The identification of a vaccine approach that successfully targets liver stage malaria is a critical step in the development of a highly protective vaccine. The number of hepatic parasites resulting from an infectious mosquito bite is relatively small, probably no more than 100,42
and development in the liver is non-pathogenic. Their elimination prior to transformation into the highly pathogenic blood stages would entirely prevent illness. Moreover, the roughly five-day span of liver stage development in P. falciparum
appears long enough for cytotoxic or IFN-γ-secreting effector cells to work, as demonstrated by the induction of sterile immunity to P. yoelii
in BALB/c mice13,14
by immunization with Py
where the duration of liver stage development is just over two days13,44
or by the induction of sterile immunity to P. knowlesi
in non-human primates (NHP) where the duration of liver stage development is just over five days and thus similar to P. falciparum
In both the mouse14
models, protection induced with gene-based vaccines was dependent on CD8+ T cells that eliminated infected hepatocytes by direct cytotoxicity or through an IFN-γ-mediated mechanism. Thus the development of protective gene-based platforms protecting humans against P. falciparum
should be feasible.
With this trial, we took the first steps to clinically develop the multi-antigen approach, combining the DNA encoding five pre-erythrocytic stage antigens into a plasmid mixture (MuStDO5) and assessing intramuscular administration via jet injection in 32 (dropping to 31) healthy, malaria-naïve adults, aiming to induce cell-mediated protection. The rationale for multiple antigens was based on the genetic restriction of adaptive cell-mediated immunity: since each malaria antigen provides only limited numbers of strongly-binding class I epitopes for the diverse MHC antigens of the human population, the combination of multiple protective malaria proteins is likely required to achieve broad population coverage.48
In prior non-clinical studies, we have shown that multivalent DNA vaccines comprising plasmid mixtures can be immunogenic in mice and NHP without significant interference,16,45,49,73
although plasmids encoding certain malaria proteins could theoretically suppress the immunogenicity of other plasmids if included in the mixture,50
as shown in vitro.51
The first objective of the clinical trial was to explore safety and tolerability. The trial demonstrated that the MuStDO5 vaccine was both safe and well tolerated and caused no severe or serious side effects. Adverse events were mostly mild, mostly local to the injection site, and rapidly resolving. Injection of DNA plasmids did not induce anti-ds-DNA antibody responses or anti-hGM-CSF responses, allaying concerns that DNA immunization might lead to auto-immunity.52,53
Several immunized volunteers developed acute elevations of the muscle enzyme CPK, but these events showed no consistent timing with respect to immunizations, occurred with equal frequency in unimmunized control volunteers, and, in each case, followed the initiation of military style physical training. Moreover, the degree of CPK elevation was consistent with that previously reported for civilians or military service members engaging in intense exercise programs,54,55
so it did not appear that the vaccine had rendered the study subjects more sensitive to exercise-induced muscle damage. None of the individuals with elevated CPK in our study had associated changes in blood creatinine levels (data not shown), indicating that there was no renal insult. Overall, MuStDO5 administration was safe, consistent with findings from other DNA vaccine trials in malaria,29
The second objective was to assess immunogenicity. The MuStDO5 vaccine successfully induced IFN-γ responses to each of the five antigens; positive responses were detected to 38.2% (13/34) of the class I-restricted peptides and to 94.1% (16/17) of the class II-restricted peptides tested.31
About half the volunteers responded to class I or class II peptides from at least one antigen, although only 6/31 (19%) demonstrated class I responses. The responses were boosted by exposure to malaria challenge, a potentially useful attribute for any vaccine used in endemic areas where frequent exposure to the parasite could boost the protection induced by such a vaccine. The results also indicated that responses to CSP were not inhibited by inclusion in the five-plasmid cocktail when historically compared with the CSP plasmid administered alone in earlier studies.31
However, cellular responses were generally of low magnitude, ranging from 12 to 96 net spot forming cells (sfcs)/106
PBMCs (geometric mean, 23.4), and likely much higher responses are needed to protect. This could be achieved by a number of improvements, such as codon-optimization to enhance expression in mammalian cells,58
delivery via electroporation to improve cellular uptake or use of an adjuvant superior to GM-CSF.59
Antibody responses were also measured in this trial. As has been found with some other DNA vaccines including our earlier clinical studies with the monovalent PfCSP-encoding plasmid, these responses were negligible, with no significant pre-/post-immunization differences recorded. Like cellular responses, antibody responses, particularly against PfCSP, were boosted on challenge indicating a potential benefit of DNA vaccination for residents of endemic areas where frequent malaria exposure could boost protective responses and accelerate the development of naturally acquired immunity. However, non-immunized controls were equally boosted by exposure to CHMI. As with cellular responses, the low antibody titers indicated the need for improved DNA vaccine technologies such as the use of superior adjuvants, or the use of improved regimens such as DNA priming followed by viral vector boosting.
The third objective was to assess protection. The challenge results showed no protection, neither sterile protection nor significant delay in the onset of parasitemia. A post-hoc analysis of volunteers with A02 and/or A03 class I supertype alleles in cohorts 1–4 showed the same negative results.
