Coxiella burnetii RpoS Regulates Genes Involved in Morphological Differentiation and Intracellular Growth

The Q fever bacterium Coxiella burnetii has spore-like environmental stability, a characteristic that contributes to its designation as a potential bioweapon. Stability is likely conferred by a highly resistant, small cell variant (SCV) stationary-phase form that arises during a biphasic developmental cycle. Here, we define the role of the alternative sigma factor RpoS in regulating genes associated with SCV development. Genes involved in stress responses, amino acid transport, cell wall remodeling, and type 4B effector secretion were dysregulated in the rpoS mutant. Cellular impairments included defects in intracellular growth, cell wall structure, and resistance to oxidants. These results support RpoS as a central regulator of the Coxiella developmental cycle and identify developmentally regulated genes involved in morphological differentiation.

ABSTRACT

Coxiella burnetii, the etiological agent of Q fever, undergoes a unique biphasic developmental cycle where bacteria transition from a replicating (exponential-phase) large cell variant (LCV) form to a nonreplicating (stationary-phase) small cell variant (SCV) form. The alternative sigma factor RpoS is an essential regulator of stress responses and stationary-phase physiology in several bacterial species, including Legionella pneumophila, which has a developmental cycle superficially similar to that of C. burnetii. Here, we used a C. burnetii ΔrpoS mutant to define the role of RpoS in intracellular growth and SCV development. Growth yields following infection of Vero epithelial cells or THP-1 macrophage-like cells with the rpoS mutant in the SCV form, but not the LCV form, were significantly lower than that of wild-type bacteria. RNA sequencing and whole-cell mass spectrometry of the C. burnetii ΔrpoS mutant revealed that a substantial portion of the C. burnetii genome is regulated by RpoS during SCV development. Regulated genes include those involved in stress responses, arginine transport, peptidoglycan remodeling, and synthesis of the SCV-specific protein ScvA. Genes comprising the dot/icm locus, responsible for production of the Dot/Icm type 4B secretion system, were also dysregulated in the rpoS mutant. These data were corroborated with independent assays demonstrating that the C. burnetii ΔrpoS strain has increased sensitivity to hydrogen peroxide and carbenicillin and a thinner cell wall/outer membrane complex. Collectively, these results demonstrate that RpoS is an important regulator of genes involved in C. burnetii SCV development and intracellular growth.

IMPORTANCE The Q fever bacterium Coxiella burnetii has spore-like environmental stability, a characteristic that contributes to its designation as a potential bioweapon. Stability is likely conferred by a highly resistant, small cell variant (SCV) stationary-phase form that arises during a biphasic developmental cycle. Here, we define the role of the alternative sigma factor RpoS in regulating genes associated with SCV development. Genes involved in stress responses, amino acid transport, cell wall remodeling, and type 4B effector secretion were dysregulated in the rpoS mutant. Cellular impairments included defects in intracellular growth, cell wall structure, and resistance to oxidants. These results support RpoS as a central regulator of the Coxiella developmental cycle and identify developmentally regulated genes involved in morphological differentiation.

INTRODUCTION

Coxiella burnetii is a wide-ranging zoonotic pathogen that causes Q fever in humans. Reservoir hosts are numerous, including ticks and many species of mammals (1). Typically, humans are infected when encountering contaminated aerosols generated by infected domestic livestock, with dairy cattle, sheep, and goats being the primary reservoirs for human transmission (2). Abortion waves in sheep and goats caused by C. burnetii deposit heavily contaminated birth products into the environment that, when desiccated, can easily transmit disease (3). In humans, acute Q fever generally manifests as severe flu-like illness. Localized chronic infections, such as endocarditis or vascular disease, can occur, especially in immunocompromised individuals (1, 4, 5).

In the environment, C. burnetii is a naturally obligate intracellular bacterium that must invade host cells to replicate. C. burnetii preferentially targets primary mononuclear phagocytes, such as alveolar macrophages, to cause disease (6, 7). Following host cell internalization, C. burnetii directs biogenesis of a specialized membrane-bound compartment called the Coxiella-containing vacuole (CCV). Formation of the CCV is facilitated by the PmrAB-regulated Dot/Icm type 4B secretion system (T4BSS) that secretes effector proteins into the host cell cytoplasm (8–11). C. burnetii effectors manipulate several host cell processes, such as autophagic, secretory, and endolysosomal trafficking (12–15). Subsequently, the CCV resembles an archetypical autophagolysosome, with traditional autophagy and endolysosomal markers decorating the vacuole (reviewed in reference 16). Indeed, the CCV maintains inhospitable conditions indicative of an autolysosome, including acidic pH (∼4.8) and acid hydrolase activity (17–20). Reactive oxygen species (ROS) generated by infected macrophages can also inhibit C. burnetii growth (21).

A hallmark of C. burnetii during in vitro and axenic (host cell-free) growth is a biphasic developmental cycle where bacterial cells morphologically transition from exponential-phase (replicating) large cell variants (LCVs) to stationary-phase (nonreplicating) small cell variants (SCVs) (22). Differentiation of the LCV to the SCV is evident after approximately 4 to 5 days of growth, with nearly homogenous SCVs observed following 10 days of growth (22–24). The different cell variants are largely defined by ultrastructure. SCVs (0.2 to 0.5 μm in size) have condensed chromatin and a thick cell wall/envelope. LCVs (>0.5 μm in size) have dispersed chromatin and a more characteristic Gram-negative cell wall/envelope (22, 23, 25, 26). In addition, SCVs are less metabolically active in axenic buffers (27). In contrast to life cycles of Legionella pneumophila and chlamydial species, where stationary-phase cell types are necessary for perpetuating infection (28, 29), C. burnetii LCV and SCV forms have similar infection rates in cell culture (23, 24). Thus, the role of the SCV appears to be long term viability in the environment, which enables dissemination of C. burnetii to naive hosts. This hypothesis is supported by the pronounced resistance of the SCV to osmotic and mechanical stress relative to the LCV (26, 27, 30, 31).

