| Home | About | Journals | Submit | Contact Us | Français |
Among chlamydial virulence factors are the type III secretion (T3S) system and its effectors. T3S effectors target host proteins to benefit the infecting chlamydiae. The assortment of effectors, each with a unique function, varies between species. This variation likely contributes to differences in host specificity and disease severity. A dozen effectors of Chlamydia trachomatis have been identified; however estimates suggest that more exist. A T3S prediction algorithm, SIEVE, along with a Yersinia surrogate secretion system helped to identify a new T3S substrate, CT082, which rather than functioning as an effector associates with the chlamydial envelope after secretion. SIEVE was modified to improve/expand effector predictions to include all sequenced genomes. Additional adjustments were made to the existing surrogate system whereby the N terminus of putative effectors was fused to a known effector lacking its own N terminus and was tested for secretion. Expansion of effector predictions by cSIEVE and modification of the surrogate system have also assisted in identifying a new T3S substrate from Chlamydia psittaci. The expanded predictions along with modifications to improve the surrogate secretion system have enhanced our ability to identify novel species-specific effectors, which upon characterization should provide insight into the unique pathogenic properties of each species.
The Chlamydiaceae are obligate intracellular bacteria distinguished by a unique biphasic developmental cycle. Each cycle begins when the infectious, metabolically-arrested elementary body (EB) attaches to a eukaryotic host cell. Once inside the host cell, the bacterium differentiates into the non-infectious, metabolically-active reticulate body (RB). Throughout the developmental cycle, chlamydiae exist and multiply within an intracellular vacuole called an inclusion.
As a group, these bacteria are capable of infecting a wide range of hosts including humans, animals, and protozoa. While the method of infection and the developmental cycle for all Chlamydia spp. are rather conserved, the characteristic disease manifestations resulting from infection by each species are unique. There are at least three species of Chlamydia that cause disease in humans. Chlamydia trachomatis causes both genital and ocular infections. It is the most common sexually-transmitted bacterial pathogen with an estimated 101 million cases worldwide in 2005 (WHO, 2011). The ocular disease, trachoma, is the world’s leading cause of infectious blindness, with more than 500 million people at high risk, 140 infected, and about 6 million with permanent blindness (WHO, 2013). Chlamydia pneumoniae is responsible for an acute respiratory tract infection (Grayston, et al., 1993). Additionally this species has been associated with stroke, atherosclerosis and Alzheimer’s disease (Saikku, et al., 1992, Wimmer, et al., 1996, Balin, et al., 1998). Psittacosis caused by the zoonotic agent, Chlamydia psittaci, is spread to humans by infected birds (Heddema, et al., 2006). Downstream sequelae include severe pneumonia and occasionally endocarditis and hepatitis.
Regardless of the species or resulting disease, infecting chlamydiae cause disease by impeding proper function of their host cell to aid in their own development. This includes altering the host cytoskeleton, inhibiting lysosomal fusion, sequestering host nutrients, and preventing or inducing apoptosis (Byrne & Ojcius, 2004, Ying, et al., 2007). Throughout the developmental cycle, all Chlamydia species are able to perform many of these functions along with disease-specific tasks using secreted virulence factors. Many of these factors, termed effectors, are translocated from the chlamydiae directly into the cytosol of the infected cell through type III secretion (T3S).
While these effectors likely play a critical role in Chlamydia pathogenesis, identification and therefore characterization of these proteins has been a challenge. Though many attempts have been made, the identification of a common, recognizable signal that targets these proteins for secretion has remained indistinct. Additionally, most chlamydial effectors lack sequence homology to other known effectors and do not reside in pathogenicity islands. In order to streamline the process of identification, a variety of methods for targeting putative effectors have been employed. While most effectors do not share similar sequences to other known effectors, a few are well-conserved across many species. These have been identified through homology searches. Additionally, the presence of characteristic secondary structures has aided in the identification of the Inc family of proteins, many of which have been demonstrated to be type III secreted. While these methods of searching have resulted in the identification of new effectors, they yield limited success since the majority of effectors are species-specific. That being the case, hypothetical proteins unique to Chlamydia have been targeted as potential effectors. Chlamydia proteins found within the cytosol of infected cells have also been examined for T3S. Regardless of the method used to identify putative effectors, these proteins must then be directly tested for T3S. Because the construction of a chlamydial isogenic T3S mutant-parent pair for Chlamydia remains elusive, currently the best method consists of using a T3S surrogate system.
