Quantitative analysis of morphogenesis and growth dynamics in an obligate intracellular bacterium

Obligate intracellular bacteria of the order Rickettsiales include numerous arthropod-borne human pathogens. However, our understanding of the basic biology of Rickettsia species is limited by technical challenges imposed by their obligate intracellular lifestyle. To overcome this roadblock, we developed quantitative methods to assess the cell wall composition, intracellular growth, and morphology of Rickettsia parkeri, a human pathogen in the Spotted Fever Group of the Rickettsia genus. Analysis of the cell wall composition of R. parkeri revealed unique features including a high M3 monomer fraction and absence of LD-crosslinks. Using a novel fluorescence microscopy approach, we quantified the cell morphology of R. parkeri in live host cells and found that bacterial morphology is maintained stably during exponential growth in two different epithelial cell lines. To assess population growth kinetics in a high-throughput and high-resolution manner, we developed an imaging-based growth assay and applied this to determine the growth rate of up to 24 infected cultures at a time. We also sought to gain insight into the cell cycle regulation of R. parkeri. To this end, we developed methods to quantify the fraction of the population preparing to divide as well as those undergoing active constriction. These approaches permitted a quantitative analysis of cell cycle status across a population of R. parkeri. Finally, as a proof of concept, we applied the above tools to quantitatively determine how MreB, a bacterial actin homolog, contributes to the growth and morphogenesis of R. parkeri. Inhibition of MreB with the small molecule MP265 led to cell rounding and slowed growth, suggesting that MreB is required for the growth and shape maintenance of R. parkeri. Collectively, we developed a toolkit of high-throughput, quantitative tools to understand intracellular growth and morphogenesis of R. parkeri that is translatable to other obligate intracellular bacteria. AUTHOR SUMMARY The obligate intracellular lifestyle of members of the bacterial order Rickettsiales, which includes important human pathogens, has hindered our progress in understanding their biology. Here we developed and applied high-throughput, quantitative tools to analyze essential features of rickettsial cell biology such as morphology and growth in living host cells. By applying these tools in a proof of concept, we showed that the bacterial actin homolog, MreB is required for the regulation of rod shape and intracytoplasmic growth.

growth and morphogenesis of R. parkeri that is translatable to other obligate 44 intracellular bacteria. 45

AUTHOR SUMMARY 46
The obligate intracellular lifestyle of members of the bacterial order Rickettsiales, which 47 includes important human pathogens, has hindered our progress in understanding their 48 biology. Here we developed and applied high-throughput, quantitative tools to analyze 49 essential features of rickettsial cell biology such as morphology and growth in living host 50 cells. By applying these tools in a proof of concept, we showed that the bacterial actin 51 homolog, MreB is required for the regulation of rod shape and intracytoplasmic growth. 52

