The expression, processing and localization of polymorphic membrane proteins in Chlamydia pneumoniae strain CWL029

Background Chlamydiae are obligate intracellular bacteria, which are important human pathogens. Genome sequences of C. trachomatis and C. pneumoniae have revealed the presence of a Chlamydia specific gene family encoding polymorphic outer membrane proteins, Pmps. In C. pneumoniae the family comprises twenty-one members, which are all transcribed. In the present study, the expression, processing and localisation of the sixteen full-length Pmps in C. pneumoniae strain CWL029 have been further investigated by two-dimensional gel electrophoresis and immunofluorescence microscopy. Results Ten Pmps were identified in elementary bodies (EBs). Eight of these were investigated with respect to time dependent expression and all were found to be up-regulated between 36 and 48 hours post infection. Antibodies against Pmp6, 8, 10, 11 and 21 reacted with chlamydiae when infected cells were formalin fixed. Pmp6, Pmp20 and Pmp21 were found in cleaved forms, and the cleavage sites of Pmp6 and Pmp21 were identified. Conclusions The Pmps are heavily up-regulated at the time of conversion of RB to EB, and at least ten Pmps are present in EBs. Due to their reaction in formalin fixation it is likely that Pmp6, 8, 10, 11 and 21 are surface exposed. The identified cleavage sites of Pmp6 and Pmp21 are in agreement with the theory that the Pmps are autotransporters.


Background
Chlamydiae are pathogenic gram-negative bacteria of which C. pneumoniae causes upper and lower respiratory tract infections in humans [1]. Going through a developmental cycle the chlamydiae alternate between infective elementary bodies (EBs) and replicative reticulate bodies (RBs) [2]. The bacteria are obligate and intracellular, residing inside a specialized phagosome, named the chlamydial inclusion. The duration of the developmental cycle for C. pneumoniae cultivated in cell culture is about 72 hours [3].
The C. pneumoniae CWL029 genome sequence revealed the presence of a gene family, the pmp family, consisting of 21 members [4] that were paralogous to the pmps found in C. trachomatis [5] and C. psittaci [6][7][8][9]. The C. psittaci Pmps have been analysed by two-dimensional electrophoresis in an earlier study [37]. The Pmps are two-domain proteins with similarity to autotransporter proteins [4,10,11]. They are characterized by a high frequency of the two sequences FxxN and GGAI in the N-terminal part (twelve and seven repeats on average, respectively) [4], and their C-terminal part shows the characteristics of a βbarrel [10]. The GGAI motif and the prediction of a C-terminal β-barrel suggest that the Pmps are autotransporters, transporting an N-terminal passenger domain through a pore formed by their C-terminal part [10,12]. Seventeen C. pneumoniae Pmps contain a signal peptidase I cleavage site directing transport over the inner membrane, and two Pmps contain a signal peptidase II cleavage site suggesting lipid modification [4].
The Mw of most Pmps in C. pneumoniae strain CWL029 is predicted to be just below 100 kDa, but three are larger: Pmp6 is 142 kDa, Pmp20 is 178 kDa and Pmp21 is 167 kDa and Pmp12 is only 56 kDa. Four genes (pmp3, 4, 5 and 17) contain a mutation resulting in premature stop. In C. pneumoniae strain AR39 [13] and J138 [14]pmp6 contains 393 less base pairs, and in strain J138 Pmp2 and Pmp4 contain frame shift mutations [14].
Variation of membrane protein expression is thought to provide protection against the immune system of the host [15] and it has been suggested that the Pmps may provide such a protection of chlamydiae. This was indicated by the findings of Birkelund et al. [16] and Pedersen et al. [17] who observed differential expression of Pmp10.
The detection of Pmp6 forty kDa below the predicted value has been interpreted by Henderson and Lam [10] as an indication that an N-terminal passenger domain is cleaved off, which is frequently seen for autotransporters. In strains AR39 and TW183, sharing the 393 base pair deletion, Pmp6 was detected at the resulting theoretical fulllength position of 130 kDa by Grimwood et al. [19].
In the study described in this paper, 2-D PAGE was used to elucidate the expression pattern of the C. pneumoniae Pmps in CWL029 at different times during the developmental cycle. Pmp6, Pmp20 and Pmp21 were found to be cleaved and the cleavage sites of Pmp6 and Pmp21 were identified. IMF was used to determine whether Pmps could be detected at the surface of the bacteria.