Another means of streamlining the search for effectors is to use methods involving computational predictions of T3S effector proteins (Arnold, et al., 2009, Lower & Schneider, 2009, Samudrala, et al., 2009, McDermott, et al., 2011). These methods rely on obtaining data from the protein sequences of known secreted effectors and then applying the data to a machine-learning algorithm. The trained program can then be used to make predictions from an uncharacterized set of proteins. In our analyses we have focused on predictions made by the SVM-based Identification and Evaluation of Virulence Effectors (SIEVE) program (Samudrala, et al., 2009). This program was trained on known effectors from the plant pathogen Pseudomonas syringae by analyzing effector features such as sequence, conservation in other organisms, G+C content of the gene, amino acid composition, and the first 30 residues in the N terminus. Once trained, SIEVE’s accuracy was validated by predicting known secreted effectors from Salmonella enterica serovar Typhimurium. The method was also applied to predict effectors from C. trachomatis.
In this report we analyzed the accuracy of the Chlamydia effector predictions made by SIEVE first through experimental validation of T3S in a Yersinia pseudotuberculosis surrogate system to verify that a predicted protein is indeed a T3S substrate. To assess whether each T3S substrate identified is truly an effector (a virulence factor secreted into the host cell), we proceeded to characterize the protein during a chlamydial infection. Additionally, results from the Y. pseudotuberculosis surrogate system were also used to retrain SIEVE into cSIEVE, a Chlamydia-specific prediction program that provides new effector predictions that extend beyond C. trachomatis to other species of the genus. To improve the secretion assay results and to help eliminate false negatives, a construct permitting fusion of the 25 amino terminal residues of predicted effectors to a known secreted effector was used to alleviate the need for Chlamydia-specific chaperones in the Y. pseudotuberculosis surrogate system. Using this strategy, we report here the identification of several new chlamydial T3S substrates and potential effectors including a late-expressed, T3S substrate of C. trachomatis that associates in clusters at the chlamydial cell surface post secretion.
The full open reading frames or the 5′ ends of specified genes were amplified from purified C. trachomatis serovar D genomic DNA or C. psittaci Cal10 genomic DNA using primer sets with engineered restriction sites (Integrated DNA Technologies) and Platinum Taq DNA Polymerase High Fidelity (Invitrogen) as described by the manufacturer. All primers and restriction sites are listed in Supplemental Table 1. The resulting PCR amplicons were purified and restriction digested as described by the manufacturer (New England Biolabs). Full-length amplicons were cloned into the pFLAG-CTC vector (Sigma Aldrich), and the 5′ gene regions were cloned into pFLAG-CTC-694 (see below). The resulting plasmids were transformed and propagated in chemically-competent Escherichia coli TOP10 cells (Invitrogen). The plasmid constructs were purified using the Qiagen HiSpeed Plasmid Midi kit as directed by the manufacturer.
Amino-terminally truncated ct694 was amplified using primers containing engineered restriction sites XmaI and SalI, and cloned into the pFLAG-CTC vector as previously described. The resulting plasmid construct was named pFLAG-CTC-694. To test putative effectors in the new plasmid, the 5′ gene fragments corresponding to the first 25 amino acids of Chlamydia effector candidates were amplified using primers containing engineered restriction sites NdeI and XhoI and cloned into pFLAG-CTC-694 using the same restrictions sites. The resulting plasmids express fusion proteins that contain a putative N-terminal signal sequence linked to the C terminus (aa 46-323) of a known effector protein, CT694, lacking its own N-terminal signal sequence.
