Transcriptome analysis of the Mg²⁺-responsive PhoP regulator in Yersinia pestis⋆

Abstract

PhoP was previously shown to be important for Yersinia pestis survival in macrophage and under macrophage-induced stresses. In this work, a phoP disruptant of Y. pestis 201 was generated using the Red cloning procedure. The transcription profile of the wild-type Y. pestis was then compared with that of the phoP mutant under Mg²⁺-limiting conditions. It was revealed that PhoP/PhoQ governed a wide set of cellular pathways in Y. pestis, especially including the positive regulation of many metabolic processes, Mg²⁺ transport, peptidoglycan remodeling, lipopolysaccharide (LPS) modification and various stress-protective functions. The Mg²⁺ transport system regulated by PhoP may make Y. pestis to maintain the magnesium homeostasis under low Mg²⁺ environments. The PhoP-controlled stress-protective functions likely constitute the molecular basis for the observation that mutation of the phoP gene rendered the bacteria more sensitive to various macrophage-induced stresses. Modification of LPS mediated by PhoP is hypothesized to not only neutralize negative charges as normally done by Mg²⁺ ions, but also mediate the resistance of Y. pestis to antimicrobial peptides. The microarray results provide a population of candidate genes or pathways, and further biochemical experiments are needed to elucidate the PhoP-dependent mechanisms by which Y. pestis survives the antibacterial strategies employed by host macrophages.

Yersinia pestis, PhoP/PhoQ, DNA microarray, Transcriptional profiling

1 Introduction

Yersinia pestis is considered as a facultative intracellular pathogen capable of surviving and growing within macrophages [1–4]. When the Y. pestis-infected flea bites a person, it regurgitates thousands of Y. pestis cells into the mammalian host. Professional phagocytes ingest these bacteria readily. Bacteria that are ingested by neutrophils are killed. However, a few bacilli are taken up by tissue macrophages, which do not eliminate the bacteria, but instead provide a supportive environment for their unwelcome guests to survive, proliferate and synthesize virulence determinants. The ability of Y. pestis to survive in macrophages is considered to be critical to the early phase of plague pathogenesis, and then bacteria are released from macrophages into the extracellular environment, where they resist phagocytosis by the polymorphs. The resulting infection quickly spreads to the draining lymph nodes, which become hot, swollen, tender, and hemorrhagic, giving rise to the characteristic black buboes responsible for bubonic plague.

When Y. pestis cells are phagocytosed by a macrophage, they are usually found in a vacuole, the phagolysosome [2]. Macrophages may limit the availability of Mg²⁺ in the phagolysosome; a low Mg²⁺ concentration is indicative of an intracellular environment, while a high Mg²⁺ concentration denotes extracellular fluids [5]. PhoQ, an inner-membrane-bound sensor-kinase, is inactive when bound with Mg²⁺; however, when the extracellular Mg²⁺ concentration drops, Mg²⁺ dissociates from PhoQ, leaving it activated. PhoQ then phosphorylates PhoP that either activates or represses transcription of specific target genes. In Salmonella, the two-component regulatory system, PhoP/PhoQ, signals the bacterium presence in an intracellular (low Mg²⁺) or extracellular (high Mg²⁺) environment, thereby inducing the transcription of genes required for survival within or entry into host cells [6]. Salmonella strains harboring null alleles of the phoP or phoQ gene are unable to survive within macrophages and are highly attenuated for virulence in mice [6].

The phoP mutant of Y. pestis showed a reduced ability to survive in macrophages and under in vitro macrophage-induced conditions of low pH, oxidative stress, high osmolarity and antimicrobial peptides [7,8]. The mean lethal dose of the phoP mutant in mice was increased 75-fold in comparison with that of the wild-type strain [8]. The protein profiles of the wild-type and phoP mutant strains, grown at either 28 or 37 °C, revealed more than 20 differences [8]. Modification of LPS in Y. pestis was found to be under the control of the PhoP/PhoQ system [7,9].

DNA microarray was successfully used in screening out various Y. pestis genes whose transcription was differentially regulated by environmental perturbations [10–14]. In the present work, the global gene transcriptional patterns were compared using DNA microarray between the wild-type (WT) strain and the phoP mutant grown under low Mg²⁺ conditions. Genes with changed transcriptional level are considered as the possible members controlled by the PhoP responsive regulator.

2 Materials and methods

2.1 Bacterial strains and mutant construction

Y. pestis WT strain 201 was isolated from Microtus brandti in Inner Mongolia, China. It has the phenotypes as F1⁺ (able to produce fraction 1 antigen or the capsule), VW⁺ (presence of V antigen), Pst⁺ (able to produce pesticin) and Pgm⁺ (pigmentation on Congo-red media). Strain 201 has an LD₅₀ of less than 100 cells for mice by subcutaneous challenge. Strain 201 belongs to a newly established Y. pestis biovar, Microtus [15]. Biovar microtus strains are supposed to be avirulent to humans, although they are highly lethal to mice [15].

