CcpA Mediates Proline Auxotrophy and Is Required for Staphylococcus aureus Pathogenesis

Abstract

Human clinical isolates of Staphylococcus aureus, for example, strains Newman and N315, cannot grow in the absence of proline, albeit their sequenced genomes harbor genes for two redundant proline synthesis pathways. We show here that under selective pressure, S. aureus Newman generates proline-prototrophic variants at a frequency of 3 × 10⁻⁶, introducing frameshift and missense mutations in ccpA or IS1811 insertions in ptsH, two regulatory genes that carry out carbon catabolite repression (CCR) in staphylococci and other Gram-positive bacteria. S. aureus Newman variants with mutations in rocF (arginase), rocD (ornithine aminotransferase), and proC (Δ¹-pyrroline 5-carboxylate [P5C] reductase) are unable to generate proline-prototrophic variants, whereas a variant with a mutation in ocd (ornithine cyclodeaminase) is unaffected. Transposon insertion in ccpA also restored proline prototrophy. CcpA was shown to repress transcription of rocF and rocD, encoding the first two enzymes, but not of proC, encoding the third and final enzyme in the P5C reductase pathway. CcpA bound to the upstream regions of rocF and rocD but not to that of proC. CcpA's binding to the upstream regions was greatly enhanced by phosphorylated HPr. The CCR-mediated proline auxotrophy was lifted when nonpreferred carbohydrates were used as the sole carbon source. The ccpA mutant displayed reduced staphylococcal load and replication in a murine model of staphylococcal abscess formation, indicating that carbon catabolite repression presents an important pathogenesis strategy of S. aureus infections.

Staphylococcus aureus is an important human pathogen that causes a wide variety of diseases, including soft tissue infections, toxic shock syndrome, and necrotizing pneumonia, in both hospital and community settings (30). This versatile microbe infects and replicates in all tissues of the human body, for which processes adequate supplies of nutrients are critical. S. aureus is generally considered a proline auxotroph and imports this amino acid via two independent uptake systems (2, 21, 36, 47, 50). Based on their substrate affinities and transport attributes, a low-affinity and a high-affinity transporter (putP) can be distinguished (2, 36, 47, 50). The low-affinity transport system (ProP, a major facilitator superfamily proline/betaine:cation symporter) is thought to be activated by osmotic stress and responsible for bacterial adaptation to stress conditions (37). On the other hand, the high-affinity system (PutP) enables proline transport and scavenging under conditions where supplies of this nutrient are limiting (2, 47). Mutations in putP diminish the virulence of S. aureus in several different animal model systems of infectious diseases, including endocarditis, wound infection, and renal abscess (8, 40).

An intriguing attribute of S. aureus proline metabolism is the organism's ability to revert from a proline-auxotrophic state to prototrophy during prolonged incubation on laboratory medium lacking proline (21). This prototrophic reversion implies that S. aureus must be endowed with abilities for proline synthesis, a conjecture that was corroborated by bioinformatic analysis of the staphylococcal genome sequence. S. aureus genomes harbor genes for two proline synthesis pathways (9, 46), the pyrroline-5-carboxylate (P5C) reductase pathway (arginase to ornithine aminotransferase to P5C reductase) and the ornithine cyclodeaminase pathway (arginase to ornithine cyclodeaminase) (Fig. 1). By performing biochemical analyses of proline-prototrophic variants and arginine catabolism mutants, Townsend and colleagues provided evidence that staphylococci synthesize proline from arginine, in agreement with the possibility that the P5C reductase pathway is fully functional (46). On the other hand, bioinformatic reconstruction of metabolic networks via genome analysis of S. aureus N315 concluded that S. aureus must utilize the ornithine cyclodeaminase pathway (Fig. 1) (9). Historically, several models have been entertained to explain prototrophic reversion. For example, in an early model, Gladstone assumed a mechanism of bacterial “training,” the gradual activation or acquisition of proline biosynthesis function (21). On the other hand, Townsend et al. observed elevated arginine uptake in proline-prototrophic variants and predicted the occurrence of mutations that increase arginine transport and thereby enable reversion (46). The genetic determinant(s) for the observed increase in arginine transport, however, has not yet been identified. In sum, previous work has left unresolved by what mechanism staphylococci generate prototrophic reversion and which of their two proline synthesis pathways is being used when these variants grow in the absence of proline.

FIG. 1.

Two pathways for proline synthesis in S. aureus. The two pathways share the first step (from arginine to ornithine), but they diverge at the step(s) of ornithine's conversion to proline. Gene names and locus tags in the Newman genome are indicated in parentheses.

Carbon catabolite repression (CCR) allows microbes to select their preferred source from multiple offerings of carbon catabolism and, thereby, conserve energy (45). In Gram-positive bacteria, the catabolite control protein A (CcpA) is a key mediator of CCR (26). CcpA is a LacI/GalR family transcription factor that binds to specific cis-acting DNA sequences designated the catabolite responsive element (CRE) (25, 26, 45). Depending on the relative position of the CRE within CCR-responsive promoter sequences, CcpA may act either as a repressor or an activator (33, 34, 51). When tested in vitro, CcpA alone displays low affinity for the CRE; however, binding of its corepressor, HPr that is phosphorylated at Ser-46, induced a dramatic increase in CcpA's affinity for the CRE (14, 39).

HPr is a small phosphocarrier protein (9 to 10 kDa) encoded by ptsH. In low-GC Gram-positive bacteria, this protein plays a central role in metabolic sensing. It has two phosphorylation sites, His-15 and the aforementioned Ser-46. His-15 is phosphorylated by enzyme I (EI) of the phosphoenolpyruvate (PEP):sugar phosphotransferase system (PTS) and plays a critical role in sugar transport (49). On the other hand, Ser-46 is phosphorylated by HPr kinase/phosphorylase (HPrK/P) in the presence of a preferred carbon source, such as glucose (15, 32, 39). The HPr-Ser46∼P then binds to CcpA and enhances the DNA binding activity of CcpA by juxtaposing the DNA binding domains (39).