Results
The proteome study by Vandahl et al. [20] was not conclusive on the absence of any Pmps. Hence, we performed 2-D immnoblotting (IMB) to identify the 2-D PAGE position of expressed Pmps not earlier mapped. Pmp3, 4, 5, 12 and 17 were omitted from the study because these are truncated. Polyclonal antibodies (pAbs) were obtained by immunization of rabbits with recombinant proteins produced in E. coli (table 1). Antigens were raised against the N-terminal part of Pmp6, 7,8,10,13,14,15,16,18,19,20 and 21 because antibodies raised against full length Pmps showed cross-reactivity. The N-terminal part is the most variable and has been detected at the surface for C. psittaci Pmps [9]. All the obtained pAbs reacted in 1-D IMB with the recombinant Pmps to which they were raised (data not shown). The antigen for IMB was EB lysate. Reacting spots (figure 1A) were identified by comparison to a reference gel upon autoradiography of the blot (figure 1B) [22] or by MS analysis of spots excised from corresponding gels.
The proteins detected by each antibody are listed in table 1 using bold for strong reaction and plain for weaker reaction. PAb203-rPmp10 was mono-specific for Pmp10. The antibodies against rPmp2, rPmp7, rPmp8 and rPmp11 recognized these Pmps (respectively), and additional Pmps as listed in table 1. PAb221-rPmp6 recognized the 91 kDa spot that was earlier identified as originating from Pmp6 [20] and in addition another Pmp6 spot at 60 kDa marked X in figure 1A. PAb236-rPmp20 recognized a spot located just above the 90 kDa Pmps, which was earlier identified as originating from Pmp20 [20]. PAb237-rPmp21 recognized the 91 kDa Pmp6 spot, Pmp13 and a 48 kDa Pmp21 fragment marked Y in figure 1C. The results with antibodies against rPmp1 and rPmp9 were inconclusive and MS identification of these Pmps could not be obtained. Antibodies against rPmp16 and rPmp18 both reacted with Pmp8 and Pmp10 but no other proteins. Antibodies against rPmp15 and rPmp19 did not re- 2-D immunoblotting of EB proteins Radiolabeled proteins from EBs purified 72 hpi were separated by 2-D PAGE and electroblotted onto PVDF membranes. The membranes were reacted with polyclonal rabbit antibodies raised against recombinant Pmp. Bound antibodies were visualized by color reagents and membranes were afterwards exposed to X-ray films so that reacting spots could be identified by comparison to an annotated gel. A: PVDF blot of EB proteins reacted with Pmp6-pAb221, X shows a C-terminal fragment of Pmp6 recognized by Pmp6-pAb221; B: Autoradiography of A; C: PVDF blot of EB proteins reacted with pAb237-rPmp21, Y shows an N-terminal fragment of Pmp21 recognized by pAb237-rPmp21. Pmp10 act with any EB proteins. We were thus not able to detect Pmp 15, 16, 18 and 19 by IMB using EB lysates as antigen.

