Protective Immune Responses to Apical Membrane Antigen 1 of ...

INFECTION AND IMMUNITY, Aug. 1996, p. 3310–3317 0019-9567/96/$04.0010 Copyright q 1996, American Society for Microbiology

Vol. 64, No. 8

Protective Immune Responses to Apical Membrane Antigen 1 of Plasmodium chabaudi Involve Recognition of Strain-Specific Epitopes PAULINE E. CREWTHER,1,2* MARY L. S. M. MATTHEW,1,2 ROBERT H. FLEGG,2 1,2 AND ROBIN F. ANDERS Cooperative Research Centre for Vaccine Technology1 and The Walter and Eliza Hall Institute of Medical Research,2 Post Office Royal Melbourne Hospital, Victoria 3050, Australia Received 15 March 1996/Accepted 24 May 1996

Apical membrane antigen 1 (AMA-1), an asexual blood-stage antigen of Plasmodium falciparum, is an important candidate for testing as a component of a malaria vaccine. This study investigates the nature of diversity in the Plasmodium chabaudi adami homolog of AMA-1 and the impact of that diversity on the efficacy of the recombinant antigen as a vaccine against challenge with a heterologous strain of P. chabaudi. The nucleotide sequence of the AMA-1 gene from strain DS differs from the published 556KA sequence at 79 sites. The large number of mutations, the nonrandom distribution of both synonymous and nonsynonymous mutations, and the nature of both the codon changes and the resulting amino acid substitutions suggest that positive selection operates on the AMA-1 gene in regions coding for antigenic sites. Protective immune responses induced by AMA-1 were strain specific. Immunization of mice with the refolded ectodomain of DS AMA-1 provided partial protection against challenge with virulent DS (homologous) parasites but failed to protect against challenge with avirulent 556KA (heterologous) parasites. Passive immunization of mice with rabbit antibodies raised against the same antigen had little effect on heterologous challenge but provided significant protection against the homologous DS parasites. recovered without drug treatment after experimental challenge with P. fragile. The immunized monkeys experienced delayed onset of infection and reduced peak parasitemias compared with control monkeys, all of which required drug treatment. In vaccine trials using mice, immunization with the AMA-1 ectodomain (AMA-1B), produced in Escherichia coli and refolded in vitro, protected against challenge with Plasmodium chabaudi parasites (1). The degree of conservation of AMA-1 sequences implies a conserved function for this molecule in different species of Plasmodium. Pairwise comparisons between all the known amino acid sequences of AMA-1 homologs indicate greater than 50% identity, with 16 cysteine residues conserved in all sequences (2, 11, 17, 18, 24, 30). All of these cysteines are found in the ectodomain of the molecule, which is stabilized by eight intramolecular disulfide bonds (12). AMA-1 lacks the sequence repeats and marked polymorphisms observed in other malaria antigens such as the merozoite surface antigens MSP-1 and MSP-2. However, limited sequence variation, resulting from point mutations, is observed among alleles of AMA-1 in P. falciparum (18, 23, 28), P. knowlesi (31), and Plasmodium vivax (2). The impact of this diversity on the potential of AMA-1 to provide protection against natural infection needs to be assessed. The availability of virulent and avirulent P. chabaudi strains which infect laboratory mice provides an experimental system for examining the impact of antigen diversity on vaccine efficacy. The diversity observed among the AMA-1 alleles of P. chabaudi adami DS and 556KA and P. chabaudi chabaudi CB is similar to, but more extensive than, that previously observed among P. falciparum alleles (18). We show here that the 36 amino acid substitutions between the DS sequence and the published sequence for 556KA (17) are sufficient to make immunization with the refolded ectodomain of DS AMA-1 ineffectual against challenge with heterologous 556KA para-

Antigens of Plasmodium falciparum located on the surface or in the apical organelles of merozoites offer considerable potential as components of a vaccine against malaria. Apical membrane antigen 1 (AMA-1) of P. falciparum (also called PF83) is a type I integral membrane protein located both in the rhoptries and on the merozoite surface (23, 30). AMA-1 is synthesized in segmenting schizonts as a short-lived 80-kDa precursor which is processed N terminally to a 62-kDa fragment (5). Initially, both the 80- and 62-kDa species are located apically, possibly in the neck of the rhoptry, and the 62-kDa fragment is also released onto the surface of the merozoite (5, 19, 23). These biological features of AMA-1, together with its relative conservation within the genus and its efficacy in preclinical vaccination trials, suggest a role for this molecule in erythrocyte invasion and thus support its candidacy as a vaccine component. AMA-1 has been shown to protect against malaria infection in experimental simian and rodent systems. Monoclonal antibodies to PK66, the Plasmodium knowlesi homolog of AMA-1, or their Fab fragments blocked the invasion of rhesus erythrocytes by P. knowlesi merozoites in vitro (6, 27). Immunization of rhesus monkeys with purified PK66 conferred partial protection against experimental challenge, and antibodies raised to the purified antigen inhibited merozoite invasion in vitro. Antibodies raised to the denatured antigen were not inhibitory (7, 8). Using a different simian experimental system, Collins et al. (3) obtained partial protection by immunization with recombinant Plasmodium fragile AMA-1 produced in a baculovirus expression system. Four of five Saimiri monkeys which had been immunized with the recombinant P. fragile AMA-1 * Corresponding author. Mailing address: The Walter and Eliza Hall Institute of Medical Research, Post Office Royal Melbourne Hospital, Victoria 3050, Australia. Phone: 61-3-9345-2555. Fax: 61-39347-0852. 3310

