Host immunity modulates transcriptional changes in a multigene family (yir) of rodent malaria

Authors


E-mail jlangho@nimr.mrc.ac.uk; Tel. (+44) 208 816 2558; Fax (+44) 208 816 2638.

Summary

Variant antigens, encoded by multigene families, and expressed at the surface of erythrocytes infected with the human malaria parasite Plasmodium falciparum and the simian parasite Plasmodium knowlesi, are important in evasion of host immunity. The vir multigene family, encoding a very large number of variant antigens, has been identified in the human parasite Plasmodium vivax and homologues (yir) of this family exist in the rodent parasite Plasmodium yoelii. These genes are part of a superfamily (pir) which are found in Plasmodium species infecting rodents, monkeys and humans (P. yoelii, P. berghei, P. chabaudi, P. knowlesi and P. vivax). Here, we show that YIR proteins are expressed on the surface of erythrocytes infected with late-stage asexual parasites, and that host immunity modulates transcription of yir genes. The surface location and expression pattern of YIR is consistent with a role in antigenic variation. This provides a unique opportunity to study the regulation and expression of the pir superfamily, and its role in both protective immunity and antigenic variation, in an easily accessible animal model system.

Introduction

Plasmodium falciparum and Plasmodium vivax are the most prevalent malaria species which infect humans. Although P. falciparum is responsible for most malaria-associated mortality worldwide, P. vivax malaria, which is less susceptible to current control methods, predominates outside of Africa (Mendis et al., 2001). Even in the presence of a vigorous immune response in the infected/immune host, these parasites give rise to debilitating and chronic disease with an associated reduction in economic performance. Antigenically variable surface antigens (VSAs) expressed at the surface of the parasitized red blood cell (pRBC) help the parasite evade host immunity (Brown and Brown, 1965; Barnwell et al., 1982; McLean et al., 1982; Mendis et al., 1988) and may explain how these chronic infections are maintained. Multigene families coding for VSAs exist in several malaria parasite species (Kaviratne et al., 2003), and paralogues and orthologues are found within and across these species (al-Khedery et al., 1999; Carlton et al., 2002; Fischer et al., 2003).

Elucidation of the molecular mechanism of antigenic variation has focused until now on the var gene family of P. falciparum which encodes an immunogenic protein, PfEMP1, which mediates parasite sequestration and has been linked to the pathology of malaria. PfEMP1 undergoes clonal antigenic variation (Biggs et al., 1991; Roberts et al., 1992) thereby facilitating evasion of host immunity (Baruch et al., 1995; Smith et al., 1995; Newbold et al., 1997). While there is no homologue of var in P. vivax, the vir multigene family, with more than 600 members, has been described in this parasite and is also thought to be involved in immune evasion (del Portillo et al., 2001). Homologues of vir occur in the rodent malarias Plasmodium yoelii (yir ), P. chabaudi (cir ) and P. berghei (bir) (Carlton et al., 2002; Janssen et al., 2002), termed the pir superfamily (Janssen et al., 2004), and VIR and CIR proteins are thought to be expressed on the surface of the pRBC (del Portillo et al., 2001; Janssen et al., 2002; Merino et al., 2003). Antigenically variant parasite populations have been identified in P. chabaudi infections (McLean et al., 1986).

The availability of a good rodent model for the study of variant antigens would make a major contribution to the field, facilitating the critical analysis of the relationship between VSA, host immunity and immune evasion in malaria, an analysis difficult to perform with human parasites. Although both P. vivax and P. falciparum do infect New World monkeys (Hommel et al., 1983; Collins et al., 1994), and, indeed, for the latter, serologically distinct parasite populations have been isolated from primary and convalescent peaks, the use of these models is currently severely limited. Initial investigations of phenotypic antigenic variation in Plasmodium knowlesi-infected rhesus monkeys, in P. falciparum-infected Aotus monkeys and subsequently in humans, were facilitated by the early recognition and characterization of the antigens involved (Brown and Brown, 1965; Leech et al., 1984; Marsh and Howard, 1986). In vitro assays were available to detect the differential expression of members of these highly immunogenic, immunodominant yet relatively small protein families. Such an in vitro assay for antigenic variation is not currently available for P. yoelii.

We performed transcriptional analysis to determine whether sets of yir genes were transcribed and to enable us to develop reagents to focus on those antigens most likely to reflect VSA changes at the surface of infected erythrocytes. The study of the regulation and expression of the pir/PIR superfamily will be important to the understanding of the mechanisms of antigenic variation.

Here, using a reverse genetics approach, we show that many yir genes are transcribed during the erythrocytic cycle of P. yoelii in infected mice. We demonstrate changes in yir transcription which are dependent on, and modulated by, the presence of a host immune response and establish that YIR proteins are expressed on the surface of live P. yoelii infected cells. These data suggest that antigenic variation and evasion of host immunity may be an important function for this enormous family of variant proteins.

