Efficient Method of Cloning the Obligate Intracellular Bacterium Coxiella burnetii

Abstract

Coxiella burnetii is an obligate intracellular bacterium that replicates in a large lysosome-like parasitophorous vacuole (PV). Current methods of cloning C. burnetii are laborious and technically demanding. We have developed an alternative cloning method that involves excision of individual C. burnetii-laden PVs from infected cell monolayers by micromanipulation. To demonstrate the cloning utility and efficiency of this procedure, we coinfected Vero cells with isogenic variants of the Nine Mile strain of C. burnetii. Coinhabited PVs harboring Nine Mile phase II (NMII) and Nine Mile phase I (NMI) or Nine Mile crazy (NMC) were demonstrated by immunofluorescence. PVs were then randomly excised from cells coinfected with NMI and NMC by micromanipulation, and PVs harboring both strains were identified by PCR. Fresh Vero cells were subsequently infected with organisms from coinhabited PVs, and the PV excision and PCR screening process was repeated. Without exception, PVs obtained from second-round excisions contained clonal populations of either NMII or NMC, demonstrating that micromanipulation is an efficient and reproducible procedure for obtaining C. burnetii clones.

Coxiella burnetii is an obligate intracellular bacterium that is highly infectious for both animals and humans. In humans, C. burnetii causes Q fever, a disease that normally manifests as an acute influenza-like illness. The agent is typically transmitted via contaminated aerosols in association with domestic livestock operations (reviewed in reference 21). In all host cell systems, C. burnetii conducts its infectious cycle and replicates to high numbers in a parasitophorous vacuole (PV) with characteristics of a phagolysosome (36). The PV enlarges concomitant with C. burnetii growth to occupy nearly the entire cytoplasm. At this point, the large and spacious PV is easily visible by light microscopy at a low magnification (e.g., ×200).

The study of C. burnetii is hindered by the many limitations imposed by its obligate intracellular lifestyle (2). In fact, to date the only defined virulence factor of the organism is lipopolysaccharide (LPS). The organism undergoes an LPS phase variation reminiscent of that observed in the Enterobacteriaceae. Transition from a smooth (full-length O-side chains) phase I to a rough (truncated O-side chains) phase II LPS structure occurs upon serial passage of the organism in nonimmunocompetent hosts, such as embryonated eggs or tissue culture cells. Phase II organisms are avirulent for guinea pigs, whereas phase I C. burnetii causes disease and is always associated with naturally infected mammals and ticks (12). Isogenic LPS variants of the Nine Mile strain exist, with the Nine Mile phase I (NMI), Nine Mile crazy (NMC), and Nine Mile phase II (NMII) strains producing full-length, moderately truncated, and severely truncated LPS, respectively (22). NMC and NMII contain nearly the same large chromosomal deletion that eliminates genes involved in O-antigen biosynthesis (16, 35).

The identification of additional C. burnetii virulence factors would be dramatically aided by a system for genetically manipulating the organism. Progress toward this goal was made by Suhan and coworkers (33), who in 1996 successfully transformed C. burnetii to ampicillin resistance by using electroporation. Since then, transformation by electroporation has been described for the obligate intracellular bacterium Anaplasma phagocytophilum (7) and four species of rickettsia: Rickettsia conorii (29), R. prowazekii (26, 27), R. typhi (34), and R. monacensis (3). While an efficient method of allelic exchange for generating isogenic mutants of these organisms has not been described, random transposon mutagenesis has been achieved in Rickettsia spp. (3, 25) and A. phagocytophilum (7) by using transposome (11) and Himar1 (17) transposon technology, respectively.

