Pathogenic Chlamydia Lack a Classical Sacculus but Synthesize a Narrow, Mid-cell Peptidoglycan Ring, Regulated by MreB, for Cell Division

The peptidoglycan (PG) cell wall is a peptide cross-linked glycan polymer essential for bacterial division and maintenance of cell shape and hydrostatic pressure. Bacteria in the Chlamydiales were long thought to lack PG until recent advances in PG labeling technologies revealed the presence of this critical cell wall component in Chlamydia trachomatis. In this study, we utilize bio-orthogonal D-amino acid dipeptide probes combined with super-resolution microscopy to demonstrate that four pathogenic Chlamydiae species each possess a ≤ 140 nm wide PG ring limited to the division plane during the replicative phase of their developmental cycles. Assembly of this PG ring is rapid, processive, and linked to the bacterial actin-like protein, MreB. Both MreB polymerization and PG biosynthesis occur only in the intracellular form of pathogenic Chlamydia and are required for cell enlargement, division, and transition between the microbe’s developmental forms. Our kinetic, molecular, and biochemical analyses suggest that the development of this limited, transient, PG ring structure is the result of pathoadaptation by Chlamydia to an intracellular niche within its vertebrate host.


Introduction
Chlamydia is an obligate intracellular pathogen and the single most prominent cause of bacterial sexually transmitted infections and infectious blindness worldwide. Frequently referred to as the 'silent epidemic', chlamydial infections are often asymptomatic, which results in a lengthy delay between infection and the onset of disease symptoms [1]. Approximately 1.4 million Chlamydia infections are reported in the United States annually [2,3] and an estimated 90 million individuals are believed to be infected globally [4]. Untreated chlamydial genital infections can result in cervicitis, pelvic inflammatory disease, and ectopic pregnancy in women and urethritis in men.
Chlamydia has undergone a lengthy (>700 million year) adaptation to an intracellular environment in addition to its more recent co-evolution with humans and other vertebrate hosts possess almost all of the genes of the PG biosynthesis pathway [23]. Numerous studies have since shown that the vast majority of the proteins encoded by these genes are functional in vitro or when expressed in E. coli [24][25][26][27][28][29]. The chlamydial genome appears to lack only a few PG synthesis genes, such as glutamate / alanine racemases and transglycosylases, which are essential for making PG subunits and polymerizing these subunits into PG chains, respectively [23] (Fig 1a). In addition the Chlamydia genome lacks a gene encoding FtsZ, a cytoskeletal cell division initiation protein that organizes numerous PG biosynthetic enzymes around the cell division plane in almost all bacteria [7].
Recently, a method was developed to label bacterial PG using the inherent promiscuity of PG biosynthetic enzymes for tagged, fluorescent D-amino acid (FDAA) probes as substrate analogs [30,31]. Our work with the next generation of bioorthogonally tagged D-amino acid dipeptide (DAAD) probes that mimic DA-DA during PG synthesis (Fig 1a and 1b) revealed PG in C. trachomatis [9], providing the first direct evidence of its existence in these organisms. Another study confirmed the presence of PG in an environmental strain of Parachlamydiaceae, the amoebae-symbiont, Protochlamydia amoebophila [6,10], providing evidence that both related phyla synthesize PG. Strikingly, while PG in P. amoebophila forms a typical, cell-encompassing sacculus, PG in C. trachomatis forms only a distinct ring-like band at its mid-cell.
Despite these advances, many questions remain concerning the function of PG in Chlamydiae and the significance of its ring-like structure. Through the use of 3D super resolution structured illumination microscopy (SIM) and clickable DAADs we define the ring-like PG structure of C. trachomatis as a 140 nm wide, dynamic ring that forms immediately after the previous cell division and follows cell constriction at the division septum. We show that this limited PG ring is also present in other pathogenic chlamydial species; C. muridarum, C. caviae, and C. psittaci. Formation of the ring is non-uniform and directly linked to the bacterial actin-like cytoskeletal protein MreB. When MreB polymerization is inhibited the PG ring is rapidly and non-uniformly turned over, suggesting competition between two coordinated, but separable processes: MreB-linked PG synthesis and an unknown turnover mechanism. We propose a reshaping model to explain how this narrow PG ring facilitates both cell enlargement and division. We also propose that Chlamydia limits the timing of PG ring assembly and dissociation to the intracellular replicative phase, allowing the pathogen to moderate its detection by the host immune system. These results suggest that the absence of a PG sacculus by pathogenic Chlamydia is the result of pathoadaptation to its intracellular niche within vertebrate hosts.

