S528

Potential Relevance of Chlamydia pneumoniae Surface Proteinsto an Effective Vaccine

Gunna Christiansen,1 Anna-Sofie Pedersen,3

Karin Hjernø,1 Brian Vandahl,1,2 and Svend Birkelund1

Departments of 1Medical Microbiology and Immunologyand 2Molecular and Structural Biology, University of Aarhus,

and 3Loke Diagnostics, Science Park, Aarhus, Denmark

The surface of Chlamydia pneumoniae is covered with proteins but their exact identificationis not known probably because of the presence of conformational epitopes. A family of 21pmp genes has been found by DNA sequencing. In common, these genes have the capacityto encode the amino acid motif GGAI. Several of the genes have the capacity to encode outermembrane proteins of about 100 kDa. Thus, they are candidate genes to encode the protein(s)present in the 98-kDa protein band of the C. pneumoniae outer membrane complex. Theproduction of recombinant GGAI proteins is described as is the use of polyclonal antibodiesraised against the recombinant GGAI proteins to determine their expression in C. pneumoniaeelementary bodies. At least three of the proteins, Omp4, 5, and 11, are expressed.

The respiratory pathogen Chlamydia pneumoniae causes up-per and lower respiratory infections [1, 2]. The diagnosis of theinfection is infrequently verified by isolation of the microor-ganism, but frequently it can be confirmed by serology by anincrease in antibody titer [3]. The method of choice for serologyis microimmunofluorescence (MIF), which allows comparisonof antibody binding to C. pneumoniae and Chlamydia trachom-atis elementary bodies (EBs). This method showed an associ-ation between C. pneumoniae and development of atheroscle-rosis [4]. The MIF test depends on binding of human antibodiesto surface-exposed C. pneumoniae components. In order tomodify the serodiagnostic test it is important to identify thesurface-exposed components.

Common to all Chlamydia species is the biphasic life cycleand the genus-specific lipopolysaccharide (LPS) epitope. LPSis present on the surface of both C. pneumoniae and C. tra-chomatis [5–7]. The human humoral immune response readilyproduces antibodies binding to Chlamydia LPS. The presenceof LPS epitopes on both C. trachomatis and C. pneumoniaenecessitates the presence of both organisms in the MIF test.The other major C. trachomatis immunogen is the major outermembrane protein (MOMP). In C. trachomatis, MOMP is sur-face exposed [8] but apparently is not in C. pneumoniae [9].This may explain why MOMP is not a major C. pneumoniaeimmunogen [10].

The surface of C. pneumoniae EB is covered by protein(s).Their exact identification has not been made, probably due to

Grant support: Danish Health Research Council, Danish Veterinary andAgricultural Research Council (12-0850-1, 12-0150-1, 20-3503l), AarhusUniversity Research Foundation, and Novo Nordisk Foundation.

Reprints or correspondence: Dr. Gunna Christiansen, Dept. of MedicalMicrobiology and Immunology, Bartholin Bldg., University of Aarhus, DK8000 Aarhus C, Denmark (gunna@medmicro.au.dk).

The Journal of Infectious Diseases 2000;181(Suppl 3):S528–37q 2000 by the Infectious Diseases Society of America. All rights reserved.0022-1899/2000/18106S-0033$02.00

the presence of conformational epitopes [6]. This is reflected bythe fact that C. pneumoniae–specific monoclonal antibodies thatrecognize the C. pneumoniae surface do not react in immuno-blotting, making the identification of the proteins difficult [6,7, 9, 11]. An indication of the possibility that several differentproteins may be present or be interchangeable came fromKnudsen et al. [11]. They showed that at least 4 genes, omp4-7, in a gene family encode proteins with a repeated motif con-sisting of 4 amino acids (aa): glycine, glycine, alanine, and iso-leucine (GGAI proteins). The proteins were detected in anexpression library screened with a polyclonal antibody directedagainst the sarcosyl-insoluble, SDS-denatured C. pneumoniaeouter membrane complex (OMC). Sequencing of the C. pneu-moniae genome [12, 13] revealed that this GGAI family has 21members, the pmp/omp genes, that use ∼5.5% of the codingcapacity of the genome. The presence of such a large gene familyis an indication of their potential importance for C. pneumoniae.A similar gene family was identified in the C. trachomatis ge-nome [13, 14], where 9 genes with the capacity to encode GGAIproteins were identified. Their potential function and locali-zation have not been determined. Five genes of Chlamydia psit-taci encoding GGAI proteins have been sequenced [15, 16].These proteins are major immunogens recognized by post-abortion sera from experimentally infected C. psittaci ewes.These proteins are synthesized early in the developmental lifecycle and are present in the C. psittaci OMC [17], and their N-terminal part is surface exposed [18].

Materials and Methods

Microorganisms. C. pneumoniae VR1310 was cultivated inHEp-2 cells in RPMI 1640 (Gibco BRL, Gaithersburg, MD) con-taining 25 mM HEPES buffer and 10% fetal calf serum (GibcoBRL; inactivated by heating at 567C for 30 min and sterile filtered)and purified as described [11, 19]. For cloning and expression of

JID 2000;181 (Suppl 3) C. pneumoniae Surface Proteins S529

recombinant (r) proteins, we used Escherichia coli BL21(DE3)(NovaGene, La Jolla, CA).