The phoP mutant of Y. pestis was generated by using the one-step inactivation method based on the lambda phage recombination, as previously described for Escherichia coli [16]. Briefly, the helper plasmid pKD46 was first transformed into Y. pestis 201. The phop::Cm mutagenic cassette was amplified by PCR from the pKD3 plasmid with the primers phoP-CM-F (5′-atgcgggttctggttgtggaagataacgcgttgttgcgtcaagattgcagcattacacg-3′) and phoP-CM-R (5′-ctagttgacgtcaaaacgatatccctgaccacgaatagtcgtgtaacgcactgagaagc-3′), and then transformed into strain 201/pKD46. Mutants were selected by plating the electroporated cells on the agar plates containing chloramphenicol. Colonies of resistant transformants were subsequently selected. Chromosomal integration of the mutagenic cassette was confirmed by PCR and sequencing. The mutants were incubated at 37 °C overnight and then tested for the loss of the temperature-sensitive pKD46 plasmid by looking for penicillin sensitivity. The elimination of the helper plasmid was verified by PCR.

2.2 Bacterial growth and RNA isolation

A chemically defined TMH medium [17] was used for cultivating the bacteria. Both the WT strain and the phoP mutant were pre-cultivated in the TMH media containing 20 mM MgCl₂ to an OD₆₂₀ of about 0.6. Bacteria were harvested, washed twice with TMH medium containing 10 μM MgCl₂, and then resuspended in the 10 μM Mg²⁺ TMH media. The cell cultures were 1:20 diluted in the fresh 10 μM Mg²⁺ medium. Y. pestis grows much more slowly in vitro at 37 °C than 26 °C, so we cultivated the bacteria to middle exponential phase (OD₂₆₀ about 0.6) at 26 °C, and transformed the cell cultures to 37 °C, the temperature encountered by the pathogen during infection, for 1 h before harvested for RNA isolation. A previous microarray analysis indicated that the upshift of temperature from 26 to 37 °C for 1 h readily induced the expression of the virulence-related genes in Y. pestis (data not shown).

Immediately before being harvested, bacterial cells were mixed with RNAprotect Bacteria Reagent (Qiagen) to minimize RNA degradation. Total cellular RNA was isolated using the MasterPure™ RNA Purification kits (Epicenter). RNA quality was monitored by agarose gel electrophoresis and RNA quantity was measured by spectrophotometer.

2.3 DNA microarray

Based on the genomic sequences of Y. pestis CO92 [18] and 91001 [19], a total of 4005 open reading frames (ORFs) representing about 95% of non-redundant annotated genes of the two Y. pestis strains were successfully amplified by PCR using gene-specific primers [20]. Each of the purified amplicons was spotted in duplicate on the CSS-1000 silylated glass slides (CEL) by using a SpotArray72 Microarray Printing System (Perkin–Elmer Life Sciences) to construct the DNA microarray.

2.4 Probe synthesis and microarray hybridization

Fifteen to twenty micrograms of RNA was used to synthesize cDNA in the presence of aminoallyl-dUTP, genome directed primers (GDPs) and random hexamer primers with the Superscript II system (Invitrogen). The reverse transcription of bacterial RNA by the mixture of GDPs and random hexamers has been proven to be more effective and reliable than with either GDPs only or random hexamers only [11]. The aminoallyl-modified cDNA was then labeled by Cy5 or Cy3 monofunctional dye (Amersham) as described previously [11]. The Cy3 and Cy5 reaction mixtures were combined, and the unincorporated dye was removed using QiaQuick columns (Qiagen). The purified probes were dried in SpeedVac.

The slides were crosslinked by using a UV Stratalinker (Hoefer). NaBH₄ was used to block the free aldehyde groups on the slide surface. The slides were prehybridized in a buffer containing 5 × SSC, 0.1% SDS and 0.1% BSA, and then washed and blown to dry by hot air. The Cy3/Cy5 differentially labeled cDNA samples were resuspended in hybridization solution (50% deionized formamide, 5 × SSC, 0.1% SDS, 5 × Denhardt's solution, and 0.5 μg/μl of sheared salmon sperm DNA), and then hybridized with the slides at 42 °C for 18–20 h. After hybridization, the slides were washed in 1 × SSC with 0.06% SDS for 2 min, then in 0.06 × SSC for 2 min and finally in ethanol for 2 min. The slides were blown to dry by hot air and were scanned by using a GenePix Personal 4100A Microarray Scanner (Axon Instruments).

