A Novel Serine/Threonine Protein Kinase Homologue of Pseudomonas aeruginosa Is Specifically Inducible within the Host Infection Site and Is Required for Full Virulence in Neutropenic Mice

Abstract

A genetic locus of Pseudomonas aeruginosa was identified that is highly and specifically inducible during infection of neutropenic mice. This locus, ppkA, encodes a protein that is highly homologous to eukaryote-type serine/threonine protein kinases. A ppkA null mutant strain shows reduced virulence in neutropenic mice compared to the wild type. Overexpression of the PpkA protein greatly inhibited the growth of Escherichia coli or P. aeruginosa. However, a single amino acid change at the catalytic site of the kinase domain eliminated the toxic effect of PpkA on bacterial cells, suggesting that the kinase domain of PpkA is functional within bacterial cells.

We have previously reported a method for the isolation of genes induced upon infection of neutropenic mice, using Pseudomonas aeruginosa PAK. After five rounds of selections, 22 different genetic loci were identified through characterization of 45 randomly picked isolates (10). To identify a locus that is the most highly inducible in vivo, two additional rounds of selection were conducted with mice as described earlier (10). A total of 48 colonies were picked and analyzed by Southern hybridization followed by DNA sequencing, as described previously (10). Fusion sites in 29 of them were identical to the np6 locus and the remaining 19 were identical to the np1 locus from our initial selection study (10). We focused our attention on the np1 locus since it appears to encode a previously uncharacterized gene product. The strains and plasmids used in this study are listed in Table 1.

TABLE 1.

Strains and plasmids used

Strain or plasmid	Genotype or description	Source or reference
E. coli DH5α	endA1 hsdR17 supE44 thi-1 recA1 gyrA96 relA1 Δ(lacZYA-argF)U169 λ-φ80 dlacZΔM15; recipient for recombinant plasmids	Bethesda Research Laboratories
P. aeruginosa
PAK	Wild type	David Bradley
PAK-AR2	PAK strain with purEK gene deleted	10
PKN-A	PAK strain with ppkA gene mutated by Ω insertion	This study
Plasmids
pTZ18R	E. coli cloning vector	U.S. Biochemicals
pGEX5x-1	GST fusion vector	Pharmacia Biotech
pUC19Ω	pUC19 plasmid carrying a 2-kb Ω fragment	8
pVK-np1	Cosmid clone containing the ppkA gene in a 15-kb insert	This study
pNP1	np1::purEK fusion rescued from the chromosome of NP1	This study
pPKN-Sα	4-kb SalI fragment containing 5′ two-thirds of ppkA gene in pTZ18R, ppkA in the same direction as lacZα′	This study
pSJ9711	2-kb Ω fragment inserted into the ppkA gene in pPKN-Sα	This study
pHJY9	In-frame fusion of the ppkA gene behind gst of pGEX5x-1	This study
pHJY10	D-to-N change at position 129 of PpkA encoded in pHJY9	This study

The np1 locus is highly and specifically inducible in host tissue.

To confirm the in vivo inducibility of the np1 locus, we compared the in vivo (neutropenic mice) and in vitro (minimal medium A [MinA]) (1a) replication rates between the original isolate, NP1, and the parent purEK deletion strain, PAK-AR2. Since purines are limited under either the in vivo or in vitro assay conditions, the growth rate of the NP1 strain should be proportional to the strength of the np1 promoter, which controls purEK gene expression. Equal numbers of the two bacterial strains were mixed and injected intraperitoneally into six neutropenic mice (2 × 10⁵ cells per mouse). Bacterial cells were recovered from livers of the mice 24 and 48 h after inoculation. The numbers of each bacterium were determined under conditions that allowed the growth of both strains (L agar) or NP1 only (L agar containing 150 μl of carbenicillin per ml). Assuming that the two bacterial strains were cleared equivalently by the host defense system, the ratios of the two bacterial strains in the animal tissue reflect their relative replication rates in vivo, which is indicative of the np1 promoter strength. As an in vitro control, the same bacterial mixture was inoculated into 100 ml of MinA medium, with a final concentration of 10⁵/ml, and incubated at 37°C. As shown in Fig. 1, by 24 and 48 h of infection (in vivo), the NP1 strain had overgrown PAK-AR2 by 84- and >6 × 10⁵-fold, respectively, whereas in vitro the approximate 1:1 ratios were maintained at all times. These data clearly indicate that the np1 locus is specifically and highly expressed in the in vivo environment.

FIG. 1.

Comparison of replication rates between NP1 and PAK-AR2 under in vivo and in vitro growth conditions. Numbers of each bacterial strain recovered from livers of neutropenic mice (A) or from MinA medium (B) after 24 and 48 h are shown. The in vivo data represent an average from three mice at each time point. The in vitro data represent an average of three independent mixed-culture tests.

