|Home | About | Journals | Submit | Contact Us | Français|
Variants of an ilp (invasin-like protein) gene have been identified previously in Chlamydia caviae and in Chlamydia suis. The C. caviae ilp gene is interrupted by two frame shift mutations while the C. suis gene is intact. Characterization of the ilp gene in C. caviae passaged minimally in vitro showed that the two frameshift mutations were present in the original isolates. The gentamicin protection assay was used to determine if E. coli bacteria expressing the intact C. suis ilp could adhere to or invade HEp-2 cells. While inv+ clones showed increased adherence and invasion, no increase in adherence or invasion was observed for ilp+ clones. However, these clones were found to produce detectable amounts of ilp transcript. In a 48 hour time course of C. suis culture, ilp transcript was initially detected at 8 hours, peaked at 16 hours, and declined subsequently. Antibodies specifically recognizing the putative functional domain of Ilp failed to detect any ilp -specific gene product in either E. coli or C. suis cultures. These data suggest that ilp does not encode a functional protein and raise questions about how ilp was introduced and maintained in Chlamydia.
Invasins and intimins comprise a family of proteins associated with tight adherence to and/or invasion of eukaryotic cells. The initial two members of this family, Yersinia pseudotuberculosis invasin and E. coli intimin, show significant sequence homology, mostly within the amino terminus (1,2). An ORF encoding an invasin-like protein (Ilp) was identified in the genome of C. caviae strain Guinea Pig Inclusion Conjunctivitis (3). Although the ORF includes two frameshift mutations, the predicted Ilp amino terminus shows significant homology to invasin and intimin. The identification of this gene within Chlamydia sp. was intriguing because the Chlamydiales are a group of obligate intracellular bacteria characterized by a developmental cycle alternating between the metabolically active, non-infectious reticulate bodies and the infective, metabolically inactive elementary bodies within the sequestered environment of an inclusion inside the host cell. However, a complete ilp gene has not been found in any of the other ten published Chlamydiales genomes (3-11), although partial ilp sequence is found in C. muridarum (12,13). Furthermore, the identification of this mostly intact, relatively large gene (4 kb) was surprising in the context of the proposed reductive evolutionary path of the Chlamydiales to a niche-specific minimal genome configuration (6). The minimal genome concept and evidence that nearly all genes are transcribed in C. trachomatis suggest strongly that no gene would be maintained without a benefit (14,15). Recently, an uninterrupted version of ilp was identified in a mini-plasticity zone of Chlamydia sp. downstream of rrn in C. suis (13,16). This work attempts to characterize ilp more fully and determine if it is functional in C. suis.
HEp-2 cells were cultured in RPMI-1640 with glutamine, plus 5% FBS. GST fusion constructs, made using pGEX-2T (Amersham), pGEX-mIlp and pGEX-YscU (17), were expressed in E. coli TB1. Plasmid pCR2.1/5-1 contains full length ilp with its putative promoter cloned into pCR2.1 (13). Plasmids pRI203, encoding full length invasin, and pBR322 were obtained from Ralph Isberg and Mark Strauch respectively. These plasmids were transformed into E. coli DH5α and HB101. C. suis S45, encoding full length ilp, was obtained from Dan Rockey. Several C. caviae isolates were used to characterize the mutations identified (Table 1).
Female Hartley strain guinea pigs, (Simonson Laboratories, Gilroy, CA), were inoculated in the conjunctiva with 25 μl SPG containing 106 IFU of C. caviae onto each eye. Following infection of the first animal with tissue culture grown chlamydiae, C. caviae was passaged directly from animal to animal four consecutive times with 7-day intervals. Briefly, a swab was used to collect C. caviae from the conjunctiva when the gross pathologic response was beginning to decline and then to inoculate the next, naïve animal by rubbing the conjunctiva. A second swab was collected from the contralateral eye of each guinea pig, placed in transport medium and frozen at −70°C.
DNA was extracted from swabs using cold phenol:chloroform-isoamyl alcohol followed by isopropanol precipitation. The final DNA pellet was resuspended in dH2O. PCRs to amplify the frame-shift region of ilp of C. caviae archived specimens and passages in animals were performed using primers GPIC Invasin F & GPIC Invasin R and Promega Taq DNA Polymerase (Table 2) (18). The PCR products were directly sequenced, using the Big Dye Terminator method and ABI sequencing (Perkin Elmer, Applied Biosystems) following the manufacturer’s instructions.