2.5 Microarray data analysis

The scanning images were processed and the data were further analyzed by using GenePix Pro 4.1 software (Axon Instruments) combined with Microsoft Excel software. Spots were analyzed by adaptive quantitation, and the local background was subsequently subtracted. Spots with background-corrected signal intensity (median) in both channels lower than 2-fold of background intensity (median) were rejected from further analysis, and then the remaining data points were normalized by total intensity normalization methods. The normalized log₂ ratio of test/reference signal for each spot was recorded. The averaged log₂ ratio for each gene with at least four data points was finally calculated. Significant changes of gene expression were identified with Significance Analysis of Microarrays (SAM) software [21].

2.6 Computational search for PhoP-binding sites

The 500-bp-long promoter regions upstream from the start codon of the 706 PhoP-regulated genes, revealed by the microarray analysis, were retrieved with the retrieve-seq program at http://rsat.ulb.ac.be/rsat/. The (T/G)GTTTA-N₅-(T/G)GTTTA motif was searched for within their sequences, with the dna-pattern program at http://rsat.ulb.ac.be/rsat/, setup with a 2-bp mismatch.

3 Results and discussion

3.1 Overview of microarray analysis

The phoP gene was disrupted successfully by a λ Red recombination system-based approach in which PCR primers provide the homology to the targeted gene(s) [16]. For the identification of the members of genes under the control of the PhoP regulator, the transcriptional profile of the phoP disruptant (test sample) was compared to that of the WT strain (reference sample) grown in the presence of 10 μM MgCl₂. Competitive hybridization of fluorescently labelled cDNA was performed to determine the relative ratios of mRNA levels under the paired (test/reference) conditions.

Several controls in the microarray procedures were conducted to ensure the reliability of the data. First, each gene or ORF was present in duplicate on the printed slides. Second, for each condition, total RNA was extracted from two independent biological replicates. Third, each RNA preparation was used to make probes for two separated slides as technical replicates, for which the incorporated dye was reversed. Fourth, significant changes of gene transcription level were identified with SAM software using one class mode by choosing a delta value based on the false discovery rate. A few genes whose transcription levels showed a change less than 2-fold, but were identified to be differentially regulated by SAM, were discarded. In addition to the genes identified by SAM, we still included genes, belonging to an operon in which most gene members were identified by SAM to be differentially regulated, which showed 2-fold variation at mRNA level at least, but did not fulfill the criteria of SAM [22]. Finally, no obvious transcriptional change was detected for phoQ that is the very first gene downstream of phoP according to the microarray results, indicating that this is no polar effect of the phoP mutation.

Transcription of 403 genes was repressed in the phoP mutant (PhoP-activated), while 303 genes induced (PhoP-repressed). Fig. 1 shows the functional classification of all the differentially regulated genes, giving an overall picture of the alteration of the global gene transcription pattern of Y. pestis affected by the PhoP responsive regulator under the Mg²⁺-limiting conditions (see Supplementary Tables 1 and 2 for a complete list of up- and down-regulated genes).

1

Differentially regulated genes grouped by functional classification according to Y. pestis CO92 genome annotation. Gene expression in the phoP mutant (test sample) is compared with that in the WT strain (reference sample) under the low (10 μM) Mg²⁺ stimulating condition. The differentially regulated genes on the chromosome are divided into various functional categories, and those on the three plasmids are listed additionally. The number of genes up- or down-regulated for each functional group is represented.

