The Cyclic AMP Receptor Protein, CRP, Is Required for Both Virulence and Expression of the Minimal CRP Regulon in Yersinia pestis Biovar microtus

Abstract

The cyclic AMP receptor protein (CRP) is a bacterial regulator that controls more than 100 promoters, including those involved in catabolite repression. In the present study, a null deletion of the crp gene was constructed for Yersinia pestis bv. microtus strain 201. Microarray expression analysis disclosed that at least 6% of Y. pestis genes were affected by this mutation. Further reverse transcription-PCR and electrophoretic mobility shift assay analyses disclosed a set of 37 genes or putative operons to be the direct targets of CRP, and thus they constitute the minimal CRP regulon in Y. pestis. Subsequent primer extension and DNase I footprinting assays mapped transcriptional start sites, core promoter elements, and CRP binding sites within the DNA regions upstream of pla and pst, revealing positive and direct control of these two laterally acquired plasmid genes by CRP. The crp disruption affected both in vitro and in vivo growth of the mutant and led to a >15,000-fold loss of virulence after subcutaneous infection but a <40-fold increase in the 50% lethal dose by intravenous inoculation. Therefore, CRP is required for the virulence of Y. pestis and, particularly, is more important for infection by subcutaneous inoculation. It can further be concluded that the reduced in vivo growth phenotype of the crp mutant should contribute, at least partially, to its attenuation of virulence by both routes of infection. Consistent with a previous study of Y. pestis bv. medievalis, lacZ reporter fusion analysis indicated that the crp deletion resulted in the almost absolute loss of pla promoter activity. The plasminogen activator encoded by pla was previously shown to specifically promote Y. pestis dissemination from peripheral infection routes (subcutaneous infection [flea bite] or inhalation). The above evidence supports the notion that in addition to the reduced in vivo growth phenotype, the defect of pla expression in the crp mutant will greatly contribute to the huge loss of virulence of this mutant strain in subcutaneous infection.

Yersinia pestis is the causative agent of bubonic and pneumonic plague, a zoonotic disease that still poses great threats to public health. Y. pestis is a dangerous pathogen that not only is able to parasitize the flea in part of its life cycle but also is highly virulent to rodents and humans, causing epidemics of a systemic and often fatal disease (33). Plague is transmitted between mammals by the bite of an infected flea, although infection can also occur following inhalation of aerosols containing Y. pestis and even by direct contact or ingestion. The pathogenicity of Y. pestis involves a diverse array of virulence determinants that are coordinately expressed during different stages of infection, including colonization, invasion, early intracellular growth, avoidance of host defense, and extracellular proliferation (33).

During infection, Y. pestis must survive in different host response milieus that make bacterial living conditions far from optimal, through appropriate adaptive/protective responses that are primarily reflected by changes in expression of specific sets of genes. Thus, it can rationally be said that regulatory networks that govern a complex cascade of cellular pathways provide Y. pestis with pathogenic mechanisms that operate in a concerted manner.

Studies on the regulation of gene expression in bacteria hint at the existence of global regulators, where a single regulatory determinant controls the expression of many seemingly unlinked target genes that belong to multiplex cellular pathways (20). The cyclic AMP receptor protein (CRP) is an important global regulator that controls transcription initiation for more than 100 genes/operons in Escherichia coli (30). CRP is active only in the presence of cyclic AMP (cAMP), which behaves as a classic small-molecule inducer (4). It functions as a dimer in the form of a cAMP-CRP complex (4). The presence of glucose in the growth medium results in decreased levels of cAMP and CRP (13).

E. coli catabolizes other sugars when the supply of glucose has become depleted, whereas the presence of glucose prevents E. coli from catabolizing alternative sugars. This process is referred to as catabolite repression and is mediated mainly by the cAMP-CRP complex through regulation of the expression of specific sets of genes (4). The cAMP-CRP complex is able to stimulate or repress its direct target genes by binding to a symmetrical consensus DNA sequence, TGTGA-N₆-TCACA (known as the CRP box sequence), located within the upstream promoter regions (9).

The Y. pestis crp gene is composed of an open reading frame containing 633 nucleotides with a G+C content of 46.9% (21). It encodes a deduced protein of 210 amino acids, with a calculated molecular mass of 23,627.41 Da, which is 98.6% identical to the E. coli protein in amino acid sequence, with the same length. There are only three differences in amino acid residues between them (S119 versus A119, N123 versus R123, and I127 versus V127 for Y. pestis and E. coli, respectively).

In this study, we present evidence showing that CRP is important for virulence of Y. pestis and that the crp mutant has divergent degrees of attenuation by subcutaneous and intravenous routes of infection. In addition, the cAMP-CRP complex plays roles in global transcriptional regulation of genes, including two laterally acquired plasmid genes, pla and pst. Our study on the regulation of pla by CRP is similar to that of Kim et al. (15), but we confirmed this regulation in an ancient strain of Yersinia pestis bv. microtus, indicating that this regulation evolved early during speciation of Y. pestis from Yersinia pseudotuberculosis.

