The putative outer membrane location of the OMP90 (formerly POMP) family from the ovine abortion strain of Chlamydia psittaci was investigated by immunoelectron microscopy. Using a non-embedding technique, antigens were shown to be localised on the outer membrane surface of both elementary and reticulate bodies, the infectious and non-infectious forms of Chlamydiae respectively. Antibodies affinity-purified against the expressed amino- and carboxy-terminal halves of one of the family members, OMP90A, demonstrated that the amino half is surface-exposed while the carboxyl half is most probably localised internally. Surface localisation on elementary bodies indicates the importance of these proteins as protective antigen candidates.
The Chlamydia spp. are obligate, intracellular Gram-negative bacteria responsible for a broad spectrum of disease in both humans and animals [1–3]. The bacteria undergo a unique life-cycle consisting of two predominant developmental forms: the infectious elementary body (EB) and the non-infectious, metabolically active reticulate body (RB). The infection cycle starts with the attachment and uptake of the small EB (approx. 0.3 μm diameter) by the target host epithelial cell. After gaining access to the intracellular environment within a phagocytic vacuole, the EB is able to inhibit phagosome-lysosome fusion and convert to the much larger RB (0.5–1.3 μm diameter) which undergoes division by binary fission. The duration of the cycle varies from approx. 48 to 72 h depending on species, growth conditions, etc. In the latter part of the cycle RBs reorganise back into EBs and are released into the extracellular environment ready to infect other cells.
Evidence suggests that protein components exposed on the EB outer membrane surface are crucial for the successful infection of host cells [4, 5] and are therefore appropriate targets for vaccine development and molecular/functional analysis. However, many chlamydial antigens which elicit an immune response still remain uncharacterised and the major outer membrane protein (MOMP) remains the only characterised antigen unequivocally demonstrated to be exposed on the EB outer membrane surface [6, 7]. Recently, genes coding for a group of highly immunogenic envelope proteins of approximately 90 kDa, the putative outer membrane proteins (POMPs), from the ovine enzootic abortion (OEA) subtype of C. psittaci have been identified, isolated and characterised [8, 9]. OEA is a major cause of abortion in sheep in several European countries, and represents a proven zoonotic risk to pregnant women in whom infection can be life-threatening . The POMPs have been extensively analysed by 2-D electrophoretic analysis  and have been suggested to be important serodiagnostic antigen candidates [9, 12]. They may also play a role in protective immunity since they are present in chlamydial outer membrane complexes [9, 13, 14] which have been shown to be protective . However, such a role is not supported by the recent results of Buendıća et al.  who failed to demonstrate localisation to the EB surface.
The aim of this study was to resolve the conflicting results on the surface localisation of this family of proteins by electron microscopy, using both serotype 1- and subspecies-specific monoclonal antibodies and using affinity-purified antibodies to one of the recombinant POMPs, to determine their suitability for further studies as potential vaccine candidates.
Materials and methods
Cell culture, infection and Chlamydia purification
The OEA C. psittaci strain S26/3 was grown in McCoy cells . Infected cells were harvested using sterile glass beads at 48 h post infection, and EBs/RBs purified by density gradient centrifugation through urografin, as described previously . Purified EBs/RBs were stored in 20 mM Tris, pH 7.4/0.2 M KCl at −70°C until use.
McCoy cells, grown on sterile plastic coverslips in trac bottles (Bibby Sterilin Ltd, Stone, UK) and in 225-cm3 tissue culture flasks for post-embedding and cryo-sectioning, respectively, were infected with OEA C. psittaci as previously . At 48 h post infection coverslips and cells harvested from flasks were washed with phosphate-buffered saline, pH 7.4 (PBS). Coverslips were air-dried and stored at −70°C until use, whereas pelleted cells were used immediately.
Monoclonal and affinity-purified antibodies
The POMP (181 and 0040) and MOMP (4/11) monoclonal antibodies (mAbs) used in this study have been described previously [9, 14, 17]. Affinity-purified antibodies to the recombinant amino- and carboxy-terminal halves of POMP90A  were prepared as described previously , with the following modifications. Purified recombinant protein (30-μg aliquots)  was subjected to preparative sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions. Protein was transferred to nitrocellulose, and strips containing the recombinant amino and carboxyl fragments were excised and incubated with pooled post-abortion sheep sera [8, 9] and sera from ewes vaccinated with the purified recombinant carboxy-terminal protein , respectively. The optimal dilution for each antibody used (ranged from 1/2 for affinity-purified antibodies to 1/200 for mAbs) was that which showed maximum specific labelling and minimum background. Antibody and conjugate negative controls for non-specific binding were provided by an anti-Pasteurella haemolytica leukotoxin mAb  and Tris-buffered saline, pH 7.4 (TBS)/1% normal donkey serum (NDS), respectively.
