Knockout of Babesia bovis rad51 ortholog and its complementation by expression from the BbACc3 artificial chromosome platform

Babesia bovis establishes persistent infections of long duration in cattle, despite the development of effective anti-disease immunity. One mechanism used by the parasite to achieve persistence is rapid antigenic variation of the VESA1 cytoadhesion ligand through segmental gene conversion (SGC), a phenomenon thought to be a form of homologous recombination (HR). To begin investigation of the enzymatic basis for SGC we initially identified and knocked out the Bbrad51 gene encoding the B. bovis Rad51 ortholog. BbRad51 was found to be non-essential for in vitro growth of asexual-stage parasites. However, its loss resulted in hypersensitivity to methylmethane sulfonate (MMS) and an apparent defect in HR. This defect rendered attempts to complement the knockout phenotype by reinsertion of the Bbrad51 gene into the genome unsuccessful. To circumvent this difficulty, we constructed an artificial chromosome, BbACc3, into which the complete Bbrad51 locus was inserted, for expression of BbRad51 under regulation by autologous elements. Maintenance of BbACc3 makes use of centromeric sequences from chromosome 3 and telomeric ends from chromosome 1 of the B. bovis C9.1 line. A selection cassette employing human dihydrofolate reductase enables recovery of transformants by selection with pyrimethamine. We demonstrate that the BbACc3 platform is stably maintained once established, assembles nucleosomes to form native chromatin, and expands in telomere length over time. Significantly, the MMS-sensitivity phenotype observed in the absence of Bbrad51 was successfully complemented at essentially normal levels. We provide cautionary evidence, however, that in HR-competent parasites BbACc3 can recombine with native chromosomes, potentially resulting in crossover. We propose that, under certain circumstances this platform can provide a useful alternative for the genetic manipulation of this group of parasites, particularly when regulated gene expression under the control of autologous elements may be important.


Introduction
Babesiosis is a tick-borne disease caused by apicomplexan parasites of the genus Babesia. Humans are not the natural host for any babesial parasite but may be an incidental host, acquiring zoonotic infections with a variety of different species. Babesia microti is the most common species of Babesia to infect humans, although in western Europe infections commonly occur with Babesia divergens. In the U.S. infections have been observed with Babesia duncani and B. divergens-like organisms, as well as the unspeciated WA1 and MO1 isolates (reviewed in [1]). Many individuals may carry asymptomatic infections [2], including as a result of inadequate drug treatment of acute parasitemia [3,4], posing a serious risk to the blood supply [5]. In cattle babesiosis may be caused by at least five different species, with Babesia bovis generally considered the most virulent. B. bovis shares many parallels with the human malarial parasite, Plasmodium falciparum, including immune evasion via cytoadhesion and antigenic variation, and the capacity for development of a lethal cerebral disease [6].
Bovine babesiosis caused by B. bovis is quite severe. Cattle that survive the acute disease develop a strong anti-disease immunity, but remain persistently infected for periods of at least several years. This parasite makes use of at least two mechanisms to effect persistence: (i) cytoadhesion of infected red blood cells (iRBCs) to capillary and post-capillary venous endothelium, presumably in order to avoid splenic clearance; and (ii) rapid antigenic variation of the cytoadhesion ligand, VESA1, to avoid antibody-mediated forced re-entry into the circulation [7]. Cytoadhesion is therefore a behavior that is immunologically sensitive, and which the parasite correspondingly has evolved means to protect. The development of the ability to abrogate antigenic variation would allow for ready elimination of the parasite; this capability is a desirable goal for parasite control through disruption of an essential aspect of its biology.
There is a great need to understand the molecular underpinnings of different virulence mechanisms in parasites. One informative way to facilitate such studies is through the ability to genetically manipulate the organisms. To date, the tools developed for the genetic manipulation of B. bovis are limited and remain most appropriate for use in characterizing individual targets, although there is a need for high-throughput methodologies [8]. The expression of exogenous genes has been demonstrated, using babesial promoter sequences to drive transcription of the target sequences. This has been achieved transiently through transfection with circular plasmids [9][10][11][12][13]. The longevity of this class of vector has been improved by the inclusion of a centromeric segment, improving the segregation of the circular molecules [12]. Long-term, stable expression has been achieved by single-crossover insertional mutagenesis of chromosomal loci [12,14]. This approach demonstrates the capacity for creating targeted gene disruptions and the potential for the use of promoter trapping strategies. Significantly, doublecrossover knockout of the thioredoxin peroxidase (tpx) gene has been achieved, providing complete gene knockouts [12]. Complementation of tpx was achieved by re-integration of the gene ectopically into the EF1α locus, resulting in high level expression. However, ectopic expression of exogenous genes in this manner may suffer from improper expression due to local chromatin modifications, including silencing or significant overexpression, compromising interpretation of genetic contributions to phenotype. Despite the available tools there remains a need for convenient platforms that are stable, efficiently and accurately segregated during mitosis, allow the transcription of large sequences, and which may be properly regulated relative to metabolic need. Artificial chromosomes represent a potentially useful platform in which genes could be presented in a chromosome-like context, including assembly as chromatin and even the ability to adopt localized higher-order DNA structure when appropriate, something that is unlikely in circular molecules. Moreover, artificial chromosomes might reasonably be expected to be replicated in proper stoichiometry with the natural chromosomes, and to segregate properly during binary fission.
Antigenic variation in B. bovis is known to heavily utilize segmental gene conversion (SGC) to modify the VESA1 ligand [15], and may participate in in situ transcriptional switching (isTS) as well. During SGC sequences within a gene are replaced in short segments with sequences duplicated from elsewhere in the genome. Over time, genes that are so modified, such as the ves genes encoding the VESA1 ligand of B. bovis [6] or the VSG genes of Trypanosoma spp. [16,17], become mosaics comprised of segments of many different genes in the family. SGC is thought to be a form of homologous recombination (HR), much the same as gene conversion, a process that is dependent upon the activities of Rad51 proteins [18][19][20][21]. To initiate investigation of the enzymatic underpinnings of antigenic variation by SGC in B. bovis, we describe here the knockout of the B. bovis rad51 gene by double crossover replacement [22]. Given the potential for adverse effects if expressed at inappropriate levels, we did not want to overexpress Bbrad51 during complementation. Moreover, because of the knockout of the Bbrad51 gene, homologous recombination was lost and complementation of the knockout by reintroduction of the gene into the genome was not possible. We have addressed this issue through artificial chromosome technology, an approach that allows the complementation of disrupted genes by reintroduction of the entire locus, providing gene control through its autologous regulatory elements in a chromatin setting. An artificial chromosome platform has already been developed for Plasmodium berghei, enabling transformation efficient enough for shotgun screening to recover genes associated with drug resistance [23,24]. We have created an analogous platform for use with B. bovis through the assembly of an artificial chromosome, BbACc3. BbACc3 is comprised of B. bovis centromeric and telomeric ends in a modified pBluescript-KS(+) backbone. This construct employs the human dihydrofolate reductase (hDHFR) gene as selectable marker, conferring resistance to pyrimethamine. BbACc3 was constructed and is manipulated as the circular plasmid, pBACc3, which is linearized for transfection. Linearization with PmeI removes a spacer sequence and exposes legitimate telomeric ends. Here, we demonstrate that BbACc3 confers pyrimethamine resistance to B. bovis, is maintained as a distinct linear chromosome that assembles into chromatin, and undergoes expansion of its telomeres. Significantly, the Bbrad51 gene expressed from BbACc3 under regulation by its autologous elements provided normal levels of complementation of the Bbrad51 knockout upon challenge with methylmethane sulfonate (MMS). Although further optimization is desirable, this platform may facilitate long-term complementation of essential genes for studies connecting genotype and phenotype, particularly where improper regulation would be problematic.