MATERIALS AND METHODS

Bacterial strains.

The wild-type (WT) Y. pestis strain 201 belongs to a newly established Y. pestis biovar, Y. pestis bv. microtus (34), which was thought to be avirulent to humans but highly virulent to mice. It was grown in Luria-Bertani (LB) broth or chemically defined TMH medium (26) at 26 or 37°C. E. coli was grown in LB broth at 37°C. When needed, antibiotics were added at the following concentrations: 100 μg/ml for ampicillin, 50 μg/ml for kanamycin, and 34 μg/ml for chloramphenicol.

Disruption of the crp gene.

An in-frame deletion of the crp gene was constructed by using a one-step inactivation method (see Table S1 in the supplemental material for all oligonucleotide primers used in this study), based on lambda phage recombination, in which PCR primers provide the homology to the target gene, as described previously by Datsenko and Wanner (7). The entire coding region of crp was replaced by a kanamycin resistance (Kn^r) cassette, which was verified by PCR and DNA sequencing. The resulting mutant strain was referred to as the Δcrp mutant.

Complementation of the crp mutant.

A PCR-generated DNA fragment containing the crp coding region plus its promoter-proximal region (∼500 bp upstream of the coding sequence) and transcriptional terminator (∼300 bp downstream) was cloned into the pACYC184 vector (GenBank accession number X06403) and verified by DNA sequencing. The recombinant plasmid was subsequently introduced into the Δcrp mutant, giving the complemented mutant strain C-crp. Real-time reverse transcription-PCR (RT-PCR) experiments were performed to assess the crp mRNA levels in the WT, Δcrp, and C-crp strains. For the crp gene, the transcript was lacking in the Δcrp mutant and restored in the C-crp strain relative to that in the WT (data not shown), indicating successful mutation and complementation.

Determination of LD₅₀ values.

Y. pestis strains were grown in LB broth to an optical density at 620 nm (OD₆₂₀) of about 1.0. Bacterial cells were harvested by centrifugation, washed twice with phosphate-buffered saline, and diluted in phosphate-buffered saline. Fifty percent lethal dose (LD₅₀) determination was performed by injecting four to six groups of five or six 6- to 8-week-old female BALB/c mice with serial dilutions of bacterial cultures, via the subcutaneous or intravenous route. Mice were monitored daily for 14 days to calculate the numbers of live mice, and the LD₅₀ was calculated using the Probit method (5).

Determination of bacterial growth in mice.

One hundred microliters containing 765, 5,000, or 515 CFU of the WT, Δcrp, or C-crp strain, respectively, was inoculated into each BALB/c mouse by the intravenous route. At 24 and 48 h postinfection, livers, spleens, and lungs were collected for homogenization. At each time point, four or five mice were tested for each bacterial strain. Serial dilutions of the resulting homogenates were plated onto brain heart infusion (BHI) agar to calculate the numbers of CFU after incubation at 26°C for 48 h. For the Δcrp and C-crp strains, tissue homogenates were plated on BHI agar with kanamycin, but for the WT, BHI agar without antibiotic was used.

According to preliminary experiments, when the Δcrp strain was injected at a dose of <10³ CFU, most of the bacterial cells delivered were rapidly cleared in mice. For determination of the numbers of bacterial cells in mice, the number of initially inoculated CFU of the Δcrp strain used was greatly larger than that of the WT or C-crp strain (see above). Given the dramatic difference within the above CFU counts, the CFU values at each time point were log₁₀ transformed and then normalized. For normalization, the initially inoculated CFU of WT, Δcrp, and C-crp cells were set to be 10³ (a normalized log₁₀ value of 3), and the resulting normalization factors were then applied to all the time points.

Microarray expression analysis.

WT and Δcrp cells were grown at 26°C in TMH medium with the addition of 1 mM cAMP (TMH-1 mM cAMP) to an OD₆₂₀ of about 1.0, diluted 20-fold into fresh TMH-1 mM cAMP for cultivation at 26°C until an OD₆₂₀ of about 1.0, and finally transferred to 37°C for 3 h. Bacterial cells were harvested for isolation of total RNA. Immediately before being harvested, bacterial cultures were mixed with RNAprotect bacterial reagent (Qiagen) to minimize RNA degradation. Total RNA was isolated using a MasterPure RNA purification kit (Epicenter), with removal of contaminated DNA. RNA quality was monitored by agarose gel electrophoresis, and RNA quantity was determined by spectrophotometry.

Gene expression profile differences between the WT and Δcrp strains were compared by using a Y. pestis whole-genome cDNA microarray as described previously (31). RNA samples were isolated from four individual bacterial cultures, as biological replicates, for each strain. Dual fluorescently (Cy3 or Cy5 dye) labeled cDNA probes, with a reversed incorporated dye, were synthesized from the RNA samples and then hybridized to four separated microarray slides. A ratio of mRNA levels was calculated for each gene. Significant changes in gene expression were identified with SAM software (27). After the SAM analysis, only genes with at least a twofold change in expression were collected for further analysis.