Results/discussion

Plasmodium yoelii yir coding sequences are located subtelomerically and are predicted to consist of three exons: 3–15 bp, 700–900 bp (polymorphic) and 30–90 bp (highly conserved) (Fig. 1A; http://www.tigr.org/tdb/e2k1/pya1/) flanking two highly conserved introns (80–150 bp) (Fig. S1). Using primers designed within conserved regions of the gene (Fig. 1A), reverse transcription polymerase chain reaction (RT-PCR) from RNA obtained from ring-schizont stages confirmed the deduced intron/exon structure and demonstrated that yir transcripts, like vir (del Portillo et al., 2001) and cir (Janssen et al., 2002), are detectable throughout the intraerythrocytic cycle (Fig. 1B, and data not shown). Sequencing of RT-PCR products from individual samples indicated that many yir genes within the genome are transcribed. The location of yir transcripts, produced with primer sets designed to conserved regions of the genes (green ◆), in a bootstrapped (1000 replications) neighbour-joined unrooted tree is shown in Fig. 2, with bootstrap numbers shown on the branches of interest. Both neighbour-joining and minimum evolution tests of phylogeny yielded similar results, thus validating the phylogenetic tree. [Fig. S2 shows expanded branches with yir loci and bootstrap values; numbers refer to unique loci within the YIR database (http://www.tigr.org/tdb/e2k1/pya1/). Table S1 lists the investigated transcribed genes in order of appearance in the phylogenetic tree]. The repertoire of transcribed genes presented here is not exhaustive, as the number of different transcripts detected almost equalled the number of clones sequenced. If the majority of transcripts are translated into protein, then the sheer number of YIR proteins expressed, in itself, could provide an efficient mechanism of evasion of host immunity.

Figure 1.

A. Schematic depicting predicted yir intron/exon structure. Shaded boxes show the major conserved regions of the gene. Location of RT-PCR primers (P1–P4) and the predicted product sizes are indicated. P3 amplifies a broad transcript repertoire. P4 amplifies a more limited transcript repertoire. The location of peptides 1 and 2, the conserved cysteines and the major conserved amino acid blocks and polymorphic regions within the protein are shown.
B. Transcription of yir genes in asexual erythrocytic stages of P. yoelii. RNA was extracted from pRBC fractionated into predominantly schizonts (Sz), early (ET), middle (MT) and late (LT) trophozoites and rings (R) by differential gradient centrifugation. RT-PCR was performed, using P3 or P4 primers, in the presence (+) or absence (–) of reverse transcriptase. Product identity was verified by sequencing (results not shown).

Figure 2.

Phylogenetic unrooted tree of 720 yir sequences. A total of 720 sequences were aligned with clustalw. Based on the alignment, a neighbour-joined unrooted phylogenetic tree was calculated, bootstrapped using MEGA version 2.1 (Kumar et al., 2001). Genes transcribed in the starter population and transcriptional differences between primary and secondary infections are shown. In the full tree transcribed sequences are represented by green ◆ and red circles/purple squares (listed in order of appearance on tree in Table S1 in Supplementary material, keeping their relative rank in the multiple sequence alignment). Transcribed genes amplified with primers P3 (green diamonds) and P4 (coloured circles/squares) are indicated. Transcripts found in the starter population (used to initiate all further experiments) [green diamonds and purple squares (set 1)], showing the transcription of a large number of genes during the primary infection: set 1 shows a set of genes transcribed in all infections of immunocompetent and immunodeficient hosts, amplified using P4 primers. One of these transcripts (PY01810) was amplified from all samples tested. Set 2 (red circles) shows a different set of related transcripts, detected using P4 primers, found in re-infections of semi-immune hosts only, with transcript PY02298) being most common. Figure S2 shows expanded branches containing sets 1 and 2 with bootstrap values.

To investigate whether the transcriptional profile of yir changes during infection, and whether it is influenced by an acquired immune response, restriction fragment length polymorphism (RFLP) analysis of RT-PCR products was performed on parasite RNA from primary and secondary infections of immunocompetent and immunodeficient mice. Initially, mice were infected with the starter population and parasite RNA was obtained at days 12 and 18 post infection (p.i.). The RFLP pattern in parasite RNA from the starter population and from these infected mice was essentially unchanged (Fig. 3A), indicating that the yir transcription pattern is relatively stable in a primary infection. The appearance of a different band in mouse 3 RNA would indicate, however, that minor transcriptional changes of yir occur during the course of an infection Fig. 3A.

Figure 3.

RFLP analysis of primary and secondary infections in immunocompetent and immunodeficient mice. RFLP analysis of Ssp 1 or Dra 1 digested RT-PCR products, obtained using primer set P4 (Fig. 1A) (end-labelled by incorporation of a terminal 32P label into the reverse primer), was carried out on samples from P. yoelii-infected BALB/c and RAG2–/– mice. RFLP experiments were repeated to minimize or eliminate RT-PCR artefacts.
A. RFLP analysis of a primary infection in BALB/c mice. Samples were collected on day 12 and 18 p.i. with 5 × 103 starter population parasites. The RFLP pattern from RNA of the starter (a) population, and from pRBC of three different mice (1–3) at days 12 and 18, are shown. Arrows indicate the dominant transcripts.
B. RFLP analysis of primary and secondary infections in BALB/c mice. Mice were infected as described above. Thirteen days p.i., blood samples were collected and aliquots immediately dissolved in lysis buffer for RNA extraction (b). The mice were then treated with pyrimethamine (days 13–19) and no further parasites were observed. On day 34 the mice were re-challenged with 109 homologous parasites i.p. The ensuing low parasitaemia was cleared by days 3 and 4 after challenge when blood was subinoculated into RAG2–/– mice to expand the subpatent parasite populations and RNA extracted (c). The RFLP patterns from the primary (b) and secondary infections (c) are shown (right). Arrows indicate the dominant bands. Expansion of parasite populations into RAG2–/– mice (c) does not change the RFLP pattern, as the RFLP pattern of the similarly subinoculated sample (b) parasites remained stable (data not shown).
C. RFLP analysis of primary, recrudescent and secondary infections in RAG2–/– mice. Animals were infected as in (A) and treated daily with 5 mg per kilogram of pyrimethamine between days 7 and 13 (shaded area). In three mice the parasitaemia spontaneously recrudesced on days 20 and 21; RNA was extracted from pRBC (day 24) for RFLP analysis and sequencing. In the four remaining mice parasites were not detected up to day 32, when these mice were re-infected with the starter population (109 parasites) and parasitaemia was monitored. Parasites were detectable over the ensuing 3 days and were then collected for RNA analysis. The RFLP patterns obtained from infecting (a), two of the recrudescing (d) and three of the re-infection (e) parasite populations are shown. Arrows indicate the dominant transcripts.

