Annotation of the sequenced R. buchneri ISO7^T genome [GenBank: JFKF01000000.1; (18)] identified the presence of two clusters of genes with similarity to bacterial genes for aminoglycoside, polyketide, and non-ribosomal peptide synthesis. This strain of R. buchneri was isolated from the ovaries of an I. scapularis female collected from a dog in Minnesota (18). To determine potential functions of proteins in the two putative antibiotic clusters, amino acid sequences from the R. buchneri ISO7^T genome were searched against other available sequences using the NCBI protein–protein BLAST algorithm, performed with default parameters. Putative protein function was determined by examining data for each protein in the InterPro database as well as performing literature searches. The gene clusters were also compared to the R. buchneri genome derived from the I. scapularis genome sequence, REIS (Wikel) [GenBank: CM000770.1; (54)]. Protein sequences obtained from annotated genomes were aligned using ClustalW (75) and MUSCLE (76) in MacVector version 18.1.5.

Embryonic tick cell lines ISE6 (77), IRE11 (78), and AAE2 (79), derived from I. scapularis, I. ricinus, and A. americanum, respectively, were maintained at 34°C in L15C300 medium supplemented with 5% heat-inactivated fetal bovine serum (FBS), 5% tryptose phosphate broth (TPB), and 0.1% lipoprotein concentrate (LPC; MP Biomedicals, Irvine, CA, USA), adjusted to pH 7.2–7.5 with 1 M NaOH, as previously described (77). Rickettsia buchneri ISO7^T [(18); hereafter referred to as Rb-WT] and R. buchneri expressing GFPuv from the plasmid pRAM18dRGA [(80); hereafter referred to as Rb-GFPuv] were maintained in IRE11 at 28°C in a modified L15C300 medium containing 10% FBS, 5% TPB, 0.06% NaHCO₃, 6 mM HEPES, and 0.1% LPC; pH was not adjusted. Rickettsia parkeri Tate's Hell expressing mKate from plasmid pRAM18dSFA (Rp-mKate) (80, 81), R. peacockii-GFPuv [pRAM18dSGK; (80)], R. monacensis IrR/Munich with mKate (pRAM18dSFA), and A. phagocytophilum HGE1 expressing mCherry from an intergenic Himar1 transposon insertion (82) (A. phagocytophilum-mCherry) were grown in ISE6 cells, and R. amblyommatis Darkwater (kindly supplied by Chris Paddock, CDC) were grown in AAE2 cells. Infected tick cell cultures were maintained at 34°C in L15C300 with 10% FBS, 5% TPB, 0.1% LPC, 0.25% NaHCO₃, and 25 mM HEPES, adjusted to pH 7.5 (83). The infection level of cell cultures was assessed by Giemsa staining. Vero cells (African green monkey kidney) were grown in Gibco RPMI 1640 (Thermo Fisher, Waltham, MA, USA) supplemented with 10% FBS and 2 mM L-glutamine at 34°C following established methods (84). All cultures were grown in 25-cm² culture flasks (CELLSTAR, Greiner Bio-One, Monroe, NC, USA).

Host cell-free bacteria were prepared as previously described (83); heavily infected tick cells were added to tubes containing rock tumbler grit (60/90 coarse silicon carbide, Lortone, Mukilteo, WA, USA), vortexed for 30 s, then passed through a 2 μm filter to remove cellular debris. Cell-free bacteria were then collected by centrifugation at 13,200 × g for 5 min at 4°C.