Computer analyses. For computer analysis, we used GeneticsComputer Group (GCG 8.1 and 9.1; Madison, WI) packages.Multiple alignment was performed with the PILEUP program.To search databases for sequences with the motif FxxNx(4,14)GGA[ILV], we used the FINDPATTERN program. Data-base search for protein homologs was done with the similaritysearch programs FASTA and BLAST (release 37) available atSWISSPROT (http://www.expasy.ch) and the GCG package. Wedid structural analyses with two programs: ProDom (update 23July 1999; http://protein.toulouse.inra.fr/prodom.html) and PEP-TIDESTRUCTURE (GCG). The Fortran program AMPHI (1988MPI Biology, Tuebingen, Germany) [20] was used for predictingthe transmembrane b-strands. The program Compute pI/Mw (http://www.expasy.ch/tools/pi_tool.html) [21] was used for determina-tion of the isoelectric point and molecular weight (GenBank ac-cession numbers: AJ001311, AJ133034, AJ133035, U72499,U65943, U65942, P45508, and Q03155).

Construction of rOmp4-hybrid protein. The recombinant hybridprotein between E. coli AIDA (Q03155) and C. pneumoniae Omp4(AJ001311) was constructed by polymerase chain reaction (PCR)as described [11]. The upstream primer was constructed to encom-pass the pET30 LIC plasmid LIC sequence, two stop codons, andthe E. coli Shine Delgarno sequence followed by 22 nucleotides ofthe omp4 sequence beginning with ATG (fMet). The omp4 down-stream primer contained the omp4 sequence (aa 525–531) tailed bythe AIDA b-barrel sequence (aa 839–848). To amplify the b-barrelpart of E. coli AIDA, we used plasmid pIB264 (provided by M.Alexander Schmidt, Westfalische Wilhelms-University, Munster,Germany). The upstream primer was complementary to the tail ofthe omp4 downstream primer. The downstream E. coli AIDAprimer encompassed the region downstream of the C-terminal partof the gene (3981–4005 bp; GenBank accession number, Q03155)tailed by a LIC site. Each primer set was used for PCR of C.pneumoniae and E. coli pIB264 DNA [15]. The two PCR productswere mixed and amplified by two LIC primers. The hybrid genethus obtained was cloned into the pET30 LIC vector (Novagen,Madison, WI) and transformed into E. coli NovaBlue. For ex-pression, the plasmid was transformed into E. coli BL21 (DE3)after control sequencing. The hybrid protein thus consisted ofOmp4 aa 1–531 and AIDA aa 839–1287.

Antibodies. Antibodies to C. pneumoniae OMC were obtainedby purification of C. pneumoniae OMC [8]. Rabbits were immu-nized as described with purified C. pneumoniae OMC [11], with orwithout solubilizing the proteins in SDS sample buffer before im-munization (K151 to solubilized native OMC and K150 to SDS-denatured OMC). Recombinant proteins to the C. pneumoniaeOmp4, 5, 6, 7, 9, 10, 11, 13, and 15 were obtained as fusion proteinsfrom LIC cloning prepared as described [11] (K195, Omp4; K179,Omp5/K203; K204, Omp6; K207, Omp7; K208, Omp10; K201,Omp11; K202, Omp13; and K209, Omp15). Human serum sampleswere from healthy Danish blood donors.

Immunoblotting. Immunoblotting was performed as described[22]. Purified C. pneumoniae EBs were suspended in SDS samplebuffer, loaded to a 10% SDS polyacrylamide gel, or boiled for 5min before being loaded onto the gel to fully denature the proteins.The proteins were electrotransferred to nitrocellulose. Immuno-

blotting was done with K150 and K151 diluted 1:1000 or withhuman sera diluted 1:200. Antibody binding was detected withalkaline phosphatase–conjugated goat-anti human IgG (BioRad,Richmond, CA).

MIF microscopy. HEp-2 cells cultivated on coverslips wereinfected with 0.7 inclusion-forming units (ifu) per C. pneumoniaecell. After infection for 72 h, the infected monolayer was fixed withmethanol for 1 h as described [19]. The infected monolayers werethen incubated with primary antibodies diluted 1:50 for 30 min,washed, and incubated with fluorescein isothiocyanate–conjugatedsecondary antibodies (goat anti-rabbit IgG 1:100, JacksonImmunoResearch Laboratories, West Grove, PA). MIF microscopywas done as described [19].