Identification of cleaved Pmps
The antibodies against the high molecular weight Pmps (Pmp6, Pmp20 and Pmp21) recognized protein spots at lower molecular weight than expected. PAb221-rPmp6 recognized a set of spots, X, at 60 kDa (figure 1A) in addition to the earlier identified 91 kDa fragment of Pmp6. MS analysis of a tryptic digest of X revealed peptides from the C-terminal part of Pmp6 (singly underlined peptides, figure 2). In contrast, the mass spectra from the Pmp6 spot at 91 kDa was found to contain peptides solely from the N-terminal part of Pmp6 (doubly underlined peptides, figure 2). To verify the identifications, gels were electroblotted onto PVDF membranes and N-terminal amino acid sequences of the proteins were assessed by Edman degradation. The sequence obtained from X (xVPVVPVAP) confirmed that it was a C terminal fragment of Pmp6 as it matched an internal sequence of Pmp6      Cleavage at this site results in a fragment of a theoretical Mw of 51 kDa, which is in good agreement with the observed Mw of 48 kDa. For the protein spot at 66 kDa containing the N-terminal peptides, cleavage at the signal peptidase cleavage site predicted by Signal-P was confirmed at the amino acid 30 ( 30 AHSLHSSELD). The theoretical molecular weight of a fragment starting at this position and ending at the cleavage site at amino acid 1185 is 116 kDa. As no peptides were identified at the Cterminal side of amino acid 548 in the N-terminal fragment, further cleavage or degradation from the C-terminal end will be a likely explanation of the discrepancy between expected and observed values.
Pmp20 was identified just above the 90 kDa Pmps, but as no molecular weight markers were present above 90 kDa, the exact molecular weight of the protein cannot be deduced. However, it is found much lower than the expected 178 kDa and the earlier obtained MS identification of this protein spot was based on peptides located in the C-terminal part of Pmp20 (residues 1057-1069, 1387-1400, 1421-1432, 1506-1527, 1557-1571, 1587-1597, and 1644-1654) indicating that also Pmp20 is cleaved.

2-D PAGE analysis of time dependent expression
The time of expression of the Pmps identified in the 2-D protein profile was investigated by pulse labeling with MOMP is expressed during the growth fase of Chlamydiae [25]. DnaK is an early protein [26] and has been suggested to be RB specific and hence to show decreased expression late in the developmental cycle [27]. The expression of MOMP was found to increase at the same rate from 24 to 36 hpi as from 36 to 48 hpi ( figure 4). The expression of DnaK was similar to that of MOMP at 24 hpi, but showed no increase from 24 to 36 hpi and decreased from 36 to 48 hpi ( figure 4).
It can be concluded that all Pmps investigated with respect to time dependent regulation (Pmp2, 6,7,8,10,11,13,14) were found to be strongly up-regulated between 36 and 48 hpi, coinciding with the conversion of RB to EB. This is in agreement with what has been reported for C. trachomatis and C. psittaci POMPS [28,29]. At 48 hpi Pmp8 and Pmp10 were several fold higher expressed than any other Pmp.

Immunofluorescence microscopy
In order to investigate whether the Pmps absent from EBs could be detected during growth in cell culture, the antibodies against these Pmps were used in IMF on C. pneumoniae infected HEp-2 cells. Cells cultivated on coverslips were infected and fixed in methanol at 72 hpi. The investigated antibodies were visualized with a FITCH conjugated secondary antibody. The antibodies against rPmp15, 16, 18 and 19, which did not react in IMB, did not react in IMF either (table 1). Furthermore, Pab236-rPmp20 was found not to react in IMF (table 1) although Pmp20 was identified in EB gels by IMB and MS. Examples are shown in figure 5.
All Pmp antibodies were used in IMF of infected cells that were formalin fixed at 72 hpi. Formalin cross-linking of outer membrane proteins prevents antibodies from penetrating the bacteria so that only proteins in the outer membrane are accessible. MAb18.1-DnaK was included as a control of the impermeability as described earlier [30].
As the DnaK epitope recognized by MAb18.1 is formalin insensitive [30], the lack of reaction of MAb18.1 (figure 5, row 1) must be due to the prevention of reaction of interior proteins by the fixation.
The pAbs raised against recombinant Pmp1, 2, 7, 13, 14, 15, 16, 19 and 20 did not react after formalin fixation. Due to the described cross-reaction of the Pmp antibodies, the reacting antibodies were absorbed with recombinant proteins. The Pmp antibodies were absorbed with rPmps corresponding to the Pmps to which they cross-reacted in IMB and with the rPmps to which they were raised, respectively. The reaction of pAb208-rPmp9 could be prevented by addition of any of the Pmps 6, 8, 10, 11 or 21 (table 2), and hence it must be considered non-specific. The pAbs raised against rPmp6, 8, 11 and 21 were found to react specifically in formalin fixation (table 2), as their reaction was prevented by the recombinant Pmp that they were raised agains, but not by the ones to which they cross reacted in IMB. Pmp8 is shown as an example in figure 5 (6). Pab203-rPmp10 was not investigated by competition IMF as it was shown to be mono-specific for Pmp10 by IMB. Pmp10 [30,17] and Pmp11 [30] have earlier been shown to be surface exposed and the results indicate that this is also the case for Pmp6, 8 and 21. Other Pmps may also be at the surface, but in that case the epitopes recognized by our antibodies have been destroyed by the fixation or in some way masked.
PAb116-rMOMP did not react after formalin fixation (figure 5, row 2), which is in agreement with earlier reports on the lack of surface exposed linear epitopes of C. pneumoniae MOMP [31]. Pab110-rOmp2 did react (figure 5, row 3) indicating the presence of surface exposed epitopes of Omp2 in C. pneumoniae in opposition to C. trachomatis where Omp2 is probably located at the inner surface of the outer membrane complex [32,33].

