Research Article

Fusion-Competent Vaccines: Broad Neutralization of Primary Isolates of HIV

See allHide authors and affiliations

Science  15 Jan 1999:
Vol. 283, Issue 5400, pp. 357-362
DOI: 10.1126/science.283.5400.357


Current recombinant human immunodeficiency virus (HIV) gp120 protein vaccine candidates are unable to elicit antibodies capable of neutralizing infectivity of primary isolates from patients. Here, “fusion-competent” HIV vaccine immunogens were generated that capture the transient envelope-CD4-coreceptor structures that arise during HIV binding and fusion. In a transgenic mouse immunization model, these formaldehyde-fixed whole-cell vaccines elicited antibodies capable of neutralizing infectivity of 23 of 24 primary HIV isolates from diverse geographic locations and genetic clades A to E. Development of these fusion-dependent immunogens may lead to a broadly effective HIV vaccine.

The expanding epidemic of HIV infection threatens to engulf more than 40 million persons worldwide by the year 2000 (1). The need for an effective HIV vaccine is urgent, but progress toward this goal has been slowed in part by the inability of any vaccine candidate to elicit antibodies capable of neutralizing infectivity of primary HIV isolates (PIs) from infected individuals (2, 3).

Because the HIV envelope protein mediates the early binding and entry steps in infection, many vaccine strategies have focused on this target. In 1993, two recombinant forms of the surface gp120 subunit of the HIV envelope protein (rgp120) were advanced as candidate vaccines for a large-scale efficacy study sponsored by the National Institutes of Health (NIH). In previous clinical studies, these rgp120 vaccines had been shown to be safe and to elicit antibodies capable of potently neutralizing related laboratory-adapted isolates of HIV (4). The inability of rgp120 vaccine sera to neutralize PI viruses, however, interrupted this momentum (5) and highlighted many central questions that have continued to impede development of an HIV vaccine candidate.

In contrast to antibodies elicited by rgp120 vaccines, antibodies from persons actively infected with HIV are able to neutralize infectivity of PI viruses, albeit incompletely (6). Surveys of patient sera typically report low-level neutralization of 30 to 50% of PIs, although breadth and titers vary. We speculated that this ability to neutralize PI viruses might be related to the presentation of functioning envelope protein in active infection, as compared with the static, nonfunctioning presentation of the envelope protein in rgp120 vaccines.

HIV Fusion and Fusion-Competent Vaccine Immunogens

The HIV envelope protein orchestrates a complex series of protein-protein interactions and structural changes that ultimately result in fusion of the virus and cell membranes and infection of the cell. Upon binding to CD4, the envelope protein undergoes conformational change that facilitates subsequent interaction with one of several coreceptor molecules, predominantly the CC chemokine receptor 5 (CCR5) or the CXC chemokine receptor 4 (CXCR4) (7). Interaction with either coreceptor induces further conformational change in the envelope protein and exposure of the hydrophobic fusion domain of the transmembrane gp41 subunit, which then mediates fusion of the apposed cell and virus membranes. On the basis of this dynamic model of HIV binding and entry, we undertook to develop HIV vaccine immunogens that explicitly incorporate these functional intermediate structures.

One measure of envelope protein function is the ability to mediate cell-cell fusion. When cells expressing envelope protein are cocultured with cells expressing CD4 and coreceptor, multinucleated syncytia form over the course of 6 to 24 hours. For our vaccine studies, we asked whether this process of binding and fusion might be captured in progress by formaldehyde cross-linking before extensive syncytium formation. In these studies, the functioning envelope protein was derived from a T lymphocytotropic PI virus obtained from the Amsterdam Cohort (ACH168.10; 168P) (8). The molecularly cloned envelope protein, as well as the parental syncytium-inducing (SI) virus, uses both CCR5 and CXCR4 coreceptors. COS-7 cells were transfected to express the envelope protein (COS-env) and subsequently cocultured with human U87 glioma cells that express CD4 and CCR5 coreceptor (U87-CD4-CCR5) (9). To capture transitional intermediates during the process of binding and fusion, we fixed cocultures in 0.2% formaldehyde after 5 hours (10) when few if any multinucleate cells were evident. This inactivated whole-cell preparation was used as the fusion-competent (FC) immunogen.