Two-dimensional (2-D) PAGE and immunoblotting. 2-D PAGEwas done as described [23]. Semiconfluent monolayers of HeLacells were infected with 1 ifu per cell of C. pneumoniae VR1310.For detection of chlamydial protein synthesis, we used a methio-nine/cysteine-free RPMI 1640 medium containing 10 mg/mL gen-tamicin, 20 mg/mL cycloheximide, and 100 mCi/mL [35S]methio-nine/cysteine (Promix; Amersham Laboratories, Amersham, UK).At the end of the labeling period, cells were washed in PBS, loos-ened with a rubber policeman in PBS, and (after centrifugation at16,000 g for 30 min) resuspended in 30 mL of 10% SDS and boiledfor 5 min. The sample was then resuspended in 1 mL of lysis buffercontaining 9 M urea, 4% CHAPS, 40 mM Tris base, 65 mM di-thioerythriol (DTE), and pharmalyte 3-10 (Pharmacia Biotech,Uppsala, Sweden) and sonicated and centrifuged at 16,000 g for10 min. Supernatants containing the labeled chlamydial proteinswere stored at 2707C until use.

For isoelectric focusing, we used 18-cm-long linear immobilized(pH 4-7) gradient drystrips (Pharmacia). Each strip was soakedovernight with unlabeled C. pneumoniae proteins with addition ofan 35S-labeled sample in lysis buffer in a reswelling tray (Phar-macia). The amount of 35S-labeled chlamydial protein was adjustedto 200,000 cpm per strip. Rehydrated strips were run in the firstdimension at 300, 350, 400, 450, and 500 V each for 0.5 h followedby 3500 V for 3 h and 5000 V for 20 h at 157C. After the focusingwas completed, the strips were equilibrated for 15 min in a buffercontaining 6 M urea, 30% wt/vol glycerol, 2% wt/vol SDS, 0.05 MTris-HCl, pH 6.8, and 2% wt/vol DTE. The strips were subse-quently equilibrated in a buffer in which DTE was replaced by2.5% wt/vol iodoacetamide. In the 2-D run, the proteins were sep-arated on 9%–16% linear gradient SDS-polyacrylamide gels (18

mm) until the bromophenol blue front reached thecm 3 20cm 3 1bottom of the gel.

Gels were soaked in deionized water and then equilibrated in abuffer containing 13 mM Tris, pH 7.2, 100 mM glycine, and 10%vol/vol methanol for 30 min. The polyvinylidene difluoride (PVDF)membranes (Immobilon-P; pore size 0.45-mm; Millipore, Bedford,UK) were soaked in methanol for 1 min and equilibrated as forthe gels. After electroblotting, the membranes were immunostainedas described [22].

Electron microscopy (EM). EM was performed as described[24]. Recombinant E. coli Omp4 hybrid cells or plasmid-free E.coli cells, cultivated in Luria broth and induced with 1 mM iso-propyl b-D-thiogalactopyranoside in the presence of 5 mM b-mer-captoethanol for 2 h, were mounted on carbon-coated, glow-dis-charged grids. After adsorption, immunodetection was done using

S530 Christiansen et al. JID 2000;181 (Suppl 3)

Table 1. Characterization of GGAI proteins.

Omp Pmp

aa(including

leader)

Mw(withoutleader) pI GGAI Characteristics

4 11 929 97.0 5.91 6 Lipid modified5a 10 928 95.6 5.22 6 Lipid modified6 1 923 97.7 6.34 67 2 842 87.5 5.84 48 3 177 18.4 8.85 0 Frameshift9 4 591 59.3 5.01 6 Frameshift

10 9 928 95.5 5.91 611 8 931 95.5 5.16 612 7 937 97.5 5.66 513 12 515 53.9 4.94 2 Truncated14 13 974 100.7 6.52 615 5 494 48.4 4.76 4 Frameshift— 6 1408 143.0 5.31 16— 14 979 101.2 6.79 3— 15 939 100.2 6.48 6— 16 935 102.1 6.04 9— 17 275 26.6 5.69 2 Frameshift— 18 893 97.7 6.16 8— 19 948 101.1 8.56 9— 20 1724 117.4 5.36 11— 21 1610 117.8 4.94 5

NOTE. aa, amino acid; Mw, molecular mass.a Frameshift was found in genome sequence [12].

the polyclonal antibody K151 (diluted 1:500) or patient serumsamples (diluted 1:50). As secondary antibodies, we used goat anti-rabbit or goat anti-human conjugated with 15 nm of gold (BioCell,Cardiff, UK). EM was done with an electron microscope (Jeol1010) equipped with a Kodak megaplus 1.4 slow scan camera. Theimage was transferred to a SUN SparcStation10 with an SDVdigital video camera interface.

Results

GGAI gene family. A homology search with the aa se-quences of Omp4 and Omp5 [11] against the deduced open-reading frames within the C. pneumoniae genome [12] revealed21 genes with the potential to encode proteins. The predictedsizes and structures of these genes are shown in table 1. Acharacteristic of the hypothetical proteins was that they couldbe divided into two potential domains: an N-terminal domainconsisting of a region with the motif GGAI present in a variablenumber and a C-terminal part. Ten GGAI genes were se-quenced by our laboratory in parallel with the genome se-quencing project (table 1). omp4–15 genes were sequenced byour laboratory and the pmp1-21 genes were from the genomesequence [12]. Only one difference was found in the nucleotidesequences (at position 912 after the ATG in Omp5, there is aC residue less in the genome sequence than described by Knud-sen et al. [11]). This changes the open-reading frame to a pre-mature stop in the genome sequence. Omp4, 5, and 10–14 werepositioned in a cluster on the genome together with Pmp6 andPmp15–18. In this cluster, all genes except Omp5 were in thesame direction of transcription. Similarly, the Omp6–9 and 15were localized in a cluster and all had the same direction oftranscription. Finally, the genes encoding Pmp19 and 20 andthe gene encoding Pmp21 were localized at different positionson the genome [12].

