Yersinia pestis pH 6 antigen forms fimbriae and is induced by intracellular association with macrophages

Ability to express pH 6 antigen (Ag) is necessary for full virulence of Yersinia pestis; however, the function of the Ag in pathogenesis remains unclear. We determined the nucleotide sequence of a 4232 bp region of Y. pestis DNA which encoded the pH 6 Ag structural gene (psaA) and accessory loci necessary for Ag synthesis. Protein sequences encoded by the Y. pestis DNA were similar to accessory proteins which function in the biosynthesis of Escherichia coli fimbriae Pap, K88, K99 and CS3 as well as the molecular chaperone for the Y. pestis capsule protein. Electron microscopy and immunogold labelling studies revealed that pH 6 Ag expressing E coU or Yersinia produced flexible‘fibrillar’organelles composed of individual linear strands, multiple strand bundles or wiry aggregates of PsaA. Y. pestis associated with the murine macrophage‐like cell line, RAW264.7, expressed pH 6 Ag in an intracellular acidification‐dependent manner. Together with an earlier study showing that a Y. pestis psaA mutant was reduced in virulence, these results demonstrate that the expression of fimbriae which are induced in host macrophages is involved in plague pathogenesis.


Introduction
Yersinia pestis is a facultative intracellular parasite which can survive and multiply inside macrophages (Straley andHarmon, 1984a: 1984b;Charnetzky and Shuford, 1985). The interaction of /. pestis with host macrophages has long been thought to be important in the pathogenesis of bubonic plague (Cavanaugh and Randall. 1959). although little is known about this interaction. Many of the known virulence determinants of Y. pestis are induced when the organism is cultivated under conditions that simulate the environment encountered by fhe organism inside the mammalian host (Straley and Brubaker. 1981). However, only two of the Yersinia outer membrane proteins (Yops) have been reported to be expressed by bacteria associated with macrophages (Pollack et ai. 1986;Straley, 1991).
Approximately 30 years ago, Ben-Efraim et ai (1961) described an antigen (Ag) that was produced only when y, pestis was cultured at temperatures above 36''C and pH values below 6.7. This new Ag was designated pH 6 Ag, More recently, we cloned a Y. pestis Ag that was similarly regulated (Lindler ef ai. 1990). Although antisera from the previous study (Ben-Efraim etai.. 1961) were not available, we also designated the cloned protein as pH 6 Ag because of its similar regulation and biochemical characteristics (Lindler ef at.. 1990). Furthermore. pH 6 Ag was found to be necessary for virulence of Y. pestis in the mouse model by the intravenous route of infection (Lindler ef a/.. 1990).
in the previous study (Lindler etai, 1990). we isolated several Escherichia coli clones which encoded loci necessary for the synthesis ot the Y. pestis pH 6 Ag. However, pH 6 Ag was not regulated by pH or temperature in these E. co//clones (Lindler et ai. 1990). The structural gene designated psaA mapped within a 1.7kb EcoRI-SamHI fragment of Y. pestis chromosomal DNA. We also isolated a single TniOlacZ insertion 1.2 kb upstream of psaA which greatly reduced the expression of pH 6 Ag by mutant Y. pestis (Lindler et ai, 1990). This transposon mutation defined an auxiliary locus designated psa£ and mapped within a 0,9 kb HcoRI fragment of DNA, Gene fusions to both psaE and psaA allowed us to determine the direction of transcription for both of these loci. Further, a third locus was suggested to be involved in pH 6 Ag expression by transposon mutagenesis and deletion analysis (Lindler ef ai, 1990). The third locus was downstream from psaA and was located near a SamH! restriction enzyme recognition site. The expression of the precursor and processed form of pH 6 Ag was altered by a single transposon mutation within 60 bp of the BamH\ site (Lindter ef ai, 1990). We now designate this third locus psaB.
The expression of pH 6 Ag at acidic pH and 37°C suggested that it might be synthesized in host macrophage phagolysosomes or extracellularly in abscesses such as buboes (Lindler et al. 1990). Here, we present results which demonstrate that /. pestis pH 6 Ag is a fibriilar structure and is induced by Y. pestis present inside cultured macrophages.

