CteG is a Chlamydia trachomatis effector protein that associates with the Golgi complex of infected host cells

Chlamydia trachomatis is a bacterial pathogen causing ocular and genital infections in humans. C. trachomatis multiplies exclusively inside host cells within a characteristic vacuole, from where it manipulates host cells by injecting them with type III secretion effector proteins. Here, we identified CteG as the first C. trachomatis effector associated with the Golgi. For this, C. trachomatis strains expressing candidate effectors fused to a double hemagglutinin (2HA) tag were constructed. Then, among these strains, immunofluorescence microscopy revealed that CteG-2HA was delivered into the cytoplasm of infected cells. Between 16–20 h post-infection, CteG-2HA mostly associated with the Golgi; however, CteG-2HA also appeared at the host cell plasma membrane, and at 30 or 40 h post-infection this was its predominant localization. This change in the main localization of CteG-2HA was independent of intact microfilaments or microtubules. Ectopic expression of different regions of CteG (656 amino acid residues) in uninfected cells revealed that its first 100 residues contain a Golgi targeting region. Although a C. trachomatis cteG mutant did not display a defect in intracellular multiplication, CteG induced a vacuolar protein sorting defect when expressed in Saccharomyces cerevisiae. This suggested that CteG might function by subverting host cell vesicular transport.

deubiquitinating enzymes 16,17 , which localize within the cytoplasm of host cells and that have not been shown to be T3S substrates, as well as chlamydial T3S substrates secreted into the inclusion lumen 18,19 . Some of the non-Inc chlamydial effectors localize at the inclusion membrane 17,[20][21][22] , at the host cell plasma membrane 22 , or at the host cell nucleus [23][24][25] , while others are membrane-associated 11,26 or have undefined localization.
In this work, following the identification of candidate chlamydial T3S substrates using Yersinia enterocolitica as a heterologous host 27,28 , we show that the C. trachomatis CT105 protein (CTL0360 in C. trachomatis serovar L2 strain 434/Bu; L2/434) is delivered into host cells during infection. In infected cells, bacterially-delivered CT105 initially mainly localized at the Golgi complex and then at the plasma membrane. CT105 is the first Chlamydia protein described to localize at the Golgi in infected cells, and we identified a Golgi-targeting region within its first 100 amino acid residues. Using Saccharomyces cerevisae as model, we also show that CT105 can modulate eukaryotic vesicular trafficking.

CT105-2HA is delivered by C. trachomatis into the cytoplasm of infected cells.
To test if the candidate chlamydial T3S substrates CT053, CT082, CT105, CT429, and CT849 27,28 can be transported by Chlamydia into the cytoplasm of host cells, C. trachomatis strain L2/434 was transformed with plasmids encoding these proteins with a double hemagglutinin (2HA) epitope tag at their C-termini. Protein production was confirmed by immunoblotting of extracts of HeLa cells infected for 40 h with C. trachomatis strains harboring plasmids encoding CT053-2HA (predicted molecular mass of 17 kDa), CT082-2HA (60 kDa), CT105-2HA (68 kDa), CT429-2HA (39 kDa), or CT849-2HA (18 kDa) (Figs 1A and S1). The strains producing CT053-2HA, CT082-2HA and CT105-2HA also showed species migrating on SDS-PAGE at a molecular mass different from the one predicted for the full-length proteins (Figs 1A and S1), as previously observed when identical 2HA-tagged versions of the proteins were produced in Y. enterocolitica 27,28 . Overall, these experiments confirmed that the constructed C. trachomatis strains expressed the expected 2HA-epitope tagged proteins.
Full-length orthologues of CT105 are only present in C. muridarum and C. suis, and ct105 expression is mostly restricted to C. trachomatis LGV strains. CT105 is a protein of 656 amino acid residues whose sequence does not show significant similarity to other proteins except for potential orthologues in other Chlamydia spp. However, full-length orthologues of CT105 were only found in C. muridarum and C. suis (Supplementary Table S1). In C. suis and in other Chlamydia spp. different open reading frames might encode proteins with some identity (between 30-22%) to only some parts of the amino acid sequence of CT105 (Supplementary Table S1). As previously noted 29 , analysis of the mRNA levels of ct105 by reverse transcription quantitative PCR (RT-qPCR) in different C. trachomatis strains revealed that the gene is only significantly expressed in LGV strains (serovars L1-L3) ( Supplementary Fig. S3). Furthermore, also as previously noted 29 , considering the promoter region of ct105 based on the transcription start site identified in C. trachomatis LGV strain L2b/UCH-1/proctitis 30 , C. trachomatis non-LGV strains (serovars A-K) lack 74 nucleotides upstream from the putative −10 region recognized by C. trachomatis σ 66 ( Supplementary Fig. S4). It has also been previously shown that ct105 is a pseudogene in C. trachomatis ocular strains (serovars A-C) 31 . Therefore, among C. trachomatis serovars, active CT105 is mostly produced by LGV strains.
