Received 30 June 2009/Returned for modification 24 July 2009/Accepted 21 October 2009

INFECTION AND IMMUNITY, Jan. 2010, p. 145 153 Vol. 78, No. 1 0019-9567/10/$12.00 doi:10.1128/iai.00740-09 Copyright 2010, American Society for Microbiology. All Rights Reserved. Prime-Boost Immunization with Adenoviral and Modified Vaccinia Virus Ankara Vectors Enhances the Durability and Polyfunctionality of Protective Malaria CD8 T-Cell Responses Arturo Reyes-Sandoval, 1 * Tamara Berthoud, 1 Nicola Alder, 2 Loredana Siani, 3 Sarah C. Gilbert, 1 Alfredo Nicosia, 3 Stefano Colloca, 3 Riccardo Cortese, 3 and Adrian V. S. Hill 1 The Jenner Institute, University of Oxford, Old Road Campus Research Building, Roosevelt Drive, Oxford OX3 7DQ, United Kingdom 1 ; Centre for Statistics in Medicine, Wolfson College Annexe, Linton Road, Oxford OX2 6UD, United Kingdom 2 ; and Okairòs, Via dei Castelli Romani 22, 00040 Pomezia, Rome, Italy 3 Received 30 June 2009/Returned for modification 24 July 2009/Accepted 21 October 2009 Protection against liver-stage malaria relies on the induction of high frequencies of antigen-specific CD8 T cells. We have previously reported high protective levels against mouse malaria, albeit short-lived, by a single vaccination with adenoviral vectors coding for a liver-stage antigen (ME.TRAP). Here, we report that primeboost regimens using modified vaccinia virus Ankara (MVA) and adenoviral vectors encoding ME.TRAP can enhance both short- and long-term sterile protection against malaria. Protection persisted for at least 6 months when simian adenoviruses AdCh63 and AdC9 were used as priming vectors. Kinetic analysis showed that the MVA boost made the adenoviral-primed T cells markedly more polyfunctional, with the number of gamma interferon (INF- ), tumor necrosis factor alpha (TNF- ), and interleukin-2 (IL-2) triple-positive and INF- and TNF- double-positive cells increasing over time, while INF- single-positive cells declined with time. However, IFN- production prevailed as the main immune correlate of protection, while neither an increase of polyfunctionality nor a high integrated mean fluorescence intensity (imfi) correlated with protection. These data highlight the ability of optimized viral vector prime-boost regimens to generate more protective and sustained CD8 T-cell responses, and our results encourage a more nuanced assessment of the importance of inducing polyfunctional CD8 T cells by vaccination. The feasibility of developing a malaria vaccine is widely supported by evidence showing that immune responses are important to limit infectivity by the plasmodial parasite. A key observation came from experiments showing that irradiated sporozoites induce protective immunity in animals and humans (6, 14), suggesting at the same time that the pre-erythrocytic stage of the parasite life cycle could be susceptible to the immune attack. In addition, targeting the parasite at the liver stage (LS) can be crucial to stop the infection before the symptoms appear during the blood stage (15). Induction of CD4 and CD8 T cells producing gamma interferon (IFN- ) is an important goal in malaria vaccination due to their major role in mediating protection during the liver stage (2). T-cell responses can effectively be induced by immunization with viral vectors such as fowlpox 9 (FP9) and modified vaccinia virus Ankara (MVA) (1, 10). However, a single administration of these vectors has given unsatisfactory results in animals in terms of protection against malaria. This has prompted the development of alternative methods of vaccination to enhance protection, such as the sequential administration of two vectors spaced by an interval of days or weeks, known as heterologous prime-boost regimens (8). * Corresponding author. Mailing address: The Jenner Institute, University of Oxford, Old Road Campus Research Building, Roosevelt Drive, Oxford OX3 7DQ, United Kingdom. Phone: 441865617621 or 447988880366. Fax: 441865617608. E-mail: arturo.reyes@ndm.ox.ac.uk. Published ahead of print on 26 October 2009. The authors have paid a fee to allow immediate free access to this article. It has recently been reported that a single vaccination with adenoviral (Ad) vectors coding for a pre-erythrocytic malarial antigen (Ag) elicited high CD8 T-cell numbers and outstanding short-term protection against murine malaria (17, 24). However, protective efficacy decreased substantially over a period of 8 weeks. Here, we report high levels of long-term sterile protection against malaria for up to 6 months after vaccination with prime-boost regimens involving an initial prime with either simian or human Ad vectors followed by MVA (Ad- MVA) coding for the pre-erythrocytic-stage malaria antigen ME.TRAP, a transgene that has been used in clinical trials and contains TRAP from Plasmodium falciparum fused in-frame to a multiepitope (ME) string with multiple B-cell, CD4, and CD8 epitopes, including the BALB/c H-2K d -restricted epitope from P. berghei known as Pb9 (SYIPSAEKI) (11). Previous studies with adenoviral vectors have been performed using the human serotype 5 (AdH5), to which most humans have been exposed during life, prompting the development of neutralizing antibodies that in turn impair the transgene-specific B- and T-cell responses elicited by vectors from the same serotype (12). We now extend those observations by assessing the performance of prime-boost regimens with alternative adenoviral serotypes of simian (chimpanzee) origin (AdC) that do not circulate in human populations and to which, therefore, neutralizing antibodies are rarely found (26). High protective levels were maintained during a long period of time of 26 weeks for two regimens that used simian Ad vectors (SAds) as priming agents: AdCh63 and AdC9. Flow cytometry combined with Boolean analysis revealed that polyfunctional CD8 T-cell responses initially primed by Ad can be Downloaded from http://iai.asm.org/ on April 22, 2018 by guest 145