DNA sequence oftheV. pestis psaE, psaA, psaB and downstream partial open reading Irame
We determined the nucieotide sequence of the region of Y. pestis chromosomal DNA encoding pH 6 Ag and the accessory loci {Fig, 1), Four significant open reading frames (ORFs) were detected which would be transcribed in the direction identified previously from m-Mu dl1734 and Tn ?D/acZgene fusions (see above and Lindler et at.. 1990). The first ORF began at bp 397 and ended with a TGA nonsense codon at bp 1041 (Fig.2). This ORF includes the site of the TnTO/acZ mutation which defined psaE (Lindler et al.. 1990). We determined the position of the transposon insertion in pPSN1 {psaE::Tr\10lacZ) by comparing the nucieotide sequences of the mutant plasmid with the non-mufagenized DNA. The Tn lOlacZ mutation occurred by insertion of the transposon 3' to the guanosine residue at bp 736 ( Fig. 2), The second ORF began at bp 2056, 2071 or 2077 and terminated in a nonsense codon at bp 2547. The predicted protein sequence which was measured from bp 2071 is shown in Fig. 2 because this coding region Is preceded by a weak Shine-Dalgarno (1974) sequence 9bp upstream of the ATG translation initiation codon. This second ORF is located in the region previously mapped as the psaA structural gene (Lindler et al, 1990). The 158-amino-acid PsaA protein (Fig, 2) has a predicted pi of 5.8. This predicted isoelectric point is in agreement with our previous findings that the pH 6 Ag was acidic when analysed by two-dimensional protein electrophoresis (Lindler era/,, 1990). A third ORF was found between bp 2674 and bp 3492 (Fig. 2). The third ORF designated psaB spanned the region previously found to be necessary for the expression of the mature 15 kDa pH 6 Ag polypeptide (Lindler etal.. 1990). A partial ORF began at bp 3579 and continued past the end of the DNA for which we determined the nucieotide sequence ( Fig. 2), Because of the incomplete sequence information, we designated this putative protein coding region asORF4'.

Predicted protein sequence similarities
We analysed the EMBL and Genbank Data Libraries for similarity with our predicted PsaE, PsaA, PsaB and 0RF4' sequences. No significant similarity was found with the PsaE protein. The pH 6 Ag structural gene product (PsaA) had limited similarity to the £ coliP pilus adhesin. PapG (Lund etal.. 1987), and an influenza virus haemagglutinin (InfH; Air, 1981). The protein alignments are shown in Fig. 3. Considering highly conservative amino acid substitutions, these regions of PsaA (Fig. 3) were 40% similar to PapG and 54% similar to tnfH. The PsaB and 0RF4' predicted proteins had a high degree of similarity to several accessory proteins involved with pilus or capsule expression by enteric bacteria (Fig, 4; 0RF4' data not shown). The 273-amino-acid PsaB protein was similar to fhe E. coli chaperone proteins for CS3 (Jalajakumari ef al.. 1989). Pap (Holmgren and Branden, 1989) and K88 (Bakker et al.. 1991) pili as well as the KlebsiellapneumoniaepWus. Mrk (Allen etal., 1991) and the molecular chaperone for the Y. pestis capsule protein fraction 1 (Fl; Galyov et al, 1991). The protein sequence alignment of these molecular chaperones is shown in Fig.  4. The identity with PsaB ranged from 41% to 22% for the Fl and K88 chaperones, respectively. Considering conservative amino add substitutions, the similarity with PsaB ranged from 61% for Fl chaperone to 47% with the K88 chaperone. Holmgren et al (1992)   Rg. 2. Nucleotide sequence of V. pes/is pH 6 Ag including psaE, psa^, psafl and 0RF4V Nucleotide residue numbers are shown on Ihe left. Putative protern products are labelled and shown below the DNA sequence. Predicled secretion signal peptides are underlined below Ihe amino actd sequence of the proteins (Von Heijne. 1986). Restriction enzyme recognition sites are overlined and labelled above the DNA sequence. Potential Shine-Dalgarno (Shine and Dalgarno, 1974) sequences are underlined betow the DNA sequence. The asterisk above Ihe nucleotide sequence indicales the pofnt of insertion of Tn lOlacZin pPSNI lpsaE::Tn10tacZ). These sequence data appear in the EfylBUGenBank/DDBJ Nudeotide Sequence Data Libraries under the accession number U86713. 4). Similar results were obtained from database searcbes with the 0RF4' amino acid sequence (data not shown). Primary protein sequence identity with ORF4' ranged from 29% to 22% for the E. coli Pap (Norgren er ai. 1987). K88(Mooie/a/., 1986), and K99 ) pilus accessory proteins PapC, FaeD and FanD. respectively. Tbe FaeD, FanD, and PapC proteins are members of a family of proteins each having a molecular mass of approximately 85 kDa (de Graaf, 1990). The PapC family is a group of outer membrane proteins necessary for the transport and assembly of pilin subunits into mature pill in £. coli(6e Graaf, 1990).

