|Home | About | Journals | Submit | Contact Us | Français|
Babesia bovis is an intraerythrocytic hemoparasite of widespread distribution, which adversely affects livestock production in many regions of the world. This parasite establishes persistent infections of long duration, at least in part through rapid antigenic variation of the VESA1 protein on the infected-erythrocyte surface. To understand the dynamics of in vivo antigenic variation among the parasite population it is necessary to have sensitive and broadly applicable tools enabling monitoring of variation events in parasite antigen genes. To address this need for B. bovis, “universal” primers for the polymerase chain reaction have been designed for the ves1α gene, spanning from exon 2 to near the 3′ end of cysteine-lysine-rich domain (CKRD) sequences in exon 3. These primers robustly amplified this segment, with minimal bias, from essentially the entire repertoire of full-length ves1α sequences in the B. bovis Mexico isolate genome, and are equivalently present in other isolates. On purified genomic DNA, this primer set can achieve a sensitivity of 10 genome equivalents or less. When applied to the amplification of cDNA derived from the B. bovis C9.1 clonal line evidence consistent with mutually exclusive transcription of the ves1α gene was obtained, concomitant with detection of numerous mutational events among members of the parasite population. These characteristics of the primers will facilitate the application of polymerase chain reaction-based methodologies to the study of B. bovis population and antigenic switching dynamics.
The bovine hemoparasite, Babesia bovis, uses a combination of cytoadhesion and antigenic variation to persistently infect cattle, a combined mode of survival that is also used by the human malarial parasite, Plasmodium falciparum . In B. bovis both processes rely upon the same parasite-derived protein, the variant erythrocyte surface antigen-1 (VESA1), expressed on the infected-erythrocyte (IE) surface [2-5]. VESA1 is a heterodimeric protein comprised of two structurally similar but distinct subunits, referred to as VESA1a and VESA1b [2,6-8]. VESA1a is encoded by members of the 1α branch of the ves multigene family , whereas VESA1b is thought to be encoded by members of the 1β branch (; additional unpublished data). When recognized by antibody cytoadhesion of parasitized erythrocytes is prevented [5,9]. However, antigenic variation renders existing immune responses ineffective at recognizing parasitized erythrocytes in vitro  and, it is thought, in vivo . Clearly, the kinetics of the immune response and antigenic variation will significantly impact upon the dynamics of parasite population growth and persistence. Although the capacity for rapid antigenic variation in the VESA1 protein is well established, essentially nothing is known regarding the dynamics with which the structure of each subunit is altered during infection in vivo, or even during in vitro growth. Knowledge of variation dynamics will improve our understanding of parasite persistence, and impact upon the development of any strategies targeting antigenically variant components for immunoprotection.
One robust approach to understanding better the dynamics of in vivo antigenic variation would be to directly monitor changes in VESA1 structure over time during an infection. This could be achieved by sampling the VESA1 isoforms transcribed at different phases of infection, establishing both the apparent rate of in vivo variation in the parasite population within a single host, and the complexity of isoforms produced under different circumstances. Given that a primary mechanism of antigenic variation in B. bovis is segmental gene conversion [7,8], there is considerable potential for the creation of expressed, unique, mosaic copies of the gene not present in the genome prior to the time of their formation.
While it has been possible to characterize changes that occur to expressed ves genes, it has not been feasible to follow the dynamics of their change in real time because the tools to follow in vivo behavior do not currently exist. We therefore initiated the development of such tools. Polymerase chain reaction (PCR) primers suitable for use either in PCR or reverse transcriptase-PCR (RT-PCR) reactions were developed to facilitate the acquisition of this information for the ves1α gene. We describe here the development and characterization of a pair of “universal” primers encompassing a part of exon 2 and most of the CKRD domain-encoding region of exon 3 of the ves1α gene. Within the VESA1a polypeptide this region is extremely rich in cysteine and lysine residues. Although full-length CKRD domains contain a long, highly conserved, complex motif ; additional unpublished results), it is overall highly polymorphic in sequence, and the CKRD region of each of the approximately 72 different ves1α genes within the B. bovis genome appears to be unique . Modifications by segmental gene conversion result in further expansion of the repertoire of unique expressed variants . Thus, the CKRD domain-encoding region of the ves1α gene should provide a sensitive estimation of structural, and presumably antigenic, changes occurring during the course of an infection. The ability to monitor changes in the CKRD region of the transcribed ves1α gene would facilitate study of the dynamics of antigenic variation in vivo, both within the individual host and the parasite population as a whole.
