Brucella ovis Cysteine Biosynthesis Contributes to Peroxide Stress Survival and Fitness in the Intracellular Niche

ABSTRACT

Brucella ovis is an ovine intracellular pathogen with tropism for the male genital tract. To establish and maintain infection, B. ovis must survive stressful conditions inside host cells, including low pH, nutrient limitation, and reactive oxygen species. The same conditions are often encountered in axenic cultures during stationary phase. Studies of stationary phase may thus inform our understanding of Brucella infection biology, yet the genes and pathways that are important in Brucella stationary-phase physiology remain poorly defined. We measured fitness of a barcoded pool of B. ovis Tn-himar mutants as a function of growth phase and identified cysE as a determinant of fitness in stationary phase. CysE catalyzes the first step in cysteine biosynthesis from serine, and we provide genetic evidence that two related enzymes, CysK1 and CysK2, function redundantly to catalyze cysteine synthesis at steps downstream of CysE. Deleting cysE (ΔcysE) or both cysK1 and cysK2 (ΔcysK1 ΔcysK2) results in premature entry into stationary phase, reduced culture yield, and sensitivity to exogenous hydrogen peroxide. These phenotypes can be chemically complemented by cysteine or glutathione. ΔcysE and ΔcysK1 ΔcysK2 strains have no defect in host cell entry in vitro but have significantly diminished intracellular fitness between 2 and 24 h postinfection. Our study has uncovered unexpected redundancy at the CysK step of cysteine biosynthesis in B. ovis and demonstrates that cysteine anabolism is a determinant of peroxide stress survival and fitness in the intracellular niche.

INTRODUCTION

Brucella spp. are intracellular pathogens that have numerous mechanisms to contend with host-generated stressors and exploit host resources for growth. Within the host, brucellae are subject to nutrient limitation (1), phagosomal acidification (2), and direct attack from reactive oxygen and reactive nitrogen species (3) originating from the host-derived respiratory burst (4, 5). Dozens of genes involved in oxidative stress responses, acid stress responses, nutrient assimilation, and respiration have been implicated in the biology of Brucella infection (6). More recent studies have defined a role for the general stress response pathway in mitigation of multiple chemical stressors in vitro and in maintenance of chronic infection in vivo (7, 8). However, relatively little is known about the mechanisms Brucella spp. use to adapt to stresses encountered in axenic cultures during stationary phase. The study of stationary phase has the potential to inform the discovery of genes that influence infection, intracellular replication, and survival (9), as there are postulated parallels between stationary-phase physiology and the physiologic state of Brucella in the intracellular niche (1).

We sought to develop an approach to identify genes involved in stationary-phase physiology in the ovine pathogen Brucella ovis. B. ovis (10) is an understudied member of the Brucella clade that has a number of distinguishing genomic features (11). It is one of two naturally rough species among the classical Brucella group (12) and is the only species of this group that is nonzoonotic. The host environment inhabited by B. ovis is quite restricted: the pathogen is typically sexually transmitted and has a specific tropism for the male genital tract in rams (13–15). We previously constructed a randomly barcoded (RB) library of B. ovis Tn-himar mutants (31), and in the present study, we set out to develop this barcoded mutant library as a tool to identify B. ovis genes with fitness defects in stationary-phase culture. We measured the relative fitness of RB Tn-himar mutants as a function of growth phase in a complex medium and discovered that disruption of the cysteine biosynthesis gene, cysE, resulted in the largest stationary-phase fitness defect in our experiment. Thus, our screen provided evidence for a link between a sulfur assimilation/cysteine biosynthesis pathway and stationary-phase physiology.

Recent studies have significantly advanced our understanding of the roles of carbon and nitrogen metabolism in Brucella physiology and infection (17, 18), but our knowledge of sulfur metabolism in Brucella spp. is comparatively limited. This study defines the role of a sulfur metabolism pathway in the growth, stress survival, and infection biology of Brucella. Specifically, we show that cysE-dependent biosynthesis of cysteine influences culture yield and stationary-phase entry in B. ovis. We further provide genetic evidence that two related CysK-family enzymes, CysK1 and CysK2, function redundantly at a step downstream of CysE to produce cysteine. B. ovis strains lacking cysE or both cysK1 and cysK2 are compromised in peroxide stress survival and in colonization of mammalian host cells.

RESULTS

B. ovis cysE Tn-himar mutant strains have a fitness defect in stationary phase.

We inoculated ∼1.5 × 10⁹ B. ovis RB Tn-himar organisms into brucella broth in triplicate and collected samples at intervals throughout the growth curve, at optical densities at 600 nm (OD₆₀₀) of 0.05, 0.12, 0.9, and 2.4 (corresponding to early logarithmic, logarithmic, late logarithmic, and stationary phases). Barcodes were PCR amplified, sequenced, and tallied as previously described (19) to assess the relative abundance of each mutant strain in each sample. Our analysis yielded composite fitness scores for 2,638 of 3,391 annotated genes in B. ovis (see Data Set S1 in the supplemental material). Data for 118 mutants that exceeded a t-like test significance threshold of ≥4 are presented in Fig. S1 (see also Materials and Methods). We observed the largest relative fitness score changes at an OD₆₀₀ of 2.4 (i.e., stationary phase) in this data set.