PROTECTIVE IMMUNE RESPONSES TO AMA-1 OF P. CHABAUDI

VOL. 64, 1996

3311

sites. In addition, rabbit antibodies raised against the same antigen and transferred to mice provided protection against challenge with homologous DS parasites but not against challenge with the heterologous parasites. Thus, the fine specificity of the immune response induced by immunization with AMA-1 appears to be a critical determinant for protection against challenge with P. chabaudi. These observations, together with the distribution and nature of the nucleotide and amino acid substitutions, lead us to conclude that positive selection operates on antigenic sequences in AMA-1. MATERIALS AND METHODS Parasite strains and maintenance in mice. P. chabaudi adami DS and DK and P. chabaudi chabaudi CB were obtained from D. Walliker, University of Edinburgh. P. chabaudi adami 556KA, originally from F. Cox (University of London), was obtained from I. Clark, Australian National University. DK is a cloned line derived from 556KA. Of the two strains used in vaccine trials, DS is a virulent strain and 556KA is avirulent. Parasites were maintained by routine passage in mice for no more than five passages. Cloning. Nucleotide sequences corresponding to full-length AMA-1 (AMA1A) and its ectodomain (AMA-1B) from P. chabaudi adami were amplified as fragments lacking the signal sequence from genomic DS DNA by using Pfu DNA polymerase. Similarly, the AMA-1B sequence was amplified from genomic 556KA DNA. The site for cleavage of the signal sequence was predicted by reference to the work of von Heijne (29) and the published P. chabaudi AMA-1 sequence (17). It was subsequently found, by N-terminal sequence analysis, that AMA-1 expressed in insect cells by using a baculovirus vector was cleaved after residue 21 rather than residue 24 as predicted (12). Oligonucleotide primers, consisting of nucleotides 73 to 91 (59) and 1662 to 1677 (39; AMA-1A) or 1422 to 1437 (39; AMA-1B), were used for amplification of the relevant fragments with BamHI and PstI sites at the 59 and 39 ends, respectively. The products were digested with BamHI and PstI, purified after agarose gel electrophoresis with the Magic PCR Preps Purification System (Promega), ligated into pDS56/RBS11/ 6XHis (Roche), and transformed into E. coli JPA101. Bacterial colonies expressing AMA-1A and AMA-1B as fusion proteins with 6 histidine residues at the N terminus were identified by the reactivity of the recombinant protein with hyperimmune mouse serum raised by repeatedly infecting mice with P. chabaudi adami. DNA sequencing and sequence analyses. The sequences of selected recombinants were established by using the PRISM DyeDeoxy Terminator Cycle Sequencing Kit (Applied Biosystems, Inc., Foster City, Calif.) and a series of overlapping oligonucleotides designed by reference to the published sequence from 556KA (17). Sequencing reaction products were analyzed on a model 373A automated DNA sequencer (Applied Biosystems). To verify that the cloned DS AMA-1 sequences did not contain any PCR misincorporations, the correct sequence was established by direct sequencing of the full-length DS AMA-1 gene or fragments thereof, amplified from genomic DS DNA by the PCR using Pfu DNA polymerase. The complete sequence of the AMA-1 gene from strain 556KA and partial sequences from isolates CB and DK were also established in the same way. All sequences were determined for both strands. The DS sequence requires verification of the extreme 59 and 39 ends, which were inevitably fixed in the PCR by the sequence of the oligonucleotide primers. The SeqEd editing program (version 1.0.3; Applied Biosystems) and the Sequence Analysis Software Package, version 8 (September 1994), of the Genetics Computer Group (Madison, Wis.) were used for sequence analysis. In pairwise comparisons between alleles of AMA-1, the rates of nucleotide substitution were estimated as the numbers of synonymous substitutions per 100 synonymous sites (dS) and nonsynonymous substitutions per 100 nonsynonymous sites (dN) between different alleles of AMA-1 by the methods of Nei and Gojobori (20) and Ina (15). The standard errors associated with the Nei and Gojobori substitution rates were calculated by the method of Nei and Jin (21). For this analysis, the amino acid sequence was divided into five domains based on the predicted cleavage pattern and the cysteine connectivities established for AMA-1B expressed in insect cells by using a baculovirus expression vector (12) (Fig. 1A). The prosequence (residues 22 to 82) is the N-terminal proteolytic cleavage fragment without the signal sequence (Fig. 1A). Domains I (residues 83 to 253), II (residues 254 to 365), and III (residues 366 to 480) constitute the mature ectodomain. The transmembrane and cytoplasmic domains have been treated as a single domain for the purpose of these calculations, as there are no substitutions in either region. Purification and refolding of the antigen. The procedures for the purification and refolding of recombinant AMA-1B will be described in detail elsewhere. Briefly, washed inclusion bodies were solubilized in 6 M guanidine-HCl–10 mM Tris-HCl–0.1 M sodium phosphate (pH 8.0) and the recombinant protein was purified by chromatography on Ni-nitrilotriacetic acid resin (Qiagen). To initiate refolding, the eluted protein solution was diluted 1:100 in ice-cold, vacuumfiltered 100 mM Tris-HCl–1 mM reduced glutathione (pH 8.0). After incubation on ice for 2 min, oxidized glutathione was added to a final concentration of 0.25 mM and the protein was allowed to refold under nitrogen at room temperature