The orthologous Bbrad51 gene was identified
The B. bovis rad51 gene (Bbrad51) was identified by first searching the genus Babesia nonredundant proteome (S1 Table). The search was initiated with the Saccharomyces cerevisiae Rad51 sequence (CAA45563) as bait, using BLAST, with default parameters [25]. In this initial search, proteins XP_001609877 (E = 2e-103) and XP_001609660 (E = 4e-15) were identified as candidates. To cast a wider net and ensure capture of all candidates, CAA45563 was again used to initiate an iterative search of the genus Babesia non-redundant protein database, using Psi-BLAST [26,27] with default parameters [26]. After three iterations the top B. bovis candidate, XP_001609877, had a significance score of E = 3e-156, and three additional proteins of the Rad51/DMC1/RadA superfamily were also identified. Each of these proteins then was used individually in reciprocal iterative searches of the NCBI non-redundant protein database, excluding the genus Babesia, using Delta-BLAST and composition-based statistics to provide their presumptive identities [26,27]. These proteins included the second-best fit, XP_001609660 (E = 3e-86; XRCC3-like), XP_001610815 (E = 8e-16; Rad51 homolog 2-like), and XP_001609995 (E = 0.077; XRCC2-like). Genes encoding putative orthologs of all four proteins are found among various other piroplasms (S1 Table; adapted from PiroplasmaDB). For XP_001609877 this search yielded over 200 significant hits of E = 0.0, all of which were confirmed or putative eukaryotic Rad51 molecules sharing 47-49% identity. Alignment of the four candidates with a number of confirmed or predicted Rad51 proteins demonstrated excellent alignment of all functional domains and motifs (S1 Fig). An alignment tree was created by the Neighbor-joining method, using Jukes-Cantor protein distance measures and 100 bootstraps to assess robustness. The tree shows that, among the apicomplexan parasites Rad51 proteins cluster in apparent clades (Fig 1A, colored blocks), suggesting shared selection pressures within individual groups. P. knowlesi was the only exception, failing to cluster with other Plasmodium spp. XP_001609877 clusters with other eukaryotic Rad51 proteins, forming an apparent clade with other piroplasmids, yet is clearly quite different. By contrast, the other three B. bovis RecA/RadA/Rad51-related proteins cluster more closely with the Escherichia coli RecA protein included as outlier. An analysis of XP_001609877 for probable structural domains by the Conserved Domains Database yielded specific hits of PTZ00035 ("Rad51 protein, provisional"; residues 6-343; E = 0e+00) and cd01123 ("Rad51_DMC1_radA, P-loop NTPase superfamily"; residues 105-339; E = 3.33e-111), as well as a lower-confidence hit in the N-terminus for pfam14520, a small "helix-hairpin-helix" domain (residues 39-83; E = 2.02e-03). The pfam14520 "helix-hairpin-helix" domain, which is common to Rad51 proteins, was lacking from the other three candidates. Further support for the putative assignment of this protein as a legitimate Rad51 ortholog was obtained by ab initio prediction of its three-dimensional structure, using Robetta software (http://robetta.bakerlab.org/) [28,29]. The predicted structure models were superimposed onto the crystal structure of the S. cerevisiae Rad51 H352Y mutant protein (PDB accession 3LDA) [30], using the Chimera Matchmaker algorithm [31]. The predicted three-dimensional structure of XP_001609877 (model 4) was found to coordinate spatially extremely well with the 3LDA crystal structure, including Walker A and B motifs, residue Q298 regulating access of ATP to the active site [30], and the interaction domain sequences pivotal to filament formation ( Fig 1B). Based on these results we refer to XP_001609877 as BbRad51, and BBOV_II003540 as the Bbrad51 gene encoding this protein. The Bbrad51 gene was annotated with two predicted introns [32], a structural model that was confirmed by RT-PCR amplification, cloning, and sequencing of the full-length transcript. Flanking sequences were recovered through 5´-and 3´-RACE reactions (see Methods). The results exactly matched the published B. bovis T2Bo isolate genome sequence, including predicted start and stop codons, and intron boundaries (Fig 2A; [22,32]). Active transcription of Bbrad51 by asexual stages has been reported previously [33], and was confirmed here by RT-PCR (S2 Fig, panel A).
For brevity, CE11Δrad51 ko1 H5 is hereafter referred to as " ko1 H5". The structure of the modified Bbrad51 locus and loss or gain of Bbrad51 or gfp-bsd transcription, respectively, was confirmed by diagnostic PCR, RT-PCR, and Southern blotting, confirming the replacement of Bbrad51 with the gfp-bsd selection cassette from pRad51KO ( Fig 2B; S2 and S3 Figs). PCRmediated cloning and sequencing of the modified locus confirmed that error-free doublecrossover homologous recombination (HR) had occurred. Two additional independent knockout clonal lines, CE11Δrad51 ko2 and CE11Δrad51 ko3 (for brevity, referred to as "ko2" and "ko3", respectively), were established approximately one year later. These two lines were confirmed by diagnostic PCR and sequencing of the insertion sites (not shown). Because of the lag in obtaining the latter two lines, tests of phenotype initially were performed with ko1 H5 and subsequently confirmed with the latter two. As the results were generated in different experiments they are presented separately.

Bbrad51 knockout had no effects on growth or morphology
Knockout of Bbrad51 had no observable effects on parasite morphology at the level of light microscopy (not shown). Despite the expected significance of Rad51 function to DNA repair and genome stability, no significant differences were observed in the growth rates of wild type and Bbrad51 knockout lines (Fig 3). The growth data shown was obtained with a SybrGreenbased assay of cumulative DNA content to avoid the subjectivity associated with microscopic assays, although microscopic assessments of the percentage of parasitized erythrocytes on Giemsa-stained smears of cultured parasites confirmed these results (not shown). Growth was followed for 48 hours only, to avoid any artifactual differences caused by a need for media changes or cell manipulations. However, as these parasites are not developmentally synchronous in their growth, and have an 8-10 hour asexual cell cycle [34], this period represents approximately 5-6 complete cell cycles. Mean differences in growth rate of as little as 2% per generation would result in larger differences than were observed. Thus, within the limits of sensitivity of our assay no differences in growth rates were observed between wild type and knockout lines.