Purified pFLAG-CTC or pFLAG-CTC-694 plasmids containing inserts consisting of full-length or 5′ fragments of a putative Chlamydia effector or control genes were transformed into Yersinia pseudotuberculosis pIB29MEKBA (MEKBA) (Hakansson, et al., 1996) or YPIII pIB868 (ΔyscS) (Bergman et al., 1994). Transformants were cultured and T3S assays were performed as described by Hower et al (Hower, et al., 2009) with slight modification. In brief, cultures were grown at 26°C in LB medium containing either 5mM CaCl2 or 5mM EDTA and 20 mM MgCl2 to an OD600 of 0.2. The temperature was increased to 37°C, and 0.1 mM isopropyl-B-D-thiogalactopyranoside (IPTG, American Bioanalytical) was added to each culture. Growth continued for 4 hours. Pellet and supernatant samples were prepared from each culture for immunoblot analysis as described (Fields, et al., 2003).
Samples prepared from the heterologous T3S assay were subjected to SDS-PAGE in 12.5% polyacrylamide gels. The proteins were transferred to PVDF membranes by electroblotting. All membranes were blocked for 1 hour in blocking buffer (phosphate-buffered saline containing 0.2% Tween 20 and 5% nonfat dry milk). The immunoblots were screened with the primary antibody, rabbit anti-FLAG tag (1:5,000; Cell Signaling Technology) followed by an HRP-conjugated goat anti-rabbit IgG secondary antibody (1:25,000; Kirkegaard & Perry Laboratories) or primary antibody, mouse anti-β-lactamase (1:5,000; Thermo Scientific) followed by an HRP-conjugated goat anti-mouse IgG secondary antibody (1:25,000; Kirkegaard & Perry Laboratories). Bound HRP-conjugated antibody was detected using Supersignal West Dura Extended Duration Substrate (Thermo Scientific).
SIEVE was retrained using a set of 15 proteins we knew to be secreted in this or another similar surrogate system as positive examples for training (CT082, CT089, CT118, CT223, CT232, CT233, CT456, CT578, CT621, CT694, CT695, CT696, CT847, CT861, Cpn0809, Cpn1020, Cpn0677, CCA00062, this work and (Fields & Hackstadt, 2000, Clifton, et al., 2004, Lugert, et al., 2004, Fields, et al., 2005, Subtil, et al., 2005, Chellas-Gery, et al., 2007, Hobolt-Pedersen, et al., 2009, Hower, et al., 2009)) and the remainder of the proteins in the Chlamydia proteome (including those thought to be secreted effectors, but not secreted in the heterologous system) as negative examples. Leave-one-out cross-validation was used to obtain scores reported for all proteins as previously reported (Samudrala, et al., 2009).
The full-length ct082 gene was cloned into the pET-32 expression vector (Invitrogen). E. coli BL21 cells were transformed with the plasmid construct and protein expression was induced with IPTG. The recombinant CT082 protein contained an N-terminal His-tag that allowed for purification using TALON Metal Affinity Resin (Clontech) as described by the manufacturer. Two Hartley Strain female guinea pigs (Charles River, Wilmington, MA) weighing approximately 500g were immunized subcutaneously at four separate sites with 0.25 ml each of a 1:1 suspension of CT082 in PBS (200 μg/ml) and Freund’s complete adjuvant (Sigma, St. Louis). Two weeks and four weeks later, each animal was similarly immunized using Freund’s incomplete adjuvant. The animals were exsanguinated two weeks after the last booster immunization and the serum obtained. The immunization protocol was approved by the University of Arkansas for Medical Sciences Institutional Animal Care and Use Committee.