In this study, we used genetic approaches to characterize the mechanisms of proline-prototrophic reversion in S. aureus and identified mutations in ccpA and ptsH. In wild-type staphylococci, CcpA and HPr mediate CCR repression of rocF and rocD, encoding the first and second enzymes in the P5C reductase pathway. Ornithine cyclodeaminase (ocd) does not appear to be involved in proline-prototrophic reversion.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

S. aureus Newman and its proline-prototrophic variants were used in this study (16). For routine cultures, tryptic soy broth (TSB) was used; however, for proline prototroph emergence tests, chemically defined medium (CDM) as described by Schwan et al. (40) was used. All bacterial cultures were carried out at 37°C with shaking at 200 rpm. For strains with antibiotic resistance, appropriate antibiotics were added to the culture medium at the following concentrations: chloramphenicol, 5 μg/ml (for the integration plasmid) or 10 μg/ml (for the multicopy plasmid), and erythromycin, 10 μg/ml. For strains without plasmid or transposon, nalidixic acid (5 μg/ml) was added to prevent contamination by Gram-negative bacteria.

Isolation of proline-prototrophic variants.

S. aureus Newman was grown overnight at 37°C in 3 ml of CDM. The next morning, the cells were washed once with CDM deficient in proline and arginine (CDM-PR), and then the cell density was adjusted to an optical density at 600 nm (OD₆₀₀) of 1. One hundred microliters of the cell suspension was spread on a CDM agar plate deficient in proline, and the plate was incubated at 37°C for 3 days. The proline prototrophy of the colonies was verified by growing them in 5 ml CDM-P at 37°C overnight; all of the colonies grew to an OD₆₀₀ of 3 to 4 within 24 h.

Construction of plasmids for the complementation tests.

In this study, to construct plasmids for the complementation tests, we used two different plasmids, the multicopy plasmid pOS1 (38) and the integration plasmid pCL55 (28). First, ccpA was amplified by the primers P503 and P523 (Table 1), and then the amplified fragment was cloned between the BamHI and SmaI sites of pCL55. The resulting plasmid was named pccpA. To clone ccpA in pOS1, pccpA was digested with BamHI and EcoRI, and then the resulting ccpA fragment was cloned in pOS1 digested with the same enzymes, generating pOS1-ccpA. Because of its in vivo stability, pccpA was used for transcript analysis (see Fig. 5) and animal tests (see Fig. 8), while pOS1-ccpA was used for growth pattern analysis (see Fig. 2).

TABLE 1.

Primers used in this study

Primer	Sequence	Application
attB1-1503-up	5′ GGGGACAAGTTTGTACAAAAAAGCAGGCTATCGCCTAAAATATCCAGCAC 3′	proC deletion
1503-R-KpnI	5′ TCCGGTACCTACGAGTTTCATTAACGTCACA 3′	proC deletion
attB2-1503-down	5′ GGGGACCACTTTGTACAAGAAAGCTGGGTCATTTTTTGGAACGAGTGCAG 3′	proC deletion
1503-F-KpnI	5′ GTAGGTACCCAATAAAAACAAACCCGCCAAC 3′	proC deletion
P392	5′ CTAAATCTCTCATACCAATTAGTAC 3′	rocF RT-PCR
P336	5′ TAAGGATCCGGGGGACGCTTATGACAAAGAC 3′	rocF RT-PCR
P342	5′ TAAGGATCCGATATGATGACTAAATCTGAAAAAATT 3′	rocD RT-PCR
P393	5′ ATCTCCAAAATCAACTTTTCTAAATC 3′	rocD RT-PCR
P344	5′ TAAGGATCCCGTTAATGAAACTCGTATTTTATGG 3′	proC RT-PCR
P394	5′ ATATAAAAATGCTGGGCCGCTTC 3′	proC RT-PCR
P934	5′ CCTTACCAAATCTTGACATCC 3′	16S rRNA RT-PCR
P935	5′ GTGTAGCCCAAATCATAAGG 3′	16S rRNA RT-PCR
P503	5′ GCGAGTTGGTACGAATCTACTCGTC 3′	ccpA cloning in pCL55
p523	5′ TAAGGATCCAAGGAAACTATAGACTACTTAAC 3′	ccpA cloning in pCL55
NdeI-ccpA-N	5′ GAGGAACATATGACAGTTACTATATATGATGTAGC 3′	ccpA cloning in pET15b
BamHI-ccpA-C	5′ GGGGGATCCTTATTTTGTAGTTCCTCGGTATTCAA 3′	ccpA cloning in pET15b
NdeI-hpr-N	5′ AATGTACATATGGAACAAAATTCATATGTAATC 3′	ptsH cloning in pET15b
BamHI-hpr-C	5′ ACAGGATCCTTTAGTCAATCCTTCTTTTGATAAGAC 3′	ptsH cloning in pET15b
NdeI-hprK-N-F	5′ TTTGGGCATATGTTAACGACAGAAAAACTAGTT 3′	hprK cloning in pET15b
BamHI-hprK-C-R	5′ GGGGGATCCCTACTCCTCACTCTTATGACTG 3′	hprK cloning in pET15b
P1060	5′ TATGTAATTAAAGAATAATGTATGCGCTTACCATTATCAAGCAATAGCTA 3′	rocF EMSA
P1061	5′ TAGCTATTGCTTGATAATGGTAAGCGCATACATTATTCTTTAATTACATA 3′	rocF EMSA
P1062	5′ TAATTATGTCATTATAAATGTAAGGGTTTTCAAAATAGACAAATTTAATG 3′	rocD EMSA
P1063	5′ CATTAAATTTGTCTATTTTGAAAACCCTTACATTTATAATGACATAATTA 3′	rocD EMSA
P1064	5′ TATATATACAATAATAAGAAAATATAACATACAAATCAAAAACTAAAGGG 3′	proC EMSA
P1065	5′ CCCTTTAGTTTTTGATTTGTATGTTATATTTTCTTATTATTGTATATATA 3′	proC EMSA

FIG. 5.

