Cynomolgus macaques (Macaca fascicularis) or rhesus macaques (M. mulatta) weighing 4 to 6 kg were used. For vaccine studies with VEE replicons, EBOV GP or NP genes were introduced into the VEEV RNA as described (10). Groups of three cynomolgus macaques were vaccinated with VRP that expressed EBOV GP, EBOV NP, a mixture of EBOV GP and EBOV NP, or a control antigen (influenza hemagglutinin) that has no effect on EBOV immunity. Animals were vaccinated by subcutaneous injection of 10⁷ focus-forming units of VRP in a total of 0.5 mL at one site. Vaccinations were repeated 28 days after the first injection and 28 days after the second.

In conducting research with animals, the investigators followed the Guide for the Care and Use of Laboratory Animals prepared by the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, National Research Council (1996). The animal facilities and animal care and use program of the U.S. Army Medical Research Institute of Infectious Diseases are accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International.

For vaccine studies using primates, we adapted the optimal immunization regimens determined from the rodent studies. For the vaccine based on recombinant VACV, the EBOV GP gene was inserted into a VACV transfer vector plasmid, and recombinant VACV expressing EBOV GP were isolated as reported (13). Three cynomolgus macaques were injected subcutaneously with the EBOV GP-expressing VACV vector. Injections were repeated at 28 and 53 days after the first injection.

For vaccine studies with inactivated EBOV whole-virion preparation, viral particles were concentrated from Vero cell culture fluids by ultracentrifugation in a sucrose density gradient. The infectivity titer of the preparation was approximately 8.0 log₁₀ PFU/mL. The preparation was inactivated by exposure to ⁶⁰Co gamma rays (6 x 10⁶ rads). The absence of residual infectivity was proven by exhaustive testing for residual infectivity in assays in Vero cells (15,16). Two cynomolgus monkeys and two rhesus monkeys were injected subcutaneously with a 50-µg dose of the gamma-irradiated virion preparation in RIBI adjuvant (Corixa, Hamilton, MT). As a further check on complete viral inactivation, blood samples taken from the monkeys 3 and 5 days after they received the vaccine were free of infectious viremia. Injections were repeated at days 7 and 35 after the initial injection.

For vaccine studies using a liposome formulation, three cynomolgus monkeys were vaccinated with gamma-irradiated virus encapsulated in liposomes containing lipid A, as described for previous studies in mice (14). Animals received 1.0 mL of the liposome preparation by intravenous injections that were repeated at 28 and 55 days after the initial vaccination. Four macaques (two cynomolgus and two rhesus) served as unvaccinated controls for the VACV, gamma-inactivated virion, and liposome studies.

Anti-EBOV neutralizing antibody titers were monitored by measuring plaque reduction in a constant virus:serum dilution format (15). All macaques received intramuscular injections in the leg with 1,000 PFU of the Zaire subtype of EBOV, which was isolated from a human patient in 1995 (16). Blood was obtained from all monkeys under Telazol anesthesia (Fort Dodge Laboratories, Fort Dodge, IA) at 2- or 3-day intervals postinfection to determine infectious viremia, neutralizing antibody titers, and standard hematologic and clinical pathology parameters. All terminally ill monkeys were killed and necropsied for pathologic examination. Virus infectivity assays on plasma and tissue homogenates were done by forming plaques on Vero cell monolayers as described (15,16).

Tissues were immersion fixed in 10% neutral-buffered formalin and processed for histopathologic and immunohistochemical characteristics as described (17-19). Replicate sections of spleen were stained with phosphotungstic acid hematoxylin to demonstrate polymerized fibrin. Sections of spleen from five EBOV-infected guinea pigs and five mice from previous studies (20,21) were similarly stained for polymerized fibrin. Portions of selected tissues from 11 monkeys were also immersion fixed in 4% formaldehyde and 1% glutaraldehyde and processed for transmission electron microscopy according to conventional procedures (17-19).