FIG. 1. Schematic diagrams showing structural features of AMA-1. (A) The cysteine connectivities and putative domain structure (12). The arrow indicates the predicted cleavage site of the N-terminal fragment (prosequence) based on the observed mass of the cleaved molecule. (B) Alignment of sites of nonsynonymous amino acid substitution with sites correlated with synonymous codon changes. (C) Alignment of sites of amino acid substitution between P. falciparum AMA-1 molecules (18) with sites of substitution between AMA-1 molecules of P. chabaudi DS and 556KA.

for 16 h. The refolded AMA-1B was concentrated by chromatography on DEAESepharose. Reduction and alkylation of AMA-1B. Purified and refolded AMA-1B was reduced by incubation at 378C for 30 min in the presence of 10 mM dithiothreitol and alkylated by the addition of iodoacetic acid to a final concentration of 50 mM. Production of antibodies. Rabbits were immunized with 200 mg of refolded AMA-1B emulsified in complete Freund’s adjuvant and injected intramuscularly into four sites. Three subsequent boosters of 200 to 270 mg of antigen with incomplete Freund’s adjuvant were given at intervals of 1 to 2 months. Immunoblots. The immunoblot procedure was essentially as described by Crewther et al. (4). In some experiments ECL (Amersham Life Science) was used instead of 125I-protein A for the detection of antigen-antibody complexes, according to the manufacturer’s instructions. Vaccine trials. Refolded or reduced and alkylated AMA-1B was emulsified with Montanide ISA 720 adjuvant (Seppic) and used to immunize female BALB/c (H-2k) mice. Each mouse was injected intraperitoneally with a total of 20 mg of protein on three occasions at 4-week intervals. Mice were challenged with 106 infected erythrocytes (IRBC) 10 days after the third immunization. IRBC for challenge were obtained from mice with parasitemias of between 5 and 20%. Prior to challenge, mice in the immunized and control groups were ear clipped with a code number and randomly assigned to groups of five to eight in different boxes. Parasitemias were monitored by microscopic examination of Giemsa-stained thin blood films. When mice exhibited symptoms of malaria, they were examined twice daily and moribund mice were killed. Despite this policy, some mice died in the period between observations. For the transfer of antibodies to parasite-infected mice, immunoglobulin G (IgG) was isolated from the sera of rabbits immunized with refolded DS AMA-1B or from the sera of control rabbits by affinity chromatography on protein A-Sepharose (Pharmacia/Amrad). Three milligrams of IgG was administered to mice, intravenously or intraperitoneally, at day 5 after challenge, when the parasitemias were between 2 and 15%. Nucleotide sequence accession numbers. Nucleotide sequence data reported in this paper have the following Genbank accession numbers: U49743 (DS), U49744 (CB), U49745 (DK), and M25248 (556KA).

RESULTS There is limited diversity among P. chabaudi sequences. The complete sequences of the AMA-1 genes from strains DS and 556KA and partial sequences from the variable regions of the

3312

CREWTHER ET AL.

INFECT. IMMUN.

FIG. 2. Alignment of predicted amino acid sequences of AMA-1 molecules from P. chabaudi adami DS, 556KA, and DK and P. chabaudi chabaudi CB. The 16 conserved cysteine residues are boxed. Residues correlated with synonymous codon changes are denoted by lowercase letters. #, DK start; p, CB start; §, CB and DK end.

genes from the DK and CB strains are shown in Fig. 2. These results indicate that the published sequence is from strain 556KA rather than from DS as reported (17). There are 79 single base substitutions in the nucleotide sequence of the DS AMA-1 gene compared with that of 556KA, all in the sequence encoding the ectodomain. Of the 55 resulting codon changes, 36 give rise to amino acid substitutions, a degree of divergence slightly greater than that seen between the two most different P. falciparum alleles (18). The partial sequence for the DK AMA-1 gene (nucleotides 112 to 1133) is identical to the 556KA sequence. The CB sequence (nucleotides 151 to 1133) differs from the DS sequence in 64 sites and from the 556KA sequence in 23 sites, resulting in 45 and 19 codon changes, respectively. A comparison of the deduced amino acid sequence of AMA-1 from strain CB with those of the proteins from the DS and 556KA strains revealed 30 and 13 amino acid substitutions, respectively. All of the cysteine and tryptophan residues and most of the proline and tyrosine residues are conserved in all sequences. Amino acid substitutions are clustered in the AMA-1 ectodomain. The general distribution of the amino acid substitutions in P. chabaudi AMA-1 is similar to that already observed in the AMA-1 sequences from other Plasmodium species, with the domain I sequences, particularly between the first and third cysteine residues, being the most diverse (Fig. 1C). A second relatively hypervariable region, located around