The Coxiella developmental cycle is morphologically well characterized (26). However, only a few SCV-specific proteins have been identified, including the small, basic DNA-binding proteins Hq1 and ScvA and a few antigens (32–34). Moreover, several small RNAs and the 6S RNA are developmentally regulated by C. burnetii and thought to play roles in differentiation (35). Recent transcriptional profiling revealed subsets of genes associated with SCV development, including those involved in stress responses, arginine metabolism/transport, and cell wall remodeling (25). Accordingly, C. burnetii likely requires tight regulation of these processes to generate the SCV form.

Complex regulation of genes is needed for bacterial survival during environmental stress and host cell infection (36). A prominent regulator of stationary-phase physiology and stress-related adaptation is the alternative sigma factor RpoS. RpoS (sigma S or σ^s) is well characterized in numerous bacterial species and is considered the primary stationary-phase-specific alternative sigma factor of RNA polymerase (36). RpoS is necessary for promoter recognition and transcriptional initiation of genes that encode various stress-related pathways, as well as some virulence factors (36). In fact, in Escherichia coli the RpoS regulon comprises approximately 10% of the genome (37–40). Genes that aid survival of environmental stressors, such as acid (41, 42), ROS (43, 44), heat (45), and starvation (46), are a major part of this regulon. The generally considered close relative of C. burnetii (47), L. pneumophila, undergoes an RpoS-regulated biphasic life cycle inside macrophages by generating replicative and transmissive forms reminiscent of C. burnetii LCVs and SCVs (28, 48). Like L. pneumophila, C. burnetii encodes RpoS (cbu1669). Interestingly, RpoS expression is higher in LCVs than in SCVs (49).

The stress resistance and stationary-phase appearance of the SCV suggest that RpoS regulates genes required for differentiation. To explore this possibility, we deciphered the transcriptomes and proteomes of wild-type and ΔrpoS mutant bacteria in the LCV and SCV forms. Pronounced dysregulation of genes by the SCV ΔrpoS form (SCV ΔrpoS) was revealed, including those involved in stress responses, arginine transport and metabolism, cell wall remodeling, and assembly of the Dot/Icm T4BSS. Aberrant gene regulation corresponded to defective intracellular growth, ROS resistance, and cell wall structure.

RESULTS

SCV ΔrpoS has an intracellular growth defect.

To investigate the role of RpoS in growth and viability, a C. burnetii rpoS deletion mutant was generated and subsequently complemented. Axenically grown 7-day stocks of these strains, along with wild-type bacteria, were used to inoculate ACCM-D, which was incubated for 4 and 14 days to generate LCVs and SCVs, respectively. Significant differences in growth were not observed over 14 days (Fig. 1A). Moreover, the viabilities of LCVs and SCVs were similar between the ΔrpoS mutant, complement, and wild-type strains (Fig. 1B). However, the SCV ΔrpoS mutant showed a significant intracellular growth defect in Vero cells (Fig. 1C) and THP-1 macrophage-like cells (Fig. 1D) relative to the LCV ΔrpoS mutant. Defective intracellular replication of SCV ΔrpoS in Vero cells correlated with smaller, more-condensed CCVs with significantly less area than those generated by wild-type and complement strains (Fig. 1E and F). Collectively, these data suggest that gene regulation by RpoS is important for optimal SCV infection and subsequent replication.

FIG 1.

C. burnetii SCV ΔrpoS has an intracellular growth defect. (A and B) C. burnetii wild-type, ΔrpoS, and ΔrpoS complement (ΔrpoS comp) strains were grown in ACCM-D for 14 days, and samples were harvested for GE (A) and CFU (B) measurements. Error bars represent the standard errors of the mean from three independent biological experiments. (C and D) Vero (C) and THP-1 (D) cells were infected for 7 days with axenically grown C. burnetii wild-type, ΔrpoS, and ΔrpoS complement (comp) LCVs (day 5) or SCVs (day 14). C. burnetii growth is expressed as the mean fold change between 0 and 7 days postinfection from technical duplicates from three independent experiments for Vero cells and technical quadruplicates from two independent experiments for THP-1 cells. Error bars represent the standard errors of the mean. ns, not statistically significant; *, statistically significant difference (P < 0.05); ****, statistically significant difference (P < 0.0001). (E) Confocal fluorescence micrographs of Vero cells infected for 3 days with axenically grown C. burnetii wild-type, ΔrpoS, and ΔrpoS comp LCVs (day 5) or SCVs (day 14). CD63 (red) and C. burnetii (green) are stained by indirect immunofluorescence, and DNA (blue) is stained with DAPI. Images are representative of multiple images from two independent experiments. Scale bar, 10 μm. (F) Area of CCVs generated by C. burnetii wild-type, ΔrpoS, and ΔrpoS comp strains after 3 days of growth in Vero cells, as measured by using ImageJ. Error bars represent the standard errors of the mean from at least 150 cells per strain from two independent experiments. ****, statistically significant difference (P < 0.0001).

RNA sequencing reveals the C. burnetii RpoS regulon.