Cell-free Rb-WT were prepared from infected IRE11 cultures as described above. Bacterial pellets were washed once in SPG buffer, then resuspended in 1 ml TRI Reagent (Sigma-Aldrich, St Louis, MO, USA), vortexed, and rested at room temperature for 10 min. Samples were centrifuged at 12,000 × g to remove particulates and the supernatant transferred to new tubes and mixed 1:2 with 100% ethanol, followed by vortexing. RNA was purified using a Direct-zol RNA Miniprep Kit (Zymo Research, Irvine, CA, USA). Contaminating DNA was removed by treating three times with Ambion TURBO DNA-free Kit (Thermo Fisher), followed by purification with an RNA Clean & Concentrator Kit (Zymo Research). Reverse transcriptase PCR (RT-PCR) was carried out using the Access RT-PCR system (Promega, Madison, WI, USA) in 25 μl reactions consisting of 5 × reaction buffer, 10 μM dNTPs, 10 μM of each primer, 0.5 μl Tf1 DNA polymerase, 14.5 μl nuclease-free water, and 1 μl sample. Nuclease-free water was used as a negative control, and no reverse transcriptase (no RT) controls included water instead of the DNA polymerase to confirm the absence of contaminating DNA. RT-PCR was performed in a Techne TC-312 Thermocycler with the following cycling conditions: 45 min at 45°C; 2 min at 94°C; 40 cycles of 30 s at 94°C, 1 min at 55°C or 58°C, and 2 min at 68°C; final extension 1 min at 68°C. Amplification of HTH, lagD, ppsE_1, kanC, and btrB was performed with an annealing temperature of 58°C, and amplification of glycogen synthase, homoserine kinase, and lgrB at 55°C. Primers used in this study are shown in Table 1. RT-PCR products were visualized on a 1.2% agarose gel stained with GelGreen (Biotium, Fremont, CA, USA).

Competition assays were set up in 24-well plates to compare the infection and replication of R. monacensis-mKate, Rp-mKate, and A. phagocytophilum-mCherry in tick cells with and without Rb-GFPuv infection. For the R. monacensis and R. parkeri experiments, three 24-well plates were prepared; the first contained uninfected IRE11, the second contained IRE11 heavily infected with Rb-GFPuv, and the third contained a mixture of infected and uninfected cells to give an approximate level of Rb-GFPuv infection of 45%. Cells (0.4 ml) were applied to each well, giving 1 × 10⁶ cells/well in the R. monacensis experiment or 3 × 10⁵ cells/well in the R. parkeri experiment. Plates were incubated in a humidified candle jar at 28°C. After 24 h, plates were infected with 250 μl fresh medium containing host cell-free R. monacensis-mKate or Rp-mKate, in 10-fold serial dilutions. Each dilution was applied to each plate in triplicate, and the remaining wells were used as negative controls. The number of mKate-positive colonies in the lowest-dilution wells was determined by fluorescent microscopy and used to extrapolate the number of rickettsiae in each dilution. Wells were observed on a Nikon Diaphot fluorescent microscope, and adhesion/invasion and replication were determined for each dilution on each plate by visualization of red fluorescent rickettsiae over 14 days. Additionally, the ability of Rp-mKate to infect IRE11 was assessed daily over the 14 days by observing at least 500 cells per well and determining the percentage of cells containing replicating fluorescent R. parkeri.

For A. phagocytophilum experiments, five 24-well plates were prepared; two contained uninfected IRE11, two contained IRE11 heavily infected with Rb-GFPuv, and the fifth contained a mixture of infected and uninfected cells to give an approximate level of Rb-GFPuv infection of 25%. Cells (0.5 ml) were applied to each well to give 1 × 10⁶ cells/well. One uninfected IRE11 plate and one infected IRE11 plate were incubated at 27°C in a humidified candle jar; the remaining three plates were incubated at 34°C with 4% CO₂. After 24 h, plates were infected with 250 μl medium containing 10-fold serial dilutions of host cell-free A. phagocytophilum-mCherry, added to plates in triplicate. The number of Anaplasma in each dilution was estimated by testing each dilution in a sixth 24-well plate seeded with ISE6 cells and counting the number of mCherry-positive colonies in the lowest-dilution wells and then extrapolating to each dilution. Wells were observed on a Nikon Diaphot fluorescent microscope, and adhesion/invasion and replication were determined for each dilution on each plate by visualization of red fluorescent bacteria.

A fluorescent plate reader assay was used to measure the growth dynamics of fluorescent Rickettsia in IRE11 cells. Uninfected IRE11 and IRE11 heavily infected with Rb-GFPuv (>95% cells infected) were adjusted to 1 × 10⁵ cells/ml in fresh 10% FBS medium. Heavily infected and uninfected cells were then mixed to create additional populations with 25%, 50%, and 75% cells infected (Figure 2). A volume of 200 μl of each cell population was added in triplicate to wells of a clear-bottomed black-sided 96-well plate (Falcon, Corning, NY, USA), with dH₂O added between wells to prevent drying, and incubated at 28°C in a humidified candle jar for 24 h to allow cells to settle to the bottom of wells. Host cell-free Rp-mKate were resuspended in 1 ml fresh medium and enumerated on a Petroff–Hausser chamber. The cell-free bacteria were diluted to create 1,000:1, 100:1, and 10:1 multiplicity of infection (MOI) ratios and then added to IRE11 cells in 96-well plates in a volume of 10 μl. The plate was returned to the humidified candle jar and incubated at 28°C. Fluorescence readings were taken 24 h later and then every 24 h up to 14 days postinfection. Readings were taken at room temperature (~22–25°C) on a BioTek Synergy H1 microplate reader at excitation/emission 395/509 for GFPuv and 588/633 for mKate and adjusted to uninfected IRE11 to account for background fluorescence.

