|Home | About | Journals | Submit | Contact Us | Français|
The V3 epitope is a known target for HIV-1 neutralizing antibodies (NAbs), and V3-scaffold fusion proteins used as boosting immunogens after gp120 DNA priming were previously shown to induce NAbs in rabbits. Here, we evaluated whether the breadth and potency of the NAb response could be improved when boosted with rationally designed V3-scaffold immunogens. Rabbits were primed with codon-optimized clade C gp120 DNA and boosted with one of five V3-cholera toxin B fusion proteins (V3-CTBs) or with double combinations of these. The inserts in these immunogens were designed to display V3 epitopes shared by the majority of global HIV-1 isolates. Double combinations of V3-CTB immunogens generally induced more broad and potent NAbs than did boosts with single V3-CTB immunogens, with the most potent and broad NAbs elicited with the V3-CTB carrying the consensus V3 of clade C (V3C-CTB), or with double combinations of V3-CTB immunogens that included V3C-CTB. Neutralization of tier 1 and 2 pseudoviruses from clades AG, B, and C and of peripheral blood mononuclear cell (PBMC)-grown primary viruses from clades A, AG, and B was achieved, demonstrating that priming with gp120 DNA followed by boosts with V3-scaffold immunogens effectively elicits cross-clade NAbs. Focusing on the V3 region is a first step in designing a vaccine targeting protective epitopes, a strategy with potential advantages over the use of Env, a molecule that evolved to protect the virus by poorly inducing NAbs and by shielding the epitopes that are most critical for infectivity.
In the HIV vaccine field, prime/boost regimens are among the most effective for inducing protective antiviral antibodies (Abs) (65). The mechanism by which prime/boost immunization works follows fundamental immunologic principles in which T and B memory cells are established with priming. Subsequently, activation and proliferation of Ab-producing B cells are stimulated by the boosting immunogen, working in conjunction with cytokines produced by the memory T cells. While nonprotein primes can induce excellent cell-mediated immunity, proteins are particularly effective for eliciting Ab responses, and the combination of nonprotein priming and protein boosting gives a much stronger Ab response than does delivery of protein vaccines alone (38, 44).
Most studies of various forms of envelope (Env) proteins used as immunogens without previous priming induce neutralizing antibodies (NAbs) of limited potency and/or breadth (25, 37, 46, 59, 63, 70), presumably targeting Env epitopes which are either unexposed on the virion surface or irrelevant in terms of eliciting antiviral activities. These findings suggest that epitope-targeted vaccines might be more effective than various forms of Env, as long as they elicit functional Abs that recognize epitopes that are conserved across HIV strains. However, several groups have failed to elicit epitope-focused Abs specific for epitopes in the membrane proximal external region (MPER) of gp41 or in the CD4 binding site (CD4bs) of gp120 using immunogens designed to focus the immune response on these selected Env regions. For example, the CD4bs is composed of residues in distant regions of gp120, and it is extremely difficult to recapitulate with mimotopes; consequently, attempts to date to focus the immune response on the CD4bs by various approaches have met with failure (53, 57, 58). It has also been difficult to elicit NAbs to epitopes in the MPER of gp41 (11, 15, 16, 19, 31, 41, 48, 50, 51). Though highly conserved and composed of linear stretches of amino acids, this region is poorly immunogenic (12, 54) and requires lipid reactivity as well as recognition of the MPER peptide to effect neutralization (2, 17).
Use of V3-scaffold proteins constructed without structural considerations for the site of insertion into the scaffold or the nature of the V3 portion inserted also failed to induce broadly neutralizing Abs (5, 6). In contrast, we showed that it is possible to focus the immune response on a single epitope and induce NAbs by priming with gp120 DNA and boosting with recombinant proteins that share only the V3 region with the prime. The recombinant proteins consisted of the V3 loop of gp120 inserted into “scaffold proteins” such as a recombinant truncated form of gp70 from murine leukemia virus which carries a single copy of V3 or recombinant cholera toxin B (CTB) which carries five copies of V3 and so induces a stronger Ab response (66, 72, 73). These epitope-scaffold immunogens used in a prime/boost immunization regimen induced cross-clade NAbs in rabbits; >80% of these Abs were shown to be V3 specific (73). Additionally, Moseri et al. have induced clade B NAbs by immunizing with optimally constrained V3 peptides (47). These studies underline the importance of structural elements for constructing epitope-scaffold immunogens that are rationally designed rather than empirically chosen; they emphasize the need for structural understanding of both the epitope and the scaffold, and they attest to the value of bioinformatics and epidemiologic data relevant to the existence and prevalence of shared immunogenic structures among global virus isolates.
