Chlamydia trachomatis serovars D-K account for the most prevalent bacterial sexually transmitted infections (STIs) worldwide. Despite aggressive control efforts, C. trachomatis infections have continued to constitute a serious public health risk (World Health Organization, 2011; Rekart et al., 2013). Infection may result in cervicitis, and in some women, C. trachomatis may ascend into the endometrium and Fallopian tubes, where it can establish a chronic infection leading to diseases such as pelvic inflammatory disease (PID), ectopic pregnancy, and infertility. C. trachomatis infections of women also pose a risk to infants, as infants born from mothers with C. trachomatis infections can develop conjunctivitis and/or pneumonia. Finally, epidemiological evidence indicates that C. trachomatis infection of the reproductive tract also may increase the risk of HIV transmission, making the study and understanding of the pathogenicity of this bacterium imperative.

In adapting to an intracellular niche, C. trachomatis has evolved a notably reduced genome of ~1 million base pairs that supports a unique developmental cycle (Stephens et al., 1998). This cycle typically involves two forms: infectious elementary bodies (EBs) and dividing, metabolically active reticulate bodies (RBs) (Moulder, 1991). Under stressful conditions in vitro, however, C. trachomatis can enter into an alternative viable but non-dividing growth form, termed a persistent growth form (Beatty et al., 1994b). This persistent growth state extends the host-bacterial relationship in vitro and has thus been proposed to be linked to chronic infection and adverse outcomes, including elicitation of tissue-damaging host responses, in vivo (Hogan et al., 2004; Darville and Hiltke, 2010). Persistent forms are also less responsive to antimicrobial therapy in vitro (Wyrick and Knight, 2004; Reveneau et al., 2005) and in vivo (Byrne, 2001; Phillips-Campbell et al., 2014). Several excellent reviews provide extensively insightful descriptions of the chlamydial persistent growth state and the potential to establish a chronic relationship with the host (Beatty et al., 1994b; Hogan et al., 2004; McClarty et al., 2007; Wyrick, 2010; Schoborg, 2011; Lo et al., 2012).

C. trachomatis growth exclusively takes place within an inclusion, a membrane-bound vacuole. Despite the presence of this barrier, C. trachomatis actively communicates with the host cells. One method of interaction is the coordination of trafficking specific subsets of host vesicles to and from the inclusion, enabling delivery of the inclusion components and nutrients required for infection (Fields and Hackstadt, 2002). C. trachomatis also secretes numerous effectors with host cell-modulating activities across the complex membranes of the bacterium and the inclusion or host cytoplasm to host cell compartments. Although diverse secretory pathways, including the type II (Sec), type III, and type V (autotransporter) secretion systems, have been shown to play a role in the translocation of protein effectors (Crane et al., 2006; Valdivia, 2008; Chen et al., 2010a; Mueller et al., 2014), a mechanism for robust delivery of complex bacterial components to host cells is likely to be mediated by membrane vesicles (MVs) that emerge from the envelope of growing bacteria (Giles et al., 2006; Giles and Wyrick, 2008; Wang et al., 2011a; Frohlich et al., 2012). As part of bacterial growth and/or envelope stress responses, the formation of MVs is a universal feature found in all Gram-negative bacteria (Beveridge, 1999; Kulp and Kuehn, 2010), Mycobacterium spp. (Prados-Rosales et al., 2011), and Gram-positive bacteria such as Bacillus spp. (Dubey and Ben-Yehuda, 2011). Herein, we review recent findings on the mechanistic aspects of bacterial adaptation to stress with a focus on how C. trachomatis remodels its envelope, produces MVs, and the potential important consequences of MV production with respect to host-pathogen interactions.

One of the critical developmental events common to Chlamydia spp. is the ability of the organisms to adapt their envelopes for the purpose of interacting with the host cell. Like other Gram-negative bacteria, the C. trachomatis envelope consists of an outer membrane (OM), an inner membrane (IM), and a periplasm (Figure 1A). Recent elegant studies by several groups have revealed that chlamydiae possess functional peptidoglycan necessary for bacterial division, despite the lack of the cytoskeletal protein FtsZ (McCoy and Maurelli, 2006; Ouellette et al., 2012; Pilhofer et al., 2013; Liechti et al., 2014). Nevertheless, a unique feature of the chlamydial envelope, distinguishing them from other Gram-negative bacteria, is the presence of a disulfide bond cross-linked major outer membrane protein (MOMP) with the periplasmic localized OmcB and the lipoprotein OmcA only in EBs (Hackstadt et al., 1985; Hatch, 1996). Both OmcB and OmcA contain abundant cysteines. This cross-linkage is believed to contribute to cell wall rigidity and osmotic stability of the EBs.