The increasing success in genetic manipulation of obligate intracellular bacteria highlights the need for efficient methods for cloning transformants for phenotypic analyses. Plaque assays, whereby infected cell monolayers are overlaid with agarose to confine consecutive rounds of bacterial infection and monolayer destruction to a restricted area, are useful in cloning some obligate intracellular bacteria (19, 37). This procedure requires formation of plaques visible to the naked eye (generally greater than 1 mm in diameter) before overall degradation of the monolayer. Plaque assays are most amenable to organisms having active mechanisms of cell-to-cell spread and/or a membrane-lytic method of host cell egress, such as the spotted fever group rickettsia (37). Plaque cloning has been described for C. burnetii; however, the procedure is exceedingly difficult, owing to the nonlytic nature of C. burnetii infection. Small plaques (<1 mm in diameter) form only in primary chicken embryo fibroblast following a lengthy, 16-day incubation (24, 37). Indeed, only two isogenic plaque clones of the Nine Mile isolate are described in the literature (1, 14). A variation of the plaque assay was recently used to clone nonlytic Chlamydia pneumoniae (9). In this study, infectious foci that form under an agarose overlay were identified by fluorescence microscopy and their positions marked, and then clones were picked with a sterile swab. Clones of C. burnetii (33) and other intracellular bacteria (18, 25) have also been derived by the tedious and time-consuming method of end-point-limiting dilution.

Phenotypic evaluation of genetic transformants and strain variants of C. burnetii exhibiting different biological properties would be aided by an efficient method for cloning the organism. However, current protocols for obtaining clonal populations of C. burnetii are laborious and technically challenging. Here, we describe an effective means of cloning C. burnetii that relies on excision by micromanipulation of individual PVs that are easily visible by phase-contrast light microscopy. Following excision, organisms within single PVs can be genotyped using DNA amplification methods and used to infect fresh cells for clonal expansion.

MATERIALS AND METHODS

C. burnetii.

C. burnetii NMI (RSA493), NMII (RSA439), NMC (RSA514), and Australian QD (RSA425) (Aus) variants were propagated in African green monkey kidney (Vero) fibroblasts (CCL-81; American Type Culture Collection) grown in RPMI medium (Invitrogen, Carlsbad, CA) supplemented with 2% fetal bovine serum (FBS). Organisms were purified from infected cells by Renografin density gradient centrifugation as previously described (13).

Generation of monoclonal and polyclonal antibodies.

Monoclonal antibodies (MAb) were produced from hybridomas resulting from fusion of spleen cells from female BALB/c mice and SP2/0 myeloma cells. The mice were inoculated with purified, formalin-killed NMII C. burnetii mixed with the monophosphoryl lipid A-trehalose dicorynomycolate emulsion adjuvant system (Sigma-Aldrich, St. Louis, MO) and boosted according to the manufacturer's protocol. MAb reactivity was assessed by immunoblotting. Briefly, C. burnetii NMII cell lysates were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to an Immobilon-P membrane (Millipore, Bedford, MA). The membrane was placed in a Miniblotter 25 multichannel screening apparatus (Immunetics, Cambridge, MA) and incubated with undiluted hybridoma culture supernatants. Bound antibodies were detected using horseradish peroxidase-conjugated anti-mouse immunoglobulin G (IgG; Pierce, Rockford, IL) and enhanced chemiluminescence (Amersham, Cleveland, Ohio). For generation of polyclonal rabbit sera directed against C. burnetii NMI, formalin-fixed organisms were mixed with monophosphoryl lipid A-trehalose dicorynomycolate-cell wall skeleton emulsion adjuvant (Sigma-Aldrich) and used to immunize New Zealand White rabbits. (All animal protocols described in this article were approved by the Rocky Mountain Laboratories Institutional Animal Care and Use Committee.)

LPS extraction and immunoblot analysis.

C. burnetii LPS was extracted by a modified hot phenol method that exploits its hydrophilic nature (14). Briefly, 1 mg (dry weight) of purified C. burnetii was suspended in 1 ml of 50% phenol. The sample was boiled for 10 min, incubated for 5 min on ice, and then centrifuged at 14,000 × g for 5 min. The aqueous phase was collected and the extraction repeated on the pellet. The aqueous phases from both extractions were pooled (approximately 1 ml) and vacuum dried overnight. The resulting pellet was dissolved in 100 μl of distilled H₂O, the sample dried again, and the pellet dissolved in 50 μl of SDS-PAGE sample buffer. Immunoblotting was conducted as described above.

