In vitro Wolbachia infections in D. melanogaster cell culture are stable over time

We successfully established and maintained in vitro wMel and wRi infections in two D. melanogaster cell lines, the neuroblast-like JW18 cell line [20] and the macrophage-like S2 cell line [21] (diagramed in S2 Fig). Fluorescence in situ hybridization (FISH) with 16S rRNA probes visually confirmed the presence of Wolbachia in infected cells (S1B–S1F Fig) and its absence from doxycycline-cured (DOX) cells (S3A–S3F Fig). We used whole genome sequencing (WGS) and reference genome mapping to confirm infection strain identities, estimate the genomic titer of each symbiont infected cell line, and observe fluctuations in titers over time. We consistently observe wMel at a higher titer (∼10–30) than wRi (∼0.1–3) (S3G Fig and S1 Table).

The wMel strain outcompetes the wRi strain from equal starting ratios

The wMel strain of Wolbachia outcompetes the wRi strain in D. melanogaster in vitro infections. To recapitulate the conditions of a mixed Wolbachia strain infection in vitro, we mixed wMel and wRi infected cells at approximately equal genomic titers (Fig 1A). This equal starting ratio was selected to not advantage either strain and study the differences in infected host cell and strain growth rates. Each mixed culture was split into triplicate, and passaged every seven days, with a sample collected for sequencing at each passage. We estimated the abundance of each symbiont by calculating the proportion of total coverage contributed by that symbiont, which is the average coverage of the symbiont divided by the sum of the average coverages of both symbionts and the host (S1 Table). In the first three weeks immediately following the initial mixing of the two strains, both wMel and wRi increased in frequency. However, after this phase of initial expansion, only wMel continued to increase in frequency. By week five, wMel accounted for an average proportion of total coverage of 90% (Fig 1F). During this timeframe, the JW18 neuroblast-like D. melanogaster cell culture cells exhibited adherence defects that suggested the cells were under stressful conditions, whereas the S2 cells maintained their normal phenotypes (Fig 1B–1E).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The wMel strain consistently outcompetes the wRi strain in D. melanogaster cell culture.

A) Schematic overview of the 1:1 wRi:wMel mixed infected cell line experiment. B-E) Tissue culture micrographs of the mixed cell lines at B,D) week 3 and C,E) week 4. B,C at 40x and D,E at 20x magnification. F) Proportion of Illumina whole genome sequencing coverage mapped to the wMel (blue) and wRi (red) genomes out of the total coverage mapped to all Wolbachia and D. melanogaster host genomes, plotted by replicate and host cell type (S2, dashed or JW18, solid). G) Relative growth rates of wMel compared to wRi over the first seven weeks of wMel exponential growth for the cell lines and replicates in (F). The slope of these plots was used to calculate the selection coefficients in S2 Table.

https://doi.org/10.1371/journal.ppat.1012149.g001

We used a simple haploid model of relative fitness (see Methods) to estimate the selection coefficient (ω) of wMel in the mixed infection experiments. Because we observed that wMel replaces wRi within five to seven weeks post mixing, we constrained our selection coefficient estimates to six weeks post-mixing in order to capture the early dynamics of selection acting on the two strains (Fig 1G). We estimated selection coefficients ranging from 2.81–3.06 in the JW18 cell line, and 1.80–2.36 in the S2 cell line (S1 Table). These values indicate that wMel is far fitter in D. melanogaster cells than wRi. However, this selective advantage may have been influenced by wMel’s high starting concentration. Next, we explore whether wMel outcompetes wRi when it is a minority constituent in two-strain mixtures.

