Gene Therapy (2004) 11, 448–456 & 2004 Nature Publishing Group All rights reserved 0969-7128/04 $25.00 www.nature.com/gt RESEARCH ARTICLE Reduced immunogenicity of DNA in mixtures

M Sedegah1,2, Y Charoenvit1, L Minh1, M Belmonte1, VF Majam1, S Abot1, H Ganeshan1, S Kumar1, DJ Bacon1, A Stowers3, DL Narum4, DJ Carucci1 and WO Rogers5 1Malaria Program, Naval Medical Research Center, Silver Spring, MD, USA; 2Department of Microbiology, The University of Maryland School of Medicine, Baltimore, MD, USA; 3Laboratory of Parasitic Diseases, NIAID, NIH, Bethesda, MD, USA; 4EntreMed, Inc., Rockville, MD, USA; and 5Naval Medical Research Unit 3, Ghana Detachment, c/o Department of State, Washington, DC, USA

We measured the ability of nine DNA vaccine plasmids from the mixture frequently reduced the observed suppres- encoding candidate vaccine to induce sion. Boosting with recombinant poxvirus increased the and interferon-g responses when delivered alone response in animals primed with either a single or in a mixture containing all nine plasmids. We further or the mixture, but, even after boosting, responses examined the possible immunosuppressive effect of indivi- were higher in animals primed with single plasmids than in dual plasmids, by assessing a series of mixtures in which those primed with the nine- mixture. Boosting did not each of the nine vaccine plasmids was replaced with a overcome the suppressive effect of mixing for IFN-g control plasmid. Given alone, each of the vaccine plasmids responses. Interactions between components in a multi- induced significant antibody titers and, in the four cases for plasmid DNA vaccine may limit the ability to use plasmid which appropriate assays were available, IFN-g responses. pools alone to induce responses against multiple targets Significant suppression or complete abrogation of responses simultaneously. were seen when the plasmids were pooled in a nine-plasmid (2004) 11, 448–456. doi:10.1038/sj.gt.3302139 cocktail and injected in a single site. Removal of single

Keywords: DNA vaccine; Plasmodium falciparum; multivalent vaccine

Introduction vaccine candidates from P. falciparum failed to show evidence of suppression of antibody responses to A major potential advantage of DNA is that it individual components, and indeed the response to one may be possible to mix several plasmids encoding target, the P. falciparum merozoite surface protein 1, was antigens from different stages or strains of a single enhanced in the mixture.5 In preliminary experiments pathogen, or from multiple pathogens, and to induce with P. falciparum DNA vaccines, we found more immune responses against the individual components. In complex effects. The antibody responses to each of five a number of cases, however, with mix- individual target antigens were decreased in the mixture; tures of protein or polysaccharide vaccines has led to however, the effects on cellular immune responses decreased responses to individual components of the depended both on the individual and on the mixture.1,2 It is not yet clear whether such effects will parameter measured. Cytotoxic T-cell responses to P. be important in DNA vaccines. In a mouse model of falciparum liver stage antigen 1 (PfLSA1), to PfCSP and malaria, the combination of DNA plasmids encoding two P. falciparum exported protein 1 (PfEXP1), and to P. different vaccine antigens, the Plasmodium yoelii circum- falciparum sporozoite surface protein 2 (PfSSP2) in- sporozoite protein (PyCSP) and P. yoelii hepatocyte creased, decreased, and were unchanged, respectively. erythrocyte protein-17 kDa (PyHEP17), overcame the Interferon-g ELISPOT responses to individual antigens genetic restriction of protection seen when either were either unchanged or decreased (Sedegah et al, plasmid was used alone.3 In a murine tuberculosis unpublished data). model, immunization with a mixture of four DNA We have now constructed a nine-plasmid DNA vaccine plasmids encoding different antigens induced vaccine mixture of plasmids encoding candidate vaccine comparable or modestly enhanced protection compared antigens from the sporozoite, intrahepatocytic, and to the most protective individual antigen plasmid, erythrocytic life cycle stages of P. falciparum, designated without evidence of antigenic competition.4 Immuniza- M9. We investigated whether the antibody or interferon- tion of mice with a mixture of four plasmids encoding g responses to each antigen induced by the mixture were suppressed relative to the response induced by immu- nization with single plasmids. We further attempted to Correspondence: Dr WO Rogers, Naval Medical Research Unit 3, Ghana, c/o Department of State, 2020 Accra Place, Washington, DC 20521-2020, define the role of individual vaccine plasmids in USA suppressing the responses to others, by assessing a Received 23 December 2002; accepted 07 July 2003 series of mixtures of eight plasmids from which each Low immunogenicity of DNA vaccine mixture M Sedegah et al 449 individual plasmid had been systematically excluded. of highly expressed mammalian genes, as previously Finally, we asked whether suppressive effects due to described.6 The resultant DNA sequences were synthe- mixing plasmids could be overcome by boosting with sized and cloned into the DNA vaccine plasmid VR1020,7 recombinant viral vaccine expressing the target antigen(s). designed to express cloned genes in fusion with the human tissue plasminogen activator signal sequence. The synthesis and characterization of PfMSP1 FVO and Materials and methods PfEBA-175 plasmids have been described;6 the other seven vaccine plasmids will be described separately. All Mice the plasmids used for injection were produced to GMP Female 4–6-week-old outbred CD-1 mice obtained from specifications (Vical, Inc., San Diego, CA, USA). Two Charles River Laboratories (Wilmington, MA, USA) were recombinant, attenuated viruses were used in used for the antibody response studies. Female 4–6- boosting experiments, NYVAC Pf78 and NYVAC PfCSP.9 week-old inbred BALB/cJ mice were purchased from NYVAC Pf7 expresses seven P. falciparum antigens (CSP, The Jackson Laboratory (Bar Harbor, ME, USA) and used SSP2, LSA1, SERA, MSP1, AMA1, and Pf25); NYVAC for the cellular response studies. PfCSP expresses only PfCSP. Immunogens We constructed DNA vaccine plasmids encoding nine malarial vaccine antigens. For each antigen, Table 1 The DNA plasmids were administered by intramuscular shows the plasmid designation, the stage in the injection using a 29-gauge needle and a 0.3 cm3 tubercu- Plasmodium life cycle in which the antigen is expressed, lin syringe. Animals were injected with either 50 mgof the domain of the antigen included in the vaccine antigen-encoding plasmid, 50 mg of antigen plasmid, and construct, and the in vitro expression of the antigen from 400 mg of the control plasmid VR1020, or 50 mg of each of the vaccine plasmid by transient of UM449 the nine antigen plasmids. In subsequent experiments, a melanoma cells. In each case, the amino-acid sequence of series of groups were immunized with nine-plasmid the desired domain was reverse transcribed using the mixtures from which individual antigen plasmids had most frequently used codons for each in a set been removed and replaced by VR1020. In one group,