Indirect immunofluorescence of singly and coinfected Vero cells.

For immunofluorescence labeling, Vero cells on 12-mm glass coverslips in a 24-well tissue culture plate were infected with C. burnetii, incubated for 4 or 5 days, and then fixed with 100% methanol for 5 min. NMII was selectively stained in singly infected cells by labeling cells with undiluted MAb A6 hybridoma culture supernatant followed by anti-mouse Alexa Fluor 594 IgG (Molecular Probes, Eugene, OR). All C. burnetii strains were then labeled with a 1:1,000 dilution of serum from a guinea pig infected with the C. burnetii Ohio strain (phase I) (RSA270), followed by anti-guinea pig Alexa Fluor 488 (Molecular Probes). Coinfection of Vero cells with NMII, NMC, or NMI was conducted at multiplicities of infection (MOI) of approximately 1, 10, or 100, respectively, to obtain high percentages of cells containing coinhabited PVs. MOI was based on C. burnetii genome equivalents as previously described (6). NMII was selectively stained by immunofluorescence as described above. NMI or NMC was selectively labeled with a 1:2,500 dilution of rabbit antiserum generated against formalin-fixed NMI, followed by anti-rabbit Alexa Fluor 488 IgG (Molecular Probes). Epifluorescent images were obtained using a Nikon TE-2000 E inverted microscope equipped with a CoolSNAP HQ digital camera (Roper Scientific, Tuscon, AZ). Confocal fluorescence microscopy was conducted with a modified Perkin Elmer UltraView spinning disc confocal system connected to a Nikon Eclipse TE-2000 S microscope equipped with a Photometrics Cascade:512F digital camera (Roper Scientific). Epifluorescent and confocal images were acquired using Metamorph software (Universal Imaging, Dowingtown, PA). All images were processed using ImageJ software (written by W. S. Rasband at the U.S. National Institutes of Health, Bethesda, MD, and available from http://rsb.info.nih.gov/ij/) and Adobe Photoshop (Adobe Systems, San Jose, CA).

Excision of PVs from Vero cells coinfected with NMII and NMC via micromanipulation.

Infections were conducted using Vero cells cultivated on 12-mm gridded (Eppendorf, Westbury, NY) or nongridded glass coverslips in a 24-well tissue culture plate. Cells were coinfected with NMII and NMC at MOI of approximately 1 and 10, respectively, which resulted in a high percentage of PVs harboring both variants. At 4 or 5 days postinfection, coverslips were transferred to a 90-cm-diameter petri dish containing approximately 10 ml of RPMI medium. Individual PVs were extracted from infected monolayers by using a Micromanipulator 5171 (Eppendorf) mounted to the stage of a Ziess Axiovert 25 inverted microscope, a CellTram Vario manual microinjector (Eppendorf), and a TransferTip(MDS) having a 20-μm tip diameter (Eppendorf). Individual PVs were harvested at ×320 magnification, and the TransferTip(MDS) was changed between PVs. The harvested content from individual PVs was transferred to a 1.5-ml microcentrifuge tube containing 40 μl of K-36 buffer (0.1 M KCl, 0.015 M NaCl, 0.05 M K₂HPO₄, 0.05 M KH₂PO₄, pH 7.0). An 8-μl aliquot was removed for PCR analysis (see below), and the remaining 32 μl was frozen at −80°C.

WGA of PV DNA.