To test the ability of these complex immunogens to elicit neutralizing antibodies, it was necessary to restrict the immune response to viral and virus-induced epitopes. Otherwise, antibodies to CD4 and CCR5 would be generated that would themselves block infectivity. Therefore, it was essential to use an animal model that was immunologically tolerant to the human (hu) CD4 and CCR5 components of the vaccine, as would also be the case in human immunization. Thus, immunogenicity studies were performed with transgenic mice that express hu CD4 and hu CCR5 coreceptor (11).

In pilot studies, mice were immunized with either FC immunogen or with cell controls (U87-CD4-CCR5 cells, alone or cocultured with mock-transfected COS cells) (12). Sensitivity of the homologous 168P virus to neutralization by vaccine sera was determined with U87-CD4 cells expressing either CCR5 or CXCR4 coreceptor (13). No inhibition of infectivity was observed in sera from mice immunized with cell controls, suggesting that the transgenic mice were in fact tolerant to hu CD4 and CCR5 and that other adventitious cellular reactivities did not interfere with the virus infectivity assay (Fig. 1). Sera from mice immunized with FC immunogens were able to neutralize the homologous 168P PI virus. This neutralization activity was antibody-mediated and could be adsorbed to, and subsequently eluted from, a solid support containing protein A and protein G (14). Furthermore, neutralization of the 168P virus by FC serum was observed regardless of the coreceptor used in the U87-CD4 cell infection assay (Fig. 1). Several reports have demonstrated that, in general, neutralization sensitivity is independent of specific coreceptor use (15–17). The fact that neutralization is observed in this study with CXCR4, a coreceptor to which the animal had not been exposed, argues that neutralization does not directly target the CCR5 component of the vaccine.

Figure 1

Neutralization of the homologous 168P PI virus by FC vaccine sera. Transgenic mice (hu CD4+, hu CCR5+, mouse CD4+) in studies 1 to 3 were immunized with FC immunogen (COS-env with U87-CD4-CCR5) (squares;n = 3 mice) or with cell controls (U87-CD4-CCR5 cells alone or cocultured with mock-transfected COS cells) (circles; n = 3 mice). Unimmunized mice were also used (triangles; n = 2 mice). Sera were tested for neutralization of 168P with U87-CD4 cells expressing either CXCR4 (black symbols) or CCR5 (white symbols). Data represent averages of three to six neutralization assays with serum obtained 2 weeks after the second or third immunization.

To explore the role of fusion-dependent determinants in the induction of PI virus neutralization, we expanded our studies to include fusion-incompetent (FI) immunogens—cocultures that do not undergo cell-cell fusion. These include COS-env cocultured with U87 cells (no CD4 or CCR5 coreceptor), COS-env cocultured with U87-CD4 cells, and COS-env cells to which soluble CD4 (sCD4) was complexed (18). An additional FI immunogen comprised COS-env and U87-CD4-CCR5 cells that were separately fixed with formaldehyde before mixing during the formulation of the vaccine.

In marked contrast to FC immunogens, all FI immunogens were unable to elicit significant neutralization of the homologous PI virus (Fig. 2A). These results are consistent with the well-documented failure of rgp120 vaccines to elicit PI virus neutralization. The difference in neutralization by FC and FI vaccine sera was also observed in assays with human primary blood lymphocytes (PBLs) (Fig. 2B).