Characterization of C. trachomatis expressing plasmid-encoded CT105-2HA. The L2/434 strain harboring pCT105-2HA, was then characterized by comparison to the parental strain. The profile of expression of ct105 during the developmental cycle of C. trachomatis was similar between the two strains ( Supplementary  Fig. S5). In agreement with previous observations 27 ( Supplementary Fig. S3), the highest levels of ct105 mRNA were detected at 2 h p.i. (Supplementary Fig. S5). Furthermore, at 2 h p.i., the strain harboring pCT105-2HA showed a 10-fold increase in the mRNA levels of ct105 relative to the parental strain ( Supplementary Fig. S5). However, expression of plasmid-encoded CT105-2HA in C. trachomatis had no significant impact on chlamydial growth ( Supplementary Fig. S5).
In extracts of HeLa cells infected by C. trachomatis L2/434 harboring pCT105-2HA, the 2HA-tagged protein could be detected by immunoblotting from 16 to 40 h p.i. (Supplementary Fig. S6). In addition to the band likely corresponding to full-length CT105-2HA (68 kDa), several faster migrating species of a lower molecular mass were consistently observed between 20 to 40 h p.i. (Supplementary Fig. S6). Similar observations were made by immunoblotting of extracts of HeLa cells infected by strain L2/434 harboring a plasmid (pTet-CT105-2HA) where production of CT105-2HA is driven by the tetracycline promoter (P tet ) ( Supplementary Fig. S6). This indicates that plasmid-encoded CT105-2HA is modified and/or degraded as the C. trachomatis developmental cycle progresses. Immunoblotting of extracts of HeLa cells infected by L2/434 harboring either pCT105-2HA or pTet-CT105-2HA, or of corresponding Chlamydia-enriched extracts, suggested that this modification and/or degradation of CT105-2HA occurs within the bacterium ( Supplementary Fig. S6).  Fig. S7). While at 16, 20, and 30 h p.i. CT105-2HA was detected in the host cell cytoplasm concentrated at one of the sides of the inclusion, at 30 h p.i. the protein was also detected at the periphery of the infected cells ( Supplementary Fig. S7). Furthermore, at 40 h p.i., CT105-2HA was mostly seen at the periphery of the infected cells ( Supplementary  Fig. S7).
The accumulation of CT105-2HA near the inclusion in the cytoplasm of infected cells suggested it could localize at the Golgi complex. This immunofluorescence signal of CT105-2HA also appeared less compact at longer times of infection, which was evocative of the Golgi fragmentation seen in C. trachomatis infected-cells 32 . To analyze this, HeLa cells infected by C. trachomatis L2/434 harboring pCT105-2HA were fixed at 16, 20, 30, or 40 h p.i. and then analyzed by immunofluorescence microscopy. This revealed that CT105-2HA localized in the Golgi region at 16, 20, and 30 h p.i., and that this was much less evident at 40 h p.i. (Fig. 2). There was not a perfect colocalization of CT105-2HA with cis-Golgi (GM130) and TGN (TGN46) markers, but the immunofluorescence signal of CT105-2HA near the inclusion accompanied the dispersion of the Golgi complex during infection (Fig. 2).
To further confirm the association of CT105-2HA with the Golgi, HeLa cells infected by the L2/434 strain harboring pCT105-2HA were treated with Brefeldin A (BFA), which induces reversible Golgi fragmentation 33 . For this, at 19 h p.i., the infected cells were either incubated for 1 h with BFA or with dimethyl sulfoxide (DMSO)-solvent alone and then fixed. In addition, the infected cells were incubated for 1 h with BFA, washed out from the BFA, incubated for 1 h and then fixed. Immunofluorescence microscopy revealed that in the DMSO-treated infected cells, CT105-2HA accumulated in the Golgi region (Fig. 3), similarly to untreated cells infected for 16 or 20 h p.i. (Fig. 2). In cells treated with BFA, the GM130-labeled Golgi fragmented into small vesicles and the immunofluorescence signal of CT105-2HA became dispersed and mostly below the limit of detection (Fig. 3). In cells treated with BFA and subsequently washed-out of the drug, both the GM130-labeled Golgi and CT105-2HA appeared mostly compact and near the inclusion (Fig. 3). Overall, this showed that CT105-2HA associates with the Golgi, at 16-20 h p.i., upon the initial detection of its delivery by C. trachomatis into the cytoplasm of infected host cells.