1

Selected cell processes controlled by PhoP

Fold change	Gene ID	Gene name	Product
Mg2+transport system
−4.90	YPO1660	mgtC	Mg(2+) transport ATPase protein C
−12.36	YPO1661	mgtB	Mg(2+) transport ATPase protein B
Oxidative defense
−8.80	YPO1207	katA	Catalase
−22.49	YPO3319	katY	Catalase-peroxidase
−5.06	YPO3194	ahpC	Alkyl hydroperoxide peroxidase subunit C
−3.72	YPO3375	sodC	Superoxide dismutase [Cu–Zn] precursor
−3.22	YPO2386	sodB	Superoxide dismutase [Fe]
−4.81	YPO4061	sodA	Superoxide dismutase [Mn]
−10.13	YPO2367	gst	Glutathione S-transferase
Heat shock response
−12.45	YPO0105	hslU	ATP-dependent protease
−12.17	YPO0106	hslV	ATP-dependent protease
−3.37	YPO0139	hslO	Heat-shock chaperonin
−2.77	YPO0140	hslR	Heat shock DNA/RNA binding protein
−26.32	YPO0350	groES	10 kDa chaperonin
−20.17	YPO0351	groEL	60 kDa chaperonin
−7.56	YPO0468	dnaK	Chaperone protein DnaK
−3.45	YPO0469	dnaJ	Chaperone protein DnaJ
−4.36	YPO1107	grpE	Heat shock protein GrpE
−2.71	YPO1706	htpX	Putative heat shock protein
−10.63	YPO3119	htpG	Heat shock protein HtpG
−2.98	YPO3731	hupA	DNA-binding protein HU-α
−2.59	YPO4084	ibpB	Heat shock protein
−4.96	YPO4085	ibpA	Heat shock protein
Other stress responsive genes
−11.23	YPO3970	uspA	Universal stress protein A
−3.04	YPO3969	uspB	Universal stress protein B
−8.84	YPO2510	dps	Stress responsive DNA-binding protein
−5.26	YPO2301		Putative stress protein
−2.99	YPO0495	imp/ostA	Organic solvent tolerance protein precursor
7.70	YPO2349	pspC	Phage shock protein C
8.81	YPO2350	pspB	Phage shock protein B
10.89	YPO2351	pspA	Phage shock protein A
7.36	YPO3643	cspA2	Major cold shock protein CspA2
7.46	YPO3644	cspA1	Major cold shock protein CspA1
Hyperosmolarity-inducible genes
−20.82	YPO3938	glgP/glgY	Glycogen phosphorylase
−3.78	YPO3939	glgA	Glycogen synthase
−4.23	YPO3940	glgC	Glucose-1-phosphate adenylyltransferase
−2.34	YPO3941	glgX	Putative α-amylase
−2.76	YPO3942	glgB	1,4-α-glucan branching enzyme
Peptidoglycan remodeling
−3.74	YPO1320	dacC	d-alanyl-d-alanine carboxypeptidase
−2.11	YPO0552	mraY/murX	Phospho-N-acetylmuramoyl-pentapeptide-transferase
−4.97	YPO1715	ybjR	Probable N-acetylmuramoyl-l-alanine amidase
−2.71	YPO0553	murD	UDP-N-acetylmuramoylalanine-d-glutamate ligase
3.50	YPO0549	ftsI/pbpB	Penicillin-binding protein 3
3.45	YPO1078	mltD/dniR	Membrane-bound lytic murein transglycosylase D precursor
2.62	YPO2922		Putative membrane protein
Synthesis and modification of LPS
−2.29	YPO0054	kdtX/waaE	Lipopolysaccharide core biosynthesis glycosyl transferase
−2.37	YPO0057	rfaF/waaF	ADP-heptose–LPS heptosyltransferase II
−2.97	YPO0058	rfaD/trM	ADP-l-glycero-d-manno-heptose-6-epimerase
−2.43	YPO3243	lpcA/gmhA	Phosphoheptose isomerase
−15.78	YPO2174	ugd/pmrE	Putative nucleotide sugar dehydrogenase
−2.52	YPO2418	pmrK	Dolichyl-phosphate-mannose-protein mannosyltransferase-family protein
−3.34	YPO2419	pmrJ	Conserved hypothetical protein
−9.58	YPO2420	pmrI	Probable formyl transferase
−15.60	YPO2421	pmrF	Probable glycosyl transferase
−12.36	YPO2422	pmrH	Conserved hypothteical protein
−4.52	YPO1744	pagP	Putative lipid A palmitoyltransferase
3.50	YPO1057	lpxB/pgsB	Lipid-A-disaccharide synthase
−5.29	YPO3112	prt/rfbS	Paratose synthase
−2.99	YPO3113	ddhC	putative CDP-4-keto-6-deoxy-d-glucose-3-dehydratase
2.68	YPO3116	ddhD/ascD	cdp-6-deoxy-delta-3,4-glucoseen reductase
2.50	YPO3865	wzzE/wzz	Putative lipopolysaccharide biosynthesis protein
Oligopeptide transport system
−2.27	YPO2182	oppA	Periplasmic oligopeptide-binding protein precursor
−2.18	YPO2183	oppB	Oligopeptide transport system permease protein
−3.47	YPO2184	oppC	Oligopeptide transport system permease protein
−5.46	YPO2185	oppD	Oligopeptide transport ATP-binding protein
−3.52	YPO2186	oppF	Oligopeptide transport ATP-binding protein
Regulators
−2.20	YPO0010		Putative GntR-famly transcriptional regulator
−2.63	YPO0175	crp	cAMP-regulatory protein
−3.93	YPO0414		Putative transcriptional regulator
−5.06	YPO0736		Putative regulatory protein
−2.64	YPO0849	lacI	Lactose operon repressor
−4.25	YPO1279		Putative transcriptional regulatory protein
−4.91	YPO2449		Putative LuxR-family regulatory protein
−3.75	YPO2681	celD	Cel operon repressor
−2.85	YPO3063	gcvR	Glycine cleavage system transcriptional repressor
−2.48	YPO3205	phoB	Phosphate regulon transcriptional regulatory protein
−3.11	YPO3223	crl	Curlin genes regulatory protein
−8.39	YPO3279	yfiA	Putative sigma 54 modulation protein
−4.44	YPO3589	pyrI	Aspartate carbamoyltransferase regulatory chain
−2.69	YPO3723	iclR	Acetate operon repressor
−7.98	YPO3737	rsd	Regulator of sigma D
−3.01	YPO3840		Putative TetR-family regulatory protein
−2.29	YPO3848	cyaA	Adenylate cyclase
2.43	YPO0023	ntrB	Two component regulatory protein involved in nitrogen assimilation
2.22	YPO0114	metJ	Transcriptional repressor protein
2.19	YPO0276		Putative LysR-family transcriptional regulator
2.11	YPO0543	fruR	Putative fructose repressor
2.50	YPO0985	yspR	Quorum-sensing transcriptional regulator
8.84	YPO2374	slyA	MarR-family transcriptional regulatory protein
2.05	YPO2378		TetR-family transcriptional regulatory protein
2.85	YPO2593		Putative LuxR-family regulatory proteins
3.09	YPO3143	glnK	Nitrogen regulatory protein P-II