146 REYES-SANDOVAL ET AL. INFECT. IMMUN. increased by a subsequent boost or even more after a third vaccination, especially when MVA is used as the last boost. However, no clear correlation of polyfunctionality with protective efficacy or level of cytokine secretion was evident, in contrast to results reported for protective CD4 T cells in a Leishmania vaccination model (7). Taken together, these results demonstrate the potential of Ad vectors used in combination with MVA in malaria vaccine development, providing preclinical evidence to support the use of the simian adenoviral vectors boosted by MVA in humans. MATERIALS AND METHODS Mice and immunizations. Female BALB/c mice 4 to 6 weeks of age were purchased from the Biomedical Services Unit at the John Radcliffe Hospital, Oxford. All animals were handled in accordance with the terms of the UK Home Office Animals Act Project License. We used the intradermal route of immunization, which has previously been shown to elicit similar immunogenicity compared to intramuscular administrations (17). MVA.ME.TRAP (MVA) and FP9.ME.TRAP (FP9) were administered at a dose of 1 10 7 PFU/mouse (2 10 8 PFU/ml), whereas adenoviral doses were between 5 10 9 viral particles (vp)/mouse (1 10 11 PFU/ml) and 1 10 10 vp/mouse (2 10 11 vp/ml). The interval between immunizations was 8 weeks. All vectors were administered in endotoxin-free phosphate-buffered saline (PBS). Viral vectors. All vectors express the transgene ME.TRAP that has previously been used in clinical trials (9, 11). The ME.TRAP insert is a hybrid transgene of 2,398 bp encoding a protein of 789 amino acids (aa) of P. falciparum TRAP. The ME string contains the BALB/c H-2K d epitope Pb9 among a number of other B- and T-cell epitopes (21). The simian adenoviral vectors (SAdV) were constructed and propagated as described previously (18). Construction of AdH5 (10), MVA (11), and FP9 (25) has been described earlier. AdCh63 ME.TRAP was constructed by similar methods to that for AdH5. Ex vivo IFN- ELISPOT. Ammonium chloride lysis buffer (ACK)-treated splenocytes or peripheral blood mononuclear cells (PBMCs) were cultured for 18 to 20 h on IPVH-membrane plates (Millipore) with the immunodominant H-2K d -restricted epitope Pb9 (SYIPSAEKI) at a final concentration of 1 g/ml. The enzyme-linked immunospot assay (ELISPOT) was performed as previously described (13). ICS. Intracellular cytokine staining (ICS) was performed as described earlier (17). Flow cytometric analyses were performed using a FACSCanto (BD Biosciences), and data were analyzed with either FACSDiva (BD) or Flow Jo (Tree Star) software. Analysis of polyfunctional CD8 T-cell responses was performed by using Boolean analysis in FlowJo software and SPICE 4.0, kindly provided by M. Roederer (NIH, Bethesda, MD). The mean fluorescence intensity (MFI) from the allophycocycanin (APC) fluorochrome coupled to anti-mouse IFN- was used to determine the integrated mean fluorescence intensity (imfi), which results from the multiplication of the % CD8 T cells expressing IFN- MFI, as described earlier (7). Antibody responses. Mouse serum was obtained 4 weeks after the MVA boost. Samples were diluted in microtiter plates coated with TRAP antigen (kindly supplied by GlaxoSmithKline Biologicals, Rixensart, Belgium) at a final concentration of 1 g/ml. Anti-TRAP antibody concentrations were measured by a specific IgG enzyme-linked immunosorbent assay (ELISA) to recombinant TRAP. Bound antibodies were detected by using alkaline phosphatase-conjugated antibodies specific for whole mouse IgG (Pharmingen). Serum samples from naïve mice were used as a background control. Parasite challenge. Plasmodium berghei (ANKA strain clone 234) sporozoites (spz) were isolated from salivary glands of female Anopheles stephensi mosquitoes. Parasites were resuspended in RPMI 1640 medium with each mouse receiving a total of 1,000 spz via the intravenous route. Blood samples were taken daily, from day 5 to 20; smears were stained with Giemsa and screened for the presence of schizonts within the red blood cells. Complete absence of parasites in blood following the sporozoite challenge was considered as sterile protection. Statistical analysis. The logistic regression to assess the relationship between the presence of polyfunctional CD8 T cells and protective efficacy was performed using Stata v.10. For all other analysis, GraphPad Prism version 5.0 was used. Prior to the use of a statistical analysis to compare two or more populations, the Kolmogorov-Smirnov test for normality was used to determine whether the values followed a Gaussian distribution. The kinetics of the polyfunctional T cells was analyzed by analysis of variance (ANOVA). Immune responses to the TABLE 1. Increased protective efficacy in prime-boost regimens Regimen 2wk (n 7) Ad prime % of animals with sterile protection by a : 10 wk (n 7) 2wk (n 5) AdC7-MVA and AdC7-AdC9 (see Fig. 4) regimens were compared using a t test. Protective efficacy to a challenge with P. berghei was analyzed by Kaplan-Meier survival analysis (Fig. 4 and Table 1). RESULTS Ad prime with MVA boost 8wk (n 5) 26 wk (n 3 5) AdH5 0 0 100** 60* 20 AdC6 0 0 80** 40* 0 AdC7 14 0 60* 40* 33 AdCh63 14 0 100** 40* 75*** AdC9 14 0 80** 60* 80*** FP9 0 0 40* 20 0 Naïve 0 0 0 0 0 a Shown are the sterile protection and efficacy of the vaccines to a parasite challenge. BALB/c mice were immunized with adenoviral (5 10 9 vp) and poxviral (1 10 7 PFU) vectors and then challenged 2 and 10 weeks after the prime and 2, 8, and 26 weeks after an MVA boost (n 3 to 7). Challenge was performed by intravenous administration of 1,000 sporozoites of Plasmodium berghei. Statistically significant differences are indicated by asterisks (, P 0.05; **, P 0.01; ***, P 0.001) and show comparison of individual regimens with the naïve control. Immunogenicity elicited by prime-boost regimens. We have previously reported on the immunogenicity elicited by a single vaccination with adenoviral and poxviral vectors expressing ME.TRAP (17). Our previous results also showed that heterologous prime-boost approaches can maximize the T-cell-mediated protection in animal models and that Ad/MVA regimens using P. berghei CSP can elicit complete protection against malaria (8, 10). In an attempt to enhance the immunogenicity and protection against P. berghei, we assessed the efficacy of prime-boost regimens using Ad and MVA vectors expressing ME.TRAP. Upon a single vaccination, the lowest numbers of CD8 IFN- cells were elicited by the poxviral vectors FP9 (59,000 antigen-specific cells/spleen) and MVA (480,000 cells/spleen). Adenoviral vectors elicited higher CD8 numbers when the vectors encoding ME.TRAP were administered at doses of 1 10 10 vp/mouse (690,000 Agspecific CD8 cells/spleen by AdC6, 1,300,000 by AdH5 and AdCh63, 2,000,000 by AdC7, and 2,600,000 by AdC9). These frequencies were enhanced further by a MVA boost given 8 weeks later and reached more than 4 million antigen-specific cells/spleen during the peak response. During the T-cell memory phase, this number contracted to 2 million/spleen (Fig. 1B and C). Antibody responses were also primed by all of the AdME.TRAP vectors and subsequently boosted by MVA ME.TRAP. Although all Ad-MVA ME.TRAP-encoding regimens elicited similar titers, the AdCh63-MVA regimen induced slightly better antibody responses than the rest of the Ad-MVA regimens. In addition, important differences were found when comparing the Ad-MVA to the FP9-MVA regimens; the latter regimen, which involves the successive administration of two poxviral vectors, generated barely detectable levels of antibodies (Fig. 1C), consistent with the very poor antibody immunogenicity of this regimen in clinical trials (25).