INTRODUCTION 53
The order Rickettsiales includes medically relevant, emerging bacterial pathogens that 54 grow exclusively inside a host cell and may be transmitted to mammals by ticks and 55 other arthropods. Examples of important human rickettsioses are Spotted Fever,56 To gain insight into which PG enzymes may be active during intracellular PG synthesis 139 and remodeling, we characterized the R. parkeri PG cell wall chemical composition in 140 detail by UPLC. We identified 11 different muropeptides, the MS and retention times of 141 which matched the expected values for previously described DAP-type PG (Fig 1A,  142 Supplemental Table 1). Among the identified muropeptides, most of the R. parkeri PG 143 subunits appeared as uncrosslinked monomers (~60% of the relative abundance), 144 specifically disaccharide tripeptides (M3s) (Fig 1BC, Supplemental Table 1). This 145 suggests that enzymes that cleave stem peptides between the m-DAP and D-Ala 146 positions are highly active during intracellular growth. Consistent with previous 147 bioinformatic analyses that indicated the absence of LD-transpeptidases encoded by R. 148 parkeri, our UPLC analysis showed no detectable LD-crosslinked muropeptides (Fig  149   1B), implicating LD-carboxypeptidases after DD-carboxypeptidation and/or LD-150 endopeptidases in generation of M3s (14). 151 Of note, we attempted to visualize the patterning of PG insertion in R. parkeri by 152 incubating infected cells with Fluorescent D-Amino Acids (FDAAs) (35), but this proved 153 unsuccessful. FDAAs are incorporated into PG in the periplasm, through DD-or LD-154 transpeptidase activity (36). The lack of LD-transpeptidation in R. parkeri may partly 155 explain the lack of FDAA staining in R. parkeri. In addition, the high M3 abundance we 156 observed, reflecting LD-carboxypeptidase and/or LD-endopeptidase activity, would The genome of R. parkeri encodes a putative periplasmic LD-carboxypeptidase 161 homolog, LdcA (MC1_RS02825), which we hypothesize is of primary importance in the 162 cleavage of peptide stems to generate tripeptide-containing products. The biological 163 significance for the high M3 abundance in R. parkeri remains to be uncovered; however, 164 it has been previously shown that the DAP-type PG cell wall or DAP-containing PG The second and third most abundant muropeptides identified in our analysis were the 172 the anhydromuropeptides observed. The abundance of anhydromuropeptides can be 184 used to calculate the average length PG strands, which was found to be ~20 185 disaccharide units for R. parkeri. This is similar to well-studied Gram-negative 186 organisms such as E. coli, in which the average glycan chain length is 17. 8 Muropeptide composition analysis provided fundamental insights into PG synthesis and 200 remodeling of R. parkeri. We next sought to develop tools that would allow us to 201 quantitatively probe cell shape of live R. parkeri cells to permit the dissection of the 202 pathways and parameters that influence morphogenesis. Because of the obligate 203 intracellular nature of R. parkeri, and its short reported extracellular viability window 204 (~30 min), we aimed to perform in vivo analysis of morphology in live human lung 205 epithelial cell lines (A549). To do this, we leveraged a R. parkeri strain producing 206 We next asked whether the morphology of R. parkeri varies in different host epithelial 230 cells. To do this, we infected Vero cells with Rp-GFPuv and applied our epifluorescence 231 microscopy strategy to image and quantify live intracellular bacteria at 6, 12, 24, and 48 232 hpi (Supplemental Fig 1). We found that the MdnCL in Vero cells was ~1.3 µm 233 throughout infection (MdnCL 6, 12, 24 and 48hpi = 1.32, 1.25, 1.35, 1.34 µm, 234 respectively) and the ACW in Vero cells was ~0.5 µm (ACW 6, 12, 24 and 48hpi = 0.48, 235 0.47, 0.53, and 0.54 µm, respectively) (Supplemental Fig 1AB). Previous electron 236 microscopy morphological analysis of R. parkeri grown in Vero cells showed a 237 maximum cell length of ~1.5 µm and an maximum cell width of ~0.5 µm (44), in good 238 agreement with our high-throughput, fluorescence-based approach. 239 240 Because epifluorescence microscopy measurements may be affected by the three-241 dimensional diffraction pattern of light emitted, we validated our fluorescence strategy 242 by comparing it to quantification of morphology of fixed R. parkeri imaged by phase 243 contrast microscopy, a commonly used method in the field of bacterial cell biology 244 (Supplemental Fig 2A). To do this, Vero cells were infected with R. parkeri or Rp-245 GFPuv for 48 h, quickly fixed and imaged extracellularly on agarose pads, and 246 quantitatively evaluated in MicrobeJ (Supplemental Fig 2AB). We obtained similar 247 values for median cell length (MdnCL 48hpi = ~1.4 µm) and width (ACW 48hpi = 0.55 µm) 248 using phase contrast imaging of R. parkeri (Supplemental Fig 2CD) as with 249 epifluorescence (Supplemental Fig 1). Together, our fluorescence-based morphology 250 analysis provides fundamental information about the shape parameters of this enigmatic 251 group of bacteria and provides a robust, high-throughput method for quantification of 252 morphology of intracellular live or fixed R. parkeri. Using this method, we conclude that R. parkeri cell shape is maintained stably in A549 and Vero cells throughout the early 254 and mid-exponential phases of growth. The ability to reliably quantify morphology of 255 hundreds or thousands of individual bacteria will enable studies into, for example, the Having established a method for quantification of the morphology of R. parkeri, we next 261 sought to explore the mechanisms and regulation of R. parkeri morphogenesis. As a 262 first step, we focused on cell division, which is mediated by the divisome. We noted 263 from our imaging of Rp-GFPuv the appearance of actively dividing cells, which were 264 identified as those cells with a decrease in GFPuv signal near mid-cell (Fig 3A). 265 Demograph analysis of GFPuv fluorescence in cells at 24 hpi revealed that actively 266 dividing cells comprise the longest ~10% of cells in the population, as expected (Fig 3B,  267 boxed area). To understand the impact of infection stage on R. parkeri cell division, we 268 quantified the percent of constricting cells in the population at different times post-269 infection (6, 12, 24, and 48 hpi). We found that the percentage of the population actively 270 undergoing cell division decreased by ~50% as the cells entered mid-exponential phase 271 and became less variable across replicates (% constriction at 6 hpi = 16.6% in 272 comparison to 48 hpi = 8.35%, respectively) (Fig 3C). These data raise the possibility of 273 infection-stage specific input into cell cycle progression of R. parkeri.
We next sought to visualize the division machinery, itself, in R. parkeri. To do this, we 276 focused on imaging a fluorescent fusion to ZapA. ZapA binds to the master regulator of 277 division, FtsZ, at the earliest stages of assembly of the divisome and reliably reports on 278 the localization of the divisome in other bacteria (46-48). We created a R. parkeri strain 279 bearing a plasmid encoding zapA-mNeonGreen (mNG) driven by the PzapA promoter 280 (Fig 3D). Two copies of the gene encoding a cytoplasmic blue fluorescent protein 281 (TagBFP) driven by PompA were included on the same plasmid to mark cells. We found 282 that R. parkeri ZapA-mNG displayed polar or mid-cell localization in live cells (Fig 3D). In addition to quantifying morphology, we sought to develop a high-throughput, high-293 temporal-resolution growth assay for R. parkeri. Measuring population growth kinetics of 294 rickettsial species typically requires quantifying plaque-forming units (PFUs) or genome 295 equivalents over the course of an infection, each of which are labor-intensive, low-296 throughput methods. To overcome those limitations, we developed a fluorescence-297 based imaging growth assay in a temperature-and CO2-controlled imaging plate reader. We infected A549 cells with Rp-GFPuv for 24 h and imaged GFP fluorescence 299 every 3 h over ~55 h in a plate reader (Fig 4A). We observed an increase in the Rp-300 GFPuv intensity over time (Fig 4B, Supplemental Videos 1,2). Next, we plotted the 301 total GFP intensity and used these data to calculate the doubling times for each curve 302 (tD). We found that the tD of Rp-GFPuv in A549 cells was ~4:32h ± 0:28 h (Fig 4C). 303

