) according to the manufacturer’s instructions. Presence of His-tagged protein was initially confirmed using His-Tag Protein Expression Check Kit (Abcam), before lysates were subjected to two rounds of Ni-NTA-based affinity purification (Thermo) and one round of MWCO centrifugation (Merck) following the manufacturer’s protocol. Following purification, protein solution was dialyzed into 20 mM Tris–HCl buffer (pH 7.8), and protein was quantified via BCA assay (Thermo). Protein gels (Invitrogen) were run and stained with Instablue Coomassie Blue (Abcam) to verify purified protein was of the expected size. An equal amount of each enzyme was then added to the reaction tubes to which 200 μM PLP, 1 mM EDTA, and 5 mM OAHS (Insight Biotech) were added and initiated as below.

Hydrogen sulfide utilization enzyme and growth assays

For immunoprecipitated Hsu1-GFP, beads were washed with twice with Y-PER Yeast Protein Extraction Reagent (Thermo Fisher), followed by a final wash with reaction buffer (0.1 M potassium phosphate buffer (pH 7.8), 1 mM EDTA, 200 μM pyridoxal 5′-phosphate (PLP), all Sigma). Enzyme-bead complexes were then resuspended in 100 μl of reaction buffer containing 5 mM OAS (Sigma) or OAHS and placed with cap open into Eppendorf tube holders contained in a large beaker. A smaller beaker containing 20 mM NaHS solution was also placed into the large beaker that was sealed in parafilm and made airtight. The entire setup was placed at 30°C and incubated for 24 h, during which H₂S would outgas from the NaHS solution, dissolve, and react with OAS/OAHS in the presence of the enzyme (S3C Fig). The setup was identical for reaction tubes containing recombinant enzymes opposed to enzyme-bead complexes, with the exception of EDTA and PLP, whose concentration was reduced by half to reduce background. Tubes were then retrieved from the beaker and placed in an Eppendorf Thermomixer and shaken for 2 h at RT to remove residual H₂S before quantification via Ellman’s reagent (Thermo Fisher). For growth assays, 50 μl of reaction supernatant was transferred to 15 ml round bottom tubes containing 3 ml of media and incubated with shaking for up to 5 days at 25°C before endpoint OD_600nm was measured using a cuvette spectrophotometer.

Quantification of cysteine/homocysteine

Quantification of cysteine/homocysteine resulting from the OAS/OAHS reaction, respectively, was performed using Ellman’s reagent (5,5-dithio-bis-(2-nitrobenzoic acid, ThermoFisher; [37]), which preferentially reacts with sulfhydryl groups to yield a colored product. In brief, a standard curve using cysteine (0 to 0.2 mM, Sigma) was generated and incubated with the reagent following the manufacturer’s protocol at a 1:1 ratio. Absorbance was measured on a Tecan Infinite M200 PRO spectrophotometer at λ = 412 nm, and concentration was calculated using the standard curve.

Sulfur limitation assays