As indicated in table 1, the DNA sequence of the genesshowed they belonged to the GGAI gene family. Only 1 gene,omp13, had a truncated version of the gene with only 2 GGAIrepeats. In 4 genes, omp8, omp9, omp15, and pmp17, frameshiftmutations lead to a premature stop of the proteins in an oth-erwise full-length gene. Due to such a mutation, Omp8 did notcontain the GGAI motif. The deduced aa sequence of omp/pmp genes showed that 12 of the 21 genes had the capacity toencode proteins of very similar size (95.5–102 kDa) and 1, pmp6,could encode a protein of 143 kDa (table 1). Alignment of thepredicted aa sequence showed the over all aa sequence identityto be low. The alignment of Omp4, 5, 10, and 11 is shown infigure 1. Patches of identical aas, of which the motif GGAIwas repeated 6 times in the N-terminal part of the molecules,were scattered along the protein sequence. No particularly con-served parts could be identified. The alignment showed smallgaps scattered along most of the sequence. The truncatedomp13- and omp15-containing frameshift mutations differedfrom the aligned proteins by showing large insertions in thecentral part of the GGAI domain (data not shown). In the C-

terminal part of the molecules, tryptophan residues were con-served, and in all cases where full-length molecules were pre-dicted, the last aa was phenylalanine.

Analysis of GGAI protein structures. Two GGAI proteins(Omp4 and Omp5) were predicted to have a leader sequencewith a cleavage site specific for signal peptidase II (at aa 15and 17, respectively; figure 1) and are therefore potential lip-oproteins [11]. Seventeen genes were predicted to have a leadersequence with a cleavage site specific for signal peptidase I (forOmp10 and 11 at aa 26); for 2 genes (pmp16 and 18), no suchcleavage site was predicted. Thus, 19 of the 21 hypotheticalproteins had the capacity to pass the chlamydial membrane.

To analyze for proteins with homology to the GGAI Chla-mydia proteins, we searched databases using the FINDPAT-TERN program for proteins containing x GGAI (several times)or the motifs FxxN and GGA[ILV] separated by 4 to 14 non-conserved aa. The search revealed homology to membrane pro-teins of an additional 4 bacterial species. The proteins are theRompA protein of several Rickettsia species whose function isassociated with adhesion, Bordetella pertussis adhesin FHAB,Helicobacter pylori HP1288, and E. coli Yfal proteins of un-known function. The search revealed 2 surface-exposed proteinsbut the mode by which the proteins were translocated to thesurface was unknown.

To analyze for potential features that predict how GGAIcould be incorporated into the chlamydial outer membrane, acomputer analysis was performed. The Chou-Fassmann andGarnier-Osguthorpe-Robson predictions did not reveal anyspecific structures. We then analyzed the C-terminal part of theGGAI proteins by the computer program AMPHI. This pro-

Figure 1. Alignment of deduced amino acid sequences of GGAI proteins Omp4, 5, 10, and 11. Identical amino acids are shown in reverse.

S532 Christiansen et al. JID 2000;181 (Suppl 3)

Figure 2. Immunoblotting of SDS-PAGE–separated proteins fromC. pneumoniae elementary bodies. After transfer of proteins to nitro-cellulose membrane, membrane was cut in strips and reacted with hy-perimmune sera from rabbits immunized with purified recombinantOmp4–7, 9–11, 13, and 15 (lanes are labeled by anti-Omp antibodyused). No. 198 is polyclonal antibody directed against C. pneumoniaeouter membrane complex reacting with major outer membrane protein(MOMP), Omp2, and 95- to 100-kDa protein bands.

gram predicted that the C-terminal part of the GGAI proteinswould produce a C-terminal transmembranic b-barrel consist-ing of 14 strands. A further indication that the GGAI proteinsare transmembranic is that all contained conserved tryptophansand a phenylalanine residue as the most terminal aa and thepresence of a potential amphipathic b-sheet with hydrophobicresidues at positions 1, 3, 5, 7, and 9 from the C-terminal aa(figure 1) [25]. The presence of such a potential amphipathicb-sheet localized in the C-terminal part of the protein is char-acteristically present in outer membrane proteins of gram-neg-ative bacteria [25].

