A Plasmid Encoding Murine Granulocyte-Macrophage Colony-Stimulating Factor Increases Protection Conferred by a Malaria DNA Vaccine¹

P;-5q;44qlasmodium yoelii (Py)³ infects laboratory mice and has provided a useful model for studying antimalarial immunity. Malaria parasites develop in the liver before causing blood stage infection, and immune responses directed against liver parasites can protect against disease in mice and humans (1, 2, 3). In the mouse, a number of immune effector responses can stop malaria infections at the liver stage, probably either by killing the intracellular parasite or destroying the infected liver cell. These effector responses include CD4⁺ (4, 5, 6, 7) and CD8⁺ (8, 9) T cells, Abs (8), IL-12 (10), IFN-γ (8), and nitric oxide (11, 12, 13). Several Py proteins have been identified as targets of these immune responses, among them is Py circumsporozoite protein (PyCSP) (14). This protein has well-characterized epitopes for Abs (15, 16), CD4⁺ (4), and CD8⁺ (14) T cells, making it a useful tool for monitoring a range of immune responses. In addition, one DNA plasmid vaccine encoding the PyCSP fused to 82 amino acids (aa) of IL-2 protected 54% of BALB/c mice against sporozoite infection (17).

Granulocyte-macrophage CSF (GM-CSF) is a glycoprotein of 127 aa in humans that was first described as a growth factor for stem cells of the granulocyte and macrophage lineages (18). Subsequently, GM-CSF has been found to have effects on many cell types, both bone marrow derived and somatic. In particular, GM-CSF enhances the maturation of dendritic cell precursors (19). GM-CSF recombinant protein has been used as a vaccine adjuvant with hepatitis B vaccine in humans, where it led to Ab production after a single immunization (20). GM-CSF has also been been studied as an adjuvant of DNA vaccines in rodent models (21, 22, 23, 24, 25). However, in none of the published studies was in vivo protection measured as well as the range of Ab and T cell effects of GM-CSF. We now report that mixing GM-CSF plasmid with the PyCSP plasmid vaccine enhances protection against malaria and increases B cell, CD8⁺ T cell, and especially CD4⁺ T cell responses to PyCSP.

The PyCSP encoding plasmid used in these studies, PyCSP₁₀₁₂, was created by PCR amplification of the DNA sequence encoding the PyCSP from the plasmid nkCMVintPyCSP.1 (17) and ligation into the plasmid VR1012 (26). Transfection of UM449 human melanoma cells with plasmid VR2507 led to expression of PyCSP, as assessed by Western immunoblotting analysis using an anti-PyCSP mAb NYS1 (27) (data not shown). The plasmid encoding murine GM-CSF was produced in the VR1019 plasmid, which is a version of the VR1012 plasmid with the addition of a leader element from rat preproinsulin II (28). The plasmid encoding mouse GM-CSF mutated at amino acids 15 (H to A) and 21 (E to A) was produced by cloning cDNA provided by Dr. M. Prystowsky (29, 30) into the VR1019 plasmid. Plasmids for immunization were purified by double cesium banding and diluted in normal saline. Endotoxin levels were less than 0.6 EU/mg.

UM449 cells were provided by Dr. Peter Hobart (Vical, Inc., San Diego, CA). Transient transfections with plasmid constructs for Western blot analysis were performed as described previously (31). Ab and positive control protein for GM-CSF studies were polyclonal rabbit anti-mouse GM-CSF Ab and recombinant mouse GM-CSF produced in yeast, both from (Genzyme, Cambridge, MA). We used the M-NFS-60 cell line as a assay of bioactive murine GM-CSF (32), a gift of Dr. Drew Pardoll (Johns Hopkins University, Baltimore MD).

P815, EL4, and A20 cell lines were purchased from the American Type Culture Collection (Manassas VA).

BALB/cByJ, B10.D2, and B10.Q female mice were purchased from the The Jackson Laboratories (Bar Harbor, ME). Mice received their first immunization at 4 to 6 wk of age.

Mice were immunized two times at 6-wk intervals. A total volume of 50 μl was injected into the tibialis anterior muscle of each leg. Plasmids were mixed and injected in the same syringe. Unless otherwise mentioned, immunizations were with 50 μg of PyCSP₁₀₁₂ plasmid and 50 μg of GM-CSF plasmid at each site.

Three weeks after the second DNA immunization, mice were challenged by i.v. injection with 100 Py sporozoites. To document malaria infection, Giemsa-stained blood films were examined on days 7, 11, and 14 after challenge.