Correlation of ORFs and expressed proteins
To determine if the identified ORFs produced proteins having molecular weights similar to that predicted by DNA sequence analysis, we performed in vitro transcriptiontranslation. Several linear DNA templates corresponding to various lengths of V, pestis chromosome were generated by the polymerase chain reaction (PCR) and used as templates for in vitro transcription and translation (Fig. 5   The 24 kDa size of this protein is in close agreement with the 23958 predicted molecular weight of PsaE. A slightly faster migrating species was produced by these reactions. This result suggested the removal of the protein signal sequence from PsaE that was predicted by primary protein sequence analysis (Fig, 2). Template C (Fig. 5, lane C) specifically directed the synthesis of 18, 17 and 15 kDa proteins. The 17 kDa and 15 kDa species reacted with pH 6 Ag specific antiserum when similar extracts were analysed by Western blotting (data not shown). The 18 kDa protein may be the product of an ORF which begins at bp 1038 and terminates at bp 1526 (Fig. 2). We are unsure if the 18 kDa protein is necessary for the biosynthesis of pH 6 Ag, However, the predicted translation product of this ORF (bp 1038 to bp 1526) did not have significant similarity to proteins in the current databases. A protein with a size similar to the 30 648 molecular weight predicted for the psaB gene product is specifically synthesized from the appropriate region of Y. pestis DNA (Fig. 5 , 1984). In contrast, /. pestis KIM5-3001.1 {psaA') did not cause agglutination of the SRBCs. We found that E, coli HB101 containing either pDGI or pDG27 also specifically agglutinated SRBCs when cultures were grown in Luria broth (LB) at 37°C, These results suggest that all genetic information necessary for synthesis, transport and assembly of pH 6 Ag was contained within the 6.5 kb Kpnl to Cla\ fragment of Y. pestis DNA. However, E. co//containing pDG9 did not cause agglutination of SRBCs. The plasmid pDG9 contains a 3.1 kb Kpn\ to SamHI fragment of Y. pestis DNA ( Fig. 1; Lindler et al., 1990) and therefore encodes PsaE and PsaA but only the first 133 amino acids of PsaB, Accordingly, PsaB and/or downstream genetic information was necessary for functional expression of pH 6 Ag in E. coli.

Electron microscopy and immunogold labelling ofpH 6 Ag
Protein sequence similarities and haemagglutination reactions suggested that pH 6 Ag was fimbrial. Accordingly, we examined Y. pestis and E, coli clones for the presence of fimbriae by electron microscopy after growth using the appropriate inducing conditions. We were unable to observe fimbriae associated with the ceil surface of Y. pestis KIM5-3001, This may have been because of the presence of the Fl capsular Ag produced by Y. pestis (Brubaker, 1972), Fl is a capsular protein which covers the surface and surrounding milieu of Y. pestis with 'granular particles' which form an extracellular matrix {Chen and Elberg, 1977), To visualize the location of Fl, we labelled Y. pestis KIM5-3001 grown under pH 6 Ag-inducing conditions with monoclonal antibody (mAb) 6H3. which recognizes Fl, and examined tfie grids by electron microscopy. These studies revealed the presence ot immunogold-labelled F1 capsule associated with the cell surface and surrounding environment {data not shown). Thus, it was possible that the presence of F1 on Y. pestis KIM5-3001 was obscuring the morphology of pH 6 Ag fimbriae. Consequently, we chose, first of all, to study pH6 Ag produced by E. coli HB101 containing pDGI, Uranyl-acetate-stained E. coli harbouring pDGi produced fibriilar organelles that protruded from the cell surface (Fig, 6), Fimbriae were consistently observed on the surface of E. coli clones that had been grown overnight with aeration in broth cultures; however, we had difficulty observing fimbriae on the surface of clones that had grown in logarithmic phase or on the surface of solid agar. This latter result was in agreement with our previous observation that very little pH 6 Ag could be extracted Rg. 6. Electron pholomtcrographs ol negatively stained and immunogoid-labelled E. co//containing the doned pH 6 Ag locus, A, £. coli harbouring pDGI negativeiy stained with uranyl acetate expressing the three pH 6 Ag fibriilar morpholypes, B. Higher magnification of (A) showing the ultrastructural detail of the three morphotypes of pH 6 Ag, i.e. single strands, multistranded bundles, and large aggregates, C. £ coli HB101 containing pDGI labelled by the immunogold technique. Cells were incubated wilh a 1:20 dilulion of pH 6 Ag-specific antiserum. Inset is a higher magnification of (C) (region designated by arrow) wtiich shows more clearly the binding of immunogold particles to the aggregative fibriilar morphotype. D, E. CO//HB101 containing the pHC79 cloning vector labelled by the immunogotd technique using similar conditions as in (C), In these micrographs. CS designates cell suriace, Untabelled arrowheads designate the single-stranded pH 6 Ag fimbriae morphotype. Arrows labelled with 'a' designate the aggregative fibriilar morphology, while arrows labelled with 'b' designate the multistranded bundles of pH 6 Ag fimbnae. Bar markers: a, 0,1 \im\ b, 0.05 jim; c, 0,25 |xm, (inset bar is 0,05 nm); d. 0,5 jim.
with KSCN from the surface of E coti clones that had been grown in logarithmic phase (Lindler et al.. 1990). The fimbriae were visualized as subtle, fine, singular strands (approximately 4nm in diameter), as multistranded bundles of three or more fimbriae. or as iarge aggregates (Fig. 6. A and B). The immunogold labelling technique was useful in identifying these structures as pH 6 Ag (Fig. 6C). Anti-pH 6 Ag serum coated immunogold complexes were specifically observed bound to fibrillar structures expressed on the surface of E. co//HBi 01 containing pDGi (Fig. 6C) as well as pDG5 and pDG27 (data not shown). There was no binding of pH 6 Ag serum to E CO//HB101 containing the cloning vector (Fig. 6D), These studies demonstrated that the expression of pH 6 Ag in E co//resulted in fimbriation of the host cell.