Table 1 DNA vaccine constructs

Vaccine antigen Plasmid designation Stage of expression Antigenic domain Expressiona

Circumsporozoite protein VCL-2571 Sporozoite and hepatocyte Full length (PfCSP)

Sporozoite surface protein VCL-2576 Sporozoite and hepatocyte Full length (PfSSP2)

Liver-stage antigen 1 VCL-2559 Hepatocyte NH2-terminus, two repeat (PfLSA1) units, and C-terminus Exported protein 1 VCL-2565 Hepatocyte and erythrocyte Full length (PfExp1) Liver-stage antigen 3 VCL-2573 Hepatocyte and erythrocyte Full length (PfLSA3)

Merozoite surface protein VCL-2574 Hepatocyte and erythrocyte C-terminal 42 kDa

142 3D7 (PfMSP142 3D7) fragment

Merozoite surface protein VCL-2575 Hepatocyte and erythrocyte C-terminal 42 kDa

142 FVO (PfMSP142 FVO) fragment

Apical merozoite antigen VCL-2577 Hepatocyte and erythrocyte Ectodomain (PfAMA1)

Erythrocyte-binding VCL-2568 Hepatocyte and erythrocyte Region II antigen-175 RII (PfEBA- 175) aExpression of antigen from the vaccine plasmids was assessed by transient transfection of UM449 melanoma cells, as described previously.6 The individual bands corresponding to each antigen are shown. Expression was detected with the following antibodies and sera in Western blots: PfCSP, mouse mAb NFS1; PfSSP2, mouse mAb PfSSP2.1; PfExp1, mouse mAb N1; PfLSA1, polyclonal mouse serum; PfLSA3, polyclonal mouse serum; PfMSP142 3D7 and FVO, mouse mAb 5.2, a pan-specific conformational monoclonal antibody; PfAMA1, rat mAb 4G2, a conformational monoclonal antibody; and PfEBA-175 mouse mAb R 217.