RNA sequencing (RNA-seq) was performed to define the RpoS regulon associated with C. burnetii developmental forms. LCV ΔrpoS and SCV ΔrpoS had 118 and 652 genes, respectively, dysregulated by >2-fold with P values of ≤0.05 (see Table S1 in the supplemental material). Of the SCV dysregulated genes, 416 were upregulated, indicating substantial RpoS repression of stationary-phase-specific genes or regulators of these genes. In contrast, 237 SCV genes were downregulated. The pronounced dysregulation of 652 genes (approximately 31% of the C. burnetii Nine Mile RSA493 genome) (47) during stationary phase is presumably due to nutrient depletion and other stresses.

Genes associated with SCV differentiation are regulated by RpoS.

The large cohort of RpoS-regulated genes directed us to focus on four subsets of genes that were previously shown to be developmentally regulated during SCV differentiation (25). Thirty-two SCV-associated genes were related to arginine transport/metabolism, cell wall remodeling, and responses to general and oxidative stress. Of these, RNA-seq revealed 25 genes were significantly dysregulated ≥2-fold at day 14 in SCV ΔrpoS, with 7 and 18 up- and downregulated, respectively (Fig. 2A).

FIG 2.

The majority of SCV-associated genes are regulated by RpoS. (A) Fold change in gene expression of previously identified SCV-associated genes by C. burnetii ΔrpoS compared to expression by wild-type bacteria. Transcripts were determined by RNA-seq at 5 and 14 days post-inoculation of ACCM-D, while protein synthesis was quantified by LC-MS/MS at 14 days only. Blue and red indicate a decrease and increase in expression, respectively. For protein expression, −10.00 indicates no peptide fragments were detected in C. burnetii SCV ΔrpoS. Gray indicates no peptide fragments were detected in C. burnetii SCV wild-type or ΔrpoS. (B to D) QuantiGene verification of argR (cbu0480) and artP (cbu0481) (B), oxyR (cbu1476) and ahpC2 (cbu1477) (C), and enhC (cbu1136) and and scvA (cbu1267a) (D) expression after 14 days of incubation in ACCM-D of C. burnetii wild-type, ΔrpoS, and ΔrpoS complement (ΔrpoS comp) strains. Error bars indicate the standard errors of the mean from two independent experiments performed in triplicate. (E) Immunoblots assessing EnhC and ScvA production by C. burnetii SCV wild-type, ΔrpoS, and ΔrpoS complement (ΔrpoS comp) strains grown for 14 days in ACCM-D. Blots were probed with a rabbit anti-ScvA (α-ScvA) or anti-EnhC (α-EnhC) antibody. As a loading control, blots were probed with an anti-elongation factor Ts (α-EFTs) antibody. (F) The consensus RpoS-binding site of the 32 C. burnetii SCV-associated genes was identified by searching 100 bp upstream of each translational start site for the consensus RpoS nucleotide binding site sequence of E. coli (consensus sequence; TCTATACTTAA) (41). The 11-nucleotide sequences of C. burnetii were aligned using WebLogo 3 software (50) to a create a sequence logo demonstrating common nucleotides. Note the conserved nucleotides CTA at nucleotide positions 2 through 4 and the T at nucleotide position 8.

To examine the correlation between transcription and translation, we also performed whole bacterial cell liquid chromatography with tandem mass spectrometry (LC-MS/MS) on SCV wild-type and ΔrpoS strains. In general, protein levels by LC-MS/MS correlated with gene expression by RNA-seq. Overall, 247 proteins were significantly dysregulated ≥2-fold in SCV ΔrpoS. Fifty-five of these proteins were ≥2-fold less abundant, and 58 were undetected in SCV ΔrpoS relative to SCV wild-type bacteria. Furthermore, 134 proteins were ≥2-fold more abundant in SCV ΔrpoS with 133 proteins not detected in either wild-type or ΔrpoS bacteria (Table S2). Of the 25 SCV-associated genes significantly dysregulated by RNA-seq in SCV ΔrpoS, 10 demonstrated correlative protein synthesis while two, artP (cbu0481) and pspC (cbu0774), demonstrated contradictory changes (Fig. 2A). The latter discrepancy may reflect posttranscriptional modifications that affect protein stability.

To validate the RNA-seq results, we used QuantiGene technology to measure transcript levels of independently grown C. burnetii SCVs. Probes specific to a subset of arginine metabolism, cell wall remodeling, and general stress/oxidative stress genes were used. QuantiGene analysis demonstrated similar transcriptional behavior, with argR (cbu0480) and artP (cbu0481) upregulated and enhC (cbu1136) and scvA (cbu1267a) downregulated in SCV ΔrpoS compared to wild-type and complement strains (Fig. 2B and D). In contrast, oxyR (cbu1476) and ahpC2 (cbu1477) were downregulated in the rpoS mutant, in contrast to what was seen by RNA-seq and LC-MS/MS (Fig. 2C). The magnitude of expression differences between RNA-seq and QuantiGene analysis may reflect independent sample variability and/or dynamic range signal differences between the two assays. Nonetheless, the EnhC and ScvA gene and protein expression profiles were further confirmed by immunoblotting, where a marked decrease in protein expression was observed in the SCV ΔrpoS mutant (Fig. 2E).

