Vaccination with Plasmodium knowlesi AMA1 Formulated in the Novel Adjuvant Co-Vaccine HT™ Protects against Blood-Stage Challenge in Rhesus Macaques

Plasmodium falciparum apical membrane antigen 1 (PfAMA1) is a leading blood stage vaccine candidate. Plasmodium knowlesi AMA1 (PkAMA1) was produced and purified using similar methodology as for clinical grade PfAMA1 yielding a pure, conformational intact protein. Combined with the adjuvant CoVaccine HT™, PkAMA1 was found to be highly immunogenic in rabbits and the efficacy of the PkAMA1 was subsequently tested in a rhesus macaque blood-stage challenge model. Six rhesus monkeys were vaccinated with PkAMA1 and a control group of 6 were vaccinated with PfAMA1. A total of 50 µg AMA1 was administered intramuscularly three times at 4 week intervals. One of six rhesus monkeys vaccinated with PkAMA1 was able to control parasitaemia, upon blood stage challenge with P. knowlesi H-strain. Four out of the remaining five showed a delay in parasite onset that correlated with functional antibody titres. In the PfAMA1 vaccinated control group, five out of six animals had to be treated with antimalarials 8 days after challenge; one animal did not become patent during the challenge period. Following a rest period, animals were boosted and challenged again. Four of the six rhesus monkeys vaccinated with PkAMA1 were able to control the parasitaemia, one had a delayed onset of parasitaemia and one animal was not protected, while all control animals required treatment. To confirm that the control of parasitaemia was AMA1-related, animals were allowed to recover, boosted and re-challenged with P. knowlesi Nuri strain. All control animals had to be treated with antimalarials by day 8, while five out of six PkAMA1 vaccinated animals were able to control parasitaemia. This study shows that: i) Yeast-expressed PkAMA1 can protect against blood stage challenge; ii) Functional antibody levels as measured by GIA correlated inversely with the day of onset and iii) GIA IC50 values correlated with estimated in vivo growth rates.


Introduction
The Plasmodium falciparum parasite is responsible for at least 300 million cases of malaria [1] and about 800,000 deaths every year [2]. The majority of these deaths occur in children under five and nulliparous women in sub-Saharan Africa [3]. An effective malaria vaccine combined with existing anti-malarial strategies (e.g. indoor residual spraying, bed-nets, antimalarial drugs etc.) would certainly help to reduce malaria-related morbidity and mortality, and as such would be a much needed addition to the arsenal.
The leading vaccine candidate Apical Membrane Antigen 1 (AMA1) was first discovered as a minor invariant merozoite antigen in Plasmodium knowlesi [4,5] able to protect rhesus monkeys from asexual blood stage challenge with P. knowlesi (Pk) [6]. The protective effect of AMA1 vaccination has since been demonstrated in mouse and simian malaria models [7][8][9] and has spurred the clinical development of P. falciparum AMA1 [7].
P. knowlesi causes a fulminent infection in rhesus monkeys, often resulting in high parasitaemias similar to levels observed in P. falciparum infections in humans, and ultimately in death if left untreated [10][11][12]. P. knowlesi has recently also been shown to infect humans under natural conditions and was shown to be a contributing cause of death [13,14].
The clinical development of AMA1 has resulted in several trials [15][16][17][18][19], including a phase IIa mosquito challenge of non-malaria exposed volunteers [20] and a phase IIb trials in Malian children [21,22]. The phase IIa study did show a small, but significant, delay in the pre-patent phase as determined by PCR data modelling, whilst time to patency was similar in vaccinees and controls, despite high Growth Inhibition Assay (GIA) titres in the vaccinees [20]. One phase IIb study in Malian children using Alum did not show any significant protection [22], whereas a Phase IIb study using a potent adjuvant (AS02) revealed a modest non-significant vaccine efficacy of about 17% when analysed against all AMA1 alleles in infected subjects. The number of parasites expressing the homologous 3D7-AMA1 was too low allow for efficacy to be determined, but when data were analysed for parasites expressing the homologous 3D7-AMA1 haplotype in the highly polymorphic C1 region, vaccine efficacy was estimated at 64% [23]. This has fuelled the debate whether vaccination with AMA1 would have protective potential in humans.
The problem in pre-clinical and early clinical evaluation of a vaccine is the lack of correlates of protection. This hampers vaccine development, as without these correlates it is not clear what kind of responses a vaccine needs to induce in order to protect. A recent paper by Dutta et al. suggests that high titres of, functionally active, antibodies against AMA1 are required for protection in a nonhuman primate (Aotus) model [8]. We use non-human primate models to establish correlates of protection, as an alternative to rodent malaria models, that may be less representative for the human malaria's [24]. The Pk rhesus monkey model allows proof of concept testing of P. falciparum and P. vivax vaccine candidates for which Pk orthologs exist as well as evaluation of vaccines against P. knowlesi. Humoral and cellular immune responses can be evaluated in depth. Parasite challenge can be performed either by infected mosquito bite, sporozoite injection (i.v., s.c. or i.d.) or i.v. inoculation of infected red blood cells, and followed-up beyond patency, allowing for a larger window of opportunity to detect vaccine effects as compared to human phase IIa trials. Moreover, rhesus monkeys can be infected repeatedly with P. knowlesi before immunity develops [11]. Unlike clinical trials, trials in monkeys do not require cGMP-grade materials (although at least GLP grade antigen is preferred), and formal pharmacotoxicology testing is not mandatory. In addition, preliminary safety profile of vaccines can be established in monkeys (behaviour, appetite, local reactions, clinical chemistry and haematology).
In this study, we used P. knowlesi in Macaca mulatta (rhesus monkey) as a model to test the efficacy of the blood stage vaccine candidate AMA1. The vaccine antigen PkAMA1 was expressed in the methylotrophic yeast Pichia pastoris, purified and quality controlled similar to the cGMP production of PfAMA1 [25]. The antigen used in the control group, cGMP-grade PfAMA1 [25], induces PkAMA1 cross-reactive antibodies detectable in ELISA, but results obtained in our laboratory have shown that these antibodies do not have appreciable GIA activity when tested on P. knowlesi.
A number of clinical studies, among others the one describing a phase Ia clinical study, showed that the adjuvant AlhydrogelH is not potent enough to achieve high antibody levels for AMA1 resulting in high GIA titres [15,16,22]. Therefore we used the novel adjuvant, CoVaccine HT TM that induces significantly higher levels of functional antibodies to AMA1 than a proprietary water in oil adjuvant, whilst being well tolerated by pigs and macaques [26,27]. The active ingredients of CoVaccine HT TM are sucrose fatty acid sulphate esters (SFASE) and an oil-in-water (squalane and Polysorbate-80) emulsion [26,28]. It has been shown that this adjuvant enhances both humoral and cell mediated responses when used with subunit vaccines [26]. CoVaccine HT TM is currently in clinical development (http://www.clinicaltrials. gov/ct2/results?term=CoVaccine+ht).
Here we report the parasitological and immunological results of a rhesus macaque vaccination -blood-stage challenge study with Pichia pastoris expressed PkAMA1 formulated with CoVaccine HT TM . We show that there is a correlation between functional (GIA) antibody levels and the delay of onset of parasitaemia as well as with in vivo parasite multiplication rates. The promising results obtained here warrant clinical development of CoVaccine HT TM adjuvanted P. falciparum and, possibly, P. knowlesi AMA1 as vaccines for life-threatening human Plasmodium infections.