Our studies revealed that pH 6 Ag was highly conserved between Y. pestis and Yersinia pseudotuberculosis PBi/-f at both the genetic and immunologic level (see Fig. 9 later, and below). Therefore, we chose to examine the morphology of pH 6 Ag in K pseudotuberculosis PB1/+ since this organism does not synthesize Fl capsule (Brubaker. 1972), All three morphotypes of pH 6 Ag % Rg. 7. Electron photomicrographs of negatively stained and immunogold-labelled Vers/niaspp. A, High-magnification photomicrograph of V, pseudotuberculosis PB1/+ negatively slained with uranyl acetate, B, Fibriilar pH 6 Ag on a grid atter immunogoid labelling of Y. pestis KIM 5-3001 with undiluted mAb 6H3 and negative staining with PTA, C, Y. pseudotuberculosis PB1/+ labelled by the immunogold technique. Cells were incubated with a 1 :t 000 dilution of pH 6 Ag-specific antissrum, D, Crude KSCN extract of V, pseudotuberculosis FBI/*-grown at 37 C and pH 6, Immunogold labelling was with a 1 ;1000 dilution ot anti-pH 6 Ag antiserum. No labelling was observed when KSCN extracts oi Y. pseudotuberculosis PB1/+ grown at 37 C and pH 8 were treated similarly. See the Fig. 6  were seen protruding from the surface of uranyl-acetatestained celts (Fig. 7A), Furthermore, we observed pH 6 Ag fibriilar bundles on the grids of Y. pestis which had been immunogold labelled with anti-FI mAb (Fig. 7B), These observations confirmed that pH 6 Ag fimbriae were produced by both Y. pestis and Y. pseudotuberculosis. Lastly, pH 6 Ag fimbriae were identified by the immunogold labelling ot structures with similar morphologies on the surface (Fig. 7C) or in crude KSCN preparations (Fig.  7D) of /. pseudotuberculosis PB1/-(^ grown at 37"C and pH 6. In control experiments, when Y. pseudotuberculosis was cultured at 37"C and pH 8, we did not detect any labelling (data not shown). Thus, the specificity of the immunogoid reaction in identifying structures as pH 6 Ag was confirmed.

Expression ol pH 6 Ag by Y. pestis associated with macrophages
Expression of Y. pestis pH 6 Ag in vitro at 37^0 and acidic pH suggested that the Ag may be synthesized inside phagocytic cells such as macrophages. To test this possibility, we infected the murine macrophage-like cell line, RAW264.7. with Y. pestis which was not expressing pH 6 <pH6Ag < pH 6 Ag Ffg. 6. Expression of pH 6 Ag by V. pestis Inside macrophages in the presence or absence of monensin. Macrophage cell line RAW264,7 was infected with Y. pestis which was not expressing pH 6 Ag (grown at SO^C and pH 7.2). The muttiplicity of infection was approximately one bacterium per macrophage. A. Macrophages infected with Y. pestis psaA' {+\ar,es) ot Y. pesf/s psaA~ (-lanes) strains. Samples were removed at various times after infection of the macrophages for determtnation of viable bacterial counts as well as Western biol analysis. The time, in hours, is indicated above each pair of lanes. The immunoreactive protein corresponding to pH 6 Ag is indicated by the arrow. The equivalent of approximately 5 K 10^ tsacterial cfu was loaded on each lane, B. Effect of the addition of monensin on the expression of pH 6 Ag by Y. pestis inside macrophages, Monensin was added to the infected macrophages as described in the Experimental procedures. The micromolar concentration of monensin is indicated above each lane. The time, in hours, is indicated above each sei of three lanes. The position of pH 6 Ag is indicated with the arrow. The equivalent of approximately 4 x 10* V. pestis cfu was loaded per lane.
Ag. We removed samples of macrophages at various times after infection with V. pesf/s for the determination of viable colony-form ing units (cfu) and Western blot analysis. We found no difference in the survival of Y. pestis KlM-5 3001.1 psaA mutant bacteria and the wild-type parent strain. Expression of pH 6 Ag was not observed immediately after infection of the macrophages (Fig, 8A), However after 10h in the macrophage intracellular environment, pH 6 Ag was expressed by V. pestis psaA^ bacteria (Fig. 8A). It was not likely that the expression of pH 6 Ag observed was due to acidification of the macrophage growth medium. Two lines of evidence support this: (i) the pH ot the medium removed from the infected RAW264.7 cells before sampling was approximately 7; and (ii) the inclusion of gentamicin in the macrophage growth medium to inhibit extracellular replication of the bacteria.
To determine if acidification of the phagolysosome was necessary to induce the synthesis of pH 6 Ag. we treated Y. pes//'s-infected macrophage cultures with monensin (Fig. 8B). Monensin is a carboxylic ionophore which disrupts the acidification of intracellular compartments (Horwitzand Maxfield. 1984;Wileman etal. 1985), Long-term (greater than 10 h) exposure of Y. pesf/s-infected macrophages to monensin concentrations which ranged from 5-10[,iM resulted in cytotoxicity to the infected RAW264,7 ceil line (data not shown). However, treatment of similar infected cultures with monensin concentrations of 0.1 and 1 |JM reduced the expression of pH 6 Ag by /, pestis atter 21 h of growth in RAW264.7 macrophages (Fig. 8B). In control experiments, the addition of 5|JM monensin to broth cultures of V, pestis growing at pH 6 and 37 C had no effect on pH 6 Ag expression. Thus. Y. pestis pH 6 Ag was expressed in association with macrophages in a manner which required acidification of the Intracellular environment.

