Transcriptional Regulation of the 4-Amino-4-deoxy-L-arabinose Biosynthetic Genes in Yersinia pestis*

Inducible membrane remodeling is an adaptive mechanism that enables Gram-negative bacteria to resist killing by cationic antimicrobial peptides and to avoid eliciting an immune response. Addition of 4-amino-4-deoxy-l -arabinose (4-aminoarabinose) moieties to the phosphate residues of the lipid A portion of the lipopolysaccharide decreases the net negative charge of the bacterial membrane resulting in protection from the cationic antimicrobial peptide polymyxin B. In Salmonella enterica serovar Typhimurium, the PmrA/PmrB two-component regulatory system governs resistance to polymyxin B by controlling transcription of the 4-aminoarabinose biosynthetic genes. Transcription of PmrA-activated genes is induced by Fe3+, which is sensed by PmrA cognate sensor PmrB, and by low Mg2+, in a mechanism that requires not only the PmrA and PmrB proteins but also the Mg2+-responding PhoP/PhoQ system and the PhoP-activated PmrD protein, a post-translational activator of the PmrA protein. Surprisingly, Yersinia pestis can promote PhoP-dependent modification of its lipid A with 4-aminoarabinose despite lacking a PmrD protein. Here we report that Yersinia uses different promoters to transcribe the 4-aminoarabinose biosynthetic genes pbgP and ugd depending on the inducing signal. This is accomplished by the presence of distinct binding sites for the PmrA and PhoP proteins in the promoters of the pbgP and ugd genes. Our results demonstrate that closely related bacterial species may use disparate regulatory pathways to control genes encoding conserved proteins.

Inducible membrane remodeling is an adaptive mechanism that enables Gram-negative bacteria to resist killing by cationic antimicrobial peptides and to avoid eliciting an immune response. Addition of 4-amino-4deoxy-L-arabinose (4-aminoarabinose) moieties to the phosphate residues of the lipid A portion of the lipopolysaccharide decreases the net negative charge of the bacterial membrane resulting in protection from the cationic antimicrobial peptide polymyxin B. In Salmonella enterica serovar Typhimurium, the PmrA/PmrB two-component regulatory system governs resistance to polymyxin B by controlling transcription of the 4-aminoarabinose biosynthetic genes. Transcription of PmrAactivated genes is induced by Fe 3؉ , which is sensed by PmrA cognate sensor PmrB, and by low Mg 2؉ , in a mechanism that requires not only the PmrA and PmrB proteins but also the Mg 2؉ -responding PhoP/PhoQ system and the PhoP-activated PmrD protein, a post-translational activator of the PmrA protein. Surprisingly, Yersinia pestis can promote PhoP-dependent modification of its lipid A with 4-aminoarabinose despite lacking a PmrD protein. Here we report that Yersinia uses different promoters to transcribe the 4-aminoarabinose biosynthetic genes pbgP and ugd depending on the inducing signal. This is accomplished by the presence of distinct binding sites for the PmrA and PhoP proteins in the promoters of the pbgP and ugd genes. Our results demonstrate that closely related bacterial species may use disparate regulatory pathways to control genes encoding conserved proteins.
The lipopolysaccharide (LPS) 1 is a major component of the outer membrane of Gram-negative bacteria (1,2). It consists of a poly-or oligosaccharide region that is linked to a glycolipid molecule termed lipid A (3). The lipid A is a major determinant of the interaction between Gram-negative bacteria and the host innate immune system because: 1) it is the molecule responsible for activation of host cell receptors such as TLR4 (4,5); and 2) the phosphate residues in lipid A provide the LPS with a negative charge that constitutes the target of cationic antimicrobial proteins and peptides (6,7). Certain Gram-negative species have the ability to modify the lipid A, which enables them to resist killing by cationic antimicrobial compounds and to avoid eliciting an immune response (8 -14).
