Specialized contact sites regulate the fusion of chlamydial inclusion membranes

Abstract

The intracellular bacterial pathogen Chlamydia trachomatis replicates within a membrane-bound compartment called the inclusion. Upon infection with several chlamydiae, each bacterium creates its own inclusion, resulting in multiple inclusions within each host cell. Ultimately, these inclusions fuse together in a process that requires the chlamydial protein IncA. Here, we show that inclusions form unique contact sites (inclusion contact sites, ICSs) prior to fusion, that serve as fusogenic platforms in which specific lipids and chlamydial proteins concentrate. Fusion depends on IncA clustering within ICSs and is regulated by PI(3,4)P₂ and sphingolipids. As IncA concentrates within ICSs, its C-terminus likely interacts in trans with IncA on the apposing membrane, securing a high concentration of IncA at fusion sites. This regulatory mechanism contrasts with eukaryotic or viral fusion systems that are either composed of multiple proteins or use a change in pH to initiate membrane fusion. Thus, our study demonstrates that Chlamydia-mediated membrane fusion is primarily regulated by specific structural domains in IncA and its local organization on the inclusion membrane, which is affected by the host cell lipid composition.

The bacterial pathogen Chlamydia trachomatis replicates within a vacuole known as an inclusion. Here, Linton et al. show that in multiply infected cells, apposing inclusions form unique membrane-contact sites that regulate membrane fusion.

Introduction

Membrane fusion occurs when two membranes merge into a continuous bilayer, allowing the contents of both compartments to mix^1,2. While fusion driven by eukaryotes and viruses is well described, membrane fusion mediated by bacteria is poorly characterized. One pathogen for which bacteria-driven membrane fusion is critical is Chlamydia trachomatis (Ctr). Ctr is an intracellular bacterium that replicates within a host plasma membrane-derived compartment called the inclusion. Inclusions are essential for Chlamydia’s survival, as they isolate the pathogen from the host cell cytoplasm, thus protecting it from the host defense. Importantly, inclusions also facilitate Chlamydia’s response to environmental and developmental changes³.

Within each host cell, the number of inclusions is tightly regulated. At a multiplicity of infection (MOI) > 1, each bacterium forms its own inclusion in which it begins to replicate, resulting in multiple cytoplasmic inclusions within a cell. Ultimately, these inclusions fuse into a single compartment between 12–16 h post-infection (hpi) through the process of homotypic fusion^4–6.

Homotypic fusion requires the chlamydial protein IncA, as IncA knockout (IncA_KO) Ctr cannot undergo fusion^4–6. Clinical isolates that lack IncA or express mutated IncA on the inclusion are also non-fusogenic^7,8. Interestingly, these non-fusogenic isolates are more frequently associated with subclinical infections compared to their fusogenic counterparts^7,9. While these isolates can establish infection, infected patients present with lower bacterial burdens, fewer symptoms, and fewer signs of infection, likely resulting in less tissue damage than those infected with fusogenic strains⁹. In vitro, one of these isolates also grows slower than its fusogenic counterparts¹⁰. Notably, only ~1.5% of the clinical isolates successfully cultured are non-fusogenic, suggesting that homotypic fusion may undergo positive selection within the human population. Alternatively, other non-fusogenic strains may elicit asymptomatic infections more frequently. Thus, patients do not seek treatment, and the strains are not detected. These observations suggest that IncA-mediated homotypic fusion supports Ctr pathogenicity^7–9.

IncA is expressed on the inclusion surface around 10–12 hpi¹¹. It contains a bilobed transmembrane domain (TMD) flanked by two cytosolic regions, a short N-terminal tail (N-tail), and a large C-terminal domain. Evidence suggests that IncA is a bona fide bacterial fusion protein, but the molecular mechanism of IncA-mediated fusion has yet to be established. This membrane fusion event is unique as (i) IncA is expressed on the entire inclusion surface, while fusion only takes place in specific regions, and (ii) IncA is identical on both membranes, yet there is no known activation step to control fusion. In contrast, the eukaryotic soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) fusion machinery requires a target (t)-SNARE complex on one membrane and a distinct but complementary vesicular (v)-SNARE protein on the other, allowing fusion to occur only when cognate v- and t-SNAREs interact^12–14. Viral fusion proteins are also single proteins, but they require activation by a change of pH or binding to a co-receptor to promote fusion¹⁵. These observations underscore how unique this bacterial system might be compared to other fusion machinery and raise an intriguing question: how is IncA regulated so that inclusions fuse with the appropriate compartment at the appropriate time?

Previously, we determined the atomic structure of the C-terminal domain of IncA and showed that it forms a non-canonical four-helix bundle encompassing the helices H_a, H_b, H_c, and H_d¹⁶. The crystal structure of IncA also revealed intramolecular interactions, which ensure that IncA remains in a folded monomeric conformation on the inclusion surface. In particular, IncA encodes a small clamp, H_clamp, which forms several intramolecular contacts with H_d that are central for controlling its monomeric state and fusogenic activity¹⁶. Ultimately, IncA oligomerizes using H_c and H_d, an event that is required for fusion¹⁷. Altogether, these data suggest that the activity of IncA requires robust intramolecular regulation to prevent cis oligomerization on one membrane while remaining primed for fusion in trans with the apposing membrane. How IncA switches between its cis monomeric and trans oligomeric forms to mediate fusion is unknown.

Here, we show that inclusions form unique membrane contact sites that we named inclusion contact sites (ICSs). ICSs form before fusion and are enriched in specific chlamydial proteins, including IncA and IPAM (CT223). ICSs also have a unique lipid composition compared to the rest of the inclusion membrane. Interestingly, inclusions form multiple membrane domains that serve distinct functions: microdomains containing phosphorylated forms of the Src kinase (pSrc) family are involved in the trafficking of inclusions to the microtubule-organizing center (MTOC)¹⁸, while inclusion-endoplasmic reticulum (ER) contact sites, that are enriched in the lipid transfer protein CERT, mediate lipid transfer to the inclusion¹⁹. Here, we establish that ICSs are distinct from pSrc-microdomains and ER-inclusion contact sites, and therefore constitute a novel membrane domain on the surface of the inclusion. In fact, we found that these ICSs form unique fusogenic platforms that control homotypic fusion.

IncA clustering in ICSs is critical for fusion, and preventing the enrichment of IncA significantly delays fusion despite its presence on the rest of the inclusion membrane. IncA clustering is controlled by both its C-terminal domain and the cellular lipid composition, notably PI(3,4)P₂ and sphingomyelin. Importantly, IncA must be present on both inclusions to cluster, likely so that IncA on apposing membranes can interact in trans at the level of their C-termini. Such a requirement for IncA topology precludes the inclusion from fusing with the cellular organelles since they do not express IncA on their surface, thus adding yet another regulatory layer to this process. Together, these results indicate that the local membrane environment is critical in controlling IncA-mediated homotypic fusion during Chlamydia infection. This is the first regulatory process identified for this bacterial fusion system.

Results

Inclusions form unique membrane contact sites

A hallmark of eukaryotic cells is the sophisticated endomembrane system that creates physically and functionally distinct organelles. Organelles frequently form local membrane contact sites (CSs) prior to fusion^20,21. Specific proteins and lipids cluster within these CSs, thus confining membrane fluidity and curvature changes to a defined region of the membrane. These sites are important for locally regulating fusion^20,21. To investigate the mechanism of homotypic fusion during Ctr infection, we first determined whether inclusions form CSs enriched in specific chlamydial proteins and lipids before fusion (Fig. 1a).

Fig. 1. Inclusions form unique contact sites (ICSs).

a Schematic depicts ICSs as points of contact between inclusion membranes in which specific lipids (brown) and proteins (green and blue) are clustered (see insert). Created in BioRender. Paumet, F. (2023) BioRender.com/s94s603. b–d HeLa cells were infected with IncA_KO Ctr (MOI 3). At 2 hpi, cells were transfected with plasmids expressing Akt-PH-GFP (b, green) or Btk-PH-GFP (c, green). Cells were then fixed at 24 hpi and labeled with anti-IPAM antibody (red). DNA was stained with Hoechst (blue). Enlarged ICSs (white box) are shown in the Zoom panel. Scale bar = 20 μm. Images are representative of three independent experiments. d ESs were calculated and plotted for each probe and IPAM as a control. Each dot = individual ES, black horizontal bar = mean. Gray box denotes ESs <2. n = 150 total scores per probe from three independent experiments. ns = not significant, **** denotes p-value < 0.0001 (one-way ANOVA). e–g IncA_KO Ctr infected cells were fixed at 24 hpi and labeled with anti-IPAM (green) and anti-pSrc (e), anti-IncB (f), or anti-CERT (g) antibodies (red). DNA was stained with Hoechst (blue). Enlarged ICSs (white box) are shown in the Zoom panel. Scale bar = 10 μm. Asterisks mark the inclusions. Images are representative of three independent experiments.