Immunoblotting of EB extracts
Immunoblotting of whole EB extracts (using EBs purified at 48 h post infection) with the anti-POMP mAbs, affinity-purified antibodies, and control mouse and sheep Abs was as described previously [8, 9].
Immunoelectron microscopy: non-embedding technique
Purified and unfixed EBs/RBs were immunolabelled by a modification of the method of Collett et al. , essentially as described by Salinas et al. . Briefly, carbon-formvar-coated 200-mesh copper or gold grids (Agar Scientific Ltd, Stansted, UK) were placed on 10-μl droplets of the purified chlamydial suspension described in Section 2.1 and incubated for 10 min at 37°C. Grids were then floated on a 60-μl droplet 2% bovine serum albumin (BSA)/1% NDS/TBS for 10 min to block non-specific binding. After washing twice on droplets (60 μl) of TBS, grids were placed on 60-μl droplets of the appropriate antibody in 1% BSA/1% NDS/TBS for 30 min at 37°C, washed five times in TBS and placed on a 30-μl droplet of either goat anti-mouse or donkey anti-sheep IgG conjugated to 15-nm gold particles (British BioCell International Ltd, Cardiff, UK) diluted 1/20 in 1% BSA/1%NDS/TBS for 30 min at 37°C. Grids were finally washed a further four times in TBS, four times in distilled water and air-dried before observing in a transmission electron microscope (TEM) (Philips CM120 BioTwin).
Immunoelectron microscopy: post-embedding technique
S26/3-infected McCoy cells grown on coverslips were fixed in 2% paraformaldehyde/0.5% glutaraldehyde/0.1 M PBS for 90 min at 4°C. After washing in PBS they were dehydrated in ethanol, infiltrated in LR White resin (London Resin Co., Basingstoke, UK) and polymerised under vacuum at 50°C for 24 h (no initiator was added) . Immunolabelling was carried out on ultrathin sections (cut from blocks on a Reichert Jung Ultracut) mounted on carbon-formvar 200-mesh copper grids, as described in Section 2.4. Sections were counterstained in uranyl acetate/lead citrate for examination in the TEM.
Immunoelectron microscopy: cryo-sectioning
S26/3-infected McCoy cell pellets (Section 2.1) were fixed in 2% paraformaldehyde/0.5% glutaraldehyde/0.1 M PBS for 30 min and prepared for cryo-sectioning as follows, based on the method of Tokuyasu . Cells were suspended in 4% gelatin/PBS at 30°C, centrifuged at 1300×g for 10 min at room temperature and placed on ice to solidify the gelatin. Embedded cell pellets were cut into small 1-mm3 cubes, fixed again as described above, and washed in PBS. Cubed cell pellets were infiltrated with 2.3 M sucrose/PBS to act as cryoprotectant and the preparation stored at 4°C until use. Fixation prior to sectioning was achieved by plunge-freezing the material, mounted in a droplet of 2.3 M sucrose, in liquid nitrogen. Ultrathin sections were cut using a Leica UCT ultramicrotome with FCS low temperature sectioning system (on loan from Leica) and collected on carbon-formvar-coated 200-mesh gold grids. Immunogold labelling was as described in Section 2.4, except sections were counterstained in uranyl acetate before being embedded in a 2% polyethylene glycol/0.1% methyl cellulose solution and examined in the TEM.
Results and discussion
When the non-embedding technique was performed on purified Chlamydia with the subspecies- and serotype 1-specific anti-POMP mAbs, 0040 (Fig. 1A,B) and 181 (Fig. 1C,D) respectively, labelling was observed on both EBs and RBs. Similar, but greater, labelling was observed with the anti-MOMP mAb 4/11 (results not shown). The level of immunolabelling shown in Fig. 1, calculated at approx. 400 gold particles per μm2 for EBs (B and D) and 70/140 gold particles per μm2 for RBs (A and C, respectively), was representative of over 80% of the Chlamydia, observed on two grids on at least three separate occasions for each mAb. Lighter staining was observed over 10–20%, and less than 10% showed little staining. Non-specific staining was not observed with conjugate controls or when the non-specific primary (anti-leukotoxin) mAb was used.
None of the mAbs tested resulted in immunolabelling when the post-embedding technique was used. This was presumably due to loss of antigenicity resulting from glutaraldehyde and paraformaldehyde fixation. In view of this, immunolabelling of cryo-sections was attempted in which antigenicity was better preserved. However, morphology was not as well preserved by this method, when compared to post-embedding, and so it was difficult to determine the specific location of antibody binding. Regardless of this, both the anti-POMP mAb 0040 (Fig. 2) and the anti-MOMP mAb 4/11 (not shown) showed labelling in close association with the outer membrane surface of both RBs and EBs, in agreement with the labelling observed using the non-embedding technique. No labelling was observed with the negative control anti-leukotoxin mAb (not shown).