However, when primary infections of immunocompetent mice (BALB/c) were drug-cured and the mice re-infected (Fig. 3B, I), a strikingly different RFLP pattern was observed in parasites obtained from the re-infections, compared with their primary infections, with multiple new RFLP bands appearing (Fig. 3B, II, and Fig. S3 compare lanes a and b with c).

This was in contrast to the patterns seen in immunodeficient mice (RAG2–/–) infected with the starter population where there was no change in the RFLP pattern between the starter population, recrudescent parasite populations obtained after subcurative drug treatment, or parasites obtained from re-infections of drug-cured mice (Fig. 3C).

Sequencing the RT-PCR products from the different groups confirmed these transcriptional changes (Fig. 2; Figs S4 and S5). Transcriptional changes within a subset of the total yir repertoire were investigated using a more restrictive primer set (P4), as the very large number of yir genes transcribed in an infection precluded the use of conserved (P3) primers (transcribed genes green ◆, Fig. 2). Restriction of the analysis to a smaller subset allowed transcriptional differences within a particular yir subset to be identified. Two sets of transcripts were detected. Set 1 transcripts only were detected in the starter population, primary infections in BALB/c mice and from primary, recrudescent and secondary infections in RAG2–/– mice. All clones (90 from 12 mice) from starter populations (a) in Fig. 3A–C, from primary infections in BALB/c mice (Fig. 3B, b) and from primary and secondary infections in RAG2–/– mice (Fig. 3C, d and e), belong to a set (set 1) of closely related transcripts (Fig. 2, purple ●; Fig. S2), of which one (♯PY01810) predominated (83/90 clones). Although P. yoelii infections are of limited duration in the mouse, set 1 transcripts continued to be the only transcripts detected with primer set 4 after sequential passage (six passes) of the parasite through RAG2–/– mice for 52 days (data not shown). However, in re-infected semi-immune BALB/c mice, both set 1 transcripts and a very different set (set 2) of related transcripts were detected (Fig. 2, red ●; Figs S2 and S5), with transcript ♯PY02298 being most abundant (12/45 clones from five mice). Therefore, re-infection of semi-immune animals led to the expanded expression of more diverse yir gene products. Interestingly, RT-PCR using ♯PY02298-specific primers indicated that this gene was transcribed in the starter population (data not shown). Therefore, the differences observed, both by RFLP and by sequence analysis, probably represent changes in the relative proportions of these transcripts within the transcribed repertoire. The YIR protein encoded by the ♯PY02298 transcript, detected after re-infection, shares only 37% identity (49% similarity) with the predominant YIR protein (♯PY01810) present in the starter population and differs in several predicted antigenic regions (Fig. S6; Kolaskar and Tongaonkar, 1990). Overall, these results support the view that parasites modulate yir transcription in response to host immunity. These data are consistent with earlier serological studies of P. knowlesi and P. chabaudi infections, where new variants were detectable only in recrudescent parasites (Brown and Brown, 1965; McLean et al., 1982; 1986; Jarra et al., 1986) or in parasites selected after passive transfer of hyperimmune serum (Jarra et al., 1986).

Western blot analysis (Fig. 4) demonstrated that polyclonal antisera to two non-overlapping YIR peptides (location within the protein shown in Fig. 4A and Fig. S7) reacted with parasite proteins of the predicted molecular weight (approximately 30 kDa), and these antisera also stained approximately 25% of fixed permeabilized schizont-infected RBC (Fig. 4C) in a cytometric assay.

Figure 4.