Assessment of MMS-sensitivity
One common means of assessing contributions of a protein to DNA repair is to determine sensitivity to DNA damage deliberately induced by chemical insult. In this study, we employed exposure to methylmethane sulfonate (MMS), a chemical which alkylates guanines to 7-methylguanine and adenines to 3-methyladenine [35]. To apply that approach in this situation, sensitivity of B. bovis CE11 to MMS first was titrated over a range from 0 to 2000 μM, employing a strategy of exposure to MMS for 90 minutes, with a buffer washout of the MMS, followed by growth for 72 hours (S4 Fig). Initially, we used a microscopic approach to follow parasite growth, to avoid any concerns about MMS effects on erythrocytes or residual DNA fluorescence from dead parasites. The results ranged from little or no effect at � 250 μM to complete killing by exposure to 2000 μM MMS. B. bovis CE11 and ko1 H5 lines then were compared routinely over a range from 125-1000 μM. B. bovis ko1 H5 was significantly more susceptible to DNA damage at 250 or 500 μM MMS than CE11 wild-type parasites (Fig 4). In the CE11 line there was little or no effect seen at 250 μM. B. bovis CE11 parasites exposed to 500 μM MMS showed a severe reduction in growth. However, by 48h the rate of growth had increased significantly, approaching the growth rate of the 0 μM controls between the 48 and 72-hour time-points. In contrast, ko1 H5 parasites were significantly reduced in growth at 250 μM, and 500 μM-treated parasites failed to regain a normal growth rate even by 72 hours post-treatment. The differences in parasitemia are likely due to initiation of the samples at slightly different ppe values. These results indicate that BbRad51 plays a role in recovery from DNA damage due to nucleotide alkylation [36][37][38][39].

Re-integration of Bbrad51 gene
To confirm the contribution of BbRad51 to this phenotype, we attempted to complement B. bovis ko1 H5 parasites by replacement of the gfp-bsd selection cassette integrated at the Bbrad51 locus with the vector, pBbrad51wt_comp. pBbrad51wt_comp was designed to integrate by double crossover HR (S5 Fig), replacing the knockout construct with the intact Bbrad51 wt gene, along with a selection cassette expressing the hDHFR gene [40]. CE11 wild-type parasites also were transformed with this vector to control for any effects on parasite viability or sensitivity to MMS caused by manipulation of the Bbrad51 locus. Importantly, successful complementation by double crossover HR would have resulted in identical Bbrad51 locus structures in both CE11 and ko1 H5 parasites. CE11 parasites incorporated this vector on the first attempt. A similar vector, differing only by the addition of enhanced green fluorescent protein (EGFP) sequences to the 5´end of the Bbrad51 open reading frame, also successfully integrated into the CE11 parasite genome without error in two of two attempts, although each crossed over 3t o the EGFP sequences and failed to incorporate the fluorescent tag. However, HR readily occurred at this locus. In contrast, ko1 H5 parasites could not be complemented in this manner in 11 transformation attempts, 7 performed with linearized vector (favoring double crossover homologous recombination) and 4 with circular vector (favoring single-site integration of the entire plasmid). No integration of the transfection plasmid occurred at the Bbrad51 locus through the intended double-crossover HR, or by single-site integration at the Bbrad51 locus or ectopically at any other site. Our inability to achieve integration via double-crossover HR in ko1 H5 suggests that HR is significantly impaired in the absence of BbRad51, consistent with BbRad51 playing a role in HR-based DNA repair. When tested for MMS sensitivity, CE11/ pBbrad51wt_comp parasites were indistinguishable from CE11 wild-type (Fig 4), indicating that modification of the Bbrad51 locus in this way had not affected the ability of the parasite to repair DNA damage caused by MMS. While we recognize that CE11/pBbrad51wt_comp does not represent the complementation of the defect on a ko1 H5 background that we had attempted to achieve, we suggest that this represents the probable outcome of successful complementation of the knockout at the Bbrad51 locus.

Identification of putative B. bovis centromeres
Because we were unable to complement the knockout phenotype through re-integration of the Bbrad51 gene into the genome, we chose instead to attempt its stable expression from an episomal location. Prior work with an artificial chromosome platform in P. berghei [23] suggested this could be a productive approach for expression of exogenous genes in an HR-compromised parasite line. To create one for B. bovis we first had to recover native B. bovis centromeric and telomeric sequences. Putative B. bovis centromeres were identified by focusing on key features common to centromeres in other organisms, including very high percent A+T content, internal repetitive sequences, and a size large enough to mediate attachment of a kinetochore [41,42]. In this study we searched the genome for A+T-rich regions that are � 2.5 S.D. below the mean G+C content, using a computational window of 1000 bp and length cutoff of � 2 Kbp. Each region so identified was further analyzed by dotplot analysis for internal repetitive structure. Four regions fitting these criteria were identified in the B. bovis C9.1 line nuclear genome [43], one per region corresponding to the four B. bovis T2Bo chromosomes [32]. The internal repeat structures of the B. bovis T2Bo chromosome 2 centromere and of the C9.1 line chromosome 3 centromere were confirmed by dotplot analyses (S6 Fig). This was done by plotting each sequence against itself, using a 21 nucleotide sliding window (not shown). Each candidate matched well with the predicted centromeric regions of the B. bovis T2Bo genome [32], and used by Kawazu and coworkers in constructing the circular centromere-containing plasmid vector, pDHFR-gfp-Bbcent2 [12]. We elected to use the putative centromere from the B. bovis C9.1 line chromosome 3 due to its shorter length of 2.3 Kbp, as compared with 3.5-5 Kbp of the other three, and to ensure full compatibility with derivatives of the B. bovis Mexico isolate, such as the C9.1 and CE11 lines.

Stability of BbACc3
Parasites transformed with BbACc3 ( Fig 5) could be selected with 10 μM pyrimethamine, the approximate IC 90 for this drug, after approximately three weeks. Parasite growth was somewhat slower than non-transformed parasites, with parasites typically reaching a maximum PPE of approximately 3%. To assess stability of BbACc3, established parasites that had been maintained under pyrimethamine selection for 49 days were split into two. One culture was maintained in the presence of 2 μM pyrimethamine, whereas the second was grown in the absence of drug for 99 days. Each then was grown for 72h in the presence of a range of pyrimethamine concentrations to measure effective IC 50  BbACc3 that had been grown in the absence of drug pressure was indistinguishable from wild type CE11 parasites (IC 50 = 2.7 μM). These results demonstrate that significant expression of the exogenous gene, hDHFR, occurs from BbACc3, resulting in a ten-fold increase in IC 50 for the population. The drop to wild type levels of drug resistance in parasites grown in the absence of continued selection suggests that either silencing of the hDHFR marker or loss of BbACc3 occurred. There seems to be a fitness cost associated with maintenance of BbACc3 and derivatives that could result in its loss over time without persistent drug pressure. Evidence of this was also observed in complementation experiments. We did not pursue definitive identification of the cause of this fitness cost, but it is clear that when maintained under drug pressure BbACc3 was maintained and that the hDHFR marker remained actively expressed.