The covalent modification of the phosphate residues in lipid A with 4-amino-4-deoxy-L-arabinose (4-aminoarabinose) confers resistance to the cationic peptide antibiotic polymyxin B (14 -21). The enzymes mediating the biosynthesis of 4-aminoarabinose are encoded by the pbgP operon (also referred to as pmrHFIJKLM, Ref. 19 and arn, Ref. 22) and the ugd gene (19,23,24). In Salmonella enterica serovar Typhimurium, expression of these genes is controlled by the PmrA/PmrB two-component system (25). Transcription of PmrA-activated genes is induced by Fe 3ϩ , which is the signal sensed by the PmrA cognate sensor PmrB (26), and by low Mg 2ϩ , in a mechanism that requires not only the PmrA and PmrB proteins but also the Mg 2ϩ -responsive PhoP/PhoQ system and the PhoP-activated PmrD protein, a post-translational activator of the PmrA protein (27,28) (Fig. 1A).
The PmrA protein binds to the pbgP and ugd promoters at regions that include the hexanucleotide repeat (C/T)TTAAT separated by 5 bp, which has been termed the PmrA box (29 -32). The PmrA protein promotes transcription of the pbgP and ugd genes using the same start sites whether the inducing condition is Fe 3ϩ , which is a PhoP/PhoQ-and PmrD-independent process, or low Mg 2ϩ , which is dependent on both PhoP/ PhoQ and PmrD (28). Thus, wild-type Salmonella expresses PmrA-regulated genes and is resistant to polymyxin B following growth in the presence of Fe 3ϩ (26) and/or low Mg 2ϩ (27), whereas pmrD and phoP mutants are resistant to polymyxin B if grown in the presence of Fe 3ϩ but sensitive if the inducing condition is low Mg 2ϩ (26,27).
Salmonella modifies its lipid A with 4-aminoarabinose during infection of murine macrophages (33). This modification appears to be PhoP/PhoQ-dependent because it was detected only under PhoP-inducing conditions when bacteria were grown in defined media (33), and because expression of the ugd gene inside macrophages required a functional PhoP/PhoQ system even though ugd transcription can be promoted by other two-component systems independently of PhoP/PhoQ (34). Escherichia coli cannot modify its lipid A with 4-aminoarabinose in response to the low Mg 2ϩ signal that induces the PhoP/PhoQ system (33,35) because its highly divergent PmrD protein fails to activate the PmrA protein, which prevents E. coli from expressing PmrA-dependent genes in low Mg 2ϩ (36).
The plague agent Yersinia pestis can promote PhoP-dependent modification of its LPS with 4-aminoarabinose (14). This is surprising because Yersinia lacks a PmrD protein. Moreover, it indicates that Salmonella and Yersinia must use different strategies to promote expression of lipid A modifying genes under PhoP-inducing conditions. Here we report the mechanism by which Y. pestis regulates expression of the 4-amino-arabinose biosynthetic genes mediating resistance to polymyxin B. Our results demonstrate that closely related bacterial species adopt distinct regulatory strategies for expression of conserved genes encoding structural proteins, and highlight the difficulty of deducing bacterial behavior solely on the basis of gene content.
Construction of Strains with Defined Deletions-All mutant strains were constructed according to the one-step disruption protocol (41) using the following primer pairs: 3387 (5Ј-GCCGCCTCTATAAAGATA-TTTATCAGAATAACTGAGAGCTCATATGAATATCCTCCTTAG-3Ј) and 3388 (5Ј-ATCAGCGCCAGCATTAACAGCAAACGACGGCGCATG-CTGAGTGTAGGCTGGAGCTGCTTC-3Ј) for pmrA; 3385 (5Ј-TATCCT-GTTATCCGGTTAACGTTTTATCAAGGATTGGTGTCATATGAATAT-CCTCCTTAG-3Ј) and 3386 (5Ј-CGAAGGGAGAAGGGTTTATTATTTT-TTCTGAGCATAGTGTGTGTAGGCTGGAGCTGCTTC-3Ј) for phoP; 4340 (5Ј-ATAGTTAATAGTCCATGAAGGTGTCCTAAGGGATTTATT-AGTGTAGGCTGGAGCTGCTTC-3Ј) and 4341 (5Ј-GTTATAGACAGGA-ATAACAATGGAGACCTTCTTAATTGGTCATATGAATATCCTCCTT-AG-3Ј) for pbgP. The Y. pestis phoP pmrA double mutant was constructed using the one-step method to delete the pmrA gene from the phoP mutant strain EG14738. The chromosomal structure of the mutated loci was verified both by PCR as described (41) and by Southern hybridization using probes specific to: (i) the antibiotic resistance genes used during the construction of the chromosomal deletions, and (ii) sequences flanking the inactivated loci (data not shown). S1 Nuclease Assay-To prepare RNA, overnight cultures grown in the defined medium described above containing 10 mM Mg 2ϩ , washed and diluted 1:50 into 50 ml of the defined medium containing either 10 M MgSO 4 , 10 mM MgSO 4 , or 10 M MgSO 4 ϩ 100 M FeSO 4 . Because the ⌬phoP mutant was unable to grow in defined medium containing 10 M MgSO 4 , strains were not washed prior to subculture into defined medium containing 10 M MgSO 4 , which allowed growth. Total RNA was extracted from early-logarithmic phase cultures (A 600 , 0.250) with the MasterPure RNA purification kit (Epicenter Technologies) according to the manufacturer's protocol.