Figure 2

Neutralization of 168P by FC, but not FI, vaccine sera. (A) Transgenic mice in study 4 were immunized with FC immunogen (black squares, n = 4), FI immunogens (COS-env with U87 cells, gray circles, n = 4; COS-env with U87-CD4 cells, gray diamonds, n = 3; COS-env with sCD4, white diamonds n = 2; COS-env with U87-CD4-CCR5 cells, each fixed separately before mixing for immunization, gray squares,n = 2), or mock-transfected COS cell immunogen (cocultured with U87-CD4-CCR5 cells) (white circles; n = 2). Unimmunized control mice (white triangles; n = 2) were also used. Neutralization was independent of specific coreceptor use (Fig. 1), and data here represent averages of three to six neutralization assays in U87-CD4-CXCR4 or -CCR5 cells. In some cases, sera that had been analyzed individually were pooled in equal proportions in order to conserve limited amounts of sera. (B) Neutralization by FC and FI vaccine sera was also determined in human PBL culture. PBLs were isolated, stimulated with phytohemagglutinin, and grown in the presence of interleukin-2; neutralization was determined as described (13). HIV p24 antigen was determined after 5 days of culture by ELISA, and values were normalized to the virus control (36 ng/ml). Asterisks indicate p24 antigen concentrations below the limit of detection at the dilution used in the ELISA. Vaccine groups are as defined above, and sera were pooled for this assay.

Despite the apparent requirement for fusion competency in the immunogen, we wanted to exclude the possibility that FC vaccine sera inhibited viral infectivity in a nonspecific manner. Thus, FC vaccine sera were shown not to inhibit infectivity of pseudotyped HIV virions bearing an amphotropic murine leukemia virus (MLV) envelope protein (19), nor to neutralize a PI of the simian immunodeficiency virus SIVmac251 (20) (Fig. 3).

Figure 3

FC vaccine serum does not neutralize pseudotyped HIV virions bearing amphotropic MLV envelope protein (19) or primary SIVmac251 (20). For HIV bearing an amphotropic MLV envelope protein (ampho MLV pseudotype), neutralization sensitivity with pooled FC and FI antisera was determined in U87-CD4-CXCR4 cells. For primary isolate SIVmac251, neutralization was determined in U87-CD4-CCR5 cells. Symbols are as defined in Fig. 2: FC immunogen (black squares) and FI immunogen (COS-env + U87 cells, gray circles).

As with conventional rgp120 vaccines, FI vaccines were able to elicit neutralization of a related laboratory-adapted isolate of HIV, the T cell line–adapted derivative of 168P, 168C (2,15) (Fig. 4). Neutralization titers of the TCLA 168C virus were comparable between FC and FI vaccine sera, as were titers of antibodies to gp120 (anti-gp120), suggesting a similar degree of inherent immunogenicity among the vaccines.

Figure 4

Neutralization of TCLA 168C virus by FI vaccine sera. Neutralization sensitivity of the 168P PI virus and its TCLA derivative 168C were tested in U87-CD4-CXCR4 cells with pooled sera: FC immunogen (black squares), FI immunogens (COS-env + U87 cells, gray circles; COS-env + U87-CD4 cells, gray diamonds; COS-env + sCD4, white diamonds), and mock-transfected cell controls (white circles).

The failure of FI vaccines to elicit PI virus neutralization in the transgenic mouse model highlights the specificity of the neutralization elicited by FC vaccines (21). Furthermore, the consistent failure of FI vaccine sera to inhibit PI virus infectivity strongly indicates that the immune response is not directed to adventitious human cellular targets, such as those that confounded early studies of inactivated SIV vaccines (22). Rather, we suggest that FC immunogens present unique fusion-dependent determinants that mediate neutralization of PI viruses.

Broad Neutralization of Primary Isolates

A critical issue in HIV vaccine development centers on the ability of vaccine antisera to neutralize a broad range of diverse PI viruses. To determine the breadth of PI virus neutralization elicited by FC immunogens, we examined the sensitivity of a panel of representative PI viruses from five prevalent and geographically diverse phylogenetic clades (23). As depicted in Fig. 5, FC sera elicited by a functioning clade B envelope protein were able to neutralize 23 of 24 PI viruses tested—monocytotropic (non-SI; NSI) and T lymphocytotropic (SI viruses) from North America and Europe (clade B), Africa (clades A and D), Thailand (clades B and E), and India (clade C). Despite the sequence diversity among these isolates, most were similarly sensitive to neutralization by FC vaccine sera. One isolate (92RW008) failed to attain >50% neutralization and two others (93IN904 and 92UG024) showed limited neutralization beyond 50%; these exceptions to the otherwise broad pattern of neutralization further indicate that FC immunogens target primarily viral, rather than cellular, determinants. FI sera were uniformly unable to neutralize these heterologous PI viruses, in keeping with the failure of rgp120 immunogens. The broad and uniform neutralization of diverse PI viruses suggests that the critical determinants presented by FC immunogens are highly conserved and may be intimately tied to the basic functioning of the envelope protein in binding and fusion.

