Molecular Basis of Yersinia enterocolitica Temperature-Dependent Resistance to Antimicrobial Peptides

Abstract

Antimicrobial peptides (APs) belong to the arsenal of weapons of the innate immune system against infections. In the case of Gram-negative bacteria, APs interact with the anionic lipid A moiety of the lipopolysaccharide (LPS). In yersiniae most virulence factors are temperature regulated. Studies from our laboratory demonstrated that Yersinia enterocolitica is more susceptible to polymyxin B, a model AP, when grown at 37°C than at 22°C (J. A. Bengoechea, R. Díaz, and I. Moriyón, Infect. Immun. 64:4891–4899, 1996), and here we have extended this observation to other APs, not structurally related to polymyxin B. Mechanistically, we demonstrate that the lipid A modifications with aminoarabinose and palmitate are downregulated at 37°C and that they contribute to AP resistance together with the LPS O-polysaccharide. Bacterial loads of lipid A mutants in Peyer's patches, liver, and spleen of orogastrically infected mice were lower than those of the wild-type strain at 3 and 7 days postinfection. PhoPQ and PmrAB two-component systems govern the expression of the loci required to modify lipid A with aminoarabinose and palmitate, and their expressions are also temperature regulated. Our findings support the notion that the temperature-dependent regulation of loci controlling lipid A modifications could be explained by H-NS-dependent negative regulation alleviated by RovA. In turn, our data also demonstrate that PhoPQ and PmrAB regulate positively the expression of rovA, the effect of PhoPQ being more important. However, rovA expression reached wild-type levels in the phoPQ pmrAB mutant background, hence indicating the existence of an unknown regulatory network controlling rovA expression in this background.

INTRODUCTION

Antimicrobial peptides (APs) are the front line of defense against infections in virtually all groups of organisms. Nearly all APs have a net positive charge, and the three-dimensional folding results in an amphipathic structure (12, 33, 54). The action of APs is initiated through electrostatic interaction with the bacterial surface (12, 33, 54, 71). For some peptides, lethality is linked to membrane perturbations, although there is an increasing body of evidence to indicate that APs may have intracellular targets (12). Bacteria have developed diverse strategies to resist APs, and it is generally accepted that resistance to APs is a virulence phenotype. The strategies for resistance include the alteration of cell surface charge, the active efflux of APs by energy-driven transporters, and the proteolytic degradation of APs (55, 62). Recently, we have uncovered a new strategy of resistance based on the neutralization of the bactericidal activity of APs by anionic bacterial capsule polysaccharides from Klebsiella pneumoniae, Streptococcus pneumoniae, and Pseudomonas aeruginosa (14, 48).

Pathogens respond and adapt to the host environment by upregulating those genes necessary for growth and survival, whereas genes deleterious for infectivity might be downregulated. This is achieved by the concerted action of global regulatory networks, and evidence supports the notion that bacterial systems implicated in AP resistance have come under transcriptional control by such global systems. A well-studied example is the PhoPQ two-component regulatory system of Salmonella enterica, which regulates genes necessary for intracellular survival and cellular invasion and is required for resistance to a subset of APs (4, 23–26). However, it is poorly understood how pathogens coordinate the expression of systems mediating AP resistance with those of virulence factors such as adhesins, toxins, or even type III secretion systems.

Yersinia enterocolitica is a human pathogen that causes a broad range of gastrointestinal syndromes (11). To infect humans, Y. enterocolitica must adapt to the host environment, and the bacterium possesses a number of virulence factors that help it colonize the intestinal tract and resist host defense mechanisms (49, 69). Temperature regulates most, if not all, virulence factors of yersiniae (49, 69). One example of a temperature-dependent trait is the expression of the lipopolysaccharide (LPS) O-polysaccharide (OPS). The optimum expression occurs when bacteria are grown at room temperature (RT), 22 to 25°C (5, 10). In contrast, when they are grown at 37°C, the host temperature, only trace amounts of OPS are produced (5, 10). Another example is the expression of invasin (Inv), the major adherent factor encoded by Y. enterocolitica, whose expression in vitro is higher at RT than at 37°C (57–59). The means used by Y. enterocolitica to sense temperature changes and to transduce them into gene regulation are not fully understood. Hitherto, it has only been shown that the global regulator H-NS mediates inv repression whereas RovA acts as a derepressor, hence relieving the negative regulation of H-NS (22, 66). Moreover, Herbst and coworkers (34) showed that high temperature induces conformational changes of RovA that impair its DNA-binding capacity and also render the protein more susceptible to degradation by the ATP-dependent protease Lon.

Various reports suggest that yersiniae modify their lipid A's in response to growth temperature (3, 6, 39, 60, 64, 65), and this is particularly clear for Yersinia pestis. Thus, at 37°C, Y. pestis synthesizes a tetra-acyl lipid A lacking any secondary acylation (39, 64). At 21°C the lipid A is mainly hexa-acylated (39, 64). The substitution of the lipid A with aminoarabinose is also temperature regulated, being higher at 21°C than at 37°C (1, 64). In other bacteria, the lipid A modification with aminoarabinose is linked to the resistance to polymyxin B, commonly used to demonstrate AP resistance in laboratory settings. Not surprisingly, Anisimov and coworkers found a correlation between susceptibility to polymyxin B and the aminoarabinose content of Y. pestis lipid A (1). And recently, the contribution of aminoarabinose to polymyxin B resistance in Y. pestis has been conclusively demonstrated (19).

Previous studies from our laboratory demonstrated that Y. enterocolitica is more susceptible to polymyxin B when grown at 37°C than at 21°C (8). Therefore, we hypothesized that temperature may regulate the expression of Y. enterocolitica lipid A modifications linked to AP resistance. In this work, first we identified Y. enterocolitica lipid A modifications implicated in AP resistance. Second, we showed that temperature regulates their expression. Third, we assessed the role of these lipid A modifications in virulence. And fourth, we studied the regulatory networks controlling the expression of these modifications.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains and plasmids used in this study are listed in Table 1. Unless otherwise indicated, Yersinia strains were grown in lysogeny broth (LB) medium at either 21°C or 37°C. When appropriate, antibiotics were added to the growth medium at the following concentrations: ampicillin (Amp), 100 μg/ml for Y. enterocolitica and 50 μg/ml for Escherichia coli; kanamycin (Km), 100 μg/ml in agar plates for Y. enterocolitica, 50 μg/ml in agar plates for E. coli, and 20 μg/ml in broth; chloramphenicol (Cm), 20 μg/ml; trimethoprim (Tp), 100 μg/ml; tetracycline (Tet), 12.5 μg/ml; and streptomycin (Str), 100 μg/ml.

Table 1.