Phosphoinositides (PIs) are phosphorylated derivatives of the lipid phosphatidylinositol. Despite constituting <1% of cellular lipids, PIs play a disproportionally prominent role in regulating membrane dynamics²². Local PI metabolism allows for the recruitment of specific proteins to intracellular membranes that create functionally distinct domains²³. PIs co-segregate with other lipids, such as cholesterol and sphingomyelin, which are essential for lipid raft formation²⁴. Significantly, PIs also influence membrane curvature and interact with the eukaryotic tethering and fusion machinery, ultimately impacting the fusion of cellular membranes²⁵. Here, we tested several fluorescent lipid probes that bind specific PIs. HeLa cells were transfected with the probes following infection with IncA_KO Ctr at an MOI of 3 to promote the formation of multiple inclusions. In these conditions, inclusions cannot fuse, and ICSs become enriched and readily detectable. At 24 hpi, IncA_KO Ctr–infected cells were stained with an antibody raised against another chlamydial inclusion membrane (Inc) protein, IPAM, to label the inclusion membrane^26,27.

Akt1-PH, which binds PI(3,4,5)P₃ and PI(3,4)P₂, was found to be enriched within ICSs in both fixed (Fig. 1b) and live cells (Fig. S1a), indicating that PI(3,4,5)P₃ and/or PI(3,4)P₂ clusters within these sites. To discriminate between these two lipids, we used the probe Btk-PH, which targets only PI(3,4,5)P₃. Figure 1c and S1b show that the Btk-PH probe is not concentrated in ICSs, suggesting that PI(3,4,5)P₃ does not cluster in ICSs. Using anti-PI(3,4)P₂ antibodies, we confirmed that PI(3,4)P₂ clusters in ICSs (Fig. S1c, d). To assess the degree of PI(3,4)P₂ clustering, the enrichment score of Akt1-PH versus Btk-PH in ICSs was quantified by measuring the ratio of their fluorescence intensities at ICSs versus the periphery of the apposing inclusions (See Fig. S2 for the description of the strategy). An enrichment score (ES) of 2 represents the sum of two membranes, an ES > 2 indicates protein enrichment, and an ES < 2 denotes exclusion from ICSs. Using this strategy, ES_Akt-PH scores 3.6 on average, indicating that this probe is enriched in ICSs, as opposed to ES_Btk-PH, which only scores an average of 1.5, indicating that this probe is depleted from the ICSs (Fig. 1d). Akt1-PH clustering is specific as PLCδ-PH labeling PI(4,5)P₂ (Fig. S3a), FAPP1-PH labeling PI4P (Fig. S3b), and EEA1-FYVE labeling PI3P (Fig. S3c) are not enriched in ICSs. Surprisingly, we observed that the chlamydial inclusion protein IPAM also strongly localizes within the ICSs (Fig. 1b, c, red) with an ES_IPAM scoring an average of 10.6 (Fig. 1d). IPAM exclusively and consistently labels ICSs, regardless of the cell type infected or the Ctr strain used (Fig. S4a–b). Therefore, we used IPAM as an ICS marker throughout this study.

Inclusions form specific microdomains in which cholesterol concentrates along with active members of the Src kinase family and chlamydial proteins, including IPAM and IncB¹⁸. These microdomains serve as nodes for interactions between dynein and the inclusions and allow the inclusions to migrate to the MTOC to complete the developmental cycle^18,28. To determine whether ICSs constitute a variant of microdomains or are, in fact, unique, we labeled IncA_KO Ctr-infected cells with anti-pSrc and anti-IncB at 24 hpi. In parallel, cells were fed TopFluor-cholesterol at 4 hpi. Neither pSrc (Fig. 1e) nor IncB (Fig. 1f) are localized at the ICSs, and while cholesterol is present, it is not enriched (Fig. S3d), thus indicating that ICSs are distinct from chlamydial microdomains.

Inclusions also form contact sites with the ER within which CERT concentrates to facilitate lipid transfer between the ER and the inclusion¹⁹. As shown in Fig. 1g, CERT is excluded from the ICSs, indicating that ICSs also do not overlap with ER-contact sites. Altogether, these data show that the ICS constitutes a novel membrane domain on the inclusion, which likely has a distinct function.

IncA clusters in ICSs before fusion

Since IncA is critical for homotypic fusion^4–6, we tested whether this chlamydial protein clusters in ICSs prior to fusion. Wild-type (WT) Ctr expresses all the elements necessary for fusion. As such, these inclusions undergo fusion as soon as IncA is expressed ~10–12 hpi, making it a challenge to enrich and visualize ICSs. Thus, to delay fusion and increase the number of ICSs, we used nocodazole, which blocks microtubule polymerization and delays homotypic fusion²⁹. We first infected cells with WT Ctr for 5 h to allow the infection to proceed normally. We then added nocodazole to the infected cells for 16 h, before fixing and labeling them with anti-IncA antibody (Fig. S5). Anti-CT813/InaC (hereafter referred to as CT813) antibody, which labels an Inc protein involved in cytoskeletal remodeling^30–32, was used to mark the inclusion membrane, while anti-IPAM antibody was used to mark the ICSs. As expected, 46% of infected cells treated with nocodazole display multiple inclusions at 21 hpi (Fig. S6b, c) compared to 11% in DMSO-treated cells (Fig. S6a, c). In these conditions, endogenous IncA is enriched in ICSs, where it colocalizes with IPAM (Fig. 2a, Fig S6b). CT813, on the other hand, is excluded from these points of contact (Fig. 2a, green). Although the upper 10% values average 2.8 (Fig. 2b), the overall average ES_IncA is modest at 2.1, reflecting that only a limited number of IncA-enriched ICSs can be observed in the overall nocodazole-treated cell population. This observation is consistent with ICSs rapidly undergoing fusion once IncA reaches a critical threshold. Furthermore, nocodazole asynchronously delays but does not block homotypic fusion, resulting in significant variation in IncA clustering at each time point, i.e., individual ES_IncA scores vary from 1.1 to 7.3. In contrast, CT813, which is involved in cytoskeletal reorganization but has no role in homotypic fusion^30–32, displays an ES_CT813 consistently ranging from 0.5 to 2.5, with an average of 1.5, indicating that this protein is primarily excluded from ICSs. In contrast, IPAM, which predominantly localizes in the ICS, displays an average ES of 5.7, indicating significant enrichment in ICSs before fusion. Similar to ES_IncA, though, ES_IPAM also shows a substantial variation from 2 to 32, averaging lower than the usual ES_IPAM average of 11.5 (Fig. S4a), further supporting that ICSs asynchronously and transiently form in nocodazole-treated cells.