Cloning of Plasmodium knowlesi ama1
A synthetic gene, codon optimised for expression in Pichia pastoris, comprising domain I-II-III of Plasmodium knowlesi H strain AMA1 (amino acids 43 to 487) (Genbank accession number XM_002259303), was constructed by DNA2.0 (Menlo Park, CA). Five Potential N-glycosylation sites were removed, by replacing one of the amino acid in the NX[T/S] fingerprint with an amino acid present in an AMA1 orthologue, analogous to what was done for P. falciparum AMA1 [29]. Asn 107 was replaced by Lys, Ser 178 was replaced by Asn, Asn 190 was replaced by Glu, Ser 240 was replaced by Arg and Asn 441 was replaced by Gln. A PCR forward primer (PkF: 59GGAATTCCCAATCATTGAGAGATC39) with a 59 EcoR1 restriction site and two reverse primers with 39XbaI restriction sites (PkRIII: 59CGTCTAGACCTTCCTAGTAACA-TTTTCTTC39; PkRII: 59GCTCTAGACCTGGAAACTCGT-TGTCTAC 39) were used to amplify PkAMA1 domain I-II-III (aa 43-487) and PkAMA1 domain I-II (aa 43-387), respectively. PCR products were cloned into EcoRI-XbaI sites of the pPicZaA vector (Invitrogen, Leek, The Netherlands) in frame with the vectorencoded myc-epitope and hexa-Histag. After transformation of E. coli DH5a cells, plasmids were isolated, the sequence was confirmed by DNA-sequencing (Baseclear, Leiden, The Netherlands), and used to transform P. pastoris KM71H following the manufacturer's protocols.

Selection of Pichia clones
Transfected P. pastoris were tested for protein production. Cells were cultured in 10 mL glycerol-containing medium for 24-30 h (30uC, 200 rpm). Cells were harvested by centrifugation and resuspended in 4 mL methanol-containing medium, and then cultured for 24 h. After low-speed centrifugation, the culture supernatant was harvested and 20 mL of the culture supernatant was analyzed by non-reducing SDS-PAGE with coomassie brilliant blue staining or Western blotting with the reduction sensitive monoclonal antibody R3/1C2 [4], which also served to confirm the identity of the proteins.

Fermentation of Pichia pastoris
Fermentation runs for either construct were done in a 7 L fermentor (Applikon, Schiedam, The Netherlands) with a starting volume of 2 L. Fermentations were essentially performed according to the instructions provided by the manufacturer of the yeast strain and the vector (Invitrogen, Leek, The Netherlands) for P. pastoris Mut S cells [25]. PkAMA1 DI-II-III and PkAMA1  DI-II proteins The methodology used for the production of PkAMA1 proteins was similar to what was used for PfAMA1 [25]. Proteins were purified using two chromatography purification steps, Nickel IMAC affinity chromatography followed by size exclusion chromatography. In brief, filtered culture supernatant was loaded on a Ni-IMAC-Sepharose column (GE Healthcare, Diegem, Belgium). Washing was done with 10 column volumes of 20 mM Na-phosphate, pH 7.8, 500 mM NaCl and 10 mM imidazole to remove contaminating proteins. PkAMA1 proteins were eluted with 20 mM Na-phosphate pH 7.8, 500 mM NaCl and 100 mM imidazole. This fraction was exchanged into PBS using ultrafiltration concentrator tubes (10 kDa cutt-off; Millipore, Amsterdam, The Netherlands). Protein fractions from different fermentations were pooled and concentrated to 1 mL. Subsequently this fraction was injected on a Superdex 75 preparative size exclusion chromatography column (GE healthcare, Diegem, Belgium). Fractions containing PkAMA1 proteins of 50 kDa (PkAMA1 D I-II-III) or 40 kDa (PkAMA1 DI-II) were pooled, concentrated and filter sterilised and then stored at 220uC.

SDS-PAGE
SDS-PAGE (NuPAGE Novex; Invitrogen, Leek, The Netherlands) 4-12% gradient gels were used with MES-buffer according to the manufacturer's instructions. For densitometry coomassie blue stained gels were scanned with a GS-800 densitometer (Bio-Rad) and analyzed with Quantity one software package version 4.5.

Conformational integrity assay for PkAMA1
Quantitative binding of PkAMA1 to the conformation-dependent MAb [4] was determined as follows: MAb R3/1C2 (2 mg mL 21 ) was coupled to 1 mL cyanogen bromide (CNBr) activated Sepharose 4 Fast Flow according to the manufacturer's instructions (Amersham Biosciences). Two-hundred mg of PkAMA1 DI-II-III, PkAMA1 DI-II and BSA were diluted in 3 volumes binding buffer (ImmunoPure A, Pierce, Rockford, IL). Samples were separately passed over the immobilized R3/1C2 affinity column and the flow through was collected, and samples containing AMA1 were confirmed to be negative for protein by mBCA protein determination. After extensive washing with 10 bed volumes PBS, bound proteins were eluted with 5-10 volumes 4.0 M NaSCN, pH 7.0. The eluted proteins were exchanged into PBS and concentrated using ultrafiltration concentrator tubes (10 kDa cut off; Millipore, Millipore, The Netherlands). The protein contents in the eluate, flow-through and wash fractions were estimated by mBCA. Lyophilised cGMP PfAMA1 [4] was reconstituted with saline for injection to a concentration of 200 mg mL 21 and then formulated as above.