Pathogenic Chlamydia lack a classical PG sacculus
Despite the demonstration of a conventional PG sacculus in P. amoebophila [10], similar PG isolation and labeling techniques (e.g. using FDAAs) proved unsuccessful in studies of pathogenic Chlamydia [9,16,19]. We hypothesized that distinct morphological differences in PG configuration and structure might exist between pathogenic Chlamydia and the distantly related environmental endosymbionts. In our previous study we used DAADs (Fig 1b) with diffraction-limited conventional fluorescence microscopy to show that labeling of PG is constrained to a thin band in C. trachomatis [9]. This is in stark contrast to the uniform labeling of cell periphery observed in all other PG-containing bacteria that produce peripheral PG sacculi [9]. 3D super-resolution microscopy confirmed that PG from C. trachomatis grown in the presence of DAADs for several generations formed as a narrow ring (Fig 2).
This unique PG localization is common to the Chlamydia genus, as the DAAD-labeled PG of three evolutionarily representative, pathogenic, veterinary chlamydial species, C. muridarum, C. caviae, and C. psittaci, also localized to a single, ring-like structure (Fig 2) and none Fig 2. Structured illumination microscopy of DAAD PG labeling in pathogenic Chlamydia. Structured illumination microscopy (SIM) was conducted on four pathogenic Chlamydia species. EDA-DA was added 2 hours post infection (hpi) and coverslips were fixed at 18 hpi. PG labeling (represented by EDA-DA) is shown in green, the major C. trachomatis outer membrane protein (MOMP) is shown in red, and cell nuclei are in blue (this labeling scheme is maintained in all subsequent figures, unless otherwise stated). MOMP staining is not shown for the three other species. Images are representative of~20 inclusions viewed per strain. Scale bar = 1 μm. of these species had peripheral PG labeling. In contrast, DAAD labeling of the coccus-shaped Staphylococcus aureus, which synthesizes PG predominantly at the septum to produce a sacculus [30], resulted in uniform cell surface labeling (S1 Fig). Thus, we conclude that the restriction of DAAD labeling to a narrow ring at mid-cell is a unique and defining characteristic of pathogenic chlamydial species.
Non-replicative chlamydial EBs do not retain PG Evasion of the immune response is a powerful evolutionary driver for many bacterial pathogens. Because the human innate immune system recognizes and responds to PG [8], we hypothesized that Chlamydiaceae may limit PG synthesis to where and when it is absolutely needed, i.e. the division site of actively replicating cells. Since the immune system is most likely to interact with the infectious, extracellular EBs, we reasoned that Chlamydia may exclude PG from EBs and restrict its synthesis to the metabolically active intracellular RBs, as has been suggested previously [19,21,32]. This hypothesis is bolstered by the fact that the labeling of chlamydial PG was previously shown to occur no earlier than 8 hpi, coinciding with the EB to RB transition of C. trachomatis [9].
The use of asynchronous infections allow for the visualization of Chlamydia at various stages of the developmental cycle simultaneously, and the diameters of MOMP-labeled cells can be used to distinguish between RBs ( 1.0 μm diameter) and EBs (~0.3 μm diameter). When asynchronous infections are carried out in the presence of PG-labeling probe EDA-DA, PG labeling localizes to chlamydial RBs and appears absent in the much smaller, non-replicative EBs (Fig 3a). Given that the EBs present at 22 hpi may be noninfectious and thereby not accurately reflect the labeling potential in normal, healthy EBs, we conducted additional labeling experiments for extended durations (40 hrs), allowing enough time for completion of the Chlamydia developmental cycle. By 40 hpi, the vast majority of RBs containing labeled PG within mature inclusions have differentiated back into infectious EBs. In mature inclusions there was no detectable PG label in these smaller, MOMP-labeled particles (Fig 3b). We also made use of a simple, biological separation method for establishing whether Chlamydia EBs retained labeled PG after transitioning from replicating RBs. We infected cells with C. trachomatis for 18 hours and then incubated them in the presence of a DAAD (4 mM) for an additional 25 hours. This method allowed labeling of the bacteria in the presence of the DAAD for an entire developmental cycle, i.e. RB to EB. Chlamydia EBs were then harvested from mature inclusions 43 hours post infection (hpi) and used to infect a fresh cell monolayer. DAAD (4 mM) was present during all steps of collection and infection. Infected cells were then fixed at 3 hpi to ensure that only EBs were observable (RBs are incapable of invading host cells [16]) and that invading EBs did not have sufficient time to differentiate into RBs (EB to RB transition for C. trachomatis L2 strain BU/434 occurs~8 hpi). None of the newly invading chlamydial EBs (identified by MOMP labeling) contained labeled PG (Fig 3c). These results suggest that EBs do not synthesize PG nor do they retain any PG synthesized during the RB stage.
Replicating RBs maintain a narrow ring of PG that aligns with the septal division plane The use of confocal microscopy in our previous study [9] severely limited resolution of the PG rings. In order to better characterize the localization of PG rings within chlamydial RBs, we utilized SIM microscopy to examine fluorescently labeled PG in intracellular RBs labeled for MOMP. We found that due to the compact nature of RBs within chlamydial inclusions, defining individual cell boundaries in 3D SIM projections was often challenging (Fig 4a). We decided to augment our visual characterization by also examining individual imaging planes (as opposed to rendered 3D projections). By eliminating foreground and background fluorescence contamination (Fig 4b-4e, S2 Fig, S1 Video) we were able to generate additional images that clearly delineate individual chlamydial RBs, and thereby more accurately report localization of their PG (S3 Fig, S2, S3 and S4 Videos).
Detailed, high-resolution SIM analysis of DAAD-labeled PG in C. trachomatis at 18 hpi, when the majority of Chlamydia within inclusions are in the actively growing/dividing RB phase, revealed the following features: 1) PG rings were confined to mid-cell: in dividing bacteria in which a membrane invagination was clearly visible, PG localized to the division septum (Fig 4a and 4b   RBs take up and incorporate DAADs rapidly and processively during exponential growth In rod-shaped bacteria, assembly of the complete division complex (and accompanying septal PG) occurs at the mid-cell just prior to the initiation of cell division, and only after a lengthy cell elongation phase [7]. Similarly, the model coccus-shaped organism, Staphylococcus aureus, undergoes a period of peripheral growth before onset of septal PG synthesis [33]. In contrast, although C. trachomatis strain L2/434 has a generation time of~2.5 h to 3 h [34,35], experiments with long DAAD incubation times (e.g. 18 h) revealed complete septal PG rings in the vast majority of RBs present within the developing inclusion (Fig 4a and 4b, S3 Fig). This suggests that synthesis and assembly of the septal PG begins unusually early in the cell cycle compared to other bacteria. In order to investigate the dynamics of PG ring formation, we varied the time that RBs were exposed to DAADs. Pulse labeling experiments on actively dividing RBs (18 hpi) confirmed that complete PG rings were detectable with as little as 10 min of DAAD labeling (Fig 5a), indicating that the basic PG ring assembly is also rapid. Although there was no significant change in the overall ring structure (i.e. a relatively constant 140 nm PG ring width) after labeling pulses longer than 10 min, the DAAD signal intensity gradually increased with increasing pulse duration. We later quantified this, allowing us to conduct the first kinetic analysis on the uptake and transport of any substrate by intracellular Chlamydia.
The close proximity of individual RBs within chlamydial inclusions and the difficulty in comparing fluorescence intensities of PG rings in dissimilar spatial orientations make single cell PG signal quantification practically impossible. Therefore, as a general quantitative approach, we tracked average epifluorescence values for individual inclusions using maximum intensity projections and the MOMP signal to normalize the average cell density per inclusion. We utilized an asynchronous infection model, allowing us to view a broad range of inclusion sizes (representing various stages of maturation / development) at each given time point. The range of inclusion sizes was similar for all time points, and the average inclusion size did not significantly differ between time points analyzed. Measurement of new DAAD incorporation over 2 h (from 18 to 20 hpi) showed that while mean MOMP fluorescence remained relatively constant (Fig 5b), mean PG fluorescence per inclusion gradually increased over the same time span (Fig 5c). No correlation was detected between inclusion size and average pixel intensity for either labeled PG or the major chlamydial outer membrane protein (MOMP) (S4a- S4c  Fig), indicating that asynchronous chlamydial infections did not significantly affect the experimental outcome. Together, these results indicate that DAADs are continuously incorporated into one basic PG ring structure over several hours (Fig 5b and 5c) during the replicative phase of the chlamydial developmental cycle.

