While cloning pRAM18 and pRAM23 from a genomic library of R. amblyommii AaR/SC, we obtained 15 kbp of a provisional third plasmid ^[4]. We PCR-amplified the remainder of the plasmid using end sequence complementary primers and sequenced the overlapping 18,408 bp amplicon. The third plasmid, pRAM32, was determined to be 31,972 bp in length with a G/C content of 34%. Twenty-two genes or pseudogenes were predicted by comparison of pRAM32 nucleotide or translated sequences with sequences in GenBank using blastn or blastx (NCBI). The degree of congruence between pRAM32 and other known rickettsial plasmids is low. Approximately 18% of its sequence has 74% or higher homology to pRAM18 and 22% of pRAM32 displays 79% or greater homology to pRAM23 and pRF (Rickettsia felis) sequences. Regions of similarity include genes encoding the DnaA-like replication initiator protein, transposases and the conjugal transfer protein TraA_Ti. The pRAM32 dnaA-like gene has homologs on other rickettsial plasmids but its C-terminal domain is separated from the rest of the gene by 2,027 bp of sequence containing a transposase gene and a split transposase gene. Other features of interest in pRAM32 include a patatin-like phospholipase gene implicated in virulence ^[8], a 1,870 bp sequence with homology to a cluster of three transposase genes that recurs six times in the R. felis genome, a traW homolog encoding a type F conjugative transfer system protein and a pseudogene similar to traU encoding a conjugal DNA transfer protein on REIS, the rickettsial endosymbiont of Ixodes scapularis, the black-legged tick.

Purified AaR/SC was embedded in agarose, lysed, divided into three lanes and separated by pulsed-field gel electrophoresis (PFGE) (Figure 1 panels B, D and F). Southern analysis was performed to illustrate the existence of three distinct plasmids by hybridization with one of three digoxigenin-labeled probes: a DNA Invertase gene probe specific for pRAM18, an hsp2 probe specific for pRAM23, or a recD probe specific for pRAM32. The probes hybridized in distinctly different patterns exhibiting three main forms for each of the plasmids (Figure 1 panels C, E and G). The smallest and least abundant plasmid forms were putative linear monomers that were not visible on SYBR Green (Lonza, Rockland, ME) –stained gels but hybridized at approximately 22 kbp, 27 kbp and 34 kbp (indicated in Figure 1 with asterisks on panels C, E and G) relative to linear markers (panel A). The putative supercoiled forms of pRAM18 and pRAM32 co-migrated between the 65 and 70 kbp markers, while the putative pRAM23 supercoiled form migrated between the 55 and 60 kbp markers. This form of the plasmids appeared as a doublet on the SYBR Green-stained pulsed-field gels. Additional plasmid isomers approximately twice the size of the supercoiled forms migrated between the chromosomal and host cell mitochondrial DNAs. An uncharacterized fourth form was present in pRAM32 but absent from pRAM18 and pRAM23 while the abundance of supercoiled pRAM32 was quite low relative to supercoiled pRAM18 and pRAM23.

Real time PCR primers were designed for single copy genes identified on each of the three plasmids. The relative ratio of spoT, pRAM23 parA or pRAM32 parA to chromosome-encoded gltA gave relative copy number values of 2.0, 1.9 and 1.3 for pRAM18, pRAM23 and pRAM32, respectively.

A 27.6 kbp recombinant plasmid pRAM18/Rif/GFPuv (Figure 2A) was used to derive two smaller shuttle vectors pRAM18dRG (Figure 2B) and pRAM18dRGA (Figure 2C). Portions of pRAM32 were used to construct the vector pRAM32dRGA (Figure 2D). All shuttle vectors were constructed to include the parA and dnaA-like genes and intervening sequences, and a selection cassette encoding rifampin resistance and a GFPuv fluorescent marker.