Adaptation of BbAcC3 derivatives over time
For BbAcC3 and modified derivatives to be used successfully for many purposes would require that this construct behave like a real chromosome, including assembly into chromatin and expansion of telomeric ends for stability. To test for such adaptation over time, we performed partial micrococcal nuclease (MNase) digestion of isolated chromatin. Digestion products were probed by staining with SybrGreen to detect total chromatin, and by Southern blotting to detect BbAcC3_Bbrad51-derived sequences. BbAcC3_Bbrad51wt sequences provided a ladder of digestion products with a nucleosomal periodicity of approximately 157-158 bp, matching that of total native chromatin ( Fig 7A) and previously reported values [44], indicative of assembly on nucleosomes. Southern blot probing of isolated plasmid or genomic DNAs, either uncut or cut with informative restriction endonucleases, revealed the expansion of both ends of established BbAcC3 and BbAcC3_Bbrad51 chromosomes (Fig 7B). We interpret this to reflect expansion of the telomeric repeats at each end due to (i) the ability to amplify internal fragments of each pBAC derivative after establishment, with no evidence of internal amplification or rearrangements of sequences (S7 Fig), and (ii) precedence in other systems, such as Trypanosoma brucei [45]. Assessment of the retention of BbACc3_Bbrad51wt under selection pressure was accomplished by quantitative PCR. Primer pairs recognizing the 5´and 3´end of the Bbrad51 gene yielded a ratio of 0.94 ± 0.01 5´ends to 2.88 ± 0.13 3´ends (S3 Table). Accounting for the 3´end sequences present in the artificial chromosome yields a ratio of 1.0 BbACc3_Bbrad51wt per 2.06 native genomes. This was supported by copy number measurements of hDHFR (representing artificial chromosomes) relative to mean copy numbers of the single-copy genes, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and thioredoxin peroxidase-1 (TPX-1) [46]. Interestingly, this same assay provided evidence that BbACc3 is maintained at essentially one copy per parasite (S3 Table), as would be expected. These results support the likelihood of a fitness cost in the retention of the BbACc3_Bbrad51wt chromosome, and indicates a need for further optimization to overcome this.

Complementation of Bbrad51 knockout phenotype
Because we were unable to complement the loss of Bbrad51 by knock-in of the gene via double crossover HR, we attempted complementation with the entire Bbrad51 locus provided via the BbACc3 platform. CE11 wild type and ko1 H5 knockout parasites were transformed by transfection with the artificial chromosome constructs, BbACc3 (as negative control) and BbACc3_Bbrad51wt. pBbACc3_Bbrad51wt is built upon pBbACc3 but also contains the entire uninterrupted Bbrad51 locus, including promoter and 3´-termination sequences to facilitate proper regulation (Fig 5B). ko1 H5/ BbAC_Bbrad51wt parasites displayed reduced sensitivity to MMS compared with ko1 H5 or ko1 H5/ BbACc3 parasites, that was not significantly different from wild-type. While not quite reaching wild type levels, this result indicates successful complementation by providing the Bbrad51 locus on the BbACc3 platform. The contribution of ko1 H5/ BbBACc3_Bbrad51wt is especially clear when compared with ko1 H5/ BbACc3 parasites, which appear to be somewhat less robust overall than ko1 H5 (Fig 8).