the eighth cysteine residue in domain II, appears to be unique to P. chabaudi (Fig. 2). An alignment of AMA-1 sequences from five species indicates that only certain sites in the first hypervariable region will tolerate an amino acid substitution and that at most of these sites only two alternative amino acid residues are found within a species (data not shown). In the domain II hypervariable region unique to P. chabaudi only 4 of 15 consecutive codons are conserved in all three P. chabaudi sequences examined. There are two potential N-linked glycosylation sites in this region of the 556KA sequence, and the clustered substitutions abolish both of these sites in the CB and DS sequences. Synonymous and nonsynonymous substitutions have similar distributions. In order to gain insight into the evolutionary mechanisms responsible for the observed pattern of substitutions, the rates of synonymous and nonsynonymous substitution were calculated in pairwise comparisons between the different alleles of P. chabaudi AMA-1 for various regions of the molecule (Table 1). This analysis did not show a bias towards nonsynonymous substitutions, as has been reported for the P. falciparum AMA-1 (14). The synonymous rate, dS, was greater than the nonsynonymous rate, dN, in all regions except the prosequence, for which a single nonsynonymous change was recorded, and the transmembrane and cytoplasmic domains, in which there were no mutations. The distributions of synonymous and nonsynonymous substitutions, however, were similar


VOL. 64, 1996

3313

TABLE 1. Rates of synonymous and nonsynonymous nucleotide substitution Result forb: Region(s) (no. of codons compared) and calculation methoda

Prosequence (61) NG I Domain I (171) NG I Domain II (112) NG I Domain III (115) NG I Transmembrane and cytoplasmic domains (78) NG I

DS vs 556KA

DS vs CB

dS

dN

0.0 0.0

0.7 6 0.7 0.6 6 0.6

17.8 6 4.6 13.6 6 3.6

556KA vs CB

dS

dN

dS

dN

8.2 6 1.5 9.1 6 1.7

19.0 6 4.8 14.5 6 3.7

8.5 6 1.5 9.2 6 1.6

2.0 6 0.1 0.0

0.5 6 0.7 0.0

21.5 6 6.5 14.3 6 4.3

3.8 6 1.2 4.3 6 1.4

11.8 6 4.5 8.7 6 3.4

2.6 6 1.0 2.9 6 1.1

10.8 6 4.3 8.1 6 3.2

4.4 6 1.3 4.8 6 1.4

5.1 6 3.0 4.5 6 2.7

1.8 6 0.8 1.9 6 0.9

0.0 0.0

0.0 0.0

a NG, Nei and Gojobori (20) method I, 1-parameter Jukes and Cantor model (standard errors calculated by the method of Nei and Jin [21]); I, Ina (15) method I, 2-parameter Kimura model. b Data are rates 6 standard errors.

(Fig. 1B). Both dS and dN were substantially higher for domains I, II, and III than for the prosequence or the transmembrane and cytoplasmic domains, with dS values being particularly high for domains I and II. Although domain I is the most variable region of AMA-1, the CB and 556KA domain I sequences were nearly identical, as indicated by the low values for both dS and dN for this pairwise comparison. In contrast, the CB and 556KA domain II sequences were as divergent as any of the other pairwise comparisons for this region of AMA-1. Recurrent mutations generate radical amino acid substitutions. The elevated rates of both synonymous and nonsynonymous mutation in domains I, II, and III are associated with multiple substitutions in many codons (Fig. 3). Of the 55 codons which have sustained changes in the DS sequence compared with that of 556KA, there are 5 codons in which the nucleotides in all three positions have been mutated and 14 codons in which two nucleotides have been mutated. This clearly indicates recurrent mutations within the same codon. For example, asparagine at residue 283 (nucleotides 847 to 849) is encoded by AAC in 556KA and by AAT in DS, a synonymous third-base substitution. In CB there has also been a nonsynonymous substitution in the second position of the codon, resulting in AGT, coding for serine. Similarly, residue 154 (nucleotides 460 to 462) is leucine in 556KA (TTG) and CB (TTA) and isoleucine in DS (ATA). Of the 36 amino acid substitutions between the DS and 556KA sequences, 19 result in a change in charge, and the distribution of these radical substitutions is nonrandom. For example, between the first and second cysteine residues there are 10 radical amino acid substitutions, including all six substitutions in the 14-residue sequence from position 105 to position 118 (Fig. 2, underline). Despite the radical nature of the individual amino acid substitutions, overall, the electrostatic properties of the domains were conserved. The pI of domain I, estimated from the amino acid composition, was 7.90 in 556KA and CB and 7.56 in DS. The pI of domain II was 9.13 in DS and CB and 9.26 in 556KA, and the pI of domain III was 4.98 in CB and 4.90 in 556KA. Immunization with recombinant AMA-1 protects against homologous but not heterologous challenge. Mice were immu-