Fig. 2. The bacterial fusion protein IncA clusters in ICSs.

a HeLa cells were infected with WT Ctr (MOI 3) for 5 h to allow the infection to proceed before being treated with 20 μM nocodazole. Cells were then fixed at 21 hpi and labeled with anti-IncA (magenta), anti-IPAM (cyan), and anti-CT813 (green) antibodies. DNA was labeled with Hoechst (blue). Enlarged ICSs (white box) are shown in the Zoom panel. Asterisks denote inclusions. Scale bar = 10 µm. Images are representative of three independent experiments. b ESs were calculated for each Inc. Each dot = individual ES, black horizontal bar = mean. Gray box denotes ESs <2. N = 130 total scores per Inc from three independent experiments. ns = not significant, **** denotes p-value < 0.0001 (matched-paired, one-way ANOVA). c Anti-FLAG-Frankenbody_mScarlet3 HeLa cells were infected with IncA_KO Ctr complemented with IncA_WT-FLAG (MOI 3). IncA_WT-FLAG expression was induced at 5 hpi with 20 ng/mL anhydrotetracycline. z-stacks of infected cells were acquired every 5 min as described in the Methods. Individual frames are shown beginning at 17 hpi (0 min). IncA images show the enrichment of the anti-FLAG Frankenbody_mScarlet3 signal at the ICS prior to the fusion of the two inclusions (zoom). The HiLo look-up table (HiLo LUT) was then applied to the IncA images to highlight the grayscale differences (low signal = blue; high signal = red). Scale bar = 5 µm. Blue arrows denote signal enrichment at the ICS. Images are representative of three independent experiments. d HeLa cells were infected with WT Ctr (MOI 3), fixed at 18 hpi, and labeled with anti-IPAM (cyan) and -IncA (magenta) antibodies. DNA was labeled with Hoechst (blue). Enlarged ICSs (white box) are shown in the Zoom panel. Asterisks denote inclusions; Scale bar = 10 μm. Images are representative of three independent experiments. e HeLa cells were infected, fixed, and labeled as in d. ESs for each Inc was measured. N = 15 total scores per Inc from three independent experiments. Gray box denotes ESs <2. ns = not significant, **** denotes p-value < 0.0001 (matched-paired, one-way ANOVA).

To overcome this limitation and further observe IncA_WT enrichment in infected cells prior to fusion, we developed a live cell imaging assay. To do so, HeLa cells were transfected with a FLAG-specific intracellular antibody, called a frankenbody, fused to the fluorescent protein mScarlet3. This anti-FLAG frankenbody is a chimeric single-chain variable fragment that binds to FLAG epitopes in living cells³³ (Fig. S7a). Frankenbody-transfected cells were then infected with IncA_KO Ctr expressing IncA_WT-FLAG and GFP¹⁶. As IncA_WT-FLAG is expressed during infection, the frankenbodies label the FLAG-tagged protein on the inclusion surface in living cells (Fig. S7a). Using this strategy, we monitored the distribution of IncA_WT-FLAG on the inclusion membrane over time and observed that IncA concentrates in ICSs until fusion occurs, ~10 min after initial contact (Fig. 2c, Fig. S7b, and Supplementary Movie 1). This clustering was consistently observed in infected cells before inclusion fusion.

Finally, we were able to identify instances of endogenous IncA clustering at early time points during infection (18 hpi) in fixed cells infected with WT Ctr (Fig. 2d). In these conditions, ES_IncA is an average of 3.4. As expected, IPAM is also consistently enriched with an average ES_IPAM of 6.5, while CT813 is absent with an average ES_CT813 of only 1.6. (Fig. 2e). These snapshots of IncA clustering were rare in WT Ctr–infected cells as fusion occurs rapidly after IncA reaches a critical level of enrichment in ICSs. Together, these data from multiple complementary methods demonstrate that IncA clusters in ICSs, further supporting its relevance to the fusion process.

The clustering of IncA in ICSs is necessary for fusion and requires its C-terminus

Since inclusions form ICSs in which IncA concentrates, we next assessed whether this clustering is required for homotypic fusion. IncA is an Inc protein that encodes a bilobed transmembrane domain (TMD), a small 35 amino-acid N-terminal tail (N-tail), and a large C-terminus composed of four helices H_a-d (Fig. 3a)¹⁶. Interestingly, the C-terminal region of IncA switches between a stable monomeric four-helix bundle and oligomers^16,34. In this context, IncA clustering within ICSs could be controlled by one of its protein domains. To test this possibility, we used the Ctr mutant IncA_polyA, which expresses an IncA protein in which the helix H_d is mutated, thus disrupting the monomeric four-helix bundle (Fig. 3b)¹⁶. As seen in Fig. 3c, IncA_polyA does not efficiently cluster within ICSs during infection, with an ES consistently between 1.0 and 3.3 and averaging only 1.8 (Fig. 3d). IPAM and CT813 (Fig. 3c, d and Fig. S8a) remain clustered and excluded from the ICSs, respectively, suggesting that the effect is specific to IncA. The failure of IncA_polyA to concentrate at ICSs at 24 hpi correlates with a homotypic fusion defect (Fig. 3h, blue) since 69% of IncA_polyA Ctr-infected cells display multiple inclusions at 24 hpi, compared to 3% of WT Ctr-infected cells. Fusion eventually occurs by 40 hpi as 90% of IncA_polyA Ctr-infected cells contain fused inclusions- close to that of WT Ctr-infected cells (Fig. 3i, blue). These results suggest that the C-terminus of IncA significantly contributes to its clustering within ICSs, which correlates with the timing of homotypic fusion.

Fig. 3. The C-terminal domain of IncA controls IncA clustering in ICSs and regulates homotypic fusion.

a Schematic of full-length IncA, displaying an N-terminal tail (N-tail) and transmembrane domain (TMD), as well as four C-terminal helices (H_a-d). The N-tail and the C-terminal domain face the host cytoplasm, while the TMD anchors IncA in the inclusion membrane. Created in BioRender. Paumet, F. (2023) BioRender.com/z33h938. b Schematic of IncA_polyA-FLAG in which H_d residues were mutated to alanines to disrupt critical intramolecular contacts to create the IncA_polyA-FLAG complemented mutant strain¹⁶. Created in BioRender. Paumet, F. (2023) BioRender.com/u70l441. c HeLa cells were infected with IncA_polyA-FLAG Ctr (MOI 5) for 24 h in the presence of anhydrotetracycline. Cells were fixed and labeled with anti-FLAG (magenta, labels IncA_polyA-FLAG), anti-IPAM (cyan), and anti-MOMP (green) antibodies. DNA was stained with Hoechst (blue). Asterisks denote the inclusions; Scale bar = 10 μm. Images are representative of three independent experiments. d ESs for cells infected with IncA_polyA-FLAG Ctr at 24 hpi. N = 150 total scores per Inc from three independent experiments. Gray box denotes ESs <2. ns = not significant, **** denotes p-value < 0.0001 (one-way ANOVA). e Schematic of IncA_chimera-FLAG in which IncA N-tail and TMD were swapped for those of CT813, to create the IncA_chimera-FLAG complemented mutant strain. Created in BioRender. Paumet, F. (2024) BioRender.com/f62h052. f Cells were infected with IncA_chimera-FLAG Ctr (MOI 5) for 24 h, fixed, and labeled as in c. Asterisks denote inclusions; Scale bar = 10 μm. Images are representative of three independent experiments. g ESs were calculated as in d. **** denotes p-value < 0.0001 (one-way ANOVA). h, i Homotypic fusion was quantified for the indicated strains in infected cells (MOI 7) at 24 hpi (h) and 40 hpi (i). Graphs depict the average number of cells containing multiple inclusions +/- SD from three independent experiments. ns = not significant, * denotes p-value < 0.05, **** denotes p-value < 0.0001 (one-way ANOVA).

The organization of proteins in membrane domains can also depend on their TMD through interactions with specific lipids or proteins. To exclude the involvement of IncA’s TMD in clustering, we used CT813, which is consistently excluded from ICSs in all the conditions tested (Fig. S4c–d). The N-tail/TMD of IncA was swapped for that of CT813 while keeping the C-terminus of IncA intact, thus creating IncA_chimera (Fig. 3e). We then transformed IncA_KO Ctr with this construct before infecting HeLa cells with the resulting mutant. As seen in Fig. 3f, IncA_chimera is expressed on the inclusion surface, where it clusters in ICSs, with an ES ranging from 1.8 to 10.9 with an average of 4.6 (Fig. 3g). These results indicate that the N-tail/TMD of IncA is not involved in the clustering of IncA in ICSs and further supports the importance of its C-terminus in this process. ICSs in IncA_chimera-infected cells appear otherwise unchanged as the phospholipid composition is the same as IncA_KO Ctr-infected cells (Fig. S9a–e). Similarly, IPAM remains clustered (Fig. 3f, g and S8b), and cholesterol is present but not enriched (Fig. S9f). Finally, CT813 remains excluded from the ICS (Fig. 3g and S8b), like IncB, pSrc, and CERT (Fig. S10a–c).