Plate reader experiments for R. amblyommatis used a similar protocol to the above, except that only uninfected and 25% and >95% infected cell populations were used, and only 1,000:1 and 10:1 Rp-mKate challenges were performed. Additionally, only mKate fluorescence was measured for these experiments. The R. peacockii experiments were carried out using the same protocol as for Rb-GFPuv. Prior to IRE11 plate experiments, cell-free R. amblyommatis and R. peacockii (grown in AAE2 and ISE6 cells, respectively) were transferred to IRE11 and cultured in modified L15C300 medium at 28°C for at least 2 weeks to adjust them to growing in these conditions. Both Rickettsia species were found to grow well in IRE11 cells.

To assess the effect of cell-free R. buchneri and lysate of R. buchneri on the replication of Rp-mKate, a 96-well plate was set up with 200 μl of uninfected IRE11 adjusted to 1 × 10⁵ cells/ml in fresh medium, as above. Cells were treated with either 50 μl medium (negative control), 50 μl of cell-free Rb-WT prepared from 2.5 × 10⁵ heavily infected IRE11, 50 μl of a 1:10 dilution of the cell-free Rb-WT, 50 μl cell lysate from 2.5 × 10⁵ IRE11 heavily infected with Rb-WT, or 50 μl cell lysate from 2.5 × 10⁵ uninfected IRE11. Lysates were prepared by sonicating cells on ice at full power for a total of 1 min (separated into 3 × 20-s bursts, with 20-s intervals resting on ice). After 2 h, cells were challenged with cell-free Rp-mKate at 1,000:1, 100:1, or 10:1, and the plate was incubated at 28°C in a humidified candle jar. mKate fluorescence was measured every 24 h for 14 days, as described above, and adjusted to uninfected, untreated IRE11.

To determine if R. parkeri could induce R. buchneri antibiosis activity, an additional plate reader assay was used to examine whether lysates from R. buchneri challenged with Rp-mKate for varying lengths of time exhibited inhibitory effects on the growth of Rp-mKate in IRE11 cells. Wells of a 6-well plate were seeded with 2 ml IRE11 75% infected with Rb-WT, at 1 × 10⁵/ml. One well served as a no challenge control, while to the remaining wells cell-free Rp-mKate at a ratio of 100:1 was added. The plate was incubated at 28°C in a candle jar. After 24 h, cells from the control and one of the challenged wells were collected, and rickettsiae were isolated from IRE11 as above. Cell pellets were then frozen at −70°C. At 48, 72, 120, and 168 h after infection, this was repeated for cells from each of the remaining wells. Lysates were prepared from the pellets by four freeze thaw cycles of −70 to 37°C. The rickettsial lysates from each treatment were resuspended in 120 μl medium, and then 10 μl was added to 12 wells of a 96-well plate, each containing 200 μl IRE11 at 1 × 10⁵/ml, prepared 24 h previously. Wells were then challenged with Rp-mKate at ratios of 1,000:1, 100:1, and 10:1. Medium without Rp-mKate was added to control wells for each treatment. The plate was incubated at 28°C in a humidified candle jar, and mKate fluorescence was measured every 24 h for 14 days, as described above.

To test the antibiotic activity of R. buchneri against extracellular bacteria, antibiotic susceptibility testing was performed against Escherichia coli D21 and Staphylococcus aureus MN8 using disk diffusion assays. IRE11 infected with Rb-WT and uninfected IRE11 was pelleted by centrifugation at 350 × g for 6 min at 4°C, and the pellets were frozen at −70°C. Cell-free Rb-WT was prepared by vortexing with rock tumbler grit and filtration through a 2-μm filter, as described above, then the bacteria were pelleted by centrifugation at 13,600 × g for 7 min at 4°C, and the cell pellet frozen at −70°C. Lysates of the three samples were prepared by freeze-thawing (four cycles of −70 to 37°C) and then centrifugation at 13,600 × g for 5 min at 4°C. Bacterial or cell lysates were then resuspended in 50 μl medium, of which 30 μl was added to 50 μl of absolute methanol. A volume of 20 μl of each sample was then added to separate filter paper disks. Spectinomycin at 10 and 100 μg was added to two additional filter paper disks as positive controls. Disks were air-dried in a biosafety cabinet for 20–30 min and then placed onto Mueller–Hinton agar plates streaked with E. coli D21 or S. aureus MN8. Plates were incubated for 18 h at 37°C.