VOL. 78, 2010 ADENOVIRUS, MVA, AND POLYFUNCTIONAL CD8 RESPONSES 147 FIG. 1. Kinetics of the CD8 responses induced in Ad/MVA primeboost regimens and stimulation of antibodies by prime-boost regimens. (A) Groups of BALB/c mice (n 3) were immunized with 1 10 10 vp of human (AdH) and chimpanzee (AdC) adenovirus expressing ME.TRAP and boosted 8 weeks later with 1 10 7 PFU of MVA ME.TRAP. (B) Total number of CD8 T cells per spleen expressing IFN- upon stimulation with Pb9 peptide after Ad or FP9 prime and MVA boost. The number represents the average of 3 mice per group. (C) Antibody responses against TRAP in four different heterologous prime-boost regimens. Responses were assessed on week 4 after a boost with MVA. The graph displays only three Ad regimens (high, intermediate, and low) out of five tested and one poxviral regimen (FP9-MVA). O.D., optical density. Data are represented as means standard errors of the means (SEM). Adenoviral vectors used for immunizations are from the human serotype 5 (AdH5) and simian serotypes (AdC6, AdC7, AdC9, and AdCh63). The poxviral vectors are fowlpox 9 (FP9) and modified Vaccinia virus ankara (MVA). Polyfunctional CD8 T-cell responses are increased upon an MVA boost and display a further enrichment during a secondary memory phase. It has recently been established for the parasitic disease organism Leishmania major that increased frequencies of CD4 T cells producing simultaneously several cytokines, or polyfunctional T cells, correlate better with protection, due to the greater ability to produce more IFN- on a per cell basis (7). However, the correlation of CD8 T-cell functionality with efficacy is mainly inferred from studies of human infectious diseases, where it is difficult to determine causality in correlative studies of pathogen control and CD8 T-cell polyfunctionality (3, 23). Therefore, we have analyzed the effect of prime-boost regimens on the frequencies and kinetics of polyfunctional CD8 T cells and attempted to determine whether they associate with protection against liverstage malaria. We have previously shown that a single immunization with adenoviral vectors enhances the functionality of CD8 T cells compared to vaccination with poxviral vectors (MVA and FP9) (17). We started by assessing the kinetics of polyfunctional CD8 T cells (coproducing IFN-, tumor necrosis factor alpha [TNF- ], and interleukin-2 [IL-2], designated 3 cells ) after a single prime with AdC9 ME.TRAP and compared this to a prime-boost regimen with AdC9 ME.TRAP followed by MVA ME.TRAP 8 weeks later. As shown in Fig. 2A to C, 3 cells constitute a minor component after a single prime ( 0.3% for all regimens), peaking at 2 weeks and decreasing afterwards. Interestingly, the frequency of 3 CD8 T cells was increased 2 weeks after MVA ME.TRAP boost and continued to become further enriched with time. Twenty-six weeks after the last immunization, the frequency of 3 CD8 T cells reached a relative percentage of nearly 50% of the induced epitopespecific CD8 response (Fig. 2B) and more than 10% of the absolute numbers, being significantly higher than at the other time points (Fig. 2C, purple bar). Conversely, CD8 T cells producing 2 cytokines (2 ) displayed different kinetics, increasing shortly after the prime, contracting during the primary memory response phase at 10 weeks and increasing upon MVA boost to remain stable over a period of 26 weeks (Fig. 2B). Lastly, the 1 cells producing only IFN- and none of the other cytokines initially peaked at 2 weeks postprime and decreased afterwards, and their frequency did not increase significantly as a consequence of the boost vaccine. The kinetics of polyfunctional CD8 T cells were similar for all regimens regardless of the initial vector used for priming (data not shown). Finally, we assessed the frequency and functionality of the CD8 responses in the liver, the homing organ of the sporozoites. Three vaccination regimens were included for this analysis: AdC9 ME.TRAP alone, MVA ME.TRAP alone, and a prime-boost with both vectors (Fig. 2D). We detected outstandingly high IFN- responses following the prime-boost regimen (61.6%) or a single immunization with AdC9 ME.TRAP (49.6%), while MVA elicited low frequencies of IFN- cells (9.26%). In terms of multifunctionality, liver CD8 responses consisted mainly of single-cytokine IFN- producers (MVA AdC9 AdC9-MVA), followed by double-positive CD8 cells expressing IFN- and TNF- (AdC9-MVA AdC9 MVA). As can be seen in Fig. 2D, only double-positive responses can be improved by vaccination with an adenoviral vector and further enhanced in prime-boost regimens. CD8 frequencies and functionality are increased by double and triple immunizations, mainly with MVA as a boost. Multiple immunizations were given to BALB/c mice to determine whether the frequencies and functionality of CD8 T cells could be increased using heterologous adenovirus immunization or a combination of two different Ad vectors and MVA (Fig. 3A). The overall IFN- responses were improved especially when MVA ME.TRAP was used as a final boost and were significantly higher than the frequencies elicited by heterologous Ad-Ad immunizations without MVA boost (P 0.01 for AdC7/AdC9/MVA versus AdC7/AdC9/AdH5; P 0.05 for AdC9/MVA versus AdC9/AdH5) (Fig. 3B). On the other hand, the sequential administration of three heterologous adenoviruses did not induce substantially higher frequencies than two heterologous adenoviral vectors. Polyfunctional CD8 responses were also enhanced by the use of MVA as a