Gene Therapy Low immunogenicity of DNA vaccine mixture M Sedegah et al 450 two plasmids that code for two alleles of the same or )À(% lysis with targets infected with wild-type antigen (PfMSP1 3D7 and PfMSP1 FVO) were replaced at vaccinia virus or control peptide). the same time with 100 mg of VR1020. The total dose was Spontaneous release was obtained as cpm of targets in always administered in a volume of 100 ml PBS (pH 7.4), the presence of medium, and maximum cpm was split such that 50 ml was administered in the right tibialis obtained as the cpm of targets in the presence of 5% anterior muscle and 50 ml in the left tibialis anterior Triton X-100. In some experiments, a portion of the muscle. A total of two immunizations were administered effector cells were depleted of CD8 þ T cell or CD4 þ T at 3-week intervals. In one experiment, a reduced dose of cells and retested against positive test targets. Anti- 10 mg of each plasmid was used. In some experiments, CD4 þ - or anti CD8 þ -coated Dynabeads (Dynal Inc., mice were boosted intraperitoneally with 1 Â 107 Great Neck, NY, USA) were used to deplete the effector PFU recombinant virus in a total volume of 0.2 ml in cells of CD4 þ or CD8 þ T cells according to the PBS, pH 7.2, 6 weeks after the second dose of DNA. manufacturer’s instructions. To test for genetic restriction Immune responses were assayed at 2 weeks after each of the CTL response in assays, mismatched target cells immunization. were made with EL4 cells.

Antibody assays Eight P. falciparum recombinant proteins were used for Results EIA. The expression, purification, and characterization of recombinant PfCSP, PfMSP1 FVO, and 3D7, and PfE- Effect of mixing BA175 region 2 have been described.10,11 Expression and We first asked whether the antibody response to the characterization of recombinant PfSSP2, PfLSA1 PfEXP1, individual components of M9 was reduced when they and PfAMA1 will be published separately. were included in the nine-gene mixture. We compared The optimal concentrations (0.5–4.0 mg/ml) of recom- the response to 50 mg doses of each single plasmid to the binant proteins were first determined by standard EIA response when 50 mg of each plasmid was pooled into and used in subsequent experiment as solid-phase M9. Table 2 shows the effect of mixing the M9 plasmids antigens. EIA was performed as previously described.12 on the geometric mean EIA response to individual The results are recorded as OD 0.5 U, the reciprocal of antigens. Depending on the individual antigen, EIA the serum dilution at which the mean of quadruplicate responses in the M9 group were reduced to 0.04–6.2% of OD readings was 0.5. Immunofluorescence assays were the response seen when the individual antigen plasmid carried out using air-dried P. falciparum sporozoites and was used alone. All reductions were statistically sig- air-dried parasitized erythrocytes.13,14 nificant (Po0.05 by Kruskal–Wallis ANOVA on ranks followed by Student—Newman–Keuls pairwise compari- IFN-g ELISPOT son on log-transformed data). Detection of antigen-specific IFN-g-producing cells was Figure 1 shows the effect of mixing on IFA titers performed as previously described,15 with minor modi- against sporozoites (Figure 1a) and infected erythrocytes fications. cells from immunized mice were (Figure 1b). The geometric mean antisporozoite titer incubated with previously identified H-2d restricted T- induced by immunization with the PfCSP plasmid alone cell from antigens of interest, PfCSP(7G8) 39–47 was reduced approximately 50-fold when the plasmid NYDNAGTNL, PfEXP1(3D7) 66–80 EVNKRKSKYK- was delivered as a component of M9 (Po0.05 KW LATSV, and PfLSA1(3D7) 1671–1679 YYIPHQSSL. In ANOVA, SNK, log-transformed). Anti-infected erythro- the case of PfSSP2, spleen cells were incubated with cyte titers induced by the PfEBA-175 plasmid alone were irradiated P815 cells that had been infected with a reduced eight-fold in the M9 mixture (Po0.05, KW recombinant poxvirus expressing PfSSP2,9 or with wild- ANOVA, SNK, log-transformed), in spite of the fact that type virus at a multiplicity of of 10, 18 h earlier. five of the other eight plasmid components of M9 are capable of inducing anti-infected erythrocyte antibodies Cytotoxic T-cell assay when delivered as single plasmids (data not shown). Standard 51Cr-release CTL assays were carried out as Figure 2 shows the results of IFN-g ELISPOT assays for previously described.15 Briefly, the effector cells were responses to four pre-erythrocytic stage antigens, PfCSP, stimulated by coculture with recombinant ALVAC- PfSSP2, PfEXP1, and PfLSA1. Immunization with PfCSP infected, irradiated P815 cells (for PfSSP2) or 10 mg/ml and PfLSA1 as single plasmids induced mean IFN-g peptide (for PfCSP, PfEXP1, and PfLSA1, as described responses of 60 and 250 SFC/million splenocytes; neither above) for 7 days. Targets were recombinant WR response was detectable when the plasmids were vaccinia-infected (PfSSP2) or peptide-pulsed (PfCSP, combined in the M9 mixture (Figure 2a,b). In contrast, PfEXP1, and PfLSA1), 51Cr-loaded P815 cells. Standard SSP2 IFN-g responses were only slightly decreased in the 51Cr release assay was carried out. The effector cells M9 mixture (Figure 2c), and responses to PfExp1 were were washed, counted, and suspended in complete barely detectable in response to either the individual medium at 2 Â 106/ml, and tested with 5000 target cells plasmid or to M9 (Figure 2d). Similar results were in a 96-well U-bottom plate. A 5-h standard chromium observed in CTL assays for responses to PfCSP, PfSSP2, release method was followed and per cent-specific lysis PfEXP1, and PfLSA1 (data not shown). calculated. Assays were carried out in triplicate and per cent lysis Effect of total DNA dose was defined as ((experimental cpmÀspontaneous cpm)/ Next, we attempted to determine whether the suppres- (maximum cpmÀspontaneous cpm)) Â 100%. Per cent- sion of responses in M9 was the result of antigenic specific lysis was defined as (% lysis with targets infected competition among multiple antigens or a simpler effect with recombinant virus expressing P. falciparum proteins of total DNA dose. Therefore, we compared the