Bacterial strains and plasmids used in this study

Bacterial strain or plasmid	Genotype or comments	Source or reference
Strains
E. coli
C600	thi thr leuB tonA lacY supE	2
CC118-λpir	Δ(ara-leu)7697 araD139 ΔlacX74 galE galK ΔphoA20 thi-1 rpsE rrpoB argE(Am) recA1
DH5α-λpir	ΔlacU169 (ϕ80lacZΔM15) recA1 endA1 hsdR17 thi-1 gyrA96 relA1 λpir phage lysogen
Y. enterocolitica
8081-R−M+ (YeO8)	Derivative of wild-type strain 8081; pYV+	73
YeO8-ΔpmrF	YeO8, ΔpmrF, pmrF gene is inactivated	This study
YeO8-ΔpagPGB	YeO8, ΔpagP::Km-GenBlock, Kmr, pagP gene is inactivated	This study
YeO8-ΔmanC	YeO8, ΔmanC, internal fragment deleted by double crossover	This study
YeO8-ΔpmrAB	YeO8, ΔpmrAB, internal fragment deleted by double crossover	This study
YeO8-ΔphoPQ	YeO8, ΔphoPQ, internal fragment deleted by double crossover	This study
Yvm927	rovA deletion mutant in JB580v	22
YeO8-ΔpmrF-ΔpagPGB	YeO8-ΔpmrF, ΔpagP::Km-GenBlock, Kmr, pagP gene is inactivated	This study
YeO8-ΔmanC-ΔpagPGB	YeO8-ΔmanC, ΔpagP::Km-GenBlock, Kmr, pagP gene is inactivated	This study
YeO8-ΔmanC-ΔpmrF	YeO8-ΔmanC, ΔpmrF, pmrF gene is inactivated	This study
YeO8-ΔphoPQ-ΔpmrAB	YeO8-ΔphoPQ, ΔpmrAB; internal fragment deleted by double crossover	This study
Yvm927-ΔphoPQ	Yvm927, ΔphoPQ internal fragment deleted by double crossover	This study
Yvm927-ΔphoPQ-ΔpmrAB	Yvm927-ΔphoPQ, ΔpmrAB; internal fragment deleted by double crossover	This study
Yvm927-ΔpmrF	Yvm927, ΔpmrF, pmrF gene is inactivated	This study
Yvm927-ΔpagPGB	Yvm927, ΔpagP::Km-GenBlock, Kmr, pagP gene is inactivated	This study
Plasmids
pGEM-T Easy	Cloning plasmid; Ampr	Promega
p34S-Tp	Source of Tp cassette	18
pGPL01	Firefly luciferase transcriptional fusion suicide vector; Tpr	29
pGPL01Tp	Trimethoprim resistance cassette cloned into PstI site of pGPL01; Tpr	This work
pKD4	Km cassette source for one-step mutagenesis protocol	Pharmacia
pCP20	Plasmid encoding FLP to remove cassettes between FRT sites; mobilizable, sacB for counterselection; Ampr Cmr	37
pKNG101	oriR6K Mob+, sacB for counterselection; Strr	38
pRV1	Suicide vector, derivative of pJM703.1; Cmr	67
pTM100	Mob+, derivative of pACYC184; Cmr Tetr	51
pPROBE-gfp[LVA]	pBBR1-derived plasmid, gfp promoter-probe vector, Mob+; Kmr	52
pBAD30	Mob+, expression promoter arabinose BAD; Ampr	32
pBADYeO8rovA	500-bp wild-type rovA locus cloned into pBAD30; Ampr	This study
pTMYepagP	2-kb wild-type pagP locus cloned into pTM100; Tetr	This study
pGEMTΔphoPQ	pGEM-T Easy containing ΔphoPQ deleted gene; Ampr	This study
pGEMTΔpmrAB	pGEM-T Easy containing ΔpmrAB deleted gene; Ampr	This study
pKNG101ΔpmrF	pKNG101 containing ΔpmrF; Strr	This study
pKNG101ΔpagPGB	pKNG101 containing ΔpagP::Km-GenBlock; Strr Kmr	This study
pKNG101ΔmanC	pKNG101 containing ΔmanC; Strr	This study
pKNG101ΔphoPQ	pKNG101 containing ΔphoPQ; Strr	This study
pKNG101ΔpmrAB	pKNG101 containing ΔpmrAB; Strr	This study
pPROBEYepropmrH	pPROBE containing a 677-bp DNA fragment corresponding to the pmrH promoter region; Kmr	This study
pPROBEYeprougd	pPROBE containing an 889-bp DNA fragment corresponding to the pagP promoter region; Kmr	This study
pPROBEYepropagP	pPROBE containing a 1,096-bp DNA fragment corresponding to the pmrH promoter region; Kmr	This study
pPROBEYeprophoPQ	pPROBE containing a 509-bp DNA fragment corresponding to the pmrH promoter region; Kmr	This study
pPROBEYeprodacB	pPROBE containing a 416-bp DNA fragment corresponding to the pmrH promoter region; Kmr	This study
pRVProrovAlucFF	pRV1 containing rovA::lucFF; Cmr	60
pGPL01TppropmrF	pGPL01Tp containing a 677-bp DNA fragment corresponding to the pmrH promoter region; Tpr	This study
pGPL01TppropagP	pGPL01Tp containing a 1,096-bp DNA fragment corresponding to the pagP promoter region; Tpr	This study
pGPL01Tpprougd	pGPL01Tp containing an 889-bp DNA fragment corresponding to the ugd promoter region; Tpr	This study
pGPL01TpprophoPQ	pGPL01Tp containing a 509-bp DNA fragment corresponding to the phoPQ promoter region; Tpr	This study
pGPL01TpprodacB	pGPL01Tp containing a 416-bp DNA fragment corresponding to the dacB promoter region; Tpr	This study

Y. enterocolitica mutant construction.

In silico analysis led to the identification of Y. enterocolitica 8081 homologues of phoPQ (YE1718, YE1717), pmrAB (YE0423, YE0422), manC (YE3073), pagP (YE1762), and pmrF (YE2191) (EMBL accession number, AM286415 [70]). To obtain the phoPQ, pmrAB, and manC mutants, two sets of primers (Table 2) were used for each gene to amplify two different fragments from each gene, phoPQUP and phoPQDOWN, pmrABUP and pmrABDOWN, and manCUP and manCDOWN, respectively. Both fragments of each gene were BamHI digested, purified, ligated, amplified as a single PCR fragment using Taq polymerase (Promega), gel purified, and cloned into pGEM-T Easy (Promega) to obtain pGEMTΔphoPQ, pGEMTΔpmrAB, and pGEMTΔmanC. A kanamycin resistance cassette, obtained as a 1.5-kb PCR fragment from pKD4 (17), was BamHI digested and cloned into BamHI-digested pGEMTΔmanC to obtain pGEMTΔmanCKm. DNA fragments for pagP (2 kb) and pmrF (1.7 kb) were PCR amplified, gel purified, and cloned into pGEM-T Easy to obtain pGEMTpagP and pGEMTpmrF. pGEMTpagP was amplified by inverse PCR (13) to delete an internal fragment of 250 bp in the coding region of pagP, and a kanamycin resistance cassette, obtained as a 1.4-kb PstI blunt-ended fragment from pUC-4K (Gen-Block; Pharmacia), was cloned into the plasmid obtained by inverse PCR to generate pGEMTΔpagPGB. pGEMTpmrF was also amplified by inverse PCR to delete a fragment of 360 bp in the coding region of pmrF and ligated to obtain pGEMTΔpmrF.

Table 2.