Chlamydial PG ring assembly correlates with non-uniform localization of MreB patches
Rapid, yet persistent chlamydial PG ring assembly would require tight spatiotemporal coordination of new PG incorporation along the ring. We investigated the degree to which these mechanism(s) are coordinated by the chlamydial cell division machinery. The bacterial tubulin-like protein FtsZ plays an essential role in cell division by assembling the division apparatus at the correct division plane in almost all bacteria, chloroplasts, some mitochondria, and even some archea [7,[36][37][38]. Chlamydiae [9,10] and its distant relative Planctomyces [39,40] constitute the only two phyla of PG-containing bacteria that lack an FtsZ homolog [41]. Studies now suggest that in the absence of FtsZ, MreB may act as the division plane organizer in Chlamydia, since both MreB and the septal organizer RodZ localize to the division septum [42][43][44]. MreB is an actin-like protein that controls the cylindrical nature of rod-shaped bacteria by forming dynamic patches traveling around the circumference of the side walls. Since DAADs labeled chlamydial PG at the division septum, and chlamydial MreB has previously been shown to localize to the division septum [44], it is a logical candidate for facilitating PG ring assembly. By combining anti-chlamydial MreB antibody [44] with DAADs and 3D-SIM imaging, we found that in RBs chlamydial MreB is patchy, similar to the localization patterns reported in other bacteria [45][46][47], and that these MreB patches appeared to co-localize with PG rings (Fig  6a). Inhibition of filament formation with MreB depolymerizing agents, A22 and MP265 [48,49], resulted in a loss of MreB labeling in normal chlamydial RBs (S5a  [44]. These results strongly suggest that MreB polymerization is required for new PG incorporation along the septal PG ring structure in Chlamydiae. Because MreB moves circumferentially along the lateral walls of many bacterial species [45][46][47], we hypothesized that MreB could be similarly dynamic along the chlamydial PG ring. We reasoned that we could infer MreB dynamics by labeling its downstream product, namely newly incorporated PG, with DAADs. Though indirect, a brief DAAD pulse would record the movement of MreB on the PG ring in the form of a fluorescent trace of new DAAD incorporation. A short (five minute) pulse with DAAD resulted in non-uniform, patchy PG trace distinguishable by epifluorescence (S6a Fig) and more clearly discernable at SIM resolution ( Fig  6c). Anti-MreB signal partly co-localized with short EDA-DA pulses and the remaining signal was interspersed between PG patches, appearing to complement the newly forming PG arcs (S6b Fig). The lack of complete co-localization of MreB patches and PG patches indicates that we may lack the instrumental sensitivity to follow these rapid processes with ideal spatiotemporal resolution, however, these observations are consistent with the hypothesis that dynamic MreB patches facilitate incorporation of new PG along the septal plane, eventually completing a full PG ring in~10 min (Fig 5a).
If A22 successfully blocks MreB-dependent PG incorporation in Chlamydia, we predict that inhibition of MreB polymerization should result in the accumulation of incomplete PG muropeptides, as is the case in E. coli [38]. Because classical analytical techniques [16,17,19,21] are inadequate to detect chlamydial PG in lysates of infected cells, we employed a new, highly sensitive, immunodetection approach [50]. Chlamydial PG fragments are detected by the host cell via the intracellular NOD2 receptor, which recognizes N-acetylmuramic acid (MurNAc) present on muropeptide breakdown products of bacterial PG [50]. When coupled with mass spectrometry, this then can be used to measure the relative abundance of chlamydial PG muropeptides [50]. In order to determine if incomplete PG muropeptides accumulate when MreB polymerization is inhibited, we infected HeLa cells with Chlamydia for 18 hours followed by A22 (75 μM) treatment for 2 hours. Lysates from infected cells treated with A22 resulted in a significant increase in NOD2 signaling when compared to untreated (no A22) lysates from  [50]. Subsequent analysis via tandem mass spectrometry (MS/MS) confirmed that the breakdown products of the 477 mu peak were identical to those of an MDP fragment from C. trachomatis (S7d Fig) and column retention times for 494, 666, 653 peaks correspond to those observed previously for chlamydial MDP, MTP, and the cross-linked PG fragment [50]. In addition, we found that the relative abundance of a cross-linked, PG degradation product of 653 mu (described previously, [50]) decreased significantly in A22-treated samples (S7c Fig). As the structures for the UDP-precursors of chlamydial lipid I synthesis have not yet been identified / characterized, we cannot definitely state whether the accumulation of MDP and MTP in A22-treated cells is the result of PG degradation, incomplete lipid I synthesis, or both. In an attempt to ascertain whether these fluctuations in muropeptide abundance resulting from A22-treatment were similar to those seen when PG biosynthesis is directly inhibited, we examined samples collected from Chlamydia-infected cells that had been treated with DCS, a D-Ala-D-Ala ligase inhibitor that arrests PG biosynthesis (Fig 1). We found that, similar to A22, DCS treatment resulted in an increase in the abundance of both chlamydial MDP and MTP and a decrease in 653 muropeptide species (S7c Fig). MreB is essential for chlamydial replication, differentiation, and growth In addition to blocking PG biosynthesis, inhibition of chlamydial MreB polymerization also affected Chlamydia replication and its interaction with host cells throughout its biphasic life cycle. Consistent with previous work [43,51], we found that A22 inhibition of MreB after the EB to RB transition (8 hpi) resulted in arrest of RB replication (S8a and S8b Fig). A22-treated RBs also became slightly enlarged, though significantly less than aberrant bodies induced by exposure to antibiotics targeting various penicillin-binding proteins (PBPs), i.e. β-lactams [43]. Unlike β-lactam induced aberrant bodies that still show strong DAAD labeling [9], A22-induced aberrant bodies completely lacked detectable PG (Fig 6b). This observation suggests an essential role for MreB as well as PG in maintaining chlamydial cell size and shape.
When infected cells were treated with MreB inhibitors prior to the EB-RB transition (< 8 hpi), some aberrant body formation (~4-6%) was observable, but the vast majority of bacteria appeared to remain in an EB-like state of development, despite being intracellular for more than 16 hpi (S8b, S8c and S8f Fig). Inhibition of the normal developmental cycle is reversible as these EB-sized Chlamydia appeared to differentiate into RB sized Chlamydia when the inhibitor was removed (S8b and S8f Fig). This suggests that in addition to maintaining cell size and division, functional MreB may also be critical for the EB to RB transition. In order to establish how compounds A22 and MP265 affect the ability of Chlamydia to complete its developmental cycle, infected cells were treated for the indicated times with each compound, and EBs were collected at the termination of a normal chlamydial developmental cycle (44 hpi). Numbers of inclusion forming units (IFUs) were then obtained by re-infecting monolayers to establish the number of viable EBs present in each test group. We found that when Chlamydia-infected cells were treated with either A22 (75 μM) nor MP265 (125 μM) for brief (one hour) periods, the number of IFUs recovered was only slightly fewer than in our untreated control groups (S8d In the A22-treated group,~10 4 IFUs were recovered, whereas in the MP265-treated group~10 7 IFUs were recovered. Compound A22 has previously been reported to exhibit off-target effects in other bacterial systems [49], and we suspect that this may explain the difference in IFUs recovered between A22 and MP265 treatment groups. When IFUs were collected at 52 hpi (as opposed to 44 hpi), the number of recovered IFUs increased substantially (2-10 hpi (52 hpi) column, S8d Fig). By contrast, no similar increase in IFUs recovered was observed in MP265-treated groups that were allowed an additional 8 hours of recovery time, prior to EB harvesting.
Consistent with the idea that polymerized MreB is a general facilitator of chlamydial growth and the biphasic developmental cycle, MreB inhibition also arrested inclusion formation/ maturation as evidenced by the lack of discernable inclusion membrane protein A (IncA) labeling in the inclusion membrane upon A22 treatment, while DCS-treated bacteria still developed mature inclusions (S8e Fig). Time course studies revealed that inclusion maturation continued to be suppressed when the MreB polymerization inhibitor was added as late as 12 hpi, approximately four hours after the EB-RB transition is known to occur [52] and~2-4 hours before IncA can be detected at the inclusion membrane. In contrast, DCS-treated Chlamydia were capable of expressing IncA and incorporating it on the surface of their inclusions. When the inhibitor was removed as late as 10 hpi, chlamydial inclusions resumed normal maturation as evidenced by the appearance of IncA at the inclusion membrane with no aberrant bodies visible at 22 h (S8f Fig). Taken together, these data suggest that MreB is a key regulator of chlamydial growth and development essential for the EB to RB transition (~8 hpi), cell enlargement and division, PG biosynthesis in RBs (>8 hpi), maturation of inclusions (>12 hpi), and differentiation of RBs back into EBs.
Nascent chlamydial PG assembly is D,D-transpeptidation/ carboxypeptidation independent DAAD incorporation into chlamydial PG is independent of the D,D-transpeptidation/carboxypeptidation activity of PBPs, as evidenced by the persistent and rapid labeling in ampicillintreated aberrant bodies [9]. In the presence of β-lactam antibiotics (ampicillin or piperacillin) the PG signal followed irregular branches within the enlarged, chlamydial aberrant bodies and rarely exhibited complete ring morphology (S9a and S9b