For immuno electron microscopy analysis, Chlamydia-infected cells were fixed in 4% paraformaldehyde, 0.1M PIPES buffer (pH 7.35) at various times post infection. Cells were scraped off the tissue culture vessel, washed, pelleted and enrobed in 2.5% Low Melting Temperature (LMT) agarose. Agarose blocks containing cells were trimmed into ~1 mm3 size, washed and dehydrated by using “progressive lowering of temperature” (PLT) technique as previously described (Gounon, 2002), infiltrated and embedded in unicryl at −20°C under UV from 24 to 48 hrs. Ultrathin sections were cut on a Leica UC6 ultramicrotome (Leica Microsystems, Inc., Bannockburn, IL) and collected onto formvar-coated Nickel grids. Immunogold labeling was performed as follows: grids were inverted section-side facing down onto a drop of blocking solution containing 5% BSA, 0.1% fish gelatin in PBS, pH 7.4, for 30 min. After blocking, grids were transferred onto a 10 μL droplet of primary antibody diluted in incubation buffer containing 0.2% acetylated BSA, 0.1% fish gelatin in PBS for 60min at room temperature. Grids were then washed five times in incubation buffer and incubated with 10 nm-gold-conjugated secondary antibody for 30 to 60 min, followed with five washes. Grids were then fixed with 2% glutaraldehyde in PBS for 5 min, rinsed with water, and contrasted with 1% uranyl acetate in 50% methanol. Grids were washed again with water, air dried and examined using a Technai T12 transmission electron microscope (FEI) at 80 keV. Images were acquired with an AMT digital camera.
Using Yersinia pseudotuberculosis as a T3S surrogate, we sought to experimentally validate 30 of the untested predictions made by SIEVE. Each full-length protein was expressed with a C-terminal Flag tag in the T3S-capable (MEKBA) or T3S-null mutant (ΔyscS) Y. pseudotuberculosis strains. Each transformant was cultured in medium replete or deplete of Ca2+, which suppresses or induces T3S, respectively. Upon completion of the assay, culture supernatants and cell pellets were analyzed via immunoblot analysis where the recombinant proteins were detected using anti-Flag tag antibody. To serve as a control for cell lysis, an identical immunoblot was screened with anti-β-lactamase antibody. Only effector candidates that were specifically detected in the supernatants of the MEKBA strain post-induction (temperature shift in low Ca2+) and absent in the ΔyscS mutant in both induced and non-induced conditions were considered positives. Using these rigorous criteria, of the 30 proteins tested only one, CT082, was identified as a T3S substrate (Figure 1). CT082 has been recently independently identified as a T3S substrate by Pais et al (Pais, et al., 2013).
The SIEVE algorithm integrates genomic data to accurately predict T3S substrates based on several different characteristics of their sequences: amino acid sequence biases, evolutionary relationships, C+G content, and the N-terminal amino acid sequence (Samudrala, et al., 2009). To allow better prediction of Chlamydia T3S effectors that could be secreted using the heterologous secretion system, we retrained SIEVE using a set of 15 proteins we knew to be secreted in this system as positive examples and the remainder of the proteins in the Chlamydia proteome as negative examples. The newly-trained program, cSIEVE is intended to provide more accurate effector predictions in Chlamydia trachomatis and to expand the predictions beyond effectors of C. trachomatis to all sequenced Chlamydia genomes. We show a better correlation between the scores assigned to secreted substrates versus those that are not secreted when the cSIEVE algorithm was used compared to when the original SIEVE was used (Figure 2A). The mean scores rendered by cSIEVE provide a better distinction between proteins that are secreted and those that are not; whereas there is not a significant difference between the mean scores supplied by SIEVE of secreted versus non-secreted proteins. Table 1 lists the top 100 predictions of SIEVE and cSIEVE in C. trachomatis. Using the cSIEVE predictions combined with the previously-described Y. pseudotuberculosis heterologous T3S assay, an additional chlamydial T3S substrate has been identified, C. psittaci Cal10 ORF70 (Figure 2B).
