Chlamydia trachomatis serovar L2 strain 434/Bu was provided by Dr. Anthony Maurelli (University of Florida). Chlamydial infections were carried out in HeLa cells or L2 cells (also provided by Dr. Anthony Maurelli) unless otherwise noted. The L2 mouse fibroblast cells and HeLa cell lines were passaged in high glucose Dulbecco’s modified Eagle medium (DMEM, Gibco) and 10% fetal bovine serum (FBS, HyClone). HEK-Blue-hNOD1 and -Null1 cells were purchased from InvivoGen and propagated according to the manufacturer’s instructions.

HeLa cells were plated on glass coverslips in 24 well tissue culture treated plate (Costar) at a confluency of ~70–80%. Cells were infected with C. trachomatis L2 434/Bu in cold Dulbecco’s modified Eagle medium (DMEM; Gibco; 250μl per well) and placed on a rocker in the 37°C tissue culture incubator for 2h. Subsequently, the DMEM was removed and replaced with DMEM supplemented with 10% FBS (HyClone) and 1x MEM Non-Essential Amino Acids Solution (Sigma; 250μl per well). The following aberrance inducing conditions were tested in this study and added at either the time of infection or at 22hpi: 10μM piperacillin, 10μM mecillinam, 10μM penicillin G, 125μM MP265 (polymerization inhibitor of the bacterial cytoskeletal protein MreB), and 3μM chloramphenicol.

HeLa cells were infected with C. trachomatis L2 434/Bu at an MOI of 1 unless otherwise stated. Cells and replicating intracellular Chlamydia were grown in the presence of the PG intercalating reagent ethynyl-D-alanyl-D-alanine (EDA-DA) at a concentration of 1mM. At 24 hpi, cells were washed 3 times with 1X PBS, permeabilized with methanol and 0.5% Triton X (5min each) prior to blocking with 3% Bovine Serum Albumin (BSA) for 1h and antibody labeling. Click chemistry was carried out to link the PG-associated EDA-DA with conjugated fluorophore Azide-Alexa Fluor 488, as previously described (Liechti et al., 2014, 2016). The chlamydial Major Outer Membrane Protein (MOMP) was labeled with anti-MOMP primary antibody (goat) and secondary antibody (donkey anti-goat Alexa Fluor 594) at the dilutions 1:500 and 1:1,000, respectively. All coverslips were mounted on slides with ProLong Gold Antifade Mounting Media and stored in the dark at 4°C prior to imaging.

All imaging was conducted with a Zeiss ELYRA PS.1 in Structured Illumination (SIM) mode. Settings were fixed at the beginning of image acquisition, Z-stacks were collected for all images taken, and the same parameters were applied for collecting and post-processing all images taken within each experiment. All SIM processing was conducted in RAW fluorescent units (as opposed to averages or percentile standardizations), thus maintaining readouts well below maximum threshold and eliminating the potential for post-processing steps to approach pixel saturation. All post-image processing was conducted on Zen 2012 (Carl Zeiss) software. ImageJ was used for all subsequent image analysis utilizing the FIJI addon “3D Objects Counter” to identify the 3-dimensional spatial constraints of all PG-containing objects within individual C. trachomatis inclusions. Relative object volumes were estimated based off of pixel fluorescence data present throughout a Z-stack. Pixels were assigned a “1” or a “0” value based on a common threshold setting applied to all images examined. A composite of all connected “1” pixels through x, y, and z planes was thus constructed from that imaging data. Each PG-labeled object was assigned an estimated volume (in μm³), fluorescence mean intensity (average pixel intensity), and integrated density (the sum of all pixel intensity measurements comprising an object). Object size thresholding was restricted to objects larger than 0.001 μm³, the detection limit of this approach, corresponding to a single pixel. Statistical significance values were calculated utilizing a one-way ANOVA coupled with a multiple comparisons test with each condition compared with the untreated control group. Inclusion diameters were calculated using the average of the maximum diameter visible from all corresponding z planes and the diameter of a line perpendicular to this maximum diameter.