The Escherichia coli consensus RpoS-binding sequence TCTATACTTAA was used to probe genes in Fig. 2A for possible RpoS regulated promoters (41). We searched 100 bp upstream of the translational start site of the 32 selected SCV-associated genes, with the cutoff for matching nucleotides set at 6 nucleotides. All but spoT (cbu303), htpX (cbu0546), and DNA ligase (cbu1934) had a predicted RpoS-binding site(s) (Fig. S1). Twenty-four with binding sites were dysregulated by the SCV rpoS mutant (Fig. 2A). When the sequences of the predicted RpoS-binding sites were aligned using WebLogo 3 (50), there was considerable nucleotide conservation, with a CTA sequence highly conserved at positions 2 through 4 and a thymine (T) conserved at position 8 (Fig. 2F). RpoS promoter conservation, along with transcriptional and translational data, strongly suggest that genes involved in arginine metabolism, stress responses, cell wall remodeling, and production of the SCV-specific protein ScvA are directly regulated by RpoS during C. burnetii differentiation.

C. burnetii dot/icm genes are regulated by RpoS.

RNA-seq demonstrated that 22 of the 27 dot/icm genes are downregulated ≥2-fold in SCV ΔrpoS. Of the 22 downregulated genes, LC-MS/MS revealed that six corresponding proteins were >2-fold less abundant, eight were undetected in SCV ΔrpoS only, and eight were undetected in both SCV wild-type and ΔrpoS bacteria (Fig. 3A). In corroboration, immunoblots of DotA, IcmK, and IcmX showed considerable decreases in protein levels in SCV ΔrpoS (Fig. 3B). However, minimal complementation was observed. Interestingly, these results parallel the partial complementation of SCV ΔrpoS for intracellular growth (Fig. 1C to F). RNA-seq revealed no ordered pattern of expression between putative effectors (16).

FIG 3.

C. burnetii dot/icm genes are regulated by RpoS. (A) Fold change in gene expression of dot/icm genes by C. burnetii ΔrpoS compared to expression in wild-type bacteria. Transcripts were determined by RNA-seq at 5 and 14 days post-inoculation of ACCM-D, while protein synthesis was quantified by LC-MS/MS at 14 days only. Blue and red indicate a decrease and increase in expression, respectively. For protein expression, −10.00 indicates no peptide fragments were detected in C. burnetii SCV ΔrpoS. Gray indicates no peptide fragments were detected in C. burnetii SCV wild-type or ΔrpoS. (B) Immunoblots assessing IcmX, DotA, and IcmK production by C. burnetii wild-type, ΔrpoS, and ΔrpoS complement (ΔrpoS comp) strains grown for 14 days in ACCM-D. Blots were probed with a rabbit anti-IcmX (α-IcmX), anti-DotA (α-DotA), or anti-IcmK (α-IcmK) antibody. As a loading control, blots were probed with an anti-elongation factor Ts (α-EFTs) antibody.

C. burnetii ΔrpoS mutant has increased sensitivity to hydrogen peroxide.

RNA-seq and protein expression data showed C. burnetii RpoS regulates the oxidative stress genes oxyR (cbu1476) and ahpC2D (cbu1477/1478) which encode an oxidative stress-responsive transcriptional regulator and an alkyl hydroperoxide reductase, respectively (24) (Fig. 2). To test whether RpoS plays a role in controlling the oxidative stress response, we exposed wild-type C. burnetii, the ΔrpoS mutant, and the complemented mutant to 1 mM H₂O₂ over 14 days in ACCM-D. Based on the genome equivalents (GE), the rpoS mutant was unable to grow under this condition, whereas both C. burnetii wild-type and ΔrpoS complement strains began to grow after 8 days, presumably after alleviating oxidative stress caused by the H₂O₂ (Fig. 4). These data support the idea that RpoS regulates genes involved in detoxifying ROS, thereby enabling C. burnetii to persist in the hostile CCV environment.

FIG 4.

C. burnetii ΔrpoS has increased sensitivity to hydrogen peroxide. C. burnetii wild-type, ΔrpoS, and ΔrpoS complement (ΔrpoS comp) strains were grown in ACCM-D for 14 days with or without 1 mM H₂O₂. GE were enumerated every 2 days and plotted over time. Error bars represent the standard errors of the mean from three independent experiments performed in duplicate.

C. burnetii ΔrpoS mutant has increased susceptibility to carbenicillin and a thinner cell wall/outer membrane complex.

It was recently discovered that LCV to SCV transition is associated with prominent cell wall remodeling, with a notable modification being an increase in nonclassical 3-3 cross-links between the peptide stems of peptidoglycan (25). This modification is consistent with dramatic upregulation in the SCVs of five predicted l,d-transpeptidase (Ltd) genes with YkuD domains: cbu0053, cbu0318, cbu0957, cbu1138, and cbu1394 (25). Ltds catalyzes 3-3 cross-links, a modification associated with some stationary-phase bacteria that may confer enhanced cell wall stability (51, 52). By RNA-seq and LC-MS/MS, Ltd gene expression was substantially downregulated in SCV ΔrpoS (Fig. 2A). This alteration would presumably render SCV ΔrpoS more sensitive to β-lactam antibiotics, which target d,d-transpeptidases (penicillin binding proteins) responsible for classical 4-3 cross-links (51, 52). Consistent with this hypothesis, the C. burnetii ΔrpoS mutant transitioning to the SCV form exhibited increased sensitivity to carbenicillin, an antibiotic that inhibits penicillin binding proteins (Fig. 5A). Cryo-electron microscopy revealed a thinner cell wall/outer membrane complex in SCV ΔrpoS, primarily near the cell poles, suggesting an aberrant cell wall structure (Fig. 5B). Collectively, these results demonstrate that RpoS regulates cell wall remodeling genes that modify the peptidoglycan during SCV development.

FIG 5.