Detection of YIR proteins in P. yoelii-infected erythrocytes.
A. Schematic of YIR protein structure showing conserved cysteines (C), regions conserved between YIR proteins (hatched boxes) and the location of the peptides used to raise antisera.
B. Western blot analysis. (i) Erythrocytes from RAG 2–/– mice infected with P. yoelii 17XA were lysed directly into SDS-PAGE buffer and extracted proteins from 4 µl (lane 2) or 10 µl (lane 3) blood were electrophoresed through NuPAGE 4–12% SDS denaturing gels, transferred to nitrocellulose membranes and probed with affinity-purified polyclonal serum to semi-conserved motif 1 [peptide 1 (ISAGCLYLLDEFIKDC) (α pep 1)] or purified IgG from mouse antiserum (Sigma-Aldrich) (control). Lane 1 contains proteins extracted from 10 µl uninfected blood from a RAG2–/– mouse. Size markers were broad range pre-stained protein markers Seeblue2 (Invitrogen). (ii) Proteins extracted from erythrocytes collected from BALB/c (lane 2) and C57BL/6 (lane 3) mice infected with P. yoelii 17XA were electrophoresed as for (i) and probed with unpurified antiserum to semi-conserved motif 2 [peptide 2 (KLYDALQSLCNMYNEF) (α pep 2) or unpurified normal serum (control)]. Lane 1 contains uninfected erythrocytes. Size markers were broad-range pre-stained protein markers (Bio-Rad).
C. Indirect immunofluorescence. P yoelii infected pRBC from BALB/c mice enriched for schizonts on Nycoprep gradients. Acetone-fixed thin films of these cells were stained with mouse antiserum to peptide 2 and FITC-conjugated anti-mouse IgG. Parasite nuclei were stained with DAPI.

Fluorescence-activated cell sorting (FACS) analysis showed that up to 7% of unfixed pRBC of a schizont enriched fraction (Fig. 5A) stained with two different YIR antisera, thus demonstrating expression of YIR on the surface of live pRBC. Expression occurred predominantly on schizonts, as infected cells of the ring or trophozoite fractions contained only a small number of stained infected cells (< 2%). Antibody binding to only a proportion of infected cells may be due to either the recognition of only a subpopulation of the total YIR proteins expressed or the unequal access of different anti-YIR antibodies to their epitopes. Indeed, the differences between the two anti-YIR antibodies seen in this assay could be due to restricted access of anti-YIR peptide 1 antibodies to their epitope, caused by the close proximity of peptide 1 to a predicted hydrophobic region of the protein. The binding of YIR antisera to the surface of pRBC could be inhibited with YIR peptide in a concentration-dependent manner, confirming the specificity of the antibody binding (Fig. 5B).

Figure 5.

Surface expression of P. yoelii YIR by FACS analysis and indirect immunofluorescence assay of live cells.
A. Schizont (SZ), trophozoite (T) and ring (R) stage enriched live pRBC from infected RAG2–/– mice were labelled with (i) affinity-purified mouse anti-peptide 1 serum (α-pep1) or affinity-purified control mouse serum (control 1) and (ii) mouse anti-peptide 2 (α-pep2) serum or control mouse serum (control 2). Mouse antibodies were detected with FITC-conjugated goat anti-mouse IgG+M. Parasite nuclei were stained with Hoechst dye. RBCs were gated on forward and side scatter (> 400 000 gated events collected). Numbers shown represent the percentage of YIR-positive pRBC, excluding uninfected RBCs. Stage composition in the two experiments are as follows (i) SZ fraction: 57% SZ, 41% T, 1% R, 2% uninfected RBC; T fraction: 60% T, 6% SZ, 13% R, 21% uninfected RBC; R fraction: 31% R, 3% T, 0% SZ, 66% uninfected RBC; and (ii) SZ fraction: 73% SZ, 23% T, 1.5% R; T fraction: 90% T, 5% SZ, 1.3% R; R fraction: 60% R, 0% T, 0% SZ, 40% uninfected RBC. FACS data plotted on a logorithmic scale.
B. Inhibition of surface staining by FACS with peptide. Surface staining of schizont and trophozoite enriched live pRBC by FACS was inhibited with 5 × 10−6 to 5 × 10−10 mol peptide 2. Peptide 2 and an irrelevant peptide (CNPHYHNDPELKEII) were dissolved in glacial acetic acid at various concentrations and coated to 96-well plates by evaporation. Affinity-purified mouse anti-peptide 2 serum was pre-incubated with the peptides, before 2 × 106 pRBC were added and flow cytometry was performed as described before. The reduction in numbers of YIR-positive cells is expressed as a percentage of inhibition. One representative experiment of three, performed in duplicate, is shown here. Error bars indicate standard deviation of the mean.
C. YIR surface stained live pRBCs are intact. Multiple surface staining was performed to differentiate intact (α-MSP-1-negative) and leaky (α-MSP-1-positive) pRBCs. Schizont-enriched pRBCs were labelled with affinity-purified rabbit anti-peptide 2 antibodies, followed by PE-conjugated swine anti-rabbit antibodies and mouse anti-MSP-1 antibodies (clone 25.1) followed by PE-conjugated goat anti-mouse IgG to detect intracellular MSP-1 in leaky pRBCs. As controls mouse IgG (Sigma) and irrelevant affinity-purified rabbit antibodies were used. Gate was set on forward and side scatter and HOECHST-positive pRBCs (panel 1). Only a small number of pRBC reacted with anti-MSP-1 antibody and are therefore leaky. The majority of YIR-positive pRBC are MSP-1-negative (panel 2), verifying that YIR is detected on the surface of the pRBC, and is not intracellular. Controls shown are anti-YIR peptide 2 in combination with purified mouse IgG (panel 3) and anti-MSP-1 in combination with anti-rabbit irrelevant control antibodies (panel 4).
D. Indirect immunofluorescence on live cells. Cells were stained sequentially with HOECHST and rabbit anti-YIR peptide antibodies, mouse anti-MSP-1 antibody, normal mouse serum or serum from P. yoelii immune mice (RIS), followed by FITC-conjugated anti-mouse IgG+M or FITC-conjugated anti-rabbit IgG. Images were captured on an Axioplan 2 imaging system (Zeiss). Panels 1–4 show pRBC stained with antibodies to YIR peptide 2, MSP-1, immune serum and normal mouse serum(ns), respectively, and the corresponding HOECHST-stained parasite nuclei, bright field image and FITC/HOECHST overlay, as marked. Anti-YIR antibody and immune serum only stain the surface of the intact cell.
E. Indirect immunofluorescent staining of acetone-fixed pRBC with anti-MSP-1 antibody. Acetone-fixed cells were stained with anti-MSP-1 antibody, as described above.