The disk diffusion assays were repeated using pellets of live IRE11, Rb-WT-infected IRE11 (25%, 50%, and >95% infected), and cell-free Rb-WT, as well as supernatant from IRE11 cultures at various levels of infection with Rb-WT (25, 50, >95%) or uninfected. Cells were pelleted by centrifugation at 350 × g for 6 min at 12°C, supernatant was removed to separate tubes, and then pellets were resuspended in 100 μl medium. A volume of 20 μl of resuspended cell pellets or 20 μl supernatant was added to filter paper disks. Spectinomycin was added to additional disks at 100 μg for positive controls. Disks were air-dried for 20-30 min in a biosafety cabinet, then applied to Mueller–Hinton agar plates streaked with E. coli D21 or S. aureus MN8. Plates were incubated for 18 h at 37°C.

To further test the antibiotic activity of supernatant from Rb-WT-infected IRE11 cultures against R. parkeri grown in mammalian cells, Vero cell cultures were grown at 34°C in 2.5 ml RPMI medium supplemented with 2.5 ml supernatant from either uninfected IRE11 or IRE11 heavily infected with Rb-WT. Cell-free Rp-mKate was added to Vero cultures and flasks checked daily for evidence of infection, which was assessed by the appearance of plaques in the cell layer as well as the timing and size of plaques.

To determine whether the expression of antibiotic genes by Rb-WT was induced by challenge with R. parkeri, and to determine the time of maximal expression, a time course experiment was set up comparing unchallenged and Rp-mKate-challenged Rb-WT-infected IRE11. A 6-well plate was prepared with each well containing IRE11 75% infected with Rb-WT at 1 × 10⁵/ml in 2 ml modified L15C300 medium. One well served as a no challenge control, while to the remaining wells cell-free Rp-mKate at a ratio of 100:1 in 10 μl was added. The plate was incubated at 28°C in a candle jar. After 24 h, cells from the control well were collected and centrifuged for 2 min at 500 × g, and the cell pellet was resuspended in 750 μl RNAlater solution (Qiagen, Hilden, Germany) and stored at−20°C. Cells from the challenge wells were collected at 24, 48, 72, 120, and 168 h after infection, resuspended in RNAlater solution, and stored at −20°C. RNA isolation was performed using TRIzol–chloroform extraction. cDNA was then prepared using Takara RT PrimeScript Kit (Takara Bio, Kusatsu, Japan), after a 30-min treatment with a gDNA eraser at room temperature to remove contaminating DNA. Quantitative PCR (qPCR) was performed on an Agilent Mx3005P RT-PCR system using the Agilent Brilliant II SYBR Green Master Mix, using 10-μM primers against various targets from both gene clusters (Table 1) and 1 μl cDNA preparation. Since both R. buchneri and R. parkeri were present in samples, I. scapularis GAPDH was used as a reference gene. Nuclease-free water and no reverse transcriptase controls were included on each plate. Each sample was run in triplicate. Cycling conditions were 10 min at 95°C, followed by 40 cycles of 30 s at 95°C, 1 min at 55°C (glycogen synthase, GAPDH), or 52°C (btrB, kanC, lgrE, ppsE_1), 1 min at 72°C. Results were compared to GAPDH expression to adjust for total cDNA per sample. To measure primer efficiency, standard curves consisting of cDNA from R. buchneri-infected IRE11 were also added to reaction plates. To account for differences in primer efficiency, relative quantification was calculated with a PCR efficiency correction (85).