HEK-Blue cells expressing the hNOD1 receptor and carrying the NF-kB SEAP (secreted embryonic alkaline phosphatase) reporter gene (InvivoGen) were used according to the manufacturer’s instructions and adapted to assess NOD1-specific NF-κB activity during infection with C. trachomatis, as previously described (Brockett and Liechti, 2021). Briefly, 3 × 10⁵ cells/mL of HEK-Blue hNOD1 or Null1 cells were plated in 96-well plates (total reaction volume of 200μl per well, ~6.0 × 10⁴ cells per well) and allowed to adhere for 6h at 37°C. These cells were subsequently infected with C. trachomatis L2 434/Bu at a MOI of ~1. Plates were centrifuged for 1h at 2,000g to enhance the synchronization of infections and then incubated in a CO₂ incubator at 37°C. Supernatants were collected at the designated time points for subsequent analysis of SEAP activity. A colorimetric reporter assay was then utilized in order to quantify the abundance of SEAP in cell supernatants. Twenty microliter of supernatants collected from infected cells were added to 180μl of the SEAP-detection solution (InvivoGen) followed by incubation at 37°C for ~8h. SEAP enzymatic activity was then quantified using a plate reader set to 650nm. Infected cells were compared to uninfected controls, as well as infected cells that had been treated with various PG-targeting antibiotics. Uninfected cells treated with the known NOD1 signaling ligand Tri-DAP (1μg/ml) were used as a positive control. All experiments were carried out in parallel in the HEK-Blue Null1 cell line, which contains the empty expression vector but lacks hNOD1. HEK-Blue NOD1 SEAP and Null1 reporter assays were carried out independently in biological triplicate, statistical analysis was conducted by 2-way ANOVA, and significance values were analyzed utilizing Sidak’s multiple comparisons test. Values plotted are means of the raw OD650 measurements.

To determine the extent to which PG labeling can be utilized to investigate the stages of the Chlamydia division process, we carried out a mapping study of all PG-containing objects within 100 chlamydial inclusions. Infected cells were incubated with the PG intercalating agent ethynyl-D-alanyl-D-alanine (EDA-DA; Liechti et al., 2014) and allowed to grow for 24h, at which time the cells were fixed, permeabilized, incubated with cupric sulfate and an azide-conjugated fluorophore in a click chemistry reaction. Cells were subsequently immunolabeled with a monoclonal antibody specific for the pathogen’s Major Outer Membrane Protein (MOMP). Labeled cells were then imaged by structured illumination microscopy (SIM), Z-stacks of infected cell monolayers were collected, all EDA-DA labeled objects within each inclusion were counted and assigned corresponding spatial dimensions and labeling intensities (Figures 1b–d).

We have reported previously that septal (“ring”) PG in Chlamydia species is not uniform and can differ in diameter between individual bacterial RBs (Liechti et al., 2016). We have also noted that peptidoglycan in C. trachomatis is transient and can dissociate rapidly when the cell division apparatus is inhibited (Liechti et al., 2016) or when “persistence” is induced utilizing various stressors (Brockett and Liechti, 2021). PG labeling intensity is a function of multiple cellular processes: (i) the uptake of the labeling reagent, (ii) the relative rates of its incorporation during peptidoglycan biosynthesis and dissociation during peptidoglycan degradation, and (iii) the relative density of the PG-containing structure. Having observed differences in PG labeling intensities when PG biosynthesis was directly interrupted, we questioned whether PG rings differed significantly in their labeling intensities throughout chlamydial development. We reasoned that if differences exist in either (1) the rates of PG synthesis and PG degradation or (2) the density of PG produced between bacteria at different stages of development, then we should be able to observe corresponding differences in PG labeling intensity. To investigate this question, we plotted PG-containing object volume against object fluorescence integrated density for all labeled objects within 15 chlamydial inclusions (Figure 1e). The overall data fit well within the linear range (R²=0.969), and we found that as object volume increased, so too did the variance in mean fluorescent intensity labeling, as judged by the distance of each data point relative to the calculated linear regression line for the data set. As integrated density is a function of a labeled object’s volume and mean fluorescent intensity, we reasoned that these differences in the labeling intensities of larger objects could signify differences in either object shape, the density of the labeled PG within objects, or the kinetics of PG assembly and degradation occurring in these structures.

To assess whether the divergence in PG labeling observed in the pooled inclusion data was representative of RB populations within individual inclusions, we repeated our examination for each inclusion from the previous analysis and plotted PG-labeled object volume and intensity measurements for every labeled object within each inclusion. We found similar levels of variance in object mean fluorescent intensity in all inclusions, and examining inclusions containing fewer bacteria allowed us to focus on individual PG-containing objects and compare their relative shapes and staining intensities. For both smaller (<5μm diameter, Figure 2A) and larger (≥5μm, Figure 2B) inclusions, “brighter” objects (high mean fluorescent intensity) appeared to orient as disks or smaller, compact rings (Figures 2A,B; red/yellow), whereas “dimmer” objects (low mean fluorescent intensity) varied in ring diameter and appeared thinner across the ring width (Figures 2A,B; blue/green). Striking differences in labeling intensity between objects of similar volume were found to be the result of relative differences in the apparent thickness of otherwise smaller rings (Figure 2B, lower left panel, objects 56 and 60). Based on these three defining features (shape, volume, and labeling intensity), we found that PG orientation within intracellular chlamydial RBs fit within four broad categories: (i) disks, (ii) small diameter/large volume (≥0.06μm³) rings, (iii) large diameter /large volume (≥0.06μm³) rings, and (iv) small diameter/small volume (< 0.06μm³) rings.