Transcript analysis for proline synthesis genes. (A) The wild type (wt) and proline-prototrophic variants (NMpro+5, NMpro+18, NMpro+24, NMpro+33, and NMpro+36) were grown in TSB, and total RNA was purified. Transcript levels of the proline synthesis genes in the purified RNA were measured by RT-PCR with 17 cycles of amplification. 16S RNA was used as a control. (B) The wild-type strain (wt), the ccpA transposon mutant ΦΝΞ-7791 (7791), and the ΦΝΞ-7791 mutant carrying the vector pCL55 (pCL55) or the complementation plasmid pccpA (pccpA) were all grown in TSB. After total RNA was purified from the cells, transcript levels were analyzed by RT-PCR with 18 cycles of amplification. The numbers below the images represent the results of densitometry analysis normalized to the transcript level of the wild-type strain. N, no reverse transcriptase.

FIG. 8.

CcpA's contribution to staphylococcal virulence in the murine abscess formation model. The wild-type strain Newman, the ccpA transposon mutant ΦΝΞ-7791, and the ΦΝΞ-7791 strain carrying either pCL55 or pccpA (10⁷ CFU) were administered to 10 BALB/c mice via retro-orbital injection. Four days later, the mice were sacrificed and bacterial loads in liver and kidneys were measured by counting CFU on TSA plates. Horizontal bars indicate the means of all observations. Bacterial loads were compared with those in mice infected with the wild type (NM). Statistical significance was measured by one-way ANOVA with Bonferroni's posttest. NM, Newman strain; 7791, ΦΝΞ-7791 strain; **, P < 0.05; n.s., not significant.

FIG. 2.

Growth patterns of the ccpA mutant and the complemented strain in a proline-deficient medium. The complementation plasmid, pOS1-ccpA, was constructed by cloning ccpA in the E. coli-S. aureus shuttle plasmid pOS1. Cells were grown in CDM-P at 37°C for 30 h, and their optical density was measured at 600 nm. 7791, ccpA transposon insertion mutant ΦΝΞ-7791; WT, wild type.

Deletion mutagenesis.

In-frame deletion of proC, the gene encoding Δ¹-pyrroline 5′-carboxylate reductase, was carried out using the allelic exchange plasmid pKOR1 as described previously (4). To generate flanking DNA fragments for the deletion, the following primers were used (Table 1): attB1-1503-up/1503-R-KpnI for the upstream fragment and attB2-1503-down/1503-F-KpnI for the downstream fragment.

RT-PCR.

The wild-type strain Newman and the proline-prototrophic variants were grown in TSB, and cells were harvested at mid-log phase (OD₆₀₀ of ≈0.9). From the harvested cells, RNA was purified with a FastRNA pro blue kit (Q-Biogene) in accordance with the manufacturer's recommendations. To eliminate contaminating genomic DNA, the purified RNA was treated with RQ1 RNase-free DNase (Promega) at 37°C for 30 min. From 2 μg of the purified RNA, cDNA was synthesized by using Superscript II reverse transcriptase (Invitrogen) and random primers (Applied Biosystems). The reverse transcription reaction (RT-PCR) was carried out in a total volume of 20 μl at 42°C for 50 min and, subsequently, at 72°C for 15 min. Two microliters of the cDNA reaction mixture was used for PCR amplification with Taq DNA polymerase (New England Biolabs) with the following primers: P336/P392 for rocF, P342/P393 for rocD, P344/P394 for proC, and P934/P935 for 16S rRNA (Table 1). To prevent saturation of the reaction mixture, the number of DNA amplification cycles was limited to either 17 or 18 at the following conditions: 94°C for 15 s, 55°C for 15 s, and 72°C for 30 s.

Expression and purification of His-tagged CcpA, HPr, and HPrK/P.

The genes were PCR amplified by using Phusion DNA polymerase (New England Biolabs) with the primers listed in Table 1. The amplified DNA was digested with NdeI and BamHI and cloned into pET15b (Novagen). The resulting plasmid was electroporated into Escherichia coli XL1 Blue and, subsequently, into E. coli BL21(DE3) (Invitrogen). His₆-tagged proteins were induced by 0.5 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) in Luria broth at either 37°C (His₆-CcpA and His₆-HPr) or 16°C (His₆-HPrK/P) and purified with Ni-Sepharose 6 fast flow (GE) by following the manufacturer's recommendations.

Phosphorylation of HPr by HPrK/P.

The phosphorylation assays were performed in the presence of 6 μM His₆-HPr and 2 μM His₆-HPrK/P in phosphorylation buffer (10 mM Tris-HCl, pH 7.4, 50 mM KCl, 5 mM MgCl₂, 1 mM dithiothreitol [DTT], and 10% glycerol). The final reaction mixture volume was 50 μl. [γ-³²P]ATP (0.5 μl, 10 μCi/μl) was added to initiate the reaction. The reactions were carried out at 37°C for 30 min and then stopped by the addition of 20 μl of 2× SDS loading buffer (0.5 M Tris-HCl [pH 6.8], 4% [wt/vol] SDS, 20% [vol/vol] glycerol, 2% 2-mercaptoethanol, 0.001% bromophenol blue). Samples were analyzed by 13% SDS-PAGE and then subjected to autoradiography on a phosphor screen (BAS-IP; Fuji).

EMSA.