148 REYES-SANDOVAL ET AL. INFECT. IMMUN. FIG. 2. Kinetic responses of the generation of polyfunctional CD8 T-cell responses upon a prime or a prime-boost regimen. (A) Groups of 5 BALB/c mice were immunized with 5 10 9 vp of adenovirus and 1 10 6 PFU of the poxvirus MVA. Intracellular cytokine staining (ICS) to quantify production of IFN-, TNF-, and IL-2 from CD8 T-cells was performed in blood at different time points upon stimulation with the immunodominant Pb9 peptide. Graphs were generated after performing a Boolean analysis in FlowJo (Graphpad) and data analysis in SPICE software (Mario Roederer, VRC, NIH). (B) Kinetics of the CD8 responses after a single prime with AdC9 ME.TRAP or a prime-boost with AdC9-MVA. Pie charts display relative percentages of CD8 cells producing one (1 ), two (2 ), or three (3 ) cytokines at various time points. (C) Kinetics of the polyfunctional responses in absolute numbers showing 1, 2, and 3 cells. (D) CD8 responses in liver induced by various regimens 2 weeks after immunization. Mice were vaccinated as mentioned above, and T cells were isolated from perfused livers. Plots represent the % of IFN- in the CD8 compartment upon stimulation with Pb9. Pie charts show relative percentages of CD8 cells producing one, two, or three cytokines. Plots shown are from representative mice, and multifunctionality data in pie charts were calculated from 3 mice per group.

VOL. 78, 2010 ADENOVIRUS, MVA, AND POLYFUNCTIONAL CD8 RESPONSES 149 Downloaded from http://iai.asm.org/ FIG. 3. Generation of multifunctional CD8 T-cell responses after single-, double-, or triple-vaccination regimens. Groups of 6 BALB/c mice were immunized as described (A). Intervals between prime and boost were 8 weeks, and immune responses were assessed 2 and 8 weeks after the last vaccination. CD8 T cells from blood were stimulated, stained, and analyzed as described in the legend to Fig. 2. (B) Percentage of CD8 T cells producing IFN- in different prime-boost regimens, without taking into consideration coproduction of other cytokines. (C) Percentage of CD8 T cells coproducing the three cytokines (3 ) IFN-, TNF-, and IL-2. (D) Frequencies of CD8 T cells coproducing the two cytokines (2 ) IFN- and TNF-. (E) Percentage of CD8 T-cells producing only IFN- (1 ) and none of the other cytokines. Data are represented as means SEM. on April 22, 2018 by guest final boost, especially in triple immunizations (Fig. 3C). A similar pattern was present in 2 (Fig. 3D) and 1 (Fig. 3E) CD8 cells, in which the highest frequencies were obtained after the boost with MVA. Interestingly, both 3 and 2 polyfunctional T cells increased in frequency with time, whereas 1 cells peaked shortly after the boost but did not persist or enrich over time and instead contracted to very low levels (Fig. 3B to D). In conclusion, these results show that using MVA as a boost in double- or triple-vaccination regimens can augment the proportion of persisting polyfunctional CD8 T cells as well as overall T-cell numbers. Prime-boost regimens induce long-term protection against a malaria challenge. It has previously been shown that a single dose of Ad vectors coding for ME.TRAP can elicit high levels of sterile protection to malaria in a mouse model (17). However, such protective levels can only be achieved for a short period after injecting high doses of 1 10 10 vp/mouse and decrease substantially after 8 weeks. Prime-boost regimens, on the other hand, have the advantage of permitting the use of lower doses of vectors to obtain higher and more durable T-cell frequencies. The ability of lower doses to elicit good immune responses could decrease vaccine reactogenicity when used in humans. As shown in Table 1, a single prime at the dose of 5 10 9 vp/mouse induced only modest protective levels with three regimens after 2 weeks (14% each for AdC7, AdC9, and AdCh63), which were absent by week 10. A boost with MVA substantially improved protection, displaying a range of sterile protection from 100% with AdCh63-MVA and AdH5-MVA, to 40% in the FP-MVA regimen. Protection with the latter regimen decreased with time from 40% to 20% and 0% at 2, 8, and 26 weeks postboost, respectively. Importantly,

150 REYES-SANDOVAL ET AL. INFECT. IMMUN. protection levels with most of the Ad-MVA regimens remained high at week 8 postboost (especially when primed with AdCh63 and AdH5) and strikingly so after 26 weeks, when AdCh63-MVA and AdC9-MVA were able to induce remarkable 75% and 80% rates of sterile protection, respectively (Table 1). Incidentally, this was also the time point when the polyfunctional 3 CD8 responses reached the highest levels. In conclusion, Ad-MVA prime-boost regimens, especially those using the chimpanzee-derived AdCh63 and AdC9 as a prime, can elicit high levels of sterile protection that persist over a period of 6 months after the boost. Relationship between protective efficacy and presence of polyfunctional CD8 T cells at the time of challenge. In an attempt to determine if the induction of polyfunctional CD8 responses in blood is associated with protection, a logistic regression analysis was used to assess whether sterile protection was associated with the percentage of T cells producing different combinations of cytokines in blood. This analysis was performed across three independent experiments in which mice were administered prime-boost regimens that included Ad-MVA, heterologous Ad-Ad, and FP-MVA regimens and challenged at different time points after a prime or a prime boost. An additional logistic regression was carried out including the time of challenge as a covariate using 3 possible levels for this variable: 2 weeks postprime, 8 weeks postboost, or 26 weeks postboost. The aim of this analysis was to attempt to define the relationship between CD8 functionality and protection against challenge, regardless of the vaccine regimen and the time of challenge after vaccination. The calculated odds ratios (OR) indicated that for each unit increase in the percentage of cells producing 3 cytokines (3 : IFN-, TNF-, and IL-2 ), there was a 5% reduction in the odds of being infected (OR 0.95; 95% confidence interval [CI], 0.83 to 1.08). However, this was not statistically significant (P 0.42), and the 95% confidence interval suggested that there could be as much as an 8% increase. Having