Gene Therapy Table 2 Antigen-specific antibody responses in immunized CD-1 mice

Plasmid immunogen

EIA Ag Aga alone Ag/1020b M9c M9-CSPd M9-SSP2 M9-EXP1 M9-LSA1 M9-LSA3 M9-MSP1 3D7 M9-MSP1 FVO M9-MSP1 3D7 M9-EBA175 M9-AMA1 and MSP1FVO

CSP 13 400 5e 2.3 0.8 27 1.2 3 1.2 5 8.4 11 3.2 (0.80–32) (0.30–16) (0.1–5.2) (6–120) (0.20–7.9) (0.60–15) (0.15–7.9) (1.4–18) (2.9–25) (2.5–45) (0.60–19) SSP2 1530 23 0.5 0.17 2.8 0.67 0.84 0.64 1.1 2 2 0.9 (2.4–220) (0.1–1.9) (0.05–0.50) (1.0–7.8) (0.20–2.3) (0.37–2.6) (0.15–2.7) (0.36–3.2) (0.58–6.6) (0.63–6.2) (0.25–3.1) EXP1 19 30 13 11 8.2 12 12 8.4 24 8 28 17 (4.2–214) (2.4–77) (2.0–55) (1.6–41) (2.1–71) (2.3–61) (1.7–41) (3.7–157) (1.7–38) (4.1–184) (3.2–94) LSA1 16 500 0.22 0.04 0.03 0.03 0.06 0.06 0.03 0.04 0.02 0.03 0.06 (0.03–1.7) (0.01–0.12) (0.01–0.10) (0.009–0.09) (0.02–0.20) (0.02–0.14) (0.01–0.07) (0.01–0.09) (0.007–0.05) (0.01–0.10) (0.02–0.17)

MSP1 3D7 2030 382 3.7 17 3 7.4 5.8 9.2 26 7.8 1.5 mixture Sedegah vaccine M DNA of immunogenicity Low (59–2500) (0.26–55) (1.4–210) (0.15–61) (0.58–96) (0.48–71) (0.64–130) (1.8–394) (0.84–72) (0.11–20) MSP1 FVO 870 27 6.2 21 3.3 10 6.2 10 0.17 9.4 4

(1.4–560) (0.48–80) (1.9–250) (0.15–71) (0.74–137) (0.52–75) (0.54–180) (0.03–0.84) (0.94–93) (0.33–47) al et EBA175 39 800 75 3.9 11 1.5 23 3.2 5.7 8 4.1 4.7 4 (29–194) (0.48–32) (4.7–30) (0.23–10) (4.4–124) (0.56–18) (1.0–31) (1.7–39) (0.85–20) (0.87–25) (0.6–27) AMA1 38 ,400 52 1.9 0.95 0.5 21 0.44 0.74 0.64 0.94 0.93 0.96 (30–89) (0.36–11) (0.13–7.1) (0.06–4.2) (9.5–45) (0.09–2.3) (0.13–4.2) (0.12–3.5) (0.14–6.4) (0.08–11) (0.18–5.2)

a Response to immunization with the single plasmid (50 mg) encoding the antigen assayed, geometric mean titer (OD¼0.5). b Immunization with the antigen plasmid (50 mg) and control plasmid (400 mg). c Immunization with M9. d Immunization with M9 from which the indicated antigen plasmid(s) was removed and replaced with control plasmid. e Geometric mean EIA response, 16 CD-1 mice per group, as a percentage of the geometric mean response to immunization with single plasmid (95% confidence interval for proportion). eeTherapy Gene 451 Low immunogenicity of DNA vaccine mixture M Sedegah et al 452