Recently, more detailed data have become available mapping the conserved structural elements in the V3 crown within and between HIV subtypes (3, 13, 14, 36, 71). Using these data, newly designed and optimally constructed V3-scaffold protein immunogens have now been synthesized and tested for antigenicity in vitro and for immunogenicity in vivo in rabbits. The results of these experiments show that when primed with codon-optimized clade C gp120 DNA, a V3-CTB immunogen carrying the consensus clade C V3 (V3C-CTB) induces cross-clade NAb responses; in addition, the results show that, generally, double combinations of V3-CTB boosts that include V3C-CTB give the most broad and potent NAb responses.
A codon-optimized gp120 env gene from HIV clade C primary isolate 92BR025.9 was prepared as described previously (73) with the V3 sequence of CTRPNNNTRKSIRIGPGQAFYATGEIIGDIRQAHC, which differs from the consensus clade C V3 loop at the positions underlined.
The sequence of the unmodified wild-type CTB (CTBWT) used as the scaffold for V3 inserts is MTPQNITDLC AEYHNTQIHT LNDKIFSYTE SLAGKREMAI ITFKNGATFQ VEVPGSQHID SQKKAIERMK DTLRIAYLTE AKVEKLCVWN NKTPHAIAAI SMAN. The three amino acids shown in bold and underlined were deleted and indicate the point of the V3 insertion. Genes for unmodified, wild-type CTB (CTBWT, as control) and cholera toxin B (CTB) with various full-length V3 inserts (V3-CTB) were chemically synthesized and cloned into pSUMO plasmids, followed by expression in Escherichia coli and purification by affinity chromatography, as previously described (66). The sequences of the V3 inserts in each of the five forms of V3-CTB are shown in Table 1. V3 inserts consisted of the clade C consensus V3 (V3C-CTB); the clade H consensus (V3H-CTB); the V3 from clade B strain JR-CSF (V3B-CTB), which differs from the clade B consensus by one residue at position 5; and two rationally designed V3 sequences (V32219-CTB and V33074-CTB).
Female New Zealand White rabbits 6 to 8 weeks old (with a body weight of ~2 kg) were purchased from Millbrook Farm (Amherst, MA) and housed in the animal facility managed by the Department of Animal Medicine at the University of Massachusetts Medical School in accordance with an IACUC-approved protocol. Five rabbits were included in each immunization group (with the exception of the group boosted with CTBWT, which contained 3 rabbits). All rabbits received three DNA immunizations at weeks 0, 2, and 4 using a Bio-Rad Helios gene gun (Bio-Rad Laboratories, Hercules, CA). The gp120 DNA vaccine plasmids were applied as a coating onto 1.0-μm gold beads at a ratio of 2 μg of DNA per mg of gold. Each gene gun shot delivered 1 μg of DNA to a total of 36 nonoverlapping sites on the shaved abdominal skin of each rabbit at each of the three priming immunizations. The animals then received two boosts with CTBWT or with one or more of the V3-CTB immunogens at weeks 10 and 14. A total of 100 μg/injection of the V3-CTB(s) was administered intramuscularly with incomplete Freund adjuvant (IFA). Thus, animals receiving a single form of V3-CTB received 100 μg per dose whereas, in those animals receiving double combinations of V3-CTB immunogens, the two forms of V3-CTB were mixed so that 50 μg of each was delivered at each dose. Blood was collected prior to immunization and 2 weeks after each immunization. In one experiment, rabbits were bled periodically for a total of 76 weeks after the initiation of the experiment to assess the longevity of the Ab response.
For detection of NAbs in rabbit immune sera, 14 pseudoviruses from the clade B and the clade C standard pseudovirus panels (42, 43) were used along with an additional nine tier 1A and tier 1B pseudoviruses from clades AG, B, and C. A standard pseudovirus neutralization assay was performed as previously described (42, 56). Briefly, 2-fold serial dilutions of heat-inactivated serum were prepared starting at a dilution of 1:10. The serum/pseudovirus mixtures were then incubated with the TZM.bl target cells, and luciferase activity was measured 48 h later. Pools of prebleed sera were tested as negative controls against each pseudovirus, and all sera were also tested against a negative-control pseudovirus carrying the envelope of murine leukemia virus.