Deterministic growth: wMel’s selective advantage is independent of starting infection frequency

The wMel strain outcompetes wRi when it is the minority strain in host cell culture cells, indicating that wMel is a deterministic competitor whose selective advantages are not dependent on starting infection frequency. To assess whether wMel’s competitive advantage is frequency-dependent or deterministic, we mixed wMel-infected and wRi-infected cells at approximately 1:100 and 1:1000 ratios based on the relative genomic titers of the respective strain in the stable-infected cell lines. The wRi strain is at lower titer than the wMel strain in both S2 and JW18 cells, limiting the titer mixtures to this value (∼0.3–4.1). Relative titers were measured by Illumina whole genome sequencing each week over 11 weeks (S1 Table). Similar to the equal titer mixtures, we observed a rapid increase of the frequency of wMel within five to seven weeks post-mixing in both the 1:100 and 1:1000 mixtures across both cell lines and all replicates (Fig 2A and 2B). However, in contrast to the 1:1 mixtures, the wMel strain required more time to become fixed, only reaching an average proportion of total coverage of 86% by week 10 in both the 1:100 and 1:1000 mixtures.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The wMel strain deterministically outcompetes the wRi strain in mixed infections, even when starting at only 1/100th or 1/1000th the frequency of wRi.

A) Schematic overview of the 1:100 and 1:1000 wRi:wMel mixed infected cell line experiments. B,C) Representative epifluorescence FISH images of week two of the 1:1000 wMel:wRi mixture (replicate A). B) JW18 cell line and C) S2 cell line at 20x; scale bar = 50 μm, DAPI = blue, Jupiter-GFP = green (JW18 only), and Wolbachia 16S rRNA = red. D,E) Proportion of wMel (blue) and wRi (red) genome coverage out of the total coverage of all Wolbachia strains and D. melanogaster host genomes, plotted by replicate and host cell type (S2, dashed or JW18, solid) in mixed infections started at wMel:wRi ratios of D) 1:100 and E) 1:1000. F,G) Relative growth rates of wMel compared to wRi in mixed infections started at F) 1:100 and G) 1:1000 ratios. The slopes from F and G were used to calculate the selection coefficients (ω) overlaid in the plots, also in S2 Table.

https://doi.org/10.1371/journal.ppat.1012149.g002

The frequency of wMel relative to wRi increased continually over the 11 week experiment in both cell lines in the 1:100 mixtures, allowing us to estimate the strength of selection acting on wMel over the total length of the experiment (Fig 2C). However, in the 1:1000 mixtures wMel was undetectable in week 0, highlighting the extreme disadvantage in initial frequency when compared to wRi. Therefore, we estimated selection coefficients for these mixtures from week one onwards (Fig 2D). In the 1:100 mixtures, selection coefficients (ω) for wMel ranged between 2.62–2.92 and 2.31–2.33 in the JW18 and S2 cell lines, respectively. In the 1:1000 mixtures ω ranged from 3.36–3.63 in the JW18 cell line, and 2.76–2.87 in the S2 cell line (Fig 2F and 2G). Interestingly, we found that in the 1:1000 mixtures, the wMel strain grows significantly faster than wRi and exhibits higher selection coefficients than in the 1:100 mixtures across both cell lines (S4 Fig). This suggests that wMel is able to modulate its growth rate to more efficiently populate host cells when starting at a lower initial frequency.

Given Wolbachia’s propensity for recombination [22–24], we tested for the presence of recombinant haplotypes between the competing strain genomes in the 1:1, 1:100, and 1:1000 wMel:wRi mixed infection experiments. Putative recombinant events were detected by identifying overlapping chimeric alignments to both the wMel and wRi genomes in regions of high mappability. In total, we identified 67 recombinant events across all experiments. Although these events were extremely rare, we found they were relatively well distributed across both genomes. However, we did identify enrichment of recombinants in three 20kb regions in wRi and two 20kb regions in wMel (S5 Fig). The highest number of putative recombinant events occurred when strains co-occurred the longest, in the 1:1000 S2 mixtures. (S6 Table). These results make intuitive sense, as recombination mediated through passive processes such as homology-directed repair with divergent strain eDNA requires high concentrations (equal strain mixtures) and many chances (long co-culture times).