To more rigorously assess mutants with fitness values that varied as a function of growth phase, we further filtered the genes to include only those with a fitness score of ≥|1| at at least one time point (see Materials and Methods). We hierarchically clustered the 64 genes that passed this cutoff (Fig. 1A; Data Set S1) and divided these clustered genes into four groups that displayed different fitness patterns throughout the growth curve (Fig. 1B). Mutations in group 1 genes resulted in no fitness defect during exponential growth but did result in a fitness defect at an OD₆₀₀ of 2.4 (i.e., stationary phase). Genes in group 2 had negative fitness scores throughout the growth curve. These two groups contained the majority of mutants. Group 3 (four genes) had positive fitness scores in log phase and a negative fitness score in stationary phase, while group 4 (three genes) had null or positive fitness scores at all phases of the growth curve. We clustered these genes by predicted functional category (Fig. S2A; Data Set S2) and found that genes encoding purine metabolism enzymes and tRNA modification enzymes were enriched in group 1. However, the gene with the lowest fitness score in stationary phase, BOV_RS06060 (old locus tag, BOV_1224), is annotated as serine-O-acetyltransferase (cysE) (Fig. 1; Fig. S2B). As such, we chose to further characterize the function of cysE in B. ovis.

FIG 1.

Assessment of B. ovis mutant strain fitness as a function of growth phase identifies cysE as a determinant of stationary-phase fitness. (A) Heat map showing mean fitness scores (n = 3) of B. ovis mutant strains harboring barcoded transposon insertions in nonessential genes. Each row represents a gene, and each column is a point during the growth curve in BB (OD₆₀₀). Genes with a t-like significance score of ≥|4| and a fitness value of ≥|1| at at least one time point are included in the heat map. Genes were hierarchically clustered (left), which yielded four main groups (right). The fitness profile of strains harboring RB Tn-himar insertions in cysE is marked with an asterisk on the right of the heat map. (B) Alternative representation of data in panel A, where the average fitness score for each mutant (gene) is plotted as a function of time point. Individual lines represent genes, and the shaded area delimits the maximum and minimum values within that group. cysE is presented as a thick blue line. Group 1, blue; group 2, red; group 3, yellow; group 4, green.

B. ovis ΔcysE enters stationary phase prematurely and has reduced culture yield in vitro.

B. ovis CysE has high sequence identity (52%) and similarity (73%) with the well-characterized CysE enzymes of Escherichia coli and Salmonella enterica (20) and is clearly classified as CysE in the NCBI conserved domain database (E value = 4.3e⁻¹²²; https://www.ncbi.nlm.nih.gov/Structure/cdd). This protein is therefore predicted to execute the initial step in cysteine biosynthesis, specifically, the addition of an acetyl group from acetyl coenzyme A (acetyl-CoA) to serine, producing O-acetylserine (Fig. 2A). To confirm the cysE stationary-phase phenotype observed in the RB transposon insertion sequencing (Tn-Seq) experiment (Fig. 1), we built a B. ovis strain harboring an in-frame deletion of cysE (ΔcysE). We grew the ΔcysE mutant in parallel with wild-type (WT) B. ovis ATCC 25840 (Fig. 2B) and observed that the ΔcysE mutant enters stationary phase earlier and terminates growth at a lower density than the WT, thus corroborating the Tn-Seq result. This phenotype was complemented by the addition of 4 mM cysteine to the medium (Fig. 2B) or by ectopic expression of the cysE gene from a lac promoter in the presence of IPTG (isopropyl-β-d-thiogalactopyranoside) (Fig. 2C). Ectopic overexpression of cysE in a WT B. ovis background did not modify growth kinetics or the growth curve shape compared to an empty vector control (WT/pSRK-EV) (Fig. 2C). We conclude that cysE and cysteine biosynthesis are necessary for normal B. ovis growth yield in brucella broth.

FIG 2.

The ΔcysE strain enters stationary phase prematurely; this growth defect is rescued by addition of cysteine to the growth medium. (A) Schematic of cysteine biosynthesis from serine. Teal arrows indicate cysteine biosynthesis enzymes annotated in the Brucella ovis genome (RefSeq accession no. NC_009505 and NC_009504). Enzymes: CysE (BOV_RS06060, cysE; O-acetylserine transferase) and CysK1 (BOV_RS09280, cysK; cysteine synthase A). (B) Representative growth curves of wild-type (WT; circles) and ΔcysE (triangles) strains with (gray and ochre, respectively) or without (black and teal, respectively) addition of 4 mM cysteine (Cys) to the growth medium. (C) Representative growth curves of the WT carrying the pSRK empty vector (EV; black circles) or pSRK-cysE (gray squares) and the ΔcysE strain carrying pSRK (teal triangles) or pSRK-cysE (purple diamonds) in BB with 1 mM IPTG and 50 μg/ml Kan. Error bars represent standard deviations for technical replicates in representative experiments.

cysK1 and cysK2 function redundantly in cysteine biosynthesis.