Real-time RT-PCR.

Gene-specific primers were designed to produce a 150- to 200-bp amplicon for each gene. Bacterial growth and RNA isolation were performed as described above, except that the contaminated DNA in RNA samples was removed by using an Ambion DNA-free kit. cDNAs were generated by using 5 μg of RNA and 3 μg of random hexamer primers. Using independent cultures and RNA preparations, real-time PCR was performed in triplicate as described previously (23), using a LightCycler system (Roche) together with SYBR green master mix. Based on the standard curve of 16S rRNA expression for each RNA preparation, the relative mRNA level was determined by the classic ΔC_T method (12). 16S rRNA was used to normalize gene expression of all other genes. The transcriptional variation between the WT and Δcrp strains was then calculated for each gene. A mean ratio of 2 was taken as the cutoff of statistical significance.

Preparation of His-CRP.

The entire coding region of the crp gene was cloned directionally into the BamHI and HindIII sites of plasmid pET28a and verified by DNA sequencing. The recombinant plasmid, encoding a His₆-CRP fusion protein, was transformed into BL21λ(DE3) cells. Overexpression of CRP in LB medium was induced by addition of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside). The overexpressed protein was purified under native conditions with nickel-loaded HiTrap chelating Sepharose columns (Amersham). The purified and eluted protein was concentrated to a final concentration of about 0.3 mg/ml, and its purity was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis.

lacZ reporter fusion and β-galactosidase assay.

A 400- to 500-bp promoter region upstream of each gene was cloned directionally into the EcoRI and BamHI sites of plasmid pRS551, expressing lacZ, and verified by DNA sequencing. The recombinant plasmids were introduced into the WT and Δcrp strains. The plasmid pRS551 was also transformed as a negative control. The resulting strains were grown as described for microarray analysis. β-Galactosidase activity was determined for each strain by using the Promega β-galactosidase enzyme assay system. Assays were performed in triplicate.

DNA-binding assays.

DNA-binding assays, including electrophoretic mobility shift assay (EMSA) and DNase I footprinting, were performed as described previously (23). For EMSA, a 400- to 500-bp DNA region upstream of each gene (containing a predicted CRP binding site) or the corresponding cold probe (i.e., unlabeled target DNA) was radioactively labeled, incubated with increasing amounts of purified His-CRP protein, and then subjected to 4% (wt/vol) polyacrylamide gel electrophoresis. In the DNase I footprinting experiments, coding and noncoding strands (containing the predicted CRP binding site) were labeled with γ-³²P at the 5′ end and then incubated with increasing amounts of His-CRP; after partial digestion with DNase I, the resulting fragments were analyzed by denaturing gel electrophoresis. Radioactive species were detected by autoradiography.

Primer extension analysis.

For the primer extension assay (23), we used RNA preparations as described for microarray analysis. An oligonucleotide primer complementary to a portion of the RNA transcript of each gene was employed to synthesize cDNAs from the RNA templates. Electrophoresis of primer extension products was performed with a 6% polyacrylamide-8 M urea gel. The yield of each primer extension product would indicate the mRNA expression level of the corresponding gene in each strain and, furthermore, could be employed to map the 5′ terminus of the RNA transcript for each gene.

Computational promoter analysis.

The 500-bp DNA sequence upstream of the start codon of each gene tested was retrieved with the retrieve-seq tool (11). Matching of the position-specific scoring matrix (PSSM) of E. coli CRP within the promoter sequences was conducted by using the patser-matrix and convert-matrix tools (11). The predicted CRP binding sites of Y. pestis were aligned and displayed by the WebLogo program (3) to generate a sequence logo.

RESULTS

Reduced in vitro growth of Δcrp mutant.

Deletion of the crp gene in the WT strain generated the Δcrp strain. This mutant strain was then complemented in trans to give the C-crp strain. To investigate the requirement of crp for in vitro growth, we compared the growth of the WT, C-crp, and Δcrp strains at 26°C in TMH-1 mM cAMP. The Δcrp mutant grew more slowly than the WT (Fig. 1), with doubling times of approximately 7 h for the Δcrp mutant and 3 h for the WT. The reduced in vitro growth due to deletion of crp was restored in the complemented mutant C-crp strain, indicating that the mutation was nonpolar.

FIG. 1.

Growth curves for Y. pestis in TMH. Overnight Y. pestis cultures of the WT, Δcrp, and C-crp strains (at an OD₆₂₀ of 1.0) were diluted 20-fold into 50 ml of fresh TMH-1 mM cAMP. The OD₆₂₀ values of each strain were monitored for each culture at 2-h intervals until the cultures reached the stationary phase. Experiments were repeated three times.