Discussion
With increasing human population density and environmental encroachment, as well as climate change, many zoonotic diseases are emerging and/or expanding in range, including babesiosis [47][48][49]. It is thus important that we understand well the biology of these parasites, and prepare to defend against their expansion. Key to understanding the biology of parasites at the molecular level is the ability to manipulate them genetically, something which is not yet well developed for babesial parasites. We have been especially interested in the mechanisms of immune evasion employed by B. bovis as a potential "Achilles heel" of this parasite which, if compromised, could lead to its control during infection [50]. At least two mechanisms are used to evade an ongoing immune response: cytoadhesion in the deep vasculature and rapid antigenic variation of the cytoadhesion ligand [6]. Previously, we have shown that antigenic variation relies in large part upon segmental gene conversion (SGC) for modification of the expressed member of the ves gene family [15]. With SGC assumed to be a form of homologous Asterisks demonstrate correspondence among the nucleosomal patterns generated from the samples, which form a ladder with 156-159 bp spacing between bands, as previously reported [44]. (center panel) Southern blot of the same gel as in the left panel, probed with β-lactamase sequences to detect the pBluescript backbone of BbACc3_Bbrad51wt. (right panel) The same blot stripped and re-probed with Bbrad51 coding sequences to detect endogenous Bbrad51 (CE11), both BbACc3_Bbrad51wt-provided and non-deleted 3´sequences ( ko1 H5/Br51), and remaining non-deleted sequences recombination, a process thought to be dependent upon the activities of Rad51 and related proteins [18][19][20][21]51], we knocked out the Bbrad51 gene to begin an assessment of its contribution to this phenomenon. Experience in other systems lends credence to this idea. For example, in Trypanosoma spp., which also uses SGC to form mosaic isoforms of VSG proteins [16,17], knockout of TbRad51 or TbRad51-3 expression reduced the overall rate of antigenic variation dramatically [52,53]. Here, we demonstrate that a lack of BbRad51 led to hypersensitivity to MMS, an alkylating agent that methylates adenosine bases [54]. The methyl adducts can from the 3´end of Bbrad51 ( ko1 H5). B. Southern blot probing of plasmids and genomic DNAs from B. bovis lines with Bbrad51 coding sequences (left panel). The same blot was stripped and re-probed with β-lactamase sequences (right panel). Overall telomeric lengthening of BbACc3_Bbrad51wt by 2-3 Kbp can be seen by comparing lanes 4 and 7 (both probes, arrowheads). Telomeric lengthening of 1-1.5 Kbp at each end of BbACc3_Bbrad51wt can be observed with either probe (compare lanes 11 and 13; arrowheads). Extreme lengthening of the "left" half of BbACc3 is seen (compare lanes 3 and 6, or 10 and 12, asterisks) when using the β-lactamase probe, perhaps due to crossover with a native chromosome within the telomeric repeats on that end. The 3 Kbp bands detected in lanes 9-11 with the Bbrad51 probe are due to hybridization with the remaining 3´terminal sequences that were not deleted from the endogenous Bbrad51 gene upon integration of the pRAD51KO plasmid. Lanes are: (1) B. bovis CE11 wt gDNA; (2) B. bovis ko1 H5 gDNA; (3) ko1 H5/ BbACc3 gDNA (uncut); (4) ko1 H5/ BbACc3_Bbrad51 gDNA (uncut); (5) CE11/ pBbrad51wt_comp gDNA (uncut); (6) pBbACc3 plasmid (+ PmeI); (7) pBbACc3_Bbrad51wt plasmid (+ PmeI); (8) CE11 gDNA (+ BamHI); (9) ko1 H5 gDNA (+ BamHI); (10) ko1 H5/ BbACc3 gDNA (+ BamHI); (11) ko1 H5/ BbACc3_Bbrad51wt gDNA (+ BamHI); (12) pBbACc3 plasmid (+ BamHI, PmeI); (13) pBbACc3_Bbrad51wt plasmid (+ BamHI, PmeI); (14) CE11 gDNA (+ EcoRI); (15) CE11/ pBbrad51wt_comp gDNA (+ EcoRI). Numbers to the sides of images refer to the positions of double-stranded DNA band size standards, in Kbp.
https://doi.org/10.1371/journal.pone.0215882.g007 bovis parasites of the CE11 (wild type), ko1 H5 (Bbrad51 knockout), ko1 H5/BbACc3_Bbrad51wt, and ko1 Ht/BbACc3 lines were exposed to varying concentrations of MMS for 90 minutes, followed by drug washout. Parasites then were placed back into culture for 72h, and growth assessed by the SybrGreen method. Growth characteristics were plotted as the growth of each line relative to itself, employing 0 μM as maximal growth and 1000 μM MMS as no growth. Asterisks indicate a difference from the CE11 sample, based upon a one-way ANOVA, using CE11 as control group (significance values, p = 0.05 at 125 μM, < 0.001 at 250 μM MMS). The presence of BbACc3_Bbrad51wt complemented ko1 H5 sensitivity to 250 μM MMS to a level that was not significantly different from wild-type parasites. The presence of BbACc3 sequences, however, appears to impact somewhat the overall robustness of parasites, consistent with a fitness cost in retention of the chromosome.
We confirmed in two ways that the MMS-hypersensitive phenotype observed in the Bbrad51 knockout lines was due to the loss of BbRad51 and not some artifactual secondary effect caused by modification of the genome. First, we integrated pBbrad51wt_comp into the Bbrad51 locus and demonstrated that the phenotype was identical to wt upon exposure to MMS. Secondly, we complemented the phenotype in parasites lacking the Bbrad51 gene. This was initially attempted with pBbrad51wt_comp, which was designed to replace sequences associated with pBbrad51ko and reinsert the Bbrad51 gene back into the genome at its native locus, where it would be under control of its own promoter and terminator elements. However, in 11 attempts, eight performed with linearized vector and three with circular plasmid, no successful transformations were achieved, whereas three out of three attempts were successful in wild-type B. bovis CE11 parasites, using linearized vector. We interpret this difference as support for the pivotal importance of Rad51 proteins to HR, including in B. bovis. However, while supportive, these results also left the Bbrad51 gene without complementation to provide formal proof of its involvement. Therefore, we sought to achieve complementation by another means.
In recent years, progress has been made in the ability to genetically modify B. bovis, with the development of methods for transient and stable transfection with episomally-maintained plasmids [9,11,14,[55][56][57], and by targeted integration via single- [12,58] or double-crossover recombination [22,56]. Although each approach is very useful, each also has its own limitations. Transient transfection provides a short-term phenotype only, does not affect the entire parasite population, and may not accurately reflect gene regulation. Segregation of episomes during mitosis is inaccurate, resulting in a heterogeneous parasite population and plasmid loss from the population over time. Such heterogeneity could compromise the ability to interpret complementation of a phenotype that is not dramatic, such as the MMS-sensitivity phenotype of Bbrad51 knockout parasites. This problem was partially solved by the inclusion of centromeric sequences into the plasmid which dramatically improved segregation and resulted in sufficient stability for a plasmid to be maintained in the absence of drug selection for two months [12]. However, a circular construct retains the potential to integrate by single-crossover through sequence elements within the plasmid, potentially compromising ongoing studies. We sought to address this issue through the construction of a linear mini-chromosome, BbACc3, containing centromeric and subtelomeric sequences, and telomeric repeats. BbACc3 is manipulated as a plasmid (pBACc3) in E. coli for modification and production. It is then linearized for transfection into B. bovis, where it is maintained as a linear mini-chromosome with legitimate telomeric ends. Once established, BbACc3_Bbrad51 was found to assemble with nucleosomes into chromatin (Fig 7A), and to undergo expansion of the telomeric repeat regions (Fig 7B). These behaviors suggest that, once established, the BbACc3 platform behaves in a manner consistent with a native chromosome. Here, BbACc3_Bbrad51, carrying the Bbrad51 gene and its native flanking 5´and 3´regulatory sequences, was demonstrated to complement the lack of BbRad51 in knockout parasites with regard to MMS-sensitivity. Moreover, complementation occurred at levels not significantly different from wild-type, strongly supporting the notion that BbRad51 contributes to DNA repair and consistent with a possible role in segmental gene conversion. The ability to express sequences in a regulated fashion opens the possibility of using BbACc3 as a platform to complement knockout of essential genes, or to perform studies of promoter structure and functional interactions. Moreover, constructs carrying exogenous genes could be used to engender transmission-blocking immunity by including vector proteins, to immunize against other co-transmitted parasites by inclusion of appropriate vaccine genes, or other similar applications.
In the course of these studies we found it useful to construct and employ an artificial chromosome, BbACc3, as a platform for the reintroduction of the Bbrad51 gene into B. bovis. This allowed us to circumvent the need for homologous recombination to reintegrate Bbrad51 sequences into the genome in order to acquire a uniform population. Significantly, BbACc3_Bbrad51wt enabled the complementation of the Bbrad51 knockout phenotype at near-normal levels in response to MMS challenge. Once established, the BbACc3 platform behaved like a native chromosome, forming chromatin and lengthening its telomeres. Despite these positive attributes, a cautionary note must also be provided. Southern blot results suggest that within CE11 wild-type parasites, the major parasite population to grow up during selection appears to have undergone a crossover event between BbACc3 and one of the native chromosomes (Fig 7B). This appears to have occurred through the short subtelomeric sequence or telomeric repeats at the "left" end near the β-lactamase coding sequences, resulting in significant expansion of that end of BbACc3, and an overall much longer artificial chromosome than in ko1 H5/BbACc3_Bbrad51wt, even after expansion of its telomeric repeats. An event of this nature would always be a potential hazard in any HR-competent parasite line. A useful improvement of BbACc3 could be made through the inclusion of replication origin sequences that might improve the ease of initial establishment of the chromosome, which in these experiments was no faster than selection of double crossover integration mutants. Thus, while artificial chromosome technology holds considerable promise for certain applications and was invaluable in this study, it will require optimization to meet its full potential, particularly with regard to retention of the artificial chromosome by individual parasites. Regardless, BbACc3 provides a new option in the toolkit for genetic manipulation of B. bovis and perhaps other babesial parasites where its centromeric and regulatory sequences may function, and will be made freely available to the scientific community.
From these studies we conclude that Bbrad51 encodes the B. bovis Rad51 ortholog, and that this protein is non-essential to the parasite for in vitro survival and growth. It remains to be determined whether the parasite could similarly survive during infection of a host without BbRad51. Further, we provide evidence that the encoded protein is involved in recovery from DNA damage, consistent with known functions of Rad51 proteins in other systems [59]. However, any connection of BbRad51 with SGC specifically remains to be established.