A fourth objective was to determine if immune responses and protection could be enhanced by co-administration of plasmid expressing a human cytokine. A sixth plasmid encoding hGM-CSF was added to MuStDO5 using a dose-escalation design, based on murine studies demonstrating a doubling of protection from 28% to 58% when a DNA plasmid encoding murine GM-CSF was added to a protective DNA plasmid encoding P. yoelii
In the murine model, GM-CSF DNA induced a 30-fold increase in antigen-specific antibodies, a 5-fold increase in antigen-specific IFN-γ spot forming cells and a significant (p < 0.05) increase in protection using a DNA/poxvirus prime/boost regimen.60
Most prior experience with GM-CSF has been with the recombinant protein, approved for use in stem cell and bone marrow transplant patients to reconstitute the myeloid series (sargramostim) or as part of an immunotherapy regimen for prostate cancer (sipuleucel-T). Recombinant GM-CSF enhances phagocytosis of P. falciparum
blood stage parasites by human neutrophils,61
has been shown to protect against murine malaria in combination with the synthetic peptide met-enkephalin62
and has been used to enhance immunogenicity of plasmid malaria vaccines in chimpanzees.63
The DNA encoding GM-CSF has previously been used as an adjuvant to recruit dendritic cells to the site of an immune response64
and to enhance the immunogenicity and protection in murine and NHP malaria models including in neonatal mice.36,60,65-71
GM-CSF DNA has also been shown to enhance the avidity of elicited immunoglobulin G for SIV envelope glycoproteins and to enhance antibody-dependent cellular cytotoxicity in association with significant improved efficacy against SIV challenge.72
In other models however, GM-CSF DNA has been reported to both augment and suppress cellular or humoral responses, according to the antigen, mouse strain and whether or not it is combined with other cytokines.73-76
Human studies of GM-CSF DNA have been restricted to cancer patients, where hGM-CSF-encoding plasmid has been well-tolerated77,78
and may have shown some benefit.79
This is the first study where it was used in healthy humans to enhance a prophylactic vaccine against an infectious agent.
The finding that GM-CSF offered no apparent benefit, and indeed may have diminished immunogenicity for class I responses following immunization and following boosting on exposure to malaria challenge,31
was unexpected, contrasting with our findings in animal malaria models36,80
as well as those of Li et al.70
One potential explanation could be non-optimal expression levels. When we studied the injection of murine GM-CSF DNA in mice, it was not possible to detect systemic levels of the protein.39
While a likely advantage with regard to tolerability, this may indicate that at the site of induction for the immune response, which could be at the injection site, regional lymph nodes or more distant splenic or hepatic sites, GM-CSF levels were too low or transient to impact the induction of the immune response. On the other hand, the trend toward a delay in the onset of parasitemia in cohort 2 that was not accentuated with higher hGM-CSF doses in cohorts 3 and 4 raises that possibility that tissue levels of hGM-CSF could also have been too high, and that lower doses might have been more effective.
As second explanation for our results could be that GM-CSF was expressed either too early or too late relative to the malaria antigens and that staggered administration would have been more effective (Hartikka J., unpublished data). For example, GM-CSF might trigger the early maturation of dendritic cells, without allowing the opportunity for appropriate antigen uptake and presentation. However, the injection of the plasmid encoding hGM-CSF into humans either prior to or following the injection of MuStDO5 would be impractical, even if it improved the quality of immune response.81
In addition, plasmids encoding other cytokines have significantly enhanced immunogenicity when co-administered with a DNA vaccine, including IL-15,82,83
IL-12 and IL-28B,84
high mobility group box 185
and may be preferable to GM-CSF. Alternatively, an adjuvant that acts through an immunostimulatory mechanism could bypass the need to include single cytokines as co-treatments.
In summary, the MuStDO5 vaccine trial represented a first step in the development of the DNA vaccine approach to inducing protective cell-mediated immunity targeting multiple pre-erythrocytic stage antigens. The vaccine was safe and well-tolerated and induced IFN-γ responses to all five malaria proteins, supporting the multi-antigen strategy, but it was insufficiently immunogenic, and GM-CSF did not enhance immunogenicity or protection, contrary to expectations based on animal models. Key approaches to improving DNA vaccines, already under study at the time the MuStDO5 trial was performed, include codon optimization and other enhancements to the plasmid, use of more compatible plasmid mixtures to eliminate interference, electroporated administration, use of more concentrated formulations to allow delivery of higher plasmid doses, co-administration with alternative cytokine plasmids, immunostimulatory adjuvants or boosting with recombinant viral vectors such as adenovectors. In follow-on studies of DNA vaccines for the prevention of malaria, we have used codon optimized plasmids, higher doses of DNA (1 mg per plasmid), and boosting with recombinant adenovirus vectors; these three changes, planned at the time that the MuStDO5 trial was performed, have resulted in improved immunogenicity and protection, to be reported in future publications.