adjusted for time of challenge, the odds of being infected were reduced by 3% for each unit increase in the percentage of cells producing 3 cytokines (OR 0.97; 95% CI, 0.76 to 1.22). Again, this difference was not statistically significant (P 0.77). For double-positive cells (2 : IFN-, TNF-, and IL-2 ), there was no significant change in odds of infection as the percentage of cells producing 2 cytokines increased (OR 0.99; 95% CI, 0.94 to 1.04; P 0.75). However, having adjusted for time of challenge, the effect became significant (OR 0.88; 95% CI, 0.81 to 0.96; P 0.004). The odds ratio indicated a 12.5% reduction in odds of infection for each unit increase in percentage of cells producing 2 cytokines, and the 95% confidence interval indicated that this reduction in odds could be between 4% and 20%. When assessed for the presence of single-positive cells (1 IFN-, TNF-, IL-2 ), there was a significant decrease in odds of infection as the percentage of cells producing 1 cytokine increased (OR 0.87; 95% CI, 0.80 to 0.95; P 0.002). The unadjusted odds ratio indicated that the odds decreases by 13% per unit increase in the percentage of cells producing 1 cytokine. The odds ratio adjusted for time of challenge indicated that the odds decreased by 11% per unit increase in the percentage of these cells and was statistically significant (OR 0.89; 95% CI, 0.80 to 0.97; P 0.01). The total production of INF- by CD8 was also included in the analysis, without considering coproduction of TNF- and IL-2. The odds ratio indicated that for each unit increase in the percentage of cells producing IFN-, there was a 4% reduction in the odds of being infected. This was marginally statistically significant (OR 0.96; 95% CI, 0.92 to 1.00; P 0.04), and the 95% confidence interval suggested that there could be up to an 8% decrease. Having adjusted for time of challenge, the odds of being infected were reduced by 7% for each unit increase in the percentage of cells producing IFN-. This result was statistically significant (OR 0.93; 95% CI, 0.89 to 0.98; P 0.004), and the 95% confidence interval indicated that the true decrease in odds ranged between 2% and 11%. Finally, the MFI and imfi, which is calculated as % of IFN- MFI (7), were also analyzed for correlation with protection; however, the odds ratio indicated that for each unit increase the odds of being infected remained unchanged (OR 1.00 for both). In summary, we found that increased numbers of CD8 T cells producing IFN- are associated with a decrease in the odds of infection. These results were statistically significant for total IFN-, single IFN- producers (1 ) and IFN- TNF- (2 ) CD8 T cells but not for 3 polyfunctional cells nor for overall level of MFI or imfi. Comparison of the immune responses and protective efficacy using Ad-MVA and heterologous Ad (Ad-hetAd) primeboost regimens. The availability of multiple viral vectors increases the potential combinations available for use in prime-boost regimens. Based on results from previous reports, two combinations were compared for antibody and T-cell responses as well as protective efficacy. The Ad-MVA (A-M) regimen has previously been reported to enhance immunogenicity and protection when compared to other regimens involving the use of DNA and MVA (10), heterologous Ad-Ad (Ad-hetAd) regimens have also been tested in an HIV model, showing potent T-cell responses in double and triple immunizations, in both mice and macaques, although the lack of a challenge system did not allow protective efficacy to be assessed (16). In this study, shortly after an MVA (A-M) or an AdC7 (Ad-hetAd) boost (2 weeks after the final vaccination), both groups of mice displayed similar frequencies of CD8 T-cell responses and protective efficacy in the challenge with P. berghei sporozoites. The Ad-hetAd heterologous regimens induced slightly more potent responses and better protection, but none of these differences were statistically significant when assessed by a t test (Fig. 4A). However, differences in immunogenicity and protection became evident in the long term: the A-M regimens generated CD8 responses that remained high for a longer period of time, being significantly higher than the Ad-hetAd regimen at week 8 postboost (P 0.02; Fig. 4B) but only marginally higher after 26 weeks (Fig. 4C). Protection against challenge was also better in the A-M than in the A- heta regimens at the later time points, at both weeks 8 and 26 after boost, when the A-hetA combination elicited no protection at all (Fig. 4A to C). It has previously been reported that Ad vectors can induce potent antibody responses to ME.TRAP upon vaccination but that the poxviral vectors MVA and FP9 induce generally weaker antibody responses (17). Therefore, we hypothesized

VOL. 78, 2010 ADENOVIRUS, MVA, AND POLYFUNCTIONAL CD8 RESPONSES 151 that the sequential use of two Ad vectors could induce stronger antibody responses than an A-M regimen. Our results showed that the antibody responses are similar for both A-hetA and A-M regimens, suggesting that the initial vaccination is important for priming a B-cell response that can be boosted afterwards by the use of any of these recombinant vectors (Fig. 4D). As can be seen from these results, despite the fact that the Ad vectors induce more potent T- and B-cell responses when administered alone, a combination of two Ad vectors does not give an advantage in terms of the immune responses elicited when compared to the A-M regimens. Moreover, the use of MVA as a boost affords better protective levels when the responses have been primed by the Ad vectors. DISCUSSION FIG. 4. Immunogenicity and protective efficacy of prime-boost regimens using heterologous Ad-Ad or Ad-MVA regimens. Groups of 3 to 5 BALB/c mice were immunized as described in the legend to Fig. 2 with AdC7 ME.TRAP (5 10 9 vp) and subsequently boosted after 8 weeks with MVA (1 10 7 PFU) or AdC9 (5 10 9 vp), both