Attenuation of Δcrp mutant after subcutaneous and intravenous inoculations.

The LD₅₀ values for the WT, C-crp, and Δcrp strains were <10, <10, and 91,833 CFU, respectively, for the subcutaneous route of infection, while those for the intravenous route were <10, <10, and 83, respectively. The crp mutation led to a >15,000-fold increase in the LD₅₀ by the subcutaneous route of infection, while intravenous inoculation resulted in only an about 40-fold increase in LD₅₀ for the Δcrp mutant. For both routes of infection, virulence was restored in the C-crp strain. The survival curve of LD₅₀ results gave more details (see Fig. S1 in the supplemental material). These indicated a stronger role of CRP in regulating the bubonic plague than the septicemic form or the later systemic phases of infection.

Reduced in vivo growth of Δcrp mutant.

We further assessed the ability of the WT, Δcrp, and C-crp strains to survive in the livers, spleens, and lungs of BALB/c mice over a 48-h period after intravenous injection. The Δcrp mutant showed lower bacterial loads in both the spleen and lungs than those of the WT (Fig. 2). After 48 h, >10⁴-fold fewer bacteria were recovered from the livers, spleens, and lungs of Δcrp mutant-infected mice than from those of WT-infected mice. CRP is therefore important for the ability of Y. pestis to grow and multiply in mice during infection. The in vivo growth phenotype of the C-crp strain was similar to that of the WT, ensuring that the crp disruption was responsible for the reduced in vivo growth of the Δcrp mutant.

FIG. 2.

Kinetics of in vivo growth of Y. pestis. Bacteria were inoculated into mice by the intravenous route. At 24 and 48 h postinfection, bacterial loads in livers, spleens, and lungs were determined. The data shown are normalized log₁₀ CFU values for each time point.

Transcriptome analysis of Δcrp mutant by DNA microarray analysis.

By standard dual-fluorescence microarray hybridization experiments, the mRNA levels of each gene were compared between the Δcrp and WT strains grown in TMH-1 mM cAMP. A total of 292 genes were affected by the crp disruption. These genes were distributed in 24 functional categories (see Fig. S2 in the supplemental material), according to the genome annotation of Y. pestis CO92 (21), and represented >6% of the total protein-encoding capacity of Y. pestis, indicating a global regulatory function of CRP.

Identification of direct CRP targets by EMSA and real-time RT-PCR.

The 500-bp promoter region upstream of each CRP-dependent gene disclosed by cDNA microarray analysis was scanned with a PSSM consensus representing the conserved signals for CRP recognition of promoter DNA in E. coli (35). This analysis generated a weight score for each promoter DNA. Higher scores denote higher probabilities of CRP binding.

When a score of 8 was taken as the cutoff value, 38 genes were picked from the microarray data for further investigation by EMSA to determine whether CRP would bind to their upstream promoter regions in vitro. For all of the candidate genes tested, the cAMP-CRP complex bound to the labeled DNAs in a CRP dose-dependent manner (Fig. 3). CRP could not bind to the target DNA in the absence of cAMP, demonstrating that the cAMP-CRP complex specifically recognized and bound to these promoter regions. This result also confirmed that there was no nonspecific binding of the His tag to target DNA.

FIG. 3.

EMSAs. The band of free promoter DNA disappeared with increasing amounts of CRP protein, and a retarded DNA band with decreased mobility turned up, which presumably represented the CRP-DNA complex.

To confirm the specificity of CRP-DNA association, the EMSA experiments still included DNA fragments upstream of two genes, YPO0180 and YPO1099 (gene identifiers for strain CO92) (21), as negative controls. The PCR-generated upstream DNA of YPO0180 used here did not harbor the predicted CRP binding site, while that of YPO1099 gave an extremely low score during the pattern-matching analysis using the CRP PSSM as a reference (35). As expected, both of them gave negative EMSA results, even when the CRP protein was increased to 4 μg in a single reaction mixture.

Microarray data were able to hint at whether a gene was possibly controlled by CRP. Given the limited reliability of microarray data (35), real-time quantitative RT-PCR, using the same RNA preparations as those used in the microarray analysis, was performed to validate the microarray data for the 38 genes examined by EMSA. Except for one gene (YPO2468), RT-PCR and microarray data gave a good agreement. RT-PCR indicated that the YPO2468 gene had no significant transcriptional change in the Δcrp mutant relative to the level in the WT, which is discrepant with the microarray result. Thus, this gene was not considered a direct CRP target. Accordingly, of the 38 genes or putative operons tested here, 37 were shown to be under the direct control of cAMP-CRP (Table 1) by the combined use of microarray analysis, EMSA, and RT-PCR.

TABLE 1.