To exclude the possibility that the observed staining of live cells was of intracellular origin, due to the fragility of the schizont-infected cells, these cells were stained sequentially with both anti-YIR peptide 2 antibodies (rabbit) and with a monoclonal antibody to merozoite surface protein 1 (MSP-1). MSP-1 is associated with the surface of the intraerythrocytic parasites and not the surface of the infected cell (Smythe et al., 1988). Figure 5C shows that most of the pRBC population which stained with anti-YIR antibodies did not react with anti-MSP-1 antibody and only a minor percentage reacted with both anti-MSP-1 and anti-YIR antibodies (0.9% of the total pRBC population, compared with 21.2% staining with anti-YIR antibodies) confirming that the observed YIR staining occurs on the surface of the pRBC, rather than intracellularly. In addition, live pRBC stained with anti-YIR antibodies by immunohistochemistry showed clear staining of the cell surface (Fig. 5D, FITC, α-pep2). This was similar to the staining pattern observed with immune serum (Fig. 5D, FITC, RIS) but contrasted with the complete lack of staining of these cells seen with anti-MSP-1 antibody (Fig. 5D, FITC, α-MSP-1), although this antibody stains 100% fixed merozoites within schizonts (Fig. 5D, FITC). The specific surface expression shown here indicates that YIR proteins may be targets of the antibody response.

Our results, showing that YIR is expressed at the surface of live pRBC, that many different yir are transcribed and that yir transcription patterns change in response to host immunity, are all consistent with YIR being involved in antigenic variation of P. yoelii. Two mechanisms have been proposed for antigenic variation: spontaneous switching, described both in vitro and in vivo (Roberts et al., 1992; Brannan et al., 1994), followed by reduction of antigen by antibody, or antibody-induced switching, which would involve de novo activation of silent variants, as described for SICA and SICAvar (Brown and Brown, 1965; Galinski and Corredor, 2004) of P. knowlesi. We have shown that some spontaneous change in yir transcription can occur during the course of an infection but that transcriptional changes are modulated by the host immune response. As the alternative yir variants found in the re-infected semi-immune animals were also present in low amounts in the starter population, our data suggest that selection plays a major part in yir variation in vivo. In the case of the var genes of P. falciparum, the spontaneous transcriptional switching rate is constant for a particular variant but each variant has a different rate suggesting that the switching rate is predetermined for each var gene (Horrocks et al., 2004). Here, we show multiple transcripts within the population and do not detect changes in the absence of an immune response. In the immune host a shift in the balance between the different variants occurs, although the initial transcripts are still present. This change could be indicative of a shift away from a particular transcript set to an alternative set. Mathematical models suggest that immune responses to minor epitopes, which give rise to sequential dominance of different minor epitopes, may provide a means for persistence of malaria infections (Recker et al., 2004). It is feasible that members of the yir family may constitute minor epitopes, whereas clone-specific immunity to other polymorphic antigens may ultimately result in parasite clearance (Paget-McNicol et al., 2002). In addition, while immunity to P. falciparum develops with age (Bull et al., 1998), people of all ages develop P. vivax-associated disease (Mendis et al., 2001). It is possible that the sheer size of the vir multigene family, absent from the P. falciparum genome, coupled with low transmission rates, contributes to this lack of age-related immunity, by providing a means for greater antigenic variation.

The function of these clonally variant antigens is still unknown, although the most widely studied family, PfEMP1, can mediate parasite adhesion to host endothelium, sequestration in vital organs and thus pathology (Fernandez et al., 1998; Smith et al., 1998). There is no evidence for sequestration of P. yoelii in this system, nevertheless the function of the VIR/YIR family of surface antigens can now be investigated in vivo in the rodent model, through the use of genetic manipulation, transfection methods, recombinant proteins and specific immunological reagents.