We hypothesized that objects with higher-than-average labeling intensities represented PG undergoing rapid synthesis, as would be predicted during septation, the terminal step in bacterial cell division (Figure 2C). However, because it has been reported that Chlamydia PG initially localizes to the interface between parent and daughter cell during the initial polar expansion of the cell membrane (Liechti et al., 2016; Cox et al., 2020), we wanted to know whether active PG synthesis was also occurring at this earlier stage of the division process (Figure 3a). We examined PG-labeled objects in 100 labeled inclusions and found that most inclusions larger than 5μm in diameter contained between one to three examples of RBs in which PG was asymmetrically bisecting the bacterial cell, consistent with the “polar budding” phenotype (Figures 3b–d). PG-labeled objects in cells that appeared to be undergoing polar expansion had labeling intensities higher, on average, than all other objects within the inclusions (Figures 3b–d), similar to PG-labeled objects in cells undergoing septation (Figure 3e). Taken together, these data indicate that EDA-DA incorporated PG is maximally present during both the budding and septation phases of chlamydial division.

Given the differences we observed in labeling intensities for individual PG structures within RB populations, we next wanted to examine how changes to basic molecular processes affected our PG-labeled objects. We hypothesized that object size and shape were dependent on Chlamydia’s division cycle and that object labeling intensity, as measured by our assay, is affected by two distinct processes: PG synthesis and degradation. While each of these activities requires protein synthesis, we reasoned that we could delineate which was the prevailing process by incubating cells with a short, sub-inhibitory dose of the protein synthesis inhibitor chloramphenicol (Cm) and measuring the resulting changes in mean object volume and labeling intensity. Cm treatment had no discernable effect on average object volume but did result in decreasing object labeling intensity (Figures 4A–C). By comparison, similar treatment with MP265, an MreB polymerization inhibitor that blocks MreB-dependent PG synthesis in C. trachomatis (Liechti et al., 2016), results in large reductions in both PG object volume and labeling intensity (Figures 4B,C). The decrease in mean intensity during Cm treatment suggests that the balance between synthesis and degradation shifts toward turnover when protein synthesis is reduced, however, this degradation appears to be uniform across the entire PG ring as opposed to being localized to specific points on the ring. This may be a result of the timing and kinetics of the assay as previous reports investigating the effects of the inhibition of MreB polymerization on PG in C. trachomatis noted an initial reduction in PG ring labeling intensity that was subsequently followed by PG ring degradation (Liechti et al., 2016).

Previous studies investigating the effects of the ß-lactam antibiotic ampicillin on Chlamydia ring morphology have demonstrated that while EDA-DA incorporation continues to occur, PG-labeled objects exhibit an enlarged, irregular morphology (Liechti et al., 2014, 2016; Slade et al., 2019; Brockett and Liechti, 2021). This was originally thought to be due to the fact that transpeptidation is not required for the incorporation of EDA-DA, as this process occurs in the cytoplasmic space during Lipid II biosynthesis (Kuru et al., 2019). We wanted to further examine this irregular PG phenotype and establish whether it is unique to the inhibition of specific PBPs in Chlamydia. We incubated Chlamydia-infected cells for 2h in the presence of inhibitors specific for PBP2 (mecillinam; mec), PBP3 (piperacillin; pip), or both (penicillin G; penG) and compared the relative volume and intensities of PG-positive objects with an untreated control group. When PBP2 activity was inhibited alone, or in conjunction with PBP3, we saw a significant reduction in mean object volume (Figures 4A,B), however, object mean fluorescent intensity did not decrease (Figure 4C). Conversely, inhibition of PBP3 resulted in an increase in object volume (Figure 4A, bottom left panel). The effect was so robust that our subsequent imaging analysis could not distinguish a large number of individual objects, resulting in their removal from the analysis, and a subsequent “low confidence” score in our reported average volume for this treatment condition (Figure 4B). When both PBP2 and PBP3 are inhibited together, both object volume and object labeling intensity was significantly reduced. As an object’s integrated density (IntDen) is a function of both its volume and its labeling intensity, these measurements largely tracked with our mean fluorescent intensity measurements (Figures 4C,D) with the notable exception of the Mecillinam-treated group. Given the relative expansion of the PG ring when PBP3 was inhibited and the relative disassembly of the PG ring when PBP2 was inhibited, our observations suggest that PG structure in C. trachomatis is a function of these two differential PG synthases.