CysK catalyzes the step in cysteine biosynthesis subsequent to CysE, namely, the elimination reaction in which the acetyl group on O-acetylserine is displaced by sulfide to form cysteine (20) (Fig. 2A). Given the stationary-phase phenotype of the ΔcysE mutant and the fact that this defect was chemically complemented by cysteine, we expected that mutations in the cysteine synthase gene (cysK) would phenocopy ΔcysE. However, strains with transposon insertions in the locus BOV_RS09280 (old locus tag, BOV_1893), annotated as cysK in the NCBI RefSeq database, grew like the wild type (Data Set S1). Growth of a strain harboring an in-frame deletion of BOV_RS09280 also grew the same as WT B. ovis in brucella broth (Fig. 3A), confirming the Tn-Seq result.

FIG 3.

The stationary-phase phenotype of a ΔcysK1 ΔcysK2 (BOV_RS05050) double deletion phenocopies ΔcysE and is rescued by cysteine. (A) Growth of wild-type (black circles), ΔcysK1 (black diamonds), ΔcysK2 (black squares), and ΔcysK1 ΔcysK2 (teal triangles) strains in BB. (B) Growth of the strains shown in panel A with 4 mM cysteine added to the broth. (C) Growth curves of the WT carrying the pSRK empty vector (ev; black circles) and the ΔcysK1 ΔcysK2 mutant carrying either pSRK-cysK1 or pSRK-cysK2 (purple triangles) grown with 50 μg/ml Kan and 1 mM IPTG. Error bars represent standard deviations for technical replicates and may be smaller than the symbols. Growth curves were conducted at least three independent times. A representative curve is shown for each.

We considered that the lack of an apparent growth defect in the ΔcysK strain may be due to the presence of other genes with CysK activity. A possible candidate for such a gene is locus BOV_RS05050 (old locus tag, BOV_1018), which encodes a protein with 37% sequence identity and 53% similarity to the product of BOV_RS09280. Here, we refer to BOV_RS09280 as cysK1 and BOV_RS05050 as cysK2. Strains with transposon insertions in cysK2 also yielded a wild-type phenotype in our Tn-Seq experiment (Data Set S1), and a strain harboring an in-frame deletion of cysK2 (ΔcysK2) likewise grew the same as WT B. ovis. However, a ΔcysK1 ΔcysK2 strain exhibited a stationary-phase/growth yield phenotype similar to that of the ΔcysE mutant (Fig. 3A). Like the ΔcysE phenotype, the ΔcysK1 ΔcysK2 phenotype was chemically complemented by the addition of cysteine to the medium (Fig. 3B). The growth defect of the ΔcysK1 ΔcysK2 strain was genetically complemented by expressing either cysK1 or cysK2 from a plasmid (Fig. 3C). We conclude that these two genes have redundant functions in the cysteine biosynthesis pathway.

B. ovis ΔcysE and ΔcysK1 ΔcysK2 strains are sensitive to exogenous H₂O₂ stress.

Cysteine is, of course, important for protein synthesis. It is also one of the three amino acids that comprise glutathione (GSH) (Fig. 4A), which plays a central role in the mitigation of a variety of stressors in bacteria (21), including oxidative stress. We hypothesized that defects in cysteine biosynthesis would have consequences for GSH synthesis and sensitize cells to oxidative stress. We thus attempted to complement the ΔcysE growth phenotype by adding GSH to the medium (Fig. 4B). GSH supplementation partially complemented the ΔcysE growth yield defect. GSH limitation may directly contribute to premature entry of B. ovis ΔcysE into stationary phase, or GSH addition may restore cysteine homeostasis upon GSH catabolism. Since GSH is known to be involved in decomposition of hydrogen peroxide to water (21) (Fig. 4A), we assessed whether the ΔcysE mutant was more sensitive to H₂O₂ stress. We grew the WT and the ΔcysE strain to stationary phase, washed the cells, treated them with H₂O₂ for 1 h in phosphate-buffered saline solution, and then enumerated CFU. The ΔcysE strain was ∼2,000 times more sensitive to H₂O₂ than the WT, and this sensitivity was rescued by the addition of either cysteine or glutathione to the medium during growth (Fig. 4C). The hydrogen peroxide sensitivity phenotype of the ΔcysE strain was genetically complemented by expression of cysE from the lac promoter of the pSRK plasmid (ΔcysE/pSRK-cysE) (Fig. 4D). We further tested the sensitivity of the ΔcysK1 ΔcysK2 mutant to hydrogen peroxide treatment. Like the ΔcysE mutant, the ΔcysK1 ΔcysK2 strain was highly sensitive to peroxide treatment. The ΔcysK1 ΔcysK2 peroxide survival phenotype was chemically complemented by the addition of cysteine or glutathione to the medium (Fig. 4E) and was genetically complemented by ectopic expression of either cysK1 or cysK2 (Fig. 4F).