All peptides were kindly provided by Dr. G. Corradin (University of Lausanne, Epalinges, Switzerland). Peptides corresponding to PyCSP 57–70 aa (KIYNRNIVNRLLGD), 58–67 aa(IYNRNVRL), 280–295 aa (SYVPSAEQILEFVKQI), and 280–288 aa (SYVPSAEQI) were used for in vitro T cell studies (4, 14). For capture Ags in ELISAs of serum Abs, we used the PyCSP repeat peptide (QGPGAP)×4 (27) and the PyCS.1 recombinant protein comprising PyCSP 64–321 aa fused to 81 aa of the nonstructural protein of influenza A (15).

Spleen were harvested 3 wk after the second DNA immunization unless otherwise stated. Depletion of CD4⁺ and CD8⁺ cells was accomplished by negative selection using Ab coated magnetic beads from Dynal (Lake Success, NY) or Miltenyi Biotec (Auburn CA), using the manufacturer’s methods.

Spleen cells were cultured in 96-well flat-bottom plates at 2.5 × 10⁵ cells per well at 37°C and 5% CO₂. Medium was DMEM supplemented with 10% FCS, 5000 units penicillin/streptomycin, and l-glutamine. After 5 days of culture, 1 μCi of tritiated thymidine in 10 μl volume was added to each well, and on day 6 cultures were harvested and radioactivity incorporated into DNA was measured by scintillation counting.

ELISPOT assays for production of cytokines by T cells were performed by using three different methods. 1) Ag-specific IFN-γ secreting cells were detected by culturing spleen cells overnight with P815 cells pulsed with PyCSP peptide (aa 280–288) as previously published (33). 2) Lymph node or spleen cells were taken directly from mice and tested for spontaneous IFN-γ and IL-4 production as previously published (34, 35, 36). 3) Spleen cells were cultured with soluble PyCSP peptide (aa 280–295 or aa 57–70) for 6 h, and cells were assayed by ELISPOT for production of IL-2, IL-4, IL-6, IL-10, IL-12, IFN-γ, and GM-CSF in a modification of a previously published assay (34) Briefly, serial 5-fold dilutions of a single cell suspension starting from 1 × 10⁶ cells/well were incubated with or without 3 μM of synthetic PyCSP peptide or Con A for 6 h at 37°C, 5% CO₂. Culture was in 96-well plates coated with anti-cytokine Abs. The wells were overlaid with 0.05 ml of 1 μg/ml biotinylated anti-cytokine Ab for 2 h, then developed by treated with avidin conjugated alkaline phosphatase and 5-bromo-4-chloro-3-indolylphosphate/nitroblue tetrazolium to reveal spots corresponding to the position of cytokine secreting cells.

ELISAs for detection of Abs to PyCSP repeat peptide (QGPGAP)×4 (27) and the PyCS.1 recombinant protein (15) were as previously described. ELISAs for detection of IL-4 and IFN-γ after in vitro restimulation were as previously described (37). Briefly, in a 96-well flat-bottom plate, 1.5 million spleen cells per well were cultured for 3 days with synthetic PyCSP peptides (aa 280–295 or aa 57–70) at a concentration of 3 μM. Con A or medium were used in control wells. After 72 h incubation at 37°C, 5% CO₂, supernatants were collected and diluted 1:5 with a solution of 1% BSA in PBS. These supernatants were then used in ELISA assays for IFN-γ and IL-4. In some experiments, spleen cells were depleted of CD8⁺ or CD4⁺ cells before culture.

CTL assays using the PyCSP 280–288 aa K^d-restricted epitope were performed as previously described (14). Briefly, 5 million spleen cells were cultured at 37°C and 5% CO₂ in wells of a 24 well plate in 2 ml of medium containing DMEM supplemented with 10% FCS, 5000 units penicillin/streptomycin, and l-glutamine. PyCSP 280–295 aa peptide was added to a 2 μM concentration. Forty-eight hours later, 20 units per ml of recombinant human IL-2 (Cetus, San Francisco, CA) was added, and cultures continued for an additional 5 days, when they were used as effector cells. Target cells were P815, A20, or EL4 cells cultured overnight with Cr-51 and 1 μM PyCSP 280–288 aa peptide. CTL assays using the PyCSP 57–70 and 58–67 aa epitopes were run by using these peptides in the same protocol.