Effectors predicted by SIEVE and cSIEVE combined with positive results from the Yersinia surrogate secretion system led to the identification of two new T3S substrates, CT082 of C. trachomatis and ORF70 of C. psittaci. However, the relatively low rate of positive results remained. This led us to investigate the possibility that the surrogate host was not secreting all Chlamydia effectors leading to false negative results. Characterization of several T3S effectors from many bacteria has demonstrated the presence of chaperone-binding domains (Dean, 2011). Therefore, if any Chlamydia-specific chaperone is essential for secretion of a given T3S substrate, the surrogate host would be unable to provide this requirement, and the substrate will not be secreted. To circumvent this problem, we generated a construct that enabled us to test the secretion of putative effectors using only the first 25 amino acids of any target gene, which is thought to contain the secretion signal for T3S proteins (Figure 3A) (Dean, 2011, McDermott, et al., 2011). Our construct contains ct694 lacking its own signal sequence. Chimeric proteins were generated to allow for easier detection via western blotting. CT694 is a previously-described C. trachomatis effector that is secreted by the Yersinia surrogate (Hower, et al., 2009). Secretion of full-length CT694 by Yersinia indicates that it does not contain a Chlamydia-specific chaperone-binding domain. In the final plasmid construct, expressed CT694 is missing its first 42 amino acids to remove the endogenous secretion signal (‘CT694). After absence of ‘CT694 secretion by the Yersinia surrogate was confirmed (Figure 3B), we selected the first 25 amino acids of the well-known C. trachomatis effector CT089 (CopN) as a positive control to demonstrate that generating a chimeric protein with the secretion signal from a different effector is sufficient for secretion. The first 25 amino acids of CT743 (Hc1), a histone-like protein involved in condensing the chlamydial chromosome during RB to EB transition, was selected as a negative control in the assay (Figure 3B). The results of these experiments demonstrate that an N-terminal secretion signal is both necessary and sufficient for secretion of the fusion protein in our heterologous assay.
In order to compare the effectiveness of the newly-modified secretion assay with the assay screening full-length proteins, we rescreened several putative effectors that had been negative as full-length proteins again as fusion proteins (Figure 4). Interestingly, CT157, CT365 and CT875 tested positive as T3S substrates, indicating that screening the N-terminus of predicted effectors in our secretion assay is likely more effective at identifying T3S substrates than screening them as full-length proteins, possibly due to limitations in the capability of the Yersinia surrogate to secrete full-length Chlamydia T3S substrates.
To investigate whether CT082 is a T3S secreted substrate that also plays a direct role in pathogenesis through its targeting of subcellular or cytosolic targets (i.e. is an effector), we used antibody-based methods to visualize potential targets. An immunofluorescence assay using a highly-sensitive monospecific polyclonal antibody failed to detect the presence of CT082 in the cytoplasm of infected cells at the 4, 8, 16, 24, 32, and 48 hpi developmental times (not shown). At 1 hpi, the antibody stained EBs specifically, but did not stain immediately adjacent to infecting EBs as has been observed for TARP (Clifton, et al., 2004) or CT694 (Hower, et al., 2009). In addition, CT082-expressing transfected HeLa cells were unaffected in their growth properties 48 hrs post-transfection, suggesting that essential cell functions are not targeted (not shown). Immuno electron microscopy revealed clusters of CT082 associated with the chlamydial cell envelope at 24 and 42 hpi, consistent with the ct082 transcriptional profile (Belland, et al., 2003). The number and appearance of the clusters did not change over time; however several clustering patterns were observed. In Figure 5 (panels A and I) gold particles are arranged in a circle, while in other micrographs (panels B, G, J, and L), the gold particles are arranged in a short string extending from the surface or into the chlamydiae, or appearing to bridge 2 chlamydiae. In the majority of cases, the particles were arranged in doublets or triplets. Taken together these results suggest that CT082 may not be translocated to a eukaryotic target, but rather that it oligomerizes within a complex structure that remains associated with the chlamydial cell surface at late developmental times.