Molecular epidemiology ofpH 6 Ag
To determine if pH 6 Ag gene sequences were present in other Yersinia spp,. we performed Southern blot (Southern, 1975) hybridization using a psa-A-specific probe. The pH 6 Ag gene probe was completely internal to the psaA locus (bp 2149 to bp 2547. Fig, 2) and was generated by the PCR. The probe hybridized with a 3,2 kb EcoB\ fragment of Y. pestis DNA as well as Y. pseudotuberculosis strains 7. 43, R2, MSU-D and PBI/-1- (Fig, 9). However, the psaA probe hybridized with a 9.3 kb EcoRI fragment 1 2 3 4 5 6 7 8 9 10 11 12 9.3- of Y. pseudotuberculosis 32 genomic DNA (Fig. 9, lane  3). In contrast, no psaA sequences were detected in DNA derived from Yersinia enterocolitica. Shigella flexneri and Salmonella typhi (Fig, 9) even when low-stringency hybridization conditions were used (data not shown). Western blot analysis of Y. pseudotuberculosis PB1/-H grown at 37"C and pH 6 revealed that a protein was produced which immunologicalty cross-reacted with Y. pestis pH 6 Ag (data not shown). The anti-pH 6 Ag reactive material was not produced by Y. pseudotubercuiosis PB1/-f which was cultured at 37°C and pH 8, However, neither Y. pseudotuberculosis 32 nor Y. enterocolitica V^A cultured at 37 C and pH 6 produced a protein which immunologically cross-reacted with Y. pestis pH 6 Ag sera (data not shown). These resuits indicated that all genomic DNAs of the Y. pseudotuberculosis strains we examined did include sequences homologous to V. pesf/s psaA. Furthermore, at least V. pseudotuberculosis PB1/-)produced an acid-inducible polypeptide when cultured at 37''C which immunologically cross-reacted with Y. pestis pH 6 Ag.

Discussion
Our results demonstrate that pH 6 Ag is a fibrjilar structure produced by Y. pestis and Y. pseudotuberculosis. Also, we show that pH 6 Ag is induced by Y. pestis inside macrophages in an acidic intracellular environment. Several lines of evidence from our studies support these conclusions. First, we observed a high degree of similarity between a pH 6 Ag accessory protein (PsaB) and several pilin chaperone proteins. Second, cell-free KSCN extracts of Y. pestis PsaA' bacteria specifically caused the agglutination ot SRBCs when compared with similar extracts prepared from the isogenic psaA mutant strain. Third. DNA hybridization studies and immunoblotting with V, pestis pH 6 Ag-specific reagents revealed that the Ag is expressed by /, pseudotuberculosis. Fourth, our electron microscopy studies revealed fimbriae on the surface of E. CO//HB101 containing cloned pH 6 Ag as well as Y. pseudotuberculosis expressing the Ag. Multistranded bundles of pH 6 Ag fimbriae were also present on grids prepared from V. pestis. Furthermore, these structures appeared to react with pH 6 Ag-specific antiserum. Fifth, we found that induction of Y. pestis pH 6 Ag inside RAW264.7 macrophages required acidification of the intracellular environment (Fig. 8).
The wiry morphology of Yersinia pH 6 Ag fimbriae appears to be similar to CS3. a fibriiiar component of colonization factor antigen II produced by enterotoxigenic E. coli (Levine el al. 1984). Although the primary protein sequences of PsaA and the CS3 fibrillin are not similar, the chaperone proteins for these two fibriiiar proteins share a high degree of identity (Fig. 4). Most of the Yersinia pH 6 Ag fimbriae were observed to be wiry strands which could form complex aggregates. However, these flexible organelles were also seen as single 'fibriilar' strands or as laterally associated thick multifilament bundles of three or more strands. Although lateral association of E. coli CS3 into multifilament bundles has not been reported, similar aggregation of other thin fimbriae into thicker structures has been observed (Olsen et al. 1989;Giron etal., 1991). Surface fimbriae produced by K enterocolitica. Y. pseudotuberculosis (Old and Adegbola. 1984;Skurnik. 1984) and Y. pestis (Vodopianov. 1988) have been described. Immunoblotting of pH 6 Ag clones and Y. pestis expressing pH 6 Ag with anti-K pestis fimbrial antiserum obtained from Russia (Vodopianov, 1988;1990; see the Experimental procedures) indicates that the fimbriae described by Vodopianov are at least partly composed of PsaA. The fimbriae composed of the YopA protein in Y. pseudotuberculosis and /. enterocolitica (Kapperud etal.. 1987) are not synthesized by Y. pestis because of a point mutation in the coding region of the gene (Skurnik and Wolf-Watz, 1989), Various enteric major pilus subunit proteins contain little amino acid similarity (Paranchych and Frost, 1988) over their entire protein sequence. Accordingly, PsaA did not contain any significant amino acid sequence similarity to other fibrillin proteins. However, we found a limited region of similarity near the carboxyl terminus of PsaA with the E. coli Pap pilus adhesin. PapG. Hultgren et al. (1989) demonstrated that the region of amino acids around residues 301 to 314 of PapG is necessary for the adhesin to interact with the chaperone protein, PapD, The similarity we noted between PsaA and PapG encompasses this region of the E. co//protein. Also, the similarity noted between PsaA and the influenza A virus haemagglutinin (Fig. 3) suggests that amino acids 52 to 77 of the /. pestis fibrillin may be involved in binding of the fimbriae to host cells. However, further structurefunction studies will be necessary to determine if these regions of PsaA are involved in chaperone binding and haemagglutination.