www.nature.com/scientificreports www.nature.com/scientificreports/ CT105-2HA localizes at the host cell plasma membrane in C. trachomatis-infected cells. CT105-2HA was also detected in the periphery of host cells infected by C. trachomatis at 30 or 40 h p.i. (Supplementary Fig. S7), which suggested it could localize at the plasma membrane. To analyze this, HeLa cells were infected for 30 and 40 h p.i. by C. trachomatis L2/434 harboring CT105-2HA and transfected with a plasmid encoding a palmitoylated/myristoylated peptide from Lyn kinase (Lyn11) [targeting proteins to the plasma membrane 34 ] fused to monomeric EGFP 35 (Lyn11-mEGFP). The cells were then fixed and analyzed by immunofluorescence microscopy. Although CT105-2HA did not show a perfect colocalization with either Lyn11-mEGFP or with phalloidin-stained cortical actin (Fig. 4A), this showed that at 30 and 40 h p.i., CT105-2HA localizes at the host cell plasma membrane.
The number of infected cells showing CT105-2HA outside of the inclusion and at the Golgi or at the plasma membrane were then enumerated (Fig. 4B), which confirmed that in C. trachomatis-infected cells, initial delivery of CT105-2HA into the host cell cytoplasm (first detected at 16 h p.i.) results in its accumulation at the Golgi region, but as infection progresses this localization of the protein becomes less frequent and CT105-2HA accumulates more often at the periphery of the cell. These observations were recapitulated by the analysis of HeLa cells infected by C. trachomatis L2/434 harboring pTet-CT105-2HA ( Supplementary Fig. S8). To test if CT105-2HA was transported from the Golgi to the plasma membrane in vesicles associated with the microtubule and/or actin cytoskeleton, HeLa cells were infected with C. trachomatis harboring pCT105-2HA. At 20 h p.i., when CT105-2HA mostly localizes at the Golgi region (Fig. 4B), the cells were incubated in the presence of DMSO solvent as a control, cytochalasin D or latrunculin B (to inhibit actin polymerization by different mechanisms), or nocodazole (to depolymerize microtubules) (Fig. 4C). The cells were then fixed at 30 h p.i., when CT105-2HA mostly localizes at the plasma membrane (Fig. 4B), and analyzed by immunofluorescence microscopy. We observed a statistically significant difference in the localization of CT105-2HA at the cell periphery when microfilaments were disrupted i., cells were fixed with 4% (w/v) PFA, immunolabeled with antibodies against HA (red), C. trachomatis MOMP (gray), GM130 (green), and TGN46 (blue), and appropriate fluorophore-conjugated secondary antibodies. In the area delimited by a white square (left-side panels) images were zoomed (middle and right-side panels). Immunolabeled cells were examined by confocal fluorescence microscopy, and images correspond to single z sections. Scale bars, 10 µm.
www.nature.com/scientificreports www.nature.com/scientificreports/ with cytochalasin D (Fig. 4C). However, the small difference relative to DMSO-treated cells, and the absence of a significantly different effect upon disruption of the microfilaments with latrunculin B (Fig. 4C), suggests this should have no biological significance. Furthermore, disruption of the microtubule network also did not significantly affect the localization of CT105-2HA at the cell periphery (Fig. 4C). Therefore, CT105-2HA changes its predominant localization during infection of host cells by C. trachomatis from the Golgi to the plasma membrane, but this is independent of intact host cell microfilaments and microtubules.