Parasites and culture conditions
B. bovis parasites of the CE11 clonal line were used in experiments [7]. Parasites were grown under microaerophilous stationary phase culture conditions, with an atmosphere of 90:5:5 nitrogen: oxygen: carbon dioxide [60,61]. Defibrinated bovine blood products were obtained locally from Holstein cows (approved by University of Florida Institutional Animal Care and Use Committee, protocol #201102216) or commercially (Hemostat Laboratories; Dixon, CA). Parasite sensitivity to pyrimethamine was determined as described [62]. Transformed parasites were maintained in medium containing 2.0 μM pyrimethamine.

Validation of Bbrad51 gene structure
B. bovis gDNAs were isolated as described [66,67], but with prior ammonium chloride lysis of the erythrocytes [68]. Alternatively, QIAmp Mini Spin Columns (Qiagen) were sometimes used, following manufacturer's instructions. For RNA extractions, cultures were grown in erythrocytes from which host leukocytes were removed by filtration through Whatman CF-11 cellulose (GE Healthcare Life Sciences; Pittsburgh, PA) [69]. When needed, cultures were grown to elevated levels of percent parasitized erythrocytes by "dilution enrichment" [70]. Packed erythrocytes were emulsified with either TRIzol reagent (Invitrogen; Waltham, MA) or RiboZol reagent (Amresco; Solon, OH), extracted twice with chloroform and precipitated from 2-propanol. A 2244 bp segment containing the 1134 bp Bbrad51 gene and its 5´and 3í ntergenic regions was amplified from B. bovis C9.1 gDNA by polymerase chain reaction (PCR) with Phusion High-Fidelity DNA Polymerase (New England BioLabs; Beverley, MA) using primers EAM6 and EAM18. This segment was cloned into pCR2.1-TOPO-TA (Invitrogen) to create plasmid pBbRad51. Bbrad51 structure was confirmed by PCR amplification of the 1134 bp gene within the 2244 bp amplicon, using flanking primers EAM6 and EAM18. The amplicon was cloned into pCR2.1-TOPO-TA (Invitrogen) and sequenced through the University of Florida Sanger Sequencing Core by primer-walking. To characterize Bbrad51 transcripts, cDNA was made from B. bovis C9.1 line total RNA, using M-MuLV Reverse Transcriptase (New England Biolabs; Beverley, MA) and oligo-d(T) primer. Full-length coding sequences were obtained by PCR amplification of cDNA with primers EAM50 and EAM51. The amplicon was cloned into pCR2.1-TOPO-TA and sequenced. To obtain noncoding transcript sequences, RNA Ligase-mediated rapid amplification of cDNA ends (RLM-RACE) was performed, using the FirstChoice RLM-RACE kit (Ambion; Waltham, MA). Tobacco acid pyrophosphatase was used to remove the 5´-cap structure for adapter addition, and reverse transcription using M-MuLV and random decamers. Nested PCR amplification of 5´-untranslated sequences was performed using primers EAM2 and 5´-RACE outer, and EAM4 and 5´-RACE inner primers. Products were cloned into pCR-TO-PO-Blunt and sequenced. The 3´-untranslated sequences were obtained by addition of the 3´-RACE adapter, and nested PCR amplification with primers EAM1 and 3´-RACE outer, and EAM3 and 3´-RACE inner. Products were directly cloned into pCR-TOPO-Blunt and sequenced.

Bbrad51 knockout plasmid assembly
The Bbrad51 gene was knocked out by a double crossover homologous recombination strategy [71], with plasmid pRad51KO (Fig 2A). Plasmid pRad51KO was designed to replace all but the 3´-terminal 388 bp of Bbrad51 coding sequences with a selectable gfp-bsd marker (a gift from C.E. Suarez). This region was retained because of the close proximity (320 bp) of Bbrad51 to the gene downstream encoding a conserved mechanosensitive channel-like protein, XP_001609876. To assemble pRad51KO, the EF1α gene "B" hemi-promoter for driving the bsd selectable marker was PCR-amplified from B. bovis CE11 line gDNA with primers EAM42 and EAM43. Bbrad51 3´-end targeting sequences were amplified from plasmid pBbRAD51, using primers EAM49 and EAM56. The EF1α-B segment and Bbrad51 3´targeting sequences were combined by crossover PCR [72] with EAM43 and EAM56 to form fragment 3´-rad51-EF1α-B. The 92 bp B. bovis β-tubulin terminator region (T3) was amplified from plasmid pDS-bsd [11], using primers EAM45 and EAM46. The gfp-bsd fusion gene from plasmid pGFP-Bsd [14], was combined with the T3 fragment by crossover PCR, using primers EAM44 and EAM46 and including the T3 fragment in the reaction. The Bbrad51 upstream targeting sequence was added to the gfp-bsd-T3 fragment by crossover PCR with primers EAM44 and EAM47, and including pBbRad51 in the reaction, to form fragment gfp-bsd-T3-5´-rad51. The two fragments were combined by crossover PCR, using primers EAM56 and EAM47 and inserted into pBluescript II-KS(+). A single missense mutation was identified within the bsd sequence. This was corrected by PCR-amplification of the two halves of the insert with EAM61 and EAM47, and EAM62 and EAM56. The two fragments were fused by crossover PCR, using only EAM47 and EAM56, and reinserted into pBluescript II-KS(+). The corrected plasmid, pRAD51KO, was transformed into Escherichia coli DH5α. Proper construction was confirmed by full sequencing, and the corrected plasmid used for subsequent plasmid production.

Validation of Bbrad51 knock-out
Parasite populations recovered initially were screened directly with the Phusion Blood Direct PCR Kit (Thermo Scientific), using primers EAM75 and EAM77 within the coding sequences of the immediately flanking genes. Non-mutated parasites generated a band of 2.4 Kbp, whereas a proper knockout yielded a band of 3.4 Kbp. RT-PCR with primers EAM50 and EAM51 confirmed the lack of Bbrad51 transcripts. Southern blots were performed as described [15]. DNA fragments, alkaline-transferred to nylon membranes (Amersham Hybond-N + , GE Healthcare), were cross-linked with 50 mJ ultraviolet in a GS Gene Linker UV chamber (BioRad; Hercules, CA). Oligonucleotide probes were end-labeled with γ-[ 32 P] ATP, using polynucleotide kinase.