Type III secretion of effectors is essential to the virulence of many Gram negative organisms. Most proteins making up the T3S apparatus are conserved among species possessing a T3S system (Pallen, et al., 2005). In contrast to the structural proteins, the secreted effectors show little homology with one another from one species to the next. This aspect of effectors prevents the use of previously-characterized effectors from one species to aid in identification of effectors in another species. Difficulty in identifying Chlamydia effectors is compounded by the evolutionary distance between Chlamydia spp. and other Gram negative pathogens and the fact that the T3S genes in Chlamydia genomes do not reside in pathogenicity islands, but rather are dispersed throughout the genome. Regardless, identification and characterization of effectors is critical in that these proteins give direct insight into pathogenic properties of a bacterial species, which aids in our overall understanding of pathogen virulence. To circumvent these obstacles and assist in the identification of previously-uncharacterized effectors we have exploited a previously-described in silico predictive approach (Samudrala, et al., 2009).
Several machine-learning algorithms have been developed to help pinpoint potential T3S effectors (Arnold, et al., 2009, Lower & Schneider, 2009, Samudrala, et al., 2009, McDermott, et al., 2011). Collectively, the tested results of these algorithms conclude that T3S effectors are thought to be directed to the T3S injectisome by an N-terminal signal consisting of approximately 30 amino acids that does not consist of an immediately recognizable sequence. Regardless of the training method, each of these programs has proven successful in predicting secreted effectors.
SIEVE (SVM-based Identification and Evaluation of Virulence Effectors) is a computational technique that takes into account several features of known effector proteins and uses this data to teach the program to discriminate positive from negative examples. SIEVE was trained on a set of effectors from Pseudomonas syringae and then evaluated on Salmonella enterica Typhimurium (Samudrala, et al., 2009). After eliminating effectors with sequence similarity, Samudrala et al. reported a sensitivity of 90% and a specificity of 88% when used to predict effectors in S. Typhimurium. SIEVE was then retrained on the positive and negative examples from both P. syringae and S. Typhimurium and used to provide predictions in C. trachomatis. The top 10% of predictions were provided, which included 86 proteins. Of these predicted proteins, approximately 28% had been previously reported to be secreted experimentally or by another computational approach. Using the Yersinia T3S surrogate host, we tested an additional 30 putative effectors provided by SIEVE of which only one protein (CT082) was secreted in our assay (Figure 1).
Given the accuracy reported when SIEVE was originally used to predict Pseudomonas or Salmonella effector proteins, our low rate of positive results when screening the predicted C. trachomatis effectors was surprising. Since functional analysis through genetic manipulation of the chlamydial chromosome is not yet achievable in Chlamydia, researchers are limited to using heterologous systems to directly test secretion of putative effectors. Though our best option, these systems are likely not ideal for secreting all Chlamydia effectors. The unnatural levels of expressed effectors may have a deleterious effect on translocation. Additionally, these systems lack Chlamydia-specific chaperones that may be needed for the secretion of certain effectors. Chaperones are cytoplasmic proteins that bind to effectors to maintain proper effector conformation to allow for efficient translocation through the narrow pore complex and potentially direct effectors to the T3S injectisome (Page & Parsot, 2002). Therefore, more true effectors may have been tested in our original assay, which focused on full-length proteins, but were falsely reported as negative results. Another reason we may have witnessed relatively poor outcomes could be that there are fewer effectors in C. trachomatis than what has been predicted and what is typical of other pathogens. Lastly, the low number of Chlamydia T3S substrates identified may have been due to the inability of the SIEVE program to accurately predict Chlamydia effectors when trained with Salmonella or Pseudomonas effectors. If the protein features for a subset of Chlamydia effectors were drastically different from those found in other species, they may go undetected by SIEVE as it had been designed.