BALB/c mice were immunized with the PyCSP₁₀₁₂ plasmid vaccine alone or mixed with GM-CSF plasmid. Mice were challenged with a large inoculum of 100 Py sporozoites, as in our experience 1 to 10 sporoites is enough to infect 50% of normal mice. Animals were followed through day 14 for the development of parasitemia. As is usual with Py, infected mice developed parasitemias of 20% whereas protected mice had no detectable parasites. Although the fraction of mice protected from sporozoite challenge varied between experiments (probably due to differences in parasite viability/virulence), the inclusion of GM-CSF with the vaccine invariably improved protection, nearly doubling efficacy (Table I). When the fraction of protected mice in each experimental group was summed over the six experiments, there was a statistically significant enhancement of protection when GM-CSF and PyCSP₁₀₂₀ were mixed together (p < 0.0001 χ², Yates corrected). The addition of plasmid encoding mutated GM-CSF did not increase protection compared with PyCSP₁₀₂₀ plus control (p = 0.55, Fisher’s exact test). The GM-CSF-encoding plasmid alone did not protect.

The backbone of the DNA plasmids used in this work are of bacterial origin and contain immunostimulatory sequences consisting of a central nonmethylated CpG dinucleotide flanked by two 5′ purines and two 3′ pyrimidines (38, 39, 40, 41). To determine whether the adjuvant effects of the GM-CSF plasmid were due to such DNA motifs in the backbone, a control plasmid encoding a nonactive form of GM-CSF (altered at two amino acids required for binding to the GM-CSF receptor (29, 30)) was constructed. In vitro transient transfection experiments show that both the native and mutated forms of GM-CSF were produced in transfected cells (Fig. 1,A), yet only the native form of GM-CSF showed activity in the bioassay for GM-CSF (Fig. 1,B). When coinjected with the PyCSP₁₀₁₂ plasmid, the plasmid encoding the mutated form of GM-CSF did not increase protective immunity in vivo (Table I). From these experiments we conclude that increased protection is not due to inherent stimulatory properties of the injected DNA itself, but depends on the encoded GM-CSF protein.

A dose/response experiment was conducted to determine the minimum amount of GM-CSF plasmid required to improve the protective immunity induced by PyCSP₁₀₁₂ plasmid. As seen in Figure 2 A, as little as 3 μg of plasmid enhanced protection.

GM-CSF plasmid improved the protective response to PyCSP₁₀₁₂ plasmid vaccine in both strains of H-2^d mice tested, BALB/c and B10.D2 (Fig. 2,B). B10.Q mice are known to be poor responders to the PyCSP DNA vaccine (11), although they are protected by Py sporozoite immunization and make Abs against the parasite (42). B10.Q mice were not protected by the combination of PyCSP₁₀₁₂ plus GM-CSF plasmids (Fig. 2 B), nor did they make Abs to PyCSP (data not shown).

Because GM-CSF plasmid enhanced protection, we began a series of in vitro studies to determine which aspects of the immune response were influenced by adding this plasmid. Using previously defined epitopes on the PyCSP, we looked at CD8⁺ and CD4⁺ T cell responses, cytokine production, and Ab levels.

Ag-specific CD8⁺ T cell responses were measured by two methods: IFN-γ ELISPOT and chromium release assay. In the IFN-γ assay, spleen cells from mice immunized with or without GM-CSF plasmid were cultured in vitro overnight with syngeneic P815 cells pulsed with the PyCSP 280–288 aa peptide which binds to MHC class I K^d. Mice immunized with both GM-CSF plus PyCSP plasmids had 3- to 4-fold more cells secreting IFN-γ than did mice immunized with the PyCSP₁₀₁₂ plasmid alone or mixed with control plasmid p = 0.039 (Fig. 3 A). When spleen cells were depleted before culture using magnetic beads coated with anti-CD4 or anti-CD8 Abs, IFN-γ production was entirely associated with the CD8⁺ cell population (data not shown). In all cases, cells cultured with P815 cells that had been pulsed with an irrelevant (control) peptide were not stimulated to secrete IFN-γ (data not shown).