Figure 1 Groups of 16 CD-1 mice were immunized with either the PfCSP plasmid (50 mg), the PfCSP plasmid (50 mg) þ the control plasmid VR1020 (400 mg), M9, or plasmid mixtures in which the indicated plasmids were removed and replaced with 50 mg of VR1020, as described in Materials and methods. Sera were collected from the titer determined against P. falciparum sporozoites (a) or infected erythrocytes (b). Individual mouse titers (solid dot) and geometric means (hatched bar) are shown.

responses in groups of CD-1 mice immunized with either 50 mg of a single antigen plasmid or 50 mg of the antigen plasmid and 400 mg of the control expression plasmid VR1020. As shown in Table 2, antibody responses to all but one individual antigen (PfMSP1, 3D7) induced by immunization with 50 mg of a single plasmid were substantially reduced when 50 mg of the antigen plasmid

"

Figure 2 Groups of Balb/c mice were immunized with either individual antigen plasmids ((a) PfCSP; (b) PfSSP2; (c) PfLSA1; (d) PfEXP1) or individual antigen plasmid and VR1020, or the complete M9 mixture, and the number of antigen-specific IFN-g-secreting cells was determined, as described in Materials and methods. The means of triplicate well IFN-g SFC counts from individual mice (solid dot) and group means (hatched bar) are shown.

Gene Therapy Low immunogenicity of DNA vaccine mixture M Sedegah et al 453 was coinjected with 400 mg of VR1020; the reduction, however, was in each case less than that seen when the antigen plasmid was delivered in the M9 mixture. Figure 1 shows that, similarly, adding VR1020 to either the PfCSP (Figure 1a) or PfEBA175 (Figure 1b) plasmids reduced the IFA response to sporozoites or infected erythrocytes, but reduced it to a lesser extent than did immunization with M9. The mean IFN-g responses to PfCSP and PfLSA1 induced by the single plasmid immunizations were only slightly reduced by the addi- tion of VR1020 (Figure 2a, b), and the reductions were not statistically significant. To further assess whether the observed interference was primarily due to physical effects of high DNA dose, we carried out an experiment in which the dose of each individual plasmid and the overall DNA concentration in the injectate was reduced five-fold. The antibody and ELISPOT responses to the PfCSP were significantly diminished in the mice immu- nized with M9, compared with responses seen in those immunized with the PfCSP plasmid alone (Figure 3c).

Effect of individual plasmids in M9 Since it did not appear that effects based on the total amount of DNA injected accounted for all of the reduction in response seen in M9, we next attempted to determine whether there were individual plasmids within the M9 mixture whose presence reduced res- ponses to the remaining plasmids. To do so, we compared the response to M9 to the response to a systematic series of plasmid pools from which each one of the antigen plasmids had been removed and replaced by VR1020 in turn. Table 2 shows the effect of removing individual plasmids from M9 on antibody responses to the individual antigens. In the case of responses to PfCSP, PfSSP2, PfAMA1, and PfEBA175, replacement of the PfEXP1 plasmid with VR1020 led to statistically signifi- cant increases in titer compared with the response induced by M9 (Po0.05, KW ANOVA, SNK); although the increases were not statistically significant, exclusion of the PfEXP1 plasmid from M9 improved antibody responses to all the other measured antigens. Similarly, exclusion of the plasmids encoding PfEBA175, PfMSP1 (FVO), or the combination of PfMSP1(FVO) and PfMSP1(3D7), tended to increase titers to the majority of measured antigens, although the increases were statistically significant only in the case of anti-SSP2 responses (Po0.05, KW ANOVA, SNK). In contrast, removal of the PfSSP2 plasmid never increased the Figure 3 Balb/c mice were injected either with M9 in the tibialis anterior response to other antigens. Similarly, exclusion of the muscle of one leg (same site), or with PfCSP in one leg and the remaining PfExp1 plasmid from M9 increased the titer of antispor- components of M9 in the other leg, 50 mg/plasmid. (a) Antibody titers against air-dried sporozoites (IFA) or recombinant PfCSP (EIA) from ozoite and anti-infected erythrocyte antibodies (Figure 1). individual mice (solid dot) and group geometric means (hatched bar). (b) PfCSP-specific IFN-g SFC from means of triplicate wells for individual Effect of injection at different sites mice (solid dot) and group means (hatched bar). (c) Anti-CSP EIA titer In order to understand the mechanism of the suppres- and PfCSP-specific IFN-g SFC from Balb/c mice immunized with PfCSP sion, we asked whether the response to an individual in one leg (CSP alone), or with PfCSP in one leg and the remaining antigen would be reduced by simultaneous injection of components of M9 in the opposite leg (separate sites), or with M9 in m the remaining components of M9 at a distant site. As a single site (same site), 10 g/plasmid. shown in Figure 3a, IFA titers against sporozoites and EIA titers against PfCSP were lower in mice in which the separate site (P¼0.0001, two-tailed t-test). Similar results PfCSP plasmid was injected as a component of M9, than were found when the same experiment was conducted when it was injected at a separate site (IFA P¼0.17, EIA using a five-fold lower dose of each plasmid (Figure 3c). P¼0.032, two-tailed t-test, log-transformed data). IFN-g responses (Figure 3b) were not detectable in mice in Effect of viral boosting which the M9 mixture was injected at a single site, but We next asked whether the reduction in antibody and were robust when the PfCSP plasmid was injected at a IFN-g responses could be overcome by boosting with