Neutralization of primary isolates grown in human peripheral blood mononuclear cells (PBMCs) was measured as the reduction in luc reporter gene expression after a single round of virus infection using TZM.bl cells as previously described (72, 73). Fifteen viruses from clades A, AG, B, and C were selected to represent a spectrum of neutralization sensitivity when tested with a pool of human anti-V3 monoclonal Abs (MAbs), as described below. Briefly, 2-fold serial dilutions of heat-inactivated serum were prepared starting at a dilution of 1:10. The serum/virus mixtures were then incubated with the TZM.bl target cells, and luciferase activity was measured 48 h later. Prebleed serum from each rabbit was tested as a negative control against each virus, and all sera were also tested in control cultures that contained cells only and cells plus virus. The percent neutralization was calculated relative to the effect of the preimmune serum from the same rabbit at the same dilution. All sera were assayed in duplicate in at least two experiments against each virus. The 50% neutralizing titers (NT50) were determined using the method of least squares.
Breadth-potency (B-P) curves are probability-based “survival” curves which are a modified version of the magnitude-breadth (M-B) curves described in the work of Huang et al. (35) and which had been employed by Gilbert et al. (25) for analyzing panels of neutralization data. “Vaccine scores” and individual “rabbit scores” were obtained as the areas under B-P curves, in analogy with the area-under-the curve M-B curves in reference 35. For x-y plots of the B-P curves (as well as for M-B curves), the x axis is defined as x = log10(NT50), or an average of log10(NT50) values, and the y axis is defined as the fraction of viruses, or virus-rabbit serum combinations, whose log10(NT50) values strictly exceed x.
Several aspects of the M-B curve method were modified for the present analysis with B-P curves. Discontinuous step functions were replaced with continuous curves obtained by linear interpolation of the (x,y) data points. Another modification of the M-B method consists of replacing numerical integration of curves for finding the vaccine score S by a simple intuitive formula for the area under the B-P curve. S, the area under the curve, can be based on the formula S = F × A, where F is the fraction of all serum/virus combinations tested showing positive neutralization and where A is the numerical average of the log10 of the NT50 [or of the average log10(NT50) values] of all serum/virus combinations showing positive neutralization, i.e., having NT50 values. It is straightforward to derive this formula using integration by parts of the B-P curve. Lastly, it is immaterial whether the score is based on the B-P curve of pooled data from a group of rabbits receiving the same vaccine or whether the scores from B-P curves of the individual rabbits in the group are averaged; the vaccine score, S, is the same. Pairwise comparison of vaccines was based on comparing two groups of rabbit scores by the two-sample t test.
Several factors contributed to the choice of the V3 sequences inserted into the CTB scaffold. The clade B sequence had been used in previous scaffolds (66, 72) and was used here to compare the results in the present study to previous studies. The V3 consensus C sequence was chosen because it had shown good activity in the context of another scaffold in a prior study (72) and it was a close match to the clade C DNA prime (Table 1). Next, the V3 consensus H sequence was chosen for two interrelated reasons: (i) the chimeric pseudovirus carrying the consensus H V3 sequence had a different pattern of neutralization by anti-V3 monoclonal antibodies (MAbs) (reference 14 and unpublished data), and (ii) structural analysis suggested that the crown of V3H had a preference for a different three-dimensional (3D) structural shape than the V3B crown, namely, an α-helix in its N-terminal strand (T. Cardozo, unpublished data). These findings suggested that the crown of V3H might present different epitopes compared to V3B and V3C. The final two V3 inserts, V32219 and V33074, were designed by making appropriate mutations in the consensus C V3 crown in the critical residues of the “signature motifs.” Signature motifs were derived previously based on the 3D shapes of epitopes in V3 that are recognized by human MAbs which are able to neutralize diverse strains of viruses from multiple clades (3, 14, 64). Briefly, by studying the crystal structures of MAb/epitope complexes (13, 36, 60, 61), identifying the key V3 residues that are recognized by the MAb, and determining the stringency with which these residues are required for neutralization by that MAb, a signature motif for the epitope recognized by that MAb can be defined. For example, the signature motif for the relatively clade B-restricted anti-V3 MAb 447-52D is P16R18, where proline is required at position 16 in the V3 loop and arginine is required at position 18 (14, 64). The only V3 insert that bears this motif of the five used in this study was the V3B insert (Table 1). The signature motifs recognized by two anti-V3 MAbs with cross-clade neutralizing activity (34) are R9K10[I,V]12[Y,F]21 for MAb 2219 and [I,L,M]14P16[F,W]20 for MAb 3074 (14, 64). Thus, the V32219 insert lacks the [I,L,M]14 portion of the 3074 motif, and the V33074 insert lacks the R9K10 portion of the 2219 motif (Table 1).