All statistical analyses were carried out using GraphPad Prism version 9.1.2. For plate reader experiments, the growth of Rp-mKate in IRE11 cultures infected with other Rickettsia or treated with lysates was analyzed using a two-way ANOVA with Dunnett's multiple-comparison test, using Rp-mKate growth in uninfected IRE11 wells as the control. Relative gene expression was analyzed with a two-way ANOVA with Dunnett's multiple-comparison test, comparing the expression in Rp-challenged groups to that in the unchallenged control group. Statistical significance was assigned when p < 0.05.

Analysis of the sequenced R. buchneri ISO7^T genome revealed that it contains two gene clusters encoding proteins similar to those involved in antibiotic synthesis. Neither gene cluster is present in other members of Rickettsiaceae. The first cluster contains eleven genes including those for polyketide and other non-ribosomal peptide synthesis enzymes (Table 2; Figure 3A). Biosynthesis pathways for these compounds (which include the penicillins, cyclosporin A, vancomycin, and erythromycin) involve large multi-modular enzymes that act as assembly lines for the catalysis of chain elongation and addition of modifications (86). These are usually clustered with additional genes encoding tailoring enzymes that further modify the resulting compound, for example by methylation or cyclization, and/or by releasing it from the assembly line (87, 88). Proteins in the R. buchneri polyketide cluster show similarity to sequences from the Gammaproteobacteria Legionella israelensis, Erwinia amylovora, Pantoea ananatis, and Pectobacterium spp. as well as Cyanobacteria (Supplementary Figure S1; Supplementary Data S1). Interestingly, one of the hypothetical proteins in the cluster appears to be a type IV pilin, whose sequence obtained no blastp hits except those from the two R. buchneri genomes. Only the first three genes in the cluster are conserved in the REIS (Wikel) genome, while the remaining genes are not present (Figure 3A; Table 2; Supplementary Figure S1). Instead, the region of the REIS genome downstream of ppsE_1 encodes numerous tra genes (REIS_1815 = traC, REIS_1814 = traV, REIS_1813 = truncated traB, REIS_1810 = traE). The region upstream of the cluster is homologous in both genomes with a similar arrangement of genes (Figure 3A). Interestingly, the genes following ysdC in the Rb ISO7 genome are homologous to those in a different region of the REIS genome, with the next gene after ysdC, REISMN_1205, identical to REIS_1393 (Figure 3A). Notably the area upstream of REIS_1393 is heavily populated with transposase sequences, suggesting that a transposition event may have led to the loss of the remaining polyketide cluster genes in the REIS genome.

Gillespie et al. (54) also identified an additional polyketide synthase with a putative frameshift mutation (REIS_0330). This prompted an examination of the Rb ISO7 genome for additional polyketide synthase genes. A cluster of three were identified, and these are annotated as pksL_2 (REISMN_07055), pksR (REISMN_07060), and pksN (REISMN_07065). Corresponding genes encoding identical proteins are annotated in the REIS (Wikel) RefSeq on GenBank, with corresponding locus tags REIS_RS15050 (hypothetical protein), REIS_RS10770 (methyltransferase), and REIS_RS01405 (KR domain-containing protein), respectively.

The second cluster contains fifteen genes, with eleven showing similarity to genes involved in aminoglycoside antibiotic synthesis, as well as genes coding for putative antibiotic exporters and an antibiotic resistance factor (Table 3; Figure 3B). This second antibiotic cluster was also identified in the REIS (Wikel) genome (54) and found to be associated with the Rickettsiales-amplified genetic element, RAGE-A. The aminoglycoside antibiotics include streptomycin, kanamycin, and gentamicin and are produced through complex biosynthetic pathways involving many enzymatic reactions; few of these pathways have been fully characterized (89–91). Proteins of the R. buchneri aminoglycoside cluster show similarity to those from antibiotic synthesis gene clusters from Actinobacteria and Firmicutes (54), while the putative multidrug exporter mde/mdlB, transcriptional regulator LuxR, and ABC transporter lagD show greater similarity to proteins from the Gammaproteobacteria (Supplementary Figure S2; Supplementary Data S2). Meanwhile, the nucleotide translocase tlcA1/tlc2 is related to those from other Rickettsia species (54). This cluster is conserved in both R. buchneri genomes (Figure 3B; Table 3), with the majority of sequences identical (Supplementary Figure S2).

To determine whether these genes are actively transcribed by R. buchneri, eight genes from the two antibiotic clusters were selected for RT-PCR analysis (Tables 2, 3). Transcripts for all eight genes were detected, as indicated by bands of the expected sizes observed on gel electrophoresis (Figure 3C). Negative controls and no RT controls showed no products, except for HTH where a less robust band was seen, suggesting some DNA contamination, although there is clearly less amplification than in the RT well. Smaller bands observed in the btrB, lagD, and kanC reactions likely represent nonspecific products.