Primers used in this study

Purpose and target gene	Name of primer	Primer sequence (5′–3′)a
Mutagenesis
pmrF	YeO8pmrFR	CGGATCCGGGATATTCACGCCGCTATTGC
	YeO8pmrFF	CGGATCCGCAAAACCCAGTTGATCGGTGC
	InvpmrFF	ACGCTTGGTATTTGGCCCTG
	InvpmrFR	GCGACACTCACCAACCGAGG
pagP	YeO8PagPF	GGTGGTGTTGACTGCGGCAC
	YeO8PagPR	TTCTCTCGCAGTCCCCTTCC
	InvpagPF	TGTATTGTTTACCTGGATGCGTTG
	InvpagPR	ATCGCATATACTGCGTGCCAG
manC	DELTAmancUpF	CCTGCATTGTTACGTCGGTTTC
	DELTAmancupR	CGGATCCGTGATGTGGCAAGTAGCGCAGC
	DELTAmancdownF	CGGATCCGGCATGTAACAATCAAACCGGGAC
	DELTAmancdownR	CGCATCAATTACATGCCCTGC
pmrAB	YeO8BasupF2	TCGTATGGATCACGAGCAGAGC
	YeO8BasupR2	CGGATCCGAAGTCATGGATGAAGGGCCAG
	YeO8BasdownF	CGGATCCGGGAGAGTAACAAACTATGCGCTTG
	YeO8BasdownR	CGCACCTTAGCCAAAGGCTC
phoPQ	YeO8PhoPQupF	GGATATAAAGGATATCGCCAGG
	YeO8PhoPQupR	CGGATCCGGACGATGGGCCGGGTATTCC
	YeO8PhoPQdownF	CGGATCCGGGAAATTCAATAATTTGGGATGC
	YeO8PhoPQdownR	GGTGGGCGGCAGCCGCCATG
Complementation
rovA	PBAdRovAF	GCTCTAGAGCCTTACTTTGTAGTTGAATAATGTTTCTCTCAAGC
	PBADRovAR	CGGAATTCGGTAGTTATGCTAGCACGC
Promoter region
pmrF	YeO8PpmrHR	GGAATTCCTTGCTCTGCTGGCCGCAGTG
	YeO8PpmrHF	GGGTACCCAACTGTTGATTCTGAGGGCCTG
ugd	YeO8prougdF	AGCAGCATGTGCCACGCCTG
	YeO8ProugdR	GGAATTCCGGAGCATTTGGCGCATCAAC
pagP	YeO8PProagPF	ATCGCATATACTGCGTGCCAG
	YeO8ProPagPR	CTTCTTACGTAAACCCTTCTTGG
phoPQ	YeProPhoPF	ATGGCGATATCTGGGGCATG
	YeProPhoPR	GGAATTCCGGTGGGCGGCAGCCGCCATG
dacB	ProYedacBF	GCAAGATTTGCCCCATCAGG
	ProYedacBR	GGAATTCCTGCGCCATTTACCGTCATCG

^aUnderlining indicates the restriction site.

ΔphoPQ, ΔpmrAB, and ΔpagP::GB alleles were obtained by PvuII digestion of pGEMTΔphoPQ, pGEMTΔpmrAB, and pGEMTΔpagPGB, respectively, gel purified, and cloned into SmaI-digested pKNG101 to obtain pKNGΔphoPQ, pKNGΔpmrAB, and pKNGΔpagPGB, respectively. ΔmanCKm and ΔpmrF alleles were amplified using Vent polymerase (New England BioLabs) and also cloned into SmaI-digested pKNG101 to obtain pKNGΔmanCKm and pKNGΔpmrF. pKNG101 is a suicide vector that carries the defective pir-negative origin of replication of R6K, the RK2 origin of transfer, and an Str resistance marker (38). It also carries the sacBR genes, which mediate sucrose sensitivity as a positive selection marker for the excision of the vector after double crossover (38). Plasmids were introduced into E. coli CC118-λpir, from which they were mobilized into Y. enterocolitica 8081 by triparental conjugation using the helper strain E. coli HB101/pRK2013. Transconjugants were selected after growth on Yersinia selective agar medium plates (Oxoid) supplemented with Str. Bacteria from 5 individual colonies were pooled and allowed to grow in LB without any antibiotic overnight at RT. Bacterial cultures were serially diluted in LB without NaCl containing 10% sucrose, and the plates were incubated at RT. The recombinants that survived the 10% sucrose were checked for their antibiotic resistance. The appropriate replacement of the wild-type alleles by the mutant ones was confirmed by PCR and Southern blotting (data not shown). In the case of the YeO8-ΔmanCKm mutant, the kanamycin cassette was excised by Flp-mediated recombination using plasmid pCP20 (37), and the mutant generated was named YeO8-ΔmanC.

YeO8-ΔphoPQ-ΔpmrAB double mutant was obtained by mobilizing the pKNG101ΔpmrAB plasmid into YeO8-ΔphoPQ. YeO8-ΔmanC-ΔpmrF was obtained by mobilizing pKNGΔpmrF into YeO8-ΔmanC whereas YeO8-ΔmanC-ΔpagPGB and YeO8-ΔpmrF-ΔpagPGB were obtained by mobilizing pKNGΔpagPGB into YeO8-ΔmanC and YeO8-ΔpmrF, respectively. The rovA phoPQ double mutant was obtained by mobilizing pKNGΔphoPQ into Yvm927, whereas the rovA phoPQ pmrAB triple mutant was obtained by mobilizing pKNG101ΔpmrAB into Yvm927-ΔphoPQ. Yvm927-ΔpmrF and Yvm927-ΔpagPGB were obtained by mobilizing pKNGΔpmrF and pKNGΔpagPGB into Yvm927, respectively. The replacement of the wild-type alleles by the mutant ones was done as described above and confirmed by PCR (data not shown).

Isolation and analysis of lipid A.

Lipid A's were extracted using an ammonium hydroxide/isobutyric acid method and subjected to negative-ion matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry analysis (21, 60). Analyses were performed on a Bruker Autoflex II MALDI-TOF mass spectrometer (Bruker Daltonics, Incorporated) in negative reflective mode with delayed extraction. Each spectrum was an average of 300 shots. The ion-accelerating voltage was set at 20 kV. Dihydroxybenzoic acid (Sigma Chemical Co., St. Louis, MO) was used as a matrix. Further calibration for lipid A analysis was performed externally using lipid A extracted from E. coli strain MG1655 grown in LB at 37°C. Interpretation of the negative-ion spectra is based on earlier studies showing that ions with masses greater than 1,000 gave signals proportional to the corresponding lipid A species present in the preparation (3, 46, 56, 64). Important theoretical masses for the interpretation of peaks found in this study are as follows: lipid IV_A, 1,405; C₁₂, 182; C₁₄, 210; C_16:1, 236.2; aminoarabinose (AraNH), 131.1; C₁₆, 239.

Antimicrobial peptide susceptibility assay.

Bacteria were grown either at 21°C or 37°C in 5 ml LB in a 15-ml Falcon tube with shaking (180 rpm) and harvested (2,500 × g, 20 min, 24°C) in the exponential growth phase (optical density at 540 nm [OD₅₄₀], 0.8). Bacteria were washed once with phosphate-buffered saline (PBS), and a suspension containing approximately 1 × 10⁵ CFU/ml was prepared in 10 mM PBS (pH 6.5), 1% tryptone soya broth (TSB) (Oxoid), and 100 mM NaCl. Aliquots (5 μl) of this suspension were mixed in 1.5-ml microcentrifuge tubes with various concentrations of polymyxin B or magainin II (both purchased from Sigma). In all cases the final volume was 30 μl. After 1 h of incubation at the bacterial growth temperature, the contents of the tubes were plated onto LB agar. Colony counts were determined, and results were expressed as percentages of the colony count of bacteria not exposed to antibacterial agents. All experiments were done with duplicate samples on at least four independent occasions.

Antimicrobial radial diffusion assay.

The antimicrobial activities of polymyxin B, magainin II, protamine, and HNP-1 (purchased from Sigma) were assayed using a radial diffusion method previously described (44, 48). Briefly, bacteria were grown at either 21°C or 37°C in 5 ml of LB medium in a 15-ml Falcon tube, collected in the exponential phase of growth, and resuspended in PBS. An underlay gel that contained 1% (wt/vol) agarose (SeaKem LE agarose; FMC, Rockland, ME), 2 mM HEPES (pH 7.2), and 0.3 mg of TSB powder per ml was equilibrated at 50°C and inoculated with the different bacteria to a final concentration of 6.1 × 10⁵ CFU per ml of molten gel. Thirty milliliters of this gel was poured into standard square petri dishes (10 by 10 by 1.5 cm), and after solidification of the gel, small wells with capacities of 10 μl were carved. Aliquots of 10 μl of different concentrations of each peptide were added and allowed to diffuse for 3 h at 21°C or at 37°C. After that, a 30-ml overlay gel composed of 1% agarose and 6% TSB powder in water was poured on top of the previous gel, and the plates were incubated overnight at 21°C or at 37°C. The next day, the diameters of the inhibition halos were measured to the nearest millimeter and, after subtraction of the diameter of the well, were expressed in inhibition units (10 inhibition units = 1 mm). The MIC was estimated by performing linear regression analysis (with the number of units versus the log₁₀ concentration) and determining the x axis intercepts. All the experiments were run in quadruplicate on at least three independent occasions.