During the course of their evolution Plasmodium species have developed both shared and unique mechanisms to circumvent host immunity. var and SICAvar genes are unique to their species. In contrast, it is clear from our data and from other studies (del Portillo et al., 2001; Janssen et al., 2002; Fischer et al., 2003) that vir genes and yir/cir/bir of the rodent parasites all belong to a superfamily with shared ancestry and function. It has also been suggested that the rif and stevor gene families may be related to these genes (Janssen et al., 2004). The importance of the pir superfamily to the biology of malaria parasites is supported by the proportion of their genomes dedicated to these genes. Here, we applied a focused approach to the identification of transcript changes associated with immunomodulation as, with such a large gene family, several members may be expressed at the same time in any one parasite population, as shown here at the transcriptional level. However, it is possible that not all transcribed genes are translated into protein, or that the number of genes transcribed may change throughout the life cycle, as found in the case of var, where multiple transcripts are found in early asexual stages and a dominant full-length transcript is found in late-stage parasites (Chen et al., 1998; Taylor et al., 2000). Studies of transcriptional regulation, and of transcription within single parasites at each stage of the asexual life cycle, are currently under way and will be the subject of further papers (J. Fonager et al., in preparation). However, proteomic studies have shown that several BIR proteins are indeed expressed at each stage of the life cycle of P. berghei and, furthermore, it has been shown that distinct repertoires of BIR proteins were exclusive to a particular stage of parasite development and that 7% of these proteins were found only in mosquito stage parasites, suggesting that functional differences in BIR subsets occur (Hall et al., 2005). This has parallels with the phenotypic antigenic variation shown for another multigene family in P. yoelii, py235, where sequential transcription of different sets of genes in different stages of the parasite life cycle has been demonstrated (Preiser et al., 2002). Indeed, it has been speculated that subfamilies of VIR proteins have completely different functions in P. vivax infections, i.e. in protecting parasites from macrophage destruction or as a specific receptor/ligand for barrier cells of the spleen (Del Portillo et al., 2004). However, the exact function of these genes has yet to be determined, but the differences in the repertoire of transcribed genes seen in the semi-immune host compared with the non-immune/immunocompromised host, coupled with the expression of the YIR protein on the infected cell surface, do suggest a role in antigenic variation. Also, if the majority of transcribed genes are translated into protein, the sheer number of YIR proteins expressed could provide an efficient mechanism of evasion of host immunity, particularly if each individual cell expresses (a) different gene(s). Transcriptional change is a key precursor to changes in protein expression and paves the way for the study of phenotypic variation of these antigens.

Experimental procedures

Bioinformatics

tblastn (Altschul et al., 1997) searches of preliminary P. yoelii yoelii (P. yoelii) genome sequence data at The Institute for Genomic Research (http://www.tigr.org/tdb/e2k1/pya1/) were used to identify P. yoelii vir gene homologues (yir). Open reading frames were identified using Glimmer M (http://www.tigr.org/software/glimmerm) and intron/exon structure deduced by searching for the highly conserved splice sites, excision of putative introns followed by translation into protein. The deduced sequence was verified by RT-PCR using primers designed to span the introns (P1F: ATGGATCAT CACCGG exon 1; P1R: CATTTGATAAAGTACACAATACTTG exon 2 and P2F: CGATAAAATTAATGCTGGATGTTTA exon2; P2R: GTTTTTGAAATCGTTTCCG exon3), followed by DNA sequencing, and by searching EST databases (http://www.plasmodb.org) for sequences spanning the splice sites. Sequence alignments of DNA and protein sequences were performed using clustal_x (Thompson et al., 1997).

Primer set 3 consisted of two different forward primers, used in independent polymerase chain reactions (PCRs), designed to very conserved sequences, taken from the same region of the gene {P3Fa: ATATGGTTAAGTTATATGTTAA ACC [∼300 hits (http://tigrblast.tigr.org/tdb/e2k1/pya1/) with zero to two mismatches] or P3Fb: ATGGTTAAGTTACAAAT TAAACCAAAA [∼120 hits (http://tigrblast.tigr.org/tdb/e2k1/pya1/) with zero to three mismatches]} and a reverse primer located across the exons 2–3 splice site (P3R: CCAAATAAC GAATACTTATAAGAAATTC). P3 primers have the potential to amplify a broad transcript repertoire. Primer set 4 amplifies a more limited transcript repertoire (13 exact matches and ∼27 with a single mismatch) (forward primer located across the exons 1–2 splice site, P4F: GGATAAAGACGTGTGTG; P4R: CCTATCGACGAACTTG).

Analysis of the predicted protein sequences was performed using the integrated protein analysis package PIX. Transmembrane regions were predicted using all four transmembrane prediction programs contained within PIX (tmpred, tmap, das and phd). Synthetic peptides [ISAG CLYLLDEFIKDC (peptide 1), amino acids 52–67, PY06755 and KLYDALQSLCNMYNEF (peptide 2), amino acids 149–164, PY00500) (hydrophobicity: peptide 1 > peptide 2)] were designed to conserved regions of the predicted protein sequences (Fig. 1A).

All putative yir cDNA sequences in the P. yoelii genome were downloaded from TIGR (http://www.tigr.org/tdb/e2k1/pya1/). Pseudogenes, partial genes and the highly conserved exon 3 were removed from the data set. A total of 720 YIR protein sequences were collected in a non-redundant data set. Sequences were aligned with clustalw version 1.82 (33) using the following parameters (slow/accurate alignment; weight matrix = gonnet; pairwise gap opening penalty = 10; pairwise gap extension penalty = 0.10; multiple gap opening penalty = 10; multiple gap extension penalty = 0.20). The global multiple alignment was used to construct an unrooted tree using the neighbour-joining method as implemented in clustalw version 1.82. A custom program was produced to manipulate the unrooted tree to provide information of specific sequences. The global multiple alignment was also used to reconstruct a tree using parsimony as implemented by the program phylip. Phylogenetic and molecular evolutionary analyses were conducted using MEGA version 2.1 (Kumar et al., 2001). Bootstrapped tests of phylogeny were performed (1000 replications; neighbour-joining or minimum evolution) to validate the tree.