Fourteen groups of rabbits were immunized according to the protocol described in Materials and Methods. All animals received the same prime: three doses of codon-optimized clade C gp120 DNA (C1.opt) from clade C strain 92BR025.9. Subsequently, rabbits in each group were boosted twice with CTBWT, a negative-control immunogen which contains no V3 insert; with one of the five single V3-CTB immunogens; or with various double combinations of V3-CTB immunogens. The responses of individual rabbits in each of the immunization groups against clade B pseudovirus Bx08 are shown in Fig. 1. These data show the general trend found with the larger panels of pseudoviruses and primary isolates: the poorest single V3 boosting immunogen was V3H-CTB (H), while the strongest single boosting immunogen was V3C-CTB (C). In general, the double boosts gave stronger responses, although some combinations were not better than individual V3-CTB immunogens.
The reciprocals of the geometric mean titers for 50% neutralization (GMT50) against 9 tier 1 pseudoviruses (clades AG, B, and C) and 9 pseudoviruses from the clade B and C standard panels are shown in Table 2. Five additional pseudoviruses were tested which were not neutralized (data not shown). The GMT50 values for each immunization group were calculated from the individual 50% neutralizing titers (NT50) assessed with the serum of each rabbit in each group. Sera from all animals drawn prior to immunization gave NT50 values of <1:10 (data not shown). The pseudoviruses are ordered in Table 2 based on cluster analysis which used neutralization data from a multiclade panel of plasma to assess overall sensitivity (56); the most neutralization-sensitive pseudovirus (MW965.26) is on the far left, while the most resistant pseudovirus (RHPA429.7) is on the far right. An additional five pseudoviruses from the standard clade B and C panels were tested and were not neutralized by any of the immune sera (REJO4541.67, TRO.11, SC422661.8, ZM214M.PL15, and ZM197M.PB7); these data are not included in Table 2.
The sporadic presence of very low levels of NAbs in the sera of rabbits boosted with CTBWT is due to the gp120 DNA prime, as CTBWT contains no HIV epitopes. The serum NAb levels of rabbits in this group were low, with GMT50 values generally <1:10, reaching a maximum of 1:47 only against neutralization-sensitive pseudoviruses. In contrast, animals boosted with either one or double combinations of V3-CTB immunogens mounted stronger and more potent NAb responses (see statistical analyses below). Indeed, the data show that the sera of animals receiving V3-CTB boosts displayed cross-clade NAbs and could neutralize sensitive and resistant pseudoviruses as defined by the activity of pools of clade-specific HIV-1+ plasma from individuals chronically infected with either clade A, B, or C (South Africa and Tanzania), CRF01_AE, or CRF02_AG (56). The data indicate that when individual V3-CTB immunogens were used for boosting, relatively weak responses were induced; the exception to this was V3C-CTB, which induced NAbs generally as potent as those induced with the best double V3-CTB boosting combinations (see statistical analyses below). Similarly, when individual V3-CTB immunogens were used for boosting, the response was generally more narrow than when double V3-CTB boosting combinations were used, although once again, the response to V3C-CTB was as broad as those induced with the best double combinations (see below for statistical analyses). Thus, boosting with V3C-CTB alone resulted in immune sera that neutralized 13/23 pseudoviruses. For most of these 13 pseudoviruses, the sera from 4 or all 5 rabbits in the group contained demonstrable NAbs as measured by NT50 values. As noted, the response induced by V3C-CTB was approximately comparable to the best double combinations of V3-CTB immunogens such as C + 2219 and C + 3074. (See below for statistical analyses of the relative breadth and potency of the Ab responses elicited by each boosting protocol.)