Table S1 lists rickettsiae species that were tested. Electroporation of R. parkeri Oktibbeha and R. bellii RML 369-C with the 27.6 kbp recombinant plasmid pRAM18/Rif/GFPuv (Figure 2A) and subsequent selection with rifampin yielded GFPuv-expressing rickettsiae in two weeks. A pulsed-field gel showed the presence of pRAM18/Rif/GFPuv in the R. bellii transformant, visible as a SYBR Green-stained DNA band migrating slightly below the 50 kbp linear marker (Figure 3A black arrow), and its absence in the wild type R. bellii lane. A Southern blot hybridized with gfp_uv confirmed the presence of pRAM18/Rif/GFPuv plasmid (Figure 3B). A second panel of the same pulsed-field gel demonstrated the presence of multiple plasmids in wild type R. amblyommii sample (Figure 3C, arrowheads), and the Southern blot hybridized with pRAM18-specific DNA Invertase probe showed co-localization of pRAM18 DNA Invertase with gfp_uv (Figure 3D) and identified native pRAM18 (18,422 bp) migrating in the expected pattern (Figure 3D, arrowheads) as compared to recombinant pRAM18/Rif/GFPuv (27,588 bp) (white arrows). PFGE and Southern analysis also confirmed the presence of pRAM18/Rif/GFPuv in transformed R. parkeri (Figure 4A and B; asterisks mark the predominant form of pRAM18/Rif/GFPuv).

In order to facilitate transformation, smaller shuttle vector constructs were derived from pRAM18/Rif/GFPuv (pRAM18dRG and pRAM18dRGA) and pRAM32 (pRAM32dRGA) (Figure 2B, C and D respectively). Plasmid-free R. parkeri Oktibbeha and R. bellii RML 369-C were transformed with pRAM18dRGA and pRAM32dRGA. GFPuv-expressing rickettsiae were detected in R. parkeri after 5 and 7 days, and in R. bellii after 10 and 9 days, respectively. The plasmids were visualized by PFGE (Figure 4A and C), and Southern blots hybridized with a gfp_uv probe (Figure 4B and D) indicated that the majority of the plasmids existed in the putative supercoiled state (wide arrows, arrowheads and asterisks) with very little of the linear monomer present (narrow arrows).

Rickettsia montanensis M5/6, which also lacks a native plasmid, was successfully transformed with all three of the derived constructs, resulting in brightly fluorescent, GFPuv-expressing rickettsiae 20 days after electroporation (Figure 5 panels A, B and C). Analysis of R. montanensis pRAM18dRGA transformants stained with 4′,6-diamidino-2-phenylindole (DAPI) (VECTASHIELD, Vector Laboratories, Burlingame, CA) and examined by epifluorescence microscopy using DAPI or FITC filters showed over 98% congruence between rickettsia positive for DNA and rickettsia positive for GFPuv (Figure 5 panels D, E, and F), indicating that the rickettsial population contains few, if any, non-transformants. Wild type R. montanensis showed the presence of rickettsial DNA with DAPI but displayed no GFPuv fluorescence (data not shown). PFGE (Figure 6A) and Southern analysis (Figure 6B) of the R. montanensis shuttle vector transformants with a digoxigenin-labeled gfp_uv probe demonstrated that the relative positions of the plasmid forms were congruent with the expected sizes (pRAM18dRG: 12,994 bp, pRAM32dRGA: 8,872 bp and pRAM18dRGA: 10,248 bp) (Figure 6A and 6B, white arrows).

Two rickettsial species containing native plasmids were tested with pRAM18dRGA and pRAM32dRGA: R. monacensis and R. amblyommii AaR/SC. Both constructs yielded GFPuv-expressing R. monacensis 28 days after electroporation. A pulsed-field gel comparing wild type and transformed R. monacensis showed the presence of three main forms of the native plasmid, pRM, in the wild type DNA: putative mixed circular and linear dimeric plasmid, supercoiled plasmid, and linear monomeric plasmid (Figure 7A; arrows) ^[2]. These bands were not visible for the transformants on the SYBR green stained pulsed-field gel but a Southern blot hybridized with an hsp2 probe, a gene present on pRM but not on pRAM18, showed that pRM was still present in the R. monacensis transformants populations (Figure 7B, asterisks). Hybridization of the same gel with a gfp_uv-probe demonstrated that pRM co-existed with the shuttle vectors pRAM18dRGA and pRAM32dRGA (Figure 7C, black arrows). Electroporation of R. amblyommii AaR/SC with these constructs did not produce GFPuv-expressing rickettsiae, possibly due to exclusion of the shuttle vector by native plasmids containing identical incompatibility factors (see discussion).

Plasmids were successfully recovered from pRAM18dRGA and pRAM32dRGA-transformed R. montanensis and propagated in Escherichia coli. All endonuclease restriction digests of E. coli-recovered plasmids produced patterns as predicted.