Collection of 37 members of minimal CRP regulon in Yersinia pestis bv. microtus

Gene name	Mean fold change in mRNA levela		EMSA resultb	CRP regulationc	CRP sited	Positione	Gene IDf	Protein product	Functional class
	Microarray	RT-PCR
pst	−18.28	−12.59	+	A	TAAGAGTGATTCAGATCACAAAA	D/−152/11.15	YPPCP1.05c	Pesticin	pPCP1 plasmid
pla	−8.89	−19.26	+	A	ATTATGTGACTTTGCTCACATAA	R/−148/11.95	YPPCP1.07	Plasminogen activator	pPCP1 plasmid
ypkA	3.08	14.38	+	B	TTTGTGTGATTAACACCACCTTT	D/−462/8.57	YPCD1.72c	Targeted effector protein kinase YpkA	pCD1 plasmid
yfiA	−5.57	−3.47	+	A	AATGTATGATCGAGATCACTTTT	R/−84/10.74	YPO3279	Putative sigma 54 modulation protein	Regulatory functions
rpoH	3.23	2.77	+	B	AAAACGTGATCTAGATTGAACTT	D/−152/8.78	YPO3811	RNA polymerase sigma 32 factor	Regulatory functions
	−7.03						YPO0409 ↑	Periplasmic solute-binding protein	Transport/binding proteins
	−2.14						YPO0410 ↑	ABC transporter permease protein	Transport/binding proteins
	MDg						YPO0411 ↑	ABC transporter permease protein	Transport/binding proteins
	−3.27	−7.43	+	A	AAATTGTGATCTTGGTTACGCCA	D/−118/8.76	YPO0412 ↑	ABC transporter ATP-binding protein	Transport/binding proteins
sdaC	4.84	3.20	+	B	CAATTGAGATCACGATCACGGTA	D/−308/9.19	YPO1321	Serine transporter	Transport/binding proteins
ptsG	−10.34	−2.20	+	A	AATGTTTGATTGCCGTCACGTTT	R/−166/9.35	YPO1608	PTS system, glucose-specific IIBC component	Transport/binding proteins
araF	−2.44	−3.58	+	A	AACTTTTGACTGACATCACAAAA	R/−283/8.54	YPO2255	l-Arabinose-binding periplasmic protein precursor	Transport/binding proteins
mntH	3.07	13.31	+	B	AATTAGTGGGCTTGATCACATAA	D/−321/9.24	YPO2982	Manganese transport protein MntH	Transport/binding proteins
nupC	2.26	2.55	+	B	AATTAGTGGGCTTGATCACATAA	R/−136/9.24	YPO2983	Nucleoside permease	Transport/binding proteins
	−5.14	−3.59	+	A	AAAATGAGAGGTATATCTCAATT	D/−327/9.93	YPO3633	Putative periplasmic binding protein	Transport/binding proteins
gntT	−4.00	−2.05	+	A	GAGATGTGACTTTTATCACAACA	D/−124/9.78	YPO3954	Gluconate permease	Transport/binding proteins
	−3.09	−2.08	+	A	TAATTGTGGTGTAGATCTCATAA	D/−265/9.92	YPO4037	Sugar-binding periplasmic protein	Transport/binding proteins
sdhC	−2.69	−3.11	+	A	AAATCGTGATCCTAATCACTGTT	D/−255/10.35	YPO1109 ↓	Succinate dehydrogenase cytochrome b556 subunit	Energy metabolism
sdhD	−2.77						YPO1110 ↓	Succinate dehydrogenase hydrophobic membrane anchor protein	Energy metabolism
sdhA	−3.64						YPO1111 ↓	Succinate dehydrogenase flavoprotein subunit	Energy metabolism
sdhB	−3.21						YPO1112 ↓	Succinate dehydrogenase iron-sulfur protein	Energy metabolism
sucA	−3.04						YPO1113 ↓	2-Oxoglutarate dehydrogenase E1 component	Energy metabolism
sucB	−3.30						YPO1114 ↓	Dihydrolipoamide succinyltransferase component of 2-oxoglutarate dehydrogenase complex	Energy metabolism