Interestingly, even though IncA_chimera is clustered in the ICS, homotypic fusion is completely blocked, similar to the IncA_KO Ctr, with ~100% of the infected cells still displaying multiple inclusions at 24 hpi and 40 hpi (Fig. 3h, i, red). Note that because fusion is blocked, IncA_chimera can accumulate over time in the ICS, resulting in an ES_IncAchimera that is significantly higher than ES_IncAWT (Fig. 2b), which undergoes fusion as soon as a critical threshold is reached. Altogether, these results demonstrate that (1) the N-term/TMD region of IncA is essential for membrane fusion but does not play a role in the clustering of IncA in ICSs, (2) the C-terminus of IncA is critical for clustering, and (3) a defect in IncA clustering correlates with delayed fusion.

The binding of IncA C-termini in trans is essential for IncA clustering

Thus far, we have demonstrated that the C-terminus of IncA is necessary for its clustering in ICSs. However, the mechanism by which this occurs is yet to be identified. One possibility is that the C-terminal domain of IncA promotes clustering by binding specific lipids or proteins enriched in ICSs and drives IncA clustering in cis. Another possibility is that the C-termini of IncA proteins interact with one another in trans to form multimers^16,35. Discriminating between IncA oligomerization in trans versus cis using immunoprecipitation is challenging since both types of interactions would result in the same readout¹⁶. To circumvent this limitation, we used a co-infection system. If IncA clustering does not occur when IncA is expressed on only one inclusion, then trans-oligomerization is required (Fig. 4a). However, if IncA still clusters in the absence of IncA on the opposite membrane, then cis-oligomerization is involved (Fig. 4b). To assess these possibilities, we co-infected HeLa cells with IncA_KO and IncA_chimera Ctr. As seen in Fig. 4c and S11a, IncA_chimera clusters when both inclusions express it, reaching an ES of 3.8 (Fig. 4e, IncA blue). In contrast, when an IncA_chimera inclusion faces an IncA_KO inclusion, IncA_chimera is unable to cluster, and its ES drops down to 1.0 (Fig. 4e, IncA red). We observe a similar phenomenon with IncA_WT (Fig. 4d, S11b), where the ES decreases to 1.0 when the apposing inclusion does not express IncA (Fig. 4e, IncA green), contrasting with the enrichment observed when IncA is expressed on both membranes. These observations also argue against the contribution of another yet-to-be-identified factor to promote IncA clustering since both inclusions would express such a factor on both sides. Together, these results indicate that IncA must be present on both membranes to cluster and fuse.

Fig. 4. IncA clustering in ICSs requires the presence of IncA on both membranes.

a If IncA interacts in trans, then clustering will only be observed when IncA is present on both inclusion membranes. WT = wild-type. KO = IncA knockout. Created in BioRender. Paumet, F. (2024) BioRender.com/w90v943. b If IncA interacts in cis, then IncA clustering will be observed when IncA is on one or both inclusion membranes. Created in BioRender. Paumet, F. (2024) BioRender.com/q49y738. c, d HeLa cells were co-infected with IncA_KO Ctr (MOI 2.5) and IncA_chimera-FLAG Ctr (MOI 2.5) (c), or with IncA_KO Ctr (MOI 2.5) and WT Ctr (MOI 2.5) (d) in the presence of anhydrotetracycline. Cells were fixed at 24 hpi and labeled with anti-MOMP (magenta), which labels Ctr, anti-IncA (green), and anti-IPAM (red) antibodies. DNA was stained with Hoechst (blue). For c, d: Asterisks denote inclusions; Scale bars = 10 μm. Images are representative of three independent experiments. e ES quantification for the co-infections conducted in c and d. Graph depicts the average (black line) and individual ESs (dots). n = 150 total scores per Inc from three independent experiments. Gray box denotes ESs <2. ns = not significant, * denotes p-value < 0.05, *** denotes p-value < 0.001 (two-way ANOVA).

The lipid and protein composition of ICSs controls homotypic fusion

ICSs display specific lipid and protein composition. To test whether the local membrane environment of ICSs plays a role in homotypic fusion, we first altered the lipid composition of the inclusion. PI(3,4)P₂ significantly concentrates in the ICS (Fig. 1b and S1a, c, d); thus, we determined whether its depletion would affect homotypic fusion. Two enzymes are critical for the generation of PI(3,4)P₂; SH2 domain-containing inositol 5-phosphatase (SHIP2), and class II phosphoinositide 3-kinase A (PI3K-C2). To decrease the concentration of PI(3,4)P₂ during infection, we used a potent and selective small-molecule inhibitor of SHIP2, AS1949490³⁶ (Fig. 5a). Wild-type Ctr-infected cells were treated with AS1949490 after infection, and the effect of the inhibitor was measured after 21 hpi or 40 hpi (Fig. 5b). When the concentration of PI(3,4)P₂ is significantly reduced (Fig. S12a), we observe that 56% of infected cells displayed multiple inclusions at 21 hpi, compared with only 18% in the DMSO-treated (0 µM) control cells (Fig. 5c). Fusion eventually occurs, as 93% of infected cells treated with AS1949490 have undergone fusion at 40 hpi, similar to the DMSO-treated control cells (Fig. 5b, c). Since PIs can be converted into each other, interfering with the concentration of PI(3,4)P₂ likely impacts the concentration of the other PIs, including PI(3,4,5)P₃, and PI(4,5)P₂ (Fig. 5a). However, since neither PI(3,4,5)P₃, nor PI(4,5)P₂ is present at the ICSs (Fig. 1c, S1b and S3a), it is unlikely that they influence homotypic fusion. Interestingly, IPAM and IncA are no longer enriched in the ICSs in the presence of AS1949490 (Fig. S12b), indicating that PI(3,4)P₂ depletion disrupts ICS composition, which in turn likely disrupts the kinetics of fusion.

Fig. 5. The lipid and protein composition of the ICSs regulate homotypic fusion.

a Diagram of PI(3,4)P₂ biosynthesis. The inhibitor AS1949490 used to block SHIP2 is indicated in red. Created in BioRender. Paumet, F. (2024) BioRender.com/p53j299. b HeLa cells were infected with WT Ctr (MOI 3) in the presence of 0 µM or 3 µM AS1949490. Cells were fixed at 21 hpi and labeled with anti-MOMP antibody (green) to visualize inclusions. Hoechst was used to label DNA (blue). Scale bar = 20 µm. Images are representative of four independent experiments. c Average percentage of cells treated with 0 µM or 3 µM AS1949490 containing multiple inclusions +/- SD from four independent experiments. 100 cells were counted from multiple fields of view for each experiment and each condition. **** denotes p-value < 0.0001 (two-tailed t-test). d, f CHO (control), LY-B, and LY-B/cLCB1 cells were infected with WT Ctr (MOI 3) and fixed at 24 or 32 (hpi). The average percentage of cells containing multiple inclusions was measured in WT Ctr-infected cells at 24 hpi (d) or 32 hpi (f) +/- SD from three independent experiments. For each replicate, 100 cells for each condition were counted from multiple fields of view. ns = not significant, * denotes p-value < 0.05, ** denotes p-value < 0.01, and **** denotes p-value < 0.0001 (one-way ANOVA). e LY-B cells were infected with WT Ctr (MOI 3) for 2 h followed by the addition of media supplemented with dihydroceramide (ceramide) or sphingomyelin (SM). Cells were then fixed at 24 hpi, and fusion was quantified as in d. g Western blot for the detection of IPAM in HeLa cell lysates infected with WT or IPAM_KO Ctr (MOI 1, 24 hpi) shows successful knock out of IPAM (predicted molecular weight 29.4kD) in IPAM_KO Ctr. MOMP staining is used as the loading control. Representative of two independent experiments. h HeLa cells were infected with WT or IPAM_KO Ctr (MOI 3) and fixed 21 hpi. Cells were labeled with anti-IPAM (green) and anti-IncA (magenta) antibodies. DNA was stained with Hoechst (blue). Scale bar = 10 μm. Images are representative of two independent experiments. i Fusion was quantified as in d at 21 or 40 hpi.