Whole-genome amplification (WGA) was conducted directly on C. burnetii PV suspensions without prior DNA purification, using a GenomiPhi DNA amplification kit (GE Healthcare, Piscataway, NJ). Specifically, 1 μl of C. burnetii vacuole suspension was added to 9 μl of a sample buffer, which was then heated to 95°C for 3 min and cooled to 4°C on ice. Nine microliters of a reaction buffer and 1 μl of an enzyme mix were then added to the cooled DNA samples. The samples were incubated at 30°C for 18 h, followed by inactivation of the Phi29 DNA polymerase by heat treatment at 65°C for 10 min. The samples were cooled on ice and stored at −20°C. Eight separate WGA reactions were conducted on each C. burnetii vacuole suspension. The respective reactions were pooled, and DNA was purified using n-butanol. Briefly, 40 μl of sterile distilled H₂O and 1 ml of n-butanol were added to each sample, and the mixture was vortexed for 5 s. WGA DNA was pelleted by centrifugation at 12,000 × g for 30 min, air dried, and suspended in 80 μl of distilled H₂O.

PCR amplification of NMII- and NMC-specific PCR products.

The design of PCR primers for amplification of NMII- and NMC-specific products was based on the GenBank sequence (AF387640) deposited by Hoover et al. (16) that indicates the position of unique large chromosomal deletions present in NMII and NMC. An 825-bp NMII-specific PCR product was generated using the primers NMIIF (5′-ACTTTAGATAATTCCAGCATCAATG-3′) and NMIIR (5′-GTAAATAGATGCTTACTGCAATATTAGG-3′), which are specific to CBU0678 and CBU0698, respectively. A 401-bp NMC-specific product was generated using the primers NMCF (5′-ACCAAACGATTTGATCGTATATTAG-3′) and NMR (5′-CCATCTCAAAAGCACTAATTGCC-3′), which are specific to CBU0676 and CBU0700, respectively (31). Individual NMII- and NMC-specific PCRs were conducted on 4 μl of WGA DNA from each sample by using Accuprime Taq DNA polymerase (Invitrogen). PCRs were then analyzed by gel electrophoresis using 0.8% agarose gels.

Second-round infection of Vero cells with NMII and NMC obtained from coinhabited PVs.

Vero cells were infected with C. burnetii from excised PVs identified by PCR as harboring both NMII and NMC. The remaining PV content (32 μl) was mixed with 200 μl of RPMI plus 2% FBS and then added directly to Vero cells grown to approximately 80% confluence on coverslips in individual wells of a 24-well tissue culture plate. (Prior to infection, Vero cells were exposed to 15,000 rad at a rate of 295 rad/min in a ¹³⁷Cs gamma irradiator to inhibit further cell division.) To facilitate infections, plates were centrifuged at 900 × g for 30 min (23). Following centrifugation, 1 ml of RPMI plus 2% FBS was added to each well and the cells were incubated for 5 days to allow PV development. PVs were then excised by micromanipulation and their contents analyzed by PCR as described above.

RESULTS

Generation of MAb against NMII LPS.

Hybridomas were prepared from mice immunized with fixed NMII C. burnetii. Supernatant from one hybridoma culture (A6) reacted by immunoblotting against a component in NMII whole-cell lysates having an approximate molecular mass of 4,000 Da. Because the molecular mass of phase II LPS is roughly 4,000 Da (14), we investigated whether MAb A6 was specific for this molecule. As depicted in Fig. 1A, MAb A6 reacted by immunoblotting with truncated phase II NMII LPS but not intermediate-length NMC or full-length NMI LPS. Interestingly, MAb A6 also did not react with phase II LPS of the Australian QD isolate of C. burnetii, indicating that this LPS moiety is structurally distinct from that of NMII. Immunoblotting results were confirmed by immunofluorescence staining of infected cells whereby only NMII was stained by MAb A6 (Fig. 1B).

FIG. 1.