The ilp promoter was identified with FindPatterns (GCG Wisconsin Package) using a simple pattern matching algorithm, ‘TTG-n18-24-TA-n3-T’. BLAST analysis of the protein sequence defined the first Ig-like domain based on homology to the intimin/invasin Ig-like domain. Repeat domains were identified based on sequence homology to the first Ig-like domain, homologous repetitive sequences, and similar predicted secondary structures. Repeat sequences were detected using RADAR (19). Secondary structure predictions were performed using the JPred multiple prediction algorithm (20). Alignments to determine homology were performed using ClustalX 1.81. The Ilp signal sequence was identified using SignalP 3.0 (21).
The gentamicin protection assay was performed largely as previously described (22). Aerobic, mid-log cultures of E. coli DH5α or HB101 grown at 22 °C, containing no vector, pBR325, pRI203, pCR2.1, or pCR2.1 5-1 were used in these assays. HEp-2 cell monolayers (>95% confluency) washed with PBS were incubated with 106 cfu/ml in RPMI-1640 with glutamine plus 5% FBS. The monolayers were briefly centrifuged to synchronize contact of the bacteria with the monolayer, incubated for 3 hours, and washed with PBS (22). The samples were processed either to determine adherence, by lysis in 0.1% Triton-X100, or invasion, by further incubation with gentamicin (100 μg/ml) for 2 hours. The invasion samples were washed and lysed as before. Bacterial counts were determined for input, as well as adherence and invasion samples. The percent adherent or invasive was calculated relative to input. Each strain was tested in triplicate in two independent experiments.
Hela cell monolayers (>95% confluency), grown in Dulbecco’s Modified Eagle’s Medium (DMEM) plus 10% FBS and gentamicin (25 μg/ml) were washed with PBS and incubated in 1X DEAE-Dextran After removal of the DEAE-Dextran, C. suis S45 seed (106 IFU/ml) diluted 1:10 in Hank’s Buffered Saline Solution without calcium and magnesium (BioSource), was added to the monolayers and centrifuged for 1 hour (1800 RPM/22°C). Cultures were then incubated in DMEM plus 10% FBS, gentamicin (25 μg/ml), and cycloheximide (0.5 μg/ml).
C. suis S45 was cultured as described, sampled every 8 hours for 24 hours, and at 48 hours p.i.. C. suis S45-infected, and mock-infected, Hela cells were lysed in TRIzol reagent (Invitrogen). Each sample in TRIzol was processed according to the manufacturer’s instructions, except that 500 μl chloroform was used instead of 200 μl. RNA samples (10 μg) were treated with DNase I and split (5 μg each) for reverse transcription with and without SuperScript™ II Reverse Transcriptase (RT) (Invitrogen) according to the manufacturer’s instructions with RNaseOUT™ and primers LBRT3, LB3ilp3 and LB16s3 in a single reaction tube (Table 2). The cDNA template (500 ng) was then amplified using Platinum Taq (Invitrogen) and one of three primer pairs to detect either ilp or 16s rRNA: LBRT1 and LBRT2, LB16s1 and LB16s2, or LB3ilp1 and LB3ilp2 (Table 2). The LBRT and LB3ilp primer sets were respectively directed at site 1,102 and 2,882 bp downstream from the start codon. PCR products were analyzed by agarose gel electrophoresis.
To verify transcription in E. coli, RNA was isolated as before from aerobic, mid-log phase HB101 cultures containing either pCR2.1 or pCR2.1/5-1, grown as above (2.5) or at 37°C. RT-PCR was performed as before, except that a new 16S rRNA primer set, directed against E. coli 16S rRNA, was used. LBHB3 was used during reverse transcription while LBHB1 and LBHB2 were used for PCR amplification of the cDNA template (Table 2).
Sequence encoding the predicted minimal functional domain of Ilp, mIlp, comprising the final 238 carboxy terminal residues, was amplified from pCR2.1/5-1, encoding full length ilp, using primers LBmilp1 and LBmilp2 (Table 2). The GST fusion plasmid construct, pGEX-milp, was obtained by digesting the amplified fragment and pGEX-2T with EcoRI and BamHI and ligating with T4 DNA ligase (New England Biolabs). The plasmid construct, used in the generation of Ilp-specific antibodies, was confirmed by restriction analysis and sequencing.
Insoluble inclusion bodies containing GST-mIlp fusion protein were isolated from an aerobic culture of E. coli TB1 pGEX-milp by sonication in TEN buffer with Triton-X100, followed by centrifugation and washes, in the presence of protease inhibitor cocktail III (Calbiochem). The final pellet, containing the inclusion bodies, was resuspended in PBS and used for primary subcutaneous immunization in complete Freund’s adjuvant and two boosts at 2-week intervals in the guinea pig. Antiserum, collected 45 days post-primary immunization, was tested by immunoblot to verify specificity using lysates containing GST-mIlp, GST, and GST-YscU, a GST fusion with an irrelevant antigen from the chlamydial type III secretion system (17). Immunoblots were probed with antiserum at 1:1,000 or 1:500 dilutions in TBS-T (TBS, 0.1% Tween-20) with 5% non-fat milk.