T cell cytotoxicity was then studied in a chromium release assay using a 5-day in vitro stimulation with PyCSP 280–295 aa or 57–70 aa peptide. Cells from BALB/c mice immunized with PyCSP₁₀₁₂ plasmid or PyCSP₁₀₁₂ plus GM-CSF plasmids were tested in parallel. Spleen cells cultured with PyCSP 57–70 aa did not lyse either P815 or A20 target cells pulsed with PyCSP 57–70 aa or PyCSP 58–67 aa (data not shown). In contrast, spleen cells cultured with PyCSP 280–295 aa peptide lysed target cells pulsed with PyCSP 280–288 or 280–295 peptide (Fig. 3,B). Lytic activity was almost entirely due to CD8⁺ T cells, as determined by magnetic bead depletion of effector cell cultures. Immunization with GM-CSF plasmid did not enhance the amount of specific lysis measured. Although Figure 3 B shows only data for effector:target ratio = 80:1, specific lysis at 40:1 and 20:1 also unchanged (data not shown).

T cell proliferation to the PyCSP 57–70 aa peptide was greatly increased by the inclusion of GM-CSF plasmid (Fig. 4). This proliferation was dependent on CD4⁺ T cells, as magnetic bead depletion of CD8⁺ cells at the start of culture had no effect on thymidine uptake, while depletion of CD4⁺ cells or Th1⁺ cells eliminated all thymidine uptake (data not shown). There was no detectable thymidine uptake in cultures with the PyCSP 280–295 aa peptide (data not shown).

Using ELISPOT assays for IL-2, IL-4, IL-6, IL-10, IL-12, GM-CSF, and IFN-γ, we measured cytokine secretion from unfractionated spleen cells after 6-h in vitro culture with PyCSP 57–70 aa or PyCSP 280–295 aa peptides in solution (Fig. 5). No Ag-specific responses were seen for IL-6, IL-12, or GM-CSF (data not shown). Small increases in Ag-specific IL-4 were detected only in mice immunized with PyCSP₁₀₁₂ plus GM-CSF plasmids, and only with the PyCSP 57–70 aa peptide. Immunization with PyCSP₁₀₁₂ plasmid either alone or with GM-CSF led to IL-10 production which was decreased by the restimulation with either peptide. A 2-fold increase in numbers of cells secreting Ag-specific IFN-γ to either peptide was detected in mice immunized with the addition of GM-CSF plasmid. Adding GM-CSF plasmid caused a large increase in the numbers of IL-2 secreting cells present after restimulation with either peptide.

To determine whether CD4⁺ or CD8⁺ T cells were responsible for producing the IL-2 and IFN-γ after peptide stimulation, the ELISPOT experiments were repeated after magnetic bead depletion of cell subsets (data not shown). Bead depletion of CD8⁺ cells completely eliminated IL-2 and IFN-γ producing cells after PyCSP 280–295 peptide restimulation, but only reduced by half the numbers of IL-2 and IFN-γ producers from PyCSP 57–70 stimulated cultures This is consistent with the hypothesis that 280–295 is primarily an epitope recognized by class I-restricted CD8⁺ T cells, whereas 57–70 contains both class I⁴ and class II-restricted epitopes for CD8⁺ and CD4⁺ cells, respectively.

This interpretation was confirmed by IFN-γ ELISA assay. Spleen cells from mice immunized with PyCSP₁₀₁₂ plasmid alone or with GM-CSF plasmid were depleted of CD4⁺ or CD8⁺ cells, and cultured with peptide PyCSP 280–295 aa or PyCSP 57–70 aa at 3 μM final concentration for 72 h. Culture supernatants were assayed for IFN-γ by ELISA. In accordance with the ELISPOT results, coimmunization with GM-CSF plasmid increased Ag-specific IFN-γ production after stimulation with either peptide (data not shown). Depletion of CD8⁺ cells but not CD4⁺ cells eliminated IFN-γ production with PyCSP 280–295 aa peptide. Depletion of either CD4⁺ or CD8⁺ cells eliminated only half of the IFN-γ produced after stimulation with PyCSP 57–70 aa peptide.

To examine the kinetics of T cell activity after immunization, the number of spleen cells spontaneously secreting IL-4 or IFN-γ were monitored by ELISPOT assay (Fig. 6). Mice immunized with the mixture of PyCSP₁₀₁₂ and GM-CSF plasmids had larger numbers of IFN-γ secreting cells that remained elevated longer after the second immunization. As we previously described (26), IL-4 producing cells were present after the first immunization and then declined in all groups (data not shown).