MAb A6 reacts against NMII LPS. (A) LPS was extracted from C. burnetii NMI, NMC, NMII, and Australian QD (Aus) isolates by the hot phenol water method (14). Samples were separated by SDS-PAGE and stained with silver (left) or transferred to an Immobilon-P membrane for immunoblotting (right). Only LPS from NMII reacted with MAb A6. The relative sizes of molecular mass markers in kDa are shown on the left of each panel. (B) Vero cells were infected with NMI, NMC, NMII, or Aus for 5 days, fixed with methanol, and dual stained for immunofluorescence with MAb A6 (red) and serum from an infected guinea pig (green). Guinea pig serum stained all four isolates, whereas MAb A6 stained only NMII, as indicted by the yellow color produced by overlapping green and red fluorescence. Bar, 15 μm.

NMII replicates in the same PVs as NMI or NMC C. burnetii.

As a model system for testing whether PV extraction can be used to clone C. burnetii, we devised an infection strategy that results in PVs containing two genetically distinguishable strains. The specific reactivity of MAb A6 against NMII allowed microscopic assessment of cohabitation within the same PVs of NMII and NMC or NMI in coinfected cells. To achieve a high percentage of coinhabited PVs, we infected cells with higher MOI of NMC and NMI to account for their low uptake efficiencies relative to that of NMII (22) (see Materials and Methods). At 4 d postinfection, cells were fixed and NMII was selectively stained for immunofluorescence microscopy with MAb A6, and NMC or NMI was selectively stained with a dilution of rabbit polyclonal sera that does not stain NMII. The absence of NMII staining by diluted polyclonal sera is attributable to the immunodominance of the LPS O antigen, which NMII lacks (14). In all coinfected cells, NMII was observed in the same vacuole as NMC or NMI (Fig. 2B and data not shown). A common phenotype of coinhabited vacuoles was a core of NMII surrounded by a ring of NMI or NMC (Fig. 2B and data not shown).

FIG. 2.

NMI and NMII C. burnetii bacteria coinhabit the same vacuole in Vero cells. Vero cells were infected with NMI and NMII C. burnetii bacteria at MOI of 100 and 1, respectively. Cells were fixed with methanol at 4 days postinfection and NMI (green) and NMII (red) selectively stained by indirect immunofluorescence as described in Materials and Methods. (A) Low-magnification epifluorescence micrograph showing singly (arrowheads) and coinfected (arrows) cells. Bar, 15 μm. (B) High-magnification confocal fluorescence micrograph showing the characteristic “jelly doughnut” appearance of coinhabited vacuoles with NMI forming an outer ring around an inner core of NMII. Bar, 2 μm.

Excision of C. burnetii by micromanipulation.

In Vero cells, C. burnetii PVs are easily visible by phase-contrast light microscopy concomitant with onset of the organism's exponential growth phase (approximately 2 d postinfection) (6). As shown in Fig. 3A, PVs at 4 d postinfection can be precisely excised while leaving behind the remainder of the host cell, including the nucleus. C. burnetii could also be extracted from larger PVs in such a way that the vacuole reseals after extraction to presumably allow continued growth of the remaining organisms (Fig. 3B).

FIG. 3.

Individual C. burnetii PVs can be harvested using micromanipulation. (A) A phase-translucent PV is easily visible by phase-contrast light microscopy at 4 d postinfection in panel 1 (arrow). Panels 2 and 3 depict the PV excision process. (B) PVs can reseal following extraction of vacuole contents. Vero cells on gridded coverslips were infected with NMII C. burnetii for 4 d. Panels 1, 2, and 3 are a sequential series of micrographs depicting the extraction of C. burnetii from an isolated PV (arrow). Panel 4 depicts the same PV 7 h later, which has resealed and is structurally intact. Remaining phase-dense C. burnetii bacteria are visible in the upper portion of the vacuole. (Slight movement of the PV and accompanying cell occurred during the 7-h incubation.) Bars, 15 μm.

Derivation of clonal populations by micromanipulation.