In addition to phosphoinositides, sphingolipids also play a critical role in Chlamydia infection, where they are recruited to the inclusion membrane^37–40. In particular, it has been shown that the depletion of sphingolipids impairs homotypic fusion in cells infected with Ctr serovar E⁴¹. In light of their critical role during infection and to assess the extent to which these lipids may be involved in controlling homotypic fusion, we determined whether sphingolipids were also involved in homotypic fusion in serovar L2-infected cells. To do so, we used Chinese hamster ovary-derived mutant LY-B cells that cannot synthesize sphingolipids due to a loss of the serine palmitoyltransferase (SPT) cLCB1 subunit⁴² (Fig. S13a). LY-B cells complemented with cLCB1 were used as a positive control⁴³. As shown in Fig. 5d, homotypic fusion is impaired at 24 hpi in LY-B cells, demonstrating that the absence of sphingolipids interferes with the fusion of Ctr L2 inclusions, as was previously shown for serovar E⁴¹.

Among sphingolipids, sphingomyelin is particularly important for Chlamydia development^44–46. Sphingomyelin is generated by the mammalian enzymes sphingomyelin synthase 1 and 2 (SMS 1 and 2) as well as by Ctr⁴⁷ (Fig. S13a). Ceramide is a required precursor for sphingomyelin synthesis, which can be transported to the inclusion by direct interactions with the ceramide transporter CERT^19,46,47 (Fig. S13a). Ctr can also use CERT-derived ceramide to de novo synthesize sphingomyelin⁴⁷. Therefore, we used CERT knockout (ΔCERT), SMS 1 and 2 double knockout (DKO), or CERT, SMS1, and SMS2 triple knockout (TKO) HeLa cell lines⁴⁷ to assess the extent to which sphingomyelin plays a role in homotypic fusion. Using the TKO cells, we observe that homotypic fusion is also impaired at 24 hpi (Fig. S13b, c, TKO). Both sphingomyelin and ceramide trafficking must be blocked to inhibit fusion, as ΔCERT or DKO inclusions still undergo homotypic fusion, although to varying degrees (Fig. S13b, c, ΔCERT, DKO). The homotypic fusion defect at 24 hpi in the LY-B (Fig. 5e) and TKO (Fig. S13d) cells is reversed when media is complemented with sphingomyelin, validating that the lack of sphingomyelin is responsible for the fusion defect. Furthermore, while the addition of dihydroceramide rescued the fusion defect in LY-B cells, which can use this lipid to synthesize sphingomyelin, dihydroceramide did not rescue the fusion defect in TKO that lack the enzymes needed to convert this precursor into sphingomyelin (Fig. 5e and S13d). These data support that sphingomyelin is the specific sphingolipid involved in regulating homotypic fusion.

Consistent with the fusion defect in LY-B and TKO cells, IncA is excluded from ICSs at 24 hpi (Figs. S14a and S13e), with an average ES of 1.4 and 1.3, respectively (Figs. S14b and S13f). High-resolution 3D reconstruction of inclusions formed in LY-B cells distinctly shows that IncA no longer colocalizes with IPAM but instead, surrounds it at the edge of the ICSs (Fig. S14c). This is not due to a general loss of ICS integrity since Akt-PH (Fig. S14d) and IPAM (Figs. S14a–c and S13e, f) remain concentrated within the ICSs, while CT813 remains excluded (Figs. S14a, b and S13e, f). Due to the toxicity of the fluorescent sphingomyelin probe, Eqt2-SM-oxGFP, we could not determine whether sphingomyelin is clustered within ICSs.

Interestingly, homotypic fusion ultimately occurs in the LY-B cells at ~32hpi, indicating that fusion is only delayed (Fig. 5f). This correlates with delayed IncA clustering as endogenous IncA can be detected within the ICS at around 26 hpi (Fig. S14e). The uptake of exogenous lipids from the cell culture media may facilitate the slow population of the ICS over time. Together, these data indicate that the absence of sphingolipids specifically affects the clustering of IncA within ICSs and delays homotypic fusion.

In conjunction with lipids, proteins present in ICSs could play a major role in homotypic fusion. IPAM has been found almost entirely clustered in ICSs (Fig. S4a, b). To assess whether IPAM regulates fusion, we used an IPAM_KO Ctr strain in which IPAM is depleted (Fig. 5g, h). Homotypic fusion was significantly inhibited at 21 hpi in IPAM_KO Ctr-infected cells, with 45% of the cells displaying multiple inclusions, compared to only 6% for cells infected with WT Ctr (Fig. 5i). Once again, fusion is only delayed in the absence of IPAM since 95% of infected cells have undergone fusion by 40 hpi for both IPAM_KO and WT Ctr. Although the exact role of IPAM in membrane fusion is unknown, these results indicate that inclusions form unique functional platforms at contact sites, in which lipids and proteins directly and/or indirectly control homotypic fusion.

Discussion

Reminiscent of eukaryotic cells in which fusion-competent membranes generate contact sites enriched in specific lipids and proteins^20,21, we discovered that chlamydial inclusions also form membrane contact sites where fusion occurs. These contact sites are enriched in PI(3,4)P₂ and chlamydial proteins, including the bacterial fusion protein IncA, but are different from previously identified ER-inclusion contact sites or pSrc-containing microdomains. This study establishes that the ICS is a unique molecular platform, the function of which is to regulate homotypic fusion. ICSs carry out this specific task by controlling the clustering of chlamydial proteins, notably IncA. In fact, delaying IncA clustering into these regions significantly delays fusion, indicating that a high concentration of IncA in ICSs is required for homotypic fusion to readily occur. Considering that homotypic fusion is tightly linked to the concentration of IncA in ICSs, we identified elements that control IncA clustering in these sites.

Previously, we established that the C-terminal region of IncA folds into a stable four-helix bundle, thus maintaining the protein as a monomer¹⁶. When we disrupt the formation of this four-helix bundle by introducing mutations into one of the helices, as in IncA_polyA, this IncA mutant no longer clusters in the ICS. Since IncA_polyA cannot form the intramolecular interactions that lock IncA in its monomeric conformation, it can readily self-assemble in cis on the inclusion membrane¹⁶. As such, the lack of clustering in the ICSs could be due to the decrease in free non-oligomerized IncA_polyA to interact in trans upon inclusion-inclusion contact. Supporting this hypothesis, co-infections of Ctr expressing either WT IncA or IncA_chimera with IncA_KO Ctr show that when IncA is only expressed on one membrane, it is unable to cluster. Altogether, these data suggest that the clustering of IncA is due to its oligomerization in trans, at the level of the C-terminus, between apposing inclusions. This clustering, in turn, regulates homotypic fusion.

Adding another layer of regulation, we observed that the lipid composition of the ICS plays a significant role in homotypic fusion: the depletion of PI(3,4)P₂ or sphingolipids, in particular sphingomyelin, significantly delays fusion. In other systems, sphingolipids, notably sphingomyelin, are critical for maintaining the organization of lipid rafts required for efficient membrane fusion⁴⁸. It has been proposed that sphingomyelin influences other lipids to drive its organization into lipid rafts⁴⁹. Interestingly, sphingomyelin has also been shown to act as a critical regulator of the endomembrane distribution of phosphatidylserine, which plays a critical role in membrane fusion^50–52. During Chlamydia infection, the precise mechanism by which these lipids control homotypic fusion is unclear. However, we found that their depletion disrupts the composition of ICSs and IncA clustering. Therefore, these lipids may act indirectly by influencing the localization of other lipids and/or proteins, which would then control IncA clustering. This raises the possibility of an indirect impact of sphingolipids on the distribution and organization of other lipids. We have only begun to unravel the composition of ICSs and cannot exclude these possibilities.

IPAM is another chlamydial protein clustered in the ICSs, the deletion of which impairs homotypic fusion. IPAM is thought to regulate the organization of microtubules at the inclusion surface likely directing inclusions to the MTOC⁵³. Although its exact function in membrane fusion is unknown, one possibility is that the high concentration of IPAM in the ICSs ensures that the inclusions are in a tight apposition at the level of the ICSs, thus physically securing their tethering. This is currently under investigation.