Proteins of the OM are important for various purposes including envelope architecture, virulence, transport, cell division, induction of inflammatory cytokine production, and immune evasion (Hatch, 1996; Stephens and Lammel, 2001; Abdelsamed et al., 2013). All of these tasks are presumably related to the generation and function of MVs that will be discussed in the later sections of this review. MOMP is the most abundant surface-exposed protein of both RBs and EBs (Caldwell et al., 1981; Hatch, 1996) and it functions as an adherin and a porin with its β-barrel structure (Wang et al., 2006; Sun et al., 2007). MOMP also has alternative conformations that may adapt to specific chlamydial growth stages (Feher et al., 2013) and may impart different levels of immunogenicity. Also located in the OM is a family of polymorphic membrane proteins (Pmps) or autotransporters unique to Chlamydia spp. (Stephens et al., 1998). C. trachomatis encodes nine Pmps (PmpA-I) that are either temporally or constitutively expressed (Tanzer and Hatch, 2001; Tan et al., 2010). Such temporal expression may promote antigenic variation, tissue tropism, and differential disease severity (Gomes et al., 2006; Tan et al., 2009; Taylor et al., 2011; Abdelsamed et al., 2013). Caldwell's group has reported that PmpD is a species-common, pan-neutralizing target (Crane et al., 2006). Other OM proteins studied include plasmid encoded Pgp3 (Chen et al., 2010b), PorB (Kubo and Stephens, 2000), HSP70 (DnaK) (Raulston, 1995), and Omp85 (YaeT) (Stephens and Lammel, 2001) that exists in both EBs and RBs (Figure 1B) (Liu et al., 2010). In bacteria, Omp85 facilitates the insertion of assembly intermediates from the periplasm to the OM (Ricci and Silhavy, 2012).

Many chlamydial envelope components are ligands recognized by host pattern-recognition receptors that induce inflammatory cytokine production and generate adaptive immune responses (Wang et al., 2010; Taylor et al., 2011; Abdelsamed et al., 2013). Some envelope components may also contribute to immune evasion. To this end, several chlamydial envelope components have been considered vaccine candidates (Crane et al., 2006; Schautteet et al., 2011; Hafner et al., 2014). Chlamydia spp. may quantitatively or qualitatively change their envelope structures as a result of adaptation to developmental signals or environmental cues (Beatty et al., 1994b; Hatch, 1996; Carrasco et al., 2011). The plasticity of the envelope has important implications for the design of control strategies against bacterial infection because of its importance in determining susceptibility to host defenses and antibiotics (Tan et al., 2009; Hurdle et al., 2011).