304
We noted that it took some time to detect GFPuv above background in our assay. To 305 increase the sensitivity of this assay, we generated a R. parkeri strain producing 306 cytoplasmic AausFP1 (Rp-GFPAa), the brightest GFP reported to date (49). GFPAa has 307 a high quantum yield and can be excited with low light, enabling increased sensitivity 308 and long time-lapse fluorescence imaging with limited phototoxicity. Using the same 309 strategy as for Rp-GFPuv (Fig 4A), we infected A549s with Rp-GFPAa for 24 h and 310 imaged GFP fluorescence for ~55 h. We found that Rp-GFPAa fluorescence was 311 significantly brighter, allowing us to detect growth earlier, than Rp-GFPuv (Fig 4B,  312 Supplemental Videos 3,4) and calculated a tD of 5:06 h ± 0:14 h (Fig 4C). To expand 313 our analysis, we also determined the growth dynamics of R. parkeri in Vero cells 314 (Supplemental Fig 3). We found that the Rp-GFPAa and Rp-GFPuv grow with similar 315 kinetics in Vero cells (tD ~ 5.5 h for both) (Supplemental Fig 3B). We note that we 316 tested our approach monitoring growth of 24 independent infections in the same 317 experiment, enabling high-throughput growth analysis in future studies. 318

319
To validate our fluorescence-based growth curve approach, we quantified growth using for both Rp-GFP strains in Vero host cells. Importantly, our results showed similar tD 322 using PFU growth curves (~5-6 h) as in our fluorescence-based growth assay 323 (Supplemental Fig 3C).  Together, these data demonstrate the utility of our fluorescence-based growth assay in 334 determining the growth dynamics of intracellular bacteria such as R. parkeri in a high-335 throughput manner and at high temporal resolution. 336 337

Inhibition of MreB causes cell rounding and slow growth 338
With assays for quantitative evaluation of growth and morphology in hand, we sought to 339 apply these tools to assess the contribution of molecular machinery predicted to impact 340 morphogenesis in R. parkeri. In many rod-shaped bacteria, the elongasome directs