Ethics statement
For the rabbit immunisations, all animals were handled in strict accordance with good animal practice as defined by the Belgian national animal welfare regulations, and all animal work was approved by the ethics committee of the Centre d'Economie Rurale (CER Groupe, Marloie, Belgium). All rabbit handlings were performed at Eurogentec SA, Seraing, Belgium, under permit number LA 1800104.
All rhesus monkeys used in this study were captive bred for research purposes. BPRC housing and animal care procedures are in compliance with Dutch law on animal experiments, European directive 86/609/EEC, and with the ''Standard for Humane Care and Use of Laboratory Animals by Foreign Institutions'', identification number A5539-01, provided by the Department of Health and Human Services of the US National Institutes of Health (NIH). The BPRC is compliant with the recommendations of the Weatherall report ''The use of non-human primates in research'' (http://www.acmedsci.ac.uk/images/project/nhpdownl. pdf).The independent ethics committee at BPRC, constituted according to Dutch law on animal experiments, approved the study protocol (number DEC 552) prior to start of the experiment. In order to minimise suffering, venapunctures, immunisations and passage of blood stage parasites, were performed on ketaminesedated monkeys. Animals were treated with antimalarials when the parasitaemia following challenge exceeded 2%. Every day animals were carefully observed for their general health (behaviour, alertness, appetite and stool). In the event abnormalities were observed, which did not happen during the course of this study, a veterinarian was available at all times to take appropriate action. During the course of the study all animals received a variety of enrichment items (e.g. toys, food puzzles).

Rabbit immunisations
Rabbit immunisations were conducted at Eurogentec, Seraing, Belgium under national animal welfare regulations. Two groups of rabbits (N = 5) were injected intramuscularly with 0.5 mL vaccine containing 50 mg of purified PkAMA1 protein formulated in CoVaccine HT TM . Two booster injections of 50 mg were given at days 28 and 56 after the primary dose. Bleeds were taken from each animal before immunisation as well as two weeks after the second immunisation. Exsanguination was done two weeks following the last immunisation (day 70).
Vaccination, boost, and blood sampling of rhesus macaque monkeys 12 healthy adult rhesus macaques weighing between 5 to 13 kg, were selected for the study. The animals were all shown to be negative for Plasmodium parasites, anti-PfAMA1 or anti-PkAMA1 antibodies. Animals were stratified over two groups of six animals, with similar characteristics with respect to sex, age, and weight. Treatments were then randomly assigned ( Table 1). All vaccinations and venapunctures were performed on ketamine-sedated animals. Monkeys in the experimental group (N = 6) were immunised with 50 mg PkAMA1 DI-II-III (aa43-487) formulated in CoVaccine HT TM and the control monkeys (N = 6) were immunised with 50 mg GMP-grade PfAMA1 FVO strain DI-II-III (aa25-545) formulated in CoVaccine HT TM . Vaccines were given intramuscularly in alternating upper legs on days 0, 28 and 56. Animals were bled before each immunisation and on days 1, 7, and 14 following each vaccination. After the primary parasite challenge animals were left to recover before a 50 mg booster dose of Pk or PfAMA1 was given on day 203 and animals were rechallenged two weeks later (day 217). The follow up was 35 days.
Monkeys were left to recover from the second challenge and were boosted with 50 mg Pk or PfAMA1 on day 435, and challenged again on day 450. Follow up time was 34 days ( Figure 1).

Inspection of injection sites
Injection sites were inspected on days 0, 1, 7 and 14 following each vaccination. Local reactions for oedema and erythema were scored using the Draize scale [30]. A similar scoring system was applied for the rating of induration and lymph node swelling.

Clinical chemistry and haematology
Standard parameters were determined for haematology (Sysmex XT 2000iV platform; Goffin Meyvis, Etten-Leur, the Netherlands) and clinical chemistry (Cobas Integra 400; Roche Diagnostics, Basel, Switzerland) on freshly obtained blood samples. The following biochemical measurements were performed on all blood samples taken on days 0, 1, 7 and 14 following each vaccination: alkaline phosphatase, alanine transaminase, aspartate transaminase, gamma glutamyl transpeptidase, bilirubin, cholesterol, lactate dehydrogenase), urea, creatinine, albumin, total protein, glucose, iron, calcium, sodium, potassium, chloride, phosphate and HCO 3 2 . The following haematological measurements were performed: erythrocyte count, haemoglobin, haematocrit, mean corpuscular volume), mean corpuscular haemoglobin, mean corpuscular haemoglobin concentration, white blood cell count including leukocyte differentiation, platelet count and mean platelet volume. The data were compared with reference values obtained from healthy naive rhesus monkeys. Commercial kits were used according to manufacturer's instruction.

Blood stage parasites and challenge infection of rhesus macaques
A parasite donor monkey negative for AMA1 antibody and without previous exposure to malaria (also negative for Herpes-B, Simian Retrovirus and STLV) was selected. Cryopreserved wildtype P. knowlesi H strain was thawed according to the method described by Trager et al. [31]. The parasite pellet (containing mainly surviving rings) was allowed to recover in gassed complete culture medium for 16 h at 37uC, in which time the parasites developed to late trophozoites. 10,000 Freshly prepared late trophozoites were injected i.v. into the donor monkey. Daily finger pricks were taken from day 3 onwards. Parasitaemia was monitored until it reached 1% (day 8), and 10 mL of blood was drawn. The donor monkey was subsequently cured with chloroquine (10 mg/kg i.m., ResochinH, Bayer, Leverkusen, Germany) and pyrimethamine (1 mg/kg per os, Sigma, Zwijndrecht, The Netherlands) for three consecutive days. Cells were extensively washed with RPMI 1640, and parasite infected cells were adjusted to 1610 4 mL 21 . All monkeys received 1 mL of this parasite suspension containing 10,000 trophozoites i.v. in the vena sapphena, followed by 5 mL of PBS to ensure the whole inoculum was delivered. Finger prick blood samples were taken daily starting from 3 days post infection and continued for three weeks for the first challenge and five weeks for the second and third challenge. Thin blood smears were prepared every day at approximately 15:00h when the majority of parasites were at ring stage. Parasitaemias were determined by counting at least 4,000 erythrocytes in Giemsa-stained thin blood smears. Slides with very low counts of infected red blood cells (iRBC) were arbitrarily assigned a parasitaemia of 0.005%. Animals were cured if the levels of parasitaemia reached $2% or at parasitaemias between 1 and 2% if high numbers of schizonts were observed. Treatment was by i.m. injection of chloroquine and pyrimethamine as described. At the end of each challenge phase (21 days for the primary or 35 or 34 days for the second and third challenges, respectively), all animals not yet cured were then treated with chloroquine and pyrimethamine as described.
Antibody ELISA 96-well microtitre ELISA plates were coated with 50 ng protein per well and incubated overnight at 4uC. Plates were washed and blocked. Serum samples were diluted in serial two fold dilutions, starting at 1/5,000 and were in duplicate added to antigen-coated wells and incubated for 1 h at 37uC. A duplicate control dilution series of a standard rabbit or rhesus anti PkAMA1 standard serum pool was prepared from immune serum and included on every plate. After extensive washing, the plates were incubated with secondary antibodies goat anti-rabbit or anti-human immunoglobulin G conjugated to alkaline phosphatase (Pierce, Rockford, IL). Bound antibodies were visualized by adding the substrate solution pNPP (para nitro phenyl phosphate, Sigma, Zwijndrecht, The Netherlands) and incubated for 30 minutes at room temperature. Absorbance at 405 nm was measured with a Bio-Rad Microplate Reader 3550 (BioRad, Veenendaal, The Netherlands). A standard curve was constructed by four-parameter curve fitting for every plate and titres of unknowns were calculated from the four-parameter fit. Antibody titres were expressed as arbitrary units (AU), where 1 AU corresponds to an amount of IgG theoretically yielding an OD of 1 over background. Thus the titre in AU indicates the dilution at which an OD of 1 over background is achieved.