HB101 with either pCR2.1 or pCR2.1/5-1 was lysed by sonication in STE (0.1% Sarkosyl) buffer and centrifuged. Triton X-100 (0.1%) was added to the supernatant and the pellet was resuspended in PBS with 1% SDS. Additionally, samples from C. suis S45-infected and mock-infected Hela cultures were collected in PBS at 16 hours p.i., prepared as above and analyzed by immunoblot with polyclonal anti-GST-mIlp at 1:500 dilution. Pre-adsorption with GST-YscU inclusion bodies was performed to reduce reactivity to the GST portion of the fusion protein and to contaminating E. coli proteins. Bacterial lysates containing GST and GST-mIlp served as negative and positive controls, respectively. GST-YscU inclusion bodies were partially purified from E. coli TB1 pGEXTTSorfa#7, encoding yscU as a fusion with GST, as before.
A list of predicted polypeptide sequences containing either PF02369 and/or PF02368, PFAM domains associated with intimins and invasins, was retrieved using the Comprehensive Microbial Resource (CMR) (12). BLAST and literature searches identified other potential intimins or invasins not present in CMR. The list was reduced to sequences matching at least two of the following criteria: 1, scores above the HMM trusted score for either of the domains; 2, sequences having either PFAM domain, or the COG5492 superfamily domain, identified in a Conserved Domain (CD) search; 3, sequences matching by BLAST-P to another sequence fulfilling the preceding criteria (23,24). This list was further reduced to remove pseudogenes and duplicate proteins. The sequences were aligned using ClustalX and a neighbor-joining tree using BLOSUM62 was generated in Jalview (25).
The initial characterization of the C. suis Ilp identified homology, limited to the amino terminus of the protein, to invasin and intimin, as shown for C. caviae Ilp (1,2). Analysis of the C. suis ilp sequence with FindPatterns (Sarah Mathews and Peter Timms, personal communication) identified a putative σ70-like promoter, ‘TTGCCA-n15-TAATAT’, about 100-bp upstream of the predicted start codon.
In addition to sequence homology, the general Ilp structure is similar to that of other intimins and invasins and includes an extended rod-like structure consisting of several modular repeat domains and an unrelated distal domain, presumably representing the protein active site (26,27). Both ilp genes encode 7 repeat domains (D1–D7), more than invasin or intimin. The two chlamydial Ilp proteins also include a putative signal sequence with a signal peptidase I cleavage site, supporting an extracellular localization.
Although the sequence of the extracellular carboxy terminal segment of Ilp diverges from its counterparts in invasin and intimin, it is conserved between the two Chlamydia strains with percent identity of 74.8 (D2–D7) and 34.8 (D8). Secondary structure predictions revealed no conserved salient features in D8. However, several conserved beta strands were predicted in D2-D7, consistent with the Ig-like crystal structures of intimin D0-D2 and invasin D1-D3 (26,27).
Chlamydial cultures are not routinely submitted to limiting dilution and, therefore, cannot be assumed to be clonal. Moreover, the C. caviae isolate whose genome was sequenced, although derived from the original Murray isolate, had been serially cultured in vitro for many generations (28). A distinct possibility, therefore, was that the genomic ilp sequence had been isolated from a mixed population that originally included both wild type and mutant ilp. To address this question, the rrn-nqrF segment, containing ilp, was amplified directly from ocular swabs of infected guinea pigs after 5 serial passages, without intervening in vitro culture. Another specimen was swabbed from the genital tract at day 10 p.i. In addition, the corresponding sequence was amplified from several early archived specimens of the original C. caviae isolate from the Murray laboratory (Table 1). The 1962 (first passage of the isolate) and 1973 specimens are from stocks maintained in Murray’s laboratory by passage through embryonated hens’ eggs. In all cases, the nucleotide sequence of the amplicons generated was identical to that obtained from genomic analysis, containing the same two frame-shift mutations.