The newly designated Y. pestis psaB locus encodes a protein which appears to be a member of a group of molecular chaperone proteins (see Fig, 4 and Results). This conclusion is supported by our previous observation that mutation of DNA in the psaB region resulted in decreased accumulation of the 15 kDa mature form of PsaA (Lindler et al. 1990). Bakker et al. (1991 have shown that the E. coli K88ab pilus chaperone, FaeE, is responsible for protecting the pilin protein from proteolytic degradation as well as preventing premature polymerization of pilin into pilus structures on the surface of the bacterium. Thus, the decrease in the accumulation of mature y, pestis PsaA in psaB mutants may be due to increased proteolysis of the pH 6 Ag fibrillin subunit. The induction of Y. pestis pH 6 Ag fimbriae by growth at acidic pH and mammalian body temperature is unusual among bacterial fimbriae. Expression of most E. co//fimbriae occurs when the bacteria are cultured at 37''C (de Graaf, 1990), Only the Vibrio cholera toxin-coregulated pilus (tcpA) has been shown to be induced when bacteria were grown at pH 6.5 (Taylor ef al., 1987). However. unlike Y. pestis psaA expression, tcpA synthesis is maximum when cultivation of the bacteria is at 30°C. To date, only the psaE and psaB loci (see Results and Lindler et al.. 1990) are known to affect the expression of Y. pestis pH 6 Ag. In Y. pestis psaE mutants, pH 6 Ag expression is regulated normally although the amount of expression is greatly reduced (Lindler ef al.. 1990). The lack of similarity between PsaE and other fimbrial regulatory proteins may reflect the novel regulation of expression of Y. pestis pH 6 Ag. Further studies will be required to evaluate the role of PsaE and other regulatory elements in the expression of pH 6 Ag fimbriae. /. pestis pH 6 Ag has been shown to be expressed in wVo (Ben-Efraim etal.. 1961). These studies also showed that infection of mice with Y. pestis already expressing pH 6 Ag was more rapidly fatal to animals when compared with infection with bacteria not synthesizing the Ag. Previously (Lindler ef al.. 1990), we found that mutation at the psaA locus resulted in a 200-fold increase in the LD50 (50% lethal dose) of the mutant bacteria compared with the wild-type parent Y. pestis when mice were challenged by the intravenous route of infection. Also, the Interaction of y, pestis with host macrophages is important for the pathogenesis of plague (Cavanaugh and Randall, 1959). Expression of pH 6 Ag inside macrophages was observed in our studies (Fig. 8) as well as by others (Vodopianov ef al.. 1990), Taken together, these facts may lend some insight into the function of the Ag during infection by /. pestis. Bacterial fimbriae primarily function as mediating attachment of bacteria to host cells. Induction of expression of V, pestis pH 6 fimbriae inside macrophages may allow the pathogen to interact with other uninfected macrophages or other host cells after the bacteria are released from the infected cell. Infections caused by Y. pseudotuberculosis are usually not systemic but rather are localized infections resulting in acute ileitis and mesenteric lymphadenitis (Butler, 1983). Accordingly, the high conservation of pH 6 Ag fimbriae at the genetic and immunologic levels we observed between Y. pestis and Y. pseudotuberculosis supports the possibility that these fimbriae facilitate the initial stage of pathogenesis. Further investigation of the role of pH 6 Ag in the pathogenesis of plague infection will require the construction of Y. pestis pigmentation-positive psaA" mutants. Also, the expression of pH 6 fimbriae on the surface of bacteria could facilitate their entry into macrophages or other host cells. Experiments towards these ends are currently under way.
Piasmids pPSN1, pDGI. pDG4, pDG6 and pDG9 containing the cloned pH 6 Ag of V, pestis KIM5 were described previously (Lindler e/a/., 1990). Plasmid pDG5 contains a 9 kb C/al fragment of Y. pestis KIM-5 chromosomal DNA present in pDG4 (Lindler et al. 1990) cloned into the vector plC20R (Marsh et al.. 1984), All of the genetic material shown in Fig, 1 is present in pDG5 plus 2,5 kb of DNA to the left of the Kpn\ site, Plasmid pDG27 contains the 6,5 kb Kpn\ to C/al DNA fragment from pDG6 in the pSK+ vector (Stratagene), The direction of transcription of the V, pestis genes psaE. psaA. psaB and ORF4' in pDG27 is the same as that of the T7 promoter present in pSK+, For selection of antibiotic resistance phenotypes. the following antibiotic concentrations were used ((igml"'): chioramphenicol. 25: ampicillin, 100; streptomycin. 100; and tetracycline, 25.