The first 100 amino acid residues of CT105 are sufficient to target the protein to the Golgi in uninfected mammalian cells. To map the regions of CT105 that could determine its different localizations within the cytoplasm of mammalian cells, HeLa cells were first transfected for 24 h with plasmids encoding full length CT105 fused to the C-or N-termini of monomeric EGFP (mEGFP-CT105 FL and CT105 FL -mEGFP, respectively) or with C-terminal 2HA-epitope tag. Immunoblotting of whole cell extracts confirmed the production of proteins of the predicted molecular mass ( Supplementary Fig. S9). Furthermore, fluorescence microscopy showed that mEGFP-CT105 FL , CT105 FL -mEGFP, or CT105-2HA localized predominantly at the periphery of the cell ( Supplementary Fig. S9), recapitulating a main localization of CT105-2HA in cells infected by C. trachomatis for 30 h or 40 h (Fig. 4B). Next, HeLa cells were transfected with plasmids encoding mEGFP-CT105 FL (predicted molecular mass of 96 kDa) and different truncated versions of CT105 fused to the C-terminus of mEGFP: mEGFP-CT105 1-100 (38 kDa), mEGFP-CT105 Δ100 (86 kDa) mEGFP-CT105 1-320 (60 kDa), mEGFP-CT105 Δ320 (65 kDa), or mEGFP-CT105 Δ100-320 (74 kDa) (Fig. 5A). As the primary structure of CT105 revealed no predicted transmembrane domains or other obvious targeting motifs, these truncations were designed only considering the predicted secondary structure of CT105 [deduced using JPred4 36 ]. Immunoblotting of whole cell extracts confirmed the production of mEGFP-CT105 FL , mEGFP-CT105 1-100 , mEGFP-CT105 Δ100 , or mEGFP-CT105 Δ100-320 proteins of the predicted molecular mass (Fig. 5B). The production of mEGFP-CT105 1-320 and mEGFP-CT105 Δ320 proteins of the predicted molecular mass was not detected (Fig. 5B), and they were not further analyzed. HeLa cells ectopically expressing mEGFP-CT105 FL , mEGFP-CT105 1-100 , mEGFP-CT105 Δ100 , or mEGFP-CT105 Δ100-320 were fixed and analyzed by immunofluorescence microscopy for localization of the fusion protein at the cell periphery (as an indication of plasma membrane localization) and at the Golgi. The mEGFP-CT105 FL protein localized only at the cell periphery in 68 ± 8% of the transfected cells, and only at the Golgi in 7 ± 2% of the transfected cells (Fig. 5C,D). However, the fluorescent signal was generally weak in the Golgi region (Fig. 5C). Strikingly, mEGFP-CT105 1-100 localized at the Golgi region in almost all transfected cells (98 ± 1%) and was not detected at the cell periphery (Fig. 5C,D). In contrast, mEGFP-CT105 Δ100 was mostly cytosolic and was never Fixed cells were immunolabeled with antibodies against HA (red), C. trachomatis MOMP (gray), and GM130 (green), and appropriate fluorophore-conjugated secondary antibodies. In the area delimited by a white square (left-side panels) images were zoomed (right-side panels). All immunolabeled cells were examined by confocal fluorescence microscopy, and images correspond to single z sections. Scale bars, 10 µm.
To confirm the association of mEGFP-CT105 1-100 with the Golgi, and to analyze its localization within the Golgi complex, HeLa cells that had been transfected for 24 h were either incubated for 1 h with BFA or DMSO solvent, and then fixed. In addition, the transfected cells were incubated for 1 h with BFA, washed out from the BFA, incubated for 1 h, and fixed. Immunofluorescence microscopy revealed that in the control DMSO-treated cells, mEGFP-CT105 1-100 localized at the Golgi region, and showed significant colocalization with TGN46 ( Fig. 5E) but not with GM130 ( Supplementary Fig. S9). BFA-induced Golgi fragmentation resulted in near complete dispersion of the fluorescence signal of mEGFP-CT105 1-100 (Figs 5E and S9). However, BFA-induced Golgi fragmentation followed by BFA wash out revealed again a compact Golgi complex and mEGFP-CT105 1-100 in that region mostly colocalizing with TGN46 ( Fig. 5E) but not with GM130 ( Supplementary Fig. S9). www.nature.com/scientificreports www.nature.com/scientificreports/ In summary, ectopically expressed full-length CT105 (mEGFP-CT105 FL , CT105 FL -mEGFP, and CT105-2HA) localized at the cell periphery and mEGFP-CT105 1-100 at the Golgi, generally recapitulating the localization of CT105-2HA in infected cells. As mEGFP-CT105 1-100 showed a striking association with the TGN in uninfected cells, this indicated that the first 100 amino acids of CT105 contain a Golgi-targeting region. Data are mean ± standard error of the mean of three independent experiments (n = 100). (E) Cells were treated for 1 h with DMSO or 1 µg/ml BFA. Then, the cells were either fixed with 4% (w/v) PFA (upper and middle panels) or washed with complete medium lacking BFA and incubated for an additional 1 h and then fixed with 4% (w/v) PFA (lower panels). The fixed cells were immunolabeled with antibodies against TGN46 (red), and the appropriate fluorophore-conjugated secondary antibody, and examined by confocal immunofluorescence microscopy. Images correspond to single z sections. In the area delimited by a white square (left-side panels) images were zoomed (right-side panels). Scale bars, 10 µm. www.nature.com/scientificreports www.nature.com/scientificreports/ CT105 is not essential for intracellular multiplication of C. trachomatis in infected tissue culture cells. To examine the importance of CT105 during infection of tissue culture cells by C. trachomatis, a L2/434-derived strain where the ct105 gene was inactivated by an insertion between nucleotides 261 and 262 of a modified group II intron containing a spectinomycin-resistance gene (aadA) was constructed ( Supplementary  Fig. S10). To confirm that insertion of the intron prevented production of CT105, in the absence of a specific anti-CT105 antibody, a C. trachomatis strain harboring a plasmid (pCT105::aadA-2HA) was constructed with the ct105::aadA mutant allele and the sequence encoding 2HA fused to the 3′ end of ct105. HeLa cells were then infected for 40 h with C. trachomatis L2/434 harboring either pCT105-2HA or pCT105::aadA-2HA and analysed by immunoblotting. This confirmed that intron insertion within ct105-2HA prevented production of a protein detectable with anti-HA antibodies (Fig. 6A).