Chlamydial PG rings lose DAAD signal evenly and slowly
Our data strongly suggest that chlamydial MreB (in combination with PBPs) is responsible for the assembly of a properly sized PG ring localized at mid-cell and the division septum in actively dividing bacteria. However, these data do not explain how PG rings follow the cell constriction. A degradation and/or reshaping mechanism must be acting on this structure, as we observe rings of different diameters but not a full division plane until the very end of cell division (Fig 4 and S3 Fig). Based on our observation that the chlamydial PG biosynthesis machinery rapidly synthesizes and maintains a defined PG ring structure along the constriction plane, we reasoned that the older PG might be degraded homogenously around the ring as new PG is synthesized; we will refer to this as the 'homogeneous model'. Alternatively, new PG could be synthesized from the center of each PG ring band and push old PG outward towards the cell poles, as occurs in PG elongation of cocci and oval species, and be degraded at the edge of the DAAD-labeled ring band thereby restricting PG to the mid-cell (S10a Fig). For simplicity, we will refer to this putative mechanism as the 'bidirectional model'.
To explore this reshaping mechanism, we first measured how PG rings age. Chlamydia inclusions (18 hpi) were labeled with a DAAD for one hour (pulse) after which cells were washed in fresh medium and then allowed to incubate in medium without probe (chase). Loss of PG fluorescence signal from labeled chlamydial inclusions over time was then quantified following appropriate controls similar to earlier experiments (S10b and S10c Fig). Average fluorescence pixel intensities of whole inclusions, calculated over a four hour chase, indicated that the mean PG signal per inclusion (Average [Total PG signal per inclusion / area of inclusion]) decreased by 50% over approximately three hours, while the mean cell density (represented by Average [Total MOMP signal per inclusion / area of inclusion]) did not appear to change (Fig  7a and 7c). In contrast, a complementary analysis of the same data set showed that while the average integrated MOMP signal per inclusion (Average [Total MOMP signal per inclusion]) steadily increased (due to an increase in the number of cells per inclusion during chases), the average integrated PG signals per inclusion (Average [Total PG signal per inclusion]) did not change (Fig 7b and 7c). This suggests that the pulsed DAAD within an inclusion is distributed between daughter cells during the chase, though it is not currently possible to ascertain whether all of the free DAAD is fully incorporated into chlamydial PG during the hour-long pulse. Some amount of unincorporated DAAD may be simply retained by replicating RBs and incorporated into newly synthesized PG at later time points.
Although the time required for mean PG signal reduction by 50% is close to the estimated doubling time of C. trachomatis strain L2/434 (2.5-3 hours) [52], the overall thickness of the rings did not appear to change with longer chase times, and PG ring width (y, in Fig 4f) remained relatively constant with an average width of~156.9 nm (sd 12.8 nm; n = 10) at t = 1 h, an average of 147.9 nm (sd 20.7 nm; n = 19) at t = 3 h and an average of 155.9 nm (sd 18.1 nm; n = 18) at t = 4 h, compared to the aforementioned 138.6 nm ± 18.6 nm at t = 0. Inherent to the design of our pulse-chase experiment, if a bidirectional mechanism existed, older DAAD signal would split in two and become thinner at the edges as unlabeled new PG is incorporated in the middle (S10a Fig). However, as we are currently operating at the limits of 3D structured illumination microscopy, future studies conducted at higher resolutions will be needed to definitely refute the bidirectional synthesis model.

Inhibition of MreB-dependent PG synthesis leads to rapid PG degradation
We previously showed that DAADs can substitute for natural DA-DA in C. trachomatis [9] and that aberrant bodies induced by β-lactam antibiotic inhibition of chlamydial periplasmic D,D-carboxypeptidases/transpeptidases (i.e. PBPs) are still capable of incorporating DAADs ( [9], and S9a, S9b and S9d Fig). These results demonstrate that chlamydial PBP2 and PBP3, which are both inhibited by ampicillin, are not required for DAAD incorporation and that incorporation of label into chlamydial PG likely occurs in the cytoplasm. The slow decrease of the PG signal and the distribution of the signal between the newly formed daughter cells observed during pulse-chase experiments (Fig 7a and 7b) could therefore be due to incorporation of DAADs trapped within the inclusions and/or the cytoplasm of chlamydial RBs. This continuous DAAD incorporation could obscure the true kinetics of the observed PG degradation mechanism. We reasoned that if incorporation of new material was arrested by inhibition of MreB, we could uncouple chlamydial PG synthesis from degradation.
When RBs (18 hpi) were pulsed with DAADs and chased with fresh medium containing A22, the signal intensity of labeled PG rings began decreasing in a non-uniform manner within 15 minutes (Fig 8a) with no significant change in PG ring width, with an average width of 151.7 nm (sd 19.9 nm; n = 17) at t = 15 min, an average of 153.9 nm (sd 16.4 nm; n = 17) at t = 30 min and an average of 142.2 nm (sd 19.6 nm; n = 9) at t = 45 min, compared to 156.7 nm (sd 17.2 nm; n = 9) at t = 0. At 30 minutes, all PG rings were almost completely dissociated with only punctate labeling present on a handful of RBs within any given inclusion. After one hour of A22 or MP265 treatment, no labeling of chlamydial PG was discernable (A22 / MP265 chase) compared to the control (first chase) (Fig 8b). Taken together, our data indicate the presence of two separable and highly active mechanisms critical for the reshaping of the chlamydial PG ring as RBs constrict during cell division: MreB-dependent PG synthesis and an as yet uncharacterized degradation mechanism(s).