Mice and parasites

Male and female BALB/c mice, BALB/c mice with a targeted disruption of the RAG2 gene (RAG2–/–) (Shinkai et al., 1992) and female C57BL/6 mice were bred and maintained in filter racks on sterile bedding, food and water at the National Institute for Medical Research (NIMR).

Plasmodium yoelii 17XA parasites were derived from a cloned line (A) provided by Dr D. Walliker, as described previously (Jarra and Brown, 1989). Parasitaemia was monitored by daily microscopic analysis of methanol-fixed Giemsa-stained thin blood smears. BALB/c and RAG-2–/– mice were infected intraperitoneally (i.p.) with P. yoelii. The different erythrocytic stages of P. yoelii were obtained from a discontinuous Nycoprep gradient (Nycomed Oslo, Norway) composed of equal volumes of 50%, 60%, 70% and 80% (in RPMI 1640 medium) of a 27.6% Nycodenz working solution in Tris-HCl pH 7.4. The relative numbers of cells, by stage, in each band was determined from Giemsa-stained thin blood films.

Antisera

Peptides were conjugated to KLH (Imject® Maleimide Activated Mariculture keyhole limpet haemocyanin; Pierce) and to cBSA (Imject® Maleimide Activated SuperCarrier® Immune Modulator) according to the manufacturer's instructions and polyclonal antisera to these peptides were raised in BALB/c mice using standard procedures [i.e. five mice immunized i.p. with 50 µg of KLH- or cBSA-conjugated peptide in a 200 µl solution of MPL® + TDM adjuvant (Sigma-Aldrich)] five times at 3-week intervals. Control mice were immunized with adjuvant only. Similarly, anti-YIR peptide antiserum was raised in New Zealand White rabbits immunized subcutaneously (Harlan Sera-Laboratory, Loughborough, UK) with 200 µg of KLH-conjugated peptide. Monoclonal antibody to P. yoelii MSP-1 (clone 25.1) (Holder and Freeman, 1981) was a gift from Dr A.A. Holder. P. yoelii immune serum was obtained from BALB/c mice infected with 1 × 104 pRBC derived from a standard starter population of P. yoelii 17XA, spontaneous recovery followed by re-infection with 1.3 × 108 pRBC 11 weeks later and collection of serum 9 days after the re-infection.

Antibody purification

HiTrap NHS-activated HP columns from Amersham Biosciences were coupled with 2 mg of peptide according to the manufacturer's instructions. Serum was diluted (1:5) with PBS and bound to the column. The column was washed with 10 volumes of PBS and antibodies were eluted with 0.2 M Glycine, 0.15 M NaCl, pH 2.7. Antibody containing fractions were dialysed against PBS and concentrated with Vivaspin concentrators (Vivascience).

RNA extraction and RT-PCR

Total RNA was extracted (Kyes et al., 2000) and cDNA synthesised using superscriptTM II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. PCR amplifications were in 20 mM Tris pH 8.4, 50 mM KCl, 200 µM dNTP mix, 10 µM each primer, 2 µ Amplitaq Gold® with 4 mM MgCl2, 11 min 95°C, 40 PCR cycles (94°C 50 s, 56°C 50 s, 67°C 60 s) (primer set 3), or 2 mM MgCl2 (94°C 50 s, 44°C 50 s, 67°C 60 s) (primer set 4) followed by 15 min at 72°C.

SDS-PAGE and Western blotting

Parasites were solubilized in Triton X-100 and SDS buffer in the presence of protease inhibitors (Blackman et al., 1994). Proteins were resolved by electrophoresis through NuPAGE 4–12% acrylamide gels in MES buffer (Invitrogen) under reducing conditions. Markers were broad-range pre-stained protein standards (Bio-Rad) and Seeblue2 (Invitrogen). Proteins were transferred onto Hybond C extra nitrocellulose membrane (Amersham Pharmacia), as described previously (Kaviratne et al., 2002). Specific proteins were detected using (i) affinity-purified antibody peptide 1 or (ii) unpurified antibodies from mouse polyclonal antisera raised against peptide 2, followed by horseradish peroxidase-conjugated goat anti-mouse IgG (1:2000) (Bio-Rad) and enhanced chemiluminescence (Pierce). Controls were (i) purified IgG from mouse serum (Sigma-Aldrich) or (ii) unpurified normal mouse serum.

Immunofluorescence

Indirect immunofluorescence assays were performed on acetone fixed thin blood smears of a schizont-enriched preparation, prepared on 50% Nycoprep. Primary antibodies were anti-peptide antisera or normal plasma. Secondary antibodies were goat anti-mouse IgG-FITC (Sigma). Parasite nuclei were stained with 4′-6-Diamidino-2-phenylindole (DAPI). The slides were viewed at a magnification of ×100 with an Olympus Delta Vision imaging system. Surface immunofluorescent staining was performed as described by Gilks et al. (1990). Erythrocytes from P. yoelii 17XA-infected or normal RAG2–/– mice were collected into heparin and washed twice in cold Krebs solution (PBS containing 110 mM NaCl, 4.5 mM KCl, 1.1 mM MgSO4, 11 mM glucose). Cells (2 × 106 ml−1) were centrifuged (85 g, 3 min, 4°C) and incubated in HOECHST33342 [10 µg ml−1 in PBS containing 1% BSA, 5 mM EDTA and 0.01% NaN3 (PBNE)] for 10 min at 4°C. After three washes in PBNE, stained cells were incubated sequentially with 20 µl of anti-YIR peptide antibodies, anti-MSP-1 antibody, normal mouse serum or immune serum followed by FITC-conjugated anti-mouse IgG+M (BD) appropriately diluted in PBNE. Final cell pellets were mounted under coverslips in Citifluor (Citifluor, London).