Growth inhibition assay
The in vitro anti-parasite activity of total IgG was determined by P. knowlesi growth inhibition assays (GIA). Total IgG from serum or plasma was purified on Protein A columns using binding and elution buffers supplied by the manufacturer (Pierce, Rockford, IL), exchanged into RPMI 1640 and concentrated to either 12 mg mL 21 for rabbit IgG or 20 mg mL 21 for monkey IgG using ultra centrifugal devices (Millipore, Etten-Leur, The Netherlands), and subsequently filter sterilised through a 0.22-mm filter (Millipore, Etten-Leur, The Netherlands). IgG antibodies were aliquoted and frozen at 220uC until use. An alanine-synchronised late stage (schizont) P. knowlesi (H-strain) culture was adjusted to 1% parasitaemia and 4% haematocrit in 40% normal human serum and 30 mg mL 21 gentamycin. Purified IgG was added in a volume of 30 ml in a 96-well flat-bottomed tissue culture plate (Greiner, Alphen a/d Rijn, The Netherlands), to which 30 mL of the parasite culture was added. Samples were tested in triplicate at four IgG concentrations; viz. 6, 3, 1.5 and 0.75 mg mL 21 for rabbit or 10, 5, 2.5 and 1.25 mg mL 21 for rhesus IgG. The highest IgG concentration (10 mg mL 21 ) used for rhesus GIA corresponds to the total IgG concentration in undiluted rhesus serum, thus GIA activity at 10 mg mL 21 can be interpreted as an estimate of GIA activity in undiluted rhesus serum. The inhibitory mAb R3/1C2 (1.5 mg mL 21 ) and EDTA (4 mM) were also included in every test plate as positive controls. The final concentration of the culture was 1% parasitaemia at 2% haematocrit and 20% normal human serum. Plates were incubated at 37uC under a 5% O 2 , 5% CO 2 , 90% N 2 moist atmosphere for one cycle (26-30 h). Every assay was routinely performed in triplicate. Parasite specific lactate dehydrogenase (pLDH) activity, a measure for the total number of parasites, was measured using standard protocols [32].
Inhibition was calculated as 1006[12((OD 650 of infected RBCs with tested IgG2OD 650 of uninfected RBCs)/(OD 650 of infected RBCs with pre-immune IgG2OD 650 uninfected RBCs))]. GIA assays were performed in two separate batches; all assays up to and including the second challenge were done in one batch. GIA assays for the third challenge were done independently from the aforementioned series.

Lymphocyte stimulation assays
PBMCs were isolated from heparinised peripheral blood by density gradient centrifugation using lymphocyte separation medium (Invitrogen, Leek, The Netherlands) at 2100 rpm for 15 minutes without brake at room temperature. Cells at the interphase were collected and washed twice with RPMI 1640 medium. The PBMCs were resuspended in 8% foetal calf serum (FCS) (ICN, The Netherlands) in RPMI 1640 medium and adjusted to 2610 6 cells mL 21 . The cell suspension (100 mL) was added in triplicate to 96 well round bottom plates (2610 5 cells well 21 ). One-hundred mL culture medium containing AMA1 antigen was added to a final AMA1 concentration of 10 mg mL 21 . Concanavalin A was added as positive control at a final concentration of 5 mg mL 21 . Cells were incubated at 37uC in 5% CO 2 atmosphere for 3 days. Supernatant was harvested and stored at 220uC until further use.

Cytokine measurement by ELISA
Culture supernatants were thawed at room temperature and assayed for IFN-c and IL-13 using macaque IFN-c and IL-13 ELISA kits (U-Cytech Biosciences, Utrecht, The Netherlands) according to the manufacturer's instructions. Recombinant IFN-c or IL-13 were used as standards. Concentrations of unknowns were estimated using a four-parameter fit on the standard curve.
Data are presented as background corrected values, i.e. background (medium only) is subtracted from the stimulated values.

Relation between GIA and in vivo parasite growth
In vivo growth was estimated for all animals. The initial parasitaemia was calculated by dividing the inoculum (10,000) by the total numbers of red blood cells (estimated by weight and average RBC) and multiplied by four (assuming four infectious merozoites arise from one trophozoite). The parasitaemia observed at patency was then divided by the initial parasitaemia, yielding the multiplication factor between infection and day of patency plus one. The average per day in vivo growth rate was then calculated by taking the day of patency th root of the multiplication factor. In the event animals became patent, without increasing parasitaemia in the days following patency, the last occurrence of patency observed was used for calculations. This yields a conservative estimate, as it could be argued that these animals did not require treatment and thus were protected and could be analysed similar to animals that did not become patent at all. Animals that did not become patent were arbitrarily assigned the last day of observation. Average per day in vivo growth rates in controls for the three separate challenges were calculated, excluding the non-responder animal for the first two challenges, and were used for subsequent in vivo growth inhibition (IVGI) calculations. IVGI was calculated as follows: 100 6 (12(growth in PkAMA1 animal/average growth in PfAMA1 group)).

Statistical analysis
All statistical analyses were performed with the R language and environment for statistical computing (R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http:// www.R-project.org). Evaluation of statistical significance for IgG levels was performed on log-transformed titres and IgG titres are presented as geometric means with 95% confidence intervals (95% CI). Differences in GIA titres were analysed by t-test and GIA titres are presented as arithmetic means with 95% CI (the percent sign used in conjunction with GIA titre differences throughout the paper should be interpreted as percent points). IFN-c and IL-13 levels are presented as medians with P 25 and P 75 percentiles, between group comparisons were done using the non-parametric Mann-Whitney test and paired data (between time points) were analysed using Wilcoxon signed rank test. GIA and logtransformed IgG between time-point comparisons were done using paired t-tests. Survival data (time to patency or time to drug cure) were evaluated using log-rank tests.
The relation between IgG and the GIA titres was investigated by non linear least squares regression using the following formula: where IgG is the antibody titre (PkAMA1 specific IgG in AU mL 21 or total IgG in mg mL 21 ) and IC 50 represents the IgG concentration yielding 50% inhibition. Slope is a parameter indicating the steepness of the curve, with steeper curves at higher values. As Slope is rather difficult to interpret it is expressed as the foldincrease in IgG required to raise GIA levels from 10 to 90%. (It can be shown algebraically that this is: e [ln(9) -ln(1/9)]/Slope , which is approximately e (4.394/Slope) and for an increase from 10 to 50% or 50 to 90% this is approximately e (2.197/Slope) ). IC 50 and Slope values were estimated from GIA data for each animal at 4 different total IgG concentrations (viz. 1.25, 2.5, 5 and 10 mg mL 21 ) using a non-linear mixed effects model employing the GIA parameterisation described above. This method reduces 4 GIA data points at different total IgG concentrations to two parameters (viz. IC 50 and slope) and thus makes full use of the GIA titration data. The IC 50 value represents the amount of total IgG required to obtain 50% growth inhibition and can thus be considered as a measure of GIA activity in a serum, with lower IC 50 values representing more potent GIA activity. The results of the fitting procedure are depicted in Figure S1, and IC 50 values are indicated.