Discussion
In the present study, cleavage sites of Pmp6 and Pmp21 have been identified, and it is suggested that also Pmp20 is cleaved. Cleavage of the these Pmps explain why they are found at different molecular weight in different studies. Grimwood et al. [19] detected the N-terminal fragment of Pmp6 at 90 kDa that was also identified by Vandahl et al. [20], whereas Montigiani et al. [21] found the lower molecular weight C-terminal fragment of Pmp6. Only a C-terminal fragment of Pmp21 has been described previously [20,21].
There is no similarity in amino acid sequence between the cleavage sites in Pmp6 and Pmp21, but both sites are located between the C-terminal predicted β-barrel [10] and the N-terminal predicted parallel β-helix fold [34]. This position is in agreement with the theory that the Pmps may be autotransporters; as such often cleave off their Nterminal part [10]. However, if the N-terminal part is liberated it would not be detected in IMB using proteins from purified EBs as antigens, and as the N-terminal fragment is detected in EB gels for Pmp6 and Pmp21 it must be concluded that it remains bound to the EB.
Pab236-rPmp20 was raised against a fragment of Pmp20 covering the N-terminal part and some of the C-terminal fragment, but only the C-terminal part could be detected in IMB (data not shown), suggesting that the N-terminal part may be liberated. Pab236-rPmp20 did not react in IMF. It was raised against unfolded recombinant protein and may not recognize the correctly folded protein, especially not if the N-terminal part is liberated or degraded.
The fragments of Pmp6, Pmp20 and Pmp21 were detected after a labeling period of two hours, and no increase in the amount of cleavage products was observed after a chase period of six hours (data not shown). This suggests that the cleavage occurs rapidly after synthesis. We consider it unlikely that the cleavage should be an artificial phenomenon as the infected cells were harvested in a lysis solution containing 7 M urea, 2 M thiourea and reducing agents, and as the results were highly reproducible. The lack of detection of full-length products of the high molecular weight Pmps may be caused by low resolution in 2-D PAGE of such proteins, but full-length products of these were also absent in 1-D IMB by Grimwood et al. [19]. Interestingly, Grimwood et al.
[24] detected Pmp6 from strains TW183 and AR39 at a Mw of 130 kDa, which is the theoretical Mw of Pmp6 in these strains, due to a deletion of the bases encoding amino acids 429 to 559, which does not include the cleavage site.
In the present study we did not detect any of the Pmps15-18, originating from cluster III. Pmp16 and Pmp18 are the only Pmps lacking a leader sequence. Pmp17 is truncated Reaction with pAb201-rPmp8 in the presence of: rPmp8: rPmp8, to which the antibody was raised, rPmp10: rPmp10, to which the antibody cross reacted in IMB, control: no recombinant proteins. As rPmp8 prevents reaction of pAb201-rPmp8 whereas rPmp10 does not affect the reaction, we conclude that the observed reaction is specific for Pmp8. and was thus not investigated. Montigiani et al. [21] have reported identification of Pmp16 in C. pneumoniae from a clinical isolate, but Grimwood et al. [19] found that both Pmp16 and Pmp18 were unstable in CWL029, which may explain that these Pmps were not observed in our study. However, it could also be speculated whether differences have been introduced through laboratory passages. We observe expression of Pmp10 in strain CWL029 whereas it was not detected in strain CWL029 by Grimwood et al. [19] due to a frame shift mutation. Grimwood et al. [19] have described differences in the expression of Pmp1 and Pmp3 between strains CWL029 and TW183 as determined by immunoblotting, meaning that at least some Pmps are differentially expressed between strains.
Pmp6, Pmp8, Pmp10, Pmp11 and Pmp21 were detected at the surface of the bacteria by formalin fixation IMF. Apart from Pmp21, all the Pmps detected at the surface show greatest similarity to PmpG [4], a constituent of the outer membrane of C. trachomatis L2 [35,28]. An explanation of the expansion of the number of C. pneumoniae Pmps similar to PmpG could be that C. pneumoniae varies its surface to escape the immune defence of the host by changing the expression of Pmps. However, if Pmp expression is changed in order to vary the surface, this is most likely a consequence of the surface localization of Pmps rather than the very reason for the existence of Pmps.
The most highly expressed among all the investigated Pmps at all points in time is Pmp10, which is known to contain surface exposed epitopes [30]. Pmp10 is differentially expressed in infected cell culture [17] and mice [16]. Pmp10 still being the most highly expressed Pmp suggests that it carries out an important function. This function would presumably have to be supplemented in bacteria not expressing Pmp10. Pmp10 contains a predicted signal peptidase II cleavage site directing lipid modification and pmp11, which is located next to pmp10 (but in the opposite direction), encodes the only other predicted lipid modified Pmp. If the function of Pmp10 depends on lipid modification, and this function is needed in bacteria lacking Pmp10, Pmp11 would be the obvious alternative.
The finding that all Pmps are heavily upregulated at the time of conversion of RBs into EBs indicates that the function of these is structural but they may also be needed with respect to attachment or entry of EBs. An N-terminal triangular beta-layer motif could provide the bacteria with a shielding lattice and ensure proper spacing to a host cell or an epitope exposed to the complement system. If the lipid modifications of Pmp10 and Pmp11 are used as anchors inserted into the host cell membrane, subsequent action of other entry molecules would probably depend on proper spacing. However, all theories on functions of the Pmps remain very speculative.