Genomic DNA (gDNA) was isolated from in vitro-cultured B. bovis parasites by the use of sodium dodecylsulfate, Proteinase K, and phenol-chloroform extraction as described . Prior to extraction, erythrocytes were treated on ice with 0.05% (w/v) saponin in VYM buffer  to permeabilize the erythrocyte membranes and release hemoglobin. Prior to isolation of RNA, the B. bovis C9.1 line was enriched once for parasites reactive with monoclonal antibody 4D9.1G1, as described . When assayed by live-cell immunofluorescence  with 4D9.1G1 the selected parasitized erythrocytes were at least 70% reactive. Rather than being antigenically diverse at the 4D9.1G1 epitope, this primarily reflects disparate levels of VESA1 expression, with some cells being below the level of assay sensitivity (; additional unpublished observations), and varies somewhat from day to day. Total RNA was isolated from approximately 1 × 106 B. bovis C9.1 parasitized erythrocytes, by using the ToTALLY RNA kit (Ambion, Inc.; Austin, TX). The RNA was digested with Turbo DNase (Ambion) for 30 min at 37°C, an additional 1 U DNase was added, and incubation continued for 30 min. The enzyme was then inactivated with Inactivation Reagent, according to manufacturer's protocol. The RNA was reverse transcribed with Superscript II reverse transcriptase (Invitrogen; La Jolla, CA), using 1 μg of random hexamer primer.
To design “universal” primers targeting the ves1α gene, optimized alignments were prepared of all ves1α gene sequences available at the time this work was first initiated [7,8]. Focusing on exons 1 and 2, and the 5′ end of exon 3, it was determined that exon 2 is very highly conserved among ves1α genes. Four forward and three reverse primers were designed, based upon highly-conserved sequences of exon 2 and near the 3′ end of sequences encoding the CKRD domain within exon 3. Based upon initial experiments we determined that the optimal combination was comprised of primers PD2Fa (AATGYATATGTGGCCTGG) and PD2Fb (AATGYATATGTGGCCTTG) as forward primers, mixed together in a 4:1 molar ratio (hereafter referred to as “PD2Fa,b”), coupled with primer PD1R (TACAANAACACTTGCAGCA) as the reverse primer. Primers were synthesized by Sigma-Genosys (Woodlawn, TX). As an internal control for reverse transcription and amplification, the β-tubulin gene was amplified, using forward primer AZT01 (TGCCAAATTCTGGGAAGTC) and reverse primer XW10 (CATACCACCAAAGGGACTCAA).
Sequences encompassing the 3′ end of exon 2, intron 2, and most of the CKRD-encoding sequences from exon 3 (Figure 1) were amplified by the polymerase chain reaction (PCR). PCR was carried out using either iProof Polymerase (BioRad; Hercules, CA) or Phusion Polymerase (New England Biolabs; Beverly, MA). In each case, the manufacturer's “high fidelity” buffer was used, employing 1.75 mM Mg2+, along with 0.2 mM deoxyribonucleotide-triphosphates, and 250 nM primers. Both polymerases gave comparable results in most respects. PCR reactions were conducted in an MJ Research PTC-200 thermal cycler as 50 μl reactions. Amplification involved an initial 150 sec. denaturation at 98°C, followed by 35 cycles of 30 sec. at 98°C, 30 sec. at 58.5°C, and 1 minute at 72°C. This was followed by a 7 minute incubation at 72°C to enhance completeness of products prior to cloning. Amplification of the B. bovis β-tubulin gene (accession no. L00978) was included as an internal positive control, using primers AZT01 and XW10. Amplified products were cloned either by using A overhangs, vector pCR4, and the Topo-TA cloning kit (Invitrogen) for iProof polymerase, or by topoisomerase-mediated blunt-end cloning into vector pCR2.1, using the Topo Zero-Blunt cloning kit (Invitrogen) for Phusion polymerase.