Production, purification and quality control of PkAMA1
The Pichia pastoris expression system was used to produce AMA1 from P. knowlesi H strain (amino acids 43-487 and aa 43-387), with potential N-glycosylation sites removed. Several runs in a 7-Liter fermentor were performed and subsequently the protein was purified using Ni-IMAC, as the protein contains a hexa-histidine tag. Gel filtration was used in the second step, essentially for buffer exchange and for the removal of excess of dimers. Yields were approximately 2 and 40 mg mL 21 for PkAMA1 D I-II-III and PkAMA1 D I-II, respectively.
table-2-captionThe apparent molecular masses of the two constructs analyzed by 4-12% gradient SDS-PAGE were 50 kDa and 40 kDa for PkAMA1 DI-II-III and PkAMA1 DI-II respectively, in accordance with theoretical mass calculated from their respective amino acid sequences. Figure 2 shows that both products are near to apparent homogeneity. Reduction of PkAMA1 D I-II-III protein samples resulted in the appearance of two additional low molecular weight bands (Figure 2). The apparent size of the two smaller products is in accordance with the location of a proteolytic cleavage site in DII, observed in PfAMA1 FVO and 3D7 [25,32]. Densitometric analysis of SDS-PAGE on the non-reduced samples revealed that 96% of the protein was in the main band for PkAMA1 DI-II-III and 99% for PkAMA1 DI-II. Densitometry on the reduced samples showed that less than 10% and 3% of the protein is internally cleaved, for PkAMA1 DI-II-III and PkAMA1 DI-II, respectively. Both proteins react with anti hexa-histidine antibodies, anti mycantibodies and with the PkAMA1 specific antibody R3/1C2 (data not shown), as expected.
As it has been shown that immunisation with reduced and alkylated AMA1 results in low functional antibody titres [33], we designed an assay to assess the conformational intactness of the protein, by a quantitative binding assay to immobilized R3/1C2, essentially as described before [25]. R3/1C2 recognizes a reduction-sensitive epitope [4]. Eighty-eight (88%) percent of PkAMA1 (DI-II-III) was bound to immobilized R3/1C2.

Comparative immunogenicity studies and growth inhibition in rabbits using the novel adjuvant CoVaccine HT TM
The protein products made were first tested for immunogenicity in rabbits. These experiments were, given the low yield of the full length PkAMA1, also used to select the antigen of choice for the rhesus study, as unpublished studies by ourselves and those of Lalitha and co-workers [34] suggest that PfAMA1 domain I and II is nearly as good as PfAMA1 domain I, II and III at inducing growth inhibiting antibodies. To this end, two groups of five rabbits were immunised with 50 mg of either PkAMA1 DI-II-III or PkAMA1 DI-II adjuvanted with CoVaccine HT TM . Immunisations were given at days 0, 28

Safety and local reactions in rhesus macaques
All 12 rhesus monkeys in the study were carefully observed in their behaviour (general, vividness, appetite) and monitored for their general health (weight, stool) during the full course of the study. At day 1, 7 and 14 following the vaccinations at days 0, 28 and 56, blood samples were taken and analysed for haematological and clinical chemistry parameters. Both vaccine formulations were well tolerated. No vaccine related serious adverse events (i.e. an event which hypothetically might have caused death if it were more severe) were observed. For all but two animals, body weights remained unchanged during the observation period. One animal in the PkAMA1 group had a decreased appetite between days 8 and 15. No oedema, erythema or hypersensitivity reactions were observed at the injection sites. Following vaccination three animals in PkAMA1 group and one animal in PfAMA1 had a slight swelling in the draining inguinal lymph node that resolved within two weeks. Three animals in the PfAMA1 group developed a transient muscle induration at the site of injection following vaccination. After each immunisation, one animal in the PfAMA1 group experienced muscle induration at the point of injection with a maximum size of 2.5 cm. These indurations resolved within two to four weeks and did not influence the animals' behaviour.
In general, none of the animals showed persistent deviations from the reference values during the course of the study. Acute effects of sedation and sampling were evident, and are generally observed during vaccination studies. Elevations following vaccination were observed in white blood cell count, basophil count and monocytes. These returned to baseline values within 14 days after each vaccination. Neutrophil levels increased to above normal values with a concomitant drop in free serum iron. Both resolved within 14 days after immunisation.