The DNA probes for electrophoretic mobility shift assays (EMSA) were generated by annealing two 50-bp DNA primers (Table 1) which encompass the CRE site of rocF or rocD. To anneal the primers, equimolar concentrations of each primer were mixed together and incubated at 95°C for 5 min and then at room temperature for 60 min. The resulting probe was phosphorylated by using polynucleotide kinase (New England Biolabs) and [γ-³²P]ATP and kept on ice until used. The His₆-CcpA protein (1 μM) was preincubated with 1 mM ATP, 1.8 μM His₆-HPr, and 0.6 μM His₆-HPrK/P in a buffer containing 10 mM Tris-HCl, pH 7.4, 50 mM KCl, 5 mM MgCl₂, 1 mM DTT, 10% glycerol, and 3 μg/ml sheared salmon sperm DNA at 37°C for 30 min. As a control, the His₆-CcpA protein (1 μM) alone was incubated in the same conditions in the absence of ATP, His₆-HPr, and His₆-HPrK/P. Both resulting CcpA samples were serially diluted to 0.5 μM, 0.25 μM, 0.125 μM, and 0.0625 μM, followed by the addition of 2 ng of radiolabeled DNA probe. The reaction mixtures (20 μl each) were incubated at room temperature for 20 min and then were analyzed by 8% polyacrylamide gel electrophoresis (100 V for the prerun and 85 V for 85 min for sample separation). The gels were dried and subjected to autoradiography. The dissociation constant (K_d) was calculated from densitometry measurements of the DNA bands. A repeat of the assay produced similar results.

Gel-filtration chromatography.

In gel-filtration analysis, His₆-HPr was treated with thrombin (Sigma-Aldrich), and the resulting HPr without the His₆ tag was used. His₆-CcpA (5.8 μM) was incubated with HPr (8.3 μM, no His₆ tag) and His₆-HPrK/P (1.1 μM) in either the absence or presence of ATP (1 mM) for 2 h at room temperature. The reaction buffer contains 10 mM Tris-HCl, pH 7.4, 50 mM NaCl, 5% glycerol, 0.1 mM EDTA, and 5 mM MgCl₂. The rocF promoter DNA (0.17 μM) was added to the resulting mixture and incubated at room temperature for another 20 min. The DNA-protein mixture was applied to a fast protein liquid chromatography (FPLC) gel-filtration column (HiLoad 16160 Superdex 75; GE Healthcare) equilibrated with buffer A (10 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM MgCl₂). The column was eluted with the same buffer at a flow rate of 1 ml/min, and fractions (2 ml each) were collected. Aliquots (600 μl) of the fractions corresponding to the peaks (peak A, fractions 9 to 11; peak B, fractions 12 to 14; peak C, fractions 15 to 17; and peak D, fractions 24 to 25) were concentrated down to 20 μl by using an Ultrafree-0.5 (Millipore) centrifugal filter device and then subjected to OD₂₆₀/OD₂₈₀ analysis. The samples were further analyzed by SDS-PAGE (13%) followed by Coomassie staining.

Carbon source effects on the proline synthesis.

The wild-type strain Newman was grown in 3 ml TSB at 37°C overnight. The next day, cells were collected by centrifugation. The collected cell pellet was washed once with 5 ml CDM deficient in proline and carbon source (CDM-PC) and then suspended in 5 ml of CDM-PC. The cell suspension (30 μl) was added to 3 ml of CDM or CDM-P that contained 1 of 11 different carbohydrates (0.5% final concentration) as the sole carbon source. The resulting cultures were incubated at 37°C for 16 h, and cell growth was assessed by measuring the OD₆₀₀.

Murine intravenous infection and abscess formation.

An overnight culture at 37°C in TSB was diluted 100 times with fresh TSB and further incubated at 37°C until an OD₆₀₀ of 1.0. Bacteria were collected by centrifugation, washed, and suspended in phosphate-buffered saline (PBS) to an OD₆₀₀ of 0.4. To quantify the infectious dose, viable staphylococci were enumerated by colony formation on tryptic soy agar (TSA) plates. One hundred microliters (∼10⁷ CFU) of the bacterial suspension was administered via retro-orbital injection to 6- to 8-week-old BALB/c mice. Four days after the injection, mice were euthanized by CO₂ asphyxiation, and the kidneys and liver were removed. The organs were homogenized in 1 ml of PBS, and dilutions of the homogenates were spread on TSA to enumerate CFU. For statistical analysis, one-way analysis of variance (ANOVA) with Bonferroni's posttest was carried out by using Prism (GraphPad Software). This animal study was approved by the Institutional Animal Care and Use Committee at Indiana University School of Medicine—Northwest, which complies with guidelines of the National Institutes of Health.

RESULTS

Proline-prototrophic variants harbor mutations in ccpA and ptsH.

We hypothesized that mutations in regulatory genes may be responsible for proline-prototrophic reversion and tested this hypothesis by screening 1,700 nonredundant transposon mutants in S. aureus Newman (3) by plating each variant on chemically defined agar medium lacking proline (CDM-P). A single mutant, ΦΝΞ-7791, with a transposon insertion 7 nucleotides downstream of the ATG start codon of the carbon catabolite control protein A (ccpA), formed colonies within 24 h. The transposon mutation of ΦΝΞ-7791 was transduced into the wild-type strain Newman, and the transductant was tested again for growth in CDM-P. As shown in Fig. 2, the ccpA mutant grew within 24 h in CDM-P, whereas the wild-type parent strain Newman did not (Fig. 2, 7791 and WT, respectively). Wild-type ccpA was cloned in the E. coli-S. aureus shuttle vector pOS1 (38). Transformation of the resulting plasmid, pOS1-ccpA, but not of the vector control suppressed the ability of S. aureus ΦΝΞ-7791 to grow in CDM-P. Thus, insertional mutation in ccpA enables proline-prototrophic growth of S. aureus Newman.