The C. burnetii ΔrpoS mutant has increased sensitivity to carbenicillin and a thinner cell wall/outer membrane complex. (A) C. burnetii wild-type, ΔrpoS, and ΔrpoS complement (ΔrpoS comp) strains were grown in ACCM-D over 14 days. At day 4, cultures were adjusted to 200 μg/ml of carbenicillin, and the CFU were enumerated at 4, 8, and 14 days postinoculation. Data were plotted over time as the percent survival of day 4 CFU counts. Error bars indicate the standard errors of the mean from two independent experiments performed in duplicate. (B) Cryo-electron microscopy showing representative morphology of SCV wild-type, ΔrpoS, and ΔrpoS comp strains after 14 days of growth in ACCM-D. Note the thinner, less-dense cell wall/outer membrane complex in the ΔrpoS strain. Scale bar, 200 nm.

DISCUSSION

In the present study, we demonstrate that the alternative sigma factor RpoS globally regulates genes within the C. burnetii genome. A major subset of the RpoS regulon is associated with SCV production during the Coxiella biphasic development cycle. More specifically, genes associated with oxidative stress responses, amino acid acquisition, cell wall remodeling, and Dot/Icm T4BSS assembly are targets of RpoS regulation.

The observation that the C. burnetii ΔrpoS mutant has no growth defect or loss of viability when grown in synthetic medium is interesting considering rpoS mutants in multiple bacterial species lose viability in the stationary phase (53). However, these results phenocopy growth characteristics of a L. pneumophila rpoS mutant, which does not have defects in growth or stationary-phase viability when grown in a rich medium (54). Given these similarities, it is also not surprising that the transcriptional expression profiles of C. burnetii and L. pneumophila rpoS mutants parallel each other. In fact, both C. burnetii and L. pneumophila RpoS globally regulate gene expression of more than 25% of their respective genomes, with similar control of regulators, such as ArgR and CsrA-2, and stress-related genes (55).

Scavenging arginine from the environment is essential for the survival of several bacterial species (56). Typically, genes involved in arginine transport are encoded in an operon that is divergently transcribed from argR, encoding the transcriptional regulator ArgR. Under conditions of excess arginine, ArgR binds arginine and either acts as a repressor or activator of gene expression. It acts as a direct repressor of genes involved in arginine transport or biosynthesis (57). In L. pneumophila, the expression of argR is RpoS regulated and is essential for intracellular growth in amoebae (58). Furthermore, an L. pneumophila argR mutant shows dysregulated expression of >120 genes (58). Given the parallels between the C. burnetii and L. pneumophila developmental cycles and that C. burnetii is an arginine auxotroph (59), it is logical to assume that RpoS regulates C. burnetii arginine metabolism during stationary-phase persistence under the nutrient-limited conditions of the CCV.

In macrophages, C. burnetii persists within a CCV containing ROS. Superoxide anion is generated by phagocyte NADPH oxidase (21). Dismutation of O₂^– into H₂O₂ is catalyzed by superoxide dismutase, with decomposition of H₂O₂ to O₂ + H₂O occurring via catalase activity (60). Interestingly, the catalase (katE) gene of several C. burnetii strains, including the Nine Mile strain used in this study, contains a frameshift mutation that renders the encoded protein nonfunctional (47). Thus, alkyl hydroperoxide reductase (Ahp) systems, such as AhpC2D (CBU1477/1478), presumably play critical roles in degrading H₂O₂, as well as organic peroxides and peroxynitrite. The ahpC2D operon is immediately downstream of the oxidative stress regulator oxyR (cbu1476) and divergently transcribed (47). In L. pneumophila and Mycobacterium tuberculosis, Ahp systems play an integral role in combating ROS, although they are independently regulated from OxyR (61, 62). The C. burnetii ΔrpoS mutant will not grow in H₂O₂-containing medium over a 14-day time course, while wild-type and complemented bacteria, commence growth at roughly 8 days postinoculation. Thus, dysregulated antioxidant systems, such as AhpC2D, of the rpoS mutant presumably fail to detoxify the medium. This is opposed to detoxification by wild-type and complemented bacteria, which eventually generate a medium permissive for growth.

Expression of scvA is drastically downregulated in SCV ΔrpoS. ScvA is a small, basic protein that is considered the gold standard marker for SCV development, with protein expression increasing during stationary phase (24). ScvA binds double-stranded DNA and predominantly associates with the chromatin of SCVs. Consequently, ScvA is proposed to contribute to SCV DNA condensation and to protect DNA from environmental insult (34). ScvA may also be involved in epigenetic regulation, rather than simple DNA packaging. Further study of ScvA is needed to define its function.

A prominent finding of this study is the drastic downregulation of nine genes encoding cell wall remodeling enzymes in SCV ΔrpoS. In wild-type C. burnetii, the expression of these genes considerably increases during SCV differentiation (25). One of the cell wall-associated genes, cbu0419, encodes a predicted peptidoglycan polysaccharide deacetylase (63). This enzyme deacetylates N-acetylglucosamine of peptidoglycan, which renders the peptidoglycan resistant to lysozyme. In turn, cell wall fragments are less likely to be released and detected by cytosolic immune cell sensors (64–67). The expression of enhC (cbu1136) is also greatly decreased in the rpoS mutant. While the function of EnhC in C. burnetii is unknown, in L. pneumophila it is thought to aid intracellular replication by controlling lytic murein transglycosylase activity and the subsequent release of peptidoglycan peptide fragments into the cytosol (68–70). C. burnetii also encodes a lytic murein transglycosylase (CBU0925) that is downregulated in the rpoS mutant. Lytic murein transglycoslases are also known to aid penetration of the cell envelope by large, complex structures, such as type III and type IV secretion systems (71). Thus, C. burnetii RpoS may coordinately regulate these enzymes to evade cytosolic innate immune detection and to facilitate penetration of the cell envelope by the Dot/Icm T4BSS.