Individual cloned PCR products were cycle-sequenced using dye terminator technology and BigDye v1.1 reagents (Applied Biosystems; Foster City, CA), with analysis on an Applied Biosystems 3130 Sequencer, as described elsewhere . Alternatively, high-throughput sequencing was performed by robotic picking of primary recombinants, rolling-circle amplification of plasmid DNAs with Phi29 polymerase, and dye terminator sequencing of amplified products with analysis on an Applied Biosystems 3730XL Sequence Analyzer. Analysis of sequences and high-throughput sequencing were performed by the DNA sequencing facilities of the University of Florida Interdisciplinary Center for Biotechnology Research. Sequences were trimmed, aligned, and relational trees drawn using CLC Sequence Viewer, v5.0.1 (CLC Bio; Aarhus, Denmark), with manual optimization of alignments. Relational trees were drawn employing the neighbor-joining method , using 200 boostrap iterations. The nearest-neighbor method makes no assumptions regarding the rate of evolution of each sequence, and no evolutionary relationships are implied from our analyses. Alignments of cDNAs were performed including the already published p9.6.2 cDNA ves1α sequence (accession number DQ267461) obtained from the C9.1 clonal line , as “type sequence” for comparison. Genomic data for the unrelated B. bovis T2Bo isolate were downloaded from Genbank (National Center for Biotechnology Information), using genomic coordinates provided by Brayton et al. . For comparison of cDNA and gDNA sequences, intron 2 was identified by its position relative to the p9.6.2 cDNA, and removed from all gDNA sequences prior to alignment. Incomplete sequences were removed from the alignment. Sequence recovery data were plotted and analyzed by non-linear regression analysis, using SigmaPlot v11 (Systat Software, Inc.; Point Richmond, CA).
All sequences presented herein are deposited in GenBank, with accession numbers FJ536943 through FJ536983 (gDNAs) and FJ536984 through FJ537067 (cDNAs).
Expressed, full-length copies of the ves1α gene have a consistent 3-exon 2-intron structure ([7,8,11]; additional unpublished data]. Primers were designed, based upon highly conserved sequences present in available ves1α genes, which spanned across intron 2 to enable the use of these primers for analysis of both gDNA and transcripts. A schematic diagram illustrating the overall structure of ves1α genes and the locations of the primers used in this study is given in Figure 1. Preliminary experiments revealed that a (4:1) molar ratio of PD2Fa and PD2Fb (PD2Fa,b) as forward primer, coupled with PD1R as reverse primer, consistently provided very sensitive and reproducible amplification of ves1α sequences from genomic DNA. We therefore focused in subsequent experiments upon optimizing the PCR reaction for annealing temperature, Mg2+ concentration, and primer concentration, using this mixture of primers and gDNA as template.
The amplicons recovered from gDNA were a diverse mixture of major products ranging in size from approximately 550-700 bp, with a lesser smearing of higher mass products (Figure 2). Reproducible and robust amplification could be achieved with as little as 100 fg of genomic DNA, although amplification could often be achieved with as little as 10 fg, or roughly one genome equivalent of total DNA. This level of sensitivity is likely possible only because the B. bovis genome contains approximately 100 ves1α genes, scattered over the four chromosomes [8,11,16]. However, amplification from such a small quantity would seriously skew representational aspects of the results. The high mass smears may be a result of crossover PCR amplification products resulting from the presence of conserved sequences within these variant genes, including the segment under study, and/or the predominant organization of this gene family as clusters of divergently-oriented gene pairs [7,8,11]. These characteristics of the ves gene family could result in formation of some unanticipated products. Despite this, no such products were cloned, nor do any identifiable genomic copies of ves1α fall into that size range .
To determine how broadly this primer set would function for the recovery of any ves1α sequence present, ves1α sequences were amplified from 10 ng of genomic DNA, representing approximately 1,000 genome equivalents, and the amplicons were cloned and sequenced. Of 192 colonies picked for sequencing 176 provided useable, high quality sequence. The sequences recovered represented 58 unique CKRD sequences, of which 41 represented full-length sequences from the CKRD domain (Figure 3A and Supplementary Figure S1). The other 17 were incomplete sequences, presumably due to deletion of end sequences during cloning. This compares favorably with the 47 full-length CKRD domain sequences present among the 72 ves1α genes identified in the B. bovis genome  (Figure 3B). Sequences were considered to be unique if they had three or more nucleotide differences from other sequences. Of the 41 full-length CKRD sequences observed, 28 were recovered once only. All others were recovered from two to 33 times, with the distribution shown in Figure 4. The frequency distribution of individual sequence recoveries was fit well (R2= 0.9506) by a non-linear regression curve describing a dual parameter-exponent logarithmic distribution. Given this excellence of fit, we interpret this to indicate that the limiting factors in sequence detection are the points during the PCR process at which individual sequences begin to be amplified logarithmically, and whatever amplification bias exists for each gene copy once logarithmic amplification is initiated. If this distribution were due solely to initiation of logarithmic amplification, the recovery range from 1 to 33 copies would imply a difference of approximately 5.04 cycles of amplification between the latest and earliest to amplify sequences. Alternatively, if this were due solely to primer bias for specific loci, this would maximally represent a bias of approximately 10.5% per cycle for the most abundant products. It is likely that both factors come into play, particularly as the capacity for some quasi-palindromic ves loci to self-anneal when denatured (unpublished observations) is likely to be greater than others. This feature of these loci undoubtedly affects the ability of primers to anneal with one or the other of the sites, affecting the point at which logarithmic amplification predominates over linear amplification. Regardless, this primer set clearly amplifies a very wide variety of the ves1α repertoire efficiently. When applied to the study of antigenic variation dynamics at the transcript level, where the apposing ves sequence is not physically contiguous and continuous, it is apparent that it should be possible to capture most, if not all, full-length variant transcripts.