If ccpA is responsible for the proline-auxotrophic phenotype, spontaneously isolated prototrophic variants may carry mutations in ccpA. This was tested by spreading S. aureus Newman on CDM-P agar; spontaneous variants formed colonies within 48 h at a frequency of 3 × 10⁻⁶. The ccpA genes of nine of these proline-prototrophic variants were sequenced (Fig. 3 ). Eight of these variants harbored mutations in ccpA. Four variants (NMpro+5, NMpro+18, NMpro+24, and NMpro+33) carried frameshift mutations (either a deletion or a nucleotide insertion), whereas four variants (NMpro+14, NMpro+19, NMpro+26, and NMpro+36) harbored a G-to-T transversion that converted glycine (GGA) or glutamic acid (GAA) codons to stop codons (TGA or TAA, nonsense mutation) (the converted amino acid is in boldface). To confirm that these mutations are indeed responsible for proline-prototrophic reversion, the wild-type ccpA gene was provided in trans to each of the four mutants (NMpro+5, NMpro+18, NMpro+24, and NMpro+36) by transformation with pOS1-ccpA. Plasmid transformation abolished the growth of ccpA variants in CDM-P (data not shown), thereby confirming that the mutational lesions in the ccpA gene are indeed responsible for proline-prototrophic reversion.

FIG. 3.

Identification of mutations in ccpA (A) and ptsH (B). ccpA and ptsH were amplified from nine proline-prototrophic variants by PCR, and the products were sequenced. Insertion mutations are shown by inverted triangles, while other types of mutations are indicated by vertical arrows; a delta indicates a deletion. The variant identification numbers are also shown, to the right of the mutations. Amino acid changes are indicated in parentheses; an asterisk indicates a stop codon.

The variant NMpro+37 did not harbor a mutation in ccpA. We asked whether the variant carried a mutation in ptsH, a gene that encodes HPr, the corepressor of CcpA (15, 39), and therefore, PCR amplified the ptsH gene from the chromosomal DNA of NMpro+37. The PCR product of the variant strain was 1.2 kb larger than that of the wild-type strain, suggesting an insertion of a DNA segment. Subsequent DNA sequencing analysis revealed that the insertion sequence IS1181 was located in the ptsH gene of strain NMpro+37 (Fig. 3B). In summary, the observed proline-prototrophic reversion of staphylococci is due to mutational lesions that disrupt the function of the CCR transcriptional regulators CcpA and HPr.

The P5C reductase pathway is required for the emergence of proline-prototrophic variants.

The genome of S. aureus strains harbors genes that, at least in silico, could provide two independent mechanisms of proline synthesis (Fig. 1). Ornithine, the initial substrate for the two parallel pathways, is generated from arginine by the product of rocF, encoding arginase. rocD, specifying ornithine aminotransferase, and proC, encoding Δ¹-pyrroline 5-carboxylate (P5C) reductase, convert ornithine to proline via the Δ¹-pyrroline 5-carboxylate intermediate. As a second, independent pathway, ocd, encoding ornithine cyclodeaminase, is expected to allow the conversion of ornithine directly to proline (Fig. 1). To determine which, if any, gene of these two pathways is required for proline synthesis, we carried out proline prototroph emergence tests with mutants of the aforementioned genes. If an enzyme was essential for proline synthesis, a mutation in its structural gene would not allow ccpA or ptsH mutations to restore proline prototrophy. Suitable mutations were obtained from the Phoenix transposon library (rocF, rocD, and ocd) (3) or were generated by in-frame deletion mutagenesis (proC). Staphylococci were grown in a chemically defined medium lacking proline (CDM-P) for 48 h, and the optical density of the culture was measured at 600 nm. The data in Fig. 4 demonstrate that mutants with defects in the P5C reductase pathway (rocF, rocD, and proC) failed to grow in the absence of proline (i.e., no prototrophic variant), whereas the mutation in ocd and the negative-control argD mutant had no impact on the emergence of proline-prototrophic variants. These data suggest that S. aureus utilizes the P5C reductase pathway for proline synthesis in proline-prototrophic variants.

FIG. 4.

A proline prototroph emergence test. Proline synthesis mutants were grown in CDM-P for 48 h, and then the emergence of proline prototrophs was measured by the optical density at 600 nm. Mutated genes are shown under the graph. A mutant of argD encoding acetylornithine aminotransferase (NWMN_0129) involved in arginine synthesis was used as a negative control. Standard deviations are also indicated on the graph. wt, wild-type.

CCR blocks proline synthesis via transcriptional repression of proline synthesis genes.

To further characterize the molecular mechanism for the prototrophic reversion, we used RT-PCR to compare the transcript levels of proline synthesis genes in prototrophic variants with those of wild-type S. aureus strain Newman. As can be seen by the data in Fig. 5A, all prototrophic variants harbored elevated transcript levels of the first two proline synthesis genes (rocF and rocD), indicating that ccpA mutations affect the transcriptional repression of proline synthesis genes at the first two steps. The transcript level of the last gene, proC, however, was not significantly altered by the mutations, suggesting that proC is not subjected to CCR.

To further verify the CCR-mediated transcriptional repression of rocF and rocD, we complemented the transposon mutant ΦΝΞ-7791 with the wild-type ccpA gene cloned in the integration plasmid pCL55 (28) and then compared the transcript levels by RT-PCR and densitometry analyses. As shown in Fig. 5B, the presence of the complementation plasmid indeed reduced the transcript levels of rocF and rocD by approximately 2-fold; however, transcript reduction was not observed for either proC or 16S rRNA, confirming that CCR specifically represses the expression of the first two genes, rocF and rocD. Recently, by Northern blot analysis, Seidl et al. also clearly showed that the transcription of rocF (arg in their study) and rocD is repressed by the CcpA-mediated CCR (43).

CcpA binds to CREs in rocF and rocD.