We used our coinfection procedure to demonstrate that clonal populations of C. burnetii can be obtained by micromanipulation. Conveniently, the novel large chromosomal deletion of genes involved in LPS O-antigen biosynthesis found in NMII and NMC allows easy differentiation of these strains from one another and from the wild-type NMI strain (16). Vero cells were coinfected with NMII and NMC at MOI that result in a high percentage of PVs harboring both strains. At 5 d postinfection, nine PVs were randomly excised and their contents suspended in 40 μl of buffer. The 5-day incubation period was selected to optimize recovery of C. burnetii bacteria which are in the late log phase of their growth cycle (6). Longer incubations resulted in overgrown Vero cell monolayers containing PVs that were less phase translucent, making identification of distinct PVs more difficult. DNA from 8 μl of each PV suspension was whole genome amplified and then subjected to PCR using an amplification strategy that specifically identifies NMC and NMII C. burnetii bacteria (Fig. 4A). As depicted in Fig. 4A, two PVs each contained just NMC or NMII, representing clonal populations that were derived from a mixed infection. The remaining five PVs contained both strains. Based on the staining intensities of NMC- and NMII-specific amplicons, it was evident that coinhabited PVs contained roughly equal numbers of each strain.

FIG. 4.

Cloning of C. burnetii by micromanipulation. Vero cells were coinfected with NMII and NMC C. burnetii bacteria at MOI of 1 and 10, respectively. At 5 d postinfection, nine PVs were randomly excised and their contents suspended in 40 μl of K-36 buffer. DNA from 8 μl of the PV suspension was whole genome amplified. Samples were then subjected to PCR according to the strategy depicted in panel A to determine whether the sample contained NMC (PCR product of 401 bp), NMII (PCR product of 825 bp), or both (PCR products of 401 and 825 bp, respectively). PVs 1 and 2 contained NMII, PVs 3 and 4 contained NMC, and PVs 5 through 9 contained both NMC and NMII (B). The remaining 32-μl suspension of PVs 5 through 9, containing mixtures of NMC and NMII, was used to reinfect separate Vero monolayers. At 5 d postinfection, three PVs were randomly excised from each infected culture and the samples reanalyzed by PCR as described above. All PVs obtained from the second-round infection contained clonal populations of NMC or NMII.

To demonstrate that a clonal C. burnetii population could be derived from a mixed population originally harbored within a single PV, individual Vero cell monolayers on coverslips were infected with the remaining 32 μl of the five PV suspensions that contained both NMC and NMII. To aid in detection of the small number of vacuoles resulting from the low MOI, C. burnetii bacteria were centrifuged onto Vero cell monolayers that had been irradiated to stop cell division. Infection with PV contents consistently yielded 5 to 10 PVs that were easily visible by phase light microscopy at 5 d postinfection (data not shown). Three PVs were randomly excised from each infected culture and the samples reanalyzed by PCR as described above. All PVs obtained from this second-round infection contained clonal populations of NMC or NMII (Fig. 4B). Nine of 15 clones recovered were NMII, which may reflect the higher infectivity of this strain for Vero cells than NMC (22).

DISCUSSION

Here, we describe a method for cloning C. burnetii that exploits the phase-translucent nature of the large PVs inhabited by the pathogen in cultured cells. In our hands, an experienced user can excise one PV roughly every 3 min, with the majority of time between PV excisions required to install a fresh TransferTip(MDS). Thus, 100 independent clones can easily be obtained in a single day. A variation of this procedure was recently described by Gieffers et al. (10), who used micromanipulation not only to clone C. pneumoniae but also to rid the bacteria of mycoplasma contamination.

For second-round infections, the use of irradiated Vero cell monolayers improved our ability to find PVs by light microscopy. Irradiation maintained the flat, spread-out character of monolayers, and PVs that develop in these cells over a 5-day incubation period are more easily discernible than those that develop in an expanding monolayer. In place of irradiation, chemicals that inhibit host cell division without affecting C. burnetii growth might be used. Alternatively, use of a highly permissive cell line that displays better contact inhibition than Vero cells, such as human embryonic lung fibroblasts (28), might be an option.