Matsumoto and Manire described the first ultrastructure of an aberrant RB form induced by penicillin in C. psittaci-infected L929 cells as viewed by transmission electron microscopy (TEM) (Matsumoto and Manire, 1970). They found that, in the presence of penicillin, RBs no longer divided, the RB to EB transition was disrupted, and aberrant multinucleated RBs accumulated. Concurrently, abundant vesicles were pinched off the aberrant RBs. Removal of penicillin allowed for the chlamydiae to reenter a normal growth state. Many elegant studies since have revealed that the viable but nondividing altered persistent form occurs when Chlamydia spp. are exposed to a variety of stress conditions. These include IFNγ (Beatty et al., 1994a), sub-inhibitory concentrations of antibiotics (Engel, 1992; Wyrick and Knight, 2004), nutrient deprivation (Beatty et al., 1994a; Raulston, 1997; Harper et al., 2000), co-infection with either herpes simplex virus (Vanover et al., 2008; Prusty et al., 2012) or Toxoplasma gondii (Romano et al., 2013) in vitro, and infection with Chlamydia-phage (Hsia et al., 2000). These stressors are often encountered by pathogens during infection in vivo (Wyrick, 2010). IFNγ, one of the stressors commonly studied, is a key component of immunity to intracellular pathogens. IFNγ induces expression of indoleamine-2, 3dioxygenase (IDO) that catalyzes the initial step in the degradation of L-tryptophan to N-formylkynurenine and kynurenine in eukaryotic cells (Beatty et al., 1994a). Such tryptophan depletion profoundly interferes with chlamydial growth, which, depending on the IFNγ concentration and exposure time, induces the bacteria to enter into a persistent growth form or results in bacterial eradication. Exposure to β-lactam antibiotics is another well-studied stressor that causes chlamydiae to enter into a persistent growth state and is used in simulating an inadequate antimicrobial treatment of chlamydiae infection (Matsumoto and Manire, 1970; Gerard et al., 2001; Giles and Wyrick, 2008; Carrasco et al., 2011; Wang et al., 2011a; Ouellette et al., 2012; Phillips-Campbell et al., 2014). Chlamydia spp. have a wide range of responses to stress. A clear shift of gene expression profiles specific to each stressor has been consistently found in many host cell types (Beatty et al., 1993, 1994b; Gerard et al., 2001, 2013; Molestina et al., 2002; Belland et al., 2003a,b; Ouellette et al., 2006; Lo et al., 2012). Changes in gene expression induced by stress at the levels of transcription and translation are consistent with the nature of a persistent growth form that undergoes DNA replication, but not division by binary fission. Interestingly, IFNγ exposure resulted in a global transcriptional upregulation lacking increased translation in chlamydiae (Ouellette et al., 2006). In contrast, penicillin exposure induced an alteration in transcription coupled to a change in translation (Ouellette et al., 2006). These data support that the impact of each stressor on bacteria is mediated by different mechanisms, and the bacterial response to each stress may be distinct.

To explore the mechanisms by which C. trachomatis survives under persistence inducing conditions, we utilized human primary endocervical epithelial cells, the main site of C. trachomatis infection in vivo. We found that normal C. trachomatis forms could develop in cultured primary cells similarly to HeLa cells (Wang et al., 2011a). Figures 2A–C show that exposure to ampicillin or IFNγ led to aberrant RB phenotypes and an accumulation of abundant MVs generated by C. trachomatis in cultured cells. Additionally, confocal microscopy and cell fractionation analyses demonstrated that the secreted chlamydial T3S effectors, CopN and Tarp, and the secreted protease, CPAF (Zhong et al., 2001) were decreased in ampicillin-induced persistent forms (Wang et al., 2011a), the latter of which is consistent with previous observations using IFNγ exposure or iron-deprivation persistence models (Shaw et al., 2002; Heuer et al., 2003). Changes in the spectrum of host cytosolic chlamydial proteins may underlie the host-pathogen relationship because of their importance in modulating host cell signaling (Valdivia, 2008; Zhong, 2011; Mueller et al., 2014).

We have been particularly intrigued by the observed vesicular structures that are derived from both pathogen and host cells during C. trachomatis infection in culture, as they may serve as a vital element for host-pathogen interactions. C. trachomatis is likely to encounter numerous known and not-yet-identified conditions in vivo. Of these, varying levels of IFNγ in the endocervix during chlamydial infection in vivo are likely crucial as demonstrated by studies in vitro, in animal models, and in observational studies in humans (Arno et al., 1990; Beatty et al., 1994a; Byrne, 2001; Aiyar et al., 2014; Lewis et al., 2014). The direct involvement of persistent growth forms in pathogenesis in vivo in humans is challenging to prove, but several TEM studies have visualized atypical pleomorphic RBs and aberrant C. trachomatis forms in individuals with chronic infections, in Fallopian tube tissues, and in the synovium of reactive arthritis patients (Patton et al., 1994; Nanagara et al., 1995; Mazzoli et al., 2000; Bragina et al., 2001). Very recently, the Quayle laboratory developed methodology to sample endocervical cells and components of the endocervical environment in C. trachomatis infected women. By using a cytobrush to retrieve cells followed by immediate placement of these brushes in modified TEM fixative, the ultrastructure of “in vivo” chlamydial growth forms in endocervical epithelial cells could later be visualized by TEM (Lewis et al., 2014). These TEM analyses revealed that chlamydial infection in the human endocervix could result in a variety of inclusion types, containing normal forms, a mixture of relatively “normal” and aberrant forms, or inclusions with highly aberrant forms, and this varies between patients (Lewis et al., 2014). Since MVs are so small in size and lack unique biological markers for identification by standard immunostaining techniques, we have begun to take advantage of these samples to examine MV production in vivo. High magnification TEM images were obtained from cross sections of several inclusions and indicated the in vivo presence of C. trachomatis MVs in inclusions sampled from two different patients (Figures 3A–D). MVs appeared to be single membrane structures and apparently associated with the OM of chlamydial organisms similar to those observed in vitro (Figure 2). The double membrane structures of C. trachomatis forms are intact, which is an indication of faithful sample preservation. This preliminary but novel observation provides an important link to those observations made in vitro, indicating that MVs are a component of C. trachomatis infection in vivo in humans. Further examination of samples using this methodology is in progress to determine the relationship between MV formation and the bacterial growth state in natural human infections.