The intracellular growth of C. trachomatis L2/434 (parental strain), ct105::aadA mutant, and ct105::aadA mutant harboring CT105-2HA (complemented strain) was then monitored by quantifying the production of infectious progeny during the chlamydial developmental cycle. Based on this, no significant differences between the three strains were detected (Fig. 6B). However, when HeLa cells infected by the same strains for 24 h were fixed and analysed by immunofluorescence microscopy, this revealed that the inclusion size was smaller in the mutant and complemented strains by comparison to the parental strain (Fig. 6C,D). Because this difference in inclusion size between the mutant and the parental strain was also observed in the complemented strain, the defect is likely not due to the absence of CT105. Probably, intron insertion within ct105 might have caused a polar effect in the neighbouring genes. In summary, this showed that CT105 is not required for C. trachomatis growth in tissue culture cells. In addition, considering the localization of CteG at the Golgi, we analysed if its morphology was altered in HeLa cells infected by the ct105::aadA mutant, by comparison to cells infected by the parental L2/434 strain. However, significant differences in Golgi morphology were not detected ( Supplementary Fig. S11).

CT105 induces a vacuolar protein sorting defect when ectopically expressed in S. cerevisiae.
Given the localization of CT105 at the Golgi and plasma membrane in both infected and transfected mammalian cells, we hypothesized that the protein could interfere with eukaryotic vesicular trafficking. To test this, we asked if CT105 could cause a vacuolar protein sorting (VPS) defect in S. cerevisiae 37 , a eukaryotic model organism that has been often used to study bacterial effector proteins 38 . The VPS assay consists in a yeast reporter strain NSY01 (Supplementary Table S2) that produces a hybrid protein composed of carboxypeptidase Y and Invertase (CPY-Inv), which normally travels to the yeast vacuole but goes to the cell surface if trafficking is disrupted. As strain NSY01 does not produce endogenous Invertase, an enzyme that hydrolyzes sucrose into glucose and fructose, normal (Vps + ) or aberrant (Vps − ) trafficking of CPY-Inv can be scored by using an agar overlay solution indicating glucose production at the cell surface by formation of a brown precipitate (Vps + , white colonies; Vps − , brown colonies) 37 .
Strain NSY01 was transformed with yeast expression plasmids encoding either CT105 fused to the N-terminus or C-terminus of GFP (CT105-GFP or GFP-CT105, respectively) under the control of a galactose inducible promoter (Fig. 7A). The first controls used were a NSY01 derivative strain that expressed only GFP (Fig. 7A), which leads to a Vps + phenotype 39 , and NSY01 derivative strains expressing a dominant-negative form of the yeast ATPase Vps4 (Vps4 E233Q ) 40 or the Legionella pneumophila effector protein VipA (Fig. 7A), both known to cause a Vps − phenotype 37,39 . In the qualitative colorimetric enzymatic assay in solid media, it was consistently observed that GFP-CT105 or CT105-GFP induced a Vps − phenotype (Fig. 7B). In addition, quantitative analyses in liquid media were performed and the amounts of secreted and total invertase were assessed for each strain. In this assay, both GFP-CT105 and CT105-GFP caused an increase in the levels of secreted invertase (relative to those of the control strain expressing GFP), but the difference was only statistically significant for GFP-CT105 (Fig. 7C). The C-terminal half of CT105 (amino acid residues 351 to 656), but not its N-terminal part (amino acid residues 1-443), was sufficient to induce a VPS defect in S. cerevisiae ( Supplementary Fig. S12), indicating that the VPS-inhibitory activity of CT105 depends on its C-terminal region. Moreover, production in yeast of a fusion to GFP of another C. trachomatis type III secreted protein (CT142 27 ) did not induce a VPS defect (Supplementary Fig. S12). Finally, production of GFP-CT105 or CT105-GFP caused no toxicity in S. cerevisiae NSY01 ( Supplementary Fig. S12), indicating that the induction of the VPS defect by CT105 was not a consequence of an overall impact on yeast physiology. Thus, ectopic expression of CT105 in S. cerevisiae disrupted vesicular trafficking to the yeast vacuole.