nized with the refolded ectodomain of DS AMA-1 (AMA-1B) and challenged with either virulent DS or avirulent 556KA parasites (Table 2). Immunization with refolded DS AMA-1B conferred no protection against heterologous challenge with 556KA parasites, but it did protect against the homologous strain, DS. DS and 556KA infections in mice immunized with reduced and alkylated AMA-1B were unchanged from those in mice receiving adjuvant alone. The range of peak parasitemias observed in the protected mice in these experiments was higher than those observed in previous experiments (1), but this difference may be attributable to the higher challenge dose of parasites used here (106 IRBC compared with 105 IRBC used in previous studies). Passive transfer of rabbit anti-AMA-1 antibodies protects against homologous but not heterologous challenge. IgG isolated from the sera of rabbits immunized with refolded DS AMA-1B was transferred to mice infected with homologous (DS) and heterologous (556KA) parasites. The rabbit antibodies reacted with both DS and 556KA refolded recombinant forms of AMA-1 (Fig. 4). The antibodies also reacted with both parasite antigens, although a stronger signal was obtained with the homologous DS antigen. This may well reflect the fact that different amounts of AMA-1 were loaded onto the gel because of differences in the proportions of AMA-1-expressing parasites in the two antigen preparations. When purified IgG was administered to mice 5 days after challenge with DS parasites, when the parasitemias were between 2 and 15%, parasitemias were markedly reduced within 24 h compared with those in mice receiving control IgG (Fig. 5) and the immune IgG prevented any deaths resulting from infection with the virulent DS parasite. In contrast, the immune IgG had no apparent effect on the parasitemias in mice challenged with heterologous 556KA parasites. DISCUSSION In this study we have used the infection of laboratory mice with the virulent DS and avirulent 556KA strains of P. chabaudi adami to investigate the impact of diversity in AMA-1 on the efficacy achieved by immunization with refolded recombinant AMA-1. The sequences of the DS and

3314

CREWTHER ET AL.

INFECT. IMMUN.

FIG. 4. Immunoblots of refolded recombinant AMA-1 from strains DS and 556KA and native AMA-1 probed with rabbit antibodies raised against refolded recombinant DS AMA-1B. All samples were prepared with nonreducing sample buffer. Lanes: 1, refolded DS AMA-1B; 2, refolded 556KA AMA-1B; 3, Triton X-114 extract of uninfected BALB/c erythrocytes; 4 and 6, Triton X-114 extract of BALB/c erythrocytes infected with mature stages of DS parasites; 5 and 7, Triton X-114 extract of BALB/c erythrocytes infected with mature stages of 556KA parasites. Lanes 1 to 5 were probed with rabbit anti-AMA-1B. Lanes 4 and 5 were stripped of rabbit antibodies and reprobed with hyperimmune mouse sera raised against DS parasites to demonstrate the relative loading in the lanes (lanes 6 and 7). Lanes 1 and 2 were developed with 125I-protein A. Lanes 3 to 5 were developed with horseradish peroxidase-conjugated sheep anti-rabbit IgG and Enhanced Chemiluminescence reagent. Lanes 6 and 7 were developed with sheep anti-mouse IgG and ECL reagent.

FIG. 3. Codon changes and amino acid substitutions among different alleles of P. chabaudi AMA-1. Amino acid residues corresponding to synonymous codon changes are indicated by lowercase letters. The third base of each codon is numbered. *The DK and 556KA sequences are identical.

556KA AMA-1 molecules, which show considerable homology to those of the AMA-1 molecules of other species of Plasmodium, differ from each other at 36 residue positions, all located in the ectodomain of the molecule. The amino acid substitutions described previously for the AMA-1 alleles of P. falciparum and P. vivax are clustered, with a relatively hypervariable region being located in the ectodomain between the first and third conserved cysteine residues. This clustering of substitutions is even more marked for the AMA-1 of P. chabaudi, with 21 of the 36 amino acid differences between the DS and 556KA sequences (58%) occurring in this region, which constitutes 17.7% of the polypeptide. Although some P. falciparum AMA-1 sequences are almost as different as the two full-length P. chabaudi sequences compared here, no two P. falciparum sequences differ by more than 13 residues in this region of AMA-1 (18). Thus, the use of these two strains of P. chabaudi is a rigorous test of whether AMA-1 can induce immune responses that protect against heterologous as well as homologous parasite challenge. The active and passive immunization experiments described here extend our earlier observations that the ectodomain of

TABLE 2. Refolded AMA-1 protects against homologous but not heterologous challenge

Mean

Range

Days on which peak parasitemia was observed

Homologous challengeb Adjuvant alone Red & alk AMA-1 Refolded AMA-1

58.9 60.0 15.6

50–68 46–65 8.8–29

8, 9 7, 8 7–10

0 of 6 survived 0 of 6 survived 7 of 8 survived

Heterologous challengec Adjuvant alone Red & alk AMA-1 Refolded AMA-1

24.3 27.6 23.6

21–28 23–37 15–33

8, 9 7, 8 7–9

All 6 survived All 6 survived All 8 survived

Peak parasitemia (%)