Recently Buendıća et al.  failed to demonstrate the localisation of the POMPs on the surface of EBs, which does not agree with the results we have presented here. However, they used different mAbs, presumably recognising different epitopes to those used in this study. Then the lack of neutralisation observed with their mAbs [6, 21] coupled with the lack of surface exposure on EBs but not on RBs  could simply be due to epitopes being hidden during the condensation of RBs back into EBs. This difference in EB localisation may be resolved in the future with neutralisation studies using our panel of mAbs.
Although our results demonstrate the surface exposure of POMP epitopes on the EB this in no way discounts the possibility that these antigens also have epitopes exposed on the periplasmic surface of the outer membrane, as was suggested by Buendıća et al. . Previously, we have hypothesised that the amino-terminal half of POMP90A is exposed on the chlamydial outer membrane surface, and that the carboxyl half was localised internally with epitopes possibly exposed on the periplasmic surface and/or partially exposed on the outer membrane surface but inaccessible to antibody . We attempted to confirm or discount this hypothesis by immunoelectron microscopic analysis using affinity-purified Abs to the expressed amino- and carboxy-terminal halves of POMP90A. These monospecific antisera recognise the 90-kDa POMPs upon immunoblotting of S26/3 EBs, as judged by comparison of their reactivities with those of both the serotype 1- and subspecies-specific anti-POMP mAbs 181 and 0040, respectively (Fig. 3; lanes 1–4). Immunoreactivity with positive control pooled post-abortion sera from experimentally infected ewes, as described in Longbottom et al. , and with the anti-MOMP mAb 14/11 is shown in Fig. 3 (lanes 5 and 6). Reactivity with other positive and negative control sheep and mouse Abs was as previously shown [8, 9]. When the non-embedding technique was performed with affinity-purified Abs to the amino fragment, labelling was again observed on both RBs and EBs (Fig. 4A,B). The level of immunolabelling was much higher than observed with the mAbs, calculated at approx. 380 gold particles per μm2 for RBs (A) and 950 for EBs (B), but was still representative of over 80% of the Chlamydia, observed on two grids on at least three separate occasions. In contrast to these results, the level of immunolabelling with affinity-purified Abs to the carboxyl fragment was much reduced. Three levels of labelling were consistently observed on a particular grid: no labelling on 60–65% of Chlamydia; a low level of labelling on 25–30%; and a high level on 10–15%. Each of these different labelling patterns were not clustered but dispersed randomly throughout each grid and were observed on both RBs and EBs. Fig. 4 C illustrates examples of the three levels of labelling observed on EBs in close proximity on a grid. This labelling was judged to be non-specific since the same labelling pattern was also observed with the donkey anti-sheep IgG gold conjugate control (i.e. no primary antibody). This labelling pattern, which was also observed using a different source of conjugate and using rabbit anti-sheep IgG in combination with protein A-gold, was probably due to cross-reaction between the donkey anti-sheep IgG gold conjugate and contaminant antigens from the foetal calf serum which were present in the growth media. Low levels of such antigens are known to be present in preparations of Chlamydiae. Therefore, the labelling observed with the affinity-purified Abs does support the model previously proposed by us . However, neither of the affinity-purified Abs produced any immunolabelling when the post-embedding or cryo-sectioning techniques were used which means that we cannot confirm that epitopes are also exposed on the periplasmic surface.
In view of the results presented previously  and in this study, showing that the 90-kDa proteins are located in the outer membrane, we propose that the root of the names for the genes and corresponding proteins is changed from pomp and POMP to omp and OMP, respectively. Since members of the OMP90 family are clearly localised on the surface of EBs they remain potential vaccine candidates, possibly being involved in key stages of the infection process. The recent identification of protein homologues in the human pathogen C. trachomatis serovar D by the Chlamydia Genome Project (on the World Wide Web at URL http://chlamydia-www.berkeley.edu:4231/index.html) also adds to the importance of studying these proteins further for any role that they may play in protective immunity.
The authors thank Miss M. Livingstone for cell culture, Dr W. Donachie for the anti-leukotoxin antibody, and Mr B.J. Easter for photography. We also thank Dr C.E. Jeffree, University of Edinburgh, Science Faculty EM Facility for advice with the EM and are particularly grateful to Mr K. Jacobson from Leica U.K. Ltd, Milton Keynes for the loan of the ultramicrotome. This work was supported by the Scottish Office Agriculture, Environment and Fisheries Department.