Figure 5

Neutralization of diverse PI viruses from clades A to E. Primary isolates were expanded in human PBLs, and neutralization was determined in permissive U87-CD4-CCR5 (or -CXCR4) cells with pooled sera: FC immunogen (black squares), FI immunogens (COS-env + U87 cells, gray circles; COS-env + U87-CD4 cells, gray diamonds; COS-env + sCD4, white diamonds, and mock-transfected cell controls (white circles). Viral biotype is indicated if known.

Molecular Target for PI Virus Neutralization

We next sought to define the molecular target for neutralization by FC vaccines. We had shown that the elicitation of PI virus–neutralizing antibodies required as immunogen a functional envelope-CD4-CCR5 interaction. Nonetheless, these neutralizing antibodies might recognize native, nonfunctioning envelope protein as antigen. A similar instance has been reported for monoclonal antibody (mAb) 17b, which targets a CD4-induced epitope on the gp120-CD4 complex but which also binds isolated gp120 (24). In contrast, another gp120-CD4–specific mAb CG10 (25) does not recognize isolated gp120 or CD4. Thus, we tested whether neutralizing antibodies could be removed from FC vaccine sera on incubation with envelope protein expressed on the surface of transfected COS cells. Formaldehyde-fixed COS cells expressing 168P envelope protein were incubated with FC serum, and the recovered serum was then tested for PI virus neutralization (26). Neutralization activity in FC vaccine serum was removed by incubation with envelope-expressing cells, but only minimally reduced by incubation with COS cell controls (Fig. 6). Although the static form of the envelope protein does not function as an effective immunogen, we found that some aspects of the critical fusion-dependent epitopes are sufficiently represented on the static protein to allow binding. This initial binding to nonfunctioning envelope protein may facilitate subsequent access of antibody to transient fusion-dependent conformations that arise during binding and fusion. Importantly, these data strongly suggest that PI virus neutralizing antibodies target, at least in part, the HIV envelope protein.

Figure 6

Adsorption of PI virus neutralization activity by formaldehyde-fixed COS-env cells. FC vaccine serum was repeatedly incubated with formaldehyde-fixed COS-env (gray squares) or control COS (white squares) cells. Serum obtained before FC immunization was similarly adsorbed (gray and white circles, respectively). The starting FC and preimmunization sera are indicated as black squares or black circles, respectively. Sera were tested for neutralization of 168P with U87-CD4-CXCR4 cells.

These data also provide independent support that neutralizing antibodies do not target adventitious cellular proteins, which would be equally recognized (or not recognized) on control COS cells. To further exclude this possibility, we examined whether neutralization activity could be removed by adsorption to U87-CD4-CCR5 cells. FC serum was incubated with intact U87-CD4-CCR5 cells, and the recovered serum was then tested for PI virus neutralization (27). As demonstrated in Fig. 7, no reduction in FC serum neutralization was observed on adsorption to U87-CD4-CCR5 cells.

Figure 7

PI virus neutralization activity is not adsorbed by intact U87-CD4-CCR5 cells. Pooled FC vaccine serum was incubated with U87-CD4-CCR5 cells (white squares) or in an empty microculture well (mock; black squares). Pooled FI serum (COS-env + U87-CD4) was similarly treated (white and black circles, respectively). Sera were tested for remaining neutralization of 168P with U87-CD4-CCR5 cells.