The np1 locus encodes a putative serine/threonine protein kinase.

A cosmid clone containing the np1 locus was identified from a cosmid clone bank of the PAK chromosomal DNA (5) by colony hybridization, using a partial np1 gene fragment in pNP1 as a probe. DNA fragments surrounding the original purEK fusion site were subcloned and sequenced. An open reading frame (ORF) was identified with the predicted direction of transcription of the np1 gene, encoding a 1,032-amino-acid protein, which bears no obvious signal sequence in its N terminus.

The N-terminal third of the protein is similar to serine/threonine protein kinases found in bacteria and eukaryotes. The greatest similarity observed was to the putative kinases from Myxococcus xanthus (7), Mycobacterium leprae (2), Mycobacterium tuberculosis (1), and Streptomyces coelicolor (9). Multiple sequence alignment of the putative bacterial kinases with their better-studied eukaryotic counterparts revealed pronounced conservation of at least 10 of the known 12 motifs that define the Ser/Thr protein kinase superfamily in eukaryotes (3, 4), including an ATP-binding glycine loop in subdomain I, an invariant lysine residue involved in interaction with α- and β- phosphates in subdomain II, a “kinase loop” motif with an invariant catalytic aspartate in subdomain Vib (amino acid 129), and a threonine residue in motif VIII that is frequently autophosphorylated (Fig. 2). The C-terminal two-thirds of the np1 ORF, rich in proline, shares no similarity to any known sequences. Prediction of globular and nonglobular regions, using local sequence complexity measures (11), detects a nonglobular, elongated protein segment in the area spanning amino acids 280 to 660 and a globular structure in the remaining C-terminal portion of the protein.

FIG. 2.

Alignment of PpkA with related bacterial and eukaryotic Ser/Thr-like protein kinases. Unique identifiers in SWISSPROT or PDB databases are shown. Distances between the ungapped blocks of the highest similarity and the protein termini are indicated by numbers. Invariant residues are shown in boldface. Highly conserved bulky hydrophobic residues (I, L, M, V, F, Y, and W) and small-side-chain residues (A, G, and S) are also highlighted. Functionally important residues in motifs I, II, VIb, and VIII (see the text) are underlined. In the motif line, the conserved motifs in Ser/Thr kinases, as defined in Hanks and Hunter (3, 4), are indicated. In the secondary structure line, α-helices and β-strands in the known three-dimensional structure of twitchin (1KOA) are indicated.

The purEK gene fusion in the original isolate, NP1, occurred at amino acid 561. Further analysis of the other isolate, NP6, indicated that it had a purEK fusion to the same gene, but the fusion site resided in the N-terminal end, at amino acid position 198. These results indicate that we had actually isolated a single locus that is the most highly inducible in vivo, having the purEK gene fused at two different sites. This locus is designated ppkA (Pseudomonas protein kinase).

The ppkA locus is required for full bacterial virulence in neutropenic mice.

To investigate the role of the ppkA gene in bacterial virulence, a ppkA insertional null mutant was generated. An Ω fragment (8), coding for resistance to spectinomycin and streptomycin, was inserted into the 5′ structural ppkA gene on pPKN-Sα, resulting in pSJ9711. The mutant ppkA gene was then introduced into the chromosome of the wild-type PAK strain by electroporation (6). The ppkA null mutant strain, designated PKN-A, was confirmed by Southern hybridization of the chromosomal DNA (Fig. 3).

FIG. 3.

Southern hybridization of chromosomal DNA from strains PAK, PKN-A, and PKN-AC. (Top) Restriction map of the region containing the ppkA gene in PKN-A and PKN-AC. B, BamHI; P, PstI; R, EcoRI. (Bottom) Chromosomal DNA from strains PAK (lanes 1, 3, and 5), PKN-A (lanes 2, 4, and 6), and PKN-AC (lane 7) were digested with EcoRI (lanes 1 and 2), BamHI (lanes 3 and 4), or PstI (lanes 5, 6 and 7). A 4-kb SalI fragment containing the 5′ ppkA gene was used as a probe.

The PKN-A strain did not show any traits distinguishable from those of wild-type PAK when grown on either rich or minimal medium; however, a clear difference in virulence in neutropenic mice was observed. Tests of virulence in neutropenic mice were conducted as described earlier (10), and the number of animal deaths was observed at 6-h intervals for a total of 72 h. As shown in Fig. 4, about 10-fold-more PKN-A cells were needed to cause a similar lethal effect in neutropenic mice compared to the wild-type PAK strain. Furthermore, the ppkA mutant caused on average 8 to 12 h of delay in the times of the animal deaths compared to the wild type.

FIG. 4.