Fold change is present as the mean change of mRNA level for each gene under the paired growth conditions (phoP mutant versus WT strain). The positive number stands for fold up-regulated, while minus down-regulated.

2

Classification of metabolism-related genes whose transcription was affected by the phoP mutation

Functional category [number of genes annotated on the chromosome of Y. pestis CO92]	Number of genes
	PhoP-activated	PhoP-repressed
Degradation of small molecules
General [18]	6	1
Carbon compounds [66]	7	2
Amino acids [23]	5	0
Degradation of macromolecules
RNA [13]	0	1
DNA [16]	4	0
Proteins, peptides and glycopeptides [37]	12	5
Polysaccharides [13]	0	0
Energy metabolism
Glycolysis [12]	6	0
Pyruvate dehydrogenase [4]	1	0
Tricarboxylic acid cycle [15]	13	0
Pentose phosphate pathway [7]	4	0
Entner-Doudoroff pathway [2]	1	0
Aerobic respiration [20]	16	0
Anaerobic respiration [24]	1	4
Electron transport [33]	6	1
Fermentation [3]	2	0
ATP-proton motive force [9]	0	0
Glyoxylate bypass [3]	3	0
Central intermediary metabolism
General [63]	7	3
Gluconeogenesis [5]	3	0
Sugar nucleotides [11]	1	0
Amino sugars [5]	0	0
Sulphur metabolism [13]	2	0
Amino acid biosynthesis
General [1]	0	0
Glutamate family [19]	2	1
Aspartate family [24]	6	1
Serine family [8]	1	0
Aromatic amino acid family [19]	3	0
Histidine [8]	2	0
Pyruvate family [17]	2	2
Polyamine synthesis [5]	0	1
Purines, pyrimidines, nucleosides and nucleotides
Purine ribonucleotide biosynthesis [18]	4	3
Pyrimidine ribonucleotide biosynthesis [8]	1	1
2′-deoxyribonucleotide biosynthesis [8]	3	0
Salvage of nucleosides and nucleotides [16]	4	0
Miscellaneous nucleoside/nucleotide reactions [9]	0	1
Biosynthesis of cofactors, prosthetic groups and carriers [141]	13	4
Fatty acid biosynthesis [33]	6	2
Synthesis and modification of macromolecules
rRNA and stable RNAs [3]	0	0
Ribosomal protein synthesis and modification [59]	1	40
Ribosome maturation and modification [6]	0	2
Protein translation and modification [31]	2	6
RNA synthesis, RNA modification and DNA transcription [29]	0	7
Aminoacyl tRNA synthetases and their modification [40]	6	6
DNA replication, restriction/modification, recombination and repair [77]	5	6
Polysaccharides (cytoplasmic) [5]	5	0
Phospholipids [9]	0	2
Total [1008]	166	106

Transcriptional level of each gene in the phoP disruptant (test sample) was compared with that in the WT strain (reference sample) under the low Mg²⁺ stimulating condition, enabling us to identify genes transcribed under the control of the PhoP regulator.