FIG 4.

B. ovis ΔcysE is sensitive to H₂O₂ treatment; the ΔcysE growth defect and peroxide sensitivity are mitigated by glutathione. (A) Schematic of glutathione metabolism. Enzymes annotated in the Brucella ovis genome (RefSeq accession no. NC_009505 and NC_009504) are indicated in bold: GshA (BOV_RS13935; glutamate-cysteine ligase), GshB (BOV_RS10075; glutathione synthase), and Gor (BOV_RS04850; glutathione disulfide reductase). GSH, glutathione (reduced state); GSSH, glutathione disulfide (oxidized state). (B) Growth of wild-type (circles) and ΔcysE (triangles) strains in BB with (gray and ochre, respectively) or without (black and teal, respectively) 4 mM GSH added to the medium. Error bars represent standard deviations and may be smaller than the symbols. (C) Hydrogen peroxide survival assay showing the log₂ ratio of CFU of treated (20 mM H₂O₂) versus untreated (mock PBS) cultures. Black bars represent wild-type and teal bars represent ΔcysE strains. Addition of either 4 mM cysteine (+ Cys) or 4 mM GSH (+ GSH) to the medium is indicated by the shaded boxes. GSH and cysteine were washed away from the culture prior to peroxide treatment. (D) Same assay as for panel C but with strains carrying plasmids to test genetic complementation. Strains were treated with 15 mM H₂O₂. Black bars represent the WT carrying an empty vector (WT/pSRK-EV), teal bars represent the ΔcysE strain carrying an empty vector (ΔcysE/pSRK-EV), and purple bars represent the complemented ΔcysE strain (ΔcysE/pSRK-cysE). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. P values were calculated using one-way analysis of variance (ANOVA) (followed by Dunnett’s multiple-comparison test, relative to the ΔcysE strain [C] or the ΔcysE/ev strain [D]). (E) Hydrogen peroxide survival assay as for panel C, comparing wild-type B. ovis (black bars) to the ΔcysK1 ΔcysK2 strain (teal bars). (F) As for panel D but with strains carrying plasmids to test genetic complementation. Black bars represent B. ovis carrying the empty vector pSRK-EV (WT/ev), teal bars represent the ΔcysK1 ΔcysK1 strain with pSRK-EV (ΔΔ/ev), and purple bars represent the ΔcysK1 ΔcysK2 strain carrying either pSRK-cysK1 (ΔΔ/1, dark purple) or pSRK-cysK2 (ΔΔ/2, light purple). Error bars represent standard errors of the means for 3 or 4 independent experiments.

ΔcysE and ΔcysK1 ΔcysK2 strains have reduced viability in the intracellular niche.

Brucella spp. primarily reside inside mammalian host cells. There are many challenges to growth and survival in the intracellular niche, including nutrient limitation and exposure to stressors such as reactive oxygen species (ROS) (6). Given the in vitro growth and hydrogen peroxide sensitivity phenotypes of the cys mutants, we tested whether fitness of the ΔcysE and ΔcysK1 ΔcysK2 strains was compromised in the intracellular niche. We infected a human monocytic cell line, THP-1, that we differentiated into macrophage-like cells. Although entry (2 h postinfection [p.i.]) into the macrophage was unaffected by deletion of cysE or of cysK1 and cysK2, there was a significant loss in recoverable CFU of the cys mutant strains by 24 h p.i. relative to the WT (Fig. 5A and B). Thus, B. ovis ΔcysE and ΔcysK1 ΔcysK2 enter host cells like the WT, but their fitness is compromised after entry.

FIG 5.

B. ovis ΔcysE has reduced fitness in the intracellular niche of human macrophage-like cells and an ovine testis epithelial cell line. Log₁₀ CFU per well of the WT (black circles) or the ΔcysE strain (teal triangles) isolated from infected THP-1 (A and C) or OA3.ts (D) mammalian cells enumerated at 2, 24, and 48 h p.i. (A and D) or at 2, 4, 8, 12, and 24 h p.i. (C). (B) Log₁₀ CFU/well of WT (black circles) or ΔcysK1 ΔcysK2 (teal triangles) strains isolated from infected THP-1 cells enumerated at 2, 24, and 48 h p.i. P values comparing recovered B. ovis 24 h p.i. were calculated using an unpaired t test (A, B, and D). *, P < 0.05; ***, P < 0.001; ****, P < 0.0001. Infections were repeated 3 to 5 independent times. Error bars represent standard errors within the representative experiment.