Five predicted Ldts are downregulated in SCV ΔrpoS. Ldts catalyze the generation of nonclassical 3-3 cross-links between diaminopimelate molecules in peptidoglycan peptide stems. This contrasts with the activity of d,d-transpeptidases (penicillin binding proteins) that catalyze classical 4-3 peptide cross-links between d-alanine and diaminopimelate (72). The peptidoglycan of the SCV contains a high percentage of 3-3 cross-links relative to the LCV, a modification that is associated with increased expression of Ldts (25). Although the benefit of 3-3 cross-linking is unclear, the shorter distance between glycan strands is thought to increase structural rigidity (51, 52). Like the C. burnetii SCV, peptidoglycan of M. tuberculosis contains a high percentage of 3-3 cross-links (73). M. tuberculosis Ldt mutants have decreased virulence (74), abnormal cell morphology (74), and increased sensitivity to the β-lactam antibiotic amoxicillin (75). Similarly, the C. burnetii rpoS mutant is more sensitive to carbenicillin than wild-type bacteria and has an aberrant cell wall/outer membrane complex morphology. The robust downregulation of Ldts by SCV ΔrpoS suggests the peptidoglycan of the mutant has fewer 3-3 cross-links that decrease cell envelope stability in the presence of carbenicillin.

The C. burnetii Dot/Icm T4BSS is directly regulated by the PmrAB two-component system (TCS) (10). C. burnetii ΔrpoS demonstrated considerable downregulation of much of the dot/icm locus at both the transcriptional and the translational levels. This downregulation was minimally complemented when assessed by immunoblotting, a factor that may contribute to incomplete rescue of intracellular growth by delaying host modifications required for productive infection. The reason for failed restoration of Dot/Icm protein levels in SCV ΔrpoS is unclear, especially considering the complementation of other genes. This discrepancy may indicate multiple levels of dot/icm gene regulation. In fact, several dot/icm genes and pmrAB have putative RpoS-binding sites (data not shown), suggesting coordinated regulation between RpoS and PmrAB. In L. pneumophila, a complex interplay between PmrAB, other TCSs, and RpoS control effector expression during stationary phase (76, 77). C. burnetii dot/icm expression differs from that of L. pneumophila where PmrAB does not directly regulate dot/icm (76, 77).

The L. pneumophila developmental cycle is controlled by the stringent response, with RpoS as a central regulator. Incorporated into this response is a monofunctional RelA, which is a guanosine 3′,5′-bispyrophosphate (ppGpp) synthetase that responds to amino acid starvation, and a bifunctional SpoT, which is a ppGpp synthetase/hydrolase that responds to fatty acid and phosphate starvation (78). As nutrients become depleted, both enzymes coordinate differentiation from the Legionella replicative to transmissive form by synthesizing the alarmone ppGpp. During nutrient replete conditions, SpoT hydrolyzes ppGpp to maintain the replicative state (78). When ppGpp levels accumulate, RpoS is produced and subsequently coordinates differentiation to the transmissive form. Like L. pneumophila, C. burnetii synthesizes a predicted RelA (CBU1375) and SpoT (CBU0303). Thus, C. burnetii likely has a stringent response similar to L. pneumophila. In fact, spoT and rpoS are upregulated in the SCV (25). Thus, a functional stringent response presumably contributes to LCV-to-SCV differentiation.

The RpoS-regulated genes defined in this study provide insight into stationary-phase signals that drive C. burnetii SCV formation. Environmental signals, such as amino acid starvation, fatty acid limitation, acid stress, oxidative stress, and phosphate limitation initiate the stringent response in other bacteria (79). Given the potentially nutrient-limited environment of the late-stage macrophage CCV, it is likely that one or more of these signals contribute to stringent response activation and subsequent SCV generation, with RpoS being a central regulator.

MATERIALS AND METHODS

C. burnetii in vitro growth and viability determination.

The bacterial strains used in this study are listed in Table S3 in the supplemental material. C. burnetii Nine Mile RSA439 phase II strain (clone 4) was used throughout the study. To assess growth and viability, C. burnetii strains grown for 7 days were inoculated into 20 ml of fresh ACCM-D at 1 × 10⁵ GE/ml, unless otherwise indicated (59). Bacteria were grown microaerobically at 37°C in 5% CO₂ and 2.5% O₂. At indicated time points, 30 μl of each sample was taken, and C. burnetii GE were enumerated using quantitative PCR (qPCR) as previously described (80). The GE/milliliter were then plotted over days postinoculation. At days 4 and 14, CFU/milliliter counts were enumerated by plating on ACCM-D modified soft-agarose plates as described previously (81).

C. burnetii growth in mammalian cells.

African green monkey kidney epithelial (Vero) cells (CCL-81; ATCC 5774) and THP-1 (ATCC TIB-202) human monocytic cells were cultured at 37°C and 5% CO₂ in RPMI 1640 medium containing 2% fetal bovine serum (FBS) and DMEM medium containing 10% FBS, respectively. For infection, cells were plated at 1 × 10⁵ cells/well in 24-well plates and allowed to adhere overnight. THP-1 cells were differentiated into macrophage-like cells by overnight incubation in 200 nM phorbol myristate acetate (Sigma-Aldrich). Cells were then infected with C. burnetii at a multiplicity of infection (MOI) of 1, based on GE, by centrifuging the plates at 500 × g for 30 min at 37°C. Cells were washed twice with 500 μl of phosphate-buffered saline (PBS; 1.5 mM KH₂PO₄, 2.7 mM Na₂HPO₄⋅7H₂O, 155 mM NaCl [pH 7.2]) and then incubated in 1 ml of fresh medium. Infected cells were harvested by trypsinization at days 0 and 7, and C. burnetii GE were quantified by qPCR.