Blood-stage parasite challenge
At day 70, all animals were blood stage challenged by in vivo passage of 10,000 P. knowlesi H strain late trophozoites. The left hand panels in figure 3 show the parasitaemia in the blood of individual animals during the course of infection, both in the immunised (top panel) and control (bottom panel) group. Table 2 shows the parasitological and immunological data at the day of challenge. Five animals from the control group had to be cured within 8 days, the sixth animal, however (R01041 [ N ]) did not become patent during the challenge phase ( Figure 3, Table 2). One monkey from the PkAMA1 group, (Ri9980203 [e]) was patent at days 10 and 14 post infection, but was obviously able to control the parasitaemia ( Table 2). The other 5 animals had to be cured, but there was an average delay of about 2 days in the day of onset of the parasitaemia between the PkAMA1 group and the PfAMA1 control group, albeit not statistically significant (log-rank test, p = 0.453). The time to drug cure was also delayed for 4 out of 6 animals, but again not statistically significant (log-rank test, p = 0.28). When these analyses were repeated excluding the control monkey that did not become patent (R01041 [ N ]), both time to patency and time to drug cure were significantly longer in the PkAMA1 vaccinated group (log-rank test, p = 0.006 and 0.002, respectively).
Immunological analysis of the first phase of the study Figure 4 shows that all animals had low anti-PkAMA1-IgG levels before vaccination (14 95% CI 6 to 34 and 32 95% CI 17 to 61 AU mL 21 , for PkAMA1 and PfAMA1 groups respectively). Following vaccination, animals in the PkAMA1 group had detectable levels of antibodies to PkAMA1 two weeks after the first immunisation (939, 95% CI 358 to 2460 AU mL 21 , paired t-test with day 0, p = 0.0006). Anti PkAMA1-IgG also increased in the PfAMA1 group, but at a much lower rate (118, 95% CI 51 to 271, paired t-test with day 0, p = 0.0005). The second immunisation at day 28 boosted responses at day 56 to high levels in the PkAMA1 group (22.3, 95% CI 11.7 to 42.6 kAU mL 21 , paired t-test with day 28, p = 0.0002). PkAMA1-IgG levels in the PfAMA1 group were also boosted significantly after the second immunisation (3.2, 95% CI 1.6 to 6.5 kAU mL 21 , paired t-test with day 28, p = 0.0006). At day 56 anti-PkAMA1-IgG levels were significantly higher in the PkAMA1 group (7.0 fold, 95% CI 3.0 to 16.0 fold, t-test p-value = 0.0004). The third vaccination at day 56 boosted anti PkAMA1-IgG levels at day 70 even further; animals in the PkAMA1 group had 70.5 kAU mL 21 (95% CI 43.2 to 115.1, paired t-test with day 56, p = 0.0002) and animals in the PfAMA1 had 45.5 kAU mL 21 (95% CI 21.5 to 96.5, paired t-test with day 56, p,0.0001). This was in part due to one animal (Ri397 [e]) in the PfAMA1 group that had an unexpectedly high anti PkAMA1 titre (164.0 kAU mL 21 ) at day 70 and this value was confirmed in repeat ELISA's. In parallel, the IgG titre to the PfAMA1 vaccine antigen was also high in this animal (266.6 kAU mL 21 Figure 5 shows the GIA titres in the two treatment groups. The first two immunisations at days 0 and 28 induced clearly detectable GIA activity in the PkAMA1 vaccinated animals and GIA levels had reached 41.3% at day 42 (95% CI 15.6 to 67.0, paired t-test with day 0, p = 0.0218) between days 0 and 42 (figure 5). By contrast GIA levels in the PfAMA1 group did not change significantly (22.7%, 95% CI 22.9 to 6.6% at day 42, paired t-test with day 0, p = 0.226) (figure 5). The third immunisation increased GIA levels even further in the PkAMA1 group to 78.4% at day 70 (95% CI 61.8 to 94.9%, paired t-test with day 42, p = 0.003), while GIA levels remained unchanged in the PfAMA1 group 1.9% (, 95% CI 212.6 to 7.2%, paired t-test with day 42, p = 0.437). At the time of challenge (day 70) GIA levels were significantly higher in the PkAMA1 group as compared to the PfAMA1 control group (t-test, p,0.0001) (figure 5). GIA levels in the PkAMA1 group decreased significantly during the challenge phase to 46.7% (95% CI 15.2 to 78.3%, paired t-test with day 70, p = 0.0128) and a small non-significant decrease was observed for animals in the PfAMA1 group; 218.2% at day 91 (295% CI 235.1 to 21.4%, paired t-test with day 70, p = 0.053).
Antigen-specific lymphocyte stimulation assays showed that IFN-c levels were generally low with a median values of 25.4 (Quartile range: 17.7 to 36.3 pg mL 21 ) and 1 (Quartile range 1 to 1 pg mL 21 ) for the PkAMA1 and PfAMA1 groups, respectively. The difference in IFN-c levels was statistically significant (p = 0.025, Mann-Whitney test). The protected monkey in the PkAMA1 group produced the highest IFN-c level upon stimulation with PkAMA1, while all other samples were low to negative in this assay ( Figure 6). IL-13 levels were low; the median IL-13 values were 14.5 (Quartile range 2.8 to 84.2 pg mL 21 ) and 33. Plasmodium knowlesi blood-stage challenge after booster vaccination The results described above obtained after the primary challenge prompted us to extend the study with an additional boost and second challenge, similar to Deans and co-workers [6]. and was treated on day 17 ( Figure 3, Table 2). Time to patency and time to drug cure were significantly delayed in the PkAMA1 group (p = 0.009 and 0.0029, respectively).   Immunological analysis of the second challenge On the day of re-boost, anti-PkAMA1-IgG antibodies had decreased significantly in both groups compared to the levels after challenge (day 91) (PkAMA1 group: 11.2 kAU mL 21 , 95% CI 7.0 to 18.0, paired t-test with day 91, p = 0.0013 and PfAMA1 group:2.4 kAU mL 21 , 95% CI 1.8 to 3.1, paired t-test with day 91, p = 0.0052) (Figure 4). Boosting was effective and at the time of the second challenge (day 217) IgG levels had risen to 126.2 kAU mL 21 (95% CI 77.3 to 206.0, paired t-test with day 202, p,0.0001) and GIA levels had increased to 86.6% in the PkAMA1 group (95% CI 60.6 to 112.7%, paired t-test with day 202, p = 0.0002) (Figures 4 and 5). Anti-PkAMA1-IgG levels in the PfAMA1 group showed an increase to 26.6 kAU mL 21 (95% CI 18.8 to 37.8, paired t-test with day 202, p,0.0001). GIA levels increased only slightly in the PfAMA1 group to 23.4% (95% CI 220.2 to 13.4%, paired t-test with day 202, p = 0.0530). At the time of challenge IgG and GIA levels were significantly higher in the PkAMA1 group (IgG: 4.7-fold, 95% CI 2.8 to 8.1-fold, t-test, p,0.0001 and GIA: 90.1% higher, 95% CI 62.6 to 117.5%, p,0.0001).
For the PkAMA1 immunised group, GIAs done on IgG obtained five weeks after the second challenge did not decrease as markedly as after the first challenge (10% decrease, 95% CI 25.8 to 26.0%, paired t-test, p = 0.163) (figure 5). Conversely, GIA levels increased during the challenge in the PfAMA1 group  Protection after re-boost and heterologous blood-stage challenge with Pk Nuri strain To investigate whether the protection observed upon the second challenge was AMA1-mediated, monkeys were boosted with Pk or PfAMA1 and challenged with the Pk Nuri strain. The Pk Nuri AMA1 differs by only one amino acid from the H-strain, whilst other antigens expressed on the surface of infected erythrocytes that may contribute to protective responses (e.g. Schizont-Infected Cell Agglutination, SICA-Var) are expected to be different in a heterologous strain [11]. The heterologous challenge thus applies to parasite antigens other than AMA1, a situation considerably different from P. falciparum where the chance of finding two very closely related AMA1 molecules in different isolates is very small. The parasitaemias for the third challenge are depicted in the top right and bottom right panels of figure 3 and parasitological and immunological data are shown in table 2. All animals in the control group required treatment by day 8, while five out of six PkAMA1 vaccinated rhesus monkeys were able to control the parasitaemia, albeit not sterilely ( Figure 3, Table 2 Table 2). Time to patency and time to drug cure were significantly longer in the PkAMA1 group (log-rank test, p = 0.009 and 0.0009 for patency and cure, respectively). One monkey in the PkAMA1 group (R02033 [X]) could not control parasitaemia and required drug cure by day 9. This was the same monkey that did not show any signs of protection in the two previous challenges.