Conclusions and outlook 381
In an era of antibiotic resistance, understanding essential biological processes in 382 obligate intracellular human pathogens is of utmost importance, but requires the 383 development of new tools. In our study, we developed and applied quantitative methods 384 to assess peptidoglycan chemistry, cell morphology, cell cycle status, and growth 385 kinetics of the representative SFG pathogen, R. parkeri. Using these tools, we 386 established a fundamental understanding of R. parkeri morphogenesis in cultured 387 epithelial cells. Moving forward, we can begin to apply these tools to assess minute. The flow-through and column were discarded. Next, the DNeasy Mini spin 419 column was placed in a new 2 mL collection tube and 500 µL Buffer AW1 was added. 420 The columns were centrifuged for 1 minute at ~6,000 x g. The flow-through and 421 collection columns were discarded. Next, DNeasy Mini spin column was placed in a new 422 2 mL collection tube and 500 µL Buffer AW2 was added and centrifuged for 3 minutes 423 at ~20,000 x g. The flow-through and collection tubes were discarded. After discarding 424 the flow-through, the columns were centrigued at ~20,000 x g for 1 minute. Lastly, the 425 DNeasy Mini spin column was transferred into a clean 1.5 mL Eppendorf tube and 100 426 µL water were directly added into the DNeasy membrane, incubated at room 427 temperature for 1 minute and centrifuged for 1 minute at ~6,000 x g.

ZapA-mNeonGreen 457
To construct a plasmid encoding zapA-mNeonGreen (mNG) driven by the PzapA 458 promoter, zapA was PCR-amplified from R. parkeri genomic DNA using oEG1568 and 459 oEG1569 and digested with NdeI and NheI. A custom plasmid (pEG1959, pUC-IDT tagged genes from PompA with the ompA terminator. mNG was PCR-amplified from this 462 plasmid using oEG1644 and oEG1645 and digested with NheI and NotI. pUC-IDT 463 AmpR PompA-mNG-ompA3'UTR was digested with NdeI and NotI and was used as the 464 backbone in a triple-ligation with the zapA and mNG fragments. The resulting plasmid 465 (pEG1963) was digested with SacI and NdeI and ligated with PzapA, which was PCR-466 amplified from R. parkeri genomic DNA using oEG1672 and oEG1673 and digested 467 with the same enzymes. This plasmid was then digested with SacI and SacII to obtain 468 the PzapA-zapA-mNG-terminator fragment, which was then ligated in to similarly digested 469 pRAMEG05 to generate pRAMEG07. The homogenate was next transferred to (2) 1.5 mL Eppendorf tubes and the host cell 486 debris was pelleted by centrifuging the tubes for 5 minutes at 200 x g. To pellet the 487 bacteria, the supernatant was transferred to (2) 1.5 mL Eppendorf tubes and centrifuged 488 at max speed 16,300 x g for 2 minutes. The pellets were resuspended in 1 mL BHI and 489 frozen at -80 °C. The pellets were resuspended in 1 mL BHI and frozen at -80 or else pellet the bacteria, the supernatant was ultracentrifuged at 58,300 x g for 30 minutes at 552 brain heart infusion (BHI) broth (VWR, cat. no. 90003-040) per infected T175 cm 2 flask, 554 aliquoted, and frozen at −80 °C. Freeze-thaw cycles were minimized to one per stock. 555 556 R. parkeri strains producing TaqBFP or BFP-ZapA-mNG were expanded following the 557 initial electroporation and imaging-based screening. From the original plaque 558 "mechanical bead disruption" (200 uL), ~100µL of bacteria were used to infect (2) T75 559 cm 2 of confluent Vero cells. Infections were incubated in DMEM (high glucose, +L-560 Glutamine) medium with 2% Hi-FBS supplemented with 200 ng/mL rifampicin. Infected 561 cells were collected, mechanically disrupted, resuspended in 750µL BHI and stored at -562 80C (passage 2, P2). Next, passage 2 was further expanded as described above, 563 except P3 expansion was scaled up using larger flasks (T175 cm 2 ). Briefly, (2) 564 confluent T175 cm 2 flasks of confluent Vero cells were infected with ~100µL of bacteria 565 and incubated until ~80% of the cells were infected. Infected cells were collected, 566 mechanically disrupted, resuspended in 1mL BHI and stored at -80C. 567 568