Parasite growth assays
Parasite growth was assayed by counting Giemsa-stained smears, or by the use of a SybrGreen assay of parasite DNA. For microscopic assays, three biological replicates were established per condition. In titration experiments, cells were diluted 1/10 into a 10% packed cell volume suspension of untreated uninfected erythrocytes, and placed back into culture. Smears were made at 0, 48, and 72 hours post-treatment (approximately 0, 3, and 6 cell cycles [62]) for counting, establishing 125 μM-1000 μM as the informative range (S4 Fig). After validation of effect, growth characteristics following drug treatment were similarly performed microscopically, with sample collections at 0, 24, 48, and 72 hours post-treatment (Fig 4). Subsequent experiments were performed using a SybrGreen method, essentially as described [63,64], on parasites grown in bovine erythrocytes depleted of leukocytes [58]. SybrGreen assays were initiated with starting values of 0.2% parasitized erythrocytes, four biological replicates per condition, and terminal collection at 72 hours only. For measuring sensitivity to pyrimethamine, parasites were grown in complete medium containing serial 10-fold dilutions of pyrimethamine for 72 hours, at concentrations from 0 to 100 μM. For experiments involving MMS, parasites were exposed at 2% ppe to various concentrations of MMS (diluted in complete medium) for 90 minutes at room temperature, then were washed two times at 6000 x g.min at room temperature with 1x VYM buffer [65] to remove MMS. Parasites then were resuspended in complete medium, and placed under normal culture conditions. All experiments were performed at least three times to confirm reproducibility. Data from single time-point SybrGreen-based assays were plotted as values normalized relative to the range of control DNA values, calculated as (mean of the 0 μM control fluorescence-the value of each data point fluorescence) / mean of the 0 μM fluorescence-mean of the maximal drug concentration samples fluorescence). IC 50 values were estimated from the plots, based upon drug concentrations at which values would be expected to be 50% that of the total range.

Assembly of complementation constructs
Two different complementation strategies were used in this project. (i) In the first strategy, plasmid pBbrad51wt_comp (S5 Fig) was engineered to replace all sequences associated with the knockout plasmid by double crossover homologous recombination. The goal was to replace them with the wild-type Bbrad51 locus, concomitant with integration of a human dihydrofolate reductase (hDHFR) expression cassette, to enable selection of transformants with pyrimethamine [62,73]. pBbrad51wt_comp was made by PCR-amplifying Bbrad51 3t erminator sequences from B. bovis C9.1 line gDNA with primers DA255 + DA256. The hDHFR selection cassette was moved from pBACc3 by amplification with primers DA254 + DA253. pBluescript-SK(+) was linearized with XhoI + HindIII restriction endonucleases, and the three sequences were assembled simultaneously, using InFusion reagents (Clontech). The Bbrad51 locus, including complete promoter through immediate 3´sequences, was amplified from B. bovis C9.1 line gDNA with primers DA248 + DA252. The Bbrad51 locus was inserted, using InFusion reagents, into the intermediate construct that had been opened with XbaI + HindIII. (ii) The second strategy was to reintroduce the Bbrad51 locus in the context of the BbACc3 artificial chromosome, which did not require reintegration into the genome but allowed expression controlled by autologous regulatory elements. Construction of pBACc3 (circular plasmid form) was begun with destruction of the NotI site (nt 669-676) within the multiple cloning site (mcs) of pBluescript-KS(+), by a fill-in reaction with T4 DNA polymerase and blunt ligation [74]. This product was partially digested with RcaI, and amplified by inverse-PCR with Phusion Hotstart II polymerase (New England Biolabs; Beverley, MA), using primers DA162 and DA163 [75].  [32], whereas the 5´halves overlapped with Anaplasma marginale genomic sequences. A 483 bp stuffer fragment, corresponding to nucleotides 145014-145482 of the A. marginale Florida isolate genome (accession no. NC_012026), then was amplified with primers DA152 and DA153, and a two-step crossover PCR approach was used to assemble all three amplicons into one fragment containing PmeI restriction sites flanking the stuffer [72]. The fused amplicon was inserted into the RcaI site, using InFusion reagents (Clontech Laboratories, Inc.; Mountain View, CA). Putative centromeres were identified in the B. bovis C9.1 line genome (ftp://ftp.sanger.ac.uk/pub/pathogens/Babesia/). Artemis v. 16.0.0 (http://www.sanger.ac.uk/science/tools/artemis) was used to observe sequence %GC content, identifying stretches > 2.5 standard deviations below the mean in a running window of 1000 nt [76]. Sequences were considered candidates when the AT-rich segment was 2 Kbp or longer and possessed no annotated open reading frames. Candidate sequences were further analyzed by dot-plot to identify those containing significant internal repetitive sequence structure (S6 Fig). A single candidate was identified from chromosome 3, corresponding to position 2922284-2925108. Centromeric sequences were amplified with primers DA174 and DA175, and inserted into the NotI site by ligation with T4 DNA ligase. The hDHFR selection cassette was moved from the plasmid pDHFR-gfp-Bbcent2 (a gift from Shin-Ichiro Kawazu [12]) into the intermediate construct, in multiple steps. PCR-amplification of the hDHFR coding sequences with DA164 + DA165, the B. bovis actin promoter with DA166 + DA167, and the Rap1 gene 3´terminator sequences with DA168 + DA169 was used to recover these segments. A two-step crossover PCR was used to combine hDHFR coding sequence with promoter and terminator [72], simultaneously removing all HindIII, EcoRV, and PstI sites. The full-length cassette first was inserted into the SalI site of pBS2. The complete cassette then was removed with KpnI and PstI, and integrated into the artificial chromosome construct, using Gibson Assembly reagent mix (New England Biolabs; Beverley, MA) [77], yielding pBACc3. The final pBACc3 basic vector construct is 11,030 bp ( Fig 5A). To make BbACc3_Bbrad51wt, the entire Bbrad51 locus was amplified from pBbrad51wt_comp with primers DA248 + DA281, and inserted with NEBuilder reagents (New England Biolabs) into pBACc3 opened with BamHI and PstI. All added sequences and modified regions of the plasmid were fully Sanger-sequenced to confirm proper construction. Constructs were transfected into E. coli, using strains DH5α (Invitrogen), NEB10β (New England Biolabs), or Stellar (Clontech), by electroporation. For transfection, pBACc3_Bbrad51wt was linearized with PmeI to release the stuffer sequence and expose the telomeric ends ( Fig 5B). pBACc3 will be made available upon request.

Parasite transformation, and maintenance of BbACc3
Parasites were washed into cytomix buffer [78], then transfected with knockout or complementation vectors dissolved in cytomix. Transfection was achieved by electroporation with 10 pmol of DNA, under conditions of 1.25 kV, 25 μF, and 200 O resistance, as described [11]. Prior to transfection, pRad51ko and pBbrad51wt_comp were linearized with NotI. pBACc3_Bbrad51wt linearized with PmeI, releasing the A. marginale "stuffer" DNA and generating linear BbACc3_Bbrad51wt with telomeric ends. It was found not to be necessary to purify BbACc3 away from cleaved stuffer DNA prior to transfection. In some experiments parasites were transfected with pBACc3 as supercoiled plasmid. Following transfection, parasites were placed into culture for 24 h in the absence of drug selection. To effect selection, pyrimethamine then was added and maintained at a concentration of 2 μM (the approximate IC 90 concentration), beginning 24h post-transfection. In some experiments, drug pressure was removed for 21 days (approximately � 60 cell cycles) from already-transformed parasites, then was reapplied at varying concentrations to assess for loss of drug-resistance. During routine culture pyrimethamine pressure was maintained, as there appears to be a slight fitness cost to parasites maintaining BbACc3, and over time it may be lost.