The best-studied mechanism that is related to OM changes in Gram-negative bacteria is the formation of MVs during both physiological adaptation and responses to envelope stress. Bacterial MVs are not only compositionally similar to the OM, containing LPS, phospholipids, and OM proteins, but also contain selective periplasmic or cytoplasmic components, such as toxins, DNAs, and RNAs, depending on the strain (Kadurugamuwa and Beveridge, 1999; Beveridge, 1999; Kuehn and Kesty, 2005). Research demonstrates the potential of MVs as vehicles of pathogenicity, as they can deliver complex bacterial molecules to target cells (Kuehn and Kesty, 2005; Bomberger et al., 2009; Amano et al., 2010; Elmi et al., 2012). The MVs also involve immune activation or suppression, stress responses, and attachment and internalization of the bacteria. Because of the cargo carrier nature and potent built-in adjuvanticity of most MVs studied to date, they are being utilized as vaccines (Collins, 2011; Unal et al., 2011). Engineered MVs in E. coli have exhibited a promising robust and tunable platform for the development of recombinant multivalent vaccines (Chen et al., 2010c; Bartolini et al., 2013).

Although chlamydial MVs were first observed over four decades ago (Matsumoto and Manire, 1970; Stirling and Richmond, 1980), the implications of these multi-functional MVs are only beginning to be elucidated (Giles et al., 2006; Giles and Wyrick, 2008; Wang et al., 2011a; Frohlich et al., 2012). A puzzle remains with regard to what role chlamydial MVs play during infection. Research data, including our own, suggest that the generation of chlamydial protein containing MVs occurs during productive infection and is enhanced by stress (Figures 2, 3) (Matsumoto and Manire, 1970; Giles et al., 2006; Giles and Wyrick, 2008; Wang et al., 2011a; Frohlich et al., 2012). The presence of MVs, not only during the β-lactam induced aberrant state, but also during both tryptophan starvation via IFNγ exposure and in vivo infection demonstrate a universal mechanism that appears to be responsive to stress rather than merely a by-product of dysfunctional membrane biogenesis during β-lactam exposure.

Despite tremendous efforts, mechanisms by which bacteria produce MVs while sustaining their viability remain unclear. Based on studies in model organisms, including P. aeruginosa and E. coli, several mechanisms of MV formation have been proposed. First, the loss of OM-PG cross-links may allow excess lipid formation to induce MV formation. Second, accumulated periplasmic proteins simply push out the OM, trapping large amounts of protein inside the resulting vesicle. Third, the involvement of integral membrane proteins or signal molecules [such as Pseudomonas quorum sensing (PQS) molecules] induces membrane curvature and vesiculation (Schertzer and Whiteley, 2012). Fourth, the presence of rough LPS in bacteria contributes directly and indirectly to the formation of MVs (Kadurugamuwa and Beveridge, 1995; Sabra et al., 2003). Finally, genetic mechanisms may exist to regulate MV formation (McBroom et al., 2006; Schwechheimer et al., 2013). These MV inducing factors that have been proposed may not be mutually exclusive, and multiple dynamic surface components may contribute to MV production in bacteria (Kulp and Kuehn, 2010; Schwechheimer et al., 2013).

