Different pathogens have evolved a variety of unique strategies to establish a protective intracellular niche in which to replicate and to obtain essential nutrients from the host. For example, Legionella pneumophila
replicates within a vacuole that is closely associated with the ER, while Salmonella
species replicate within compartments that have characteristics of endosomes 
are among the few known pathogens that occupy an exocytic compartment from which they acquire host-derived nutrients, including SM and cholesterol 
. While previous work supports a role for canonical vesicular trafficking in the acquisition of SM by Chlamydia
from the host, it has remained a mystery as to why inhibitors of vesicular trafficking have no effect on Chlamydia
replication, given that host SM biosynthesis is necessary for bacterial replication 
. Here, we demonstrate that C. trachomatis
co-opts key proteins involved in both vesicular and non-vesicular lipid trafficking pathways to acquire SM for distinct roles during infection, providing an intriguing explanation for this apparent paradox. We found that C. trachomatis
recruits CERT, its ER binding partner VAP-A, and SM synthases to establish an on-site SM biosynthetic factory at or near the inclusion that is critical for C. trachomatis
replication. In addition, we show that C. trachomatis
co-opts the function of GBF1, a regulator of Arf1-dependent vesicular trafficking within the early secretory pathway, to further provide SM. This source of SM contributes to inclusion membrane growth and stability but is not essential for bacterial replication.
We found that depletion or inhibition of CERT significantly impaired production of infectious progeny and SM acquisition. While these treatments decrease SM biosynthesis and would thus be expected to affect C. trachomatis
replication, unexpectedly and most remarkably, we found that CERT was recruited to the inclusion membrane. Our results are consistent with two non-mutually exclusive mechanisms by which CERT could promote SM acquisition during infection (). CERT could transport ceramide directly from the ER to the inclusion, where this lipid would serve as a substrate for inclusion membrane localized SMS2, allowing SM biosynthesis to proceed on the inclusion membrane (, step A). SM would then be transferred from the inclusion membrane to intracellular RBs. Alternatively, or in addition, by virtue of its ability to bind to the ER and to the Golgi, CERT could help to coordinate recruitment of these organelles to the inclusion (, step B). This process would bring the trans
-Golgi localized SMS1 and SMS2 in close proximity to the ER and to the inclusion, thereby promoting efficient SM synthesis in the vicinity of the inclusion. Since CERT cannot transport SM 
, it is likely that this source of SM is subsequently transferred to the inclusion by a BFA-insensitive vesicular trafficking pathway. It is also possible that SM could be transferred from the Golgi at MCS to the inclusion by one of several mechanisms involving non-vesicular lipid exchange between membranes, such as transient hemifusion and/or stochastic collision with the Golgi 
Model for the role of CERT and GBF1 in C. trachomatis infection.
Our results suggest that both SMS1 and SMS2 contribute to infection. The involvement of SMS1 was not surprising, since this enzyme is responsible for the bulk of total host cell SM biosynthesis (60–80%) 
. However, our finding that SMS2 was localized to the inclusion membrane and played a role in intracellular growth was unexpected since SMS2 participates primarily in SM biosynthesis at the plasma membrane and plays a minor role in overall host cell SM biosynthesis (20–40%) 
. In contrast to SMS1, inclusion membrane localized SMS2 and CERT are not disrupted by BFA, GCA, or Nocodazole, which may explain why bacterial replication is not affected by these drugs. We note that it is difficult by current methods to demonstrate that SM is synthesized directly on the inclusion, since SM can also arrive via vesicular trafficking 
. We speculate that by preferentially recruiting SMS2 over SMS1 during infection, C. trachomatis
ensures a source of SM for itself without placing a huge burden on the host's ability to synthesize SM since SMS1 would still be active at the Golgi. Future studies will be required to unravel how SMS2 is recruited to the inclusion membrane.
How is CERT, a multi-domain protein, recruited to the inclusion? Although the recruitment of CERT to Golgi and ER membranes requires the PH and FFAT domains, respectively 
, we found that only the ceramide binding and/or ceramide transfer activity of CERT is required for its recruitment to the inclusion. While CKIγ2-induced phosphorylation of CERT inhibits its activity, CERT was still recruited to the inclusion, presumably because CERT is still able to bind ceramide. On the other hand, exposure to HPA-12 prevented CERT recruitment to the inclusion, possibly through its known ability to inhibit interaction of the START domain with membranes in vitro 
. An alternative scenario is that HPA-12 binding to CERT could alter its ability to bind PI4P, VAP-A, or potentially a bacterial factor at the inclusion membrane through steric hindrance or conformational change. We favor the hypothesis that CERT binds to ceramide at the ER, prior to its recruitment to the inclusion, and subsequently interacts with the inclusion by an as yet to be identified bacterial or host protein. It is also possible that a small amount of ceramide and/or VAP-A may initially be incorporated into the inclusion membrane from either the plasma membrane during endocytosis or by fusion with the ER or ER-derived vesicles containing ceramide, which would then promote CERT recruitment. These initial events could then set up an amplification mechanism for subsequent ceramide transfer and SM synthesis. Further studies are needed to determine whether CERT transfer activity or ceramide binding alone is necessary for its recruitment to the inclusion, and whether either of these functions cooperates with a bacterial protein present in the inclusion membrane.
It was surprising that PI4P binding was not required for CERT recruitment to the inclusion, given the recent observation that PI4P binding is necessary and sufficient to recruit the PH domain of OSBP 
. However, it should be noted that those studies were performed with only the PH domain of OSBP whereas our studies were performed in the context of the entire CERT protein containing a single point mutation (G67E) that abolishes PI4P binding. It is likely that there are multiple localization signals for CERT recruitment to the inclusion.