Animal experiments.

Six- to 7-week-old virus-free male BALB/c mice were inoculated orally with 100 μl of a bacterial mixture containing 10⁸ bacteria of wild-type or lipid A mutant strains. The mixture was serially diluted, and appropriate dilutions were plated to determine exact bacterial counts. At different time points after infection (3 and 7 days), five animals per bacterial strain were euthanized by cervical dislocation, and their spleens, livers, and Peyer's patches were aseptically removed, weighed, and homogenized into 500 μl of PBS. The bacterial loads recovered from the infected organs were determined by plating homogenates and serial dilutions onto Yersinia selective agar medium plates (Oxoid) to select YeO8 and YeO8-ΔpmrF and onto Km-containing LB plates to select YeO8-ΔpagPGB and YeO8-ΔpmrF-ΔpagPGB. Results were reported in log CFU per gram of tissue. Mice were treated in accordance with the European Convention for the Protection of Vertebrate Animals Used for Experimental and other Scientific Purposes (Directive 86/609/EEC) and in agreement with the Bioethical Committee of the University of the Balearic Islands.

Construction of reporter fusions.

A trimethoprim resistance cassette, obtained by SmaI digestion of p34S-Tp (18), was cloned into PstI-digested, blunt-ended pGPL01 (29) to obtain pGPL01Tp. This plasmid contains a promoterless firefly luciferase gene and an R6K origin of replication. DNA fragments containing the promoter regions of the pmrH, pagP, ugd, phoPQ, and dacB loci were amplified by PCR using Vent polymerase, EcoRI digested, gel purified, and cloned into EcoRI-SmaI digested pGPL01Tp suicide vector to obtain pGPL01TpYepropmrH, pGPL01TpYepropagP, pGPL01TpYeprougd, pGPL01TpYeprophoPQ, and pGPL01TpYeprodacB, respectively. Plasmids were introduced into E. coli CC118-λpir or DH5α-λpir, from which they were mobilized into Y. enterocolitica by triparental conjugation using the helper strain E. coli HB101/pRK2013. Strains in which the suicide vectors were integrated into the genome by homologous recombination were selected. This was confirmed by PCR (data not shown). To construct green fluorescent protein (GFP) reporters, EcoRI-digested promoter regions were ligated into the EcoRI-SmaI digested low-copy-number vector pPROBE'-gfp[LVA] (52). pPROBE'-gfp[LVA] derivatives were introduced into E. coli strains by conjugation.

Complementation of pagP and rovA mutants.

To complement the pagP mutant, a DNA fragment of 2.1 kb was PCR amplified using Vent polymerase (see Table 2 for the primers used). The fragment, containing the putative promoter and coding region of the acyltransferase, was gel purified, phosphorylated, and cloned into the ScaI site of the medium-copy-number plasmid pTM100 to obtain pTMYepagP. This plasmid was introduced into E. coli C600 and then mobilized into Y. enterocolitica strains by triparental conjugation with the helper strain E. coli HB101/pRK2013.

To construct an arabinose-inducible rovA, the gene was amplified by PCR using Vent polymerase and primers PBADRovAF and PBADRovAR (Table 2). The PCR product contained the native ribosome binding site, and it was digested with EcoRI and XbaI, purified, and cloned into pBAD30 (32) to give pBADYeO8RovA. This plasmid was mobilized by conjugation to the rovA mutant carrying the pGPL01Tp derivatives previously described.

Luciferase activity.

The reporter strains were grown at 21°C or at 37°C on an orbital incubator shaker (180 rpm) until an OD₅₄₀ of 1.6 was reached. The cultures were harvested (2,500 × g, 20 min, 24°C) and resuspended to an OD₅₄₀ of 1.0 in PBS. A 100-μl aliquot of the bacterial suspension was mixed with 100 μl of luciferase assay reagent (1 mM d-luciferin [Synchem] in 100 mM citrate buffer, pH 5). Luminescence was immediately measured with a Luminometer LB9507 (Berthold) and expressed as relative light units (RLU). All measurements were carried out in quintuplicate on at least three separate occasions.

GFP assays.

The reporter strains were grown at 21°C or at 37°C on an orbital incubator shaker (180 rpm) until an OD₅₄₀ of 1.2 was reached. The cultures were harvested (2,500 × g, 20 min, 24°C) and resuspended to an OD₅₄₀ of 0.6 in PBS. An aliquot (0.8 ml) of this suspension was transferred to a 1-cm fluorimetric cuvette, and fluorescence was measured with a spectrofluorophotomoter (Perkin Elmer LS55) set as follows: excitation, 485 nm; emission, 528 nm; slit width, 5 nm; integration time, 5 s. Results were expressed as relative fluorescence units (RFU). All measurements were carried out in quintuplicate on at least three separate occasions.

Statistical analysis.

The results were analyzed by analysis of variance (ANOVA) or the one-sample t test using GraphPad Prism software (GraphPad Software Inc.). Results are given as means ± standard deviations (SD). A P value of <0.05 was considered to be statistically significant.

RESULTS

Temperature-dependent variations in susceptibility of Yersinia enterocolitica to antimicrobial peptides.

Previous results (8) had shown that Y. enterocolitica strains were more susceptible to polymyxin B when grown at 37°C than at 21°C. Indeed, Y. enterocolitica strain 8081 serotype O:8 (here referred to as YeO8) (Table 1) was more susceptible to polymyxin B and magainin II when grown at 37°C than at 21°C (Fig. 1). The MICs of polymyxin B and magainin II for YeO8 grown at 21°C were also higher than those for bacteria grown at 37°C (Table 3). We asked whether this is also true for other APs. Results shown in Table 3 demonstrate that the MICs of HNP-1, protamine, and magainin I for YeO8 grown at 37°C were significantly lower than those for YeO8 grown at 21°C. The fact that the APs tested are not structurally related suggests that the temperature-dependent mechanism(s) of resistance might be common to all of them.

Fig 1.

Temperature-dependent susceptibility of Y. enterocolitica to antimicrobial peptides. YeO8 grown at 37°C (black symbols) and 21°C (white symbols) was exposed to different concentrations of polymyxin B and magainin II. Each point represents the mean and standard deviation of eight samples from four independently grown batches of bacteria, and significant survival differences (P < 0.05; two-tailed t test) between bacteria grown at 37°C (black symbols) and 21°C (white symbols) are indicated by asterisks.

Table 3.

MICs of antimicrobial peptides for Y. enterocolitica O:8^a

Antimicrobial peptides	MIC (μg/ml) at:
	21°C	37°C
Polymyxin B	3.2 ± 0.5	1.43 ± 0.2
Magainin II	45.4 ± 5.10	10.1 ± 3.0
HNP-1	27.10 ± 6.0	11.06 ± 0.07
Protamine	62.9 ± 12.2	8.11 ± 2.2
Magainin I	78.3 ± 1.5	6.5 ± 1.7

^aP < 0.05 (one-tailed t test) for comparison of MICs between temperatures for a given peptide.

Temperature-dependent variations in LPS structure associated to antimicrobial peptides resistance.