Since the transcription of rocF and rocD was repressed by CcpA, we searched the upstream region of the genes for CRE sequences, the binding site of CcpA (25, 26, 45). Indeed, CRE sites were found in the upstream region of rocF and rocD (Fig. 6A). We therefore sought to examine the binding of staphylococcal CcpA to the CRE sites with electrophoretic mobility shift assays (EMSA). Since the upstream sequence of proC does not contain a CRE site, we used the sequence as a negative control. For the EMSA, we purified not only His₆-CcpA but also His₆-HPr, as well as His₆-HPrK/P, because CcpA associates with HPr∼P to achieve high-affinity binding to DNA at CRE sites (26a, 39). As expected, the purified His₆-HPrK/P was able to phosphorylate His₆-HPr (Fig. 6B). In the EMSA, the purified His₆-CcpA was mixed with radiolabeled double-stranded DNA segments in either the presence or absence of His₆-HPr∼P, and then the protein-DNA complexes were analyzed by polyacrylamide gel electrophoresis and autoradiography. As shown in Fig. 6C, while His₆-CcpA did not bind significantly to the upstream DNA of proC, it did bind to the upstream DNA of rocF and rocD, showing the specific binding of CcpA to CRE sites. Intriguingly, although the addition of His₆-HPr∼P increased the binding affinity of CcpA to the CRE in rocD (from a K_d of 0.4 μM to 0.12 μM), the increase of the binding affinity was much smaller for the CRE in rocF (from a K_d of 0.78 μM to 0.64 μM).

FIG. 6.

CcpA binding to CRE sequences. (A) CRE sequences in rocF and rocD. The CRE sequence of pckA was used to identify CRE in rocF and rocD. Identical nucleotides are indicated with a black dot. The numbers flanking the CRE sequence represent the distance from the start codon of the corresponding gene. (B) HPr phosphorylation by HPrK/P. His₆-HPr (6 μM) was mixed with His₆-HPrK/P (2 μM) in the presence of [γ-³²P]ATP (5 μCi), and the reaction products were analyzed by SDS-PAGE and autoradiography. (C) EMSA analysis of CcpA binding to CRE. The 50-bp double-stranded DNAs containing CREs were mixed with various concentrations of His₆-CcpA in either the presence or absence of HPr∼P. As a negative control, the 50-bp DNA sequence upstream of proC was used. The DNA-protein complex was analyzed by PAGE and autoradiography. Lanes 1 and 6, 0.0625 μM; lanes 2 and 7, 0.125 μM; lanes 3 and 8, 0.25 μM; lanes 4 and 9, 0.5 μM; lanes 5 and 10, 1 μM. (D) Gel-filtration analysis of CcpA binding to rocF DNA. The mixture of His₆-CcpA (5.8 μM), HPr (8.3 μM), and His₆-HPrK/P (1.1 μM) was incubated at room temperature in either the absence (upper panel) or presence (lower panel) of ATP (1 mM). After addition of the rocF DNA and incubation at room temperature, the protein-DNA complexes were applied to an FPLC gel-filtration column and eluted at 1 ml/min. Gel-filtration chromatograms are shown, along with the SDS-PAGE analysis of the four peaks (A, B, C, and D). The control proteins (i.e., CcpA and HPrK/P for the upper panel and CcpA and HPrK/P and HPr for the lower panel) are also shown. The gels were stained with Coomassie dye.

Since it is well established that CcpA requires the association with HPr∼P for efficient DNA binding (14, 39), the significant binding of CcpA to CREs in the absence of HPr∼P and the minimal effect of HPr∼P on CcpA binding to rocF DNA (Fig. 6C) were unexpected. To examine the role of HPr∼P in CcpA binding to DNA more rigorously, we analyzed the CcpA binding to rocF by gel filtration assay. In the assay, His₆-CcpA, HPr, and His₆-HPrK were mixed in either the absence or presence of ATP, and then, after the addition of rocF DNA, the DNA-protein mixture was subjected to gel filtration chromatography. As can be seen in Fig. 6D, in this assay, four peaks were observed: the His₆-HPrK/P-HPr complex, most likely a hexamer (17), was eluted first (peak A), followed by rocF DNA (peak B), His₆-CcpA (peak C), and HPr (peak D). In the absence of ATP (i.e., no HPr∼P), the OD₂₆₀/OD₂₈₀ ratio of peak B (rocF DNA) was 2.28, suggesting the absence of proteins. Indeed, SDS-PAGE analysis of the eluted peaks showed that no proteins were coeluted with rocF DNA (lane B in the upper panel of Fig. 6D). On the other hand, in the presence of ATP, the OD₂₆₀/OD₂₈₀ ratio for peak B was 1.64, suggesting the presence of proteins. Subsequent SDS-PAGE analysis showed that, as expected, His₆-CcpA and HPr (probably in a phosphorylated form) were coeluted in peak B (lane B in the lower panel of Fig. 6D), showing that, in the gel filtration assay, the association with HPr∼P is essential for CcpA binding to the rocF DNA. Taken together, in S. aureus, CcpA specifically binds to CRE sites of rocF and rocD, and the binding to the CRE site is significantly enhanced by HPr∼P.

The repression of proline synthesis by CCR can be abolished by nonpreferred carbon sources.

If proline synthesis is repressed by CCR, the repression should be lifted when nonpreferred carbohydrates are used as a sole carbon source. To test this possibility, since the preferred carbon sources are not clearly defined for S. aureus, we prepared CDM-P with one of 11 different carbohydrates as the sole carbon source and examined the growth of strain Newman in the CDM-P. As can be seen by the results in Fig. 7, when any of the carbohydrates glucose, fructose, glycerol, sucrose, mannitol, maltose, and salicin was used, strain Newman did not grow significantly in CDM-P, suggesting that these carbohydrates are preferred carbohydrates and elicited strong CCR. When the other carbohydrates (i.e., gluconic acid, arabinose, sorbitol, or ribose) were used, however, the bacterium showed significant growth, suggesting that these are nonpreferred carbon sources and did not elicit CCR, resulting in restoration of the proline prototrophy. These results demonstrate that the proline auxotrophy is conditionally imposed by CCR and can be abolished by nonpreferred carbon sources.