Expression of GGAI proteins. Expression of 9 GGAI pro-teins was studied by immunoblotting of SDS-PAGE–separatedproteins from purified C. pneumoniae EBs reacted with hyper-immune serum from rabbits immunized with purified recom-binant proteins (rOmp4–7, 9–11, 13, and 15). Results are shownin figure 2. Bands of 95–100 kDa were observed when anti-bodies to Omp4–7 and 9–11 were used (lanes numbered ac-cording to the rOmp proteins, respectively), while antibodiesto Omp13 and 15 did not react. The 43-kDa band observed inlane 10 was also detected in serum obtained prior to immu-nization of the rabbit (results not shown). All bands seen inthe immunoblots had the size expected from the deduced aasequence except for the band in lane 9. According to the genesequence, Omp9 contained a frameshift mutation leading to apremature stop at 59 kDa (table 1). A band of this size wasnot observed. Instead reaction with a full-length protein wasseen. This indicated that in some C. pneumoniae EBs, the mu-tation may be repaired, leading to a full-length protein. A dif-ferent explanation for the anti-Omp9 antibody reaction couldbe that this antibody recognized a common epitope and thuscould recognize different GGAI proteins as seen in C. psittaci[15]. The truncated Omp13 and 15 of 54 and 48 kDa, respec-tively, could not be detected by immunoblotting.

Expression of the GGAI proteins in C. pneumoniae–infectedHEp-2 cells was further analyzed by MIF (figure 3). Antibodiesto Omp4–7 and 9–11 (see figure 3A) showed a bright fluores-cence staining of the C. pneumoniae inclusions independent oftheir size, but anti-Omp13 and 15 antibodies did not react (fig-ure 3B). These results thus mirror those obtained byimmunoblotting.

2-D PAGE and immunoblotting. Of the 8t Omp proteinsfound to be expressed by immunoblotting and MIF, 7 weresimilar in size when analyzed by immunoblotting of SDS-PAGE–separated C. pneumoniae proteins. The pI of the GGAIproteins varied, however (table 1). To further document whichof the Omp proteins were expressed, 2-D PAGE and immu-noblotting were performed. After separation on IPG strips andSDS-PAGE, the 35S-labeled C. pneumoniae proteins were trans-ferred to PVDF membranes and reacted with antibodies to therecombinant GGAI proteins. In figure 4A, the blot was reactedwith anti-Omp4 antibodies. Strongly reacting spots were seenat the expected size and pI (arrows). To verify the localization

of Omp4 at the C. pneumoniae proteome, an x-ray film wasplaced on top of the developed immunoblot membrane (figure4B). The arrows indicate the position of the strongly reactingspots. In addition to the labeled Omp4 spots, tracks of spotsover most of the 100-kDa pI 4–7 range could be seen. To com-pare the migration of C. pneumoniae Omp4 with rOmp4 syn-thesized in E. coli, 35S-labeled rOmp4 was analyzed by 2D-PAGE (figure 4C). Spots migrating similarly to the C.pneumoniae Omp4 are marked with arrows. Addition of 35S-labeled rOmp4 to the C. pneumoniae proteins revealed that therecombinant spots co-localized with those identified by im-munoblotting (figure 4D, arrows). By similar analyses, it waspossible to identify the expression and localization of Omp5and Omp11. It was characteristic of the reaction pattern thateach antibody recognized a track of spots of identical size andlocalized within a narrow pI range. Also, in E. coli, rOmp4 wasexpressed as a track of spots (figure 4C). This reaction patternwas similar to that found when monoclonal antibodies directedagainst the C. psittaci putative (P) OMP proteins that are ho-mologous to the Omp/Pmp proteins of C. pneumoniae wereanalyzed by 2-D PAGE and immunoblotting [26]. Antibodiesto rOmp6, 7, and 10 each recognized several tracks of spots,indicating the potential presence of a common epitope in agree-ment with that observed for C. psittaci POMP proteins [26].The 35S-labeled proteins seen in figure 4B indicated that many

JID 2000;181 (Suppl 3) C. pneumoniae Surface Proteins S533

Figure 3. Immunofluorescence microscopy. HEp-2 cell monolayers infected with C. pneumoniae for 72 h were methanol fixed and reacted withK203 against Omp5 (A) and K202 against Omp13 (B).

of the GGAI proteins of about 100 kDa may be expressed.Determination of the peptide content in each spot by massspectroscopy analysis will identify their exact origin [27].

Characterization and reaction of patient sera compared withreaction of K151 and K150 in immunoblotting. After deter-mining the expression of several GGAI proteins, we attemptedto learn whether they were recognized by the human humoralimmune response. C. pneumoniae–seropositive sera do not showstrong reactivity in immunoblotting [10, 28] and react variablywith a 98-kDa protein band [10, 28]. Also, Omp4 and 5 migratedifferently in SDS-PAGE with or without boiling of samplesprior to electrophoretic separation of the proteins and thischanges the reactivity with antibodies [11]. To determine thismigration, antibodies obtained by immunizing rabbits with C.pneumoniae OMC with or without SDS denaturation of theantigen prior to immunization (K150 and K151, respectively)were used in immunoblotting of C. pneumoniae EB proteins.In figure 5A, K150 raised against SDS-denatured C. pneumon-iae OMC is shown to react with the predicted bands of MOMP,Omp2, and the 98-kDa doublet (lane 2). With no boiling ofthe C. pneumoniae EB proteins before separation by SDS-PAGE, MOMP and Omp2 did not enter the gel when probedwith monospecific antibody (data not shown), and the 98-kDaprotein bands migrated more slowly as previously described[11]. K151 only reacted with the more slowly migrating bands(figure 5B). Immunoblotting of C. pneumoniae EB with patient

sera showed no specific major immunogen [10, 28] but 53-,60-, and 98-kDa bands were frequent.