Chlamydia accumulates and retains D-amino acid dipeptides
When cells were pulsed with DAADs and chased with fresh medium containing A22 (resulting in complete loss of PG signal), subsequent removal of A22 and cell recovery in fresh medium (in the absence of exogenously added DAAD or A22, second chase), resulted in a striking restoration of PG ring signal in chlamydial RBs (Fig 8a and 8b). This is indicative of relatively high concentrations of free (and/or cytoplasmic PG precursor-bound) DAADs being retained by the chlamydial cells/inclusions, despite media changes and wash steps subsequent to pulselabeling. Since MreB inhibition is reversible (S8 Fig), washing out A22 is expected to restart new PG synthesis and also incorporation of any residual DAADs remaining within cells/inclusions. However, when we conducted the second chase in the presence of very high concentrations of the native unlabeled dipeptide DA-DA (second DA-DA chase), we observed a slight drop in PG signal recovery compared to when no native DAAD was added (Fig 8b and 8c) which may indicate that the extra DA-DA is competing with the unincorporated probe during the recovery. In contrast, when the second chase was done in the presence of DCS, an inhibitor of DA-DA synthesis, we observed an increase in PG signal recovery (Fig 8b and 8c). This was expected, as depleting the cellular pool of the competing, native DA-DA would increase the frequency of DAAD incorporation. When the second chase was conducted in the presence of ampicillin, we observed only a slight decrease in signal recovery compared to the second cold chase (Fig 8b and 8c), indicating that probe PG incorporation (S9a and S9b Fig) and degradation are independent of D,D-transpeptidation/D,D-carboxypeptidation activity, which are both inhibited by β-lactam antibiotics. The competition of the native DA-DA with DAADs is also evident in the decrease in EDA-DA labeling when our initial EDA-DA pulse is followed by a native DA-DA chase (S11a Fig). We observed an even greater reduction of alkyne containing EDA-DA labeling when we chase with the alternative azide containing DAAD, ADA-DA (S11b Fig), however, this reduction in fluorescence is most likely the result of these two PG incorporated DAADs within close proximity forming stable conjugates with each other via azide-alkyne [3+2] cycloaddition (and not their corresponding fluorophores) upon the initiation of the click chemistry reaction. Nevertheless, the retention of components of immunogenic PG (DAADs), the absence of PG biosynthesis / retention in the extracellular form of Chlamydia, and the limiting of PG to a mid-cell ring often localizing to the division septum of its replicative form, all indicate that Chlamydia restricts its PG to where and when it is absolutely needed. In the context of an obligate, intracellular pathogen that has co-evolved with vertebrate hosts for hundreds of millions of years, this severe limitation on PG biosynthesis and maintenance is indicative of pathoadaptation by Chlamydia.

Discussion
To better understand the role of PG in the life cycle of obligate, intracellular pathogenic Chlamydia, we characterized the assembly and maintenance of this critical cell wall component. Our observations show that four species of pathogenic Chlamydia maintain a distinctive PG ring limited to a small region at mid-cell and localizing to the septum in bacteria actively undergoing division. We refer to this structure as a unique and limited PG ring based on four major observations: 1. The chlamydial cell maintains a narrow, dynamic PG ring structure restricted to the apparent division plane of the bacterium.
2. PG is found only in the actively growing replicative (RB) form of Chlamydia.
3. This limited PG is a common and unique feature of the pathogenic Chlamydiaceae family.