In an effort to distinguish the relative contributions of intracellular killing and cysteine (nutritional) limitation on reduced fitness of the cys mutants, we enumerated CFU recovered from THP-1 cells at time points between 2 and 24 h p.i. Wild-type and ΔcysE strains exhibited identical CFU loss between 2 and 8 h p.i. The WT began to replicate by 12 h, but ΔcysE CFU continued to decrease up to 24 h (Fig. 5C). The rate of increase in the number of recoverable CFU between 24 h and 48 h p.i. is similar between the WT and the cys mutants, providing evidence that B. ovis has access to cysteine in this environment (Fig. 5A and B). An attempt to interrogate a later time point (72 h) was confounded by egressing bacteria and subsequent killing by gentamicin in the tissue culture medium (Fig. S3). The intracellular infection defect we observed was complemented by expression of cysE from a plasmid (Fig. S4A). We attribute partial complementation to the fact that cysE was expressed from a lac promoter on a replicating plasmid; there are challenges with full induction of transgenes from heterologous promoters in an intracellular infection context.

Given the ability of Brucella to infect multiple mammalian cell types, we next tested whether the in vitro infection phenotype of the cys mutants was particular to macrophages. We infected a sheep testis epithelial cell line (OA3.ts) (22), which is derived from a host tissue type that is relevant to B. ovis infection. OA3.ts entry was unaffected by the lack of cysE, but numbers of recovered CFU were significantly lower for the ΔcysE mutant by 24 h p.i. (Fig. 5D). This phenotype was partially complemented by ectopic expression of cysE from a plasmid (Fig. S4B). The magnitude of the ΔcysE defect 24 h p.i. was greater in THP-1 macrophages than in the OA3.ts epithelial line (about 64-fold versus 16-fold, respectively) (Fig. S5). These in vitro infection data provide evidence that an intact cysteine metabolism system promotes B. ovis fitness in intracellular niche of multiple mammalian cell types.

DISCUSSION

A genome-scale search for B. ovis stationary-phase mutants leads to cysteine metabolism.

B. ovis is a widespread ovine pathogen that remains an understudied member of the genus Brucella. Using an RB Tn-Seq approach, we sought to identify genes that are important for B. ovis growth and/or survival in the late phase of axenic broth culture (i.e., stationary phase), with a larger goal of uncovering genes that are important for fitness in the intracellular environment. We identified multiple genes for which Tn-himar disruption resulted in reduced fitness in stationary phase. Among the expected mutants in this data set is rsh (BOV_RS03230), which controls the stringent response (23). Additionally, genes involved in purine metabolism, including purF, have reduced fitness in dense culture. In Mycobacterium smegmatis, PurF influences survival during stationary phase (24), and purine metabolism is known to be important for growth of multiple microbes in the intracellular and extracellular environments (25, 26). Multiple genes with a predicted role in tRNA modification also have diminished fitness in stationary phase. tRNA modification enzymes have roles in translation quality control and can function to direct translation of specific transcripts under particular growth conditions (27). Given the phenotypes of tRNA modification mutants in stationary phase, it may be the case that these genes play a role in regulation of Brucella ovis physiology in the intracellular niche.

Tn-himar strains with insertions in cysE had the most diminished fitness in stationary phase, and cysE was therefore selected for follow-up studies. Sulfur and cysteine metabolism are central to microbial growth and have been well studied in numerous pathogens (28). In Brucella spp., our understanding of sulfur metabolism is relatively limited, though biosynthesis of sulfur-containing amino acids—cysteine and methionine—has been implicated in Brucella melitensis 16M infection of mice (29). Based on the high level of sequence identity/similarity to well-studied CysE enzymes and established structural data on Brucella abortus CysE (30), B. ovis CysE is presumed to catalyze biosynthesis of O-acetylserine from acetyl-CoA and serine. The subsequent step in biosynthesis of cysteine from O-acetylserine requires displacement of the acetyl group by sulfide, a reaction that is catalyzed by CysK in many bacteria. Our growth data clearly implicate cysE in the cysteine biosynthesis pathway, as the in vitro growth defect of the ΔcysE strain is rescued by the addition of cysteine. These results support published data showing that Brucella spp. can assimilate cysteine as an exogenous organic sulfur source (31, 32).

Surprisingly, the growth phenotypes of strains with Tn-himar insertions in the gene annotated as cysK in the RefSeq database did not differ from that of the wild type, which suggests redundancy at this biosynthetic step. Consistent with this hypothesis, we present genetic evidence that two related enzymes, CysK1 and CysK2, function redundantly to produce cysteine. It is possible that CysK1 and CysK2 do not catalyze the same reaction but rather determine cysteine biosynthesis through two distinct routes. A recent study reported such a case for the cystathionine β-synthase of Helicobacter pylori, which retains some O-acetylserine sulfhydrylase activity (33); this enzyme shares primary structure features with B. ovis CysK2. CysK-family enzymes can also have functions beyond direct involvement in cysteine metabolism (34), which may influence interpretation of our results. Notably, B. abortus CysE (serine O-acetyltransferase) and CysK2 do not form a cysteine synthase complex (CSC) in vitro (35). This supports a model in which CysK2 participates in cysteine synthesis via a mechanism that differs from that catalyzed by the typical CysE-CysK CSC. We postulate that CysK1, rather than CysK2, binds to CysE to form the CSC in Brucella. The development of a defined medium that supports the growth of B. ovis would greatly facilitate future study of CysK1 and CysK2 functions in cells. Exploration of the possible intracellular fitness advantage gained by redundancy at the CysK step of cysteine biosynthesis is an interesting area for future investigation.