3.2 Mg²⁺ transport system

Bacteria are proficient at accumulating and concentrating Mg²⁺ from a nutrient-limiting condition. Mg²⁺ presumably enters the bacterial cell initially through porins in the outer membrane. Once in periplasm, Mg²⁺ interacts with the inner membrane Mg²⁺ transport systems. A total of four Mg²⁺ transport systems (CorA, MgtE, MgtA and MgtCB) have been characterized in bacteria [23,24]. Homologues of corA, mgtCB andmgtE, except mgtA, are predicted in Y. pestis [18]. In Salmonella, the mgtA and mgtCB loci are induced by the PhoP/PhoQ [23,24]. Our microarray analysis showed that PhoP positively controlled only the mgtCB expression in Y. pestis (Table 1).

3.3 Stress-protective functions

The mutation in phoP rendered the bacteria more sensitive to acid and oxidative killing and to high osmolarity [8]. Oyston and collaborators infected J774 macrophage cell cultures with either WT Y. pestis or the phoP mutant and observed that 34% of the WT bacteria were killed by macrophages 5 h after infection, while 81% of the phoP mutants had been killed by this time [8]. Similar results were observed in the Y. pestis strains used in this study (data not shown).

Our microarray analysis revealed a set of stress responsive genes whose transcription might be affected by the phoP mutation (Table 1). One of the obvious changes was the positive regulation by PhoP of 7 genes (sodC, sodB, sodA, katA, katY, ahpC and gst) involved in protection against oxidative stress. The first three genes encode superoxide dismutases catalyzing the dismutation of the superoxide radical to H₂O₂ and O₂, while katA, katY and ahpC encode catalases and peroxidases responsible for elimination of H₂O₂. Glutathione S-transferase encoded by the additional gst gene also has peroxidase activities that protect cells against H₂O₂-induced cell death.

PhoP appears to positively control the expression of 14 genes encoding major heat shock proteins (MHSPs) (Table 1). The major adaptive responses of bacteria to sudden increase of growth temperature involve the induction of many MHSPs, including chaperones, proteases, transcriptional regulators and other structural proteins [25]. The heat shock response is also induced by acid and oxidative stresses encountered by bacterial cells in phagocytes [26]. The MHSPs play roles in repairing and preventing damages caused by an accumulation of unfolded proteins resulted from the above environmental stresses.

The microarray data showed that PhoP positively controlled the transcription of three genes (uspA, uspB and dps) that might be involved in protecting bacterial cells against multiple stresses. The expression of the universal stress protein A (UspA) can be induced by a large variety of stress conditions [27]. UspA has given its name to a growing orthologous group of proteins, the UspA family, playing roles in defense against DNA damage [28]. Dps is a DNA-protecting protein expressed by bacteria under nutrition-limiting, acid and oxidative stresses [29,30].

It was previously demonstrated that the glgPACXB operon responsible for glycogen accumulation was up-regulated by 2.6- to 6.0-fold in response to hyperosmotic stress [13]. The induced accumulation of cytoplasmic glycogen under hyperosmotic stress is likely to assist in restoring the original volume of the cytoplasm. Thus, Y. pestis may make quick osmoadaptation by accumulation of glycogen to acclimatize itself to the external osmotic pressure. Herein, PhoP appears to play a role in the positive control of the glg operon.

Taken together, the fact that PhoP positively controlled various stress-responsive genes should account for the increased sensitivity of the phoP mutant to the macrophage-induced stresses. Thus, expression of specific stress-protective proteins controlled directly or indirectly by PhoP/PhoQ likely makes Y. pestis escape from macrophage killing.

We still observed the negative regulation by PhoP of the pspABC genes (pspD was not represented on the microarray) and the cspA1 and cspA2 genes. The pspABCD locus, initially identified in E. coli, encodes the phage shock proteins. These proteins help to ensure survival of the bacterium in late stationary phase at alkaline pH, and protect the cell against dissipation of its proton-motive force [31]. In our previous study, transcription of pspABC was induced by 4- to 7-fold in Y. pestis after the treatment by streptomycin [14]. The Y. pestis pspA was also up-regulated as well upon shift from 26 to 37 °C during steady-state vegetative growth [10]. It was assumed that the psp genes play roles in the accommodation of Y. pestis to environmental perturbations.

The major cold shock protein CspA acts as a chaperone to bind nascent mRNA transcripts for preventing the formation of mRNA secondary structure induced by cold shock. CspA1 and CspA2 are almost identical (97.4% identity in 70 amino acids) in Y. pestis and their structural genes are located 260 bp apart on the CO92 chromosome [18]. The cold and heat shock genes are often regulated antagonistically when Y. pestis experience heat or cold shock [12]. Herein, mutation of the phoP gene appears to reciprocally affect the transcription of the cspA genes and several MHSPs genes (see above).