Titration curves showing neutralizing activity against six pseudoviruses are shown for sera from each of the animals in the groups boosted with C or with C + 3074 (Fig. 2). Potent neutralization of SF162.LS and BaL.26, two neutralization-sensitive pseudoviruses, was detected in the sera of all 10 rabbits immunized in these two groups, with NT50 values ranging from 1:2,092 to 1:7,519 versus SF162.LS and from 1:141 to 1:812 versus BaL.26. Figure 2 also shows titration curves for neutralization of four pseudoviruses included in the clade B and C standard pseudovirus panels (42, 43): 6535.3 (clade B) and ZM109F, ZM135M, and CAP210 (clade C). These titration curves document the vaccine-induced activity against heterologous viruses of two disparate clades and illustrate demonstrable activity even when 50% neutralization is not achieved with sera diluted 1:10. Low levels of specific neutralizing activity were detected against SF162.LS (NT50 values of 1:31 to 1:70) in the sera of rabbits receiving the CTBWT boost, but no neutralizing activity was detected against the other five pseudoviruses (data not shown); this illustrates the critical role of the protein boost after DNA priming, as previously documented (73).
To compare the 14 immunization regimens in terms of both the breadth and potency of the NAb responses, two methods were used. First, breadth-potency (B-P) curves were generated (see Materials and Methods) to simultaneously compare the integrated breadth and potency of NAbs induced by different regimens. Thus, potency is shown on the x axis as the log10 of the NT50, and for each log10(NT50) value, x, breadth is shown on the y axis as the fraction of virus/serum pairs demonstrating positive NAb responses larger than 10x, i.e., log10(NT50) > x. Such a depiction translates the tabular data from Table 2 into graphic representations (Fig. 3 A and B). Figure 3A shows the B-P curves for six of the vaccine regimens in which a single boosting immunogen was used; the curves were generated from the NT50 data with the sera of each animal in each group tested against all 23 pseudoviruses (18 of which are listed in Table 2). The poorest response, as expected, was generated with CTBWT. Interestingly, V3C-CTB outperforms all of the other individual V3-CTB immunogens as shown in Fig. 3A and by statistical analyses (see below and Fig. 3C and and3D).3D). The B-P curves for the double combinations of boosting immunogens are shown in Fig. 3B. Here, the B + 2219 combination performs most poorly and the C + 3074 and C + 2219 combinations perform better than all other double combinations. Indeed, the three best boosting regimens were C, C + 2219, and C + 3074, and there was no statistical difference between these three groups (Fig. 3D).
We then derived numerical values, vaccine scores, for the relative breadth and potency of each vaccine regimen. The vaccine score, S, is the area under each B-P curve, calculated as described in Materials and Methods. These scores can be used to rank the NAb responses to different boosts. This ranking shows that the poorest response (rank 1) is elicited, as expected, with the boost lacking any HIV epitope, CTBWT (S = 0.392, Fig. 3C). Boosting with the single V3H-CTB is better (rank 3; S = 0.780). Although the response is weak, there is a statistical difference (P < 0.05) between these two groups (Fig. 3D). The single boosting immunogens ranked poorly (rank ≤7 of 14), with the exception of the V3C-CTB boost (rank 13). In contrast, the double-combination boosts generally ranked higher: six of the eight double-boost combinations ranked ≥8. As noted above and in Fig. 3C, the three boosting regimens that ranked highest were C + 2219 (rank 12), C (rank 13), and C + 3074 (rank 14).
The data presented above were generated in assays performed with coded sera in a standard assay by a reference laboratory. They demonstrate that cross-clade NAbs against tier 1A, tier 1B, and tier 2 pseudoviruses from standard panels were induced reproducibly in rabbits using several DNA prime/protein boost regimens. However, since no single assay is capable of detecting the entire spectrum of neutralizing activities, and it is not known which in vitro assay correlates with in vivo protection (22), we assessed the neutralizing activity of the rabbit sera with a second assay in which neutralization of 15 PBMC-grown primary isolates from clades A, AG, B, and C was tested. This assay is far more stringent, i.e., less sensitive, than the neutralization assay of pseudoviruses in the same TZM.bl target cells; for example, NT50 values of HIV+ human sera against primary isolates are often more than an order of magnitude less potent than those against the homologous Env-pseudotyped viruses grown in 293T cells (42). Therefore, the rabbit immune sera generated with the 14 vaccine regimens were tested individually against 15 primary isolates grown in PBMCs which were chosen to represent a range of sensitivities with anti-V3 MAbs. The anti-V3 MAb pool included anti-V3 MAbs 2191, 2219, 2558, 3074, and 3869 (27–30), which had previously been shown to display cross-clade neutralizing activity (34). Using this MAb pool at a final total concentration of 50 μg/ml, 7 of the 15 viruses could be neutralized strongly (99% neutralization) to weakly (34% neutralization). In contrast, 8/15 viruses were not neutralized by the anti-V3 MAb pool (<17% neutralization) (bottom row, Table 3).