At first glance, the reinfection efficiency of PV contents appears low, with usually 10 or fewer second-round PVs easily identified by light microscopy. However, this number is consistent with known infectivity and growth rates of C. burnetii. One-step growth curves in Vero cells show that, following a 2-day lag phase, there is an approximately 100-fold increase in NMII genome equivalents during an exponential phase lasting about 4 days (6). With C. burnetii's estimated doubling time of roughly 10 h (6), a 5-d-postinfection PV (approximately seven doubling periods) on average should contain about 60 to 70 organisms. With NMII, we routinely achieve ratios of C. burnetii genome equivalents to infectious foci-forming units of approximately 4 (data not shown), with higher ratios for NMC, which is less infectious for Vero cells than NMII (22). Despite the low number of second-round PVs, in every case we successfully located C. burnetii vacuoles and identified their contents as either clonal NMII or NMC. The fact that all second-round PVs contained clonal populations suggests that PVs of low-MOI primary infections already contain clones of C. burnetii that could be directly expanded. Expansion would be aided by biasing the excision of PVs to larger vacuoles that likely contain more C. burnetii bacteria (5, 30). This is especially true of virulent phase I C. burnetii, which is less infectious for cultured cells than avirulent phase II organisms (22). Clonal expansion could be also accomplished by second-round infection of embryonated hen's eggs or guinea pigs whereby the infectious dose of virulent C. burnetii is fewer than 10 organisms (22, 23).

C. burnetii derived from individual PVs can be easily genotyped prior to expansion of the clone. Genotyping can involve a single locus, as demonstrated here in our analysis of the C. burnetii LPS O-antigen biosynthesis region, or the entire genome. Indeed, we have previously shown that DNA microarrays are found to have identical hybridization profiles when probed with WGA DNA of PV-derived C. burnetii or DNA extracted from purified organisms (4). Identical results are also observed when these samples are used to probe oligonucleotide-based tiled arrays designed to detect single-nucleotide polymorphisms (unpublished observations), thereby demonstrating the high fidelity of the WGA process. We found that the contents of large PVs can be harvested such that the vacuole reseals to presumably support continued C. burnetii growth. Thus, it may be possible to temporally analyze C. burnetii populations harbored by a single PV. The mechanical stability of the C. burnetii PV relative to that of other pathogen PVs was shown in previous microinjection experiments (15).

In coinfected cells, we never observed NMII and NMI or NMC in different vacuoles. This result contrasts with a report proposing differential trafficking of these phase variants to vacuoles at different stages of endolysosomal maturation (8). Based on microscopic observations and PCR results, these strains appear to replicate at roughly equal rates in the same vacuole. NMII is hydrophobic due to the lack of O-antigen carbohydrates (38). This hydrophobicity is reflected in the strain's enhanced adherence to, and more efficient uptake by, host cells relative to that of NMC or NMI (38). While hydrophilic NMI organisms display pronounced and uniform Brownian motion within the PV, NMII tends to form stationary-phase-dense masses. Moreover, in fixed cells viewed by fluorescence microscopy, NMII organisms are not randomly dispersed with NMI or NMC in coinhabited PVs. Instead, the strain variants reside in distinct regions of the vacuole, with a common phenotype being a core of NMII organisms surrounded by a ring of NMI or NMC organisms.

With modifications, this cloning procedure could be applied to other obligate intracellular bacteria that infect monolayer-forming host cells, even those that do not produce large vacuoles. For example, the typhus group rickettsia R. prowazekii is exceedingly difficult to plaque (37). Late in the organism's infectious cycle (approximately 4 d postinfection), a microcolony consisting of >200 organisms fills the host cytosol (32). Despite the large rickettsial burden, infected cells are difficult to identify by light microscopy. Identification could be aided by staining live cells with a nontoxic fluorescent vital stain that is incorporated by intracellular rickettsia, such as rhodamine 123 (20). Infected cells containing clonal populations could be located by fluorescence microscopy and then the organisms harvested by micromanipulation under light microscopy.