3.4 Modification of LPS

LPS is a major component of the outer membrane. In general, LPS consists of three domains: the hydrophobic membrane anchor known as lipid A, the surface-exposed O-antigen polysaccharide, and the core sugar region connecting the former two blocks [32]. Y. pestis produces rough type of LPS (lipo-oligosaccharide, LOS) without O-antigen polysaccharide due to the inactivation of the O-antigen gene cluster [33].

Analysis of the purified LOS by mass spectrometry showed core oligosaccharides were structurally distinct between the WT and phoP mutant strains of Y. pestis, and the difference mainly occurred in the terminal galactose or heptose [7]. We observed the positive regulation by PhoP of the rfaD/trM, kdtX/waaE, lpcA/gmhA and rfaF/waaF genes (Table 1). The first three genes are involved in the biosynthesis of the LPS precursor ADP-l-glycero-d-manno-heptose [34–36], while the waaF gene encodes a heptosyl-transferase, which catalyzes the transfer of the l-glycero-d-manno-heptose residues to the core oligosaccharide [37]. The PhoP-dependent expression of these four genes might account for the difference in terminal heptose of the core oligosaccharides between WT and phoP mutant strains described above.

The phoP gene was also required for aminoarabinose modification of lipid A in Y. pestis [38]. The enzymes responsible for addition of aminoarabinose to lipid A in Salmonella are encoded by the pmrHFIJKLM operon (also referred to as the pbgP operon) and the ugd/pmrE gene [39]. The free amino groups of aminoarabinose lead to a reduction in the net negative charge of the outer membrane and, accordingly, to a reduced binding of antimicrobial peptides. Data present here showed that PhoP activated the transcription of the pmrHFIJK (missing data for pmr LM), andugd genes (Table 1). While this manuscript was in preparation, Winfield et al. demonstrated by biochemical experiments that Y. pestis has established a direct activation of transcription of pbgP and ugd by the PhoP protein [9].

Another antimicrobial peptides resistance mechanism involves incorporation of an additional fatty acid into lipid A. This modification reduces the permeability of the outer membrane in response to antimicrobial peptides and is thought to increase the stability of the membrane structure [40]. The pagP gene is responsible for this modification. The protein product of YPO1744 in CO92 chromosome [18] is similar to S. typhimurium PagP (52.8% identical in 159 amino acids). Our microarray data showed that the transcription of the predicted Y. pestis pagP gene might be under the positive control of PhoP (Table 1).

In Salmonella, modification of LPS that is controlled by PhoP/PhoQ results in decreased susceptibility to antimicrobial peptides, due to a reduced charge on the core oligosaccharide and lipid A and thus decreasing the initial binding of the positively charged molecules to the surface of the bacterial cell [41]. The Y. pestis phoP mutant was highly sensitive to polymyxin and the cationic antimicrobial peptides [7]. Data presented here make us postulate that PhoP/PhoQ, through activating the expression of specialized genes responsible for LPS biosynthesis, governs the modification of both lipid A and core oligosaccharides in Y. pestis. Mutation analysis of these PhoP-regulated genes should elucidate whether resistance of Y. pestis to antimicrobial peptides is mediated by structural changes in the LOS governed by PhoP/PhoQ.

3.5 Peptidoglycan remodeling

The cell wall of a Gram-negative bacterium consists of a thin peptidoglycan sheet between the plasma (inner) membrane and an outer membrane. Peptidoglycan is a giant macromolecule of periodic structure that forms the polymer to constitute the shape-maintaining structure of the cell wall. Role of PhoP/PhoQ in peptidoglycan remodeling has been observed in Salmonella, where PhoP governs transcription of pcgL and ugtL that involve in peptidoglycan metabolism [42]. We identified 7 genes responsible for peptidoglycan metabolism to be PhoP-regulated (Table 1).

3.6 Metabolism-related genes

We observed the down-regulation of a large set of genes involved in energy metabolism, central intermediary metabolism, degradation of small molecules and macromolecules, and biosynthesis of amino acid, purines, pyrimidines, nucleosides and nucleotides, cofactors, prosthetic groups, carriers and fatty acid in the phoP mutant (Table 2), indicating the retardance of most of the metabolic processes in the phoP mutant under low Mg²⁺ condition. There is the up-regulation of many genes responsible for ribosomal protein synthesis and modification, ribosome maturation and modification, protein translation and modification, and RNA synthesis in the phoP mutant (Table 2), suggesting the negative regulation by PhoP of RNA transcription and protein translation.