The GMT50 values against each of the 15 viruses generated from titration curves with sera from each animal in the 14 rabbit groups are shown in Table 3. As expected, viruses that were insensitive to anti-V3 MAbs were not neutralized by any of the rabbit sera. In contrast, the viruses neutralized by the anti-V3 MAbs, even those which were relatively resistant to neutralization by the anti-V3 MAb pool, e.g., clade B CA5 isolate and clade A VI313 isolate, were neutralized by the sera of most of the groups of immunized rabbits. NT50 values in the thousands were achieved against the sensitive viruses, and NT50 values reached into the hundreds with the viruses that were more resistant to V3 Abs. The B-P curves and vaccine scores showed that the weakest and most narrow response was again elicited with the CTBWT boost (Fig. 4 A and B). These data also show that five of the eight double-combination boosts scored better than V3C-CTB (Fig. 4C), in contrast with the data derived from neutralization of pseudoviruses (Fig. 3C). Statistically, the C + 3074 boosting combination was better than C alone (Fig. 4D). Notably, the C + 3074 combination ranked best in both pseudovirus and primary isolate neutralization assays (Fig. 3C and and44C).
Finally, experiments were performed to determine the longevity of the immune response. For these studies, sera were drawn at seven time points during and after immunization with clade C gp120 DNA and boosting with V3B-CTB. As shown in Fig. 5, when sera at a 1:20 dilution were assessed for neutralization against the clade B PBMC-grown primary isolate virus Bx08, peak titers were achieved 2 weeks after the second boost (week 16). Titers were maintained at essentially the same peak levels 11 weeks after the second boost (week 25). While falling to somewhat lower levels at weeks 58 to 76, neutralizing activity was maintained at detectable levels for more than 1 year after the last boost. The data are presented as Δ% neutralization, i.e., activity relative to that in preimmune sera from the same rabbit.
The ability of Abs to prevent HIV infection has been demonstrated repeatedly in many animal models, showing that passive immunization with monoclonal Abs can provide sterilizing immunity (4, 20, 21, 24, 45, 69). While in humans the protective role of Abs is not as firmly established, existing data again suggest that Abs, whether passively administered or induced by candidate vaccines, play an important role in preventing HIV infection (23, 26, 40). In advanced human vaccine trials, the immunogens used in all cases included various forms of gp120 proteins which elicited high titers of binding Abs but relatively low titers of Abs (25, 26, 52) that were able to neutralize only a small subset of neutralization-sensitive viruses.
Since a complex glycoprotein like gp120 induces a broad spectrum of Ab specificities, only a minority of the Abs elicited would be expected to have protective activity. Focusing the immune response on selected epitopes of the Env glycoproteins, rather than using whole Env, has the potential advantage of inducing Abs that specifically target regions known to elicit protective Abs, thus resulting in antisera with higher titers of functional Abs. The use of rationally designed epitope-scaffold immunogens that focus the Ab response on protective epitopes also has the potential advantage of inducing Abs of increased cross-reactivity between viral strains. The work described above supports these concepts: memory cells were induced by priming with gp120 DNA; subsequently, the memory B cells that are specific for a single region of gp120 were stimulated with V3-scaffold immunogens. As shown previously, this resulted in Abs >80% of which were specific for V3 (72). Thus, Ab responses were focused on a single neutralizing epitope, and cross-clade NAbs were induced. Notably, while all tier 1 pseudoviruses tested were neutralized (Table 2), 43% (6/14) of the pseudoviruses tested from the tier 2 clade B and C standard panels were neutralized. Table 2 shows the data for 9 of these 14 tier 2 pseudoviruses; the remaining five pseudoviruses that were tested were not neutralized (data not shown). Similarly, 47% (7/15) of primary isolates were neutralized (Table 3). These data contradict the oft-stated view that anti-V3 Abs can neutralize only T-cell line-adapted and tier 1 viruses (33). As expected, the NT50 values against the tier 2 pseudoviruses were considerably lower than those for tier 1A pseudoviruses but were in the same range as the titers against many tier 1B pseudoviruses (Table 2).