Cysteine, glutathione, and hydrogen peroxide stress.

The growth and peroxide survival defects of the ΔcysE mutant were partially rescued by addition of cysteine or glutathione to the medium. Though elevated intracellular cysteine enhances susceptibility to hydrogen peroxide stress in Escherichia coli (36, 37), we did not observe peroxide sensitization of WT or ΔcysE B. ovis upon addition of 4 mM cysteine. In our assay, 4 mM cysteine was consistently more protective than 4 mM GSH. GSH is an important redox control molecule, but the protective effect of GSH supplementation against H₂O₂ may be indirect. Specifically, it is possible that B. ovis transports and metabolizes some of the GSH to release cysteine, which is one of the three component amino acids of GSH. B. ovis is predicted to encode a γ-glutamylcyclotransferase (BOV_RS09395), which catalyzes the cleavage of GSH to form pyroglutamic acid and l-cysteinylglycine (38). The l-cysteinylglycine dipeptide could then be separated by peptidases to release cysteine. We nonetheless favor a model in which diminished GSH production (as a result of abolished cysteine production) in the ΔcysE strain directly affects H₂O₂ detoxification and growth yield of this strain. Glutathione metabolism is important in B. ovis: the gshA biosynthesis gene is essential, based on our previously published Tn-Seq data set (16). Moreover, Tn-himar insertions in BOV_RS04850 (old locus tag, BOV_0978), which is predicted to encode a glutathione-disulfide reductase (gor) that reduces GSSG to GSH, resulted in a significant fitness disadvantage throughout the growth curve.

Cysteine and growth and the intracellular niche.

Our study provides evidence that cysteine biosynthesis contributes to B. ovis fitness inside mammalian host cells. Strains harboring deletions of cysE or both cysK1 and cysK2 were not defective in host cell entry but had significantly reduced recoverable CFU at 24 and 48 h p.i in human macrophage-like and ovine epithelial cell lines. Reduced numbers of recoverable CFU of the cys mutants at 24 and 48 h can be attributed to defects that are manifested between 2 and 24 h p.i. It is difficult to fully discern the relative contributions of nutritional restriction and enhanced oxidative stress sensitivity to attenuation of the ΔcysE and ΔcysK1 ΔcysK2 strains in vitro. The similarity in the observed rates at which recoverable CFU increase between 24 and 48 h suggests that cysteine levels are not limiting, at least after 24 h. The endoplasmic reticulum (ER)-derived replicative Brucella-containing vacuole (rBCV) supports bacterial replication (39) and can be established as early as 12 h p.i. (40); based on our data, we conclude that this compartment contains enough cysteine or cysteine-containing peptides to support growth. Of note, the defect of the ΔcysE strain was more pronounced in macrophage-like cells than in the ovine testis epithelial line. Sensitivity of the ΔcysE and ΔcysK1 ΔcysK2 mutants to ROS may underlie this difference in fitness between cell lines, as macrophages typically have a more robust respiratory burst than epithelial cells (41). A significant decrease in recoverable CFU in the ΔcysE strain was evident by 4 h p.i. This is a time point before intracellular Brucella replication occurs (40), supporting a model in which increased sensitivity to host killing underlies the reduced fitness of ΔcysE and ΔcysK1 ΔcysK2 strains in host cells.

Previous B. abortus Tn-Seq studies by Sternon and colleagues (42) did not identify cysE as a gene that was important for infection of RAW 264.7 macrophages. We observed significantly reduced B. ovis ΔcysE CFU relative to WT in RAW 264.7 cells by 24 h (Fig. S6), but the experiment reported by Sternon et al. and our experiment differ in several ways. Indeed, the importance of cysteine metabolism in intracellular growth and/or survival may vary between B. ovis and B. abortus and between mammalian cell lines. Nonetheless, our data clearly provide evidence that a cysteine anabolism pathway in B. ovis is important for growth, stress survival, and fitness in the intracellular niche.