Since intimin and invasin have been shown to have a role in adherence and/or invasion, gentamicin protection assays were undertaken to determine if Ilp has a similar role in Chlamydia. Bacteria recovered from the HEp-2 monolayer after incubation and subsequent washes but before exposure to gentamicin were defined as adherent. Bacteria recovered from the monolayer after gentamicin treatment were defined as invasive. The entire ilp gene, including 135bp of upstream sequence containing the putative promoter, was previously cloned into pCR2.1, generating pCR2.1/5-1 (13). This construct was used to express full length Ilp in DH5α and HB101. HB101 with either pRI203, expressing the entire Y. pseudotuberculosis invasin, pBR325, pCR2.1, or pCR2.1/5-1, expressing Ilp, were studied in parallel in two independent experiments to determine adherence to and invasion of HEp-2 cells. While HB101 pRI203 showed high levels of adherence and invasion, HB101 pCR2.1/5-1 showed neither adherence nor invasion above background (HB101 pCR2.1 and HB101 pBR325) (Fig. 1A&B). E. coli DH5α was also tested and, as with HB101, while the invasin-expressing control showed significant adherence and invasion, pCR2.1/5-1-mediated adherence and invasion were similar to vector controls (Fig. 1C&D).
To verify that the lack of phenotype was not due to lack of transcription of ilp, HB101 pCR2.1 and HB101 pCR2.1/5-1 were cultured at 22°C or 37°C, aerobically. While ilp transcript was undetectable when bacteria were grown at 37°C, cultures grown at 22°C, the growth temperature prior to inoculation of the HEp-2 monolayers, were found to produce detectable levels of ilp transcript (not shown).
Since an adherent/invasive phenotype could not be associated with ilp in E. coli, ilp functionality was evaluated based on the presence or absence of an ilp transcript during C. suis infection. C. suis S45 cultures were sampled every 8 hours for 24 hours, with a final reading at 48 hours p.i., to determine if ilp was expressed early, mid, or late in the developmental cycle. Results showed an increase in ilp transcript first weakly detectable at 8 hours, with expression peaking at 16 hours and then decreasing at 24 and 48 hours p.i. (Fig. 2). In contrast, 16S rRNA transcription remained relatively constant throughout the time course. These data suggest mid-cycle expression for ilp with apparent late down-regulation (29).
To further assess Ilp expression, a second plasmid was designed to express a GST fusion containing the putative binding domain, D8, and the preceding Ig-like domain, D7, which together comprise the putative minimal functional domain. This fusion protein, GST-mIlp, was used to generate antibodies against Ilp. Whole cell lysates of Hela monolayers, either C. suis-infected or mock-infected, as well as HB101 with pCR2.1 or pCR2.1/5-1, were probed with GST-mIlp antiserum extensively pre-absorbed with GST-YscU inclusion bodies. A mid-cycle C. suis culture in Hela cells (16 hrs p.i.) was used as transcription was maximal at this time. E. coli cultures, grown as for the gentamicin protection assays, were used as ilp transcript was detected under these conditions. While the antiserum showed strong, specific reactivity against a band of apparent molecular 49 kDa, consistent with the predicted 52.3 kDa Mr of GST-mIlp, an Ilp-specific band was undetectable in ilp-expressing E. coli HB101 or C. suis S45-infected Hela cells (Fig. 3)
The proteins (n=90) used in this analysis were found in 54 different strains representing 44 species from 33 genera (Supplementary Table 1). These genera represented nine phyla, although more than 80% of the proteins came from the Firmicutes and Proteobacteria. The phylogenetic tree of these proteins can be divided into three groups (Fig. 4). The first group (Firmicutes-related) consists of Firmicutes proteins with three exceptions: a Y. pestis phage protein, a Microscilla sp. protein, and a B. bifidum protein. The second group (phage-related) contains all but one of the enteric phage proteins, as well as proteins from a variety of species from five phyla. The prevalence of phage proteins in this group could explain the wide diversity of species observed. The final group contains mostly Proteobacteria proteins, including the previously characterized intimins and invasins. The chlamydial proteins branch from the enteric intimin/invasin family and not with proteins from the more closely related Spirochaetes. This result suggests that Ilp may have arrived in Chlamydia via horizontal gene transfer (HGT) from an Enterobacteriaceae ancestor.
The ilp gene of C. caviae contains two frame-shift mutations early in the gene. The possibility that these mutations were artifacts of in vitro propagation over long periods of time was investigated by comparing homologous sequences from early archived specimens of the isolate, as well as in vivo passaged versions of the sequenced C. caviae isolate. In all cases, the recovered sequence carried both frame-shift mutations. This indicates that the mutations were most likely present in the original isolate and that Ilp is not required for C. caviae infection of guinea pigs. This is consistent with the finding of an ilp remnant in the murine pathogen, C. muridarum (7). Moreover, while full-length undisrupted ilp was isolated in C. suis S45, the finding of ilp interrupted by a tetracycline-resistance cassette in seven closely related C. suis strains indicates that this gene is not required for infection of pigs (16).