To determine whether R. buchneri might exhibit antibiosis against other bacteria infecting tick cells, cell-free red fluorescent A. phagocytophilum, R. monacensis, or R. parkeri was used to challenge tick cell cultures containing different levels of green fluorescent R. buchneri. Infectivity was measured 14 days after inoculating serial dilutions of cell-free rickettsiae/Anaplasma into replicated wells of a 24-well tissue culture plate seeded with R. buchneri-infected IRE11 cells. All three pathogens showed markedly reduced ability to infect and replicate in tick cell cultures infected with R. buchneri (Figure 4), and even a low level of infection with Rb-GFPuv led to a reduction in pathogen infection and replication. Fluorescent microscopy revealed that R. monacensis and A. phagocytophilum did not replicate in tick cells infected with R. buchneri (Supplementary Figure S3). Infectivity of R. monacensis, R. parkeri, or A. phagocytophilum in tick cell cultures infected with R. buchneri was reduced by 3–5 orders of magnitude compared to that in cells without R. buchneri (Figure 4), confirming that the presence of R. buchneri was inhibitory to the growth of other intracellular tick-borne bacteria. Furthermore, examining the percentage of infected cells by fluorescent microscopy during Rp-mKate infection showed that in IRE11 without R. buchneri, cells became completely infected at all dilutions over the 14-day period (Figure 4D). In contrast, Rp-mKate infection only reached low levels (<10%) in IRE11 cultures containing the endosymbiont in 45% cells, and almost no cells became infected with Rp-mKate in cultures with >95% cells harboring R. buchneri.

To further investigate the growth dynamics during coinfection, a fluorescent plate reader was used to measure the replication of Rp-mKate in IRE11 cells infected with Rb-GFPuv at levels of 25, 50, 75, and >95%, in comparison to uninfected IRE11. GFPuv measurement could clearly differentiate the various levels of infection with Rb-GFPuv and indicated a steady replication of the endosymbiont over the 14-day experiment (Figure 5A). No increases in GFPuv fluorescence were seen in uninfected IRE11. With a high (1,000:1) challenge, fluorescence from Rp-mKate growth indicated a rapid infection and replication of Rp-mKate in uninfected IRE11 from day 3 which then began to level off from day 6 onward (Figure 5B). In contrast, there were significantly lower rates of mKate fluorescence increase in IRE11 harboring Rb-GFPuv at all levels of infection from day 4 onward (p < 0.0001); compared to uninfected IRE11 at day 14, there was an 89% reduction in Rp-mKate in IRE11 with >95% Rb-GFPuv, 88% reduction in IRE11 with 75% Rb-GFPuv, 84% reduction in IRE11 with 50% Rb-GFPuv, and 76% reduction in IRE11 with 25% Rb-GFPuv. Similarly, 100:1 challenge of uninfected IRE11 with Rp-mKate resulted in a rapid increase in fluorescence from days 5 to 6, reaching a peak by day 12, indicating replication and spread of Rp-mKate in the cells (Figure 5C). However, in IRE11 with Rb-GFPuv, significant differences in mKate fluorescence were observed from day 6 (p < 0.001), and at day 14 the reduction was 99% in 50%, 75% and >95% infected cells, and 95% in IRE11 with 25% Rb-GFPuv. With a low challenge (10:1), mKate fluorescence increased from days 7 to 8 and reached its height at days 13–14 in uninfected IRE11, whereas in Rb-GFPuv-infected IRE11 mKate fluorescence was reduced by 99%-100% in 50%, 75%, and >95% infected cells, and by 98% in 25% infected cells (Figure 5D), with significant differences seen from day 8 (p < 0.0001) compared to IRE11 without Rb-GFPuv. No changes in mKate fluorescence were observed in control wells to which Rp-mKate was not added (data not shown). Together, these results suggest that the presence of R. buchneri in IRE11 has a significant inhibitory effect on the ability of R. parkeri to successfully infect and replicate in the culture.