coding for the same ME.TRAP transgene. CD8 T-cell responses and protective efficacy were assessed at different time points after the boost, at 2 weeks (A), 8 weeks (B), and 26 weeks (C). (D) Comparison of the antibody responses in the A-A versus A-M regimens. Sera were taken on week 4 after a boost, and IgG responses were assessed as described in Materials and Methods. The graph shows the optical density at 405 nm in serial dilutions of the sera. SFC, spot-forming cells. Data are represented as means SEM. An obligatory step in the malaria life cycle is the infection of hepatocytes by a few sporozoites, giving rise to thousands of merozoites that are released into circulation. The liver (or pre-erythrocytic) stage is one of the most appealing phases for vaccination against malaria to prevent pathology, and it requires the induction of relatively high frequencies of CD8 T cells to eliminate the infected hepatocytes. Subunit vaccination in the form of DNA or viral vectors is one of the strategies that can induce good T-cell responses. A large number of vectors have been tested for the pre-erythrocytic stage of the disease, such as DNA (22), Ty virus-like particles (9), fowlpox virus FP9 (1), and MVA and adenovirus (10). However, the use of individual vectors for vaccination elicits responses that in most cases yielded low or no protective levels in animal models, thus requiring the consecutive administration of two vectors spaced by a period of time, a procedure that has given rise to the concept of heterologous or synergistic prime-boost regimens (8). In this study, we show that prime-boost regimens using both human and chimpanzee adenovirus followed by MVA (A-M) can elicit sustained strong CD8 T-cell responses that surpass those achieved by a single prime or by an FP9-MVA regimen, which has already shown some efficacy in human clinical trials (25). Moreover, some A-M regimens were able to elicit complete, sterile protection in the short term and outstanding high levels of protection up to a period of 182 days after the last vaccination when chimpanzee adenoviruses AdCh63 and AdC9 were used as priming agents. Recently, we have achieved comparable and more durable immunogenicity and efficacy using peptide-coated dendritic cells and recombinant Listeria monocytogenes vectors in this murine malaria model, but this approach is not readily translatable to a widespread vaccination of humans (20). The regimens involving an adenovirus prime followed by an MVA boost were selected based on previous results showing that such combination could result in extremely high numbers of peptide-specific CD8 T cells and complete protection against challenge with P. berghei (10). Similar studies have been performed using the Plasmodium yoelii model in mice, in which a regimen using AdH5 followed by a boost with a recombinant attenuated vaccinia virus induced a complete, longlasting protection against malaria (5). To circumvent problems of preexisting immunity against AdH5 in humans, we have developed alternative serotypes from simian origin that do not circulate in human populations (26). By using these vectors in prime-boost strategies, we were able to amplify the CD8 T-cell responses and enhance protection for a long period of time and still decrease the dose to half the initial Ad dose that

152 REYES-SANDOVAL ET AL. INFECT. IMMUN. has been reported to elicit good, albeit short-term protection, thus augmenting the potential for safety and tolerability if the chimpanzee adenoviral vectors are to be tested in humans. Recently, there has been an increased interest in measuring polyfunctional responses in T cells in order to better identify correlates of protection for diseases such as HIV (3) and Leishmania infection (7). This has been prompted by the failure to identify conclusive T-cell correlates of protection for several disease models by measuring only the magnitude of the ELISPOT responses or the total T-cell frequencies by flow cytometry. The levels of functionality of human CD8 T cells are higher in HIV nonprogressors than in progressors, and the presence of such polyfunctional CD8 T cells correlates negatively with the viral load in progressors (3). However, it is impossible to determine from such observations whether the polyfunctionality is causally involved in protection or is a consequence of decreased viral and antigen load. Nevertheless, polyfunctional CD4 T cells are generated in mice by protective vaccines against Leishmania major (7) and tuberculosis (by the Mycobacterium bovis BCG vaccine) (4). In the case of L. major, both vaccine-induced polyfunctional CD4 T cells and CD4 T cells with enhanced IFN- production clearly correlate with protection. However, comparable evidence for CD8 T cells has been lacking. We have previously reported that vaccination with Ad vectors induces higher levels of polyfunctional antigen-specific CD8 T cells when compared to poxviral vectors (17). We now describe that CD8 functionality in Ad-primed responses can be enhanced by a booster vaccination, especially with MVA. Moreover, polyfunctional responses increased dramatically with the time, reaching values of triple-positive cells (IFN-, TNF-, and IL-2 ) of more than 10% of absolute frequency, which have not previously been reported. Such high levels might be an indicative of a generation of a secondary memory immune response, since such percentages were not reached after a single prime. Upon a search for a correlation between the generation of polyfunctional responses and protective efficacy, we found that an increase in cells producing IFN- without or with coproduction of TNF- was significantly associated with higher levels of protection, regardless of the regimen and time of challenge. However, importantly we found no evidence that cells secreting more IFN- nor 3 polyfunctional T cells were more protective. This contrasts with findings for CD4 T cells