sucC	−3.17						YPO1115 ↓	Succinyl-coenzyme A (succinyl-CoA) synthetase beta chain	Energy metabolism
sucD	−3.12						YPO1116 ↓	Succinyl-CoA synthetase alpha chain	Energy metabolism
cydA	−3.24	−5.88	+	A	AAACTGTGATTTGACTCACTCTG	D/−478/8.4	YPO1117 ↓	Cytochrome d ubiquinol oxidase subunit I	Energy metabolism
cydB	−2.03						YPO1118 ↓	Cytochrome d ubiquinol oxidase subunit II	Energy metabolism
pepT	−2.75	−2.14	+	A	AATATTTGATCATGATCACTTCC	D/−58/8.46	YPO1631	Peptidase T	Degradation of macromolecules
fadD	−2.32	−2.13	+	A	AAATAGTGATGTATATCTCAACC	D/−132/8.27	YPO2074	Long-chain-fatty-acid CoA ligase	Degradation of small molecules
dadA	−6.38	−2.03	+	A	AGATTGTGACTTGTCTCTCAATT	R/−112/9.62	YPO2147	d-Amino acid dehydrogenase small subunit	Degradation of small molecules
idnO	−5.63	−2.825	+	A	AAACTGTGATGCAGATCACTATA	R/−417/12.73	YPO2539 ↓	Gluconate 5-dehydrogenase	Degradation of small molecules
idnK	−2.25						YPO2540 ↓	Thermosensitive gluconokinase	Degradation of small molecules
ompC	−4.98	−3.13	+	A	AAACAGTGAGTTATAGCACATAT	R/−257/9.4	YPO1222	Porin OmpC	Cell envelope
	−3.57	−2.52	+	A	ACTTTGTGACTTAGATCGAATTT	R/−226/10.04	YPO1411	Porin	Cell envelope
ompX	2.41	5.75	+	B	AGTATGTGACCTCCATCACCCAA	R/−199/9.34	YPO2506	Porin	Cell envelope
aspA	−8.16	−20.61	+	A	TATCTGTGATTGCTCTCACACTT	R/−224/9.8	YPO0348	Aspartate ammonia lyase	Amino acid biosynthesis
aroH	2.01	11.92	+	B	GGGAAGTGATGGCGATCACAATA	R/−64/8.87	YPO2411	Phospho-2-dehydro-3-deoxyheptonate aldolase	Amino acid biosynthesis
cspD	−3.90	−12.93	+	A	AAAGTTTGATGTAGCTAACGCTA	R/−185/8.65	YPO1366	Cold shock protein	Adaptation to atypical conditions
aas	−2.54	−5.43	+	A	GAAATGCGATGCTACTCACGGTT	D/−358/9.38	YPO0793	Aas bifunctional protein	Fatty acid biosynthesis
	2.22	27.49	+	B	CTTGTGTGATCAATAGCACACTG	R/−50/9.16	YPO0180	Conserved hypothetical protein	Unknown
	−6.71	−3.93	+	A	AATATGTGCTGGATATAACAGTT	D/−130/9.33	YPO0400	Hypothetical protein	Unknown
	−2.24	−2.48	+	A	AAATAGTGAGCCAAGTAACAAAA	R/−134/8.2	YPO0819	Putative carbonic anhydrase	Unknown
	−11.96	−23.76	+	A	AGCAAGTGATGTAAATCACAAAA	R/−105/8.38	YPO1255	Hypothetical protein	Unknown
	−3.66	−4.73	+	A	GTTATGTGATGCAGATCAAATGT	D/−37/10.93	YPO2436	Hypothetical protein	Unknown
	−6.90	−4.79	+	A	ATTTTGTGACGTAGGTCACTGTA	R/−202/11	YPO2536	Putative d-isomer-specific 2-hydroxyacid dehydrogenase family protein	Unknown
	−3.00	−3.09	+	A	ATTGTGTGAATCATGTCACATTG	D/−84/10.71	YPO2795	Hypothetical protein	Unknown
	−3.84	−6.81	+	A	AAACTGTGATTCTGATCAAGGTT	D/−136/10.11	YPO3151	Conserved hypothetical protein	Unknown
	−4.35	−33.57	+	A	AATATGTGGGTTATATCACTTTT	D/−156/9.93	YPO3647 ↓	Conserved hypothetical protein	Unknown
	−4.09						YPO3648 ↓	Putative 2-hydroxy-3-oxopropionate reductase	Unknown
	−5.72						YPO3649 ↓	Putative gamma carboxymuconolactone decarboxylase	Unknown
	−6.01						YPO3650 ↓	Putative metabolite transport protein	Unknown