Primers 1429 (5Ј-CCACAAGAAGCGTATAACGC-3Ј) and 5298 (5Ј-GTAGCTTGAACAAGTCCCAC-3Ј) were labeled at the 5Ј-end by phosphorylation with [ 32 P]ATP (Amersham Biosciences) using T4 polynucleotide kinase (Invitrogen). Double-stranded DNA probes to the pbgP and ugd promoter regions were generated by PCR using primer pairs 1428 (5Ј-CTTCACTACCTATTGCTGGC-3Ј), 1429 and 5297 (5Ј-GCT-GATGCTTGCTGCTGAAG-3Ј), 5298, respectively. S1 nuclease reactions were performed as described (42). In brief, total RNA (30 g for pbgP and 100 g for ugd) and the labeled DNA probe were combined with 50 l of hybridization buffer (80% formamide, 20 mM HEPES pH 6.5, 0.4 M NaCl). The mixture was incubated at 95°C for 5 min and then left to cool down in an incubator at 30°C overnight. Hybridization reactions were treated with 10 units of S1 nuclease (Promega) for 30 min at 37°C. The reaction was stopped by the addition of 200 l of phenol-chloroform, and the aqueous phase was precipitated with ethanol. The precipitate was dissolved in sequence loading buffer and electrophoresed on a 6% acrylamide, 7.5 M urea gel together with a Maxam-Gilbert DNA ladder generated from the pbgP and ugd promoter DNA probes. Assays were performed in duplicate.
DNase I Footprinting-DNase I footprinting was performed as reported (31). Briefly, the pbgP promoter region was amplified as described above with labeling of primer 1429 for the coding strand and primer 1428 for the non-coding strand. The ugd promoter region was amplified using two different primer pairs, 3383 (5Ј-CACCTTGATGGA-CAGTTTCC-3Ј), 3384 (5Ј-TTCATACCAGACTTACTCCC-3Ј) for footprinting with the PhoP protein and primers 5297, 5298 for footprinting with the PmrA protein. Primers 3383 and 5297 were labeled for the coding strand and primers 3384 and 5298 were labeled for the noncoding strand. The Salmonella PhoP and PmrA proteins were purified as described (29,43). Binding reactions with the PhoP and PmrA proteins were carried out as follows. Proteins were incubated with 25 fmol of DNA probe in 100 l of 2 mM Hepes (pH 7.9), 10 mM KCl, 20 M EDTA, 500 g of bovine serum albumin, 20 g/ml poly(dI-dC), and 2% glycerol for 20 min at room temperature. DNase I (Invitrogen) (0.01 units), 100 M CaCl 2 , and 100 M MgCl 2 were added and incubated for 3 min at room temperature. The reactions were stopped by the addition of phenol-chloroform, and the aqueous phase was precipitated. Samples were analyzed by electrophoresis on a 6% polyacrylamide, 7.5 M urea gel and compared with a Maxam-Gilbert AϩG DNA ladder generated from the same DNA probe.