The competitive dynamics between wMel and wRi in our in vitro experiments offer insight into the mechanisms that might limit the frequency and stability of mixed infections in vivo, in nature. The quick and reproducible competitive exclusion of wRi by wMel in two D. melanogaster cell types across a range of starting frequencies suggests that mixed infections resolve reliably and quickly, consistent with theoretical predictions [25]. This potentially explains why unstable mixed infections (opposed to stable superinfections) are rarely observed in nature [17–19]. The selection coefficients estimated for wMel demonstrate a strong relative fitness compared to wRi across both D. melanogaster cell lines. However, wMel is natively associated with D. melanogaster, therefore this competitive advantage may reflect host-specific adaptations [26]. To explore whether the relative superiority of wMel as a cellular symbiont is specific to its native host, we immortalized a D. simulans cell line to repeat these investigations in wRi’s native host background.

Reciprocal infections: The wMel strain maintains its competitive advantage in wRi’s native host D. simulans

To assess the contribution of host-specific adaptations to the competitive advantage of wMel in Drosophila melanogaster, we immortalized a new D. simulans cell line from the white eye fly stock infected with the Riv84 wRi strain named Dsim6B. Initially, these cells were heterogeneous and infected with wRi (Fig 3A). Often the wRi-infected cells exhibited aberrant cellular morphologies. As the Dsim6B cell line stabilized and became more clonal, the infection was lost (Figs 3A and S5). Despite high wRi titers in D. simulans in vivo fly tissues (4.5x average genomic titer, Fig 3B) [6], repeated attempts to reinfect the cells with wRi from the stably infected D. melanogaster cell lines via the shell vial technique failed (Figs 3C and S6). In contrast, infections of Dsim6B cells with the wMel strain were very successful (S6 Fig), and the rate of titer increased to stable levels of 1-2x genomic titer depending on the initial input concentration (Fig 3D and S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The wMel strain is better at infecting D. simulans cells than D. simulans’ native strain, wRi.

A) The Dsim6B cell line was immortalized from D. simulans [w-] embryos infected with the Riv84 strain of wRi. The primary and early immortalized cell line was infected with wRi, but the bacteria were gradually lost as the cells increased in growth rate and clonality. By nine months post-infection the Dsim6B cell line had cured itself of its wRi infection. Repeated attempts to reinfect the Dsim6B cell line with wRi were unsuccessful. B) Bar plots of stable wMel (blue) and wRi (red) titers in D. melanogaster and D. simulans cells and flies (three bars indicated with fly icons). C) FISH widefield images of Dsim6B cell lines uninfected (0:1 titer), infected with the wMel strain (2:1 titer), and after attempts to reinfect with the wRi strain (∼0:1). D) Titer increase over time in the Dsim6B cell line infected with wMel via the shell vial technique at 1/10x the concentration in JW18 cells and at 1x, compared to stable Dsim6BwMel cell line infections (maintained for more than three months).

https://doi.org/10.1371/journal.ppat.1012149.g003

The differential success of wRi and wMel infections observed in our D. simulans cell line suggests that host developmental programs may enable the persistence of costly Wolbachia infections. Cell culture conditions are distinguished from in vivo conditions primarily by their simplicity of cell and organism types (sterile monoculture for both host and symbiont), which wRi may be poorly evolved to handle, despite its close relationship to wMel (99.91% identical across the 1.3–1.4 Mb genomes). Alternatively, wRi may be a better “developmental symbiont” than a “cellular symbiont”. The wRi strain’s in vivo high titers and promiscuity across fly species suggests that its persistence may be heavily reliant on a developmentally-constrained system in which the maintenance of specific host cell lineages is crucial for organismal survival. In a cell culture system, cells can replicate freely because they are free of the limitations placed on cell proliferation in a developing host. Consequently, if the growth of uninfected cells outpaces infected cells, then the infection will be lost. Given that we were able to establish and maintain wMel infections in both D. melanogaster and D. simulans cell lines, wMel may not rely as heavily on the developmental context of the host as wRi. To explore this idea further, we characterize the growth dynamics of each strain into uninfected host cells over time.

Infection expansion into uninfected host cells recapitulates wMel’s spread into wRi-containing cells

Successful Wolbachia cellular infections require healthy host cell growth, in addition to some rate of bacterial segregation during host cell division and cell-to-cell transfer to uninfected host cells. The weights of these three parameters are interdependent: if infections impact host cell growth, then cell-to-cell transfer rates need to be high to enable the infection of faster-growing uninfected cells. Otherwise, the infection will be lost due to uninfected cell overgrowth. Similarly, cell-to-cell transfer rates can only be negligible if the infection has minimal cost on host cell growth rates and segregation is efficient.