Recombinant DNA techniques and DNA sequencing
Restriction endonucleases, T4 DNA Iigase and frozen E. coli competent ceiis were purchased from BRL Plasmid DNA was purified from E. coli hosts with the Qiagen midi-plasmid purification kit (Qiagen), Rapid screening ot bacteria for plasmid DNA was as described previously (Del Sal et al. 1988), DNA restriction fragments were separated on 0,7% agarose geis (Maniatis et al.. 1989) and transferred to nitrocellulose filters as described by Southern (1975), A DNA probe specific for the V, pestis psaA sequence was generated by the PCR as described below. The PCR reaction was initiated by oligonucleotides which were homologous to bp 2155 to 2180 and the inverse complement of bp 2524 to 2547 (Fig, 2). The DfvIA probe was labelled using the random primers DNA labelling system (BRL), High-stringency filter hybridization and posthybridization washes were as described (Silhavy et al.. 1984), Low stringency was achieved using similar conditions except that 25% formamide was included in the hybridization solution. Post-hybridization washes under low-stringency conditions were similar to those above except that the final two washes were in 2x SSC (0.3 M NaCI, 0,03 M sodium citrate. pH 7,0) at 30"C, Washed and dried filters were autoradiographad at -70' C with X-omat AR film (Eastman Kodak Co), Initial DNA sequencing was on single-stranded templates generated from M13 derivatives (Ausubel etal. 1989), A3.1 kb Kpn\ to SamHI fragment liberated by restriction digestion of pDG9 was ligated into simiiariy cleaved M13mp18 and M13mp19 as described elsewhere (Ausubel et at.. 1989), This 3.1 kb DNA fragment included the psaE. psaA and the 5' end of the psaB loci ot Y. pestis. The nucieotide sequence of the remair>der of the psaB locus and downstream material was obtained from double-stranded DNA template, pDG27, The DNA sequence of both strands was determined using overlapping oligonucleotide primers, DNA primers were synthesized on an Applied Biosystems Incorporated (ABI) Model 380b oligonucteotide synthesizer. Nucieotide sequences were determined by the chain termination method (Sanger et at.. 1977) using Sequenase version 2,0 (United States Biochemical), DNA Sequence manipulation was with the PC/Gene software package (Intelligenetics Corp.). Protein or nucieotide database searches and alignments were with the Genetic Computer Group (GCG) sequence analysis software package for the VAX computer (Devereux etal.. 1984),

Protein gel electrophoresis. the PCR and in vitro transcription and translation
Linear DNA templates were generated by the PCR, The PCR reactions contained, in a 100|,it volume: lOng of pDG27 tempiate DNA, 50 pmoles of oligonucleotide primers and Hot Tub Polymerase (Amersham Corp,) according to the manufacturer's specifications. Oligonucleotide primers used to initiate polymerization at the 3' end of the DNA fragments were as follows: fragment a, GGACGGCTCAATAGCC; fragment b, GCTTTCATTGCTGTTTGC; fragment c, GCATAAGGTAAA-GACACC; fragment d. CCAAGGAGCAGCTATCCCGC. The DNA primer that initiated synthesis at the 5' end of the above sequences was the T7 promoter primer, TAATACGACTCAC-TATAGGG. The annealing times and temperatures were maximized for each primer combination to yield the specific synthesis of the linear fragment. Linear DNA fragments were purified by the Qiagen PCR purification kit (Qiagen Inc). tn vitro [^^S]methionine labelling of proteins encoded by the above linear and plasmid DNA templates was with a commercially available E. coli S30 extract (Promega Corp.). Reactions were according to the manufacturer's directions except that they contained 50 units of T7 RNA polymerase (BRL), In vitro [^^S]-methlonine-labelled proteins were separated on 4-20% denaturing polyacrylamide geis (SDS-PAGE) according to the method of Laemmli (1970). After electrophoresis, gels were impregnated with En^hance (New England Nuclear), dried, and fluorographed at-70'C. treated with 0,05% trypsin and 0,53 mM EDTA then dislodged from the tissue culture flask. Suspended macrophages were washed three times wHh DMEM and resuspended in like medium to a concentration of approximately 3-5 x 10''' per ml. A sample of Y. pestis which had been grown overnight at 3O'C was added to the macrophage suspension to give a multiplicity of infection (m,o.i.) of approximateiy one bacterium per macrophage. The infected mixture of cells was centrifuged at 500 ?< g for 5 min at room temperature. After centrifugation, the pelleted bacteria and macrophages were incubated at 37"C for 5 min. The supernatant was decanted and the cells were then suspended in fresh DMEM, Samples of 0.1 ml were placed into 96-well culture plates (Corning 25860, Corning Glass Works) and incubation continued at 37 C for 20 min. The medium was removed from each well and replaced with like medium containing 5i,igmr' gentamicin. In some experiments, various concentrations of monensin were added to the DMEM containing gentamicin. After 30 min further incubation at 37"C, an Initial time zero sample was taken. Samples were prepared by washing groups of 10 wells three times with Hanks' balanced salts solution (HBSS), Samples for determination of viable bacterial counts were pools of five of these wells that had been overlaid with 0,1 ml of ice-cold sterile water. After complete lysis of the macrophage cells, the remaining bacteria were appropriately diluted in 0,9% NaCI and inoculated onto BHI agar plates. The cfu ml"' sample was determined after incubation at SO'C for 48 h. Using these macrophage infection conditions, uptake of Y. pestis ranged from 30 to 60% of the input bacteria. Samples for Western blot analysis consisted of the contents of five wells extracted with a total of 0,1 ml SDS-PAGE sample buffer (Laemmli, 1970). The extracts were heated at 95'C for 15 min and stored at -2O''C, Similar extracts were made at the times indicated, Macrophage culture medium was changed with fresh medium when the growth medium became acidic.