Assembly and maintenance of this limited PG ring is controlled by MreB in conjunction
with PBPs and a yet-unknown PG turnover mechanism.
Based on current detection techniques and data from previous publications, we conclude that these PG rings constitute the only PG cell wall structure in actively growing pathogenic Chlamydia. Regardless of the duration of labeling with DAADs, rings of relatively constant width represent the sole fluorescence trace of PG observed in the replicative form of pathogenic Chlamydia, as opposed to the uniform, peripheral labeling present in other bacteria with classical, cell-encompassing PG sacculi following prolonged labeling [9]. While our experimental approach limits labeling of PG to the fourth amino acid of the stem peptide, we surmise that position 4 is very stable, as no L,D transpeptidases, which would cleave the stem peptide at this position, appear to be encoded within the chlamydial genome [23]. Our observations also match immunofluorescence data obtained by researchers utilizing antisera raised primarily against mycobacterial PG, which gave structures very similar to the DAAD-labeled PG rings present exclusively in RBs [32]. Pathogenic Chlamydia lack a peripheral peptidoglycan layer when imaged via electron microscopy [13], and sacculi have never been successfully purified from chlamydial cells [16,19]. The existence of a thin, lysozyme sensitive ( [9] and this work), MurNAc-containing [50] PG ring present only in RBs would account for the findings of past researchers, who each reported finding only trace evidence, if any, of PG in Chlamydia [14,15,18,20,32]. Additionally, the absence of PG labeling in EBs is consistent with the lack of NOD1 and NOD2 stimulatory PG fragments in lysates of chlamydial EBs, with only trace amounts of the PG-specific sugar, MurNAc, detectable in infected cell lysates after the EB-RB transition [50]. Our limited PG model is also supported by evolutionary analysis. We observed narrow PG ring structures in four different pathogenic chlamydial species that are evolutionarily representative of the Chlamydia genus. In contrast, Pilhofer et al. reported that Parachlamydia endosymbionts (environmental Chlamydia-like bacteria represented by P. amoebophila) possess a conventional, structurally supportive, shape-determining sacculus [10]. Simkania negevensis, a pathogenic Chlamydia-like bacterium associated with community-acquired pneumonia in humans [6,53], does not appear to possess a PG sacculus. This may hint at a speciation event of ancestral Chlamydiae: the pathoadaptation of ancestral Chlamydia (possibly also Simkania) to their human/animal hosts may have resulted in spatially restricting PG to a bare minimum, a small ring, and temporally limiting the period of PG synthesis to actively replicating, intracellular RBs. Future studies examining the PG of other environmental and pathogenic Chlamydia species will shed light on the prevalence of this peculiar PG architecture throughout the Chlamydiae.
We propose that pathogenic Chlamydia do not require a sacculus due to the osmotically stable environment in which they reside. It is noteworthy that E. coli, upon treatment with penicillin-based antibiotics under osmotically stable conditions does not lyse, and instead continues to expand, unable to divide (similar to ampicillin-induced, chlamydial aberrant bodies). The genomes of some obligate extracellular (Mycoplasma) and intracellular (Ehrlichia, anaplasma) pathogens lack all or the vast majority of PG synthesis genes, indicating that PG (and by extension, a sacculus) is dispensable for the survival of obligate extracellular / intracellular pathogens. While this line of reasoning does not account for the presence of sacculi in some obligate intracellular pathogens such as Coxiella, these exceptions can often be explained when their various environments and lifecycles are considered. Rickettsiae, for example, are highly pleiomorphic organisms and generally have broad host ranges that include arthropods, potentially subjecting them to less osmotically stable conditions than those found within vertebrate hosts. Coxiella in particular is among the most resistant of the Rickettsiae to adverse environmental conditions (i.e. osmotic stress) indicating that one major function of a PG sacculus (osmotic protection) potentially remains essential to the organism.
Assuming that Chlamydia has effectively dispensed with the need to withstand osmotic stresses by adapting to an obligate, intracellular niche, we speculate that PG would still be maintained by the organism i) if it was essential for cell division, and/or ii) if it acts as a signaling molecule to other microbes or to the host immune system, to the direct benefit of Chlamydia. Limiting, masking, or removing PG (and other immunostimulatory PAMPS) is likely an intrinsic adaptation of ancient, obligate intracellular pathogens. Such a reduction of its immunogenic profile is beneficial for a pathogen so long as this does not compromise core functions essential for its survival. As a logical extension of this model, the complete removal of PG from the organism would be ideal. However, while other pathogens (such as Mycoplasma) have developed cell division mechanisms that function in the absence of PG, Chlamydia has not. We propose that by limiting PG to the septum during its replicative growth phase, Chlamydia maintains a minimal PG synthesis activity essential for division while also minimizing its recognition by host innate immune receptors. PG fragments are potent signaling ligands for both NOD1 and NOD2 receptors of the innate immune system [54], and limiting their abundance may significantly mitigate Chlamydia's immunogenic profile. This hypothesis is supported by our finding that DAAD probes, unnatural substitutes for the PG D-Ala-D-Ala residue [9], appear to be retained within RBs and distributed between the daughter RBs upon division. The recent discovery [50] that pathogenic Chlamydia incorporates glycine into the first position of the PG stem peptide further supports pathoadaptation by the microbe. This specific modification of the peptide chain has only been observed in the intracellular pathogen Mycobacterium leprae [55] and studies have shown that alterations at this amino acid position can decrease immunoadjuvant capacity [56] as well as significantly affect MDP recognition by the NOD2 receptor [57].
Our observations indicate that chlamydial PG rings exclusively localize to the apparent division plane of RBs. Upon the onset of division, the ring contracts with the outer membrane and immediately upon division PG appears to localize in polar patches. PG is only ever discernable in these polar patches or complete rings, which suggests that the new chlamydial division plane forms immediately after the previous division. These polar PG patches may act as cues for priming the new division plane, which would force it to form perpendicular to the previous one (Fig 9a), similar to what occurs in other cocci [33]. Complete PG rings are also discernable prior to any apparent sign of constriction (Fig 4 and S3 Fig). This observation may indicate that the chlamydial PG ring either must first mature before driving constriction or it may confer some additional utility to the organism, such as structural support or acting as a scaffold for the division and membrane constriction machinery. Additionally, while cell division following ring constriction most often occurs symmetrically, the occurrence of bacteria with asymmetric division planes may serve to normalize cell size or to initiate the RB-EB transition.
The formation of incomplete PG rings upon short DAAD pulses, the patchy, partial colocalization of MreB on newly forming PG patches, the requirement for a functional MreB in order to synthesize and assemble PG rings, and the patchy and dynamic nature of MreB in other bacteria [45][46][47] all support a model in which new PG rings expand from polar patches with the aid of a mobile MreB upon completion of a cell division event (Fig 9a). The tendency of newly incorporated PG signal trailing the MreB patches might be due to the incomplete saturation of DA-DA pools with our modified dipeptides. Alternatively, there might also be a threshold of DAAD-labeled PG subunits that need to be built into the edge(s) of growing PG arcs by the pioneering MreB before it can be detected by our imaging conditions. Either would prevent us from unambiguously visualizing complete colocalization of MreB patches with short pulse PG patches. Therefore we suspect that these process may occur too rapidly for us to accurately visualize with technology currently available. We speculate that the MreB-facilitated PG synthesis occurs in a directional manner around the septal ring, constantly and dynamically threading the new PG into the older, ring-shaped material. While MreB co-localization with PG strands in enlarged aberrant bodies upon β-lactam inhibition of PBPs is consistent with this model, this observation also shows that transglycosylation in Chlamydia can be uncoupled from D,D-transpeptidation/carboxypeptidation. D,D-transpeptidation/carboxypeptidation may still play a role in defining the proper ring shape for division, but not for nascent PG assembly. We believe that future studies utilizing fluorescent protein fusions of relevant chlamydial proteins, such as MreB, coupled with live-cell optimized PG labeling DAADs will allow for these models to be tested in real-time via time-lapse microscopy.
We showed that inhibition of MreB arrests the EB to RB transition early in infection, the growth of RBs and the maturation of chlamydial inclusions at mid-infection, and the RB to EB transition late in infection. All of these phenotypes are reversible when the MreB inhibitor is removed. Therefore, we conclude that MreB is required not just for chlamydial replication, but also for cell differentiation and growth. There is a distinctive association between the spatiotemporal distribution of MreB, PG synthesis, and chlamydial cell growth. MreB and PG are both absent from chlamydial EBs and aberrant bodies induced by MreB inhibition. PG biosynthesis and cell division genes are both up-regulated upon initiation of the EB to RB transition [58] and we have shown that MreB localization and PG biosynthesis occur only after the transition. MreB and PG are both present and co-localize in normally growing RBs as well as in β-lactam induced aberrant bodies, which unlike aberrant bodies that form due to MreB inhibition, continue to grow at the same rate as healthy cells [59] while exhibiting long strands of nascent PG and patchy MreB (S9 Fig). Similar cell enlargement occurs in DCS-induced aberrant bodies that appear to be able to assemble an uncross-linkable, yet still polymerized nascent PG [32]. DCS treatment results in the accumulation of degraded muropeptides and/or PG precursors in Chlamydia and incomplete muropeptides have been shown to incorporate into the PG of E. coli, following DCS treatment [60]. Our observations, as well as those of others [32] indicate that PG glycan chains in ampicillin and DCS-induced aberrant bodies still form, giving weight to the hypothesis that nascent PG biosynthesis is independent of cross-linked PG. In the case of DCStreated cells, incomplete muropeptides can still transit to the periplasm and incorporate into newly forming PG glycan chains, but cannot form proper crosslinks [60]. This is in contrast to inhibition of MreB function in chlamydial RBs, which leads to arrest of cell growth and absence of PG synthesis. RBs lacking only D,D-transpeptidation/carboxypeptidation activity (inhibited either directly by β-lactams or indirectly by DCS), continue to enlarge, synthesize and assemble nascent PG, but appear to lack the ability to regulate their size, likely due to poor coordination of proper cell division and the absence of cross-linked PG. This also indicates that chlamydial PG may have a cellular organizational role other than simply facilitating cell division.
While our initial observations indicated chlamydial PG was comprised of relatively fixed and static rings [9], this PG ring has proven to be highly and uniquely dynamic, as evidenced by its rapid assembly and, in the absence of functional MreB, disassembly. Other bacteria that grow from the mid-cell synthesize new PG at the septum, splitting the older septal PG and pushing it outwards toward the poles where it is eventually modified and/or degraded. Given that Chlamydia maintain PG as a 140 nm ring, we are currently unable to test this bidirectional model of PG reshaping in Chlamydia (S10a Fig), as the width of the PG rings is currently at the resolution limits of SIM imaging systems. Assuming that A22 dispersion of MreB polymers does not significantly affect the activity of the PG degradation mechanism, we favor a model in which, as the cell grows and divides, the PG ring in pathogenic Chlamydia is constantly re-sculpted by coordinated, yet independent PG synthesis and degradation mechanisms (Fig 9b).
Our model for the maintenance of chlamydial PG rings predicts the presence of an equally dynamic and closely localized PG degradation mechanism accompanying MreB-facilitated PG synthesis (Fig 9b). This degradation mechanism would likely include a yet uncharacterized lytic transglycosylase capable of cleaving the PG glycan chains and the chlamydial amidase, AmiA [61,62], which cleaves the peptide stem from the N-acetylmuramic acid. Interestingly, the chlamydial amiA is expressed simultaneously with genes of the PG biosynthetic pathway [58] and the protein localizes diffusely in the periplasm [51]. The amidase encoded by C. pneumoniae possesses carboxypeptidase activity and lacks the regulatory domain present in the homologous E. coli protein [61]. A diffuse and constitutively active chlamydial AmiA possessing both amidase and carboxypeptidase activities could explain how a PG ring of constant width is maintained in normal RBs and how aberrant PG forms in β-lactam-induced aberrant bodies. When MreB is not functional, a diffuse and active AmiA would completely degrade older PG with no new PG to take its place. This could also explain the slight enlargement of RBs in the presence of A-22; in the absence of a strong PG belt, the RBs might expand slightly to reach the hydrostatic equilibrium within the inclusions. This MreB vs. PG degradation model also links MreB localization and movement to how, where, and when chlamydial PG will be made; directly by the incorporation of new PG material, and/or indirectly by down-regulating the activity of the degradation mechanism. While MreB co-localization with PG strands in enlarged aberrant bodies upon β-lactam inhibition of PBPs is consistent with this model, this observation also shows that transglycosylation in Chlamydia can be uncoupled from D,Dtranspeptidation/carboxypeptidation. D,D-transpeptidation/carboxypeptidation may still play a role in defining the proper ring shape for division, but not for nascent PG assembly.
In conclusion, pathogenic Chlamydiae lack a classical PG sacculus and limit PG synthesis to a narrow PG ring during the replicative phase of their developmental cycle. We propose that the expression of minimal quantities of immunostimulatory PG represents pathoadaptive evolution by Chlamydia. Immune evasion by the restriction or modification of PAMPs is well-documented in other pathogenic microbes [63]. The reduced PG ring present within pathogenic Chlamydia species is likely the product of their adaptation to an obligate, intracellular lifestyle in which a full PG sacculus is not required and may even be detrimental to bacterial survival. We propose that this PG ring is made and re-sculpted to facilitate cell division and growth by the interplay of MreB with a PG degradation mechanism. This study also established a direct visual link between nascent bacterial PG synthesis and a functional and dynamic MreB. With the availability of inducible promoters [64,65], gene inactivation systems [66] and allelic exchange [67] in Chlamydia, DAADs will allow future studies to explore the models proposed in this work by controlling various aspects of PG biosynthesis and degradation in intracellular, pathogenic Chlamydia. Our observation that inhibition of MreB has global inhibitory effects throughout the chlamydial life cycle not only makes Chlamydia an attractive organism in which to study MreB, but also presents this protein as a promising anti-chlamydial target for future studies.