To understand the cellular basis for wMel’s competitive advantage over wRi in vitro, we studied the expansion of these Wolbachia strains into uninfected host cells, revealing that wRi fails to establish when fewer than 50% of host cells are infected. We mixed JW18 and S2 cells infected with the wMel or the wRi strain of Wolbachia and uninfected at approximately equal quantities. Infection growth curves following the addition of 1:1 uninfected host cells to wMel-infected cell lines revealed a similar pattern of expansion as in the wMel-wRi competition experiments: across both the JW18 and S2 cell lines and all three replicates, wMel genomic titer increased rapidly in the first five weeks, and remained at a relatively stable frequency throughout the rest of the experiment (Fig 4A and S7 and S1 Tables). On average, wMel titer increased by 17% and 16% per week in the JW18 and S2 cell lines, respectively (Fig 4E and S3 Table). Conversely, in the wRi-DOX mixtures, we observed the continuous decline of wRi genomic titer in the JW18 cell line, with an average rate of 14% per week. Similarly, in the S2 cell line wRi genomic titer declined on average by 10% per week, despite the initial increase in the first week post co-culture (Fig 4B, 4E and S3 Table). Overall, the observed patterns of wMel’s growth in the wMel-DOX mixtures illustrate that the symbiont can effectively establish and expand an infection within the cell lines, and suggests horizontal transmission as a mechanism for infection establishment. To assess the impact of Wolbachia infection on host cell dynamics, we next compared the growth rates of infected and uninfected D. melanogaster cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The wMel strain is able to efficiently spread to uninfected cells through faithful segregation and cell-to-cell transfer, whereas the wRi strain cannot.

A) Schematic overview of the 1:1 wMel:DOX and wRi:DOX mixed infected cell line experiments. C,D) Genomic titers for C) wMel (blue) and D) wRi (red) over time in 1:1 mixtures with uninfected JW18 (solid line) and S2 (dashed line) cells. E) Fold change in symbiont titer per week in each mixture. Fold change was calculated by log-linear regression (S8 Fig and S3 Table). F) Cell growth rates measured by hemocytometer cell counts, quantified as the proportional growth per day from the starting cell count at 23°C (purple) and 26°C (orange). Wilcoxon rank sum p-values **p< = 0.01 and ***p< = 0.001.

https://doi.org/10.1371/journal.ppat.1012149.g004

Measuring the growth rate of D. melanogaster cells with and without Wolbachia infections revealed that both strains slow host cell division, suggesting that successful establishment requires cell-to-cell transfer. Both JW18 and S2 cell lines divide significantly faster when uninfected than when infected with either the wMel or wRi strain of Wolbachia (p<0.01 Wilcoxon rank sum test; Fig 4C). When uninfected, JW18 cells double in 2.09 +/- 0.31 days (a growth rate of 1.45x cells per day), whereas wMel-infected JW18 cells require 3.75 +/- 0.34 days to double (0.385x cells per day) and wRi-infected JW18 cells require a massive 25.0 +/- 29.3 days to double (0.16x cells per day). Similarly, when uninfected, S2 cells double in 2.02 +/- 0.37 days (1.73x cells per day). When infected with the wMel strain, S2 cells require 3.71 +/- 1.06 days to double (0.46x cells per day) and infected with wRi, they require 3.21 +/- 0.63 days to double (0.57x cells per day). Interestingly, wMel Wolbachia infection has minimal impact on the D. simulans Dsim6B cell line (3.11 +/- 1.26 vs 3.23 +/- 0.95 days to double and 0.86x and 0.64x cells per day, respectively). This may be due to the Dsim6B cell line’s lower growth rate: this cell line is highly adherent and fails to grow well at the ⅙ starting dilution that the uninfected S2 and JW18 D. melanogaster cell lines thrive with.