Relative ratios of the shuttle vector-encoded gfp_uv and chromosome-encoded gltA genes were used to estimate copy number of the shuttle vectors in transformed rickettsiae (Table 1). Transformation with pRAM18dRGA resulted in higher copy numbers than did pRAM32dRGA. Although the copy number in rickettsiae transformed with pRAM32dRGA was relatively low, GFPuv-expressing rickettsiae were readily visible by fluorescence microscopy. pRAM18dRG was tested only in R. montanensis where it was maintained at almost the same copy number as pRAM18dRGA. Both the 27.6 kbp full-length plasmid pRAM18/Rif/GFPuv and its derivative pRAM18dRGA (10.2 kbp) yielded similar copy numbers in transformed R. bellii and R. parkeri. However, all electroporations with pRAM18dRGA in R. bellii, R. parkeri, R. monacensis and R. montanensis were successful, whereas pRAM18/Rif/GFPuv yielded transformants in R. parkeri and R. bellii but not in R. montanensis or R. monacensis (Table 1), indicating the derivative was more effective.

The relative ratio of the pRM-encoded hsp2 and chromosome-encoded gltA genes was used to estimate the copy number of native pRM in shuttle vector-transformed R. monacensis (Table 2). The average copy number of pRM in the first subcultures of pRAM18dRGA and pRAM32dRGA-transformed R. monacensis was comparable to the previously published average of 2.7 in wild-type R. monacensis^[4]. The relative copy numbers of pRAM18dRGA ranged from about 3.6 to 5.5 during the course of 15 subcultures, the last 5 of which were done with and without rifampin selection in cultures maintained in parallel, indicating stable maintenance of both the native plasmid and the shuttle vectors during this period.

A multiple cloning site (MCS) containing 7 restriction sites was inserted into pRAM18dRGA yielding pRAM18dRGA[MCS] (Figure 8A). A 948 bp expression cassette encoding mCherry red fluorescent protein under an Anaplasma marginale promoter ^[9], ^[10], ^[11]was inserted into the MCS to test the construct as a functional shuttle vector, yielding pRAM18dRGA[AmTrCh] (Figure 8B). R. montanensis transformants expressed both fluorescent proteins (Figure 8C, D and E), were Southern-blot-positive with digoxigenin-labeled gfp_uv and mCherry probes, and contained intact plasmids that were recovered in E. coli.

Plasmid DNA for shuttle vector construction was obtained from R. amblyommii AaR/SC ^[3], ^[4]. The partial ompA (AF453408), ompB (JN378402), gltA (JN378401), and 17 kDa (JN378400) gene sequences of isolate AaR/SC are available at NCBI. R. amblyommii AaR/SC, R. monacensis (IrR/Munich), R. montanensis (M5/6), R. bellii (RML 369C), and R. parkeri (Oktibbeha) (Table S1) were used to test transformation efficiency of the shuttle vectors. Rickettsiae were grown in Ixodes scapularis cell line ISE6 in L-15B300 medium with NaHCO₃ and HEPES buffer ^[21] in 25-cm² tissue culture flasks. Rickettsiae from heavily infected cultures were purified from host cells as previously described ^[3].

The plasmid pRAM32 was cloned and sequenced as detailed in the Materials and Methods S1. Briefly, a fragment of pRAM32 previously cloned in pJazz (Lucigen, Middleton, WI) ^[4] was sequenced and primers complementary to the clone were used to obtain the remaining portion of the plasmid by PCR amplification of R. amblyommii AaR/SC genomic DNA. The resulting 18,408 bp amplicon was sequenced and annotated manually after submission to the NCBI Prokaryotic Genomes Automatic Annotation Pipeline (PGAAP). The nucleotide sequence of plasmid pRAM32 from R. amblyommii AaR/SC has been deposited in GenBank under the accession number CP002642.

Constructs were derived from two R. amblyommii AaR/SC plasmids: pRAM18 (Figure S1) and pRAM32 (Figure S2). Initially, a 27.6 kbp recombinant plasmid pRAM18/Rif/GFPuv (Figure 2A) was created by inserting a selection cassette into the entire pRAM18 plasmid cloned in the E. coli vector BACv2.0 (Lucigen) (Figure S1C). The selection cassette, derived from pMW1650 (kindly provided by D. H. Wood, University of South Alabama) ^[5], contained the R. prowazekii arr-2 rifampin resistance gene (rpsLp-arr-2_Rp, or Rif) and a gene coding for green fluorescent protein driven by the rickettsial ompA promoter (ompAp-GFPuv, or GFPuv). Smaller shuttle vectors, pRAM18dRG, pRAM18dRGA and pRAM32dRGA (Figure 2B, C and D), were constructed using the same selection cassette and the parA and dnaA-like genes from pRAM18 (Figure S1A) or pRAM32 (Figure S2A). Details are provided in Materials and Methods S1.