The dispensable requirement of PI4P binding for CERT recruitment to the inclusion likely explains why inclusions still form in LY-A cells expressing CERT (G67E) and why CERT still localizes to the inclusion in these cells. It is worth noting that recombinant CERT (G67E) is still able to transfer ceramide in an in vitro
, suggesting that although SM synthesis is reduced in the LY-A cell line due to the inability of CERT to bind the Golgi and transfer ceramide 
, SM synthesis may proceed at the inclusion because this specialized compartment has access to ceramide via CERT and at least one SMS isoform.
While the role of CERT in ceramide transfer and SM biosynthesis is clearly established, we cannot rule out the possibility CERT may have other functions in host cells relevant to Chlamydia
infection. Previous work has suggested that there is an intimate association of the ER or ER proteins with the chlamydial inclusion 
. It is possible that CERT and VAP-A together may be important for coordinating ER-inclusion membrane contact sites. These membrane contact sites could be important for the inclusion to obtain other essential lipids or nutrients for growth. Although loss of CERT function in individual cells has no effect on cell survival 
, knockout of CERT in mice is embryonic lethal as a result of severe defects in both the ER and mitochondria function 
. In addition, D. melanogaster
flies lacking functional CERT have a short lifespan and show enhanced susceptibility to oxidative stress 
Our studies provide further insights into the mechanism of BFA-sensitive SM acquisition. While it was previously suggested that Chlamydia
occupies a post Golgi compartment with the prediction that trafficking from the trans
-Golgi would be essential, our results demonstrate that acquisition of SM by Chlamydia
requires Arf1 activation within the cis
-Golgi but not Arf1 activation within the trans
-Golgi (, step C), suggesting that Chlamydia
is not necessarily co-opting the entire organelle. Loss of SM acquisition following inhibition of GBF1 function correlated with a loss of inclusion membrane integrity and alterations in the localization of actin and vimentin cytoskeletal components surrounding the inclusion (, step D). Whether the loss of inclusion integrity is a direct or indirect consequence of the collapse of the closely associated cytoskeletal cage or whether it results from the loss of SM acquisition, as suggested by Beatty and coworkers 
, remains to be determined. Of note, GCA was not reported to affect the stability of serovar E inclusions 
; however, this disparity may reflect strain differences.
We observed that GBF1 and Arf1 have little overlap at the inclusion membrane during infection even though they extensively colocalize at the Golgi surrounding the inclusion. This observation could reflect a transient and dynamic interaction at the inclusion. Alternatively, we favor the hypothesis that GBF1 activates Arf1 at the Golgi and then Arf1-vesicles containing SM traffic to and fuse with the inclusion. Under these circumstances, GBF1-dependent activation of Arf1 at the Golgi rather than at the inclusion would be important, which could explain why there is little colocalization of these proteins at the inclusion. Consistent with this notion, we found that inhibition of GBF1 activity resulted in a loss of inclusion localized Arf1, except at the region of abutting inclusion membranes. This finding suggests that GBF1 is required for Arf1 recruitment to the outer region of the inclusion but not for Arf1 maintenance at closely apposed inclusions. We are currently investigating the role of Arf1 at this region.
Recent studies have revealed that Arf1 and GBF1 also play important roles in lipid droplet biogenesis 
. As lipid droplets are translocated into the lumen of the inclusion 
, interference with Arf1 and/or GBF1 function could also affect Chlamydia
interaction with lipid droplets. Notably, GBF1 colocalizes with Lda3 (unpublished results), a chlamydial protein that localizes to lipid droplets 
now joins the growing list of pathogens that subvert GBF1 or Arf1 function to establish an intracellular replicative niche, including poliovirus, coxsackievirus, coronavirus, hepatitis C virus, and Legionella pneumophila 
. In contrast to these pathogens where depletion or inhibition of GBF1 by BFA significantly impairs replication, our findings demonstrate that GBF1 is more important for establishing a stable intracellular niche in which C. trachomatis
replicates. Both poliovirus and coxsackievirus encode a protein (3A) that binds to and co-opts GBF1 function, while L. pneumophila
secretes into the cell RalF, an Arf GEF, that regulates Arf1 localization to the L. pneumophila
-containing phagosome 
. Since GBF1 was not localized on the inclusion membrane, it is possible that GBF1 communicates indirectly with the inclusion through a variety of host proteins or lipids that have been shown to bind GBF1 and that localize at or near the inclusion, including Rab1 
, PI4P 
, and the Golgi matrix proteins, golgin 84 and p115 
. It will be important to determine whether the alterations in the actin and vimentin cytoskeleton during infection are directly or indirectly modulated by GBF1.
In summary, we show that C. trachomatis hijacks components of both vesicular and non-vesicular lipid trafficking pathways for SM acquisition, and that the SM obtained from these pathways is utilized in different ways by the pathogen. We hypothesize that SM acquired by CERT-dependent transport of ceramide and subsequent conversion to SM is necessary for C. trachomatis replication whereas SM acquired by the GBF1-dependent pathway is essential for inclusion growth and stability. Together, these observations provide an intriguing explanation for why inhibition of vesicular trafficking alone fails to affect intracellular replication. Moreover, our results describe the first example of a bacterial pathogen to co-opt CERT and reveal a novel strategy by which this organism creates its own SM biosynthetic factory. This work has identified novel targets that may prove useful in combating Chlamydia infections.