This elicitation of cross-clade NAbs with V3-scaffold immunogens stands in marked contrast to many unsuccessful attempts to induce cross-clade NAbs with epitope-scaffold immunogens designed to focus the immune response on other neutralizing epitopes of gp120 and gp41 (11, 15, 16, 19, 31, 48, 51, 53, 57, 58). Thus, when we and others have used conformationally correct V3 immunogens, designed on the basis of structural data, NAbs were induced (47, 72, 73). The success of inducing NAbs by focusing the immune response on conformational V3 epitopes may be attributable to many causes: V3 is a highly immunogenic region, it consists of a cluster of overlapping epitopes in a linear portion of the molecule which is structurally conserved despite its sequence variability (36, 71), and the anti-V3 immune response is enhanced by priming with an optimized gp120 DNA known to effectively prime for V3 and other Abs (66, 72, 73), eliciting V3 responses that are broader and more potent than those elicited with V3 immunogens alone (47, 73).
While the Ab response induced with the DNA prime/V3-scaffold protein boost regimens used here resulted in cross-clade neutralizing activity and succeeded in targeting epitopes shared by diverse viral envelopes, there were many viruses that were not neutralized. While the field currently understands neither the physicochemistry nor the biology underlying the relative sensitivity or resistance of a given HIV strain to Ab-mediated neutralization, in the case of V3, we know that Abs may fail to neutralize a virus even if the cognate epitope is present, a phenomenon attributable to conformational masking (1, 39); this precludes a clear correlation between the boosting immunogen and the V3 sequence of the viruses neutralized. Nonetheless, there is an indication that the V32219- and V33074-CTB boosting immunogens induced polyclonal Abs with specificities and activities similar to those of the MAbs that gave rise to the epitope-scaffold immunogen designs (T. Cardozo et al., submitted for publication). Another indication that the rational design of the boosting immunogens and the choice of which to combine is a productive path comes from several accurate predictions of which boosting combinations would give the weak responses. For example, we predicted that combining V3B- + V33074-CTBs as well as V3B- + V32219-CTBs would not elicit a better Ab response than that of any of these three immunogens used individually; as shown in Fig. 1 and Tables Tables22 and and3,3, this was correct. However, we did not predict, and cannot yet explain why, for example, V3C elicited such a strong response.
The V3-scaffold immunogens used here were rationally designed using a large body of structural data derived from crystals of V3 peptides complexed to V3-specific MAbs (13, 18, 36, 60, 61), bioinformatics analyses, and molecular modeling studies (3, 14, 36, 64, 71). This resulted in V3 epitopes which captured conserved V3 structures shared between the various viral subtypes and which were spliced into a structurally compatible region of a protein scaffold, cholera toxin B (66). The present study expands upon an initial study with a single form of V3-CTB (66), by testing four additional V3-CTB immunogens, individually and in combination, and showing that each of the five individual V3-CTB immunogens shows differential reactivity. Of the five V3-CTB immunogens (Table 1), when tested individually, the V3C-CTB induced the strongest and broadest response. Using various double combinations of these V3 scaffolds, generally, the combinations were more effective than the individual immunogens. Notably, however, all of the groups receiving V3-CTB boosts mounted cross-clade NAb responses (Tables (Tables22 and and33).
The potency and breadth of the NAbs elicited are comparable to the best results published to date by others using various forms of Env. For example, as shown in Fig. 2, potent NAbs against SF162.LS and BaL.26 were elicited by boosting with either V3C-CTB or V3C-CTB and V33074-CTB: the NT50 values in the sera of all 10 rabbits in these two groups ranged from 1:2,092 to 1:7,519 versus SF162.LS and from 1:141 to 1:812 versus BaL.26. By comparison, the mean NT50 values against SF162.LS and BaL.26 were reported to be 1:911 and 1:99, respectively, in guinea pigs immunized with Env trimers (49) and 1:1,431 and 1:48, respectively, in humans immunized with two gp120 proteins derived from subtype B strains MN and GNE8 (25); levels comparable to these were also reported in humans immunized with other rgp120 Env proteins (59) and in rhesus macaques primed with alphavirus replicon particles and boosted with Env (7). These data suggest that vaccines that focus the immune response on V3 are as effective as currently available whole-Env vaccines. Indeed, Nkolola et al. (49) showed that neutralizing activity to three of three pseudoviruses was partially or entirely absorbed out by V3 peptides but not by V1 or V2 peptides, and they concluded that gp140 trimers elicited NAbs that are directed in part against conserved elements in the V3 loop. The failure of other vaccine regimens using various Env proteins to induce cross-clade NAbs may, in large part, be due to the choice of the Env immunogen used, as it has been shown that some Envs are far more effective as immunogens than others (59, 67).