Discussion
We found that the C. trachomatis CT105 protein is delivered into the cytoplasm of infected host cells. At distinct times of the chlamydial developmental cycle, CT105 accumulates at the Golgi complex (16-30 h p.i.) and/or at the host cell plasma membrane (30-40 h p.i.). Because CT105 can be secreted by the T3S system of Y. enterocolitica 27 and can interfere with vesicular trafficking in S. cerevisiae (Figs 7 and S12), this indicates that CT105 is a newly identified C. trachomatis T3S effector. However, thus far, we have no direct evidence for the ability of CT105 to interfere with vesicular trafficking in infected human cells and this must be analyzed directly in future studies. Overall, this work expands the portfolio of known chlamydial effector proteins and reveals the first Chlamydia protein that associates with the Golgi complex. Moreover, it illustrates that a single C. trachomatis effector can have distinct subcellular localizations, and possibly different functions, during the chlamydial developmental cycle. Such dual localization during infection as a likely means to diversify effector function is evocative of the Salmonella SopB effector, which initially localizes at the host cell membrane to mediate bacterial invasion but is then redirected to the Salmonella-containing vacuole to modulate bacterial intracellular growth 41 .
In addition to CT105, the C. trachomatis effector CT867, a deubiquitinase enzyme known as ChlaDub2 16 or Cdu2 17 , has been detected at the inclusion membrane and at the host cell plasma membrane at distinct times of infection 22 . Such localization of CT105 and CT867/ChlaDub2/Cdu2 later in infection might indicate a function (2019) 9:6133 | https://doi.org/10.1038/s41598-019-42647-3 www.nature.com/scientificreports www.nature.com/scientificreports/ in host cell exit. It is unknown how CT105 and CT867/ChlaDub2/Cdu2 are differentially directed to different host cell localizations. In the case of CT105, it is unlikely that vesicular trafficking is involved as disruption of microfilaments and microtubules did not affect its localization at the plasma membrane in infected cells. We envision that host cytosolic CT105 is specifically directed to different subcellular localizations by covalent modifications it might be subjected to and/or by changes in the lipid and/or protein composition of host cellular membranes.
Ectopic expression of CT105 in yeast inhibited protein transport to the vacuole. While it remains to be directly tested, this observation suggested a potential capacity of CT105 to interfere with vesicular trafficking in infected cells. An appealing possibility for the function of CT105 would be that it contributes to nutrient acquisition by the inclusion and/or avoidance of fusion with hydrolytic compartments. However, the C. trachomatis ct105::aadA mutant did not reveal an obvious intracellular growth defect, which is not uncommon amongst C. trachomatis effector gene mutants characterized thus far 11,42,43 and might reflect redundancy in effector function or inexistence of adequate infection models 7 . Furthermore, a putative activity of CT105 on vesicular trafficking in infected www.nature.com/scientificreports www.nature.com/scientificreports/ cells could also affect cytokine secretion 44 or immune signaling 45 , which would not have an impact on chlamydial intracellular growth in tissue culture cells. Although the ct105 mutant revealed smaller inclusions relative to the parental strain (suggesting a possible defect in vacuolar membrane expansion), this could not be restored by reintroduction of ct105 in trans. The exact reasons for this need to be further analyzed, but the inability to complement intracellular growth-like phenotypes in intron insertional mutants of C. trachomatis was also recently described for ct101/mrcA 46 , while other specific phenotypes of this mutant strain could be complemented 46 .