Since K151 reacted differently than K150 in immunoblottingand differently regardless of whether the samples were boiledbefore SDS-PAGE (figure 5), we wondered if C. pneumo-niae–positive patient sera would react similarly. Results areshown in figure 6. The 6 sera from C. pneumoniae MIF-positivepatients had a fairly uniform reaction pattern (figure 6A) com-pared with samples boiled before SDS-PAGE (figure 6B) whereindividual sera had different reaction patterns as previouslydescribed [10, 28]. For control, 3 negative sera were included(lanes 7–9). They reacted weakly with MOMP and the 98-kDaprotein bands. Thus, in immunoblotting of C. pneumoniae pro-teins prepared without boiling before separation by SDS-PAGE, patient sera recognized a pattern of bands with migra-tion similar to that observed with K150 (figure 6A, lane 10).This reaction may indicate but not prove that Omp proteinsare recognized by the C. pneumoniae–positive human sera andthat the epitopes are destroyed by boiling of the samples priorto separation of the proteins by SDS-PAGE.

EM. The immunoblotting results indicate a possible cor-relation between MIF-reacting proteins and the unboiled (andthus not completely unfolded) proteins seen in figures 5 and 6.Since patient sera did not react with purified recombinantGGAI proteins (data not shown), it was thought that theseproteins did not fold correctly when produced in E. coli. In an

S534 Christiansen et al. JID 2000;181 (Suppl 3)

Figure 4. Two-dimensional (2-D) gel of 35S methionine/cysteine (met/cys)–labeled C. pneumoniae proteins. A, Gel transferred to polyvinylidenedifluoride (PVDF) membrane and reacted with K195 (anti-Omp4). B, Autoradiography of A. C and D, autoradiographs of PVDF membrane,2-D gel of 35S met/cys-labeled recombinant Omp4 produced in Escherichia coli, and labeled C. pneumoniae proteins (as in A) with addition ofrecombinant Omp4 (as in C). Arrows point to Omp4 (theoretical pI, 5.9).

Figure 5. Immunoblots of C. pneumoniae elementary body (EB)proteins reacted with polyclonal antibodies K150 raised against purifiedSDS-denatured C. pneumoniae outer membrane complex (OMC) pro-teins (A) and K151 raised against undenatured solubilized C. pneu-moniae OMC (B). Lanes 1 and 2, respectively, unboiled and boiled EBsamples applied for electrophoresis. Far left, standard markers in kDa.

effort to produce correctly folded recombinant GGAI proteins,we constructed a hybrid protein between a known b-bar-rel–forming protein (AIDA) from E. coli [29] and the GGAIpart from Omp4. The protein was stably expressed upon in-duction in E. coli and reacted in immunoblotting with the anti-Omp4 antibody (data not shown). To analyze whether the hy-brid protein was transported to the surface of E. coli, immunoEM with K151 was done. The reaction is shown in figure 7Aand the control, the same plasmid-free E. coli, in figure 7B.There was no surface labeling in the control, whereas the surfaceof the recombinant E. coli expressing the hybrid protein wasuniformly labeled by immunogold, indicating that the N-ter-minal part of Omp4 was exported to the surface of E. coli whereit reestablished the natively folded form. Similar results wereobtained with patient sera (figure 7C and 7D). These resultsmay be of importance for further understanding of the correctlyfolded structure of the C. pneumoniae outer membrane proteins.

Discussion

In this study, we demonstrated that at least 3 of the GGAIproteins, Omp4, Omp5, and Omp11, were expressed duringcultivation of C. pneumoniae in HeLa and HEp-2 cells. Threegene products could be identified by 2-D PAGE and immu-noblotting. By immunoblotting and immunofluorescence anal-ysis of C. pneumoniae–infected HEp-2 cells, we found that 2proteins were not expressed: the truncated gene product ofomp13 and omp15 containing a frameshift mutation in the cen-

tral part of the protein. The gene product of omp8 could notbe produced by the recombinant technique—probably due toinstability caused by the stop mutation in the N-terminal partof the protein. The proteome map produced by 2-D PAGE of35S-labeled C. pneumoniae proteins showed enough resolution

JID 2000;181 (Suppl 3) C. pneumoniae Surface Proteins S535

Figure 6. Immunoblot of C. pneumoniae elementary body proteins reacted with 9 human serum samples. A, Proteins separated by SDS-PAGEwithout boiling. B, Proteins boiled for 5 min before SDS-PAGE separation. Proteins from both gels were transferred to nitrocellulose membranes,cut in strips, and reacted with human serum samples diluted 1:200 (lanes 1–9). Serum samples 1–6 were positive by microimmunofluoresence;7–9 were negative. In lane 10, strips were reacted with K150 antibodies diluted 1:1000.