The negative impact of Wolbachia infection on host cell growth combined with wMel’s ability to rapidly increase in titer upon exposure to uninfected host cells indicates that cell-to-cell transfer is essential to the colonization process. D. melanogaster host cells require nearly twice as long to divide when infected with Wolbachia than when uninfected (Fig 4C). The loss of wRi from the 1:1 DOX-wRi mixtures is consistent with the replacement of infected cells with faster-growing uninfected cells over the ten weeks of co-culture (Fig 4B). In contrast, wMel’s increase in frequency over time after 1:1 mixture with DOX-cured host cells (Fig 4A), despite their inhibition of host cell division rates (Fig 4C), is consistent with efficient cell-to-cell transfer to uninfected host cells. This transfer process not only increases wMel frequency in the culture, but also prevents uninfected host cells from remaining uninfected and out-growing the infected cell population. Thus, despite not being able to specifically label, track, and distinguish living Wolbachia strains, we were able to detect indirect evidence of wMel’s superior cell-to-cell transfer ability, relative to wRi.

The in vitro Wolbachia infections harbored low levels of genetic variation and this variation fluctuated across experimental time points. Average genome-wide pairwise diversity ranged from zero to 8.1e-05 (S9 Fig). Across experimental time points, diversity increased with increasing titer (S10 Fig) because detection of variation was limited by the depth of Wolbachia genome sequencing coverage (S1 Table). Despite the low levels of variation and the limits of sequencing depth in low-titer infections and time points, we did detect a handful of wMel and wRi SNP and indel alleles that fluctuate in frequency, but do not reach fixation (S11–S14 Figs). These alleles were shared among replicates and cell lines, suggesting that they were introduced as standing variation from the stable cell lines, opposed to being derived from de novo mutation in the experiments. The 15 high-frequency wMel variant sites were located in hypothetical proteins, pseudogenes, ncRNAs, and protein-coding genes involved in oxidative stress and electron transfer (S7 Table). The five high frequency wRi variant sites were located exclusively in pseudogenes and intergenic regions. While some of these alleles may confer fitness benefits in vitro, either for competition or infection, the mutations in intergenic regions and pseudogenes are likely neutral and pose no impact on the fitness of Wolbachia. These results hold promise for our ability to track allelic variation under natural selection in Wolbachia in vitro infections. Future work will confirm the functional nature of these mutations with deep long read sequencing.

Cell culture maintenance and cell line generation

All Drosophila cells were maintained on either Shields and Sang M3 Insect Medium (MilliporeSigma S3652) or Schneider’s Insect Medium (MilliporeSigma S0146) supplemented with 10% v/v Fetal Bovine Serum (FBS, ThermoFisher A3160502). Cells were maintained in 4 mL of media in plug-seal T-25 flasks (Corning 430168) in a refrigerated incubator at either 25–27°C or 22.5–23.5°C, as indicated in the text. We performed weekly cell splits at a 1:6 dilution for uninfected cell lines and 1:2 or 1:3 dilutions for Wolbachia-infected cell lines, following visual inspections of cell growth and contamination. Adherent cells were removed by scraping with sterile, bent glass pipettes. Transitions between media types were performed in 25% intervals, requiring four weeks to transition to 100% Shields and Sang or Schneider’s Medium.

Drosophila melanogaster JW18 cells [20] and S2 cells (Thermo Fisher and [21]) were derived from a primary culture of 1–15 hr and 20–24 hr-old embryos, respectively. JW18 cells are naturally infected with the wMel strain of Wolbachia (from the in vivo infection in the fly line the cells were derived from) and S2 cells are naturally uninfected. We found that incubation temperature exerts an observable effect on Wolbachia density within these cell lines, supporting the well established relationship between Wolbachia density and temperature [33]. Specifically, cells cultured at 26°C in 2021 exhibited higher symbiont titers compared to the same cultures incubated at 23°C in 2023. Importantly, the relative differences between wMel and wRi titer are consistent between these temperature regimes: wMel is always at an order of magnitude higher titer than wRi. The difference in incubator temperatures was necessitated by the last author’s starting her new lab and buying a new incubator capable of maintaining 23°C. To generate uninfected JW18 cells, we treated JW18 wMel-infected cells with 10 μg/mL doxycycline in supplemented Shields and Sang media.