Immunized mice were monitored weekly for the production of Abs reactive with the CS.1 protein (Fig. 7). Administration of PyCSP₁₀₁₂ plasmid alone induced only a modest increase in IgG Abs 3 wk after primary immunization and an anamnestic response 2 wk postboost. By comparison, coadministration of the GM-CSF plasmid significantly increased total IgG, IgG1, and IgG2a anti-CS.1 serum titers. IgG2b was also increased but IgG3 and IgM were not (data not shown). A similar increase in PyCSP-specific Ab titer in mice coimmunized with PyCSP₁₀₁₂ plasmid plus GM-CSF plasmid was seen using the (QGPGAP)×4 repeat peptide in the ELISA assay (data not shown).

Our laboratory has been searching for ways to enhance the effectiveness of DNA vaccines. Protection of mice against malaria by immunization with PyCSP plasmids was first reported in 1994 (17). Using two different plasmid constructs, 54% of BALB/c mice were protected against a challenge dose of 100 Py sporozoites (commonly, a single sporozoite can cause infection). By comparison, the gold standard of protective immunity is the irradiated sporozoite vaccine, which protects 95% of BALB/c mice against challenge by >5000 sporozoites (42). Recent studies indicate that PyCSP plasmid vaccines do not protect all strains of mice from infection (11). Even for BALB/c mice, recent experience in our lab shows that protection can vary from 0 to 80% depending on the plasmid backbone, the skill of the person immunizing, the infectivity of the parasite, and other as yet undefined factors. The PyCSP₁₀₁₂ plasmid used in the studies described in this paper was constructed using a plasmid backbone (VR1012) developed for use in humans. Immunization with this PyCSP₁₀₁₂ plasmid alone gave 28% protection, and lower Ab and T cell responses than seen in other studies (11, 42).

Coadministering plasmid encoding murine GM-CSF with PyCSP₁₀₁₂ plasmid reproducibly enhanced protection in BALB/c mice. Initial screening experiments mixing PyCSP₁₀₁₂ plasmid with other plasmids encoding murine IL-2, IL-4, IFN-γ, IL-12, IL-15, B7.1, or B7.2 showed no large or consistent increases in the number of mice protected (W. R. Weiss, unpublished observations). We therefore focused our work on GM-CSF plasmid.

Our initial questions regarded the mechanism by which the addition of GM-CSF plasmid enhanced protection. One possibility was that immunostimulatory DNA sequences in the GM-CSF plasmid were increasing responses to the PyCSP₁₀₁₂ plasmid. In this case, enhancing effects would not depend on the production of bioactive GM-CSF in vivo, but only on the primary structure of the DNA. To test this hypothesis, we constructed a plasmid encoding an altered GM-CSF, identical to the first plasmid except for changes at two amino acids required for binding to the GM-CSF receptor. The mutant GM-CSF plasmid produced a protein product in in vitro transfections; however, mixing the mutant GM-CSF plasmid with PyCSP₁₀₁₂ plasmid gave no increased protection against malaria infection. We believe that these data largely eliminate the possibility that direct stimulatory effects of the plasmid DNA are responsible for enhanced protection. We infer that it is the production of bioactive GM-CSF protein after injection which is critical for the enhancing effect.

In vivo production of GM-CSF might increase protection of a PyCSP₁₀₁₂ plasmid vaccine by either of two mechanisms: GM-CSF could be killing the parasite by its direct action, or it could be increasing the immune response to PyCSP₁₀₁₂. To measure the direct effect of GM-CSF on the parasite, we injected mice with GM-CSF plasmid alone and found that it gave no protection against a challenge with 100 Py sporozoites. Although this is not definitive evidence, it suggests that GM-CSF does not directly kill the malaria parasite at the levels produced by plasmid injection. We are currently studying the effects of recombinant mouse GM-CSF on protection against malaria, but our hypothesis now is that GM-CSF has its most important effect as a modulator of the immune response to PyCSP₁₀₁₂ plasmid.