To address one of these concerns SIEVE was retrained on known Chlamydia T3S substrates. The new program was named cSIEVE. To determine if cSIEVE more accurately predicts secreted proteins, a comparison was made of mean scores from the cSIEVE and SIEVE algorithms for proteins secreted in the heterologous system versus those tested but not secreted. The results indicate that there is a discernible difference between the mean scores of secreted versus non-secreted proteins when the cSIEVE algorithm is used whereas the mean scores are less distinct between the sets of proteins when the SIEVE algorithm is used (Figure 2A). While cSIEVE outperforms SIEVE with respect to T3S of chlamydial effectors secreted by Y. pseudotuberculosis, the best scoring predictions provided by SIEVE and cSIEVE were not significantly different (Table 1). This indicates that cSIEVE assigns consistently low scores to effectors (SIEVE scoring assigns a low score to highest probability candidate) but can discriminate between those that are secreted in the heterologous system better than the original SIEVE. That both SIEVE and cSIEVE yield similar best scoring predictions based on a putative N-terminal signal suggests that the signal for Chlamydia effectors is not significantly divergent from that of effectors in other species. While much of the Chlamydia genome is distinct from other bacterial species, many genes involved in the process of T3S as well as the signal sequence for T3S effectors has remained relatively unchanged. This conclusion is consistent with what has been observed in effector signal sequences from various other species by in silico analyses (McDermott, et al., 2011). Since the cSIEVE algorithm was also applied to make predictions in many other genome-sequenced Chlamydia species, we have successfully identified an additional T3S substrate from C. psittaci Cal10, ORF70 (Figure 2B). Clearly, neither prediction algorithm is infallible. However the odds of identifying T3S substrates are greater when working with predictions than if genes were to be randomly tested throughout the genome.
An additional concern possibly contributing to false negative results is the potential unsuitability of the Yersinia surrogate T3S system to secrete chlamydial effectors. Among likely problems are the lack of Chlamydia-specific chaperones in the Yersinia surrogate host and the specific structural properties of the Yersinia channel, which may be unfavorable for the translocation of some heterologous effectors. We have taken steps to address each of these concerns. In lieu of introducing recombinant Chlamydia chaperones into this system, we opted to use an approach adopted in several other studies in which only the N terminus of potential effectors are tested for secretion. Several variations of this method have been previously applied (Subtil, et al., 2001, Subtil, et al., 2005, Zhou, et al., 2012). By using only the N terminus, the signal sequence is retained, while the functional and potential chaperone-binding domains of the protein are removed.
In our assay the N terminus of a candidate effector was fused to the Chlamydia trachomatis CT694 protein lacking its own N terminus (Figure 3). We chose this protein because it has been demonstrated to be efficiently secreted in this system without the need for Chlamydia-specific chaperones. CT694 is either secreted without the need for a chaperone or an endogenous Yersinia chaperone is able to take the place of its own natural chaperone. Additionally, the CT694 fusion increased the molecular weight of each recombinant protein, which facilitated detection via immunoblot. Using this technique, we found that several putative effectors that had previously tested negative for secretion as full-length proteins (CT157, CT365, CT875) were now readily secreted in the Yersinia system (Figure 4). This result further indicates that the N-terminal signal for chlamydial T3S is not unique and that T3S surrogate systems such a Yersinia T3S provide a suitable avenue for testing secretion of heterologous effectors. While screening the N terminus of each candidate appears to reduce the rate of false negatives, we realize this alternative method may consequently increase the rate of false positives. As such, any protein for which its N terminus is secreted via this method must then be analyzed directly in Chlamydia before it can be deemed an effector. In addition to providing insight on specific pathogenic functions, each new effector identified could be added to the training set to further improve the cSIEVE algorithm in an effort to more efficiently identify others.
Of the predicted effectors tested, C. psittaci ORF70 has been validated by the identification of its subcellular target in the host cell (Mojica & Bavoil, personal communication). However, further analysis of the T3S substrate CT082 have supported speculation made by Pais et al (Pais, et al., 2013) that while it is a type III secreted protein, it is not likely an effector. Our goal at the onset of this study was to identify T3S effectors via an N-terminal T3S signal, but in doing so we anticipate uncovering T3S substrates that do not function as effectors.