In order to begin to separate whether the observed inhibition of pathogen growth might be due to antibiosis by R. buchneri or competitive exclusion as observed previously (70–73) between various species of Rickettsia (that do not contain antibiotic synthesis gene clusters), additional plate reader experiments were performed using different Rickettsia species in place of R. buchneri as the resident bacteria. Firstly, this was assessed using the low-pathogenic species R. amblyommatis at infection levels of 25 and >95% in IRE11 cells, and uninfected control IRE11, which were then challenged with Rp-mKate at 1,000:1 (high) and 10:1 (low). In contrast to results obtained with Rb-GFPuv, the presence of R. amblyommatis in tick cells resulted in only partial inhibition of Rp-mKate replication. Growth of Rp-mKate in R. amblyommatis-infected IRE11 was inhibited in a manner that was relative to the level of R. amblyommatis infection (Figures 6A,B); i.e., at the low level of infection (25%), there was a lower inhibition of Rp-mKate (35% inhibition at 10:1 challenge, and 37% at 1,000:1, compared to IRE11 without R. amblyommatis at day 14), while at the high level of infection (>95%) there was a higher inhibition of Rp-mKate (54% inhibition at 10:1 challenge and 56% at 1,000:1).

To investigate whether endosymbiotic rickettsiae may have greater exclusionary effect on pathogenic bacteria than other pathogenic Rickettsia species, as well as to further investigate the contribution of the involvement of potential antibiotic production by R. buchneri, additional plate reader competition assays were conducted using R. peacockii (an endosymbiont of D. andersoni) in place of Rb-GFPuv. Wells contained either uninfected IRE11 or IRE11 infected with R. peacockii-GFPuv at levels of 25, 50, 75, and >95%. The wells were challenged with Rp-mKate at 1,000:1, 100:1, and 10:1. In the high-challenge (1,000:1) wells, mKate fluorescence was significantly lower than that in IRE11 without R. peacockii-GFPuv at day 4 onward in cells infected with >95% R. peacockii-GFPuv (Figure 6C). mKate fluorescence in IRE11 with 75% or 50% R. peacockii-GFPuv was significantly lower from day 8 and day 10 onward, respectively, while growth of Rp-mKate in IRE11 with 25% R. peacockii-GFPuv was not significantly different from IRE11 without the endosymbiont (Figure 6C). At day 14, compared to that in IRE11 without R. peacockii-GFPuv, there was a 31% reduction in mKate fluorescence in IRE11 with 50% R. peacockii-GFPuv, a 38% reduction in IRE11 with 75% R. peacockii-GFPuv, and a 51% reduction in IRE11 with >95% R. peacockii-GFPuv. Similarly, in the 100:1 challenge wells, mKate fluorescence in 50, 75, and >95% R. peacockii-GFPuv-infected IRE11 was significantly different from that in IRE11 without R. peacockii-GFPuv, from day 5 (>95%) or day 6 (50 and 75%) onward, while mKate fluorescence in 25% R. peacockii-GFPuv-infected IRE11 was similar to the control (Figure 6D). At day 14, compared to that in IRE11 without R. peacockii-GFPuv, there were reductions of 35%, 43%, and 43% in mKate fluorescence measured in IRE11 with 50, 75, and >95% R. peacockii-GFPuv, respectively. In the 10:1 challenge experiment, mKate fluorescence was significantly reduced in IRE11 50, 75, and >95% infected with R. peacockii-GFPuv relative to IRE11 without R. peacockii-GFPuv from day 7 onward (Figure 6E). In IRE11 25% infected with R. peacockii-GFPuv, mKate fluorescence was significantly lower than the control from days 11 to 14 (Figure 6E). At day 14, in comparison to that in IRE11 without R. peacockii-GFPuv, mKate fluorescence was reduced by 16, 47, 61, and 55% in IRE11 with 25, 50, 75, and >95% R. peacockii-GFPuv, respectively. Measurement of GFPuv fluorescence could differentiate the different populations of IRE11 infected with 25, 50, 75, and >95% R. peacockii-GFPuv and implied a steady replication of R. peacockii-GFPuv over time (Figure 6F). No increase in GFPuv fluorescence was observed in uninfected IRE11; rather, there was a decline, likely due to lysis of IRE11 by Rp-mKate in infected cells (Figure 6F). Together, these results suggest that neither R. amblyommatis nor R. peacockii possess the ability to inhibit growth of competing R. parkeri to the same extent as R. buchneri.