The precise spatial and temporal subcellular targeting of bacterial effector proteins during infection is an essential feature of their function 47 . In other bacterial pathogens at least three other effector proteins have been shown to localize at the Golgi in infected host cells. These are EspI/NleA from Escherichia coli 48 , GobX from L. pneumophila 49 , and SteD from Salmonella enterica 50 . EspI/NleA interferes with coatomer protein II complex function, inhibits host cell protein secretion 51 , and disrupts intestinal tight junctions 52 . In addition, EspI/NleA inhibits NLRP3 inflammasome activation 53 . GobX displays E3 ubiquitin ligase activity, but how this promotes L. pneumophila infection is unknown 49 . Finally, SteD interferes with the activity of a host E3 ubiquitin ligase, leading to a reduction of surface-localized major histocompatibility complex class II molecules and suppression of T cell activation 50 . This illustrates that no specific function can be predicted for CT105 based on its localization at the Golgi.
The first 100 amino acid residues of CT105 are sufficient for Golgi localization when the protein is ectopically expressed in Hela cells. However, this region does not display obvious motifs found in other relevant eukaryotic or bacterial proteins, and the exact molecular determinants of the Golgi localization of CT105 remain to be identified. Other relevant questions are whether the same determinants are responsible for Golgi localization in infected cells and which characteristics of the full-length CT105 protein enable its targeting to the plasma membrane. Other effector proteins that localize at the Golgi in infected cells use distinct mechanisms directing them to the organelle 49,50,54,55 . In the case of EspI/NleA, Golgi targeting is partially mediated by a Postsynaptic Density95/Disc Large/Zonula Occludens-1 (PDZ) domain on its C-terminus 54 . CT105 also possesses a putative PDZ domain 56 , but the relevant residues are not involved in its subcellular localization after ectopic expression in mammalian cells (data not shown).
CT105 displays a striking variability within C. trachomatis serovars, revealing LGV-specific genetic and transcriptomic traits. This suggests that CT105 could have a specific function related with the unique characteristics of infections by these strains, which, unlike ocular and urogenital C. trachomatis strains, can infect mononuclear phagocytes and disseminate into regional lymph nodes 29,57 . It also indicates that features of the infection by LGV strains should be an evolutionary pressure to maintain an active ct105 gene. Furthermore, among Chlamydia . The vacuolar protein sorting (VPS) phenotype was analyzed using a sucrose overlay to assess activity of secreted invertase. A vacuolar protein sorting defect (VPS − ) leads to the formation of a brown precipitate. (C) S. cerevisiae NSY01 strains encoding the indicated proteins were grown in liquid medium in the presence of galactose (inducing conditions) and the relative activity of secreted invertase was quantified (see Experimental Procedures). Data are mean ± standard error of the mean of five independent experiments. P-values were obtained by one-way ANOVA and Dunnett post-test analyses relative to GFP (0% of relative secreted invertase; not shown); *statistical significant (P < 0.05); ns, not significant.
www.nature.com/scientificreports www.nature.com/scientificreports/ species, full-length orthologues of CT105 are only present in C. muridarum and C. suis, consistent with the fact that these species are the closer relatives of C. trachomatis 3 .
In summary, we identified CT105 as a T3S effector of C. trachomatis that localizes to the Golgi and plasma membrane in infected cells, and which can disrupt trafficking to the S. cerevisiae vacuole. Because CT105 is the first Chlamydia effector shown to associate with the Golgi complex, we propose to name it CteG (for C. trachomatis effector associated with the Golgi). The identification of CteG also leaves many questions to be addressed related to its function, subcellular targeting mechanisms, diversity, and specificity within C. trachomatis and among Chlamydia species.

Materials and Methods
Plasmids, primers, and DNA manipulation. The plasmids used in this work, their main characteristics and construction details, are described in Supplementary Table S3. The DNA primers used in their construction are listed in Supplementary Table S4. Plasmids were constructed and purified using standard molecular biology procedures. The accuracy of the nucleotide sequence of all the inserts in the constructed plasmids was confirmed by DNA sequencing. The backbone plasmids used in this work were: pSVP247 19 , a derivative of p2TK2-SW2 58 , was used to generate C. trachomatis expression plasmids encoding proteins with a 2HA at their C-termini; pmEG-FP-N1 and pmEGFP-C1 [both constructed from pEGFP-N1 (Clontech) and pEGFP-C1 (Clontech), respectively, where the gene encoding EGFP was replaced by a gene encoding mEGFP from pLAMP1-mGFP 35 ], and pEF6/ Myc-His A (Thermo Fisher Scientific), were used to generate mammalian transfection plasmids; pYES2-GFP 59 and pKS84 60 , were used to generate yeast expression plasmids encoding fusions to the C-or N-terminus of GFP, respectively. Furthermore, pBOMB4-Tet-mCherry 61 and Lyn11-FRB-mCherry 62 were used to amplify the P tet and the nucleotide sequence encoding the Lyn11 peptide, respectively.