We next wanted to ensure that the loss of PG labeling under our treatment with ß-lactam antibiotics was the result of defects in PG assembly, rather than resulting from the complete inhibition of PG biosynthesis. Gram-negative bacteria are known to continue to synthesize lipid II and attempt to assemble PG despite a block on transpeptidation (Cho et al., 2014). This has previously been demonstrated with D-cycloserine (Packiam et al., 2015) and ampicillin (Brockett and Liechti, 2021) in C. trachomatis, utilizing a NOD reporter assay. We examined each treatment condition in a NOD1-signaling reporter cell line that reacts to PG-derived muropeptides produced by C. trachomatis that contain meso-diaminopimelic acid (mDAP; Brockett and Liechti, 2021). Similar to ampicillin, we found that C. trachomatis continues to generate mDAP-containing muropeptides in the presence of all other ß-lactam antibiotics (Figure 5), indicating that the absence of labeling in our mec and penG treatment conditions is not the result of an absence of Lipid II production. Coupled with our imaging analysis, these data suggest that chlamydial PBP2 and PBP3 activities are associated with different stages of the bacterium’s cell division process. Interestingly, we also found that only the complete interruption of PG assembly (via penicillin G) or PG biosynthesis (via D-cycloserine) significantly enhanced NOD1-ligand production and release. PG-derived muropeptide turnover was not significantly enhanced when PBP2 or PBP3 functions were selectively inhibited with mecillinam and piperacillin, respectively.

We next sought to examine PG-labeled objects in the context of the C. trachomatis developmental cycle. For this analysis, cell monolayers were infected via rocking, as opposed to centrifugation, and infection inoculates were left on the cell layers in order to achieve an asynchronous population of C. trachomatis infected cells all at different stages of the microbe’s developmental cycle. Cells were fixed, stained, imaged, and processed as before, but inclusion size measurements were also collected as an indicator of the stage of development that individual inclusions had reached by the point of fixation. Chlamydia inclusions fuse and expand throughout the pathogen’s developmental cycle and larger inclusions result from longer incubation times.

An analysis of inclusion dimensions revealed a relatively good distribution of inclusion sizes present in our study population (Figure 6A), which exhibited a slight bimodal distribution (Figure 6B). As replicating Chlamydia RBs should increase logarithmically over time, we hypothesized that PG-labeled objects should similarly exhibit a logarithmic function when plotted against inclusion diameter. We observed this exact phenomenon (Figure 6C), strengthening our confidence that inclusion diameter could be used as a readout for the relative age of the bacterial population within. In examining PG-labeled objects from individual inclusions, we determined that object volumes ≤0.1μm were of higher confidence than object volumes ≥0.1μm, due to the vast majority of the larger objects resulting from multiple, overlapping smaller objects being mistakenly reported as single large objects. We examined the presence of these larger objects throughout the range of inclusion diameters present in our data set, expecting to find more per inclusion later in development, when the inclusions contain significantly more labeled objects. Surprisingly, we found the reverse to be true, with the ratio of smaller to larger objects increasing as inclusion diameter increased (Figure 6D).

Previous reports have indicated that as Chlamydia proceeds through its developmental cycle its replicative forms (on average) decrease in size with subsequent division events (Lee et al., 2018). According to this model, a gradual reduction in RB size eventually enables C. trachomatis to drop below a critical size threshold, enabling the pathogen’s replicative form to convert to its infectious form (Figure 7A). This model has been difficult to test utilizing fluorescence microscopy, as delineating individual bacterial cell boundaries within densely packed, intracellular inclusions are challenging. However, given our success mapping PG-labeled objects associated with individual Chlamydia RBs, we sought to validate this model by conducting an analysis of the relative volume and density of PG-labeled structures throughout the course of Chlamydia’s developmental cycle.

Given the size-reduction-precedes-differentiation model’s constraints, we hypothesized that PG object volume should significantly decrease over time, as a result of Chlamydia progressing through its development, while labeling intensity should increase, as it has been proposed that C. trachomatis maximizes its replication rate by the mid-stage in its development. We plotted average object volumes, integrated densities, and mean fluorescent intensity values against the diameters of each inclusion for both sets of objects collected (< 0.1μm³ and>0.1μm³). For objects ≤0.1μm³ we found that average object mean fluorescent intensity (Figure 7B), object integrated density (Figure 7C), and object volume (Figure 7D) all decreased as inclusion diameters increased. For objects >0.1μm³, no similar decrease in average object integrated density or volume was observed as inclusion diameter increased (Figures 7C,D). Interestingly, objects >0.1μm³ did exhibit a similar trend as smaller objects in that their mean fluorescent intensities appeared to significantly decrease as inclusion diameter increased (Figure 7B). Overall, these observations appear to conform to the size-reduction-before-differentiation model in that average PG volume decreases over the span of a developmental cycle.