Further experiments were performed to examine whether the inhibitory effect of R. buchneri on Rp-mKate infection was due to the secretion of antibiotic products. Uninfected IRE11 was added to a 96-well plate then treated with either cell-free Rb-WT, a 1:10 dilution of cell-free Rb-WT, lysate from IRE11 heavily infected with Rb-WT, or lysate from uninfected IRE11. After 2 h, cells were challenged with Rp-mKate. Although there were some significant differences in the growth of Rp-mKate under some of the treatments, these were inconsistent across the three different challenge doses, suggesting that rather they were due to experimental variations, for example in challenge dose or cell density (Figures 7A–C). Taking these data together, there seems to be no obvious effect of any of the treatments on the growth of R. parkeri in comparison to untreated control cells. Furthermore, these results suggest that R. buchneri need to be intracellular to inhibit R. parkeri growth and that there is also no antibiotic effect of extracellular addition of Rb-WT lysate.

An additional experiment was performed to determine whether challenging R. buchneri-infected IRE11 with Rp-mKate would increase the inhibitory activity of lysates against Rp-mKate. Lysates were prepared from cell-free Rb-WT isolated from IRE11 challenged with Rp-mKate for 1–7 days and then used to treat IRE11 cells in a 96-well plate. The wells were then challenged as previously with Rp-mKate at challenge doses of 1,000:1, 100:1, and 10:1, and mKate fluorescence was measured over 14 days and compared to mKate fluorescence in cells treated with lysate from unchallenged Rb-WT-infected IRE11. At 1,000:1 challenge, there was no difference in mKate fluorescence between the control and any of the treatments (Supplementary Figure S5), showing a growth curve similar to that in other plate reader experiments. However, at challenge doses of 100:1 and 10:1, there were significant differences in mKate fluorescence in the wells treated with lysates from Rb-WT challenged with Rp-mKate for 7 days (168 h) in comparison to the control, with 43 and 77% reductions at day 14, respectively (Figures 7D,E). The mKate fluorescence in the other treatment groups was similar to that in the control wells. Unchallenged control wells were also incorporated into the experiment to check the viability of any rickettsiae in the lysate; mKate fluorescence was observed increasing on days 6–8, suggesting that viable Rp-mKate were present in the lysate (Figure 7F). However, their growth in wells treated with 168-h-challenged lysate was also much lower than in other treatment wells. The fact that the mKate fluorescence in the unchallenged control wells peaked later than in challenged wells suggests that the low level of viable Rp-mKate contributed little to the results seen in the Rp-mKate-challenged wells. Overall, these results indicate that after 7 days in the presence of Rp-mKate, lysates from Rb-WT were able to inhibit the replication of Rp-mKate in IRE11 cells at lower doses (100:1 and 10:1), which could be due to increased antibiotic activity at this time point.

To investigate whether the expression of genes from the putative antibiotic clusters of R. buchneri was upregulated in response to the presence of potentially competing bacteria, a time-course experiment was set up and qRT-PCR was used to examine the relative expression of selected genes from each antibiotic cluster in the presence and absence of Rp-mKate during infection. Over a 7-day time course, several of the examined genes appeared to be upregulated in response to multiple days' challenge with Rp-mKate (Figure 8); however, there was wide variation between replicate experiments and the differences were not statistically significant.

The antibiotic activity of R. buchneri against extracellular bacteria was investigated using antibiotic susceptibility assays. Filter paper disks were treated with cell-free Rb-WT, Rb-WT-infected IRE11, and supernatant from IRE11 cultures infected with various levels of Rb-WT (25%, 50%, and >95%). Spectinomycin (10 and 100 μg)-treated disks were used as positive controls, and uninfected IRE11 cells and IRE11 culture supernatant were used as negative controls. Disks were placed onto Mueller–Hinton agar plates streaked with E. coli strain D21 or S. aureus strain MN8. None of the IRE11- or Rb-WT-derived treatments resulted in any inhibition of E. coli or S. aureus growth, whereas spectinomycin-treated disks prevented bacterial growth (Table 4).

Supernatant from IRE11 heavily infected with Rb-WT was also tested for activity against Rp-mKate growing in Vero cells. Addition of the supernatant to Vero cultures prior to infection with Rp-mKate did not inhibit infection; there were no differences observed in the progression of plaque formation and size of plaques in comparison to a duplicate experiment using supernatant from an uninfected IRE11 culture (Supplementary Figure S6).