Gene Therapy Low immunogenicity of DNA vaccine mixture M Sedegah et al 454 recombinant poxvirus. To assess antibody responses, we in the groups primed with the PfCSP plasmid alone were primed groups of eight CD-1 mice with two doses of 3–4-fold higher than those in the group primed with M9. DNA, either the PfCSP plasmid alone (50 mg), the PfCSP There were no significant differences between the titers plasmid (50 mg) and VR1020 (400 mg), or M9 (450 mg produced by boosting with NYVAC-PfCSP compared total), and boosted with 1 Â 107 PFU of either NYVAC- with NYVAC-Pf7. PfCSP or NYVAC-Pf7.9 Figure 4a shows the anti-CSP EIA To assess IFN-g responses, we primed groups of eight titers before and after viral boosting. Before boosting, BALB/c mice with two doses of DNA, as above, and there was a statistically significant, approximately 50- then boosted with 107 PFU of NYVAC-PfCSP or NYVAC- fold reduction in geometric mean anti-PfCSP titer Pf7. Figure 4b shows the results of ELISPOT assays for between mice immunized with M9 and those immunized PfCSP peptide-specific IFN-g responses. Immunization with the PfCSP plasmid alone (Po0.001, two-tailed t-test, with two doses of PfCSP plasmid alone induced mean log-transformed data). Following boosting with either frequencies of 60–180 IFN-g SFC/106 splenocytes, while NYVAC-PfCSP or NYVAC-Pf7, geometric mean titers immunization with M9 failed to induce measurable IFN- increased in all groups. Although the differences were g responses. When mice were primed with two doses of not statistically significant, the geometric mean titers either PfCSP plasmid or M9 and boosted with either NYVAC-PfCSP or NYVAC-Pf7, the mice that had first received PfCSP plasmid alone produced mean frequen- cies of 200–280 IFN-g SFC/106 splenocytes, while those primed with M9 produced mean frequencies no greater than 25 IFN-g SFC/106 splenocytes.

Discussion It has been frequently suggested that DNA vaccine plasmids might be combined to produce cocktail vaccines directed against multiple pathogens or multiple strains or antigenic targets of a single pathogen.16 We evaluated a complex, multistage P. falciparum DNA vaccine composed of nine plasmids encoding candidate vaccine antigens from the sporozoite, exoerythrocytic, and erythrocytic stages of the parasite, designated M9. We found that the M9 plasmid mixture induced dramatically reduced immune responses to the compo- nent antigens compared with the responses to individual plasmids given singly. This reduction in responses against the components of a mixture may represent a serious challenge to the development of multicomponent DNA vaccines. Antibody responses to the individual targets within M9 were reduced by 8 to 2500-fold (Table 2) compared to the responses induced by individual plasmids given singly. Similarly, the moderate to very high IFN-g responses specific for PfCSP and PfLSA1 induced by the corresponding single plasmids were undetectable in mice immunized with M9 (Figure 2a, b). CTL responses to PfCSP and PfLSA1 were also greatly reduced or eliminated (data not shown). It is interesting that neither IFN-g nor CTL responses to PfExp1 and SSP2 were reduced in M9 (Figure 2c, d; data not shown). Although it was conceivable that the higher total DNA concentra- tion that had to be injected in the M9 mixture might have been responsible for reduced immunogenicity, we were unable to reproduce the negative effect of mixing the nine plasmids (50 mg/plasmid) by combining a single antigen plasmid (50 mg) with an empty vector plasmid, Figure 4 (a) CD-1 mice were immunized with the indicated DNA VR1020, (400 mg) to obtain identical total DNA concen- plasmids or plasmid pools, and then boosted with the indicated trations. Although the addition of 400 mg of VR1020 recombinant poxviruses, as described in Materials and methods. The reduced many antigen-specific immune responses com- individual (solid dot) and geometric mean (hatched bar) anti-PfCSP EIA pared to the responses seen with single plasmids, the titers are shown. (b) Balb/c mice were immunized with either the PfCSP reductions were smaller than those seen in the M9 plasmid alone or M9, or were first immunized with the indicated plasmids and then boosted with recombinant poxvirus, as described in Materials and mixture. In addition, similar reductions in immunogeni- methods. The means of triplicate well IFN-g SFC counts from individual city were seen even when all plasmids were injected at mice (solid dot) and group means (hatched bar) are shown. five-fold lower dose and concentration (Figure 3c).