to identify individual spots and to overlay the spots obtainedfrom recombinant GGAI proteins produced in E. coli. It wasunexpected that both the immunoblots and radioactive labelingof the recombinant proteins showed tracks of spots. Such tracksmay indicate the presence of posttranslational modifications.The findings are in agreement with those described by Gian-nikopoulou et al. [26], who showed a similar reaction patternof the Pomp proteins in C. psittaci. Their findings support theresults we describe. Even though the potential posttranslationalmodifications have not yet been determined, Giannikopoulouet al. [26] suggested by use of the Pro-site program that sincesignatures for glycosylation and phosphorylation were presentin all the C. psittaci pomp genes that the proteins could beposttranslationally modified. Since the proteins did not changesize (as observed within the resolution of the proteome map),it is predominantly the charge and not the mass of the proteinthat is changed and thus only modifications involving chargechanges may have occurred. Many spots were seen in the regionaround 100 kDa. Identification of each spot will be revealedby mass spectrometry [27], which will unambiguously determinetheir identity.

Serodiagnosis of C. pneumoniae infections have mostly beendone with the MIF technique. To replace this assay with anautomated ELISA requires that correctly folded epitopes arepresent for the antibodies. Since no antibody-binding proteinon the C. pneumoniae surface has been identified, a useful an-tigen has not yet been obtained and efforts using recombinantproteins have so far been unsuccessful (unpublished data). Since

the recombinant Omp proteins did not react with seropositivepatient sera, a different approach was taken. The use of the N-terminal part of the GGAI proteins fused with the C-terminalpart of a known E. coli outer membrane protein, the b-bar-rel–forming protein AIDA [29], and use of this transporterprotein to carry the N-terminal part of the C. pneumoniae Omp4to the surface of E. coli enabled us to test whether the N-terminal part of Omp4 could be recognized by antibodies. Thiswas the case as shown in figure 7. Identification of the N-terminal part of the GGAI protein to be recognized by theantibodies is in agreement with the results of Longbottom etal. [18] who, by immuno EM analysis with antibodies againstthe GGAI family member OMP90A from C. psittaci, dem-onstrated that only the N-terminal part of the protein is surfaceexposed.

With the demonstration of several expressed C. pneumoniaeGGAI proteins and the recombinant hybrid protein techniquepresented here, it may be possible to use this knowledge toproduce an antigen useful in ELISA. By dissecting the antigensrecognized by the human humoral immune response it thusmay be possible to identify immunogenic proteins. This knowl-edge can then be used in the search for protective antigens.

The requirement for a vaccine is to be able to mediate pro-tection against infections. For C. pneumoniae infections, nosuch components have been identified. In a mouse model, weshowed that both Omp4 and Omp5 are expressed in C. pneu-moniae inclusions in bronchial epithelial cells [30]. Whether im-munity to such infections may be obtained by vaccinating the

S536 Christiansen et al. JID 2000;181 (Suppl 3)

Figure 7. Immunoelectron microscopy of recombinant hybrid-Omp4 Escherichia coli (A and C) and control plasmid-free E. coli (B and D)were reacted with K151 (A and B) and serum from patient 1 diluted 1:50 (C and D). Bar = 0.5 mm.

mice with the recombinant GGAI proteins remains to be de-termined. It is, however, encouraging that it now seems possibleto obtain natively folded surface-exposed recombinant C. pneu-moniae outer membrane protein epitopes, thereby having a newtool for analysis of the structure and composition of the C.pneumoniae surface.

Acknowledgments

We thank Inger Andersen, Karin Skovgaard Sørensen, and LisbetW. Pedersen for excellent technical support and M. Alexander Schmidt

for providing the plasmid pIB264 that was used for amplification ofthe AIDA b-barrel.

References

1. Grayston JT. Infections caused by Chlamydia pneumoniae strain TWAR. ClinInfect Dis 1992;15:757–63.

2. Kuo CC, Jackson L, Campbell LA, Grayston JT. Chlamydia pneumoniae(TWAR). Clin Microbiol Rev 1995;8:451–61.

3. Schachter J. Diagnosis of human chlamydial infections. In: Proceedings 9thSymposium on Human Chlamydial Infections. 1998:577–86.

JID 2000;181 (Suppl 3) C. pneumoniae Surface Proteins S537

4. Saikku P, Matilla K, Nieminen MS, Leinonen M, Ekman MR, Makela PH,Valtonen V. Serological evidence of an association of a novel ChlamydiaTWAR with chronic coronary heart disease and acute coronary infarction.Lancet 1988;2:983–6.

5. Birkelund S, Lundemose AG, Christiansen G. Immunoelectron microscopicanalysis of LPS in Chlamydia trachomatis. Infect Immun 1989;57:3250–3.

6. Christiansen G, Østergaard L, Birkelund S. Analysis of the Chlamydia pneu-moniae surface. In: Chlamydial infections, proceedings of the Eighth In-ternational Symposium on Human Chlamydial Infections. 1994:173–6.

7. Christiansen G, Østergaard L, Birkelund S. Molecular biology of the Chla-mydia pneumoniae surface. Scand J Infect Dis Suppl 1997;104:5–10.