The interaction of APs with the anionic bacterial surface is a necessary event to exert their microbicidal action (12, 33, 54, 71). Not surprisingly, bacteria have developed strategies to limit this initial interaction, including changes in the LPS structure. The best-characterized strategies are the presence of OPS in the LPS molecule and the lipid A modifications with aminoarabinose, phosphoethanolamine, and palmitate (28, 30, 31, 43, 55). To explain the temperature-dependent susceptibility of YeO8 to APs, we reasoned that temperature may modulate LPS structure associated with resistance to APs. Indeed, previous studies from the laboratory had shown that the expression of OPS is downregulated in YeO8 at 37°C (5, 10). We sought to determine whether temperature may also affect the expression of lipid A modifications associated with resistance to APs. Lipid A species synthesized by YeO8 grown at either 21°C or 37°C were characterized by MALDI-TOF mass spectrometry. Previous data (60), further confirmed here, had shown that lipid A from YeO8 grown at 21°C contained predominantly hexa-acylated species (m/z, 1,824) corresponding to two glucosamines, two phosphates, four 3-OH-C₁₄, one C₁₂, and one C_16:1 (Fig. 2A). Other species detected were consistent with the addition of aminoarabinose to the hexa-acylated form (m/z, 1,955) and with the addition of C₁₆ to the hexa-acylated form, producing a hepta-acylated lipid A (m/z, 2,063) (Fig. 2A). Lipid A modifications with aminoarabinose and palmitate are associated with increased resistance to APs (28, 30, 31, 43, 55). Of note, these two lipid A modifications were not detected in lipid A from YeO8 grown at 37°C (Fig. 2B). Perusal of the literature shows that the products of ugd and pmrHFIJKLM (arnBCADTEF) (here referred to as pmrF operon) loci are required for the synthesis and addition of aminoarabinose to lipid A. Ugd converts UDP-d-glucose into UDP-d-glucuronic acid, which is next modified by pmrF operon-encoded enzymes to generate aminoarabinose (63). The gene encoding the acyltransferase, pagP, is required for the addition of palmitate to lipid A (31, 63). To verify that these loci were indeed necessary for YeO8 lipid A modifications with aminoarabinose and palmitate, we analyzed the lipid A structure from YeO8-ΔpmrF, YeO8-ΔpagPGB, and YeO8-ΔpmrF-ΔpagPGB. As expected, pmrF and pagP single mutants grown at 21°C lacked lipid A species containing aminoarabinose and palmitate, respectively (see Fig. S1 in the supplemental material). YeO8-ΔpmrF-ΔpagPGB mutant lacked species containing aminoarabinose and palmitate (see Fig. S1). At 37°C, the lipid A's from the mutant strains were similar to that of YeO8 (see Fig. S1). Taken together, it can be concluded that temperature regulates the expression of lipid A modifications with aminoarabinose and palmitate.

Fig 2.

Lipid A analysis from Y. enterocolitica. Negative-ion MALDI-TOF mass spectrometry spectra of lipid A isolated from YeO8 grown at 21°C (A) or 37°C (B). The results in both panels are representative of three independent lipid A extractions. (C) Proposed structures of the main molecular species follow previously reported structures for Yersinia and other Gram-negative bacteria. The possible addition of aminoarabinose or palmitate to the molecular species m/z 1824 is also depicted.

These results led us to test whether the transcription of these loci is temperature regulated. To monitor transcriptions of pmrF operon, ugd, and pagP, three transcriptional fusions were constructed in which a promoterless lucFF was under the control of each of the promoter regions. These fusions were introduced into YeO8, and the amount of light was determined. The expressions of the three transcriptional fusions were higher at 21°C than at 37°C, thereby giving experimental support to our hypothesis (Fig. 3).

Fig 3.

Temperature regulates the expression of Y. enterocolitica pmrF operon and ugd and pagP loci. Analysis of the expression of the loci implicated in lipid A remodeling by measuring luciferase activity of YeO8 carrying pmrF::lucFF, ugd::lucFF, or pagP::lucFF transcriptional fusions, which were grown at 21°C (white bars) or 37°C (black bars). Data are presented as means ± SD (n = 3). *, results are significantly different (P < 0.05; two-tailed t test) from the results for bacteria grown at 21°C.

Effects of LPS temperature-dependent variations in susceptibility of Y. enterocolitica to antimicrobial peptides.

To delineate the relative contributions of LPS OPS and the lipid A modifications to YeO8 temperature-dependent susceptibility to APs, first we tested the susceptibility of YeO8-ΔmanC (lacking OPS), YeO8-ΔpmrF, and YeO8-ΔmanC-ΔpmrF to polymyxin B (Fig. 4). When bacteria were grown at 21°C, YeO8-ΔmanC and YeO8-ΔpmrF were more susceptible to polymyxin B than YeO8 (P < 0.05 for each comparison versus YeO8-ΔmanC or YeO8-ΔpmrF [one-tailed t test]) (Fig. 4A). No significant differences were found between YeO8-ΔmanC and YeO8-ΔpmrF (P > 0.05 [one-tailed t test]), whereas the most susceptible strain was the double mutant YeO8-ΔmanC-ΔpmrF (P < 0.05 for each comparison versus YeO8-ΔmanC-ΔpmrF [one-tailed t test]) (Fig. 4A). When bacteria were grown at 37°C, no significant differences were found between YeO8, YeO8-ΔmanC, and YeO8-ΔpmrF, whereas the double mutant was the most susceptible strain (Fig. 4A). Nevertheless, the double mutant was still more susceptible to polymyxin B when grown at 37°C than at 21°C (Fig. 4B).

Fig 4.

Roles of Y. enterocolitica OPS and lipid A modifications on the resistance to antimicrobial peptides. (A and B) manC and pmrF mutants, grown at 21°C or 37°C, were exposed to different concentrations of polymyxin B. Each point represents the mean and standard deviation of eight samples from four independently grown batches of bacteria; the results shown in panel A for the double mutant are displayed on their own in panel B for the sake of clarity. *, results are significantly different (P < 0.05; two-tailed t test) from the results for bacteria grown at 21°C. (C) manC and pagP mutants, grown at 21°C (white bars) or 37°C (gray bars), were exposed to magainin II (40 μg/ml, 1 h). Data are presented as means ± SD (n = 5). *, results are significantly different (P < 0.05; two-tailed t test) from the results for wild-type bacteria.

Lipid A substitution with palmitate does not affect resistance to polymyxin B but does affect resistance to other peptides such as magainin II (31, 47). As we expected, YeO8-ΔpagPGB was as resistant as YeO8 to polymyxin B at both growth temperatures (data not shown). When bacteria were grown at 21°C, YeO8 was more resistant to magainin II than YeO8-ΔmanC and YeO8-ΔpagPGB, which displayed similar susceptibilities to the AP (Fig. 4C) (P > 0.05 [one-tailed t test]). pTMYePagP restored the resistance of YeO8-ΔpagPGB to magainin II to wild-type levels (data not shown). The double mutant was the most susceptible strain (P < 0.05 for each comparison versus YeO8-ΔmanC-ΔpagPGB [one-tailed t test]) (Fig. 4C), and complementation of YeO8-ΔmanC-ΔpagPGB with pTMYePagP restored the susceptibility of the mutant to YeO8-ΔmanC levels (data not shown). When the strains were grown at 37°C, no significant differences were found between the wild type, YeO8-ΔmanC, and YeO8-ΔpagPGB (Fig. 4C). The double mutant was the most susceptible strain. All strains were more susceptible to magainin II when grown at 37°C than at 21°C (Fig. 4C) (P < 0.05 for the comparison of the survival results between temperatures for a given strain [one-tailed t test]).

In summary, these data support that the OPS and lipid A substitutions contribute to Y. enterocolitica temperature-dependent variations in susceptibility to APs.

Aminoarabinose and palmitate lipid A substitutions contribute to Y. enterocolitica virulence.