Indirect immunofluorescence staining.

Vero cells were seeded onto coverslips in 24-well plates at 4 × 10⁴ cells/well and infected with C. burnetii at an MOI of 100. At 3 days postinfection, the cells were fixed with 4% paraformaldehyde in PBS at room temperature for 30 min and then permeabilized and blocked with 0.1% Triton X-100 plus 1% bovine serum albumin. Cells were stained for indirect immunofluorescence as previously described (82). Briefly, guinea pig anti-C. burnetii serum and a mouse monoclonal antibody directed against LAMP3 (CD63; clone H5C6; BD Biosciences) were used as primary antibodies. Alexa Fluor 488- and Alexa Fluor 568-conjugated goat IgGs (Invitrogen) were used as secondary antibodies. After staining, the cells were fixed with 4% paraformaldehyde for 30 min, and coverslips were mounted in Prolong Gold antifade (Thermo Fisher) containing DAPI (4′,6′-diamidino-2-phenylindole; Invitrogen) to visualize the nuclei.

Confocal microscopy and image analysis.

Cells were imaged using a Zeiss LSM-710 confocal fluorescence microscope (Carl Zeiss). Fiji (ImageJ; National Institutes of Health) was used for image analysis. The area (in square micrometers) of each CCV was measured using CD63 as a CCV membrane marker. At least 150 cells infected with each strain from two independent experiments were used for analyses.

Generation and complementation of an rpoS mutant.

The plasmids and oligonucleotide primers used in this study are listed in Table S3. All restriction enzymes were purchased from New England BioLabs. Oligonucleotide primers were purchased from Integrated DNA Technologies. PCR amplification was performed using AccuPrime Pfx or Taq polymerase (Invitrogen), and all cloning procedures were conducted using the In-Fusion cloning system (BD Clontech). DNA was transformed into either E. coli Stellar or PIR1 competent cells. E. coli Stellar (BD Clontech), PIR1 (Invitrogen), and BL21 (Invitrogen) cells were grown in Luria-Bertani (LB) broth. E. coli transformants were selected on LB agar plates containing 50 μg/ml kanamycin or 10 μg/ml chloramphenicol.

To generate a targeted mutation in C. burnetii rpoS (cbu1669), the 5′ and 3′ flanking regions of the gene were amplified via PCR from C. burnetii genomic DNA using the upstream and downstream oligonucleotide primer pairs CBU1669-5′-F/CBU1669-5′-R and CBU1669-3′-F/CBU1669-3′-R, respectively. The subsequent 5′ and 3′ fragments were cloned into BamHI/SalI-digested pJC-CAT (83) by In-Fusion, resulting in formation of an internal AgeI site between the 5′ and 3′ regions and the creation of pJC-CAT::CBU1669-5′3′. The P1169-Kan cassette was amplified by PCR from pJB-Kan (81) with P1169-Kan-AgeI-KO-rev-F and P1169-Kan-AgeI-KO-rev-R oligonucleotides and cloned into AgeI-digested pJC-CAT::CBU1669-5′3′ to create pJC-CAT::CBU1669-5′3′-Kan. To complement the rpoS mutant strain, the cbu1669 gene and native promoter were amplified by PCR using the oligonucleotides CBU1669comp-F and CBU1669comp-R and then cloned into EcoRI-digested pMini-Tn7T-CAT (83) by In-Fusion to create the plasmid pMini-Tn7T-CAT::cbu1669comp.

RNA extraction.

C. burnetii wild-type and ΔrpoS strains were inoculated in triplicate into 20 ml of ACCM-D at 1 × 10⁶ GE/ml. At 5 and 14 days of incubation, RNA was extracted using a modified hot TRIzol extraction. Briefly, bacteria in 8 ml of ACCM-D culture were pelleted, resuspended in 700 μl of TRIzol (Thermo Fisher), boiled at 99°C for 10 min, and vortexed vigorously. RNA was subsequently purified using a Ribopure RNA purification kit (Thermo Fisher). The RNAwiz extraction was omitted and replaced with the hot TRIzol extraction.

RNA-seq library preparation and sequencing.

Samples containing 1 μg of total RNA were subjected to rRNA removal using a RiboZero rRNA removal kit (gram-negative bacteria; Illumina). The mRNA sequencing library was prepared using a TruSeq mRNA sequencing library preparation HT kit (Illumina) without bead-based oligo(dT) purification. The resulting dual-indexed libraries were fragment-sized on a BioAnalyzer DNA1000 chip (Agilent) and quantitated using a Kapa Library Quant kit (Illumina) and Universal qPCR mix (Roche) to facilitate the creation of a normalized 2 nM multiplexed pool. This multiplexed pool was clustered across two lanes of a RAPID flow cell at a concentration of 11 pM, followed by single-read sequencing on an Illumina HiSeq 2500 for 50 cycles, plus an additional 16 cycles to sequence the dual indexes.

Bioinformatics of RNA-seq.