Conclusions
The Pmps investigated with respect to time dependent regulation (Pmp2, Pmp6, Pmp7, Pmp8, Pmp10, Pmp11, Pmp13, and Pmp14) were found to be up-regulated late in the developmental cycle. Due to their reaction in formalin fixation we propose that Pmp6, Pmp8, Pmp10, Pmp11, and Pmp21 are surface exposed. Pmp6, Pmp20 and Pmp21 were found in cleaved forms and the identified cleavage sites of Pmp6 and Pmp21 are in agreement with the theory that the Pmps are autotransporters.

Methods
Organisms and cultivation C. pneumoniae CWL029 (also known as VR1310) (ATTC) was cultivated in semi confluent monolayers of Hep-2 cells (ATCC) as described [20]. Elementary bodies were purified 72 hpi essentially as described [30] with the exception that Visipaque replaced Urografin for gradient making.

PCR and cloning
PCR enzymes were Expand™ High Fidelity (Roche) and reaction conditions were as recommended by the manufac- The results of competition IMF in which the recombinant Pmp listed in the top row was added to an IMF reaction using the polyclonal antibodies listed in the first column. The reactions were carried out only for the shaded

Production of antibodies
New Zealand White rabbits were immunized: i) intramuscularly on days 1, 8 and 15 using 10 µg of protein in PBS and 50% Freunds incomplete adjuvant; ii) intravenously on days 29, 36 and 43 using 10 µg of protein in PBS. The rabbits were bled on day 60.