3.7 Transport/binding functions

With the ability to transporting molecules such as ions, sugars, amino acids, vitamins, peptides, polysaccharides, hormones and lipids, the bacterial transport/binding proteins are involved in diverse cellular processes such as nutrient uptake, maintenance of osmotic homeostasis, antigen processing, cell division, bacterial immunity, pathogenesis and sporulation, cholesterol and lipid trafficking, etc. In this study, the phoP mutation affects a population of 81 genes with putative transport/binding functions (40 PhoP-activated and 41 PhoP-repressed, see Supplementary Tables 1 and 2).

Herein, our interest was mainly focused on the oppABCDF operon whose transcription is positively controlled by PhoP. Homologues of the PhoP motif ((T/G)GTTTA-N₅-(T/G) GTTTA) [43,44] were found in the promoter region of this operon (Supplementary Table 3). The Opp system is composed of five subunits: an extracellular oligopeptide-binding protein (OppA) which captures oligopeptides from the environment, two transmembrane proteins (OppB and OppC) forming the pore for internalization of the captured oligopeptides, and two proteins (OppD and OppF) in charge of ATP hydrolysis [45]. OppA also binds toxic peptides, for instance, the phytotoxic tripeptide phaseolotoxin that is toxic to both bacteria and plants [46]. In addition, OppA is implicated in protein folding and protection from stress in the periplasm [47]. The roles of PhoP-regulated expression of the opp genes in Y. pestis need to be elucidated.

3.8 Broad regulatory functions

Transcriptional change of a gene encoding regulatory protein may have pleiotropic effects on bacterial physiology. In our study, transcription of 26 genes (Table 1) with putative regulatory functions was affected by the phoP mutation. PhoP/PhoQ, responding the low Mg²⁺ signal, appears to be a global regulator that controls a complex regulatory cascade by a mechanism of not only directly controlling the expression of specific genes or operons, but also indirectly regulating various cellular pathways by acting on a set of local or dedicated regulators.

3.9 Computational search for PhoP-binding sites

The motif of (T/G)GTTTA-N₅-(T/G) GTTTA, determined by DNaseI footprint experiments, was found in the promoter region of many PhoP-controlled genes in E. coli and Salmonella [43,44,48,49]. Herein, the promoter-proximal regions of the PhoP-regulated genes determined by DNA microarray were scanned for the presence of the conserved motif. This analysis revealed 99 genes (both up- and down-regulated) that were likely under the direct control of PhoP (see Supplementary Table 3). For 10 of them, distinct PhoP motifs were found more than once on the same intergenic region.

Searching the upstream regulatory regions of these genes for the PhoP consensus sequences can help to identify the candidate genes directly regulated by PhoP. However this kind of in silico analysis is not definitive for two reasons. First, PhoP bind to sites that do not conform exactly to the defined consensus site [44]. Second, the mere presence of a PhoP consensus-binding site does not necessarily mean PhoP in vivo binds the site. Further biochemical experiments are required to determine whether these genes are the functional targets for PhoP-specific binding.

3.10 Final remarks

The two-component system PhoP/PhoQ has been showed to be required for the Y. pestis growth in macrophages and under macrophage-induced stresses [8]. Herein, genomic transcriptional profiling of the Y. pestis phoP mutant strain in response to low Mg²⁺ stimulus revealed at least 706 genes whose expression was dependent, directly or indirectly, on the responsive PhoP regulator. PhoP/PhoQ appears to govern a complex of cellular pathways in Y. pestis under Mg²⁺-limiting environment. Several general trends can be elicited on the basis of the functional classification of the differentially regulated genes. First, PhoP/PhoQ governs Mg²⁺ transport systems to maintain the magnesium homeostasis under low Mg²⁺ environment. Second, a population of genes responsible for stress-protective functions were PhoP-activated, which may constitute the molecular basis for the observation that mutation of the phoP gene renders the bacteria more sensitive to several macrophage-induced stresses. Third, the modification of LPS, which is hypothesized to not only neutralize negative charges under Mg²-limiting conditions, which is normally done by the Mg²⁺ ions, but also mediate the resistance of Y. pestis to cationic antimicrobial peptides that is abundant within macrophage phagosomes. Finally, expression of a wide set of metabolism-related genes were repressed in the phoP mutant, indicating that the phoP mutation appears to retard many of the cellular metabolic functions under Mg²⁺-limiting condition. Further experimental characterization of the function of these PhoP-regulated genes or pathways will provide insight into the mechanisms by which Y. pestis survives the antibacterial strategies employed by host macrophages and further indicate how the pathogen receives signals from the host and escapes from macrophage killing.

⋆

D.Z. and Y.H. contributed equally to this work.

Acknowledgements

Financial supports for this work came from the National Natural Science Foundation of China (No. 30430620).

Appendix A Supplementary data

Supplementary data associated with this article can be found, in the online version at doi:10.1016/j.femsle.2005.06.053.