8. Caldwell HC, Kromhout J, Schachter J. Purification and partial character-ization of the major outer membrane protein of Chlamydia trachomatis.Infect Immun 1981;31:1161–76.

9. Christiansen G, Madsen AS, Knudsen K, Mygind P, Birkelund S. Stabilityof the outer membrane proteins of Chlamydia pneumoniae. In: Proceedings9th Symposium on Human Chlamydial Infections. 1998:271–4.

10. Campbell LA, Kou CC, Wang SP, Grayston JT. Serological response toChlamydia pneumoniae Infection. J Clin Microbiol 1990;28:1261–4.

11. Knudsen K, Madsen AS, Mygind P, Christiansen G, Birkelund S. Identifi-cation of two novel genes encoding 97- to 99-kilodalton outer membraneproteins of Chlamydia pneumoniae. Infect Immun 1999;67:375–83.

12. Stephens R. http://chlamydia-www.berkeley.edu:423113. Kalman S, Mitchell W, Marathe R, et al. Comparative genomes of Chlamydia

pneumoniae and C. trachomatis. Nat Genet 1999;21:385–9.14. Stephens RS, Kalman S, Lammel C, et al. Genome sequence of an obligate

intracellular pathogen of humans: Chlamydia trachomatis. Science 1998;282:754–9.

15. Longbottom D, Russell M, Jones GE, Lainson FA, Herring AJ. Identificationof a multigene family coding for the 90-kDa proteins of the ovine abortionsubtype of Chlamydia psittaci. FEMS Microbiol Lett 1996;142:277–81.

16. Longbottom D, Russell M, Dunbar SM, Jones GE, Herring AJ. Molecularcloning and characterization of the genes coding for the highly immu-nogenic cluster of 90-kilodalton envelope proteins from the Chlamydiapsittaci subtype that causes abortion in sheep. Infect Immun 1998;66:1317–24.

17. Cevenini R, Donati M, Brocchi E, De Simone F, La Placa M. Partial char-acterization of an 89-kDa highly immunoreactive protein from Chlamydiapsittaci A/22 causing ovine abortion. FEMS Microbiol Lett 1991;65:111–5.

18. Longbottom D, Findlay J, Vretou E, Dunbar AM. Immunoelectron micro-scopic localization of the OMP90 family on the outer membrane surfaceof Chlamydia psittaci. FEMS Microbiol Lett 1998;164:111–7.

19. Clausen JD, Christiansen G, Holst HU, Birkelund S. Chlamydia trachomatis

utilizes the host cell microtubule network during early events of infection.

Mol Microbiol 1997;25:441–9.

20. Jahnig F. Structure predictions of membrane proteins are not that bad. Trends

Biochem Sci 1990;15:93–5.

21. Bjellqvist B, Hughes GJ, Pasquali C, et al. The focusing positions of poly-

peptides in immobilized pH gradients can be predicted from their amino

acid sequences. Electrophoresis 1993;14:1023–31.

22. Birkelund S, Lundemose AG, Christiansen G. Characterization of native and

recombinant 75-kilodalton immunogens from Chlamydia trachomatis se-

rovar L2. Infect Immun 1989;57:2683–90.

23. Shaw A, Christiansen G, Birkelund S. Effects of interferon gamma on Chla-

mydia trachomatis serovar A and L2 protein expression investigated by

two-dimensional gel electrophoresis. Electrophoresis 1999;20:775–80.

24. Christiansen G, Birkelund S. Transmission electron microscopy and immu-

nogold staining of mycoplasmal surface antigens. In: Miles R, Nicholas,

R, eds. Mycoplasma protocols. Totowa, NJ: Humana Press, 1998:309–18.

25. Struyve M, Moons M, Tommassen J. Carboxy-terminal phenylalanine is

essential for the correct assembly of a bacterial outer membrane protein

J Mol Biol 1991;218:141–8.

26. Giannikopoulou P, Bini L, Simitsek PD, Pallini V, Vretou E. Two-dimensional

electrophoretic analysis of the protein family at 90 kDa of abortifacient

Chlamydia psittaci. Electrophoresis 1997;18:2104–8.

27. Shaw AC, Larsen MR, Roepstorff P, Holm A, Christiansen G, Birkelund S.

Mapping and identification of HeLa cell proteins separated by immobi-

lized pH-gradient two-dimensional gel electrophoresis and construction

of a 2D-PAGE database. Electrophoresis 1999;20:977–83.

28. Kutlin A, Roblin PM, Hammerschlag MR. Antibody response to Chlamydia

pneumoniae infection in children with respiratory illness. J Infect Dis

1998;177:720–4.

29. Benz I, Schmidt A. AIDA-1, the adhesin involved in diffuse adherence of

the diarrheogenic Escherichia coli strain 2787 (O126:H27) is synthesized

via a precursor molecule. Mol Microbiol 1992;6:1539–46.

30. Christiansen G, Boesen T, Hjernø K, et al. Molecular biology of Chlamydia

pneumoniae surface proteins and their role in immunopathogenicity. Am

Heart J 1999;138:S491–5.