Immunological analysis of the third challenge
On the day of re-boost, anti-PkAMA1-IgG antibodies had decreased significantly to 15.0 kAU mL 21 (95% CI 6.8 to 33.4, paired t-test with day 252, p = 0.0008) for the PkAMA1 group and to 3.1 kAU mL 21 (95% CI 2.0 to 4.6, paired t-test with day 252, p,0.0001), for the PfAMA1 groups compared to the levels after the second challenge ( Figure 4). Boosting was effective and at the time of the third challenge (day 450), PkAMA1-IgG antibody levels had increased to 167.9 kAU mL 21 in the PkAMA1 group and to 23.6 kAU mL 21 in the PfAMA1 group (95% CI 87.0 to 324.0, paired t-test with day 435, p,0.0001 and 11.1 to 50.1, paired t-test with day 435, p = 0.0004 for Pk and PfAMA1 groups, respectively). At the time of the third challenge, anti PkAMA1-IgG levels were 7.1-fold higher (95% CI 3.0 to 17.0, t-test, p = 0.0005) in the PkAMA1 group as compared to the PfAMA1 group. Anti PkAMA1-IgG levels decreased significantly during the challenge period in both groups; to 89.9 (95% CI 52.6 to 153.8, paired t-test with day 450, p,0.0001) in the PkAMA1 group and to 13.2 (95% CI 7.5 to 23.3, paired t-test with day 450, p = 0.007) for the PfAMA1 group.
GIA levels appeared lower at day 450 as compared to day 217, but as these were determined in two independent runs, direct comparisons between days 217 and 450 can, unfortunately, not be made. GIA levels in the PkAMA1 group increased significantly following booster vaccination to 78.2% (95% CI 61.0 to 95.3, paired t-test with day 435, p = 0.0007). By contrast, GIA levels in the PfAMA1 group remained unchanged at 0.5% (95% CI 29.8 to 10.8%, paired t-test with day 435, p = 0.88). During the challenge phase, there was a tendency for decreasing GIA levels in the PkAMA1 group to 70.3% (95% CI 47.2 to 95.3%, paired t-test with day 450, p = 0.073), whereas GIA levels in the PfAMA1 group increased significantly to 18.3% (95% CI 4.9 to 31.7%, paired t-test with day 450, p = 0.024).
IFN-c levels increased significantly in both groups between the end of the second challenge (day 252) and the day of the second booster vaccination (day 435) (Wilcoxon signed rank test on pooled data for PkAMA1 and PfAMA1, p = 0.0005). Vaccination resulted in a significant decrease of IFN-c production in both groups (Wilcoxon signed rank test on pooled data for PkAMA1 and PfAMA1, p = 0.001) ( Figure 6). Levels of IFN-c did not change significantly during the challenge phase (Wilcoxon signed rank test on pooled data for PkAMA1 and PfAMA1, p = 0.30). IL-13 levels did not change significantly between the end of the second challenge and the day of the second booster vaccination in both groups (Wilcoxon signed rank test on pooled data for PkAMA1 and PfAMA1, p = 0.92) ( Figure 6). Booster vaccination, however, induced a significant increase in IL-13 levels in both groups (Wilcoxon signed rank test, p = 0.0024) ( Figure 6). IL-13 levels appeared to decrease in both groups during the challenge phase, but this failed to achieve statistical significance (Wilcoxon signed rank test on pooled data for PkAMA1 and PfAMA1, p = 0.11) (Figure 6). Relation between IgG and GIA The relation between IgG and GIA was investigated by modeling a sigmoid curve using non-linear least squares regression. Day 0 samples from both groups were included and PkAMA1 vaccinated animals were used for time points thereafter. Analysis was restricted to GIA values from one GIA run at 10 mg mL 21 IgG, which corresponds to the amount of total IgG present in rhesus serum (days 42, 70, 91, 202, 217 and 252). Thus the amounts of GIA and PkAMA1-specific IgG both correspond to values obtained with undiluted serum or plasma. The relationship between GIA and PkAMA1-specific IgG is graphically presented in figure 7. The correlation coefficient was 0.908 (95% CI 0.841 to 0.948, p,0.0001). The IC 50 value was estimated at 38 kAU mL 21 P. knowlesi AMA1-specific IgG (95% CI 31 to 46) and the fold difference in IgG between 10 and 90% GIA was estimated at 15.4 fold (95% CI 7.8 to 59.5) (Figure 7).

Correlates of Protection: Relation between GIA and in vivo growth inhibition
During the first challenge estimated In vivo growth rate ( Table 2 and Figure 8) was correlated with GIA IC 50 values (Spearman's Rho = 0.94, p = 0.017). This correlation did not change when recalculated with the assumption that monkey Ri9980203 was patent on day 10 or considered protected. This correlation was also independent of the assumption that the inoculation of one late trophozoite yielded four infected red blood cells (iRBC). Changing number of iRBC arising from one inoculated late trophozoite, however, did change the estimated growth rates in the vaccinated and in the control animals that developed parasitaemia (8.8, 6.7 and 5.8 for 1, 4 or 8 iRBC arising from one inoculated trophozoite, respectively) and in the vaccinated animals as well.
The significant correlation between GIA IC 50 values and in vivo growth was confirmed during the second challenge (Spearman's Rho = 0.94, p = 0.017) (Figure 8). When the correlation was recalculated using the first occurrence of patency for the estimation of in vivo growth rate, the correlation coefficient was 0.83 and not significant (p = 0.058). At the second challenge the four protected monkeys had IC 50 values ranging between 1.7 and 3.0 mg mL 21 IgG; the monkey with a delay had an IC 50 of 2.6 mg mL 21 , whereas the non-protected animal had an IC 50 value of 12.1 mg mL 21 IgG (Table 2).
At the third challenge the five protected animals all had IC 50 values lower than 3.4 mg mL 21 IgG and the non-protected animal had an IC 50 value of 11.1 mg mL 21 IgG ( Table 2). The correlation between GIA IC 50 and in vivo growth rate was, however, not observed at the third challenge (Spearman's Rho = 0.03, p = 1.0).