We generated the Drosophila simulans Dsim[w-]6B cell line from a w[–] (white eye) fly line previously infected with the Riverside 1984 strain of wRi Wolbachia [34,35] according to the method described in [36]. Briefly, 1–20 hr old embryos laid on grape-agar plates by Wolbachia-infected flies were collected, surface sterilized, homogenized, and plated in flasks on rich media containing 20% FBS and Wolbachia-resistant antibiotics, 60 and 100 μg/ml penicillin-streptomycin and 50 μg/ml gentamicin. During the next six months of maintenance, two of the initial twenty seed flasks converted into immortal tissue culture lines. The Dsim[w-]6B cell line was selected for further pursuit due to its planar growth pattern and ability to hold a Wolbachia infection. The native wRi infection is unstable in Drosophila in vitro culture systems long-term, as described in the Results and Discussion sections, and the natural infection was lost naturally over the course of the first year of culture.

Wolbachia infections were introduced by adding 1.2 μm-filtered infected cell lysate to uninfected D. melanogaster JW18 and S2 and D. simulans Dsim[w-]6B cells. Infected cell lysate was either obtained from Wolbachia-infected cell cultures or fly embryos (collected on grape plates, as described in [36]. All wMel infections were initialized with the wMel strain that naturally immortalized in the JW18 cell line [20]. This wMel strain is known to lack the octomom region [37], which is linked to high titer in the wMelPop strain [38]. Infected cells were serially passed through 5 μM and 1.2 μM syringe filters to produce Wolbachia-containing cell lysate. The wMel strain was applied directly (in 3 mL lysate) to uninfected S2 cells to produce the S2wMel cell line in 2017. To produce the wRi-infected cell lines and the Dsim[w-]6B cell lines, we applied 0.5–1 mL of Wolbachia-containing cell lysate to a monolayer of uninfected host cells in a flat-bottom shell vial and centrifuged the bacterial cells down onto the cell surface in a swinging bucket centrifuge at 2500 x g for 1 hr at 15°C (i.e., the shell vial technique [39]). We transferred these cells to T-12 flasks in a final volume of 2 mL for five days before scraping and transferring the cells to a T-25 flask with 2 mL of fresh media. These lines were maintained by weekly 1:2 “soft splits”, which removed no media.

All cell lines were validated with DNA-based probes and whole genome sequencing after construction and continuously during maintenance and experimentation. Cell line infection status was continuously monitored by PCR and fluorescence in situ hybridization (FISH) of Wolbachia-specific markers. Primers for the Wolbachia Surface Protein (WSP) gene [35] were used to confirm the presence and absence of wMel and wRi strains in infected and uninfected cell lines, respectively. Sanger sequencing of the WSP amplicons was performed by Azenta to confirm the strain-specific amplicon sequences. Oligonucleotide DNA probes complementary to the Wolbachia 16S rRNA sequence were used in FISH experiments following the protocol in [40] to confirm infections and estimate per-cell Wolbachia titer. Whole genome shotgun sequencing was performed with Illumina sequencing (see below) to confirm host species and Wolbachia strain identities and test for contamination.

Mixed cell line experiments

We used average genomic titer measurements from sequencing stable cell line infections to calculate how many cells of each infection and host cell type to mix for the desired mixture ratios. Host cell concentrations were quantified with a hemocytometer manually or with a Millicell Digital Cell Imager.

For strains A and B at titers of Y_A and Y_B symbiont cells/host cell, within host cells growing at densities of X_A and X_B cells/mL, mixed at a ratio of A/B, in a final volume of 4 mL per cell culture flask:

Volume of host cell culture infected with strain A = V_A = 4 mL/(X_A/X_B * Y_A/Y_B * 1/(A/B) + 1)

Volume of host cell culture infected with strain B = V_B = 4 mL ‐ V_A

Samples were collected prior to mixing, immediately after mixing, and weekly when splitting infected cell cultures into new flasks at 1/2 dilutions. For each culture at each timepoint, one mL of scraped and mixed cell-containing media was transferred to a 1.5 mL Eppendorf tube, the cells were centrifuged at 16,000xg at 4–10°C for 10 min, supernatant was discarded, and the cell pellet was snap-frozen and stored at -80°C until DNA extraction. Pellets were processed for library prep within one month of sample collection.