Raw NGS reads were processed by first removing any Illumina adapter sequences using Cutadapt v1.12. Reads were then trimmed and filtered for quality and length (minimum length, 35 bp) using a FastX Tool Kit v0.0.14 (Hannon Lab; Cold Spring Harbor Laboratory). Trimmed reads for each replicate were then aligned to the C. burnetii Nine Mile RSA493 genome using Bowtie2 v2.2.9 (84). Final read counts from mapped transcripts, based on the combined replicates, were normalized, and then differentials for each comparison were generated using DESeq2 (Bioconductor) for each experimental condition. The individual comparison results from DESeq2 were combined into one workbook with dynamic significance, fold change parameters, and added annotation.

RpoS-binding site identification.

One hundred base pairs upstream of the predicted translational start site of SCV-associated genes were searched for the E. coli RpoS-binding site consensus sequence TCTATACTTAA (41) using Geneious version 10.2.2 (Biomatters, Auckland, New Zealand). A match of 6 bp of the 11-bp consensus sequence was considered a putative C. burnetii RpoS-binding site. RpoS-binding site sequences were then aligned using WebLogo 3 (50) to illustrate conserved nucleotides.

Microcapillary reverse-phase high-pressure liquid chromatography nano-electrospray tandem mass spectrometry.

Wild-type C. burnetii and the ΔrpoS mutant were grown for 14 days in ACCM-D, and then equal numbers of bacteria (based on the GE) were pelleted, washed three times in PBS, and suspended in 2× Laemmli sample buffer. The samples were then processed and analyzed as previously described (10).

QuantiGene 2.0 assay.

Validation of RNA-seq results was performed using the QuantiGene 2.0 assay system (Affymetrix) and custom-designed probes (Panomics). C. burnetii was grown for 14 days in 50 ml of ACCM-D, pelleted, and then resuspended in 1 ml of QuantiGene lysis buffer. Samples were frozen at −80°C. Samples were lysed at 55°C for 30 min, mixed with blocking buffer and designated probes, loaded into a QuantiGene 96-well capture plate, and then incubated at 55°C for 24 h. RNA was detected as luminescence emission over 1,000 ms using a Tecan Saphire² microplate reader (Mannedorg) and is reported as relative light units (RLU). Deionized water (dH₂O) without bacterial lysate was used to subtract background luminescence. To enumerate the GE from bacterial lysates, bacteria from 1 ml of the 14-day culture were pelleted, washed once in PBS, resuspended in dH₂O, and frozen at −20°C until the GE were quantified using qPCR. RLU expression signals were normalized to 1 × 10⁸ GE/ml.

EnhC antiserum production.

The C. burnetii enhC (cbu1136) gene without the predicted signal peptide-coding region was amplified by PCR from C. burnetii Nine Mile RSA439 genomic DNA using the primers CBU1136-SP-F and CBU1136-SP-R. The PCR product was cloned by In-Fusion into the BsrGI-digested 6×His expression vector pEXP1 (Invitrogen), creating pEXP1::cbu11366-SP. Recombinant protein was produced and purified as previously described (8). Purified 6×His-CBU1136-SP (250 μg) in 25 mM Tris buffer (pH 7.2) with 0.05% Triton X-100 was mixed with the Sigma adjuvant system and used to immunize a New Zealand White rabbit. Rabbit antiserum was generated according to protocol 2008-32.1 approved by the Rocky Mountain Laboratories Animal Care and Use Committee.

Immunoblotting.

C. burnetii cultured in ACCM-D was lysed by boiling for 10 min in 2× Laemmli sample buffer. Proteins were separated on a precast 4% to 20% gradient gel using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Thermo) and then transferred to a 0.45-μm-pore-size polyvinylidene difluoride membrane (Millipore). Membranes were probed using primary rabbit antibodies specific for ScvA (34), IcmK, DotA (10), IcmX (85), or EnhC. Antibodies specific for IcmK, DotA, and IcmX were generously provided by Edward Shaw, Oklahoma State University. As a loading control, blots were probed with a mouse monoclonal antibody against C. burnetii elongation factor Ts (EFTs; generously provided by James Samuel, Texas A&M University). Reacting proteins were detected using anti-rabbit or anti-mouse IgG secondary antibodies conjugated to horseradish peroxidase (Pierce) and visualized by chemiluminescence using ECL Pico reagent (Pierce).

Hydrogen peroxide treatment.

Three milliliters of ACCM-D per well in a 12-well plate was inoculated with day 7 axenically grown C. burnetii strains at 1 × 10⁵ GE/ml. Cultures were adjusted to 1 mM H₂O₂ and incubated for 14 days. At 2-day intervals, the wells were mixed vigorously by pipetting, and the GE were enumerated from 30 μl using qPCR.

Carbenicillin treatment.

Three ml of ACCM-D per well in a 12-well plate were inoculated with day 7 axenically grown C. burnetii strains at 1 × 10⁶ GE/ml. At day 4 postinoculation, cultures were adjusted to 200 μg/ml of carbenicillin, followed by incubation for an additional 10 days. The CFU were enumerated at 4, 8, and 14 days postinoculation. Data were normalized to day 4 CFU.

Cryo-electron microscopy.

ACCM-D (20 ml) was inoculated with C. burnetii strains at 1 × 10⁵ GE/ml, and the cultures were incubated for 14 days. Cells were pelleted, washed once with PBS, resuspended in PBS, and then imaged as previously described (25).

Statistical analysis.

Statistical analyses were performed using a one-way analysis of variance in Prism software (GraphPad Software). P values of <0.05 were considered significant.

RNA-seq data.

RNA-seq results were deposited into the NCBI’s GEO database under accession number GSE110091.