Polymyxin B Susceptibility Assay-Strains were grown to logarithmic phase in the defined medium described above containing either 10 M MgSO 4 , 10 mM MgSO 4 , or 10 M MgSO 4 ϩ 100 M FeSO 4 , washed and incubated in the presence of 5.0 g/ml polymyxin B at 28°C for 1 h. Samples were serially diluted in phosphate-buffered saline, plated on BHI and incubated for 36 -48 h at 28°C for viability counts. Survival values were calculated by dividing the number of bacteria following treatment with polymyxin B relative to those incubated in the presence of PBS and then multiplied by 100. Assays were performed in triplicate.
Identification of Putative Transcription Factor Binding Sites-The promoter regions of the pbgP and ugd genes were examined manually, as well as using the GPS program (soar-tools.wustl.edu/) to identify putative binding sites for the PhoP (44 -46) and PmrA (29 -32) proteins.  (48), can modify its lipid A with 4-aminoarabinose in a PhoPdependent manner (14), and encodes a conserved PmrA/PmrB system (37, 47) (www.ncbi.nlm.nih.gov), we reasoned that the low Mg 2ϩ induction of the 4-aminoarabinose biosynthetic genes taking place in Yersinia must involve a mechanism different from the PmrD-dependent pathway described in Salmonella (27,28) (Fig. 1A).

Transcription of the Yersinia pbgP Gene Is Mediated by
To determine the transcription start site of the pbgP operon in Y. pestis, we conducted S1 mapping experiments using RNA harvested from organisms grown under conditions known to modulate pbgP transcription in Salmonella (25)(26)(27). Because the Yersinia strains were unable to grow in the N-minimal media (49) used to analyze pbgP transcription in Salmonella, we measured transcription of pbgP in Yersinia grown in a Yersinia-specific defined media (40), which differs from N-minimal media in its salt composition and carbon source. We identified two transcription start sites: an ORF-proximal site that was stronger in organisms experiencing low Mg 2ϩ than in those exposed to low Mg 2ϩ ϩ Fe 3ϩ , and an ORF-distal site that displayed the opposite behavior ( Fig. 2A). There was weak expression from both promoters in wild-type cells following growth in high Mg 2ϩ (Fig. 2A), a condition that represses pbgP transcription in Salmonella (27,50).
Deletion of the pmrA gene eliminated transcription from the distal but not from the proximal promoter ( Fig. 2A). In contrast, inactivation of the phoP gene abolished transcription from the proximal but not from the distal promoter ( Fig. 2A). There was no transcription from either promoter in a phoP pmrA double mutant, regardless of the growth condition ( Fig.  2A). These results demonstrate that Y. pestis harbors a Fe 3ϩresponding PmrA/PmrB system, which is consistent with the presence of the Fe 3ϩ -binding motif in the putative periplasmic region of the PmrB protein (26). Moreover, they indicate that, unlike what happens in Salmonella, the PmrA and PhoP pro-teins use different promoters to transcribe the pbgP gene (Fig.  2B). Interestingly, pbgP transcription from the pmrA-dependent promoter was still present in a phoP mutant experiencing low Mg 2ϩ ( Fig. 2A), indicative that the PmrA protein is activated in this media, albeit at lower levels.
The PhoP and PmrA Proteins Bind to the pbgP Promoter-Analysis of the Yersinia pbgP promoter region revealed the  (Fig. 2B), suggesting that the PhoP and PmrA proteins regulate pbgP transcription by binding to its promoter region. To test this hypothesis, we conducted DNase I footprinting analysis of the pbgP promoter using purified PhoP and PmrA proteins from Salmonella, which are 78.9 and 56.1% identical to the Yersinia PhoP and PmrA proteins, respectively. We established the following. (i) The PhoP protein protected nucleotides Ϫ17 to Ϫ46 and Ϫ28 to Ϫ53 from the PhoP-dependent start site on the coding and non-coding strands, respectively (Fig. 2C), which include the predicted PhoP box (Fig. 2B). (ii) The PmrA protein protected nucleotides Ϫ10 to Ϫ47 and Ϫ28 to Ϫ54 from the PmrA-dependent start site on the coding and non-coding strands, respectively (Fig.  2D), which include the predicted PmrA box (Fig. 2B). The regions protected by the PhoP and PmrA proteins are followed by distinct Ϫ10 regions (Fig. 2B), consistent with separate transcriptional control of the pbgP operon mediated by these two proteins.