The slides were examined and images captured on an Axioplan 2 imaging system (Zeiss).

Flow cytometry

Erythrocytes from P. yoelii 17XA-infected or normal RAG2–/– mice were collected into heparin and washed twice in cold Krebs solution. Erythrocytes (2 × 106) were incubated sequentially with 20 µl of anti-YIR peptide antibodies (mouse or rabbit) or control mouse sera and FITC-conjugated anti-mouse IgG+M (BD) appropriately diluted in PBNE. Sequential staining was performed with affinity-purified rabbit anti-peptide 2 antibodies followed by mouse anti-MSP-1 (clone 25.1) monoclonal antibody, in combination with FITC-conjugated swine anti-rabbit IgG (DAKO) and PE-conjugated goat anti-mouse IgG (BD). Rabbit affinity-purified antibodies raised to an irrelevant peptide and purified mouse IgG (Sigma) were used as controls. Parasite nuclei were stained with HOECHST33342 (10 µg ml−1) for 5 min. Cells were washed three times in PBNE and fixed in 2% paraformaldehyde in PBS. Cells (> 400 000) were gated on forward and side scatter. FACS acquisition was performed on an LSR-FACS (BD) and analysed using the FloJo software package (Tree Star, San Carlos, CA).

Peptide inhibition

Peptide 2 and the control peptide (CNPHYHNDPELKEII) were dissolved at a concentration of 1 × 10−1, 1 × 10−2, 1 × 10−3, 1 × 10−4, 1 × 10−5, 1 × 10−6, respectively, in glacial acetic acid (50 µl) and transferred into 96-well plates (Corning, NY). The plates were used once the glacial acetic acid had evaporated. Affinity-purified peptide 2 antiserum (50 µl) was pre-incubated in the peptide-coated plates for 20 min before 2 × 106 pRBC were added and flow cytometry was performed as described above.

Infections of mice

Primary infection in BALB/c mice.  Three female BALB/c mice were infected i.p. with 5 × 103 pRBC derived from a standard starter population (a) of P. yoelii 17XA (see Fig. 3). Blood samples for RNA extraction and transcription profiling were collected at days 12 and 18 p.i.

Infection and re-infection of BALB/c mice.  Mice were infected as described above. Thirteen days p.i., when approximately 50% of the erythrocytes were infected, blood samples were collected and one aliquot immediately dissolved in lysis buffer for RNA extraction (b) and a second subinoculated into RAG2–/– mice i.p. to expand the parasite population. The BALB/c mice were then treated with 5 mg per kilogram of pyrimethamine (Sigma-Aldrich) on days 13–19 to eliminate the infection. No further parasites were observed. On day 34 the mice were re-infected with 109 homologous parasites i.p. The ensuing low parasitaemia was monitored daily and was rapidly cleared 3–4 days after challenge at which time blood was subinoculated into RAG2–/– mice to expand the subpatent parasite populations. RNA was extracted from pRBC from these subinoculated RAG2–/– mice (c).

Infection and re-infection of RAG2–/– mice.  Mice were injected i.p. with 5 × 103 parasites derived from the starter population (a). Parasitaemia was monitored daily. Between days 7 and 13 animals were treated daily with 5 mg per kilogram of pyrimethamine. In one group of mice the parasites spontaneously recrudesced on days 20 and 21; RNA was extracted from pRBC (day 24) (d) for RFLP analysis and sequencing. In other mice parasites were not detected up to day 32 p.i., at which time these mice were re-infected with the starter population (1 × 109 pRBC). Parasites were detectable over the ensuing 3 days and were then collected for RNA extraction (e), transcription profiling and sequencing.

Transcriptional profiling by RFLP of RT-PCR products

RNA was extracted from the samples collected in all three infection series and reverse-transcribed, as described above. PCR primers were set 4, the exon 1/2 splice site set, amplifying a more limited transcript repertoire. The reverse primer was end-labelled with 32P using T4 polynucleotide kinase (New England Biolabs), purified through a microspin G25 column (Amersham Pharmacia) and used to prime a PCR amplification. PCR products were gel-purified (Qiagen gel purification kit) and digested overnight with the restriction enzymes Ssp 1 or Dra 1 (Roche) according to the manufacturer's instructions. Digestion products were electrophoresed through a 5% polyacrylamide gel at 25 V cm−1, dried onto 3M filter paper and exposed to autoradiographic film (Kodak Biomax). Experiments and digestions were performed in duplicate to ensure reproducibility of the results.

Sequencing

Gel-purified RT-PCR products were cloned and sequenced as described previously (Preiser and Jarra, 1998).

Acknowledgements

This work was funded in part by the European Commission contract QLRT-2001-00853 and the Medical Research Council UK and the authors would like to thank John Skehel, Avrion Mitchison and Michael Blackman for critical review of this manuscript.

Ancillary