in the Leishmania model and suggests that the most relevant functional type of T cells may need to be defined for each T-cell subset and each disease. Despite the lack of increased efficacy of 3 cells on a per cell basis, we found that 3 CD8 T cells were associated temporally with the well-maintained protection observed at 180 days after Ad-MVA prime-boost immunization. The finding of a marked increase in IL-2-positive cells with time using this regimen was surprising and, to our knowledge, has not been previously described in vaccination studies. However, it is consistent with the delayed increase in IL-2-positive cells observed after natural infection in a lymphocytic choriomeningitis virus (LCMV) model (19). It appears likely that the importance of inducing polyfunctionality in this malaria infection model related more to the durability of vaccine-induced protection than a greater protective capacity on a per cell basis of the more polyfunctional phenotype. In terms of clinical development, these findings in murine models suggest that polyfunctionality of induced T cells should be assessed at many time points in clinical trials, not just shortly after vaccination, and there is already evidence of a late enhancement of proliferative capacity of T cells after vector vaccination in humans (4). Moreover, as challenge studies are key to the clinical development of pre-erythrocytic malaria vaccines, the slow maturation of the cellular immune response profile after particular prime-boost regimens may lead to a reconsideration of the standard challenge time points in phase II studies, currently 2 weeks post-final vaccination. Finally, the Ad vectors are characterized not only by the strong T-cell responses elicited, but also by the potent B-cell responses that are stimulated even after a single vaccination. This prompted us to test whether the heterologous A-A regimens could offer an advantage over the A-M regimens in terms of antibody induction. Our results showed that primed B-cell responses can be boosted equally by Ad or MVA vectors, suggesting that the priming agent is important to define the type of immune response that will be expanded afterwards by a boosting agent. Additionally, heterologous regimens that involve the boost with MVA in immunizations were more protective and generated more functional CD8 T-cell responses and outstanding frequencies that reached 45% of CD8 cells producing IFN-. This is important to define due to the wider availability of Ad and poxvirus vectors to be used in vaccination. In conclusion, our results show that A-M prime-boost regimens are able to induce strong and durable CD8 T-cell responses capable of eliciting high levels of sterile protection long after the final vaccination. This was particularly clear with the use of two promising adenoviral vectors of chimpanzee origin, AdCh63 and AdC9, coding for ME.TRAP. Our results on the kinetics and relative protective efficacies of CD8 T cells expressing various cytokines encourage a more nuanced assessment of the importance of inducing polyfunctional CD8 T cells by vaccination. ACKNOWLEDGMENTS This work was supported by the Wellcome Trust. A.V.S.H. is a Wellcome Trust Principal Research Fellow. We thank Anita Milicic for critically reading the manuscript and for helpful suggestions. REFERENCES 1. Anderson, R. J., C. M. Hannan, S. C. Gilbert, S. M. Laidlaw, E. G. Sheu, S. Korten, R. Sinden, G. A. Butcher, M. A. Skinner, and A. V. Hill. 2004. Enhanced CD8 T cell immune responses and protection elicited against Plasmodium berghei malaria by prime boost immunization regimens using a novel attenuated fowlpox virus. J. Immunol. 172:3094 3100. 2. Belnoue, E., F. T. Costa, T. Frankenberg, A. M. Vigario, T. Voza, N. Leroy, M. M. Rodrigues, I. Landau, G. Snounou, and L. Renia. 2004. Protective T cell immunity against malaria liver stage after vaccination with live sporozoites under chloroquine treatment. J. Immunol. 172:2487 2495. 3. Betts, M. R., M. C. Nason, S. M. West, S. C. De Rosa, S. A. Migueles, J. Abraham, M. M. Lederman, J. M. Benito, P. A. Goepfert, M. Connors, M. Roederer, and R. A. Koup. 2006. HIV nonprogressors preferentially maintain highly functional HIV-specific CD8 T cells. Blood 107:4781 4789. 4. Beveridge, N. E., D. A. Price, J. P. Casazza, A. A. Pathan, C. R. Sander, T. E. Asher, D. R. Ambrozak, M. L. Precopio, P. Scheinberg, N. C. Alder, M. Roederer, R. A. Koup, D. C. Douek, A. V. Hill, and H. McShane. 2007. Immunisation with BCG and recombinant MVA85A induces long-lasting, polyfunctional Mycobacterium tuberculosis-specific CD4 memory T lymphocyte populations. Eur. J. Immunol. 37:3089 3100. 5. Bruna-Romero, O., G. Gonzalez-Aseguinolaza, J. C. Hafalla, M. Tsuji, and R. S. Nussenzweig. 2001. Complete, long-lasting protection against malaria of mice primed and boosted with two distinct viral vectors expressing the same plasmodial antigen. Proc. Natl. Acad. Sci. U. S. A. 98:11491 11496.

VOL. 78, 2010 ADENOVIRUS, MVA, AND POLYFUNCTIONAL CD8 RESPONSES 153 6. Clyde, D. F., H. Most, V. C. McCarthy, and J. P. Vanderberg. 1973. Immunization of man against sporozite-induced falciparum malaria. Am. J. Med. Sci. 266:169 177. 7. Darrah, P. A., D. T. Patel, P. M. De Luca, R. W. Lindsay, D. F. Davey, B. J. Flynn, S. T. Hoff, P. Andersen, S. G. Reed, S. L. Morris, M. Roederer, and R. A. Seder. 2007. Multifunctional TH1 cells define a correlate of vaccinemediated protection against Leishmania major. Nat. Med. 13:843 850. 8. Gilbert, S. C., V. S. Moorthy, L. Andrews, A. A. Pathan, S. J. McConkey, J. M. Vuola, S. M. Keating, T. Berthoud, D. Webster, H. McShane, and A. V. Hill. 2006. Synergistic DNA-MVA prime-boost vaccination regimes for malaria and tuberculosis. Vaccine 24:4554 4561. 9. Gilbert, S. C., M. Plebanski, S. J. Harris, C. E. Allsopp, R. Thomas, G. T. Layton, and A. V. Hill. 1997. A protein particle vaccine containing multiple malaria epitopes. Nat. Biotechnol. 