Shell vial experiments

To monitor how Wolbachia infections spread across uninfected host cells following introduction with the shell vial technique [39], we performed shell vial infections as described above for the creation of novel cell lines. Given the limited material at the start of these protocols (∼2 mL per experiment and < 1 million cells), we waited until the transfer step to T-25 flasks to take the first sample for Illumina sequencing and genomic titer quantification. Host cell-free wMel Wolbachia lysate was either added at the full concentration derived from host cell lysis or a 1/10 dilution to approximate the lower concentrations exhibited by wRi infections.

Cell growth rate experiments

Cell line cells were quantified upon splitting and seeding into new flasks and after a week’s incubation with a hemocytometer and Millicell Digital Cell Imager. While handling the cell lines as described above for “Cell Culture Maintenance”, we added one extra mL of fresh media to each flask so that one mL could be removed for sampling cell concentration and relative Wolbachia genomic titer (as a final step in the splits). These one mL samples were then quantified by counting cells in a 10 uL volume (X number of cells (>100) measured per Y number of boxes (>1 if <100 cells/box) * W dilution factor (2 if diluted by ½) * 10,000 mL^-1 = Z number of cells/mL). The rest of the cell suspension was pelleted by centrifugation (as described above), snap-frozen, and stored at -20-80°C until DNA extraction. This process was repeated one week later, except cells were resuspended by scraping prior to media removal so that the week’s worth of growth could be quantified. Cells were then diluted as described above for normal maintenance. This modified step was repeated every other week for six weeks, at most frequent.

Cell imaging and image analysis

Cell lines and experiments were continuously monitored with a tissue culture (TC) microscope and imaged with a monochromatic digital camera. Weekly, stable line and experiment cell splits were imaged on Zeiss Primovert TC microscope or a Leica DMi8 inverted microscope for confluency and contamination.

Infections were confirmed by fluorescence in situ hybridization (FISH) using DNA oligonucleotide probes complementary to the Wolbachia 16S ribosomal RNA sequence, following the protocol in White et al. 2017 [40]. Briefly, for each cell type, infection state, or experimental replicate, 1 mL of confluent cells were pipetted from a T-25 flask into a 6-well dish (Corning) one to three days before fixation. Upon confluency in the dish, cells were fixed in 8% paraformaldehyde in 1x phosphate-buffered saline (PBS) for 15 min at room temperature (RT). Following two washes with 1x PBS, cells were treated with prehybridization buffer, consisting of 50% deionized formamide by volume, 4x saline sodium citrate (SSC), 0.5x Denhardt’s solution, 0.1 M dithiothreitol (DTT), and 0.1% Tween 20 in deionized water, for one and a half hours. Following prehybridization, cells were incubated in hybridization buffer (prehybridization buffer without Tween 20) containing 500 nM Wolbachia W2 fluorescent DNA probe (5-CTTCTGTGAGTACCGTCATTATC-3) (Bioresearch Technologies) [41] at 37°C overnight. Wet kimwipes were added to the dish to prevent dehydration. The next day, cells were washed three times with 1x SSC with 0.1% Tween 20 at RT quickly, at RT for 15 min, and at 42°C for 30 min. Next, the cells were washed with 0.5x SSC at RT quickly, at 42°C for 30 min, and at RT for 15 min. These stringent washes aimed to remove unbound FISH probes from the cells. Finally, cells were washed three times with 1x PBS at RT before either staining with 3uM DAPI (4’,6-diamidino-2-phenylindole) in 1x PBS for 10 min or mounting in Vectashield fluorescent mounting medium with DAPI (Vector Laboratories).

FISH experiments were imaged on a Leica DM5500B widefield microscope or an inverted DMi8 equipped with LEDs for epifluorescence imaging. Raw Leica images were processed in Fiji [42] and analyzed in R [43].

Whole genome resequencing and analysis