Labeling
Pulse labeling of infected cell cultures was performed by using 100 µCi/mL radioactive methionine (Amersham Pharmacia) in a methionine-free RPMI 1640 medium as described previously [20]. 40 µg/mL cycloheximide was added to inhibit host cell protein synthesis during labeling. Cell cultures were labeled for two-hour-periods at 6, 12, 24, 26, 42, 48 and 54 hpi and harvested immediately after labeling by scraping off in lysis solution (7 M urea, 2 M thiourea, 4% w/v CHAPS, 40 mM Tris base, 65 mM DTE and 2% v/v Pharmalyte 3-10 (Amersham Pharmacia)). Before purification of EB the same labeling periods were used to ensure incorporation of radioactivity into proteins synthesized at all times during the developmental cycle.

Electrophoresis
One-dimensional SDS gels were run as described [30]. The protocol for two-dimensional gel electrophoresis was as described [20]. First dimension was carried out using non-linear Immobiline Drystrips pH 3-10 (Amersham Pharmacia). Proteins were focused in the strips using an IPGphor (Amersham Pharmacia) applying a voltage of 20 V for rehydration of strips and 120 kVh for focusing. For second dimension 9-16% T gradient polyacrylamide SDS gels were used. Comparative gels were loaded with 300.000 cpm as determined by TCA precipitation and scintillation counting. Gels for N-terminal sequencing were loaded with 600 µg protein from purified unlabeled EB. Gels for MS identification were loaded with 3.000.000 cpm and a total of 600 µg protein -all from purified EB. Gels for IMB were loaded with 75 µg protein from purified unlabeled EB and 300.000 cpm labeled EB protein. Protein spots in gels were visualized by autoradiography as described [20]. Comparative gels were treated with Amplify (Amersham Pharmacia) before exposure to BioMax MR X-ray films (Kodak).

Immunoblotting
Two-dimensional gels were washed for five min. in double-distilled water and soaked for 30 min. in transfer buffer. PVDF membranes (Immobilon-P, pore size 0.45 µm, Millipore) were soaked for one min. in methanol and then for 30 min. in transfer buffer. The transfer buffer contained: 50 mM Tris, 50 mM boric acid, 0.02% SDS and 10% methanol. The transfer was performed at 90 V and 10°C for four hours. Membranes were blocked in a buffer containing 20 mM Tris, 150 mM NaCl, and 3% gelatine, pH 7.5, washed in washing buffer (20 mM Tris, 500 mM NaCl, 0.05% tween-20) and incubated with polyclonal antibodies (pAbs) diluted 1/300 in antibody buffer (washing buffer added 0.2% gelatine) for one hour at 37°. After three washing steps membranes were incubated with secondary antibody (goat-anti-rabbit-IgG-AP-conjugate, BioRad) diluted 1/2000 for another hour at 37°C. After washing the blots, bound antibody was visualized with 270 µL NBT (50 mg/mL in 70% DMF) and 270 µL BCIP (25 mg/mL in 100% DMF) in 45 mL buffer containing 100 mM NaCl, 5 mM MgCl and 100 mM Tris-HCl at pH 9.5. As a fraction of the EB proteins was radiolabeled, reacting spots could be identified by comparison to a reference gel [20] upon autoradiography of the PVDF membrane [22].

Mass spectrometry
Protein spots excised from preparative gels were identified by MALDI TOF mass spectrometry as described [20].

N-terminal sequencing
Protein spots excised and pooled from eight Coomassie Brilliant Blue stained PVDF membranes each loaded with 600 µg of protein from purified EB were analyzed using an Applied Biosystems 404 protein sequencer (Perkin Elmer).
Immunofluorescence microscopy IMF was performed as described [36] analyzing infected cells fixed in methanol or formalin at 72 hpi. Formalin fixed cells were permeabilized with 0.2% Triton X-100 for 10 minutes at room temperature. Bound primary antibody was visualized with a secondary FITCH conjugated antibody (DAKO), which was absorbed to prevent crossreaction.