Immunization with PyCSP encoding plasmid alone induces protection dependent on CD8⁺ effector T cells (17). We have not yet defined the immune responses that are responsible for protection after immunization with PyCSP₁₀₁₂ plus GM-CSF. However, in vitro studies provide insights into the possible mechanisms of increased protection. Inclusion of GM-CSF plasmid stimulated a rapid increase in the number of spleen cells capable of secreting Ag-specific IL-2 and IFN-γ. IL-4 production was increased early and then declined, whereas IL-6, IL-10, IL-12, and GM-CSF production were not altered. Despite the enhancement of Th1-type cytokines, serum Abs of IgG1, IgG2a, and IgG2b isotypes were equally increased, indicating that the effect of GM-CSF on Ab production is complex and cannot be explained by the Th1 vs Th2 dichotomy. CD4⁺ T cell proliferation and cytokine responses to the PyCSP 57–70 aa epitope became prominent, increasing 10- to 100-fold. Finally, the addition of GM-CSF more than tripled the frequency of CD8⁺ T cells responding to the PyCSP 280–288 aa epitope as measured by IFN-γ or IL-2 ELISPOT. However, CTL activity of CD8⁺ T cells to the same 280–288 aa epitope was not changed by the inclusion of GM-CSF plasmid. If this is not a technical artifact due to the different sensitivities of the two assays, it may indicate that there are two distinct populations of Ag-specific CD8⁺ T cells, and that enhanced protection is associated with expansion of the subpopulation making cytokines- but without lytic activity. This is consistent with data showing that in many models of pre-erythrocytic malaria immunity, protection can be eliminated by neutralizing CD8⁺ T cells (8, 9), IFN-γ (8), or nitric oxide (11, 12, 13).

Kinetic studies of cytokine production in spleen showed that GM-CSF both accelerates and increases the magnitude of immune responses. Coimmunization stimulated a more rapid rise in the number of cells secreting IFN–γ in vivo, and a higher and prolonged response to in vitro restimulation with a PyCSP peptide. Similarly, Ab levels to PyCSP appeared earlier and at higher titers in animals receiving GM-CSF plasmid.

There have been several recent reports involving the immune-enhancing properties of GM-CSF plasmid. Xiang and Ertl (21) coinjected the GM-CSF plasmid with a plasmid encoding a rabies protein and found enhanced Ab production and increased protection in mice. That report did not examine CTL or Th responses. Iwasaki et al. (23) injected GM-CSF plasmid plus a plasmid encoding influenza NP and found that CTL responses were enhanced but did not report on Ab or Th responses. Kim et al. (22) mixed GM-CSF plasmid with plasmids encoding proteins from HIV-1 and found increased Ab production and T cell proliferation, but no increase in CTL. Geissler et al. (24) administered hepatitis C core protein with GM-CSF encoding plasmid and found an increase in Abs, a small increase in T cell proliferation, but no change in cytokine secretion or CTL activity. Okada et al. (25) used GM-CSF plasmid along with plasmid encoding HIV env protein given intranasally in liposomes and found enhancement of both Ab and CTL activity. Thus, the literature on GM-CSF plasmid immunization consistently describes increases in Abs and T cell proliferation, whereas the enhancement of other T cell responses has been inconsistent.

GM-CSF can act on many cell types (reviewed in 19 . Macrophages, dendritic cells, Langerhans cells, eosinophils, granulocytes, megakaryocytes, fibroblasts, and red blood cell precursors among others can respond to GM-CSF. We hypothesize that GM-CSF is activating dendritic cells, leading to the enhanced presentation of PyCSP epitopes. It is known that bone marrow-derived cells (such as the dendritic cell) are required for presentation of DNA plasmid Ags after i.m. injection (43, 44, 45). We suggest that GM-CSF-activated dendritic cells are better able to present Ag to naive T cells. We are particularly impressed by the enhanced CD4⁺ T cell responses we have measured. It is possible that better CD4⁺ T cell induction would lead to increased B cell and CD8⁺ T cell responses against some epitopes. However, we do not understand why some CD8⁺ T cell responses are boosted and some are not. Neither do we understand how MHC controls GM-CSF responsiveness, as seen in our experiments with B10.Q(H-2^q) mice. We are currently investigating these aspects of GM-CSF enhancement.

We believe that a plasmid encoding human GM-CSF may be clinically useful in enhancing DNA vaccines in humans. Recombinant human GM-CSF is used in patients to stimulate cell growth from bone marrow cells in several clinical settings (46), and its potential toxicities are well defined (20, 46, 47, 48). Human GM-CSF has already been tested as an adjuvant in the immunotherapy of human cancer (49) and to enhance Ab responses to recombinant hepatitis B vaccine (20). We have seen no evidence of ill effects in mice given GM-CSF plasmids. If our plasmid dose/response studies in mice are a guide, it should be possible to avoid systemic toxicity by using a very small amount of human GM-CSF plasmid to achieve adjuvant effects. If safety concerns related to injecting a human cDNA can be addressed, we are optimistic that DNA vaccines in humans can be enhanced by the addition of either plasmids encoding GM-CSF or recombinant GM-CSF cytokine.

We thank Dominic Gonzalez, Arnel Belmonte, and Romeo Wallace for technical assistance.