Taken together, our data provide evidence that FC serum neutralization targets highly conserved envelope protein structures that arise transiently during binding and fusion. Nonetheless, absolute exclusion of fusion-induced cellular targets, and the ultimate definition of specific envelope protein determinants, must await the analysis of PI virus–neutralizing mAbs to FC immunogens.

On the basis of the requirement for interaction with coreceptor, we speculate that neutralization may target a late event in virus binding and entry. Upon binding to coreceptor, the envelope protein must mediate fusion of the viral and cell membranes. Cryptic but highly conserved determinants may be exposed during this process. One possible target for neutralization might involve the trimeric coiled-coil structure that mediates membrane fusion and is highly conserved among Orthomyxoviridae (Influenza), Filoviridae (Ebola), and Retroviridae (HIV) (28). The fusion-active structure of the HIV envelope protein is believed to form subsequent to CD4 and coreceptor binding by the collapse of two helical coils within each gp41 monomer. This trimeric coiled-coil structure is thought to drive membrane insertion of the hydrophobic fusion domain of gp41 and to initiate membrane fusion. Synthetic peptides that comprise either of the gp41 helical coils are able to bind the cognate helical region and broadly inhibit viral infectivity (29). Broadly neutralizing antibodies to FC immunogens may likewise target structures involved in the activation of fusion.

The potency and breadth of neutralization by FC immunogens appears to surpass that observed in sera from infected individuals or in the more broadly neutralizing human mAbs (30). Perhaps formaldehyde fixation traps critical fusion-dependent structures that are only transiently presented during active infection. It remains to be determined whether FC immunogens target the same neutralizing determinants as active infection.

Implications for HIV Vaccine Development

The ability of antibodies induced by FC immunogens to neutralize a broad range of PI viruses necessitates a shift in our thinking on HIV vaccines. Whereas previous discussions regarding the possible number of HIV serotypes were moot in the absence of any vaccine-induced neutralization, it had been widely accepted that multiple envelope protein immunogens might be needed to span the range of HIV sequence diversity. We now show that an appropriately presented clade B envelope protein can elicit potent neutralization against most PI viruses from multiple HIV clades. We show that not only is PI virus neutralization achievable, we suggest that broad vaccine protection may not require an unlimited number of HIV serotypes.

Although the immunological basis for potential HIV vaccine efficacy is presently unknown and controversial, there is ample reason to believe that preexisting neutralizing antibody may offer protection against infection or disease. Passively administered neutralizing antibodies have been shown to exert potent antiviral effects in several experimental models of HIV infection, as manifested by complete or partial protection (31). Antibody-mediated protection is, of course, well accepted in other viral infections and vaccines (32). With the ability to elicit potent HIV neutralization will come the opportunity to determine the role of preexisting neutralizing antibody in HIV prophylaxis. Concerns regarding the diversity of HIV populations are lessened by the breadth of neutralization elicited by FC immunogens. Still, one may raise the concern that extant antibodies in HIV-infected persons do not protect against disease progression. Whether the ultimate disease course is modified by the ongoing antibody response is difficult to assess. In any case, an established chronic infection presents special challenges in terms of host competence and virus load and diversity, challenges that are distinct from those envisioned in prophylactic immunization. The effect of preexisting antibodies in exposure to a minimal infectious dose of virus in an immunocompetent host remains to be determined in preclinical and clinical studies.

We recognize that in its current form an inactivated whole-cell FC vaccine is not practical for clinical development. Nonetheless, FC formulations that incorporate critical fusion-dependent determinants of PI virus neutralization can be envisioned. For example, recombinant viral vectors that respectively express envelope and CD4 with coreceptor could be coadministered to drive critical fusion events in vivo. Alternatively, purified fusion-active complexes could be developed as an inactivated subunit vaccine. This concept of FC immunogens may also be applicable to other enveloped viruses where protection has been difficult to generate other than by live attenuated virus immunization.

  • * Present address: Oregon Health Sciences University, Portland, OR 97201, USA.

  • To whom correspondence should be addressed. E-mail: nunberg{at}


View Abstract

Navigate This Article