BALB/c mice were infected orogastrically, and 3 and 7 days postinfection the mice were dissected. The numbers of bacteria present in the Peyer's patches, spleen, and liver were determined by plating. Bacterial load was recorded as log CFU per gram of tissue (Fig. 5). At 3 days postinfection, all strains colonized Peyer's patches, although bacterial loads of the three mutants were lower than that of YeO8 (one-way ANOVA, P < 0.05) (Fig. 5A). A similar picture was obtained for spleen and liver. The lipid A mutants colonized the organs less efficiently than the wild type. Bacterial loads of the three mutants were not significantly different in any tissue (Fig. 5A). At 7 days postinfection, in Peyer's patches, liver, and spleen the bacterial loads of the three lipid A mutants were lower than those of the wild-type strain (Fig. 5B). Bacterial loads of YeO8-ΔpmrF-ΔpagPGB in all tissues were significantly different from those of YeO8-ΔpmrF and YeO8-ΔpagPGB (one-way ANOVA, P < 0.05), which were not significantly different between them (one-way ANOVA, P > 0.05) (Fig. 5B).

Fig 5.

Virulence of Y. enterocolitica lipid A mutants. Bacterial counts in mouse organs at 3 (A) or 7 (B) days postinfection. Mice were infected orogastrically with a bacterial mixture containing 6.0 × 10⁷ bacteria of the wild type (YeO8, □); 2.2 × 10⁷ bacteria of YeO8-ΔpmrF (ΔpmrF, ●); 4.1 × 10⁷ bacteria of YeO8-ΔpagPGB (ΔpagP, ▽); and 3.5 × 10⁷ bacteria of YeO8-ΔpmrF-ΔpagPGB (ΔpmrF-ΔpagP, ▲), respectively. Results were reported as log CFU per gram of tissue (Log CFU/g). *, results are significantly different (one-way ANOVA, P < 0.05) from the results for YeO8.

Effect of temperature on PhoPQ and PmrAB, which regulate lipid A modifications.

We sought to identify the regulatory architecture controlling temperature-dependent lipid A modifications. In S. enterica, the expression of pagP is controlled by PhoPQ whereas the expressions of pmrF operon and ugd are controlled by PmrAB, whose activity can be modulated by the PhoPQ-dependent PmrD connector protein at the posttranscriptional level (25, 27, 41). In silico analysis of YeO8 genome led to the identification of phoPQ and pmrAB homologues, but we did not find a pmrD homologue, which has been recently corroborated (16, 72). Of note, in YeO8 dacB is upstream of pmrAB and the three genes are cotranscribed (data not shown). This genomic organization is different from those found in other enterobacteria, where pmrC is the gene upstream of pmrAB. dacB encodes a d-alanyl-d-alanine carboxypeptidase, also known as pbp4, involved in peptidoglycan biosynthesis.

We sought to determine whether YeO8 PhoPQ and PmrAB may regulate the pmrF operon, ugd, and pagP. To accomplish this, we investigated the expression of pmrH, ugd, and pagP transcriptional fusions in phoPQ and pmrAB mutants. When bacteria were grown at 21°C, the expression of the three fusions was lower in phoPQ and pmrAB backgrounds (Fig. 6A to C). Unexpectedly, the expression of the fusions increased significantly in the double mutant phoPQ pmrAB background, reaching even wild-type levels (Fig. 6A to C). When bacteria were grown at 37°C, the expressions of pagP, pmrH, and ugd fusions in phoPQ, pmrAB, and phoPQ pmrAB backgrounds were similar and not significantly different from those found in YeO8 (P > 0.05 for each comparison versus wild-type levels) (Fig. 6A to C), hence suggesting that PhoPQ and PmrAB regulate the expression of pagP, pmrH, and ugd only at 21°C. Then, we asked whether the expression of phoPQ and pmrAB is temperature regulated. To explore this, two transcriptional fusions in which the promoter regions of phoP and dacB control the expression of lucFF were constructed. These fusions were introduced into YeO8, and luciferase activity was determined. Supporting our hypothesis, the expression of phoP and dacB fusions was significantly higher at 21°C than at 37°C (Fig. 6D and E).

Fig 6.

Y. enterocolitica PhoPQ and PmrAB two-component systems control the expression of pmrF operon and ugd and pagP loci. Analysis of the expression of pmrH, ugd, and pagP loci by YeO8 (WT), YeO8-ΔphoPQ (phoPQ), YeO8-ΔpmrAB (pmrAB), and YeO8-ΔphoPQ-ΔpmrAB (phoPQ-pmrAB) carrying the transcriptional fusions pmrH::lucFF (A), ugd::lucFF (B), and pagP::lucFF (C) grown at 21°C (white bars) or 37°C (black bars). Data are presented as means ± SD (n = 3). *, results are significantly different (P < 0.05; two-tailed t test) from the results for YeO8 grown at the same temperature. Analysis of the expression of phoPQ and pmrAB by measuring luciferase activity of YeO8 carrying dacB::lucFF (D) or phoP::lucFF (E) transcriptional fusions, which were grown at 21°C (white bars) or 37°C (black bars). Data are presented as means ± SD (n = 3). *, results are significantly different (P < 0.05; two-tailed t test) from the results for bacteria grown at 21°C.

RovA and H-NS regulate loci required for lipid A modifications.

RovA is one of the best-characterized transcriptional regulators of yersiniae. It has been proposed that RovA behaves as an anti-H-NS factor to alleviate H-NS-imposed repression (15, 22). Consistent with this model, in YeO8 H-NS represses inv expression at 37°C whereas at 21°C RovA overcomes H-NS-imposed repression, hence allowing inv expression. Of note, the expression of rovA is temperature regulated, being higher at 21°C than at 37°C (66). This evidence prompted us to study whether the temperature-dependent regulation of the loci controlling lipid A modifications could be explained by H-NS-dependent negative regulation alleviated by RovA at 21°C.

First, we studied whether RovA regulates the expression of pagP, pmrH, and ugd transcriptional fusions. Results displayed in Fig. 7 show that the expression of the fusions was significantly lower in the rovA mutant than in YeO8, the difference being nearly 1,000-fold for all fusions when bacteria were grown at 21°C (Fig. 7A to C). Furthermore, the expression of phoP and dacB was also downregulated in the rovA mutant background (Fig. 7D and E). These results were consistent with the absence of lipid A species containing aminoarabinose (m/z, 1,954) and palmitate (m/z, 2,063) in the rovA mutant grown at 21°C (Fig. 7F).

Fig 7.

RovA regulates Y. enterocolitica loci required for lipid A modifications. Analysis of the expression of pmrH, ugd, pagP, phoPQ, and pmrAB loci by YeO8 (WT) and the rovA mutant (ΔrovA) carrying the transcriptional fusions pmrH::lucFF (A), ugd::lucFF (B), pagP::lucFF (C), phoP::lucFF (D), and dacB::lucFF (E) grown at 21°C (white bars) or 37°C (black bars). Data are presented as means ± SD (n = 3). *, results are significantly different (P < 0.05; two-tailed t test) from the results for YeO8 grown at 21°C; Δ, results are significantly different (P < 0.05; two-tailed t test) from the results for YeO8 grown at 37°C. (F) Negative-ion MALDI-TOF mass spectrometry spectra of lipid A isolated from rovA mutant grown at 21°C or 37°C. The results in both spectra are representative of three independent lipid A extractions. (G) Complementation of rovA mutant using pBADYeO8RovA. Arabinose was added (from left to right, 0.00002%, 0.0002%, 0.002%, 0.02%, wt/vol; gray bars) or not (black bars) to the culture medium, and the luciferase activities of pmrH::lucFF, ugd::lucFF, pagP::lucFF, phoP::lucFF, and dacB::luccFF were determined. Bacteria were grown at 21°C. Data are presented as means ± SD (n = 3). *, results are significantly different (P < 0.05; two-tailed t test) from the results for YeO8.