Discussion
The rhesus Plasmodium knowlesi model enables the investigation of in vivo protection following (repeated) mosquito or blood-stage challenge with a longer observation window as compared to challenge studies in humans. The model also offers the opportunity to assess clinical safety in non-human primates. This safety evaluation includes clinical observations (behaviour, appetite injection site inspection etc.) as well as full haematology and clinical chemistry evaluations. Therefore the rhesus Plasmodium knowlesi model can be a valuable tool for the down-selection of human malaria vaccine candidates and formulations (provided a Pk orthologue is available).
The main findings of the work reported here are: i.) Vaccination with heterologously-expressed PkAMA1 formulated with a potent adjuvant protects rhesus monkeys from subsequent blood-stage challenge (i.e. animals are able to control parasitaemia). ii.) Protection appeared to be mainly antibody-mediated, although a contribution of cellular immunity cannot be excluded. iii.) High levels of growth inhibiting antibodies are required for protection. iv.) The degree of protection improves upon repeated infection and boosting. v.) Protection against blood-stage infection can be induced by the use of a single antigen.
The results presented here are in agreement with the findings by Deans et al. [6], here confirmed with heterologously-expressed antigen and an adjuvant that can be used in human trials. Similar to results obtained by Deans et al. [6], solid protection was only achieved following challenge and booster vaccination. A series of three vaccinations with protein formulated in adjuvant may not yield high enough antibody and/or T-cell responses to confer protection. It is possible that exposure to live parasites followed by booster immunisation contributed to a more balanced humoral and cellular response.
The data suggest that high levels of growth inhibiting antibodies are required for protection. Moreover, in vitro and in vivo growth inhibition are correlated. Based on the correlation between parasitological measures (IVG, day to patency or day to treatment) and GIA IC 50 values it is tempting to speculate that in this experiment growth inhibiting antibodies are the main mediators of protection. This is in agreement with the results from Dutta et al. in the Aotus P. falciparum model [8], the observation that passive transfer of hyperimmune IgG can abrogate parasitaemia [35] and that transplacental IgG provides protection in neonates [36,37].
The PkAMA1 vaccinated animal that was protected during the first challenge had high GIA and IFN-c levels. At the second and third challenge, however, IFN-c increased in all animals (including non-protected controls) to levels similar as observed in the aforementioned animal. Thus high IFN-c levels alone are not a correlate, although it cannot be excluded that Ag-specific IFN-c production enhances antibody-mediated protection. Of note is that IFN-c levels increased after challenge and decreased following booster vaccinations, whereas the inverse was observed for IL-13 levels. Overall a gradual increase in IFN-c levels was observed for the Pk and PfAMA1 groups over the 3 challenges, suggesting improved cellular immune responses. In mice it has been demonstrated that AMA1-specific T-cells play an important role in the protection against P. chabaudi [38], suggesting that the improved cellular responses may have contributed to the protection observed at the second and third challenge.
Results obtained with heterologous prime-boost regimens with 4 antigens (2 pre-erythrocytic and 2 blood-stage, including AMA1) combined show that rhesus monkeys can be protected from sporozoite challenge [39,40]. Some animals became patent, but were able to control parasitaemia, similar to what is reported here. Unexpectedly, the protected animals in the study by Weiss et al. did not have appreciable GIA levels at the time of challenge and GIA levels only developed during challenge [41], suggesting that T cells induced by vaccination provided help for antibody formation upon parasite exposure. GIA and anti AMA1 or MSP1 19 IgG levels at the time of drug cure, however, were similar in animals that were able to control the parasitaemia and animals that were not [41]. This suggest that other, non identified, cell-mediated mechanisms may account for the protection, as IFN-c levels at the time of challenge did not explain the ability to self cure [39]. The increase of GIA titres during challenge is in contrast with data presented here; GIA levels waned during the first and second challenges in all 6 and 4 out of 6 PkAMA1 vaccinated animals. During the third challenge 3 PkAMA1 vaccinated animals retained their titre (,5% drop) and 3 decreased. The number of animals able to retain GIA levels increased with each challenge thus it appears that boosting and parasite exposure improves immune responses in PkAMA1 vaccinated animals, possibly through the expansion of AMA1-specific T cell responses.
The results presented here indicate that high GIA levels are required for protection (IC 50 #3.5 mg mL 21 ), but these levels wane quickly, even when exposed to parasites. This begs the question on how long protective levels can be maintained. The vaccination regimen used here closely resembles the WHO Expanded Programme on Immunisation comprising immunisations at 6, 10 and 14 weeks, and a vaccination at 9 months of age (www.who.int). Several papers indicate that immunisation timing [42], prime boost regimes or a combination of schedule and prime boost may be more effective at eliciting both cellular and antibody responses [39,43]. It is tempting to speculate that combinations of prime with DNA or viral vectors (e.g. adenovirus, attenuated pox viruses, attenuated measles, virosomes or other virus-like particles) followed by a (heterologous) virally vectored and protein adjuvant boost would induce high humoral and cellular responses and thereby yield more solid protection. Whether such an immune response elicited by prime boost schedules is higher, persist longer and would protect upon the first challenge remains to be investigated.
The protection observed in the work reported here appears in contrast with results from a recent phase IIa study [20] where only a small, biologically non relevant, delay in pre-patent modelled-PCR data was observed despite high GIA titres to the, homologous, challenge strain. The phase IIa study involved mosquito challenge of volunteers, whereas the infection of the animals was done by intravenous passage of a low number of (viz. 10,000) iRBC. It is possible that sporozoite passage through the skin may have induced tolerance as was recently hypothesised [44]. Thus the route of infection and initial infection load may, in part, explain the differences. Moreover, human subjects have to be drug cured when positive on a thick smear, while a longer observation window (up to 2% parasitaemia) was possible for the animals, thus the control of parasitaemia as observed in the current study could not have been observed in a phase IIa setting. In any case, it will be interesting to compare mosquito challenge with blood-stage challenge in PkAMA1-vaccinated rhesus monkeys in future studies.
A recently reported phase IIb trial with AMA1 formulated in AS02 in Malian children [23] revealed a hint of protection (17%, not significant) against all P. falciparum infections, but significant protection (64%) when infections with parasites bearing an AMA1 homologous to the vaccine antigen were considered. Infections occurred against a background of pre-existing immunity which was boosted by PfAMA1 vaccination; a situation similar to that of the animals at the second and third challenges.
The protection observed in Malian children [23], the Aotus model [8] and in the current study, suggest that phase IIa studies for downselection of vaccine candidates may need to be interpreted carefully. It would be interesting to explore whether clinically relevant protection would have been observed if the subjects in the AMA1 Phase IIa study [20] were boosted and re-challenged, as a series of vaccinations followed by infection and booster immunisation could be considered representative for the field situation.
In addition, the PkAMA1 vaccine candidate described here may also be an attractive human vaccine candidate given the fact that Plasmodium knowlesi is a zoonosis that affects humans and may be a contributing cause of death [13,14].