Our results indicate that rodent models of EBOV hemorrhagic fever do not consistently predict efficacy of candidate vaccines in nonhuman primates, perhaps because the disease course in rodents differs from that reported in human and nonhuman primates (17-19,22,23). Mice do not have the hallmark disseminated intravascular coagulation (DIC) found in end-stage lesions of humans and nonhuman primates. Viremia and widespread tissue dissemination are much more apparent in nonhuman primates than in guinea pigs (20). In addition, guinea pigs have less DIC than do nonhuman primates. Lymphocyte apoptosis was not reported to be a prominent feature of EBOV infection in mice or guinea pigs (20,21) but was a consistent feature of disease in humans (24) and nonhuman primates (19). Clinical disease and related pathologic features in nonhuman primates infected with EBOV appear to more closely resemble those described in human EBOV hemorrhagic fever (22,23). Other studies have shown inconsistencies between rodent and nonhuman primate models of human hemorrhagic disease in the protective efficacy of candidate vaccines. For example, guinea pigs were protected from Lassa virus by VACV recombinants expressing the viral nucleoprotein (25,26); however, this vaccination strategy failed to protect rhesus macaques (27).

The effort to develop an EBOV vaccine began after the initial identification of EBOV in 1976, but 25 years later the goal remains elusive. Attempts to develop killed-virus vaccines against EBOV hemorrhagic fever have had inconsistent results (5-7). Recent progress in genetic vaccination strategies has demonstrated that immunity can be achieved against a low dose of EBOV. While protection against any lethal challenge dose of EBOV is a remarkable achievement, we have set the bar somewhat higher than 6 PFU, since a laboratory exposure through a needlestick and infected blood would likely entail a dose of at least 1,000 PFU. Therefore, our priority is to empirically develop a vaccine that protects against at least 1,000 PFU rather than to initiate an exhaustive investigation of protective immune mechanisms. We were encouraged by the demonstrated success of the VEEV replicon vector expressing MBGV glycoprotein in protecting cynomolgus macaques from challenge with homologous MBGV (12). No MBGV-neutralizing activity was observed at >1:20 dilutions in prechallenge sera of any of the MBGV GP VRP-vaccinated macaques (12), yet these animals did not become viremic, showed no signs of disease, and survived challenge. Historically, Filovirus-neutralizing antibodies have been difficult to demonstrate in vitro (15); while the presence of neutralizing antibodies is desirable, it is neither sufficient nor necessary to clear viral infection (16). Unfortunately, the VEEV replicon strategy that was successfully employed for MBGV in cynomolgus macaques and for EBOV in mice and guinea pigs (10) did not protect cynomolgus macaques from EBOV disease. These differences observed between EBOV and MBGV may result from differences in the course of infection. Specifically, the mean day of death for untreated cynomolgus monkeys experimentally infected intramuscularly with 1,000 PFU of EBOV (Zaire subtype) is 6.3 (n=15; data not shown), while the mean day of death for cynomolgus monkeys infected intramuscularly with a comparable dose of MBGV (Musoke isolate) is 9.1 (n=8; data not shown). Thus, macaques infected with MGBV have nearly three more days to mount an effective immune response against the challenge virus than macaques infected with EBOV (Zaire). Clearly, other variables, including differences observed between EBOV (Zaire) and MBGV with respect to GP gene expression (28), tropism, and host cell responses, may contribute to differences in disease pathogenesis and outcome of infections.

The induction of humoral and cytotoxic T-lymphocyte responses to EBOV NP and GP has been demonstrated in guinea pigs, although the relative contributions of these responses to immune protection are unclear (9). Moreover, transfer of EBOV immune serum in rodent and nonhuman primate models provided inconsistent results. Passive transfer of immune serum from VRP-vaccinated animals did not protect guinea pigs or mice against lethal challenge (10); however, transfer of hyperimmune equine immune globulin (which had high EBOV neutralization titers) to guinea pigs protected them against disease (16,29). Passive treatment of cynomolgus monkeys with the equine immune globulin delayed death but did not ultimately protect the monkeys against lethal EBOV hemorrhagic fever (16,29). In contrast, hamadryl baboons were protected against lethal EBOV challenge by passive treatment with the equine immune globulin and the use of a lower challenge dose (30). These results suggest that cell-mediated effector mechanisms may play a more important role in protection than do humoral responses. Nonetheless, the role of humoral immunity is in fact supported by studies showing consistent delay in death or protection of primates therapeutically treated with EBOV-neutralizing antibodies (16,29,30).

We conclude that, although rodent models are useful as preliminary screens for candidate vaccines and therapeutic treatments, nonhuman primates likely provide a more useful and definitive model for EBOV hemorrhagic fever in humans. Furthermore, differences in disease pathology between rodent and nonhuman primate models of EBOV suggest that protection of primates may require different protective mechanisms.