Reagents
Clickable dipeptide PG probes (EDA-DA and ADA-DA) and MreB polymerization inhibitors (A22 and MP265) were synthesized as previously described [9,49]. Alkynyl-and Azide-functionalized Alexa Fluor 488, TAMRA-5-azide, and Click-iT Cell Reaction Buffer Kit were purchased from Invitrogen. Antibodies against chlamydial MreB and IncA were generously provided by Scott Hefty (University of Kansas) and Dan Rocky (Oregon State University), respectively. The chlamydial RSGFP-expressing p2TK-SW2 plasmid was generously provided by Isabelle Derré (University of Virginia).
Bacterial strains, growth conditions, and quantification of inclusionforming units (IFUs) C. trachomatis serovar L2 strain 434/Bu was provided by H. Caldwell (Rocky Mountain Laboratories). C. muridarum strains Nigg "M9" [68], C. psittaci strain 6BC "BCRB" [69], C. caviae strain GPIC "SP6" [70] are all clonal lab strains picked and expanded from single plaques. Chlamydial stocks were generated and asynchronous infections were performed as previously described [9]. Briefly, chlamydial EBs were harvested from infected L2 (mouse fibroblast) cells at 40 hpi and stored at -80°C until use. For infections, tissue culture-treated glass coverslips were placed in 24 well plates (Costar) and L2 cells were plated so as to reach~70-80% confluence by the day infections were carried out. Cells were washed twice with warm Dulbecco's modified Eagle's medium (DMEM), infected at a multiplicity of infection (MOI) of 1 with bacteria resuspended in DMEM, plates were rocked for two hours at 37°C 5% CO 2 , unbound bacteria were subsequently removed, and medium was replaced with DMEM supplemented with 10% FBS (HyClone), 1 × MEM Non-Essential Amino Acids Solution (Sigma), and 0.2 μg ml −1 cycloheximide (Sigma). Medium was supplemented with native/modified dipeptide molecules, antibiotics, and/or MreB inhibitors as noted in the text. For quantification of the toxicity of MreB-polymerization inhibitors, infections were carried out as described above utilizing the C. trachomatis serovar L2 strain 434/Bu transformed with the p2TK-SW2 plasmid [71] for the expression of RSGFP. Inhibitors were added at the time points / durations indicated in the text. At either 44 or 52 hpi, 1 ml of sucrose/phosphate glutamate buffer (SPG) was added to each well and cells were collected via scraping with glass beads. Resuspended cells were then subjected to sonication for brief (ten second) pulses, 1:10 dilutions were prepared and were immediately used to infect 96 well monolayers of L2 cells, which were seeded the previous day with 200 μl of 200,000 L2 cells/ml per well. Plates were then spun in an Eppendorf desktop centrifuge at 3000 rpm 35C for 1 hour and then allowed to incubate at 37°C 5% CO 2 for 24 hours. Inclusions were counted in live cells via fluorescence microscopy and IFUs were calculated for each experimental group in biological triplicates.

Labeling of chlamydial PG
Experiments in which chlamydial PG was labeled with incorporated, clickable dipeptide probes (DAADs) were carried out as described previously [9]. Briefly, at designated time points post infection, treated coverslips containing Chlamydia-infected cells were washed first in warm DMEM, then with PBS, and fixed/ permeabilized with methanol for five minutes. Cells were then washed with PBS, further permeabilized with 0.5% Triton X for five minutes, and washed a final time with PBS. Coverslips were then blocked with 3% BSA (in PBS) and the click chemistry reaction was carried out utilizing the Click-iT Cell Reaction Buffer Kit (Invitrogen) with appropriate azide-modified fluorophores (Alexa fluor 488/647 and TAMRA-5). The addition of copper (cupric sulfate) results in the alkyne group on the dipeptide probe and the azide group on the fluorophore forming a stable triazole conjugate, thereby fluorescently labeling the PG in which the probe has been incorporated. Where indicated, chlamydial major outer membrane protein (MOMP) was labeled with anti-MOMP antibody (LifeSpan Biosciences, 1:500), chlamydial inclusions were labeled with anti-IncA antibody ( [72], 1:500) and chlamydial MreB was labeled with anti-chlamydial MreB diluted 1:1,000. Alexa fluor-conjugated anti-mouse, anti-goat, and anti-rabbit secondary antibodies (Invitrogen) were diluted 1:2,000.
Pulse chase experiments L2 cells were infected with C. trachomatis serovar L2 strain 434/Bu for 16 hours. Infection medium was then removed, cells were washed once with pre-warmed DMEM, and new infection medium was added containing 4 mM EDA-DA (pulse). After one hour, medium was removed, cells were washed twice with pre-warmed DMEM, and new medium was added containing a second probe, small molecule inhibitor, or neither (chase). Cells were fixed, permeabilized at indicated time points, and labeling of chlamydial PG and MOMP was conducted, as described previously. For the pulse-chase experiment conducted in Fig 8, a similar principle was followed, but with the following differences. After a 2 h EDA-DA (4 mM) pulse, cells were washed and incubated in fresh medium containing MP265 (125 μM) for an additional 1 h (1 st Chase). At this point the cells were washed again and new medium (that lacked both EDA-DA or MP265) was added containing either antibiotics, D-Ala-D-Ala, or neither (2 nd Chase).