Observation of chromosome maintenance and chromatin assembly
Maintenance of BbACc3_Bbrad51wt as a linear chromosomal element was confirmed by isolation of genomic DNA [66], with resolution of uncut and restriction-digested DNAs on a 0.7% agarose gel. DNAs were alkaline-transferred to nylon membranes as described [15], and hybridized with probes to Bbrad51, β-lactamase, centromeric, and hDHFR sequences. Probes were generated by PCR amplification of pBACc3_Bbrad51wt sequences with primers EAM8 + EAM11, DA290 + DA291, DA189 + DA190, or DA164 + DA165, etc., respectively. One μg of each amplicon was labeled with digoxigenin, and detected with anti-digoxigenin-HRP antibodies and CSPD substrate, according to supplier's instructions (Sigma Chemical; St. Louis, MO). Blots were imaged with a FluorChemR instrument (Protein Simple; San Jose, CA) by detection of luminescence. The stability of telomeric end lengths was tested by assessing the lengths of BamHI restriction fragments of PmeI-linearized BbACc3 constructs used to transfect, and BbACc3 constructs recovered after establishment of transformed parasites in culture. gDNAs were cleaved with BamHI, and probed with the above-mentioned probes. pBACc3 digested with PmeI to release stuffer sequences was used as control. Probes were labeled with 11-dUTP-digoxigenin, following manufacturer's instructions, and detected as described above. To observe assembly of BbACc3 and BbACc3_Bbrad51wt into chromatin, nuclei were isolated from transformed parasites, the chromatin partially digested with micrococcal nuclease (MNase), and the fragments isolated as described [44]. To observe overall nucleosomal assembly and spacing, digestion products were stained in-gel with SybrGold. Similar gels were also alkaline-blotted to nylon membranes and probed by Southern blotting, as described above. Alkali-labile probes were used to facilitate stripping of blots and re-probing. Absolute quantification of BbACc3_Bbrad51wt retention was performed by quantitative PCR of sequences from the 5´and 3´ends of Bbrad51 coding sequences present in genomic DNA preparations, using GoTaq qPCR BRYT Green Master Mix reagents (Promega Corporation; Madison, WI) and primer pairs qPCR-Bbrad51-5´F and R, qPCR-Bbrad51-3´F and R, qPCR-GAPDH-F and R, qPCR-BSD-F and R, qPCR-TPX-1F and R, and qPCR-hDHFR-F and R (S2 Table). Analysis was performed on a LightCycler 96 instrument running software version 1.1 (Roche Diagnostics; Mannheim, Germany). Purified pBbACc3_Bbrad51wt plasmid sequences were used to establish standard curves.
Supporting information S1 Fig. Similarity of B. bovis RecA superfamily-related proteins to known or putatively annotated Rad51 proteins from other species. Those proteins for which there is experimental support for catalysis of canonical Rad51 functions are indicated in blue in Fig 1. The Walker A and Walker B motifs are indicated here by red and blue box overlays, respectively. This alignment provided the basis for the phylogenetic tree shown in Fig 1A. (PDF)

S2 Fig. RT-PCR amplification of Bbrad51 and gfp-bsd transcripts.
Total RNAs were isolated from B. bovis CE11 subclones B8, C2, and C5, the initial uncloned CE11ΔBbrad51 knockout and three clonal lines derived from it ( ko1 C3, ko1 H5, and ko1 H6), and lines CE11Δrad51 ko2 and CE11Δrad51 ko3 . A. cDNAs made with oligo[dT] primer were amplified with primers EAM8 and EAM11 [11,62] for detection of Bbrad51 transcripts. B. cDNAs were amplified with XW119 and XW121 [11,62] for detection of gfp-bsd transcripts. Results demonstrate Bbrad51 transcription by wild type lines but not by knockouts. Conversely, gfp-bsd transcripts are present in knockout but not wild type lines. (TIF)

S3 Fig. Southern blot analysis of Bbrad51 locus. A.
Schematic diagram of the Bbrad51 locus in B. bovis CE11 wild type (bottom) and knockout (top) parasites, and the locations of restriction endonuclease and primer binding sites. The tables provide the anticipated sizes (in bp) of specific fragments detected with the indicated probes, based upon the genome sequence. B. Southern blots of B. bovis CE11 wild type, and the initial CE11ΔBbrad51 knockout parasite population both early and late (12 months later) in selection prior to cloning (left panel), after probing with Bbrad51-specific oligonucleotide probe, EAM10. Initial B. bovis CE11ΔBbrad51 clonal lines ko1 C3, ko1 H5, and ko1 H6 are shown in the right panel. C. The same blots shown in panel B are shown after being stripped and re-probed with oligonucleotide DA101R, specific for gfp sequences. The numbers above the bands indicate the sizes of the bands in bp. (TIF) S4 Fig. Titration of B. bovis CE11 wild type parasite sensitivity to MMS. In initial experiments to determine useful concentrations of MMS to observe phenotype, parasites were exposed to varying concentrations of MMS for 90 minutes at room temperature, then washed and placed back into culture for 72h. Smears were made at 0, 48, and 72h post-treatment, fixed and stained with Giemsa stain, and manually counted by microscopy in a blinded fashion. Growth is indicated as percent parasitized erythrocytes (PPE). This experiment indicated that a range from 250 μM to 1000 μM MMS provided useful results. (TIF)

S5 Fig. Structure of the pBbrad51wt_comp double crossover complementation plasmid.
This plasmid was designed to integrate into the genome by double-crossover homologous recombination via sequences upstream and downstream to the Bbrad51 coding sequences. In doing so, the strategy was to recreate the Bbrad51 locus, but with a short 3´-untranslated region for regulation, followed by the hDHFR selection cassette. The plasmid was introduced into parasites after NotI linearization, and selected for growth in the presence of pyrimethamine. This occurred as intended in three of three attempts with wild type parasites, but failed in 11 attempts with B. bovis ko1 H5 parasites. (TIF) S6 Fig. Identification of the B. bovis C9.1 chromosome 2 centromere. A. B. bovis T2Bo isolate genomic sequences [32] were scanned with the "GC Content (%) with 2.5 SD Cutoff" subroutine of Artemis v. 16.0.0 [76], using a sliding window of 1000 bp. Stretches of sequence > 2.5 standard deviations below the mean G+C content, and � 2 Kbp in length, were identified. Shown is the region of chromosome 3 identified by this strategy. Sequence comprising nucleotides 2922285-2925108 of the B. bovis C9.1 line genome [43], corresponding to nucleotides 29912-32736 of the B. bovis T2Bo isolate chromosome 3 [32], were recovered by PCR and used in the construction of pBbACc3. B. Dot-plots of the putative B. bovis C9.1 line chromosome 3 (left plot) and T2Bo isolate chromosome 2 (right plot) centromeres against themselves. The internal repeat structure of each is apparent from the plots, with the chromosome 2 centromere having a larger major repeat domain. The darkness of spots indicates the degree of similarity, with the dark blue diagonal lines indicating identity. Values within the range from 40-100% identity is shown. When plotted against one another there is no evidence for any specific sequence relationship (not shown).