The MIC of polymyxin B for the rovA mutant was significantly lower than that of YeO8 (mean ± SD, 1.8 ± 0.4 and 3.2 ± 0.5, μg/ml, respectively; P < 0.05 [one-tailed t test]) when bacteria were grown at 21°C, whereas at 37°C, the MICs of polymyxin B for the two strains were similar (1.3 ± 0.5 and 1.4 ± 0.4 μg/ml, respectively; P > 0.05 [one-tailed t test]). To explore whether RovA might control other loci necessary for polymyxin B resistance but not implicated in aminoarabinose biosynthesis, we determined the MIC of polymyxin B for the double mutant lacking rovA and pmrF. The MIC (1.7 ± 0.5 μg/ml) was not significantly different from those of rovA and pmrF single mutants (1.8 ± 0.1 and 1.7 ± 0.2 μg/ml, respectively), hence suggesting that RovA does not control additional loci involved in polymyxin B resistance. At 21°C, the MIC of magainin II for the rovA mutant was also lower than that for the wild type (28.4 ± 0.4 and 45.4 ± 5.10, μg/ml, respectively, P < 0.05 [one-tailed t test]), whereas at 37°C the MICs were not significantly different (11.5 ± 0.9 and 10.1 ± 3.0, μg/ml, respectively, P > 0.05 [one-tailed t test]). As in the findings reported for polymyxin B, the MIC of magainin II for the double mutant lacking rovA and pagP (26.8 ± 0.9 μg/ml) was not significantly different from those of the rovA and pagP single mutants (28.4 ± 0.4 and 26.4 ± 2 μg/ml, respectively).

Complementation of the rovA mutant using an arabinose-inducible rovA construct (pBADYeO8RovA) restored the expression of pagP, pmrH, ugd, phoP, and dacB to wild-type levels when the arabinose concentration in the culture medium was 0.002% (wt/vol) (Fig. 7G). Collectively, these findings indicate that RovA is a positive regulator of the loci implicated in lipid A remodeling including the two-component systems PhoPQ and PmrAB.

Second, we sought to determine whether there is cross talk between RovA and the two-component systems PhoPQ and PmrAB. To explore this, the expression of rovA::lucFF was measured in phoPQ, pmrAB, and phoPQ pmrAB backgrounds. Data shown in Fig. 8 revealed that the expression of rovA was downregulated in phoPQ and pmrAB backgrounds although the expression was significantly lower in the phoPQ mutant (120-fold) than in the pmrAB mutant (3-fold; P < 0.05 versus phoPQ values). Interestingly, the expression of rovA in the phoPQ pmrAB double mutant was not significantly different from that of the wild type (Fig. 8A). Therefore, we sought to determine whether this underlies the increased expression of pmrH, ugd, and pagP found in the phoPQ pmrAB mutant background at 21°C. Indeed, the expression of the three loci was downregulated in the rovA phoPQ pmrAB triple-mutant background (Fig. 8B). In the case of pmrH and pagP, the expression levels in the triple-mutant background were significantly lower than those found in the rovA single mutant, whereas the expression levels of the ugd transcriptional fusion in rovA and rovA phoPQ pmrAB were not statistically different (Fig. 8B, insets).

Fig 8.

PhoPQ and PmrAB regulate the expression of rovA in Y. enterocolitica. (A) Analysis of the expression of rovA by YeO8 (WT), YeO8-ΔphoPQ (phoPQ), YeO8-ΔpmrAB (pmrAB), and YeO8-ΔphoPQ-ΔpmrAB (phoPQ pmrAB) carrying the transcriptional fusion rovA::lucFF. Data are presented as means ± SD (n = 3). *, results are significantly different (P < 0.05; two-tailed t test) from the results for YeO8. (B) Analysis of the expression of pmrH, ugd, and pagP loci by YeO8 (WT), YeO8-ΔphoPQ-ΔpmrAB (phoPQ pmrAB), Yvm927-ΔphoPQ-ΔpmrAB (ΔrovA phoPQ pmrAB), and rovA mutant (ΔrovA) carrying the transcriptional fusions pmrH::lucFF, ugd::lucFF, and pagP::lucFF grown at 21°C. Data are presented as means ± SD (n = 3). *, results are significantly different (P < 0.05; two-tailed t test) from the results for YeO8 grown at the same temperature.

Finally, we asked whether H-NS regulates negatively the expressions of pmrH, ugd, pagP, phoP, and dacB by using a surrogate E. coli system because hns is an essential gene in YeO8 (22, 61). This experimental strategy has been used by us and others (22, 61). pPROBE derivatives were transformed into wild-type or Δhns strains, and GFP fluorescence was measured (Fig. 9). All fusions appeared to be negatively regulated by H-NS, since fluorescence was higher in the Δhns strain than in the wild type (Fig. 9). This was true at both growth temperatures, although the relative differences between wild-type and Δhns strains were always higher at 21°C than at 37°C (Fig. 9). For example, pmrH::gfp expression was 350-fold higher in Δhns than in the wild type when strains were grown at 21°C and only 80-fold higher when the bacteria were grown at 37°C.

Fig 9.

H-NS regulates Y. enterocolitica loci required for lipid A modifications. Analysis of the expression of pmrH, ugd, pagP, phoP, and dacB loci by E. coli strains (wild-type [WT], white bars; hns mutant [Δhns], black bars) carrying the transcriptional fusions pmrH::gfp, ugd::gfp, pagP::gfp, phoP::gfp, and dacB::gfp grown at 21°C and 37°C. *, results are significantly different (P < 0.05; two-tailed t test) from the results for the WT grown at the same temperature. Data are presented as means ± SD (n = 3).

To analyze whether RovA relieves the H-NS-mediated negative regulation of the transcriptional fusions, we asked whether the arabinose-inducible rovA construct (pBAD30RovA) increases the fluorescence of the reporter fusions in wild-type E. coli. As we expected, all reporter fusions showed an increased expression upon induction of P_BAD promoter of pBADYeO8RovA plasmid with arabinose (Fig. 10).

Fig 10.

Expression of Y. enterocolitica loci required for lipid A modifications by induction of rovA. Analysis of the expression of pmrH, ugd, pagP, phoP, and dacB loci by wild-type E. coli strain carrying the transcriptional fusions pmrH::gfp, ugd::gfp, pagP::gfp, phoP::gfp, and dacB::gfp, and pBADYeO8RovA. Arabinose was added (from left to right in each panel, 0.00002%, 0.0002%, 0.002%, wt/vol; gray bars) or not (white bars) to the culture medium. Bacteria were grown at 21°C. Data are presented as means ± SD (n = 3). *, results are significantly different (P < 0.05; two-tailed t test) from the results for WT E. coli without arabinose.

DISCUSSION

A hallmark of bacterial pathogens is the adaptation to the host environment by changing gene expression in response to signals encountered within the host such as the temperature shift to 37°C. In the case of the pathogens belonging to the genus Yersinia, most, if not all, virulence factors are temperature regulated. Studies from our laboratory demonstrated that the outer membrane fluidity, the permeability to hydrophobic compounds, and the susceptibility to APs are also temperature-dependent traits in yersiniae (6–9). In our previous studies, we demonstrated that pathogenic Y. enterocolitica strains are more susceptible to polymyxin B, a model AP, when grown at 37°C than at 22 to 25°C (8, 9), and in the present study, we have extended this observation to other APs, not structurally related to polymyxin B. However, the underlying molecular explanation of this temperature-dependent susceptibility to APs was unknown.