15:1280 1284. 10. Gilbert, S. C., J. Schneider, C. M. Hannan, J. T. Hu, M. Plebanski, R. Sinden, and A. V. Hill. 2002. Enhanced CD8 T cell immunogenicity and protective efficacy in a mouse malaria model using a recombinant adenoviral vaccine in heterologous prime-boost immunisation regimes. Vaccine 20: 1039 1045. 11. McConkey, S. J., W. H. Reece, V. S. Moorthy, D. Webster, S. Dunachie, G. Butcher, J. M. Vuola, T. J. Blanchard, P. Gothard, K. Watkins, C. M. Hannan, S. Everaere, K. Brown, K. E. Kester, J. Cummings, J. Williams, D. G. Heppner, A. Pathan, K. Flanagan, N. Arulanantham, M. T. Roberts, M. Roy, G. L. Smith, J. Schneider, T. Peto, R. E. Sinden, S. C. Gilbert, and A. V. Hill. 2003. Enhanced T-cell immunogenicity of plasmid DNA vaccines boosted by recombinant modified vaccinia virus Ankara in humans. Nat. Med. 9:729 735. 12. McCoy, K., N. Tatsis, B. Korioth-Schmitz, M. O. Lasaro, S. E. Hensley, S. W. Lin, Y. Li, W. Giles-Davis, A. Cun, D. Zhou, Z. Xiang, N. L. Letvin, and H. C. Ertl. 2007. Effect of preexisting immunity to adenovirus human serotype 5 antigens on the immune responses of nonhuman primates to vaccine regimens based on human- or chimpanzee-derived adenovirus vectors. J. Virol. 81:6594 6604. 13. Moore, A. C., A. Gallimore, S. J. Draper, K. R. Watkins, S. C. Gilbert, and A. V. Hill. 2005. Anti-CD25 antibody enhancement of vaccine-induced immunogenicity: increased durable cellular immunity with reduced immunodominance. J. Immunol. 175:7264 7273. 14. Nussenzweig, R. S., J. Vanderberg, H. Most, and C. Orton. 1967. Protective immunity produced by the injection of x-irradiated sporozoites of Plasmodium berghei. Nature 216:160 162. 15. Prudencio, M., A. Rodriguez, and M. M. Mota. 2006. The silent path to thousands of merozoites: the Plasmodium liver stage. Nat. Rev. Microbiol. 4:849 856. 16. Reyes-Sandoval, A., J. C. Fitzgerald, R. Grant, S. Roy, Z. Q. Xiang, Y. Li, G. P. Gao, J. M. Wilson, and H. C. Ertl. 2004. Human immunodeficiency Editor: J. F. Urban, Jr. virus type 1-specific immune responses in primates upon sequential immunization with adenoviral vaccine carriers of human and simian serotypes. J. Virol. 78:7392 7399. 17. Reyes-Sandoval, A., S. Sridhar, T. Berthoud, A. C. Moore, J. T. Harty, S. C. Gilbert, G. Gao, H. C. Ertl, J. C. Wilson, and A. V. Hill. 2008. Single-dose immunogenicity and protective efficacy of simian adenoviral vectors against Plasmodium berghei. Eur. J. Immunol. 38:732 741. 18. Roy, S., G. Gao, Y. Lu, X. Zhou, M. Lock, R. Calcedo, and J. M. Wilson. 2004. Characterization of a family of chimpanzee adenoviruses and development of molecular clones for gene transfer vectors. Hum. Gene Ther. 15:519 530. 19. Sarkar, S., V. Teichgraber, V. Kalia, A. Polley, D. Masopust, L. E. Harrington, R. Ahmed, and E. J. Wherry. 2007. Strength of stimulus and clonal competition impact the rate of memory CD8 T cell differentiation. J. Immunol. 179:6704 6714. 20. Schmidt, N. W., R. L. Podyminogin, N. S. Butler, V. P. Badovinac, B. J. Tucker, K. S. Bahjat, P. Lauer, A. Reyes-Sandoval, C. L. Hutchings, A. C. Moore, S. C. Gilbert, A. V. Hill, L. C. Bartholomay, and J. T. Harty. 2008. Memory CD8 T cell responses exceeding a large but definable threshold provide long-term immunity to malaria. Proc. Natl. Acad. Sci. U. S. A. 105:14017 14022. 21. Schneider, J., S. C. Gilbert, T. J. Blanchard, T. Hanke, K. J. Robson, C. M. Hannan, M. Becker, R. Sinden, G. L. Smith, and A. V. Hill. 1998. Enhanced immunogenicity for CD8 T cell induction and complete protective efficacy of malaria DNA vaccination by boosting with modified vaccinia virus Ankara. Nat. Med. 4:397 402. 22. Sedegah, M., R. Hedstrom, P. Hobart, and S. L. Hoffman. 1994. Protection against malaria by immunization with plasmid DNA encoding circumsporozoite protein. Proc. Natl. Acad. Sci. U. S. A. 91:9866 9870. 23. Seder, R. A., P. A. Darrah, and M. Roederer. 2008. T-cell quality in memory and protection: implications for vaccine design. Nat. Rev. Immunol. 8:247 258. 24. Sridhar, S., A. Reyes-Sandoval, S. J. Draper, A. C. Moore, S. C. Gilbert, G. P. Gao, J. M. Wilson, and A. V. Hill. 2008. Single-dose protection against Plasmodium berghei by a simian adenovirus vector using human cytomegalovirus promoter containing intron A. J. Virol. 82:3822 3833. 25. Webster, D. P., S. Dunachie, J. M. Vuola, T. Berthoud, S. Keating, S. M. Laidlaw, S. J. McConkey, I. Poulton, L. Andrews, R. F. Andersen, P. Bejon, G. Butcher, R. Sinden, M. A. Skinner, S. C. Gilbert, and A. V. Hill. 2005. Enhanced T cell-mediated protection against malaria in human challenges by using the recombinant poxviruses FP9 and modified vaccinia virus Ankara. Proc. Natl. Acad. Sci. U. S. A. 102:4836 4841. 26. Xiang, Z., Y. Li, A. Cun, W. Yang, S. Ellenberg, W. M. Switzer, M. L. Kalish, and H. C. Ertl. 2006. Chimpanzee adenovirus antibodies in humans, sub- Saharan Africa. Emerg. Infect. Dis. 12:1596 1599.

INFECTION AND IMMUNITY, May 2011, p. 2131 Vol. 79, No. 5 0019-9567/11/$12.00 doi:10.1128/iai.00185-11 Copyright 2011, American Society for Microbiology. All Rights Reserved. ERRATUM Prime-Boost Immunization with Adenoviral and Modified Vaccinia Virus Ankara Vectors Enhances the Durability and Polyfunctionality of Protective Malaria CD8 T-Cell Responses Arturo Reyes-Sandoval, Tamara Berthoud, Nicola Alder, Loredana Siani, Sarah C. Gilbert, Alfredo Nicosia, Stefano Colloca, Riccardo Cortese, and Adrian V. S. Hill The Jenner Institute, University of Oxford, Old Road Campus Research Building, Roosevelt Drive, Oxford OX3 7DQ, United Kingdom; Centre for Statistics in Medicine, Wolfson College Annexe, Linton Road, Oxford OX2 6UD, United Kingdom; and Okairòs, Via dei Castelli Romani 22, 00040 Pomezia, Rome, Italy Volume 78, no. 1, pages 145 153, 2010. Page 152, Acknowledgments, line 1 should read: This work was supported by the Wellcome Trust and the NIHR Oxford Biomedical Research Centre program. 2131