Proteins from the above extracts were electrophoresed and transferred to nitrocellulose (Towbin et al. 1979) then processed as described previously (Lindler et at.. 1990), Primary antibody (Ab) was pH 6 Ag-specific rabbit polyclonal antiserum diluted 1:1000 (Lindler et at.. 1990). To determine if pH 6 Ag was immunologically cross-reactive with the Y. pestis EV76 temperature-and pH-induced fimbriae previously described (Vodopianov. 1988). primary antibody was rabbit anti-Y.-pestis fimbrial antiserum kindly provided by Boris Mishankin, Research Anti-plague Institute, Rostov-on-Don, Russia. The latter antiserum was reacted with Western blots of crude whole-cell extracts of E. coti pH 6 Ag clones and V. pestis expressing the Ag or negative controls. Secondary antibody was biotinytated donkey-anti-rabbit serum (Amersham) diluted 1:1000, Immunoreactive protein was visualized by reaction with streptavidin horseradish peroxidase (Amersham) and the 3.3'5,5'-tetramethylbenzidine dihydrochloride (TMB) horseradish peroxidase substrate system (Kirkegaard and Perry Labratories).

RA W264.7 infection and Western blotting
Murine macrophage cell line RAW264.7 was cultured in Dulbecco's modified Eagle's medium (DMEM) at 37"C in an atmosphere of 7,5% CQg as described previously (Kelly et at.. 1991). Before infection with Y. pestis. macrophages were

Haemagglutination assay
Heparinized SRBCs were washed three times in normal saline (NS; 0,9% NaCI) and suspended in the wash solution to 0.3% (v/v), E. coli expressing pH 6 Ag or negative controls were grown overnight at 37''C in LB. Y. pestis strains were grown overnight in SBHl pH 6 at37'*C. E. coll and V. pestis were aerated by agitation at 120 r,p,m. in a New Brunswick Innova Model 4300 shaking incubator (New Brunswick Scientific). For haemagglutination assay, 0.1 ml of washed SRBCs was mixed with an equal volume of NS-washed overnight bacterial culture. Bacteria were mixed with SRBCs in a 1,5 ml microcentrifuge tube followed by incubaiion for 2 h at 37'C, NS was included as a negative control. After incubation, tubes were observed macroscopically for agglutination of the SRBCs. Tubes which appeared negative by macroscopic examination were also examined microscopically.
Electron microscopy and immunogold labelling E. coti was grown in LB overnight at 37'C, Y. pestis 1 (psaA) or the isogenic wild-type strain were grown at 37''C and pH 6 as described (see above and Lindler et at.. 1990). V. pseudotuberculosis PB1/+ was grown at 37''C in SBHl adjusted to either pH 8 or 6 until cultures reached mid-log phase. Bacteria were washed twice and suspended in distilled water. The bacterial suspension was placed on carbon coated 300-mesh copper grids then negatively stained with 0.5% uranyl acetate or 1% phosphotungstic acid (PTA) pH 7,2 and examined directly with a Phillips 400 HM transmission electron microscope operated at an accelerating voltage of 80 kV.
Immunogold labelling was as described elsewhere (Beesley. 1989), Briefly, bacteria suspended in phosphate-buffered saline (PBS; 8g NaCI. 0.2g KCI. 1.44g Na^HPO^, 0.24g KH2PQ4 per litre, pH adjusted to 7.4 with HCI) were deposited on 0.25% formvar carbon coated 300-mesh copper grids and partially dried. Primary antibody (Ab) was polycional Y. pestis pH 6 Ag-specific sera (Lindler ef ai, 1990) diluted as indicated in fig, legends 6 and 7, In some experiments, Y. pesfe-containing grids were reacted with mAb 6H3, which recognizes the Fl capsular Ag; 6H3 was kindly provided by Dr John Ezzet, Department of Bacteriology, United States Army Research Institute of Infectious Disease, F. Detrick, MD. USA. The secondary Ab was goat anti-rabbit or anti-mouse IgG labelled with lOnm gold particles diluted t:10. After final washing with distilled water, the labelled bacteria were stained with 1% PTA pH 7.2, and examined as described above.