Transcription of the Yersinia ugd Gene Is Mediated by Distinct PhoP-and PmrA-dependent Promoters-The ugd gene product mediates an earlier step in the biosynthesis of 4-aminoarabinose than those catalyzed by the enzymes encoded in the pbgP operon (51). Thus, we reasoned that, similar to Salmonella (25)(26)(27), Yersinia was likely to coordinate transcrip-tion of the ugd gene with that of the pbgP operon. Therefore, we conducted S1 mapping experiments to identify the ugd transcription start site(s) in Yersinia grown under different conditions. We identified two distinct transcription start sites: an ORF-distal site that was induced to maximal levels during growth in low Mg 2ϩ and required a functional phoP gene (Fig. 3A), and an ORF proximal site that was maximally induced in response to low Mg 2ϩ ϩ Fe 3ϩ and required a functional pmrA gene (Fig. 3A). No ugd transcription was detected in the phoP pmrA double mutant (Fig. 3A). Weak expression was detected from both start sites in wild-type cells grown in high Mg 2ϩ (Fig. 3A). These data indicate that the PhoP and PmrA proteins use distinct promoters to transcribe the Yersinia ugd gene. Moreover, they mimic the results described above for the pbgP promoter, indicative that Yersinia coordinates the expression of the pbgP operon and ugd gene.
The PhoP and PmrA Proteins Bind to the ugd Promoter-Consistent with the presence of distinct PhoP-and PmrA-dependent transcription start sites for the ugd gene (Fig. 3A), we identified PhoP and PmrA boxes in the Yersinia ugd promoter (Fig. 3B). DNase I footprinting carried out with the Salmonella PhoP and PmrA proteins demonstrated that the PhoP protein protected nucleotides Ϫ4 to Ϫ41 and Ϫ16 to Ϫ46 relative to the PhoP-dependent start site on the coding and non-coding strands, respectively (Fig. 3C), and that the PmrA protein protected nucleotides Ϫ16 to Ϫ34 and Ϫ14 to Ϫ54 relative to the PmrA-dependent start site on the coding and non-coding strands, respectively (Fig. 3D). These sequences include the predicted PhoP and PmrA boxes, respectively, which are followed by distinct putative Ϫ10 regions (Fig. 3B). In sum, these results indicate that the PhoP and PmrA proteins bind to the ugd promoter to control its transcription.
A phoP pmrA Double Mutant Exhibits Hypersusceptibility to Polymyxin B-Inactivation of the pmrA gene renders Salmonella extremely sensitive to killing by polymyxin B following growth in either low Mg 2ϩ or low Mg 2ϩ ϩ Fe 3ϩ (26,27). This is in contrast to inactivation of the phoP gene, which makes Salmonella sensitive to polymyxin B when grown in low Mg 2ϩ but not following growth in low Mg 2ϩ ϩ Fe 3ϩ (26,27). These phenotypes reflect transcription of the pbgP and ugd loci in Salmonella, which is entirely dependent on PmrA under both inducing conditions, but dependent on PhoP only during growth in low Mg 2ϩ (25)(26)(27).
Because Yersinia strains lacking either the phoP or pmrA gene can still transcribe pbgP and ugd when grown in either low Mg 2ϩ or low Mg 2ϩ ϩ Fe 3ϩ ( Figs. 2A and 3A), we reasoned that phoP and pmrA single mutants would be resistant to polymyxin B under either inducing condition. As predicted, these mutants displayed wild-type resistance to polymyxin B following growth in low Mg 2ϩ ϩ Fe 3ϩ (Fig. 4A). When the inducing condition was low Mg 2ϩ , the pmrA mutant displayed wild-type levels of resistance, whereas the phoP mutant was ten times more sensitive than the wild-type strain (Fig. 4B). These results suggest that the levels of transcription of the 4-aminarabinose biosynthetic genes originating from the PmrAdependent promoter are not sufficient to confer full polymyxin B resistance to the phoP mutant when the inducing condition is low Mg 2ϩ . On the other hand, a pmrA phoP double mutant was as susceptible to polymyxin B as the pbgP mutant under all tested conditions (Fig. 4).