Gene Therapy Low immunogenicity of DNA vaccine mixture M Sedegah et al 455 This suppressive effect of mixing DNA vaccines has occur at several of these levels simultaneously. Pre- not been consistently observed in the literature. In a liminary results from in vitro transfection experiments previous study,17 we immunized rhesus monkeys either suggest that generally similar effects of plasmid mixing with a mixture of plasmids encoding PfCSP, PfSSP2, occur at the level of both mRNA and protein expression PfExp1, and PfLSA1, with the combinations of PfCSP (unpublished data). The finding that injecting a single and PfSSP2 or PfLSA1 and PfExp1 delivered as single plasmid at a separate site from the remainder of the plasmids at separate sites, or with PfSSP2 alone. No clear mixture avoided the suppressive effect of mixing suppressive effect of mixing was seen on CTL responses suggests that suppression is not an effect resulting from for the plasmids contained in the mixture relative to the immune competition at a systemic level. It is possible, other groups. Geometric mean antibody responses to a but remains to be shown, that the observed competition PfSSP2 peptide were reduced approximately 20-fold in requires that multiple plasmids be delivered to a single the four-plasmid mixture compared to the responses cell. It might then be possible to devise delivery systems elicited by the PfSSP2 plasmid alone. A recent study and formulations that minimize the chance of delivering using a mixture of three plasmids encoding P. falciparum many different plasmids to a single cell, even within a blood-stage antigens, PfMSP1, PfEBA-175, and PfAMA1, single injection site. A thorough understanding of the found no reduction in immunogenicity when the mechanisms responsible for the observed suppressive plasmids were combined in a single injection.18 A recent effect of plasmid mixing will be critical for the design of experiment using a mixture of four plasmids encoding future multicomponent DNA vaccines. Mycobacterium tuberculosis proteins in mice found no reduction in responses to the individual antigens as a result of mixing,4 and the best protection was observed in the group that received the four-plasmid mixture. No Acknowledgements direct comparison of the response to individual plasmids This work was supported by the Naval Medical Research given alone or in a mixture has been reported in humans. Center (Military Infectious Diseases Research Program) However, preliminary results of immunization of volun- work units 61102A.S13.F.A0009 and 62787A.870.F.A0010, teers with a mixture of five plasmids encoding and by the Office of Naval Research Grant N00014-89-J- P. falciparum antigens including PfCSP suggest that IFNg 1856 to The University of Maryland at Baltimore (MS). responses to PfCSP were lower than those seen in earlier The assertions herein are the private ones of the authors trials in which the PfCSP plasmid was given alone and are not to be construed as official or as reflecting the (unpublished data). views of the US Navy or the Naval service at large. In It is not yet clear how great an obstacle the suppressive conducting the research described in this report, all effects of mixing plasmids may be to the development of aspects involving animal use were conducted in an multicomponent DNA vaccines. First, it is possible that AAALAC accredited facility, and in accordance with the although responses to individual components are re- Animal Welfare Act implementing the instructions (9 duced, the total to the pathogen is CFR, Subchapter A, Parts 1–3), Department of Defense usefully increased. However, in the current experiments, regulations, and recognized standards relating to the the total antibody response to infected erythrocytes care and use of laboratory animals. We thank David induced by M9 was lower than that induced by the Lanar (Walter Reed Army Institute for Research) for PfEBA175 plasmid delivered alone (Figure 1b), suggest- recombinant PfCSP, PfSSP2, and PfLSA1; Jennifer Meeks ing that the suppression of individual responses may be and Autumn Ramirez (Vical, Inc.) for assistance greater than can be compensated by the summation of in preparing plasmid mixtures; R Ridley (Hoffman- responses to several targets. Second, recent experiments La-Roche) for anti-Exp1 mAb N1, Tony Holder (NIMR, have shown marked improvements in immunogenicity Mill Hill) for anti-MSP1 mAb 5.2, and Alan Thomas for and efficacy when DNA vaccines are used as part of anti-AMA-1 mAb 4G2; and HM1 Robert Arcilla, HM1 a heterologous prime-boost strategy.19,20,15 It Arnel Belmonte, HM2 Thomas Smalls, and Patricia de la was possible that boosting with recombinant virus Vega for excellent technical assistance and providing would overcome whatever suppressive effect of mixing P. falciparum parasites. was observed with DNA vaccines alone. However, in the current experiments, boosting with either of two recom- binant vaccinia expressing PfCSP only partially over- came the suppressive effect of mixing on antibody References responses to PfCSP (Figure 4a), and had no effect on the suppression of PfCSP-specific IFN-g responses in M9 1 Fattom A et al. Epitopic overload at the site of injection may (Figure 4b). It therefore appears that at least for this set result in suppression of the immune response to combined 17 of nine plasmids, as currently delivered, a suppressive capsular polysaccharide conjugate vaccines. Vaccine 1999; : 126–133. effect of mixing represents a significant problem for 2 Hunt JD et al. Antigenic competition in a multivalent foot rot development of a multicomponent DNA vaccine for vaccine. Vaccine 1994; 12: 457–464. malaria. 3 Doolan DL et al. Circumventing genetic restriction of protection Several mechanisms might explain the observed against malaria with multi-gene DNA immunization: CD8+ T suppressive effects of mixing in M9. Multiple plasmids cell, interferon-gamma, and nitric oxide dependent immunity. might compete for uptake by host cells; plasmids taken J Exp Med 1996; 183: 1739–1746. up by a single cell might compete for 4 Morris S et al. The immunogenicity of single and combination factors or for the apparatus; multiple proteins DNA vaccines against tuberculosis. Vaccine 2000; 18: 2155–2163. and derived from the plasmids might compete 5 Grifantini R et al. Multi-plasmid DNA vaccination avoids for antigen presentation pathways; or competition might antigenic competition and enhances immunogenicity of a