The longevity of the Ab responses to HIV induced with Env has been a continuing problem, with titers falling to background levels within ~10 to 20 weeks after each immunizing dose (8–10, 32). The data presented above (Fig. 5) demonstrate that the use of epitope-scaffold boosting immunogens, rather than various forms of Env, induces NAb responses which are demonstrable more than 1 year after the last boost. Such a long-lived Ab response will be a needed hallmark of a clinically relevant prophylactic vaccine.
In general, the more potent the response demonstrated against pseudoviruses, the greater was its breadth. Thus, vaccine regimens that ranked lowest (ranks 1 to 7) neutralized 6 to 11 pseudoviruses, while the regimens that ranked highest (ranks 8 to 14) neutralized 11 to 13 pseudoviruses (Table 2 and Fig. 3C). Though less marked, the trend was similar for primary isolates: most of the sera from the groups receiving the low-ranking regimens neutralized <7 viruses while sera from most of the groups receiving the high-ranking regimens neutralized >7 viruses (Table 3 and Fig. 4C). These data suggest that the use of adjuvants, formulations, and/or immunization protocols that increase titers will also increase breadth. Such improvements may also increase affinity maturation, isotype switch, and FcR-mediated functions, all of which would enhance the efficacy of this and other vaccines. Thus, the titers and breadth of cross-clade NAbs achieved with the DNA prime/V3-scaffold boosting immunogens reported here should be considered a baseline that can be substantially improved.
As increasing numbers of immunization regimens are being tested in vivo and analyzed by standardized neutralization assays (49, 56, 59), it becomes important to develop methods to assess both the breadth and potency of the elicited response. In this study alone, we tested 14 different immunization regimens. To compare simultaneously the breadth and potency of each, we have established methods to graphically compare the responses (Fig. 3A and B and Fig. 4A and B), to assign vaccine scores that incorporate both of these variables (Fig. 3C and and4C),4C), and thereby to rank the regimens from worst to best. New approaches, described in Materials and Methods, were also developed to establish whether statistical differences exist between vaccine regimens, a task that will increasingly be needed to identify the most efficacious of the candidate vaccines.
Comparing the ranking of the 14 vaccine regimens (Fig. 3C and and4C),4C), it is clear that the efficacy of one vaccine regimen relative to the next is at least partially dependent upon the panel of viruses used to measure efficacy. Thus, for example, the V3C-CTB immunogen was ranked as one of the most potent immunogens when tested against the pseudovirus panel (rank 13 of 14); in contrast, it ranked 9th out of 14 when tested against the primary isolate panel. Regardless of the panels used for testing, however, the control boosting immunogen, CTBWT, ranked worst (1/14); V3C-CTB was the best of the single V3-scaffold immunogens tested; and V3C- + V33074-CTB ranked highest (14/14).
In this study, we examined the efficacy of vaccine-induced Abs directed against a single region of the virus Env, the V3 loop, and showed that cross-clade NAbs were induced and that pseudoviruses that were not previously shown to be neutralized by various anti-V3 monoclonal Abs could be neutralized by the polyclonal immune sera (34, 42, 56). Nonetheless, only about half of the viruses and pseudoviruses tested were neutralized by the sera induced by the best of the immunizing regimens. While even a modestly efficacious first-generation vaccine could have a profound effect on the AIDS epidemic (62), to achieve more complete coverage, Abs to additional epitopes will need to be induced. Existing data suggest that Abs to many Env epitopes, including V3, will work additively to provide that breadth (55, 68, 74). Focusing on a single epitope, as we did in this study, is a first step in designing a panel of epitope-scaffold immunogens that will specifically target protective epitopes. This may prove to be an effective vaccine strategy and one with advantages over the use of various forms of Env, a molecule which has evolved to protect the virus by poorly inducing NAbs and by shielding the epitopes that are most critical for the infectivity of the virus.
This work was supported by grants from the NIH (R01 AI36085 [S.Z.-P.], R01 AI065250 [S.W.], P01 AI082274 [S.L.], and U19 AI082676 [S.L.]), the Bill and Melinda Gates Foundation (38631 [S.Z.-P.] and 38619 [M.S.S.]), and research funds from the Department of Veterans Affairs.
We acknowledge the contribution of Youyi Fong of the Gates Foundation-supported Vaccine Immunology Statistical Center for his assistance in the statistical analyses of the data.
Published ahead of print on 27 July 2011.