Conservation of Direct Regulation of the Lipid A-modifying Genes by the PhoP and PmrA Proteins in the Genus
Yersinia-The genus Yersinia comprises three species that are pathogenic to man: the blood-borne Y. pestis, and the enteropathogenic Y. pseudotuberculosis and Y. enterocolitica (52). Analysis of the pbgP promoter region of Y. pestis strain CO92 (47), Y. pestis biovar Medievalis str. 91001 (www. ncbi.nlm.nih.gov), Y. pseudotuberculosis and Y. enterocolitica (www.sanger.ac.uk) revealed the presence of both PhoP and PmrA boxes followed by distinct putative Ϫ10 regions (Fig.  5). PhoP and PmrA boxes with appropriately spaced Ϫ10 regions were also identified in the ugd promoter of all examined genomes (Fig. 5) except that of Y. enterocolitica, the sequencing of which has not been completed. (ORFs with sequence identity to the PmrD protein were not present in any of the Yersinia genomes.) This suggests that these organisms may use the same regulatory mechanism described above for Y. pestis strain KIM6 to control expression of the 4-aminoarabinose biosynthetic genes. DISCUSSION We have determined the mechanism responsible for transcription of the 4-aminoarabinose biosynthetic genes pbgP and ugd in Y. pestis, establishing a direct activation by both the PmrA and PhoP proteins (Fig. 1B). This allows Yersinia to express these lipid A-modifying determinants in response to two signals: Fe 3ϩ , which is likely sensed by the PmrB protein as in Salmonella (26), and low Mg 2ϩ , possibly sensed by the PhoQ protein. This model is supported by the following data: (i) transcription of the pbgP and ugd genes is initiated from two different promoters depending on the inducing condition: a PhoP-dependent promoter that is maximally induced during growth in low Mg 2ϩ , and a PmrA-dependent promoter that is maximally induced in the presence of Fe 3ϩ ( Figs Yersinia and Salmonella Utilize Different Regulatory Strategies to Control the Genes for 4-Aminoarabinose Biosynthesis-Salmonella relies on the shunt protein PmrD to convey the low Mg 2ϩ signal that activates the PhoP/PhoQ system to the PmrA/ PmrB system (27,28) (Fig. 1A), whereas Yersinia PhoP protein activates the pbgP operon and ugd gene directly, using a different promoter from that utilized by the PmrA protein in response to Fe 3ϩ (Fig. 1B). The Salmonella PmrD protein promotes the phosphorylated state of PmrA (28), which binds to its target promoters with higher affinity than unphosphorylated PmrA (29,53) and results in global activation of the whole PmrA regulon (27). On the other hand, the Yersinia strategy demands that binding sites for the PhoP protein be present in the regulatory region of PmrA-regulated genes for expression to occur in low Mg 2ϩ . This raises the possibility that other genes regulated by the PmrA protein in Salmonella might harbor binding sites for both the PmrA and PhoP proteins in Yersinia, which would allow for expression in response to both low Mg 2ϩ and Fe 3ϩ signals. These results demonstrate that closely related bacterial species may respond to the same signals by using different regulatory mechanisms to control expression of conserved structural proteins.

Role of the PmrA/PmrB and PhoP/PhoQ Two Component Systems in the Yersinia Life Style-Our results indicate that
Yersinia harbors an Fe 3ϩ -responding PmrA/PmrB system because the PmrB protein has a Fe 3ϩ -binding motif in its predicted periplasmic region (26) and the PmrA protein directly promotes expression of the pbgP operon and ugd gene when the organism experiences Fe 3ϩ (Fig. 2 and 3). The conservation of both the PmrB protein amino acid sequence and the binding sites for the PmrA protein in the promoter regions of the pbgP and ugd loci across the genus Yersinia (Fig. 5) suggests all members of this genus experience environments denoted by Fe 3ϩ .