Altogether, our results support a model (Fig. 6) in which the concentration of specific proteins and lipids in the ICSs plays a critical role in controlling homotypic fusion through IncA clustering: ❶ IncA is first expressed on the entire inclusion membrane and maintained in a stable monomeric conformation to prevent cis self-assembly on the inclusion membrane. ❷ Once the inclusions are in close proximity, lipids, including sphingomyelin and PI(3,4)P₂, facilitate IncA clustering at the ICS prior to fusion. ❸ As IncA concentrates within ICSs, it can then undergo conformational changes and oligomerize in trans with IncA on the apposing membrane, further securing its clustering and ensuring its high local concentration at these points of contact. ❹ Finally, as the trans-oligomerization of IncA increases, it promotes homotypic fusion of inclusions and the mixing of the inclusion contents.

Fig. 6. Regulatory model of IncA-mediated membrane fusion.

❶ IncA (pink) is homogeneously expressed on the inclusion membrane where it is folded as a monomer to prevent cis self-assembly. IPAM (green) is distributed throughout the inclusion membrane. ❷ Lipids (blue), in particular sphingolipids and PI(3,4)P₂, trigger IncA clustering within ICSs. PI(3,4)P₂ also controls IPAM recruitment into the ICSs. ❸ As IncA concentrates in the ICS, it then oligomerizes in trans, further enhancing its clustering. ❹ At its optimal local concentration, IncA drives the homotypic fusion of inclusions. Ct = Chlamydia trachomatis bacteria. Created in BioRender. Paumet, F. (2023) BioRender.com/c27a297.

In addition to the local regulatory elements, we discovered that the N-tail/TMD of IncA itself is necessary to complete this fusion event, as IncA_chimera cannot promote fusion despite its high local concentration. The importance of the membrane anchor for fusion has been previously established for viral and eukaryotic fusion machinery. A fusion protein must be linked to a membrane anchor that spans the bilayer to trigger membrane fusion^54–56. The TMDs of IncA and CT813 are both bilobed, have a similar size, span the bilayer, and likely act similarly as membrane anchors. Therefore, the exchange of the TMDs is likely not responsible for the inhibition of fusion seen with the IncA_chimera. Another possibility is that the N-tail directly influences the fusogenicity of IncA by modifying its conformation. Note that the N-tail of IncA is significantly enriched in proline, with five prolines present in a seven amino-acid residue stretch. Proline residues can engage aromatic residues within or between peptides in a complex⁵⁷. As such, the N-tail could interact with the phenylalanine, tyrosine, and histidine residues present in H_a and H_b to influence the folding of IncA, thus priming it for fusion. Since the N-tail of CT813 is not as proline-rich, it may not prime IncA for fusion. As a result, IncA would not adopt a fusogenic conformation, leading to the complete inhibition of fusion. Experiments are ongoing to identify the role of each region in IncA-mediated fusion.

In the context of Chlamydia infection, this study opens new avenues of research. Could fine-tuning homotypic fusion via lipids allow Chlamydia to modulate its pathogenicity? In fact, the intracellular environment of infected cells is constantly adapting as the host mounts an immune response and responds to Ctr growth. This ever-changing environment may impact the lipid composition of the inclusion membrane, thus influencing the clustering of IncA, which would then modulate homotypic fusion. As a result, Chlamydia infection could be altered since IncA-mediated homotypic fusion is critical for its pathogenesis. Supporting this possibility, it has been shown that sphingomyelin depletion, which impacts homotypic fusion, also prevents Chlamydia from exiting persistence, a state during which Chlamydia stops replicating and goes dormant until the environment becomes favorable again⁴¹. Altogether, deciphering the regulatory mechanisms of IncA-mediated homotypic fusion will help us better understand disease progression.

Overall, these data identify how a bacterial fusion system is regulated at the molecular level. These results could inform the discovery and characterization of new prokaryotic fusion machinery. Ultimately, the mechanisms we identify here may reveal shared or divergent mechanisms of regulating membrane fusion across kingdoms.

Methods

Cell culture

HeLa 229 cells (ATCC) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Corning) supplemented with 20 mM HEPES/NaOH (pH 7.5), nonessential amino acids (HyClone), 10% fetal bovine serum (Corning), 10 μg/mL gentamicin (Thermo), and 2 mM L-glutamine (Gemini Bio). CHO-K1, LY-B, and LY-B/cLCB1 cell lines, obtained from RIKEN BioResource Research Center, were cultured in HAM-F12 media (Gibco) supplemented with 2 mM L-glutamine, 10% fetal bovine serum, and 10 μg/mL gentamicin. The parental HeLa cell line (mCAT#8), as well as HeLa knockout cell lines ΔCERT, double-knockout (DKO) for the sphingomyelin synthetases (SMS) 1 and 2, and triple knockout (TKO) for SMS1, SMS2, and CERT, were grown in DMEM supplemented with 20 mM HEPES/NaOH (pH 7.5), nonessential amino acids, 20% fetal bovine serum, 10 μg/mL gentamicin, and 2 mM L-glutamine⁴⁷.

Antibodies

An anti-IncA antibody was generated by immunization of goats (Thermo) with purified soluble recombinant IncA_87-273-His₆ protein produced in-house¹⁷. The specificity of the antibody was validated by western blot analysis of infected cell lysates (WT and IncA_KO Ctr L2), as described below. This analysis revealed a single band at the appropriate size (30.3kD) in the WT L2 Ctr-infected lysate, which is absent in the IncA_KO Ctr-infected lysate (Fig. S5a). Using immunofluorescence microscopy, we observed IncA staining on the inclusion membrane in WT-infected cells, while this staining is absent in IncA_KO Ctr-infected cells (Fig. S5b), further validating the specificity of the anti-IncA antibody. For western blot, this goat anti-IncA antibody was used at a dilution of 1:12,000 and for immunofluorescence microscopy at 1:4000. The following primary antibodies were also used in immunofluorescence: anti-FLAG (rabbit, 600-401-383; Rockland Immunochemical, 1:1,600 dilution), anti-CT813 (rabbit; T. Hackstadt, NIH, 1:300 dilution), anti-IncB (rabbit; T. Hackstadt, NIH, 1:300 dilution), anti-pSrc (rabbit, ab40660; Abcam, 1:250 dilution), anti-CERT (chicken, GW22128B; Sigma, 1:200 dilution), anti-MOMP (goat, 1621; ViroStat, 1:800 dilution), anti-EB Ctr (rabbit, 1611; Virostat, 1:500 dilution), anti-PI(3,4)P₂ (mouse, Z-P034, Echelon, 1:100 dilution), anti-PI(3,4)P₂ (mouse, Z-P034B, Echelon, 1:100 dilution), and anti-CT223/IPAM (mouse; D. Rockey, Oregon State University, 1:200 dilution). The following fluorescent secondary antibodies were used in immunofluorescence microscopy: donkey anti-chicken IgY, or donkey anti-mouse, -goat, and -rabbit IgG conjugated to Alexa Fluor −488, −555, or −647 (Thermo, all used at a 1:500 dilution). Hoechst 33342 was used to label DNA. For western blot, HRP-conjugated donkey anti-goat, anti-mouse, and anti-rabbit IgG antibodies were used (Thermo, 1:25,000 dilution).

Gene cloning and mutagenesis

Standard PCR and cloning procedures were used. Primers used are listed in Table 1. DNA sequences were confirmed by Sanger sequencing. All enzymes were purchased from Thermo. CT813_1-94IncA_85-273-FLAG (IncA_chimera) was constructed by Overlap Extension PCR. In the first step, the CT813 portion (Met1 to Leu94) was amplified using FO507/FO1307 and full-length CT813 as the template, while the IncA portion (Gln85 to Ser273) was amplified using FO1306/FO511 and full-length IncA as the template. In the second step, the CT813 and IncA products were allowed to prime each other briefly. The primers FO507/FO511 were then used to amplify the final construct using the second product as the template. The final PCR product was digested with NotI and SalI and ligated into pBOMB3-tet. The pBOMB3-Tet plasmid was constructed by removing the tetracycline promoter from pBOMB4-Tet with XhoI and NotI, followed by ligation into pBOMB3⁵.

Table 1.