FIG. 7.

Carbon source effects on proline synthesis. The wild-type strain Newman was grown in CDM or CDM-P with each of the indicated carbohydrates as the sole carbon source at 37°C for 16 h. Then, the proline synthesis was measured by the cell growth in CDM-P. Error bars show standard deviations.

Contribution of ccpA to staphylococcal virulence.

Recent studies showed that, in S. aureus, CcpA affects not only carbon metabolism but also resistance to antibiotics, production of virulence factors, and biofilm formation (41, 42, 44). To test the contribution of CcpA toward staphylococcal virulence and disease pathogenesis, we examined the ccpA mutant ΦΝΞ-7791 in a murine model of abscess formation, along with its wild-type parent strain (Fig. 8). Compared to mice infected with the wild-type Newman, mice infected with the ΦΝΞ-7791 mutant and the mutant carrying the vector pCL55 displayed significantly reduced bacterial loads in liver tissue, revealing a defect in abscess disease pathogenesis. On the other hand, when complemented with the wild-type ccpA, the bacterial load of the mutant in mice was not significantly different from that of the wild type [Fig. 8, 7791 (pccpA)], suggesting that virulence was restored by the introduction of the wild-type ccpA gene. When tested in the same murine model, however, disruption of proC did not reduce the bacterial loads (data not shown), suggesting that proline synthesis is not required for abscess formation in liver or kidneys. In sum, CcpA-mediated CCR plays an important role in the pathogenesis of S. aureus in murine liver abscess formation.

DISCUSSION

In this study, we provide genetic evidence that (i) S. aureus synthesizes proline via the P5C reductase pathway, (ii) the proline auxotrophy of S. aureus strain Newman is conditionally imposed by CcpA/HPr-mediated CCR, and (iii) the proline-prototrophic reversion is caused by mutational disruptions of the CCR effector genes ccpA and ptsH. Finally, we also showed that, in a murine abscess formation model, the CcpA/HPr-mediated CCR is required for bacterial survival in the liver.

In the genome-scale reconstruction of metabolic networks with the S. aureus N315 genome, Becker and Palsson noted that the amino acid requirement is one of the primary differences between the prediction from the genome sequence and the experimental data (9). In their analysis, S. aureus is predicted to grow in the presence of any one of nine amino acids: alanine, arginine, aspartic acid, glycine, glutamic acid, ornithine, proline, serine, and threonine (9). In contrast, Kuroda et al. reported that alanine, arginine, glycine, isoleucine, proline, and valine are absolutely required for the growth of S. aureus N315, albeit the N315 genome contains complete gene sets for the biosynthesis pathway for all of those amino acids (27). To explain this apparent inconsistency, both studies and a recent metabolic reconstruction study (29) presumed that biosynthetic pathways for amino acids must be controlled by negative regulatory mechanisms. Our results here demonstrate that the synthesis of one of these amino acids, proline, is indeed controlled by CcpA-mediated repression. We also show that proline-prototrophic variants of S. aureus Newman use only the P5C reductase pathway, whereas the microbe's ornithine cyclodeaminase appears dispensable under the conditions tested (Fig. 4). Then what could be the function of ornithine cyclodeaminase in S. aureus? In staphylococcal genomes, ocd is located in an operon with a putative cysteine synthase and siderophore biosynthesis genes. A recent study showed that ocd is repressed by the ferric uptake regulator (Fur) and iron (20), implying that the transcription of this operon is turned on under low-iron conditions, as occurs during host infection. If so, the ornithine cyclodeaminase pathway may be controlled by two distinct regulatory systems in S. aureus, rocF under CcpA-mediated CCR and ocd under Fur/iron-mediated regulation. It would be intriguing to test whether low-iron conditions without preferred carbon sources can indeed induce the ornithine cyclodeaminase pathway and lead to proline synthesis in the proC mutant S. aureus.

Several data support our conclusion that, in S. aureus, proline synthesis is conditionally repressed by CCR. First, the mutational disruption of the CCR components (i.e., CcpA and HPr) converted S. aureus Newman to a proline prototroph (Fig. 2 to 4). Second, CCR mutants showed elevated levels of transcription of two proline synthesis genes, rocD and rocF (Fig. 5). Third, CcpA/HPr∼P bound to CRE sites of two proline synthesis genes, rocD and rocF (Fig. 6). Finally, nonpreferred carbon sources can convert S. aureus to a proline prototroph (Fig. 7). The existence of CRE sites in two proline synthesis genes (i.e., rocD and rocF) and the binding of CcpA/HPr∼P to the CRE sites in EMSA experiments clearly demonstrate that CcpA/HPr∼P represses proline synthesis by direct binding. Interestingly, however, in the EMSA experiments, CcpA showed a significant binding affinity to CRE sites even in the absence of HPr∼P (Fig. 6C). Since the disruption of ptsH, the gene encoding HPr, converted the mutant to a proline prototroph, it is unlikely that, under physiological conditions, CcpA alone (without HPr∼P) can bind to and repress the proline synthesis genes. Indeed, in the gel-filtration chromatography experiment, where the CcpA concentration was 5.8 times higher than the highest concentration in EMSA (5.8 μM versus 1 μM), CcpA did not bind to rocF DNA in the absence of HPr∼P (Fig. 6D, peak B in −ATP condition). When HPr∼P was provided, however, significant fractions of CcpA and HPr (probably in a phosphorylated form) were found in the rocF DNA fraction (Fig. 6D, peak B in +ATP condition), suggesting the formation of a CcpA/HPr∼P/rocF DNA complex. Therefore, unlike the EMSA, the gel-filtration experiment showed the essential role of HPr∼P in the DNA binding of CcpA. Then what would explain the apparent DNA binding activity of CcpA without HPr∼P in the EMSA as shown in Fig. 6C? Gel electrophoresis has been known to protect protein-DNA complexes such that interactions that are short lived in free solution can persist in gel electrophoresis (18, 19, 48). In gel-electrophoresis conditions, the collision frequency between protein-nucleic acid complexes and competing reactants is thought to be reduced, resulting in protection of protein-nucleic acid complexes (48). We suspect that this so-called “molecular sequestration effect” is responsible for the apparent HPr∼P-independent DNA binding activity of CcpA shown in EMSA.