At both mid and late developmental times, CT082 is undetectable in the cytosol of infected cells, but rather it is associated with the chlamydial cell envelope. IFA analyses also confirm exclusive localization to the inclusion of infected cells. Immuno electron microscopy was applied to assess more accurately the location of CT082 within the chlamydial inclusion. At both 24 and 42 hpi CT082 exclusively resides on the chlamydial membranes of both RBs and EBs, respectively. Rather than dispersing evenly over the surface, the protein appears to congregate as isolated clusters, suggesting that it is part of a complex structure. Highly conserved orthologs of CT082 (80-99% identical) are present in all C. trachomatis serovars and C. muridarum while less conserved othologs are present in C. pneumoniae and C. caviae (approximately 40 and 50% identity, respectively). Current analysis is ongoing to determine if CT082 contributes to the assembled injectisome, if it interacts with unrelated molecules on the surface of Chlamydia, or if it functions alone. Although our analysis suggests a role of CT082 as a type III secreted substrate that may be a component of the T3S injectisome, it does not preclude a possible role of CT082 as an effector during infection, as cytosolic levels of effectors are often below detectability which may not be associated with overt pathogenic consequences.
To date, most Chlamydia effectors that have been identified and characterized derive from C. trachomatis. However, it is known from comparing effectors from various species that the majority of T3S effectors are species-specific. This feature of effectors necessitates tools to help identify putative effectors, which are scattered throughout Chlamydia genomes. We have chosen to use the SIEVE and subsequent cSIEVE algorithms for this purpose. In addition to facilitating the identification of Chlamydia trachomatis effectors, our effort to produce cSIEVE was also directed towards expanding the predictions to all species of Chlamydia in hopes of identifying both common and species-specific Chlamydia effectors, which will ultimately provide insight into the shared and unique pathogenic properties of each Chlamydia species. cSIEVE has been successful at the onset of achieving this goal. Improving predictive algorithms is an ongoing process. Each new effector identified can then be used to better train the algorithm that predicted it. As time goes on improvements to the algorithm paired with improvements to the heterologous secretion system will likely contribute to the identification of new T3S effectors in Chlamydia.
In total, we have been successful in identifying several new Chlamydia T3S substrates by testing in silico predictions directly with the heterologous T3S assay. Further proof of secretion during Chlamydia infection followed by the identification of subcellular targets is required to determine if the T3S substrates are indeed T3S effectors or if they are T3S substrates that play a distinct role in chlamydial biology. As expected, the Yersinia pseudotuberculosis surrogate system is not ideal for allowing the secretion of all Chlamydia effectors, but this issue may be resolved by testing only the N terminus of each putative effector. Using this technique, the rate of false negative results has decreased. Interestingly our data coincides with what others have witnessed in that the N-terminal T3S signal present on Chlamydia effectors is not significantly divergent from that of other species which allows for secretion via other systems. Our goal is to implement computer prediction methods to identify putative effectors from all sequenced Chlamydia genomes and then directly test for the secretion of these proteins using a modified secretion assay in a surrogate host. The results obtained from these tests can then be used to further improve the accuracy of the computer prediction method. Ultimately the characterization of newly-identified effectors will aid in our understanding of Chlamydia pathogenesis.
Research reported in this publication was supported by the National Institute of Allergy and Infectious Diseases (NIAID) of the National Institutes of Health (NIH) under award number U19AI084044 to RR, GM, JR and PB. JM was supported by the NIH-NIAID through interagency agreement Y1-A1-8401-01 and by grant NIH/NIAID A1022933-22A1. KH was supported in parts by NIH-NIDR T32 training grant no. DE007309. SM was supported by NIH-NIAID T32 training grant no. AI007540. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. CS was supported by the Summer Research Training Program of the UMB School of Dentistry.
PubMed Central Canada is a service of the Canadian Institutes of Health Research (CIHR) working in partnership with the National Research Council's national science library in cooperation with the National Center for Biotechnology Information at the U.S. National Library of Medicine(NCBI/NLM). It includes content provided to the PubMed Central International archive by participating publishers. |