Primers

Name	Sequence (3’−5’)
FO507	TACCACTCCCTATCAGTGATAG
FO511	ACGCAGTCGACCTACTTGTCATCGTCATCCTTGTAGTCGGAGCTTTTTGTAGAGGG
FO786	GGACACTCTAGAATGGACAACACCGAGGACG
FO787	ATATATGGGCCCTCACTGGGAGCCGGAGTGGCG
FO1306	GTATCAGCTTTATGCTTGCAGAAAACCGCTAATCTACATCTATAC
FO1307	GTAGATTAGCGGTTTTCTGCAAGCATAAAGCTGATACCAAGCAG
FO1333	ATTCTCGAGCACCTTGCCGAAAGTGCC
FO1334	CTTGGAGCCGTACTGGAACTG
FO1336	ATTGGTACCGCCACCATGGACTCGGGCCGGGACT
FO1337	ATTCTCGAGCTGAAGGAGCTCAACATCCAG
FO1338	ATTGGTACCGCCACCATGGAGGGGGTGTTGTACAAGTGG
FO1339	ATTCTCGAGTCTTCAGGAAGAGAAGGATGGA
FO1344	ATTGGTACCGCCACCATGGCCGCAGTGATTCTG
FO1345	ATTCTCGAGCGGTTTTAAGCTTCCATTCCTG
FO1348	ATTGGTACCGCCACCATGGTGAAGAAGGAGTGGCAATCTAGT
FO1349	ATTCTCGAGTCCTTGCAAGTCATTGAAACAT
FO1759	GTATACAAGCTTATGGCCGAGGTGAAGCTG
FO1762	GATCCTCTAGATTACTTGTACAGCTCGTCCATCC
FO1789	GTGCTGGGACCAAGCTGG
FO1790	GCGGGTTTAAACGGGCCC

Anti-FLAG-Frankenbody_mScarlet3 was generated as follows. Anti-FLAG-Frankenbody_mRuby2 was PCR amplified from pCMV-anti-FLAG-Frankenbody_mRuby2 (Addgene #175294) with the primers FD1759 and FO1762. The PCR product was inserted into EF1α-pcDNA3.1 (gift from Mark Tomishima) using the HindIII and XbaI restriction sites to generate EF1α-anti-FLAG-Frankenbody_mRuby2. Next, mRuby2 was exchanged for mScarlet3. To do so, mScarlet3 was PCR amplified with the primers FO1789 and FO1790 and inserted into EF1α-anti-FLAG-Frankenbody_mRuby2 using the Oli1 and XbaI restriction sites to generate EF1α-anti-FLAG-Frankenbody_mScarlet3.

Plasmids encoding Akt-PH-GFP, Btk-PH-GFP, and EEA1-FYVE-GFP were provided by T. Balla (NIH). PLCδ-PH-GFP (#21179) was purchased from Addgene. FAPP1-PH-GFP was provided by P. Wedegaertner (Thomas Jefferson University). Lipid probes with C-terminal mDsRed tags were generated as follows. mDsRed was PCR amplified using primers FO786 and FO787 and cloned into pcDNA3.1⁺ (Invitrogen) via XbaI and ApaI restriction sites to generate mDsRed pcDNA3.1. Akt-PH, Btk-PH, PLCδ-PH, FAPP1-PH, and EEA1-FYVE were PCR amplified using the primer sets FO1333/1334, FO1344/1345, FO1336/1337, FO1338/1339, FO1348/1349, respectively. The PCR products were then ligated into mDsRed pcDNA3.1 using the restriction sites KpnI and XhoI.

Chlamydia transformation

IncA_chimera-FLAG-expressing Ctr was generated by transforming IncA_KO Ctr L2 with IncA_chimera-FLAG pBomb4S-Tet encoding IncA_chimera-FLAG under the control of a tetracycline promoter following the protocol described in ref. ¹⁶. GFP IncA_KO Ctr L2 was generated by transforming IncA_KO Ctr L2 with empty pBomb3-Tet. Individual clones were isolated by limiting dilution and purity was assessed by immunofluorescence microscopy.

Chlamydia strains and infections

Wild-type Ctr serovar L2 (LGV 434/Bu) was provided by T. Hackstadt (NIH). Ctr IncA_KO and IncA_KO complemented with IncA_WT-FLAG or IncA_polyA-FLAG Ctr L2 strains were generated previously¹⁶. IPAM_KO Ctr L2 was provided by R. Bastidas (Duke University). All strains were propagated and density gradient-purified as described^16,58. For infections, cells were seeded on glass coverslips in 24-well plates 24 h before infection. HeLa cell lines were seeded at a density of 5 × 10⁴ cells per well, while CHO-derived cell lines were seeded at a density of 2.5 × 10⁴ cells per well. Cells were infected with 500 µL of inoculum in growth media. Anhydrotetracycline (Tet) was added at the time of infection. Plates were then immediately centrifuged at 1000 × g for 1 h at 4 °C. The MOI for each cell line and Ctr strain was determined empirically and is noted in the figure legends. The appropriate concentration of Tet for each IncA_mutant-FLAG Ctr strain was determined by immunofluorescence microscopy. The Tet concentrations used to achieve an expression level on the inclusion membrane matching that of IncA on the inclusion membrane of WT Ctr L2 were IncA_WT-FLAG (5 ng/mL), IncA_chimera-FLAG (0.4 ng/mL) and IncA_polyA-FLAG (5 ng/mL). If cells were infected for >24 h, the growth media was replaced at 24 hpi with media containing fresh Tet at the appropriate concentration.

Recombinant protein expression and purification for IncA antibody production

IncA_85-273-His₆ was expressed and purified as described¹⁶.

Lipid probe transfection

HeLa and CHO cell lines were transfected at 2 hpi with 25-50 ng of purified biosensor plasmid DNA per 2.5-5×10⁴ cells. This was mixed with an empty vector pcDNA3.1 to achieve a total of 250 ng DNA per transfection. Transfection was performed using the Continuum transfection reagent (Gemini Bio) according to the manufacturer’s instructions. For HeLa and CHO cell lines, a 1:1 or 1:1.5 DNA to Continuum ratio was used, respectively.

Live-cell imaging of lipid probes and fluorescent cholesterol

HeLa cells were infected with the indicated strains at an MOI of 3 in glass bottom plates. After centrifugation, 5 µM TopFluor or TopFluor-TMR cholesterol (Avanti) was added to the cell culture media and maintained throughout the infection. Lipid probes were transfected as described above. At 22 hpi, the cells were washed twice with imaging media (FluorBrite DMEM supplemented with 20 mM HEPES/NaOH [pH 7.5], nonessential amino acids, 10% fetal bovine serum, 10 μg/mL gentamicin, and 2 mM L-glutamine). To label nuclei, cells were incubated with 10 µg/mL Hoechst in imaging media for 15 min at 37 °C followed by several washes with imaging media. The media was then replaced with imaging media containing a 1:50 dilution of ProLong Live antifade reagent (Invitrogen), and the cells were incubated at 37 °C for 1 h before imaging. Cells were imaged on a Nikon TiE inverted fluorescence microscope with a 60x oil immersion lens and NIS-Elements software (Nikon). Images were then deconvolved using Legacy deconvolution (NIS-elements) and processed using ImageJ (NIH).

Immunofluorescence microscopy

Cells transfected with lipid biosensors, and those labeled with anti-pSrc and -CERT antibodies were fixed at the indicated time points, with 4% paraformaldehyde (PFA) in HBSS containing calcium and magnesium for 15 min at room temperature (RT). For all the other experiments, infected cells were fixed with ice-cold methanol for 10 min at RT. Fixed cells were then washed with IF-G buffer (25 mM HEPES/NaOH [pH 7.5], 150 mM NaCl, 900 nM CaCl₂, 500 nM MgCl₂, 100 mM glycine) for PFA fixation, or IF buffer (25 mM HEPES/NaOH [pH 7.5], 150 mM NaCl, 900 nM CaCl₂, 500 nM MgCl₂) for methanol fixation. PFA-fixed cells were permeabilized with 0.1% Triton X-100 in IF buffer for 15 min before blocking. Fixed and permeabilized cells were blocked in IF buffer containing 10% donkey serum, 0.1% bovine serum albumin, and 0.05% sodium azide for 1 h at RT. After blocking, cells were incubated with the primary antibodies diluted in labeling buffer (10% donkey serum, 0.1% bovine serum albumin, 0.05% sodium azide, and 0.1% Triton X-100 in IF buffer) for 1 h at RT. Next, cells were washed with 0.1% Triton X-100 in IF buffer and then incubated with Alexa Fluor-conjugated secondary antibodies (1:500 dilution) and Hoechst diluted in labeling buffer for 1 h at RT. Cells were washed with 0.1% Triton X-100 in IF buffer, followed by washes with IF buffer, and finally, briefly washed with purified water immediately before mounting. Coverslips were mounted with ProLong Glass antifade mountant (Invitrogen) and allowed to cure overnight before imaging. Cells were imaged on a Nikon TiE inverted fluorescence microscope with a 60x oil immersion lens and NIS-Elements software (Nikon). Images were processed using ImageJ (NIH). To analyze the detailed localization of IncA in LY-B-infected cells, z-stacks were acquired on a Nikon A1 Confocal Laser Scanning Microscope at 60x oil immersion in 0.085 µm sections and deconvolved using the Landweber method. 3D reconstruction was performed using the NIS-Elements software.