In S. aureus, CcpA appears to function as a global regulator, affecting not only carbon metabolism but also the production of virulence determinants, antibiotic resistance, biofilm formation, and nitrogen metabolism (41-44). Recent transcriptome and proteome analysis confirmed the broad effects of CcpA on gene expression in S. aureus (43). Therefore, the attenuation of the ccpA mutant in murine abscesses was not surprising (Fig. 8). Interestingly, however, the attenuation was observed only in the liver; in the kidney, no virulence attenuation was detected. This organ-specific virulence reduction might be caused by the unique roles the liver plays in carbohydrate metabolism and innate immune response. Storage and metabolism of carbohydrates is one of the main functions of the liver. In addition, the liver is the main organ where invading bacteria in blood are taken up and eliminated (11). The liver sinusoid is lined with Kupffer cells, the resident tissue macrophages in the liver, which constitute 80 to 90% of tissue macrophages in the body (12). The Kupffer cells not only eliminate invading bacteria by phagocytosis but also play a role in the development of adaptive immunity by interaction with neutrophils (22-24). Therefore, compared with the kidney, the liver seems to provide much harsher challenges, such that unregulated carbon metabolism and the resulting fitness loss can be detrimental to bacterial survival.

Broad effects of CcpA on gene expression were also demonstrated in other Gram-positive bacteria. For example, microarray studies suggested that CcpA regulates approximately 10% of all genes in Bacillus subtilis (34) and 347 genes (14% of all genes) in Lactococcus lactis (51). In B. subtilis, although CRE was only found upstream of rocD, the genes for arginase (rocF) and ornithine aminotransferase (rocD) are also repressed by CcpA (34). These results suggest that the P5C reductase pathway of B. subtilis is also under the CCR control of CcpA (10), although B. subtilis has been reported to synthesize proline from glutamate, not from arginine (10). Mutational disruption of the glutamate pathway renders bacilli proline auxotrophic (35), in agreement with a model whereby the P5C reductase pathway of B. subtilis is nonfunctional under experimental conditions. Unlike S. aureus, B. subtilis does not require disruption of CCR to assume proline prototrophy, as its glutamate pathway is not regulated by CcpA (34). It remains to be determined, however, whether the disruption of CCR can enable B. subtilis to synthesize proline from arginine and what roles the apparently nonfunctional P5C reductase pathway plays in B. subtilis.

Among nine analyzed variants, NMpro + 37 acquired proline prototrophy via IS1181 insertion in ptsH. IS1181 is an insertion sequence identified first in the methicillin-resistant S. aureus strain BM3121 (13). S. aureus Newman harbors two copies of IS1181 with high sequence identity (99.7%) which are positioned on the chromosome between NWMN_1751 (encoding the phosphotransferase system EIIC) and NWMN_tRNA24 (tRNA-Leu) and between rocF and NWMN_tRNA51 (tRNA-Lys). Sequence comparison revealed that the IS1181 insertion into ptsH originated from the latter site. The insertion pattern of IS1181 is heterogeneous among different S. aureus isolates. For example, S. aureus strains N315 and Mu50 contain 8 and 10 copies of IS1181, respectively, while strain MW2 lacks this element altogether (1, 27). Further, the IS1181 copies in the genome of strains N315 and Mu50 are not located in the vicinity of rocF and tRNA-Lys. The observation that IS1181 transposes into ptsH (NMpro+37), together with the uneven distribution of IS1181 among different isolates, illustrates that this mobile element introduces plasticity to staphylococcal genomes.

An interesting aspect of staphylococcal proline auxotrophy is that it is conditionally imposed by CCR and can be reversed by nonpreferred carbohydrates (e.g., gluconic acid, arabinose, sorbitol, and ribose) (Fig. 7). This conditional nature of proline prototrophy suggests that proline synthesis may play a role in staphylococcal growth under environmental conditions where proline is not in abundant supply (outside mammalian hosts). On the other hand, it also raises the question of why, in S. aureus, proline synthesis is coupled to carbon catabolism and how the coupling can be beneficial for staphylococcal survival. We think the conditional proline auxotrophy by CCR is a staphylococcal adaptation to its mammalian hosts. Since mammalian body fluids contain not only the preferred carbon source glucose but also proline (31), by shutting down proline synthesis, as well as the nonpreferred carbon source metabolism, the bacterium can conserve more energy, resulting in a fitness gain. Indeed, during in vivo growth, amino acid auxotrophs of other bacterial pathogens display an advantage over prototrophic strains. In patients with cystic fibrosis, amino acid-auxotrophic variants of Pseudomonas aeruginosa and Burkholderia cepacia were frequently isolated, and they appeared to have been selected from prototrophic wild-type strains (5-7). In addition, in a murine abscess model, the proC mutant S. aureus Newman did not show any reduction in its survival, suggesting that shutting down proline synthesis would not hamper its growth in animal hosts. Certainly, more studies are needed to identify the exact roles of the coupling of carbon metabolism and proline synthesis in S. aureus pathogenesis.