We know surprisingly little about OM biogenesis, and even less is known about the related vesicle formation in C. trachomatis. It is likely that after synthesis in the bacterial cytoplasm, chlamydial protein translocation to the OM involves complex steps conserved in Gram-negative bacteria (Hagan et al., 2010; Silhavy et al., 2010). These include: (i) translocation across the IM via the Sec pathway; (ii) folding facilitated by periplasmic chaperones/proteases, and (iii) assembly and insertion into the lipid phase of the OM, a process facilitated by a β-barrel assembly machine (BAM) complex that consists of Omp85 and its interacting lipoproteins (Ricci and Silhavy, 2012). Also necessary for maintenance of EB envelope integrity is the disulfide bond cross-linking of MOMP with OmcA and OmcB (Hatch, 1996).

We hypothesize that both bacterial and host factors contribute to the generation and release of MVs from C. trachomatis. We envision different mechanistic scenarios involved in MV formation during productive infection or the persistent growth state induced by ampicillin. In the case of productive infection, MV generation may reflect local OM deformation and/or a physiological turnover of envelope components because of envelope remodeling, a critical process in chlamydial development. In fact, RB multiplication and division require a rapid surface expansion or alteration achieved by the synthesis, assembly, and insertion of lipid or non-lipid OM components, while transition from a large RB to a small EB undergoes the opposite. In contrast, ampicillin induced hyper-vesiculation phenotypes may be the result of an envelope stress response. Exposure to ampicillin blocks RB division, although DNA replication remains. These can, in turn, result in attenuation of OM protein expression, accumulation of misfolded proteins in the periplasm, activation of proteolytic processes, and interference with the correct assembly and insertion of proteins in the OM. All of these are likely to induce heightened vesiculation. MV formation may offer a means to remove stress caused by the accumulated “toxic” waste or unfolded proteins. Chlamydial HtrA, a key player in envelope stress response (Huston et al., 2007, 2008; Zhong, 2011), is likely to play a role in controlling the formation of MVs, as is observed in other bacteria (McBroom et al., 2006). Given the strong capacity of chlamydial exploitation of host cellular machinery, MV cargo delivery may partially depend on a yet to be defined host trafficking pathway. Whether or not a potential host factor contributes to the formation of chlamydial MVs remains to be determined. Nevertheless, studying molecular mechanisms underlying envelope adaptation to development or stress offers a powerful tool and new route to understanding how C. trachomatis adapts to and survives in nutrient rich but hostile intracellular niches.

The production of MVs in bacteria is a universal mechanism whereby virulence factors, signaling molecules, and genetic materials can be packaged and effectively transported to target cells. Major challenges in the field of vesicle research with obligate intracellular bacteria include (i) developing new comprehensive approaches for vesicle isolation and characterization, (ii) isolating pure populations from bacteria or host cells, and (iii) monitoring vesicle dynamics under physiologic relevant conditions. Nevertheless, the information already available indicates that MVs produced during chlamydial infection are present in cultured cell lines, primary human endocervical epithelial cells infected with C. trachomatis, and more importantly, in clinical specimens from C. trachomatis infected patients. These data provide preliminary evidence that the MV is a component of chlamydial infection in vivo. We hypothesize that C. trachomatis varies OM organization and produces MVs during the developmental cycle and in response to stress. Many mysteries remain with regard to how MVs contribute to host-pathogen interactions, the relationship of developmental envelope biogenesis within the context of MV formation and regulation, and the modes of MV cellular targeting and delivery. Improved understanding of the mechanisms of MV mediated mass material exchanges and the effects on C. trachomatis survival may have relevance to several important aspects of chlamydial biology, including the C. trachomatis persistent growth state in vivo. Recent progresses in chlamydial genetics may have great potential to inspire significant advance in MV studies in C. trachomatis (Wang et al., 2011b; Gérard et al., 2013; Gong et al., 2013; Song et al., 2013; Bauler and Hackstadt, 2014). Since global control of C. trachomatis infection will best be achieved with a vaccine, developing a greater understanding of the functional role and the mechanisms of envelope modification and MV production may open new therapeutic avenues against this medically important intracellular bacterium.