An alternative indication of the universality of the primers is provided by comparison of the relational trees made for non-redundant full-length sequences recovered from B. bovis C9.1 line and unrelated T2Bo isolate gDNAs (Figure 3). It is unfortunately not possible to directly compare the sequences as the B. bovis genome was sequenced from the T2Bo isolate. The T2Bo isolate was obtained from Texas in 1978 , whereas this work was done using the C9.1 clonal line, a derivative of the Mexico isolate , and the two original parasite populations were segregated both geographically and temporally. Being highly variant genes overall there is no direct ves1α gene-to-gene correspondence between isolates. Indeed, construction of a relational tree from an alignment containing C9.1 and T2Bo sequences together results almost exclusively in isolate-specific segregation and clustering of sequences (data not shown). However, it is apparent that the ves1α sequences in each isolate similarly clustered into four major groups (Figure 3, dashed boxes) with numerically similar distributions of members. This finding supports the probability that there is not a significant bias toward recovery of the members of one grouping relative to another, an important characteristic of primers intended for this usage. As no specific sequence motifs appear responsible for these groupings, it is assumed they simply reflect overall similarities among members. The cause of this similarity is unclear, but in the T2Bo isolate it is unrelated to their chromosomal origins (not shown). It is possible that sequences within groups are favored interaction partners for other sequences within the same groups during gene conversion events, resulting in overall partial homogenization of sequences. This possibility remains to be explored.
The primers were tested for their intended application in a pilot experiment involving RT-PCR amplification of cDNA made from total B. bovis C9.1 line RNA. Prior to isolation of RNA the parasite population was selected once for binding of mAb 4D9.1G1 to provide as antigenically homogeneous a population as possible. This enabled us to ask whether transcripts appeared to arise from a single ves1α locus or from multiple loci. Of the 96 sequencing reactions performed, 84 gave useable results. Alignment of these sequences revealed that 67 of the sequences were identical or sufficiently similar to one another (tree distance ≤0.001), to clearly reflect transcription from a single ves1α locus that had undergone little mutation within the population (Figure S2). When aligned with the amplified gDNA squences, all but seven of the 17 remaining cDNA sequences were more closely related to the major cDNA form, to one another, and to the p9.6.2 cDNA form of the ves1α gene than to any other genomic copy, strongly suggesting these, too, were derivatives of the same locus (Figure 5). The p9.6.2 form was that present in the B. bovis C9.1 line LAT at the time when this gene family was first described; it has since undergone some modification and continues to do so due to ongoing stochastic variation. Further evidence for a common locus origin of these transcripts is provided by the extreme conservation of sequences in the region from position 244-333 of the cDNA alignment (Figure S2), in contrast with the extreme diversity present in the corresponding region among gDNA sequences (Figure S1, positions 317-453 of alignment). This result virtually precludes a multi-locus origin for these transcripts. Six of the remaining seven cDNAs more closely resembled but were significantly different from other genomic copies of the ves1α gene, indicating either that they represent unique genes not captured during gDNA amplification, or that they have been so extensively modified as to no longer resemble any specific genomic locus. On the other hand, one cDNA, SSII-G09, perfectly matched another genomic locus represented by sequence DS2-B08. This cDNA was either transcribed from that locus, perhaps as “leaky transcription”, or sequences from that locus completely replaced the amplified region within the known LAT. Alternatively, this cDNA may reflect an in situ transcriptional switch from the known LAT to the DS2-B08 locus in a small fraction of the population. These data do not discriminate among these possibilities. Differences among the various p9.6.2-like cDNAs recovered were due to scattered point mutations or replacement of short sequence patches, consistent with the previously demonstrated pattern of sequence changes observed during segmental gene conversion . Since the parasites used in this experiment had been growing in culture for several months without selection for a particular CKRD sequence, it would be expected for mutations to have accumulated, as modification of the ves1α gene occurs over time during in vitro culture . This is true despite the selection of these parasites for immunoreactivity with monoclonal antibody 4D9.1G1 prior to RNA isolation, since the epitope recognized by this antibody maps to the VDCS domain encoded by sequences approximately 2000 nucleotides downstream . Significantly, these results are consistent with and provide experimental support for mutually exclusive transcription of the ves1α gene. However, they do not imply that a single ves locus is competent for transcription, only that a single ves locus appears to be transcriptionally active at one time. It should be noted, too, that RT-PCR recovers RNA species roughly in proportion to their relative steady-state abundance, and short-lived RNAs that do not accumulate to significant levels may not be detected in a small scale analysis.