Cells lines and transient transfection. HeLa 229 and Vero cells (from the European Collection of Cell
Culture; ECACC) were maintained in high-glucose Dulbecco's modified Eagle Medium (DMEM; Thermo Fisher Scientific) supplemented with heat-inactivated 10% (v/v) fetal bovine serum (FBS; Thermo Fisher Scientific) at 37 °C in a humidified atmosphere of 5% (v/v) CO 2 . Cells were checked for Mycoplasma by conventional PCR either using the Venor GeM Advance kit (Minerva Biolabs) or as described 63 . HeLa cells were transfected by using the jetPEI reagent (Polyplus-Transfection) according to the instructions of the manufacturer.
Bacterial strains and growth conditions. E. coli TOP10 (Thermo Fisher Scientific) or NEB 10β (New England Biolabs) were used for construction and purification of plasmids, and E. coli ER2925 (New England Biolabs) was used to amplify and purify plasmids for transformation of C. trachomatis. E. coli strains were routinely grown in liquid or solid lysogeny broth (LB) medium (NZYTech) with the appropriate antibiotics and supplements at 37 °C.
C. trachomatis prototype strains B/Har36, C/TW3, E/Bour, L2/434, and L3/404 (from ATCC) and clinical strains F/CS465-95 and L2b/CS19-08 (from the collection of the Portuguese National Institute of Health) were propagated in HeLa 229 cells using standard techniques 64 . Transformation of C. trachomatis was done essentially as described by Agaisse and Derré 58 . The optimal antibiotic concentrations to select transformants were 1 U/ml of penicillin G, or 250 µg/ml of spectinomycin. Once established, the transformed strains were cultured in the presence of 10 U/ml of penicillin, or 500 µg/ml of spectinomycin, and plaque purified using Vero cells, as described 65 . Infection of mammalian cells by C. trachomatis and quantification of infection progeny was done as previously described 19 . Infected cells were harvested at the indicated times post-infection and analyzed by immunoblotting or immunofluorescence microscopy.
Construction of a C. trachomatis ct105::aadA mutant strain. A C. trachomatis ct105::aadA mutant was generated using group II intron-based insertional mutagenesis, as previously described 66,67 . Briefly, intron-insertion sites in the C. trachomatis L2/434 ctl0360 gene (orthologue of ct105 in C. trachomatis strain D/UW3) were identified using the TargeTron algorithm (Sigma). Then, the intron in pDFTT3aadA 67 (Supplementary Table S2) was retargeted for ctl0360 using standard molecular biology procedures and the DNA primers listed in Supplementary Table S4. The ctl0360 mutator plasmid (pDFTT296; Supplementary Table S3) was then used to transform C. trachomatis L2/434. Table S2) used in this work were grown as previously described 37,39 . For the invertase assays, S. cerevisiae expressing GFP fusion proteins were grown in plates with yeast nitrogen base uracil dropout (YNB-Ura) supplemented with 2% (w/v) fructose for 3 days, at 25 °C. Cells were then streaked on YNB-Ura with 2% (w/v) fructose (non-inducing media) or 2% (w/v) galactose (inducing media) and grown for 3 days at 25 °C. Assays for qualitative and quantitative detection of invertase enzymatic activity were performed as described 37,68 . Relative secreted invertase was normalized by setting GFP as the minimum value and Vps4 E233Q as the maximum value.

Yeast strains and invertase assays. S. cerevisiae strains (Supplementary
Antibodies, fluorescent dyes, and drug treatments. The list of antibodies and fluorescent dyes used is described in Supplementary Materials and Methods. To induce Golgi fragmentation, HeLa 229 cells were incubated with 1 µg/ml of BFA (Sigma; stock solution at 5 mg/ml in DMSO) for 1 h and then either fixed immediately or washed with medium without BFA and incubated for 1 h before fixation. To disrupt microtubules, HeLa 229 cells were incubated in the presence of 1 µg/ml nocodazole (stock solution at 5 mg/ml in DMSO; Sigma). To disrupt the actin cytoskeleton, C. trachomatis-infected cells were incubated in the presence of 2 µM cytochalasin D (stock solution at 5 mg/ml in DMSO; Sigma) or 500 nM latrunculin B (stock solution at 1 mg/ml in DMSO; Sigma) in serum-free DMEM.