Groupa

a

Red, reduced; alk, alkylated. Strain DS (virulent). c Strain 556KA (avirulent). b

Outcome

VOL. 64, 1996


3315

FIG. 5. Transfer of rabbit antibodies to mice infected with P. chabaudi. Mice were infected with P. chabaudi adami DS (1 3 106 IRBC injected intraperitoneally) or 556KA (5 3 106 IRBC injected intraperitoneally) and then injected on day 4 or 5 with 3 mg of IgG isolated by affinity chromatography on protein A-Sepharose from the serum of a rabbit immunized with refolded DS AMA-1B (Immune IgG) or from that of a nonimmunized rabbit (Control IgG).

AMA-1, expressed in E. coli and refolded in vitro, induces immune responses that protect against challenge with homologous DS parasites. The degree of protection against homologous challenge achieved was in stark contrast to the failure of both active and passive immunization to protect against heterologous challenge. Clearly, at least some of the 36 amino acid substitutions which distinguish the 556KA AMA-1 sequence from the DS sequence are within epitopes recognized by antibodies capable of inhibiting parasite development, presumably at the stage of merozoite invasion. Given the clustering of the majority of mutations in domain I, it appears likely that this region of AMA-1 is a major target of immune responses that protect in a strain-specific manner. Although it is possible to induce protection in animal models by immunization with AMA-1, there is no direct evidence to date that AMA-1 is a target of naturally induced protective immune responses. However, our analysis of the observed pattern of substitutions in P. chabaudi AMA-1 indicates that these sequences have been the target of positive selection (12). It seems very likely that selection pressure exerted by protective immune responses has played a role in generating the observed substitution pattern. A bias towards nonsynonymous substitutions is regarded as an indication of positive selection operating at the level of the protein sequence. Such a bias has been observed in viral antigens, e.g., the hemagglutinin of influenza virus (reviewed in reference 32), that change as a result of selection pressure exerted by protective immune responses, as well as in malaria merozoite surface antigens, including MSP-1, MSP-2, and AMA-1 of P. falciparum (14). A clustering of substitutions along the polypeptide, as seen in the AMA-1 sequences, is another indication of positive selection. In contrast to the case with P. falciparum AMA-1, there does

not appear to be a bias towards nonsynonymous substitutions in P. chabaudi AMA-1, although the rates of both synonymous substitution (dS) and nonsynonymous substitution (dN) in the three putative subdomains defined by the disulfide bonding pattern are high. The clustering of mutations in P. chabaudi AMA-1 is more marked than that in P. falciparum AMA-1, and this observation, together with the nature of the codon changes, i.e., the fact that they result in many radical amino acid substitutions, argues strongly for positive selection operating on AMA-1 despite the lack of a bias towards nonsynonymous substitutions. This apparent paradox may have been caused by an overestimation of dS and an underestimation of dN for the P. chabaudi sequence comparisons. Of the two methods that we have used, that of Nei and Gojobori (20) overestimates dS compared with the method of Ina (15). Neither method, however, takes account of recurrent nonsynonymous substitutions within a codon, and therefore, both estimates could be significantly less than the true dN. We propose that many apparently synonymous substitutions in P. chabaudi AMA-1 are the result of a series of nonsynonymous substitutions (eventually generating an alternative codon for the same amino acid) which would be favored if strong selection pressure such as that exerted by a protective antibody response is directed against AMA-1. The observation that the synonymous substitutions show the same distribution as nonsynonymous substitutions and the large number of codons in which two or three bases have been substituted strongly support this hypothesis. Also consistent with this hypothesis is the finding of a bias towards nonsynonymous substitutions in P. falciparum but not in P. chabaudi. The much-longer evolutionary history shared by P. chabaudi and its host, Thamnomys rutilans, compared with that shared by P. falciparum and humans would allow more time for

3316

CREWTHER ET AL.