Imaging and analysis
All confocal imaging was conducted with a Zeiss 710 laser scanning microscope utilizing Zen 2012 (Carl Zeiss) software and all epifluorescence imaging was conducted by a Nikon Ti-E inverted fluorescence microscope equipped with a Plan Apo 60x/1.40 Oil Ph3 DM objective and a DAPI/GFP/Cy3/Cy5 filter cube and an Andor DU885 EMCCD camera. Settings were fixed at the beginning of image acquisition and for experiments in which different samples/ time points were to be compared, the same parameters were applied for collecting and post processing all images taken. Deconvolution and maximum intensity projection (when used) was conducted utilizing AxioVision (Carl Zeiss) software employing the inverse filter setting or ImageJ, respectively. ImageJ was used for all subsequent image analysis. For generating the data presented for the kinetic analysis of the average fluorescence of chlamydial inclusions, confocal Z-stacks were taken using the 40x objective and maximum intensity projections were generated from those stacks and pixels assigned a brightness level between 0 and 255. Basic intensity quantification was conducted by using the MOMP-labeling (red) channel to define the area which encompassed individual inclusions, and this was then used as an overlay to subsequently quantify the level of fluorescence present within each inclusion area present within the EDA-DA (green) channel.
3D SIM super-resolution microscopy was performed on a Delta Vision OMX microscope equipped with an Olympus 100X/1.40 Oil PSF objective and a Photometrics Cascade II EMCCD camera. The samples were excited with lasers at 405 nm, 488 nm, 561 nm, 642 nm and the emission was detected through 419 nm -465 nm, 500 nm -550 nm, 609 nm -654 nm, 665 nm -705 nm emission filters. The image processing was conducted by SoftWorx imaging software. Further image analysis was conducted via ImageJ including the measurements of PG ring widths. For PG width measurements, rings that were aligned perpendicular to the x,y axis (for the maximum resolution) were used. SIM imaging Z stacks of chlamydial inclusions were obtained via a Zeiss ELYRA PS.1 utilizing Zen 2012 (Carl Zeiss) software for image processing.

Chlamydial muropeptide isolation
Muropeptide fragments of PG were isolated from Chlamydia-infected cells as previously described [50]. Two 175 cm 2 flasks of confluent HeLa cells were either infected with C. trachomatis L2/434 at an MOI of 1 or mock infected and incubated for two hours at 37°C in 5% CO 2 with rocking. Infection medium was then removed and replaced with DMEM with heat inactivated fetal bovine serum (10%), and cells incubated for an additional 16 hours. Medium was then removed and replaced with fresh DMEM (10% FBS) with or without 75 μM compound A22 and cells were then allowed to incubate an additional two hours. Medium was removed, cells were washed once with warm DMEM (to remove any residual A22) and then cells were harvested using glass beads, resuspended in DMEM, and sonicated at 40 amps with one-minute pulses, repeated five times. Supernatants were centrifuged at 4,000 g for five minutes to remove cellular debris, lysates were passed through a 0.22 μm filter, and then 3 kDa centrifugal filters (Amicon UFC900324) at 4000 x g for one hour at 37°C. Flow-through fractions from the 3 kDa centrifugal filters were then heat inactivated at 95°C for six minutes and assayed for activity in an HEK NOD reporter cell line (see below).

NOD1 / NOD2 NF-κB reporter assay
The NOD signaling assay was conducted on lysates of Chlamydia-infected cells as previously described. Briefly, HEK cells overexpressing either the NOD1 or NOD2 receptors, as well as the NF-κB-SEAP reporter gene (Invivogen, CA), were used to quantify the immunostimulatory potential from cell lysate fractions taken from a reverse phase C18 HPLC column. Twenty μl of cell lysate fractions from mock-infected and Chlamydia-infected cells (in the presence or absence of compound A22) were added to~5 x 10 4 HEK-Blue NOD1 or NOD2 cells in 96 well plates (total reaction volume 200 μl/well) and incubated for 24 hrs at 37°C. Secreted alkaline phosphatase (SEAP) was measured by adding 20 μl of the supernatant from lysate fractiontreated wells to 180 μl of QUANTI-blue substrate (Invivogen) in a separate 96 well microtiter plate. Untreated or uninfected cell supernatants were used as negative controls. The reaction was incubated at 37°C for 30 min and SEAP activity was assessed by reading OD at 650 nm.

PG analysis
HPLC was carried out on filtered, Chlamydia-infected cell lysates (and controls) as previously described [50]. LCMS experiments were performed on an Agilent 1200 Series liquid chromatography system coupled to an AB Sciex Q-Trap 4000 mass spectrometer with a Turbo V electrospray ionization source, as previously described [50]. Data from control/experimental groups was analyzed and overlaid using Analyst Software (v1.5.1) with all experiments and subsequent analysis conducted at least twice. (a) Schematic representation of the homogenous vs. bidirectional removal of the older PG. In contrast to the homogeneous degradation of PG about the entire ring, the bidirectional model predicts splitting of the older ring (green) into two with new PG (pink) being inserted in the middle. (b-c) Inclusion area (as measured by MOMP labeling) graphed for all groups and PG (black) / MOMP (red) labeling re-plotted against inclusion size. Inclusion size was measured to ensure that changes in fluorescence intensity were not simply attributable to inclusion growth over time. The total inclusion size trended upward three hours after the initial DAAD pulse (two hours after the beginning of the chase portion of the experiment (b)), however, this did not appear to affect average fluorescence intensities within any given experimental group (c). The same pulse chase experiment presented in panel a, but carried out over one hour and comparing the effects of exogenously added ADA-DA on loss of EDA-DA fluorescence. The addition of ADA-DA results in a drastic decrease in EDA-DA signal, most likely because incorporated azide containing ADA-DA outcompetes the Alexa Fluor 488 azide during the click chemistry reaction, i.e. most of the EDA-DA labeled PG is captured by ADA-DA labeled PG instead of the Alexa Fluor 488 that is used for the read-out of the EDA-DA labeled PG. For both experiments, average fluorescence values were calculated for each treatment group at the indicated time points, subsequent to chases. Each bar represents the average fluorescent pixel intensities of~150 chlamydial inclusions pooled from two independent experiments. Error bars represent the standard deviation of the mean for each sample group. Chlamydia PG and MOMP were labeled as described in Fig 2. (TIF) S1 Video. SIM Z-stacks (presented as single imaging planes in S2 Fig) of Chlamydia inclusions 18 hpi taken via Zeiss ELYRA PS.1. PG is labeled in green and MOMP in red. Movies begin with 3D-rendered projections of each Z stack rotated about the X or Y axis followed immediately by the presentation of each individual imaging plane successively through the Z stack of the inclusion. Single channels are presented first and followed by a merged compilation of the two. S4 Video. SIM Z-stacks of Chlamydia inclusions 18 hpi taken via Zeiss ELYRA PS.1. PG is labeled in green and MOMP in red. Movies begin with 3D-rendered projections of each Z stack rotated about the X or Y axis followed immediately by the presentation of each individual imaging plane successively through the Z stack of the inclusion. Single channels are presented first and followed by a merged compilation of the two. (WMV)