The Mg 2ϩ -responding PhoP protein was previously implicated in the regulation of modifications to the core polysaccharide (54) and lipid A (14) portions of the LPS of Y. pestis, which are important for reduction of the endotoxic properties of the LPS and for resistance to antimicrobial peptides. The conserved ability of Salmonella and Yersinia to promote expression of the 4-aminoarabinose biosynthetic genes in low Mg 2ϩ suggests that these two pathogens may encounter similar low Mg 2ϩ environments. One such environment could be the mammalian macrophage, where the Salmonella PhoP-activated genes are highly induced (12) and where 4-aminoarabinose modification of its lipid A takes place (33). Furthermore, the Mg 2ϩ -responding PhoP protein is required for replication of Y. pestis (48) and Y. pseudotuberculosis (55) inside macrophages and for virulence in mice, as reported for Salmonella (56,57), suggesting that PhoP-dependent lipid A modification may be a common strategy used by these pathogens in their interactions with mammalian hosts. Alternatively or in addition, these modifications may be carried out by Y. pestis when in non-mammalian host environments such as the flea (14), or in soil.
Members of the genus Yersinia demonstrate resistance to antimicrobial peptides (14,54,58). We have shown that deletion of both the phoP and pmrA genes was necessary to mimic  (29 -32), the sequence in red indicates the putative PhoP boxes, (44 -46), and the putative Ϫ10 regions are underlined in black. The first four amino acids of the pbgP or ugd ORF are indicated below the nucleotide sequence. the hypersensitivity of a strain lacking the pbgP gene when Yersinia experiences low Mg 2ϩ ϩ Fe 3ϩ (Fig. 4A). This likely reflects the fact that sufficient expression of the 4-aminoarabinose biosynthetic genes is attained by pmrA and phoP single mutants from the PhoP-and PmrA-dependent promoters, respectively, for wild-type levels of resistance to polymyxin B. However, we suspect that there are natural environments where only PhoP or PmrA is active, resulting in LPS modification and resistance to antimicrobial peptides mediated by the either PhoP-or PmrA-dependent control of the 4-aminoarabinose biosynthetic genes. Similar to what has been reported previously (14,54), we observed that the phoP mutant was more sensitive to polymyxin B than the wild-type strain when the inducing condition was low Mg 2ϩ (Fig. 4B). Yet, the susceptibility levels were not the same, possibly reflecting differences in strain backgrounds, phoP alleles, growth media, and assay conditions.
The modification of lipid A with 4-aminoarabinose taking place in low Mg 2ϩ may also help Yersinia and Salmonella cope with low Mg 2ϩ stress, possibly by making the Mg 2ϩ present in the LPS available to other compartments in the bacterial cell (12,18). Indeed, even organisms that lack the PmrA/PmrB system, such as the insect pathogen Photorhabdus luminescens, use the low Mg 2ϩ -inducible PhoP protein to control transcription of the pbgP gene directly (20).
The Need for Biochemical Verification of Bioinformatic Predictions-The acquisition and loss of genes is thought to be the major force driving the evolution of bacterial species (59). Therefore, attempts to determine the genetic basis for the diverse ecology of closely related organisms have focused on the identification of species-specific genes. This pursuit, which is greatly facilitated by the completion of increasing numbers of bacterial genome sequencing projects, has resulted in the identification of genes that are critical for the lifestyles of several organisms (60 -63). This premise would have led to the prediction that Yersinia should not be able to promote transcription of the pbgP and ugd genes in low Mg 2ϩ environments because it lacks the pmrD gene, which is essential for this property in Salmonella (27,28). Likewise, techniques that rely on the conservation of cis regulatory motifs across species, such as phylogenetic footprinting (64), would not have predicted that the PhoP protein controlled pbgP transcription given the absence of PhoP boxes from the pbgP promoter of several species, such as E. coli and Salmonella. Ultimately, our work emphasizes the importance of biochemical and molecular verification of regulatory pathways as changes in cis and/or trans regulatory sequences may have profound effects on the behavior of a bacterial species.