recurrent nonsynonymous substitutions within individual codons to generate what appear to be synonymous substitutions. An analogous situation may occur in the antigen-binding cleft of class I major histocompatibility complex molecules. Hughes et al. (13) showed that positive selection promotes allelic diversity by changing the charge profile in the binding cleft of the antigen recognition site. They have suggested that the enhanced rate of forward and backward substitution leads to a levelling off of dN in comparisons of distantly related species. Although the accumulation of point substitutions appears to have been a major cause of the existing diversity in AMA-1, the conservation of domain I sequences between the P. chabaudi strains, CB and 556KA, may be the result of an intragenic recombination event, a mechanism which appears to contribute to the generation of diversity in other merozoite surface antigens (16, 22, 26). In cases in which point substitutions have been selected, this may not necessarily always be due to the direct effect of these substitutions on epitopes recognized by protective antibodies. For example, two potential N-linked glycosylation sites in the 556KA sequence are abolished by the cluster of substitutions unique to domain II of P. chabaudi. Although N-linked glycosylation has not been detected in P. falciparum (9, 10), it may occur in P. chabaudi and altered levels of glycosylation could result from positive selection. Alternatively, variability in this region associated with different degrees of glycosylation may be related to some other parasite characteristic, such as virulence. Recently, Shiels et al. (25) have described sequence diversity in the immunodominant merozoite surface antigen of Theileria species which is associated with changes in the number of putative N-linked glycosylation sites. The failure of the recombinant P. chabaudi AMA-1 to protect against heterologous challenge makes it important to consider the likely effect of variation in P. falciparum AMA-1 on vaccine efficacy. Although no two P. falciparum AMA-1 sequences are as diverse in the ectodomain as are the two P. chabaudi sequences studied here, it seems likely that immunization with one form of P. falciparum AMA-1 will vary in efficacy depending on the genotype of subsequent challenge infections. The pattern of substitutions indicates that more than one region of the molecule is a target of protective immune responses. Thus, it may be possible to use a fragment of the molecule which is relatively conserved in a vaccine. Alternatively, a cocktail of a limited number of diverse forms of AMA-1 may be an effective vaccine against all genotypes of P. falciparum. Asymptomatic episodes of infection following immunization may broaden the specificity of the protective immune response induced by immunization with one or a small number of forms of AMA-1. ACKNOWLEDGMENTS We thank Etty Bonnici for preparation of the manuscript. We are grateful to Jenny Favaloro for providing the DNA samples from P. chabaudi strains; Tracey Baldwin for assistance with the animal experiments and maintenance of parasites; and Tony Hodder, Vikki Marshall, Ross Coppel, Vladimir Brusic, and Eddie Holmes for helpful discussions. This work was supported by the Cooperative Research Centre for Vaccine Technology, the UNDP/World Bank/WHO Special Program for Research and Training in Tropical Diseases, and the National Health and Medical Research Council (Australia). REFERENCES 1. Anders, R. F., P. E. Crewther, M. S. M. Leet, and D. Pye. Unpublished results. 2. Cheng, Q., and A. Saul. 1994. Sequence analysis of the apical membrane antigen 1 (AMA-1) of Plasmodium vivax. Mol. Biochem. Parasitol. 65:183– 187.