^aThe mRNA expression in the crp mutant was compared with that in the WT strain grown in TMH-1 mM cAMP. Positive numbers show increases, while negative numbers show decreases.

^b+, positive result, indicating that the His-CRP protein binds in vitro to the upstream promoter DNA.

^cA, positive and direct control by CRP; B, negative and direct control.

^dMatching of the PSSM consensus of E. coli CRP within the 500-bp promoter regions was performed by the program patser-matrix (see text). The most strongly matched CRP site was indicated with a weight score for each promoter DNA.

^eStrand/distance of CRP site upstream of transcriptional start site/matching score.

^fGene IDs were derived from the genome annotation of Y. pestis CO92. Putative multigene operons are shown in bold, and the arrows indicate the transcriptional organization.

^gMD, missing data.

CRP was previously shown to modify the plasmid copy number in E. coli (10). Herein, real-time quantitative PCR was conducted on selected genes on plasmids pPCP1, pCD1, and pMT1 (Table 2). The data indicated that the crp mutation had no effect on the copy numbers of these three plasmids (Table 2). Thus, CRP-dependent expression of plasmid-borne genes (pla, pst, and ypkA) was not due to a modification of the copy numbers of the corresponding plasmids.

TABLE 2.

Copy numbers of specific genes in the WT relative to those in the Δcrp mutant^a

Plasmid and gene	Fold change in copy number (WT value/Δcrp mutant value)
pPCP1
pst	0.94
pla	1.12
pCD1
ypkA	1.30
yopD	1.35
pMT1
caf1	1.08
YPMT1.33	1.25

^aThe genomic DNA of the WT or Δcrp strain was used as the template. Real-time quantitative PCR was done as described in Materials and Methods. The standard curve was constructed on the basis of the copy number of the 16S rRNA gene. Data for the 16S rRNA gene were used to normalize the data for all other genes. Each experiment was done in triplicate.

CRP greatly activates pla and pst promoters.

The microarray, RT-PCR, and EMSA data showed that CRP stimulated the pla and pst genes. To validate the effect of CRP on the promoter activities of these two genes, we constructed pla::lacZ and pst::lacZ fusion promoters, each consisting of an upstream DNA of the corresponding gene, and then transformed each of them into the WT and Δcrp strains. The β-galactosidase production of these lacZ fusions, which represented the promoter activity of the corresponding gene in each strain, was measured in both the WT and Δcrp strains.

It should be noted that the crp mutation had an effect on the copy number of recombinant or empty pRS551 plasmid (Table 3). Accordingly, a normalized value for the change in the activity of each fusion promoter in the WT relative to that in the Δcrp mutant was calculated to avoid the influence of the copy number of pRS551 (Table 3). For both genes, there was a huge decrease of β-galactosidase activity in the Δcrp mutant compared to that in the WT when they were grown in TMH-1 mM cAMP (Table 3). These observations showed that CRP greatly stimulated the promoter activities of pla and pst.

TABLE 3.

Promoter activity determined with LacZ reporter fusions^a

LacZ fusion	Fold change (WT value/Δcrp mutant value)		Normalized fold change of promoter activity in WT relative to that in Δcrp mutant
	Plasmid copy no.	Miller units
plap-lacZ	19.16	434.79	22.69
pstp-lacZ	7.75	107.72	13.90
Empty pRS551	3.15	6.29

^aThe upstream promoter DNA region of pla or pst was cloned into the pRS551 plasmid to fuse with the promoterless lacZ gene. β-Galactosidase activity (Miller units) was detected as the promoter activity. The copy number of recombinant or empty pRS551 was determined by real-time quantitative PCR, as described in Table 1, with primers specific for the pRS551-borne lacA gene. The detected change in plasmid copy number was set to 1 to generate a normalization factor that was subsequently used to generate the normalized change in promoter activity in the WT relative to that in the Δcrp mutant.

Structural organization of CRP-dependent pla and pst promoters.

The transcription start sites of pla and pst were determined by primer extension experiments (Fig. 4). A strong primer extension product was detected for both pla and pst. Thus, a single CRP-dependent promoter was transcribed for both of them. The −10 and −35 core promoter elements recognized by sigma factor 70 were predicted upstream of the transcription start sites.

FIG. 4.

Primer extension analysis. Electrophoresis of the primer extension products was performed on a 6% polyacrylamide-8 M urea gel. Lanes C, T, A, and G represent the Sanger sequencing reactions. The transcriptional start sites are underlined.

EMSA demonstrated that the cAMP-CRP complex could bind to the promoter regions of pla and pst. In order to locate the precise CRP binding sites within these two promoter regions, DNase I footprinting experiments were performed with the His-CRP protein plus 2 mM cAMP. As shown in Fig. 5, His-CRP protected a distinct DNA region against DNase I digestion in a dose-dependent pattern for both genes. The detected footprint regions were considered CRP binding sites.

FIG. 5.

DNase I footprinting assays. A labeled DNA probe was incubated with various amounts of purified His-CRP (lanes 1, 2, 3, 4, and 5 contained 0, 500, 1,000, 2,000, and 3,000 ng, respectively) and subjected to DNase I footprinting assay. Lanes G, A, T, and C represent the Sanger sequencing reactions. The protected region (bold line) is indicated on the right side of each gel. The DNA sequences of footprints are shown from the bottom (5′) to the top (3′).

The determination of CRP binding sites, transcription start sites, and core promoter elements (−10 and −35 regions) prompted us to depict the structural organization of CRP-activated pla and pst promoters (Fig. 6), giving a map of CRP-promoter DNA interactions for both genes.

FIG. 6.

Structural organization of CRP-activated pla and pst promoters. The top panels show the genomic organization of the pla and pst genes on the pPCP1 plasmid. Transcription/translation start sites, core promoter −10 and −35 elements, and CRP binding sites are depicted for each promoter in the bottom panels to give a map of CRP-promoter DNA interactions.