Gene Therapy Low immunogenicity of DNA vaccine mixture M Sedegah et al 456 poorly immunogenic plasmid. Eur J Immunol 1998; 28: 13 Szarfman A et al. Mature liver stages of cloned Plasmodium 1225–1232. falciparum share epitopes with proteins from sporozoites and 6 Narum DL et al. Codon optimization of gene fragments asexual blood stages. Parasite Immunol 1988; 10: 339–351. encoding Plasmodium falciparum merzoite proteins enhances 14 Charoenvit Y et al. Characterization of Plasmodium yoelii DNA vaccine protein expression and immunogenicity in mice. monoclonal antibodies directed against stage-specific Infect Immun 2001; 69: 7250–7253. sporozoite antigens. Infect Immun 1987; 55: 604–608. 7 Luke CJ et al. An OspA-based DNA vaccine protects mice 15 Sedegah M et al. Improving protective immunity induced by against infection with Borrelia burgdorferi. J Infect Dis 1997; 175: DNA-based immunization: priming with antigen and GM-CSF- 91–97. encoding plasmid DNA and boosting with antigen-expressing 8 Stoute JA et al. Long-term efficacy and immune responses recombinant poxvirus. J Immunol 2000; 164: 5905–5912. following immunization with the RTS,S . J Infect 16 Doolan DL, Hoffman SL. Multi-gene vaccination against Dis 1998; 178: 1139–1144. malaria: a multistage, multi-immune response approach. 9 Tine JA et al. NYVAC-Pf7: a poxvirus-vectored, multiantigen, Parasitol Today 1997; 13: 171–178. multistage vaccine candidate for Plasmodium falciparum malaria. 17 Wang R et al. Simultaneous induction of multiple antigen- Infect Immun 1996; 64: 3833–3844. specific cytotoxic T lymphocytes in nonhuman by 10 Kaslow DC, Hui G, Kumar S. Expression and antigenicity of immunization with a mixture of four Plasmodium falciparum Plasmodium falciparum major merozoite surface protein DNA plasmids. Infect Immun 1998; 66: 4193–4202. (MSP1(19)) variants secreted from Saccharomyces cerevisiae. Mol 18 Jones TR et al. Absence of antigenic competition in Biochem Parasitol 1994; 63: 283–289. Aotus monkeys immunized with Plasmodium falciparum 11 Liang H et al. A recombinant baculovirus-expressed Plasmodium DNA vaccines delivered as a mixture. Vaccine 2-22-2002; 20: falciparum receptor-binding domain of erythrocyte binding 1675–1680. protein EBA-175 biologically mimics native protein. Infect 19 Schneider J et al. Enhanced immunogenicity for CD8+ T cell Immun 2000; 68: 3564–3568. induction and complete protective efficacy of malaria DNA 12 Charoenvit Y et al. CD4(+) T-cell- and gamma interferon- vaccination by boosting with modified vaccinia virus Ankara. dependent protection against murine malaria by immunization Nat Med 1998; 4: 397–402. with linear synthetic peptides from a Plasmodium yoelii 17- 20 Sedegah M et al. Boosting with recombinant vaccinia increases kilodalton hepatocyte erythrocyte protein. Infect Immun 1999; 67: immunogenicity and protective efficacy of a malaria DNA 5604–5614. vaccine. Proc Natl Acad Sci USA 1998; 95: 7648–7653.

Gene Therapy