INFECT. IMMUN. 3. Collins, W. E., D. Pye, P. E. Crewther, K. L. Vandenberg, G. G. Galland, A. J. Sulzer, D. J. Kemp, S. J. Edwards, R. L. Coppel, J. S. Sullivan, C. L. Morris, and R. F. Anders. 1994. Protective immunity induced in squirrel monkeys with recombinant apical membrane antigen-1 of Plasmodium fragile. Am. J. Trop. Med. Hyg. 51:711–719. 4. Crewther, P. E., A. E. Bianco, G. V. Brown, R. L. Coppel, H. D. Stahl, D. J. Kemp, and R. F. Anders. 1986. Affinity purification of human antibodies directed against cloned antigens of Plasmodium falciparum. J. Immunol. Methods 86:257–264. 5. Crewther, P. E., J. G. Culvenor, A. Silva, J. A. Cooper, and R. F. Anders. 1990. Plasmodium falciparum: two antigens of similar size are located in different compartments of the rhoptry. Exp. Parasitol. 70:193–206. 6. Deans, J. A., T. Alderson, A. W. Thomas, G. H. Mitchell, E. S. Lennox, and S. Cohen. 1982. Rat monoclonal antibodies which inhibit the in vitro multiplication of Plasmodium knowlesi. Clin. Exp. Immunol. 49:297–309. 7. Deans, J. A., and W. C. Jean. 1987. Structural studies on a putative protective Plasmodium knowlesi merozoite antigen. Mol. Biochem. Parasitol. 26:155– 166. 8. Deans, J. A., A. M. Knight, W. C. Jean, A. P. Waters, S. Cohen, and G. H. Mitchell. 1988. Vaccination trials in rhesus monkeys with a minor, invariant, Plasmodium knowlesi 66 kD merozoite antigen. Parasite Immunol. 10:535– 552. 9. Dieckmann-Schuppert, A., E. Bause, and R. T. Schwarz. 1994. Glycosylation reactions in Plasmodium falciparum, Toxoplasma gondii, and Trypanosoma brucei probed by the use of synthetic peptides. Biochim. Biophys. Acta 1199:37–44. 10. Dieckmann-Schuppert, A., S. Bender, M. Odenthal-Schnittler, E. Bause, and R. T. Schwarz. 1992. Apparent lack of N-glycosylation in the asexual intraerythrocytic stage of Plasmodium falciparum. Eur. J. Biochem. 205:815– 825. 11. Dutta, S., P. Malhotra, and V. S. Chauhan. 1995. Sequence analysis of apical membrane antigen 1 (AMA-1) of Plasmodium cynomolgi bastianelli. Mol. Biochem. Parasitol. 73:267–270. 12. Hodder, A. N., P. E. Crewther, G. E. Reid, M. Leet, R. L. Moritz, R. J. Simpson, and R. F. Anders. Disulphide bond structure of the malaria vaccine candidate apical membrane antigen 1. J. Biol. Chem., in press. 13. Hughes, A. L., T. Ota, and M. Nei. 1990. Positive Darwinian selection promotes charge profile diversity in the antigen-binding cleft of class I major-histocompatibility-complex molecules. Mol. Biol. Evol. 7:515–524. 14. Hughes, M. K., and A. L. Hughes. 1995. Natural selection on Plasmodium surface proteins. Mol. Biochem. Parasitol. 71:99–113. 15. Ina, Y. 1995. New methods for estimating the numbers of synonymous and non-synonymous substitutions. J. Mol. Evol. 40:190–226. 16. Marshall, V. M., R. L. Coppel, R. K. Martin, A. M. J. Oduola, R. F. Anders, and D. J. Kemp. 1991. A Plasmodium falciparum MSA-2 gene apparently generated by intragenic recombination between the two allelic families. Mol. Biochem. Parasitol. 45:349–352. 17. Marshall, V. M., M. G. Peterson, A. M. Lew, and D. J. Kemp. 1989. Structure of the apical membrane antigen I (AMA-1) of Plasmodium chabaudi. Mol. Biochem. Parasitol. 37:281–283. 18. Marshall, V. M., L.-X. Zhang, R. F. Anders, and R. L. Coppel. Diversity of the vaccine candidate AMA-1 of Plasmodium falciparum. Mol. Biochem. Parasitol., in press. 19. Narum, D. L., and A. W. Thomas. 1994. Differential localization of fulllength and processed forms of PF83/AMA-1 an apical membrane antigen of Plasmodium falciparum merozoites. Mol. Biochem. Parasitol. 67:59–68. 20. Nei, M., and T. Gojobori. 1986. Simple methods for estimating the numbers of synonymous and nonsynonymous nucleotide substitutions. Mol. Biol. Evol. 3:418–426. 21. Nei, M., and L. Jin. 1989. Variances of the average numbers of nucleotide substitutions within and between populations. Mol. Biol. Evol. 6:290–300. 22. Peterson, M. G., R. L. Coppel, P. McIntyre, C. J. Langford, G. Woodrow, G. V. Brown, R. F. Anders, and D. J. Kemp. 1988. Variation in the precursor to the major merozoite surface antigens of Plasmodium falciparum. Mol. Biochem. Parasitol. 27:291–302. 23. Peterson, M. G., V. M. Marshall, J. A. Smythe, P. E. Crewther, A. Lew, A. Silva, R. F. Anders, and D. J. Kemp. 1989. Integral membrane protein located in the apical complex of Plasmodium falciparum. Mol. Cell. Biol. 9:3151–3154. 24. Peterson, M. G., P. Nguyen-Dinh, V. M. Marshall, J. F. Elliott, W. E. Collins, R. F. Anders, and D. J. Kemp. 1990. Apical membrane antigen of Plasmodium fragile. Mol. Biochem. Parasitol. 39:279–283. 25. Shiels, B. R., C. d’Oliveira, S. McKellar, L. Ben-Miled, S.-i. Kawazu, and G. Hide. 1995. Selection of diversity at putative glycosylation sites in the immunodominant merozoite/piroplasm surface antigen of Theileria parasites. Mol. Biochem. Parasitol. 72:149–162. 26. Tanabe, K., M. Mackay, M. Goman, and J. G. Scaife. 1987. Allelic dimorphism in a surface antigen gene of the malaria parasite Plasmodium falciparum. J. Mol. Biol. 195:273–287. 27. Thomas, A. W., J. A. Deans, G. H. Mitchell, T. Alderson, and S. Cohen. 1984.

VOL. 64, 1996


The Fab fragments of monoclonal IgG to a merozoite surface antigen inhibit Plasmodium knowlesi invasion of erythrocytes. Mol. Biochem. Parasitol. 13: 187–199. 28. Thomas, A. W., A. P. Waters, and D. Carr. 1990. Analysis of variation in Pf83, an erythrocytic merozoite vaccine candidate antigen of Plasmodium falciparum. Mol. Biochem. Parasitol. 42:285–288. 29. von Heijne, G. 1986. A new method for predicting signal sequence cleavage sites. Nucleic Acids Res. 14:4683–4690. 30. Waters, A. P., A. W. Thomas, J. A. Deans, G. H. Mitchell, D. E. Hudson,

Editor: S. H. E. Kaufmann

3317

L. H. Miller, T. F. McCutchan, and S. Cohen. 1990. A merozoite receptor protein from Plasmodium knowlesi is highly conserved and distributed throughout Plasmodium. J. Biol. Chem. 265:17974–17979. 31. Waters, A. P., A. W. Thomas, G. H. Mitchell, and T. F. McCutchan. 1991. Intra-generic conservation and limited inter-strain variation in a protective minor surface antigen of Plasmodium knowlesi merozoites. Mol. Biochem. Parasitol. 44:141–144. 32. Wilson, I. A., and N. J. Cox. 1990. Structural basis of immune recognition of influenza virus haemagglutinin. Annu. Rev. Immunol. 8:737–771.