It is not unexpected that transcription among ves1α genes would be mutually exclusive. The success of antigenic variation depends upon the infectious agent not revealing more of its antigen repertoire than necessary at any given time . Otherwise, a broadly reactive immune response could ensue that might eliminate the parasite before it could respond. Mutually exclusive transcription of variant antigen genes from within a multigene family has been demonstrated in other infectious agents, including among the 1000 or so variant surface glycoprotein (VSG) genes of the African trypanosomes  and the 60 var genes of P. falciparum [21,22]. B. bovis shares with these two parasitic protozoa genetic mechanisms of antigenic variation and lifestyle within the host, respectively, as well as the ability to establish persistent infections.
A key characteristic of the primer set described herein is that it was designed to span intron 2 of the ves1α gene. This facilitates assessment of the nucleic acid source being amplified during RT-PCR experiments, a fundamental characteristic for interpreting the results of such studies. During numerous attempts to perform single-cell RT-PCR on B. bovis IRBCs this proved crucial in establishing that, with rare exception, amplification of transcripts was not being detected. Rather, an apparent “reassembly” of gDNA fragments during the process of reverse transcription seems to have occurred, despite consistent failure to amplify products in control samples lacking reverse transcriptase (unpublished results). A similar observation has been reported in the amplification of mammalian transcripts . Unfortunately, similar studies performed on P. falciparum in the study of var gene transcription dynamics employed primer sets which could not provide such information due to the structure of var genes and the necessary location of primer binding within the gene [22,24]. Interestingly, ves1α intron 2 was observed to exist in two discrete size ranges among the ves1α sequences, 38-39 bp and 72-74 bp, yet to share essentially invariant sequences at the 5′ (GTAAGTA) and 3′ (TAG) ends and some internal sequences (Figure S3). The significance of this dichotomy is not clear.
Our results demonstrate that the PD2Fa,b combination of forward primers with the PD1r reverse primer provides for extremely sensitive amplification of the ves1α gene repertoire of B. bovis, with little bias. These primers also allow discrimination of transcript versus gDNA amplification, although unspliced precursor transcripts may be misinterpreted. When applied to RNA obtained from “bulk culture” of a clonal parasite line the results strongly suggest that transcription among the ves1α genes occurs in a mutually exclusive manner. Because of the characteristics and performance demonstrated by this primer set, we propose that it may provide a suitable tool to follow variation within this region of the ves1α gene during in vivo infection with B. bovis, and may provide a basis for the development of sensitive PCR-based diagnostics. Given their broad specificity, it is unlikely that such diagnostics would be adversely affected by the identity of the isolate under study or by antigenic variation within that isolate.
Figure S1. Alignment of non-redundant sequences obtained from one amplification of B. bovis C9.1 line gDNA, using the PD2Fa,b and PD1R “universal” primers. Nucleotides are individually color-coded for clarity. The consensus plot beneath the alignment provides a graphical representation of positional nucleotide conservation.
Figure S2. Alignment of cDNA sequences recovered by RT-PCR from the B. bovis C9.1 clonal line. The 84 full-length cDNA sequences obtained in a single amplification of cDNA made from bulk RNA were aligned. Most are either identical or extremely similar, or differ from the major population primarily by the exchange of a short sequence patch, as previously reported [7,8]. Each unique sequence is shown once only, with the number of times each was isolated indicated in square brackets following the sequence name. The p9.6.2 cDNA characterized in the original description of the ves multigene family was included in this alignment as a “type” sequence for comparison. The consensus plot beneath the alignment provides a graphical representation of positional nucleotide conservation.
Figure S3. Alignment of intron 2 sequences excised from non-redundant B. bovis C9.1 line gDNA CKRD sequences. Sequence conservation is apparent at both the 5′ and 3′ ends, likely for proper splicing of primary transcripts. In addition, an internal block of degenerate sequence bearing conserved nucleotides can be observed. The consensus plot beneath the alignment provides a graphical representation of positional nucleotide conservation.
The authors thank Yu-Ping Xiao for excellent technical assistance, and Alexia Berg and Allison Vansickle for assistance with animal care and parasite maintenance. This work was supported by grant #R01 AI055864 from the National Institute of Allergy and Infectious Diseases.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.