The fact that all APs tested interact with the LPS molecule led us to postulate that temperature-regulated LPS structural changes could account for the temperature-dependent variations in AP resistance. The LPS OPS and the modifications of the lipid A with aminoarabinose, palmitate, and phosphoethanolamine have been linked to resistance to APs in Enterobacteriaceae and other bacteria (28, 30, 31, 43, 55). In previous studies, we did show that the expression of Y. enterocolitica OPS is downregulated at 37°C (5, 10), and here, for the first time, we demonstrate that the expression of the loci responsible for the lipid A modifications with aminoarabinose and palmitate is downregulated at 37°C. The survival assays confirmed that the OPS and the lipid A substitutions contribute to Y. enterocolitica AP resistance. It should be noted that these Y. enterocolitica countermeasures are not redundant, since double mutants lacking OPS and lipid A modifications were more susceptible to APs than the single mutants. Interestingly, although the double mutants were the most susceptible strains, still they displayed temperature-dependent variations in the susceptibility to APs, hence suggesting that there are additional temperature-regulated traits, not necessary related to the LPS, underlying Y. enterocolitica temperature-dependent susceptibility to APs. The aminoarabinose content of Y. pestis lipid A is also temperature regulated, being higher at 21°C than at 37°C (1, 64), and in good agreement with our findings, Y. pestis is more susceptible to APs at 37°C than at 21°C (1). The picture with Y. pseudotuberculosis is a bit different. We and others have reported the presence of lipid A species containing aminoarabinose and palmitate at both temperatures, and the susceptibility to APs seems to be temperature independent (50, 64). This emphasizes that caution should be taken when extrapolating findings obtained with one species to others, no matter how closely related they are.

Our analysis of the regulatory network controlling the temperature-dependent modifications of lipid A with aminoarabinose and palmitate revealed that the two-component systems PhoPQ and PmrAB are necessary for the expression of the pmrF operon, ugd, and pagP only at 21°C because at 37°C the expression of the three loci was similar in the wild type and in the phoPQ and pmrAB single mutants. These observations prompted us to evaluate whether the expression of phoPQ and pmrAB could be temperature regulated, as indeed was the case. This is in sharp contrast to the scenario observed in other pathogens such as Salmonella, where PhoPQ and PmrAB systems do regulate the expression at 37°C and it has not been reported that their expression is temperature regulated. Another difference with the Salmonella model is that both YeO8 two-component systems regulate pmrH, ugd, and pagP loci, which is consistent with the presence of PhoP and PmrA boxes in the promoter regions of the three loci (72; see also Fig. S2 in the supplemental material). In Salmonella, under the growth conditions used in the present study, the expression of pagP is only PhoP dependent and the expression of pmrH and ugd is dependent on PmrAB via activation of PhoPQ and PmrD (27, 28, 30, 41). Nevertheless, it is not unprecedented in enterobacteria that both regulatory systems govern the expression of loci implicated in lipid A remodeling. Thus, in Y. pestis and K. pneumoniae both PhoPQ and PmrAB regulate pmrH and ugd (16, 47, 72).

One of the most striking findings of our work is that systems necessary for lipid A remodeling are under the positive control of RovA. This regulator was identified as an activator of YeO8 invasin, but soon it was perceived that RovA could regulate additional genes because the 50% lethal dose (LD₅₀) of the rovA mutant is significantly higher than those of the inv mutant and the wild type, which are not significantly different from each other (20, 66). Indeed, Cathelyn and coworkers demonstrated that 63 genes were differently regulated in the Y. enterocolitica rovA mutant (15). Our results expand this list with the pmrF operon, ugd, pagP, and the two-component regulators phoPQ and pmrAB. Our findings showed that all of them are downregulated in the rovA mutant background. At present, we can only speculate as to why these loci were not selected in the analysis conducted by Cathelyn and coworkers (15). Among other possibilities, the growth temperature is lower in our study than in the other (21°C versus 26°C, respectively), and our data indicated that there is an inverse correlation between the expression of the loci investigated and temperature (unpublished data).

In yersiniae, evidence supports the model for a dynamic regulation of genes by RovA and H-NS. In most cases, H-NS behaves as a transcriptional repressor of RovA-activated loci. Further supporting this model, our data suggest that RovA alleviates H-NS repression on the expression of the pmrF operon, ugd, pagP, and the two-component systems phoPQ and pmrAB. Therefore, the temperature-dependent expression of rovA, higher at 21 than at 37°C, underlies the temperature-dependent regulation of these loci responsible for lipid A remodeling. The presence of phoP and pmrA boxes on the promoter regions of the pmrF operon, ugd, and pagP warrants biochemical analysis to decipher the interplay of RovA, H-NS, PhoP, and PmrA to promote efficient transcription of these loci.

In Y. enterocolitica and Y. pseudotuberculosis it has been demonstrated that H-NS and RovA regulate rovA expression; H-NS represses rovA, and RovA is required for maximal expression (15, 22, 36, 42). Additional regulators of rovA are RovM and LeuO (35, 42). RovM represses rovA transcription most likely by direct binding to the rovA promoter region (35, 42). LeuO is a positive regulator, although it is unclear whether regulation occurs through direct interaction between LeuO and the rovA promoter or if LeuO modulates the expression of another regulator that, in turn, might act directly on the rovA promoter (35, 42). Our findings demonstrate that PhoPQ and PmrAB also regulate positively the expression of rovA, with the effect of PhoPQ being more important. In silico analysis of the rovA promoter region revealed the presence of phoP and pmrA boxes, thereby suggesting a direct interaction of both proteins with the rovA promoter (see Fig. S2 in the supplemental material). Nevertheless, it cannot be ruled out that the expression of other rovA regulators, such as RovM and LeuO, or others yet to be identified, could be affected as well in the phoPQ and pmrAB mutant backgrounds. Interestingly, the expression of rovA reached wild-type levels in the phoPQ pmrAB mutant background, which in turn underlined the increased expression of pmrH, ugd, and pagP found in this mutant background. Studies are ongoing to decipher the molecular basis of rovA upregulation in the phoPQ pmrAB mutant.

RovA is a member of the SlyA/MarR regulatory family, which contains homologues in several species of bacteria. We were intrigued to find out whether any other member of this family of transcriptional regulators also modulates the expression of loci required for the modification of lipid A. Indeed, Salmonella SlyA, the closest homologue to RovA, participates in a positive- feedback loop that facilitates the expression of phoPQ by antagonizing the inhibitory action of H-NS in the phoP promoter region (40, 53, 68). In addition, the PhoPQ system activates slyA transcription (40, 53, 68). Similarly to the rovA mutant, the expression of several loci necessary for LPS remodeling is downregulated in the slyA mutant, and furthermore, the Salmonella slyA mutant is more susceptible to APs than the wild-type strain (45, 53). Therefore, and despite the existing differences between Yersinia and Salmonella, it is tempting to formulate that a major role of SlyA/RovA homologues could be to alter the cell surface to protect the bacterium from toxic compounds produced by the host such as APs. As a consequence, the mutants lacking these regulators should be attenuated, as indeed has been shown for rovA and slyA. Studies in other bacterial models are required to further validate our hypothesis.

The fact that rovA regulates the expression of several genes including those implicated in the resistance to APs makes it almost impossible to attribute the virulence attenuation of this mutant to a certain RovA-regulated locus. Our virulence analysis revealed that the mutants lacking the lipid A modifications with aminoarabinose and palmitate are attenuated. It should be noted that besides inv these are the first RovA-regulated loci that have been shown to play a role in Y. enterocolitica virulence. There are several APs in the tissues infected by Y. enterocolitica, and therefore, the mutants' susceptibility to APs could explain the decreased bacterial loads of these strains in Peyer's patches, spleen, and liver. However, the in vivo scenario is complex and the final outcome of the infection is the combination of the action of antimicrobial factors (among others, complement and APs) and several types of cells, including macrophages and neutrophils, that might be activated by chemokines and cytokines. In turn, all these host defense mechanisms are counteracted by the action of Y. enterocolitica virulence factors, chiefly the pYV-encoded type III secretion system. It might be then also possible that bacterial systems necessary to dismantle host defense mechanisms are not fully functional in the mutants, like the pYV-encoded Yop arsenal of antihost proteins, and hence, the role in virulence of these lipid A modifications would be indirect. Studies are ongoing to test this possibility.