Peptidoglycan analysis 569
For the muropeptide composition analysis (Fig 1), Vero cells were grown in four groups 570 of 20 T175 cm 2 flask. Two groups were infected with parental WT 571 R. parkeri at an MOI of about 0.05-0.1 and incubated at 33 °C with 5% CO2 for ~4 days. 572 The other two uninfected groups were also incubated at 33 °C for 4 days. Cells were 573 harvested using the cell propagation method described above. Briefly, the cells were 574 scraped into the media, collected in centrifuge bottles, and centrifuged at 12,000 x g for 575 transferred to a cold dounce homogenizer, and dounce-homogenized for 40 strokes. 577 The homogenate was centrifuged at 200 x g for 5 min at 4 °C to remove the eukaryotic 578 cell debris. To pellet R. parkeri, the supernatant was ultracentrifuged at 58,300 x g for 579 30 minutes at 4 °C in an SW-27 swinging-bucket rotor. Each of the resulting five cell 580 pellets were resuspended in 3 mL 0.9% NaCl. 581 582 UPLC analysis was performed as described previously (31,32). In brief, pellets 583 resuspended in 3 mL 0.9% NaCl and were boiled in 3 mL SDS 5% for 2 h. Sacculi were 584 repeatedly washed with MilliQ water by ultracentrifugation at ~540,000 x g for 15 min at Quantitative evaluation of the morphology of live or fixed R. parkeri (Supplemental 619 Fig. 1, Fig. 5B To evaluate growth kinetics of R. parkeri, A549 (Fig. 4) and Vero (Supplemental Fig.  663 3) cells were grown in 24-well MatTek dishes (No. 1.5 uncoated coverslip, 13 mm glass, 664 cat no.) for 1-2 days (or until reaching 95% confluency) and infected with R. parkeri 665 producing GFPuv, AausFP1, or TagBFP at an MOI of 0.5-1.2 for 24 hours before imaging. To run the growth curves, a multi-mode imaging plate reader method was 667 developed using the Cytation1 imaging reader (Agilent, Biotek). Phase-contrast and 668 fluorescence images were captured through the Olympus Plan Fluorite 20× objective 669 (numeric aperture, 0.45) with a 465 nm LED GFP or 365 nm LED BFP filter cubes, and 670 focused using a laser autofocus cube (Agilent BioTek 1225010). Images were collected 671 every 3 hours for ~55 h and the infection was incubated at 33 ºC and 5% CO2. Using 672 the Gen5 software, an automated protocol was developed in which 9 images and 10 673 slices were obtained at the center of wells. These images were subsequently crop 674 stitched, linearly blended, transformed by background flattening, and Z-projected by 675 using a focus stacking method. Next, a cellular analysis was conducted on the Z-676 projected images to allow for calculation of the sum of GFP or BFP intensity. 677

PFU growth curves 679
To evaluate growth kinetics of R. parkeri producing GFPuv or GFPAa in Vero cells by 680 plaque assays (Supplemental Fig. 3C), cells were grown in T75 cm 2 flasks and 681 infected with Rp-GFPuv or Rp-GFPAa. To quantify the number of infectious bacteria, 682 infected Vero cells were harvested from the T75 cm 2 flasks, and serial dilutions were 683 performed in cold BHI. Vero cells grown in 6-well plates were then infected and the 684 infection was incubated for 30 minutes at 37 o C. DMEM 2% Hi-FBS 0.5% Ultrapure 685 agarose (Invitrogen) was used to overlay the wells. A second layer of DMEM 2% Hi-686 FBS 0.5% Ultrapure agarose containing 200 µg/mL rifampicin was added ~16 h after. 687 The plaques were quantified 5-7 days post infection.

Global complex cell shape analysis 690
For the cell shape analysis on Fig 5B and Supplemental Fig 5, binary masks of R. 691 parkeri producing cytoplasmic GFP were made and loaded into the Celltool software 692 (55). Cell shapes or "polygonal contours" of control cells as well as cells treated with 0, 693 25, 50 or 100 µM MP265 were extracted from the binaries. To take into consideration 694 the shape variability in the MP265-perturbed cells, a model for all the contours was 695 made. 95% of the shape variance was accounted to "shape mode 1", which roughly 696 reflects cell length and "shape mode 2", which roughly reflects cell width. To understand 697 the impact of increasing concentrations of MP265 on R. parkeri cell shape, a two-698