Received 6/18/97, Accepted 7/22/97

Almost all the current designs for an HIV-1 vaccine are based on one of the following: inactivated whole virus; live-vector-driven subunit or recombinant subunit virus vaccines; chemically-synthesized peptides; viral pseudo types; or DNA vaccines (reviewed in 53-54). Yet none of the experiments to date in either humans or primates have produced satisfactory results (reviewed in 26-28, 48). If partial protection is achieved, it is produced with either low doses of the pathogenic virus itself, or with a strain of genetically-closely-related simian immunodeficiency virus (SIVs) with a low pathogenicity (41-43). In addition, inactivated and recombinant approaches to vaccine design leave us with other concerns—can these vaccines prevent the cell-to-cell transmission of HIV-1 or entry of virions through the mucosal route, since these vaccines do not generally provoke an IgA-type of response? (49-56).

All of these designs of vaccine are directed towards arming the hosts with HIV-1-specific humoral immunity (HI) or cell-mediated immunity (CMI). This focus is due largely to history and momentum—HI and CMI are two well-studied and well-understood immunologic phenomena, and agents stimulating these responses have led to many successful vaccines in the past. However, new data suggest that there are problems in relying upon these approaches for a vaccine against retrovirus like HIV-1.

Most common attempts to develop a vaccine against HIV-1 or SIV (the best existing primate model for AIDS) focus on inactivated virus or various forms of recombinant vaccines (48-51). Despite extensive research, these attempts have yet to yield successful results. If some sort of protection is reported, it is only under very specific conditions (i.e. with small infectious doses or with secondary challenge with a less pathogenic virus, etc.) (38-43, 54). With SIV, the numerous attempts to develop protection with viral subunits, whole inactivated SIV, recombinant vectors expressing viral proteins and various other combinations have met with almost total failure. Even when various vaccines induced high levels of SIV-neutralizing antibodies or CMI, no protection was observed against the pathogenic strain of simian immunodeficiency virus (SIV) (SIVmac or SIVsm) (56-59).

More surprisingly, infection of a chimpanzee—our closest primate relative—with HIV-1 results in an initial viremia, but without the manifestation of any subsequent disease. Somehow this primate possesses a natural ability to prevent replication of HIV-1 and adverse consequences of the infection (56-59). However, in various trials, vaccines have failed to prevent this initial viremia in immunized chimpanzees—even in the presence of high HIV-1-specific neutralizing antibodies and CMI responses (55-59). Yet other non-vaccinated chimpanzees (such as the control animals in the above trials) are able to contain the initial viremia of HIV-1 equally well in the absence of either of these immune responses and neither of these animals develop any evidence of the HIV-1 infection (60-64). The correct interpretation of these observations may hold the key to a successful vaccine against HIV-1.

On the contrary, the development of so called "enhancing antibodies" which actually increase viral replication have been well documented in both human and animal studies. Furthermore, in certain animal trials, specific vaccines have resulted in the development of antibodies which markedly enhanced viral replication and disease progression (reviewed in 65). Currently, there are numerous suggestions to target beta-chemokines or their receptors (by synthetic or pharmacological agents) to block HIV-1 entry. We believe that such suggestions should be viewed with great care, since previously soluble CD4+ antibodies have suffered total failure (26-27, 50-52). Also, one should keep this fact in mind that the affinity for gp120 to CD4 receptors in their native form is very strong—the Ka association constant gp120-CD4 binding is about 2x10^9-11. However, the affinity of gp120 to soluble CD4 molecules produced by hybridomas or recombinant antibodies are usually much lower (2-3 log lower). In addition, receptor-ligand interactions are dynamic and they are always in a state of flux between association and dissociation, providing the high affinity HIV-1 to latch on to the target cells and enter the cells if co-receptors are present.

This "CMI hypothesis" was based upon several observations which supported the protective role of CMI. For example, at an earlier stage of the AIDS epidemic, Clerici et al (1) had shown that a majority of individuals exposed to HIV-1 were still seronegative for HIV-1, and a small percentage of unexposed or low-risk subjects showed evidence of pre-existing HIV-1-specific CMI. Lymphocytes (PBMCs) from these individuals released IL-2 and exhibited evidence of proliferation when exposed to the gp120 envelope antigen of HIV-1. Furthermore, macaque monkeys (which develop an AIDS-like immunodeficiency upon infection with SIVmac) that are exposed to low-doses of live SIVmac, usually exhibit CMI responses to SIVmac, without producing antibodies specific to SIVmac or developing any evidence of infection (reviewed in 67). On the other hand, all but one of the macaques in this study became infected when injected with high doses of SIVmac, and these animals developed SIVmac antibodies but showed no evidence of CMI. From these observations, Salk et al deduced that cell-mediated immunity (TH1) can protect against pathogenic retroviral infections (i.e. HIV-1 and SIVmac) whereas humoral responses (TH2) makes the host susceptible to infection with these types of viruses (66,68).

In a nutshell, the CMI-hypothesis proposed by Salk et al (66) holds that a patient's fate is determined by which of two types of immune effector cells-TH1 or TH2 is predominant in responding to HIV-1. According to this hypothesis, HIV-1-infected subjects switch from TH1 protective to a TH2- disease enhancing response. Since the above-referenced study was published, there have been many studies which do not support the CMI-hypothesis. For example, Romagnani et al (68) tested this hypothesis and failed to confirm the hypothesis of TH1/TH2 switch during the progression of HIV-1 infection. They did not observe any increase of IL-4, IL-5 and IL-10 production during progression of disease. Since then, it has been documented with HIV-1-infected humans that rapid progressors as well as long-term nonprogressor (LTNP) both exhibit good HIV-1-specific humoral and cell-mediated immunity, suggesting that neither humoral nor CMI response provides protection against HIV-1 (68). Furthermore, in chimpanzees, the natural protection against HIV-1 can occur in the absence of either CMI or HI responses (69).

In the simian models of AIDS, the role of humoral vs. CMI response against retroviruses remains unclear. In the African green monkeys, neither whole inactivated virus (designed to produce strong CMI nor HI have managed to protect these animals against SIVagm infection. In addition, several excellent studies of humans have reported that the decline in the initial HIV-1 viremia occurs before either neutralizing antibodies or HIV-1-specific CMI responses appear in the infected individual (69-70).