The loss of ilp in C. muridarum, caviae, and certain suis isolates indicates that, although these species may have had intact ilp in their evolutionary past, they have now dispensed of it and appear to be doing well without it. Nonetheless, it remains possible that ilp is a necessary virulence factor under specific circumstances, such as in an alternate host or site of infection. Functional studies presented in this work do not support a role in invasion or adherence, as E. coli expressing ilp showed no detectable adherence or invasion phenotype. Furthermore, while ilp transcript could be detected in E. coli and C. suis, Ilp protein was not detectable in either species. Although it is possible that Ilp is present in amounts that are below the immunoblot detection threshold, our results suggest that translation may be blocked or the protein may be rapidly degraded, rendering Ilp nonfunctional. These observations conflict with a strict model of reductive evolution for C. suis and, by extrapolation, for all Chlamydia sp.
It is unlikely that ilp was part of the genomic content of the last common ancestor preceding the Chlamydiales, since phylogenetic analysis suggests that ilp arose from an ancestor of enteric intimins and invasins rather than from the closest relatives of the Chlamydiales, the Spirochaetes. The possibility of an HGT is reinforced by multiple comparative genomic evidence supporting relatively recent HGT events in Chlamydia. Examples are the acquisition of tetracycline resistance genes in C. suis and fifteen other chlamydial genes identified as originating by HGT (16, 30). Moreover, the occurrence of an HGT at the rrn-nqrF site is consistent with the genomic plasticity of this site (13).
An intriguing possibility is that this gene arrived by HGT after major genome loss had already occurred. The donor may be an inv ancestor, based on similar temperature-dependent expression of cloned ilp and Ilp’s phylogenetic grouping with enteric intimins and invasins. That the only Chlamydia sp. carrying ilp sequence, suis, muridarum and caviae, are also possible gut pathogens, adds weight to the possibility of the donor being a gut pathogen. However, examination of the genome of the gut pathogen C. pecorum did not identify any ilp remnant. The maintenance of recently acquired ilp may be paradoxically explained by the extensive genomic loss that preceded it, causing the loss of insertion sequences and transposon elements that normally facilitate large scale deletion events (31). Different levels of maintenance in different species could be explained by subtle selective advantages or disadvantages afforded to descendants by Ilp. As Chlamydia species diverged to infect different hosts, ilp may have been lost rapidly in some, such as C. pneumoniae and trachomatis ancestors, due to a significant detrimental effect. Other species, such as C. muridarum, may have retained ilp longer because no selective disadvantage was present leading to slow decay, while C. caviae and suis may have maintained the gene because of a benefit provided in ancestral hosts that has since been lost.
A variant possibility is that ilp was introduced by a later HGT event to a putative common ancestor of the closely related C. suis and muridarum. The ilp gene may then have been transferred to C. caviae in a second, independent HGT event. This alternative possibility would explain the lack of ilp in C. trachomatis and pneumoniae and would also reduce the amount of time for gene decay to occur. However, it still suggests that C. muridarum and suis followed different patterns of gene loss, suggesting that ilp was at some point more beneficial to C. suis than C. muridarum.
In summary, the Ilp protein’s role in Chlamydia remains a mystery. It is still possible that Ilp is expressed below immunoblot detection levels and that the functional assays were irrelevant to Ilp function. However, the apparent disparity between ilp transcription and translation, coupled with the odd transcriptional upregulation in E. coli at 22°C, a temperature non-permissive to chlamydial growth, suggest that ilp is a gene in evolutionary transition. Remnants of ilp sequence in various Chlamydia species may represent time-lapsed images of the decaying ilp gene.
Supplementary Table 1. Putative Inv/Int family members based on domain and BLAST homologies. aSpecies abbreviation used in the phylogenetic tree is shown. bSupporting data for PFAM homology is indicated with % aligned for CD results or HMM score and trusted score for CMR. cDomain repeats are indicated from N-terminus to C-terminus separated by a semicolon for differing domains. dBLAST results are given as % identity to referenced ORF related to the number of aligned amino acids. A summary of repeat structures are given. eHomology to known phages by BLAST-P is indicated. fThe possibility of the ORF having undergone HGT, based on either Garcia-Vallve et al. or Tsirigos et al., is indicated.
We thank James Kaper, Mark Strauch, Ralph Isberg and Dan Rockey for the gift of strains and plasmids. This work was supported by a postdoctoral fellowship to Laurel Burall from the Training Program in Oral and Craniofacial Biology NIH/NIDCR T32 DE07309-08.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.