Cysteine and methionine metabolic pathways are attractive targets to combat various pathogens (28) because, in mammals, these amino acids must be acquired from diet. Thus, compounds that disrupt cysteine metabolism are not predicted to have direct negative effects on mammalian metabolism. In fact, O-acetylserine sulfhydrylase (OASS; i.e., cysK) inhibitors are under investigation as therapeutics for Mycobacterium tuberculosis infections (43). Our work shows that genetic disruption of cysteine biosynthesis leads to a significant defect in B. ovis fitness within host cells. This pathway is therefore a possible target for combating brucellosis.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Brucella ovis was grown on tryptic soy agar (TSA; Difco Laboratories) plates supplemented with 5% sheep blood (Quad Five) or in brucella broth (BB; Difco Laboratories) (dissolved in tap water) for liquid cultures. Cells were incubated at 37°C with 5% CO₂ supplementation. Kanamycin (Kan; 50 μg/ml), sucrose (5% [wt/vol]), or isopropyl-β-d-1-thiogalactopyranoside (IPTG; GoldBio) at 1 mM or 2 mM was added when required.

Escherichia coli strains were grown in lysogeny broth (LB; Fisher Bioreagents) or on plates containing LB plus 1.5% agar (Fisher Bioreagents) at 37°C with Kan added at a concentration of 50 μg/ml when required. E. coli WM3064, used for conjugation, was grown in the presence of 300 μM diaminopimelic acid (DAP; Sigma-Aldrich), as it is a DAP auxotroph.

Plasmid and strain construction. (i) Deletion plasmid construction.

To build the deletion strains, fragments of approximately 500 bp upstream and downstream of target genes were amplified with KOD Xtreme Hot Start polymerase (Novagen). These fragments were built so that 9 bases at both the 5′ and 3′ ends of the gene were maintained, keeping the gene product in frame to minimize polar effects. Purified DNA from Brucella ovis ATCC 25840 was used as a template. Amplified fragments were gel purified (Thermo Fisher Scientific) and assembled into the pNPTS138 suicide deletion vector (digested with HindIII and BamHI restriction enzymes; New England Biolabs) using Gibson assembly (New England Biolabs).

(ii) Complementation plasmid construction.

To build plasmids for genetic complementation, cysE, cysK1, and cysK2 were PCR amplified from B. ovis ATCC 25840 with KOD Xtreme Hot Start polymerase, gel purified (Thermo Fisher Scientific), and Gibson assembled into pSRK (44) that had been digested with NdeI and KpnI restriction enzymes (New England Biolabs). cysE, cysK1, or cysK2 was cloned downstream of P_lac (lactose; IPTG-inducible promoter).

(iii) Delivery of plasmids to B. ovis.

Constructed plasmids were transformed into chemically competent E. coli Top10 strains for plasmid maintenance. All plasmid inserts were confirmed by PCR and Sanger sequencing, and plasmids were delivered to B. ovis by conjugation using E. coli WM3064 as a donor strain. For conjugation, WM3064 donor strains were mated with B. ovis strains and spotted on TSA blood plates plus DAP and incubated overnight at 37°C in a 5% CO₂ atmosphere. Mating spots were spread on TSA blood plates plus Kan (without DAP) to select for B. ovis plasmid acquisition. When genes were deleted using the pNPTS138 plasmid, merodiploid clones were inoculated in BB overnight to allow a second crossover event and then spread on TSA blood plates plus sucrose (5% [wt/vol]) for counterselection. Single colonies harboring gene deletions were identified by patching clones on TSA blood plates with or without Kan. The putative deleted locus was PCR amplified using gene-flanking primers in Kan-sensitive clones, and the PCR fragment was resolved by gel electrophoresis to test whether the gene had been deleted. For a complete list of strains, plasmids, and primers, see Data Set S3.

Growth curves.

Cells were inoculated from ∼48-h-old TSA blood plates into BB at densities ranging from an OD₆₀₀ of 0.08 to an OD₆₀₀ of 0.2. Growth was assessed spectrophotometrically to measure OD₆₀₀. Growth curves were conducted at least three independent times with two or three technical replicates in each experiment. Representative curves are shown for each set of strains. Where indicated, cysteine (4 mM), GSH (4 mM), Kan (50 μg/μl), or IPTG (1 mM) was added to the liquid medium upon the start of the growth experiment.

H₂O₂ survival assays.

Cells were grown overnight in BB to stationary phase (OD₆₀₀ of ∼2). Cells were pelleted and resuspended in phosphate-buffered saline (PBS; Sigma) to achieve an OD₆₀₀ of 0.15. A 200-μl portion of cells was added to 1.8 ml of PBS or PBS supplemented with fresh H₂O₂ (15 or 20 mM final concentration), bringing the final OD₆₀₀ to 0.015. Cells were then incubated at 37°C in 5% CO₂ for 1 h before aliquots of a 10-fold serial dilution series were spotted on TSA blood plates. CFU were enumerated after 48 h incubation at 37°C in 5% CO₂. Experiments were repeated at least three times with each sample in duplicate or triplicate in each experiment.

DNA extractions.