DISCUSSION

Y. pestis CRP is a global regulator.

At least 6% of the Y. pestis genes were affected by the crp mutation. A collection of 37 genes or operons were under the direct control of CRP; all of them were considered the CRP regulon members of Y. pestis (here the regulon was defined as the collection of direct targets of a specific transcription regulator). They encoded a wide range of proteins, including plasmid-borne proteins, regulators, transport/binding proteins, proteins involved in energy metabolism, proteins involved in degradation of small molecules and macromolecules, proteins involved in fatty acid biosynthesis, porins, cold shock proteins, and various proteins with unknown functions. Among them, nine (rpoH, sdaC, pstG, araF, nupC, gntT, sdhCDAB-sucABCD, pepT, and dadA) have been proven previously to be under the direct control of CRP in E. coli, indicating a considerable conservation of the CRP regulatory cascades between these two bacteria. Taken together, Y. pestis CRP is a global regulator (both an activator and a repressor) that directly controls a complex regulatory cascade. We aligned the predicted CRP binding sites of the 37 direct CRP targets identified in the present study. This generated a PSSM in which each row represented a position while each column represented a nucleotide (see Fig. S3 in the supplemental material). This gave a full description of the uneven composition in each position, i.e., some nucleotides occur much more frequently than others.

CRP activates two laterally acquired genes, pla and pst, in plasmid pPCP1.

Our microarray expression analysis and real-time RT-PCR disclosed that CRP could regulate several laterally acquired genes, including pla, pst, ypkA, and yopD, and both assays demonstrated that CRP activated pla and pst while it repressed ypkA. The microarray data also indicated that CRP might regulate caf1. Our study then focused on two direct CRP targets, pla and pst. Both of them are harbored on the pPCP1 plasmid, which was acquired by Y. pestis through horizontal gene transfer during its speciation (32). The plasminogen activator (Pla) encoded by the pla gene has a combination of proteolytic, adhesive, and invasive functions, which promote Y. pestis dissemination from peripheral infection routes (6, 16-18). The pst gene encodes pesticin, which is a unique bacteriocin. It has muramidase activity that converts cells into stable spheroplasts by slowly degrading murein (28).

Given the facts that the CRP protein is extremely conserved between E. coli and Y. pestis and that the CRP proteins from these two bacteria share an identical consensus box sequence (see below) that represents the conserved signals for CRP recognition of promoter DNA, it can thus be concluded that the laterally acquired genes pla and pst have evolved to integrate themselves into the ancestral CRP regulatory cascade.

The direct regulation of pla by CRP was reported recently by Kim et al. (15). They demonstrated that activation of pla expression required a CRP binding site within the pla promoter region and that alteration of the CRP binding site nucleotide sequence prevented in vitro formation of CRP-DNA complexes and inhibited in vivo expression of pla. The CRP binding site and transcription start site for pla in a Y. pestis bv. medievalis strain (15) were confirmed by our study with a Y. pestis bv. microtus strain, indicating that control of pla by the CRP system should have evolved early during Y. pestis evolution because the Y. pestis bv. microtus strain diverged earlier than the Y. pestis bv. medievalis strain (1).

Y. pestis CRP is a virulence regulator.

Y. pestis is a clone evolved from Y. pseudotuberculosis 1,500 to 20,000 years ago (2). Compared with its progenitor, Y. pestis is highly dangerous, causing lethal diseases, although it is a newly emerged species from the view of evolution. The essential role for CRP in Yersinia enterocolitica virulence was established several years ago (22). Herein, the crp deletion in Y. pestis led to a >15,000-fold loss of virulence by the subcutaneous route of infection, in contrast to the fact that an intravenous inoculation resulted in an about 40-fold increase in LD₅₀ for the crp mutant. Therefore, it seems that CRP is more important for bubonic plague (or infection at subcutaneous sites and in the lymph) than for septicemic plague (or later systemic infection). The reduced in vivo growth phenotype of the crp mutant should contribute, at least partially, to its attenuation by intravenous infection.

Pla is essential for bubonic and primary pneumonic plague (but not for primary and secondary septicemic forms), and it specifically promotes Y. pestis dissemination from peripheral infection routes (subcutaneous infection [flea bite] or inhalation) (19, 24, 25, 29). Since the disruption of the crp gene leads to a great defect in pla expression, it can rationally be said that the defect in pla expression in the crp mutant should contribute to the >15,000-fold loss of virulence in subcutaneous infection.

CRP directly repressed the transcription of ypkA (Table 1), which was supported not only by microarray expression analysis and real-time RT-PCR but also by our preliminary results provided by EMSA, DNase I footprinting, primer extension, and lacZ reporter fusion assay (unpublished data). ypkA and yopJ constitute an operon on the pCD1 plasmid and encode two different Yop effectors of the Yersinia type III secretion system (8). According to the microarray data, CRP activated caf1R, whereas it repressed the caf1 operon. This is surprising because caf1R encodes the positive regulator of the caf1 operon that encodes the F1 capsule antigen (14). To determine whether caf1 is regulated directly or indirectly by CRP, a detailed dissection of CRP-dependent expression of the caf1 gene needs to be done. Taken together, the data reported here show that CRP appears to regulate a wide set of virulence factors in Y. pestis.