A multiple cloning site (MCS) was inserted into pRAM18dRGA yielding pRAM18dRGA[MCS] (Figure 8A). A 948 bp AmTr/mCherry cassette containing the Anaplasma marginale promoter tr^[9], ^[10], ^[11] driving expression of mCherry encoding a red fluorescent protein was inserted into the MCS yielding pRAM18dRGA[AmTrCh] (Figure 8B). Construction of the vectors is described in Materials and Methods S1.

Rickettsiae purified ^[3] from one 25-cm² flask were sufficient for two electroporation trials. Host cell-free rickettsiae were washed 3 times in 250 mM sucrose, resuspended in 250 mM sucrose and held on ice. Endotoxin-free plasmid DNA (1 µg) was placed in a chilled 0.1 cm gap electroporation cuvette (Bio-Rad, Hercules, CA) with 50 µl of rickettsiae in 250 mM sucrose and pulsed once at 1.8 kV, 200 ohms and 25 µF in a Gene Pulser II electroporation system (Bio-Rad). Electroporated rickettsiae were recovered in 1 ml of tick cell culture medium and transferred onto a confluent layer of ISE6 cells in a 25-cm² flask. The volume of medium was brought up to 5 ml and cultures were incubated at 34^oC for 24 h before adding 1–2 µg/ml of rifampin. Growth medium was changed every 3 or 4 days while maintaining continuous rifampin selection. Cultures were monitored weekly for presence of green fluorescent rickettsiae by epifluorescence microscopy using an inverted Nikon Diaphot or an upright Nikon Eclipse E400 (Nikon, Melville, NY) using Sapphire GFP or fluorescein isothiocyanate (FITC) filters, respectively.

Purified wild type and pRAM18dRGA-transformed R. montanensis were affixed to slides by centrifugation using a Cytospin (Thermo Scientific, Pittsburgh, PA), air dried, and stained with VECTASHIELD mounting medium containing DAPI (Vector Laboratories). Slides were examined by epifluorescence microscopy using DAPI or FITC filters to determine the percentage of GFPuv-expressing transformants in the population.

Purified rickettsial pellets from one 5 ml culture were lysed with 300 µl Cell Lysis Solution and genomic DNA was purified with the Puregene Core A kit (Qiagen, Valencia, CA) according to the manufacturer's gram negative bacteria protocol.

224 ng and 177 ng of genomic DNA isolated from pRAM18dRGA and pRAM32dRGA-transformed R. montanensis, respectively, was used to transform E. cloni 10G elite electrocompetent cells as per manufacturer's protocol (Lucigen). Transformed E. coli were selected on YT/Rif/Amp plates, and plasmids isolated from clones were compared to the original plasmid by endonuclease restriction digest with SpeI, EcoRI, HindIII, PstI and XhoI (pRAM18dRGA clones) or SpeI, EcoRI, HindIII, PstI, XbaI and PvuII (pRA32dRGA clones).

Purified rickettsiae were separated on PFGE gels, transferred to Zeta Probe GT genomic membranes and hybridized with digoxigenin-labeled probes ^[2], ^[3] (Materials and Methods S1).

Real-time quantitative PCR (qPCR) and the relative quantification method ^[22], ^[23] were used to determine shuttle vector and native plasmid copy number. Species-specific plasmids were used to generate standard curves with primers designed to target specific single copy genes (Tables S2 and S3) found on the shuttle vectors, native plasmids, and rickettsial chromosomal DNA (Materials and Methods S1).

R. monacensis transformed with pRAM18dRGA was subcultured 1∶100 every 10 days for 15 serial transfers (50 µl of cell suspension from a heavily infected culture inoculated into an uninfected ISE6 culture) with continuous rifampin selection (0.8 µg/ml). At the first, eighth and fifteenth transfer the relative copy numbers of the shuttle vector (pRAM18dRGA) and native plasmid (pRM) was quantified. To determine the stability of pRAM18dRGA in transformed R. monacensis maintained in the absence of antibiotic selection, rifampin was withheld from the medium of a subculture at the tenth serial transfer. After 5 additional serial transfers in rifampin-free medium (15^th in vitro passage) the relative copy numbers of the shuttle vector (pRAM18dRGA) and native plasmid (pRM) were determined.