Cells from 1 ml of stationary-phase culture were pelleted by centrifugation, washed once in PBS, and resuspended in 100 μl TE buffer (10 mM Tris-HCl, 1 mM EDTA; pH 8.0) supplemented with 1 μg/ml RNase A. Cells were lysed by addition of 0.5 ml GES lysis solution (5 M guanidinium thiocyanate, 0.5 M EDTA [pH 8.0], 0.5% [vol/vol] Sarkosyl) and 15 min incubation at 60°C. Cold 7.5 M ammonium acetate (0.25 ml; Fisher Bioreagents) was added, and mixture was incubated on ice for 10 min. Chloroform (0.5 ml; Fisher Bioreagents) was added to separate the DNA, and samples were vortexed and centrifuged. The aqueous top phase was moved to a fresh 1.5-ml centrifuge tube and 0.54 volume of cold isopropanol was added to precipitate the DNA. After centrifugation, isopropanol was discarded, and pellets were washed three times in 70% ethanol before being resuspended in TE buffer. The concentration and purity of the extracted DNA were determined spectrophotometrically (NanoDrop One; Thermo Scientific).

Barcoded Tn-Seq.

A B. ovis RB Tn-himar library was built and mapped as described previously (31). Briefly, E. coli APA752 (a WM3064 donor strain carrying a pKMW3 mariner transposon library) was conjugated into B. ovis bcaA1 (16) under atmospheric CO₂ conditions. Kan-resistant colonies were collected, grown to an OD₆₀₀ of 0.6, and frozen in 1-ml aliquots. Genomic DNA was extracted and the Tn insertion sites mapped as previously described (19).

To identify genes that confer a fitness advantage in stationary phase, the B. ovis Tn-himar library was inoculated in BB in a 5% CO₂ environment in triplicate at an OD₆₀₀ of 0.0025. An aliquot of each initial culture was collected as the reference time point. Cultures were then grown to stationary phase, with samples harvested throughout the growth curve at OD₆₀₀ of 0.05, 0.12, 0.9, and 2.4. Cells from each sample were pelleted by centrifugation and resuspended in water. Barcodes were amplified from approximately 1.5 × 10⁸ cells per PCR (see Table S1) with primers that both amplified the barcodes and added indexed adaptors (19). Amplified barcodes were then pooled, purified, and sequenced on an Illumina HiSeq 4000 system.

Fitness scores for each gene were calculated following the protocol of Wetmore and colleagues (19) using scripts available at https://bitbucket.org/berkeleylab/feba. Genes for which 2 of 3 samples had at least one time point with a t-like statistical significance score (19) of ≥|4| were included in subsequent analyses. A heat map of fitness scores of genes passing this filter is shown in Fig. S1, and the raw fitness data are in Data Set S1.

Finally, we averaged the fitness values of the three replicates and kept mutants that had an average fitness score of ≥|1| at at least one time point. Mutants in this group with a standard deviation of ≥0.75 were manually inspected, and extreme outlier points were removed from a total of six genes. The genes with adjusted average and standard deviation values are shown in red in Data Set S1. A heat map of averaged fitness values is shown in Fig. 1.

Tissue culture.

All tissue culture cells were grown at 37°C with 5% CO₂ supplementation. THP-1 cells (ATCC TIB-202) were cultured in RPMI 1640 medium (Gibco) containing 10% fetal bovine serum (FBS; Fisher Scientific). The RAW 264.6 (ATCC TIB-71) and the OA3.ts (ATCC CRL-6546) cells were grown in Dulbecco’s modified Eagle medium (DMEM; Gibco) supplemented with 10% FBS.

Infection assays.

THP-1 cells were seeded at a concentration of 10⁵ cells/well in 96-well plates, and phorbol myristate acetate (PMA) at a final concentration of 50 ng/μl was added to induce differentiation into macrophage-like cells for 48 to 96 h prior to infection. OA3.ts cells were seeded at a density of 5 × 10⁴ cells/well in 96-well plates for 24 h prior to infection. B. ovis cells were resuspended from a 48-h-old plate in RPMI plus 10% FBS or in DMEM plus 10% FBS and added to tissue culture plates on the day of infection at a multiplicity of infection (MOI) of 100 for THP-1 and at an MOI of 1,000 for OA3.ts cells. For infections with the complementation strains carrying the pSRK plasmid, the Brucella strains were streaked on TSA blood plates with Kan and IPTG 48 h prior to infection, and 2 mM IPTG was added to the tissue culture medium throughout the duration of the experiment. Plates were spun for 5 min at 150 × g and incubated for 1 h at 37°C in 5% CO₂. Fresh medium was supplied containing 50 μg/ml of gentamicin and incubated for another hour. Cells were then washed once with PBS and once in H₂O and then lysed with H₂O for 10 min at room temperature at 2 h, 24 h, and 48 h postinfection. Lysates were serially diluted, spotted on TSA blood plates, and incubated at 37°C in 5% CO₂ for 48 h to enumerate CFU. Experiments were repeated at least 3 times with three technical replicates.