PI(3,4)P₂ labeling with antibody

Infected cells were fixed with 4% PFA in HBSS containing calcium and magnesium at RT for 20 min, followed by three washes with TBS. Cells were permeabilized with 0.2% Triton X-100 in TBS for 15 min, then washed thrice with TBS. Samples were blocked for 1 h at RT in a blocking buffer (10% goat serum in TBS). Antibodies (Z-P034 and Z-P034B) were diluted 1:100 in blocking buffer and incubated with the samples for 2 h at RT. Following several washes in TBS, cells were incubated with secondary antibodies diluted 1:500 in a blocking buffer for 45 min at RT. Cells were washed three times with TBS, mounted with ProLong Glass antifade mountant, and imaged as described above.

Generation of stable anti-FLAG-Frankenbody_mScarlet3 HeLa cell line

HeLa 229 cells (5 × 10⁵) were transfected with EF1α-anti-FLAG-Frankenbody_mScarlet3 DNA (2 µg) by nucleofection using kit R and program I-013 (Amaxa). Antibiotic selection was initiated 24 h after transfection with 0.7 mg/mL active G-418. After 14 days of selection, individual clones were isolated by limiting dilution. Purity was assessed by fluorescence microscopy.

Live-cell time-lapse imaging

Anti-FLAG-Frankenbody_mScarlet3 HeLa cells (4×10⁴) were seeded in glass bottom plates 24 h before infection with IncA_KO Ctr L2 complemented with IncA_WT-FLAG at MOI 3. At 5 hpi, the monolayers were washed, and the media was replaced with imaging media containing ProLong Live antifade reagent (1:50 dilution) and 20 ng/mL anhydrotetracycline. Images were acquired on a Nikon A1 Confocal Laser Scanning Microscope outfitted with a Tokai-Hit temperature and CO₂-controlled chamber and a Plan Fluor 40x/1.3 NA oil immersion objective. Nine-micron z-stacks with 0.5 µm slices were acquired using the resonant scanner every 5 min beginning at 8 hpi until 24 hpi. Images were deconvolved using the Richardson-Lucy method and processed with Nikon NIS-Elements software.

SHIP2 inhibitor, nocodazole, and lipid complementation

To inhibit SHIP2, 3 µM of AS1949490 (Echelon) or DMSO (control) was added to the cell culture medium after infection and centrifugation, and maintained throughout the infection. To complement LY-B and TKO cells with lipids, the cells were first infected for 2 h followed by a brief wash in HBSS with calcium and magnesium. Cell culture media containing 5 µM C6 dihydroceramide (Enzo), TopFluor- Sphingomyelin (Avanti), or ethanol (control) was added to cells and maintained throughout infection. For nocodazole treatment, cells were first infected for 5 h, then washed briefly in HBSS with calcium and magnesium. 20 μM nocodazole or DMSO (control) in cell culture media was then added to cells and maintained throughout infection.

ICS enrichment score quantification

IPAM labeling was used as a proxy to mark the ICSs (Fig. S4a, b). Using NIS-Elements software (Nikon), a fluorescence intensity profile line was drawn across touching inclusions that formed IPAM-positive ICSs. Using the associated fluorescence intensity profile trace, three peaks were measured for each ICS reading: two for the two outer membranes of touching inclusions, called outer periphery (O_p), and one for the ICS (I_cs). Peak intensity measurements were obtained using the measure horizontal line tool drawn from baseline to maximum peak intensity to provide a numerical intensity for each peak. The enrichment score was calculated by dividing the fluorescence intensity peak of the ICS by the average of the two outer periphery fluorescence intensity peaks. A total of 150 ICS enrichment scores (50 scores from 3 independent experiments) were calculated and plotted for each inclusion membrane protein (see Fig. S2 for a schematic of this method).

Homotypic fusion quantification

Homotypic fusion was measured as previously described¹⁶. Briefly, cells were infected at the indicated MOIs, fixed, and labeled at the indicated time. For each of the three independent experiments, 100 cells from each sample were classified as containing one or multiple inclusions. The percentage of total infected cells with a single inclusion (having undergone homotypic fusion) or multiple inclusions (impaired fusion) was quantified.

Western blotting

For Fig. 5g and Fig. S5a, protein lysates were generated from monolayers of HeLa 229 cells infected at an MOI of 1 with WT Ctr, IPAM_KO Ctr (Fig. 5g) or IncA_KO Ctr (Fig. S5a). To prepare cell lysates, monolayers were washed briefly with cold HBSS without calcium or magnesium and lysed at 24 hpi with cold lysis buffer (1x NuPAGE LDS Sample Buffer [Novex] containing protease and phosphatase inhibitor cocktails [Apex Bio], and 250 units/mL of turbonuclease [Accelagen]) for 10 min. β-Mercaptoethanol was added to a final concentration of 360 mM. Samples were incubated at 95 °C for 10 min followed by centrifugation at 20,800 × g at 4 °C for 10 min. The supernatant was collected, and the total protein concentration was measured using a 660 nm protein assay reagent kit with the ionic detergent compatibility additive (Pierce). Absorbance measurements were obtained at 660 nm on a SpectraMax M2 plate reader (Molecular Devices). Lysates were diluted in LDS-Sample buffer for gel loading and separated on 4–12% Bis-Tris SDS-PAGE gels (Invitrogen). The amount for each sample was normalized to MOMP prior to loading. Samples were transferred to polyvinylidene difluoride membranes for 1 h at 90 volts, at 4 °C, in transfer buffer (25 mM Tris/HCl, 192 mM glycine, 10% methanol). Membranes were blocked for 1 h at RT in blocking buffer (3% bovine serum albumin and 0.05% sodium azide in TBST [25 mM Tris/HCl, pH 7.5, 150 mM NaCl, 0.1% Tween 20]). After blocking, membranes were incubated in primary antibody overnight at 4 °C in the blocking buffer. Goat anti-IncA was used at a 1:12,000 dilution, mouse anti-IPAM was used at a 1:1000 dilution, and goat anti-MOMP was used at a 1:20,000 dilution. Following primary antibody incubation, membranes were washed with TBST and then incubated with HRP-conjugated secondary antibodies (Thermo, 1:25,000 dilution) for 1 h at RT in 5% milk diluted in TBST. Membranes were then washed with TBST, followed by several washes with TBS. HRP-positive bands were detected using SuperSignal West Dura extended duration substrate (Thermo). Blots were imaged on a FluorChem R system (ProteinSimple).

Statistical analysis

All data was analyzed in GraphPad Prism v.10. For ICS enrichment score analyses, the same ICS was analyzed for all three Incs (IncA, CT813, and IPAM), and a one-way ANOVA using a matched-paired design was used. In experiments where Inc or lipid probe enrichment was measured across different ICSs, Grubb’s test was applied to detect single outliers followed by a non-matched/non-paired one-way ANOVA analysis. For homotypic fusion quantification, analysis was done with non-matched/non-paired one-way ANOVA. Significance was assumed at a p-value < 0.05.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

C.L., J.W., A.L., and F.P. contributed to the experimental design, data acquisition, and manuscript writing and editing. T.Y. and K.H. generated the HeLa mutant cell lines and contributed to the editing of the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

The authors declare that all relevant data supporting the findings of this study are included in this published article and its Supplementary Information files, or from the corresponding authors upon request. Source data are provided in this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-024-53443-7.