Survival rates of neutropenic mice infected with strain PAK, PKN-A, or PKN-AC. Bacteria were injected intraperitoneally at doses of 10³, 10⁴, and 10⁵ cells, and animal death was observed at 6-h intervals for a total of 72 h. The numbers of dead animals by 72 h over the total numbers of animals tested are shown next to the strains used.

To see whether the delay in the times of animal deaths caused by PKN-A is a direct result of the ppkA mutation or of a polar effect on downstream genes, the pPKN-Sα plasmid, containing a 3′-end-truncated version of the ppkA gene, was electroporated into PKN-A cells and a single crossover through the left arm (5′ to the “Ω” insertion site of the ppkA gene in PKN-A) was selected for, as depicted in Fig. 3. The resulting strain, PKN-AC, has a single-copy, stably maintained ppkA gene (Fig. 3 and data not shown). As shown by the virulence test results in Fig. 4, PKN-AC caused an animal death rate similar to that caused by the wild-type PAK strain. These results indicated that the N-terminal two-thirds of the PpkA protein is sufficient to complement the PKN-A mutant and that the ppkA gene is solely responsible for the reduced bacterial virulence of the mutant strain PKN-A. Furthermore, searching the database of the unfinished contigs from the Pseudomonas genome projects (accessible at http://www.ncbi.nlm.nih.gov/BLAST/pseudoabl.html) revealed that, in addition to PpkA itself, there are at least three other PpkA-related sequences, which may account for the moderate reduction of virulence of PKN-A.

The kinase domain of PpkA affects growth of E. coli and P. aeruginosa.

To confirm the size of the PpkA ORF as well as to study the biochemical properties of PpkA, we attempted to overproduce the PpkA protein in E. coli. The first ATG codon of PpkA was fused in frame behind the glutathione S-transferase (GST) gene in pGEX5x-1, resulting in pHJY9. This construct was sequenced to confirm the in-frame gene fusion. E. coli harboring the fusion construct, pHJY9, grows slowly and forms small colonies on media even in the absence of isopropyl-β-d-thiogalactopyranoside (IPTG), compared to E. coli harboring the fusion vector only. In the presence of IPTG (>0.1 mM), E. coli containing pHJY9 hardly grows on minimal or rich medium and prolonged incubation in liquid medium leads to bacterial lysis, indicating that the PpkA portion of the fusion protein is toxic to E. coli.

We next asked if the kinase activity of PpkA plays any role in toxicity. Since the conserved aspartic acid residue (amino acid 129), residing within the catalytic loop of the kinase domain, is required for the catalytic activity of the enzyme in all characterized Ser/Thr protein kinases (3, 4), it was mutated to an asparagine by site-directed mutagenesis. The mutant gene, ppkA(D129N), was then fused behind the gst gene as in pHJY9, resulting in pHJY10. In contrast to E. coli harboring pHJY9, E. coli harboring pHJY10 grows normally on media in the presence or absence of IPTG. Furthermore, a 150-kDa GST-PpkA(D129N) fusion protein was highly and specifically produced in the presence of IPTG (Fig. 5), demonstrating that the toxic effect of PpkA is due to its kinase domain. Excluding the 25-kDa GST portion, the size of the PpkA protein in the above fusion construct is in good agreement with the molecular mass predicted from the DNA sequence. By using antibody against GST, an IPTG induction-specific 150-kDa GST-PpkA fusion protein was also detected in E. coli containing pHJY9 (data not shown), including a series of smaller bands, mainly the visible 40-kDa protein band (Fig. 4, lanes 3 and 4), representing breakdown products.

FIG. 5.

Coomassie blue-stained sodium dodecyl sulfate-polyacrylamide gel of total bacterial cell extracts. E. coli DH5α harboring GST fusion vector pGEX5x-1 (lanes 1 and 2), GST-PpkA fusion construct pHJY9 (lanes 3 and 4), or GST-PpkA(D129N) fusion construct pHJY10 (lanes 5 and 6) was grown in the presence (lanes 2, 4, and 6) or absence (lanes 1, 3, and 5) of 0.1 mM IPTG. Cell extracts from the same number of bacterial cells were loaded in each lane.

Although the tac promoter that drives the expression of the gst gene is not as strong a promoter in P. aeruginosa as it is in E. coli, the same toxic effect of GST-PpkA on P. aeruginosa was observed when pHJY9 was introduced into the chromosome of PAK by single crossover (PAK×pHJY9), whereas pHJY10 had no toxic effect. Taken together, the above observations clearly indicate that the kinase domain of PpkA has an enzymatic function within bacterial cells and high-level substrate phosphorylation might have led to the inhibitory effect on bacterial growth.

Nucleotide sequence accession number.

The nucleotide sequence of ppkA has been submitted to the GenBank databases under accession number AF035395.