HIV-1 gp120 and Modified Vaccinia Virus Ankara (MVA) gp140 Boost Immunogens Increase Immunogenicity of a DNA/MVA HIV-1 Vaccine

ABSTRACT An important goal of human immunodeficiency virus (HIV) vaccine design is identification of strategies that elicit effective antiviral humoral immunity. One novel approach comprises priming with DNA and boosting with modified vaccinia virus Ankara (MVA) expressing HIV-1 Env on virus-like particles. In this study, we evaluated whether the addition of a gp120 protein in alum or MVA-expressed secreted gp140 (MVAgp140) could improve immunogenicity of a DNA prime-MVA boost vaccine. Five rhesus macaques per group received two DNA primes at weeks 0 and 8 followed by three MVA boosts (with or without additional protein or MVAgp140) at weeks 18, 26, and 40. Both boost immunogens enhanced the breadth of HIV-1 gp120 and V1V2 responses, antibody-dependent cellular cytotoxicity (ADCC), and low-titer tier 1B and tier 2 neutralizing antibody responses. However, there were differences in antibody kinetics, linear epitope specificity, and CD4 T cell responses between the groups. The gp120 protein boost elicited earlier and higher peak responses, whereas the MVAgp140 boost resulted in improved antibody durability and comparable peak responses after the final immunization. Linear V3 specific IgG responses were particularly enhanced by the gp120 boost, whereas the MVAgp140 boost also enhanced responses to linear C5 and C2.2 epitopes. Interestingly, gp120, but not the MVAgp140 boost, increased peak CD4+ T cell responses. Thus, both gp120 and MVAgp140 can augment potential protection of a DNA/MVA vaccine by enhancing gp120 and V1/V2 antibody responses, whereas potential protection by gp120, but not MVAgp140 boosts, may be further impacted by increased CD4+ T cell responses. IMPORTANCE Prior immune correlate analyses with humans and nonhuman primates revealed the importance of antibody responses in preventing HIV-1 infection. A DNA prime-modified vaccinia virus Ankara (MVA) boost vaccine has proven to be potent in eliciting antibody responses. Here we explore the ability of boosts with recombinant gp120 protein or MVA-expressed gp140 to enhance antibody responses elicited by the GOVX-B11 DNA prime-MVA boost vaccine. We found that both types of immunogen boosts enhanced potentially protective antibody responses, whereas the gp120 protein boosts also increased CD4+ T cell responses. Our data provide important information for HIV vaccine designs that aim for effective and balanced humoral and T cell responses.

KEYWORDS DNA/MVA, HIV vaccine, T cell response, antibody response, boost immunogen, nonhuman primate G OVX-B11, a subtype B DNA prime-modified vaccinia virus Ankara (MVA) boost vaccine, elicits higher response rates and titers of antibodies to the gp41 transmembrane than the gp120 receptor binding subunit of Env (1). Both components of GOVX-B11 express virus-like particles (VLPs) displaying membrane-bound Env. The current study was undertaken to test the ability of gp120 protein or MVA-expressed secreted gp140 (MVAgp140; the ectodomain of Env) to boost the ability of GOVX-B11 to elicit gp120 antibodies. Increasing the titers of antibodies to gp120 and, in particular, to the V1V2 region of gp120 is considered desirable because gp120 is the major target for neutralizing antibodies for HIV (2) and because binding antibodies for V1V2 were a correlate for reduced HIV-1 risk in the partially efficacious RV144 vaccine trial in Thailand (3,4).
The GOVX-B11 vaccine expresses the native ADA (clade B) gp160 Env, a clade B CCR-5-tropic Env from a chronic infection (5). For the gp120 boost, B.63521Δ11mutC, a clade B transmitted founder (T/F) Env, was chosen because of its favorable antigenicity and immunogenicity (6) and its projected availability as a current good manufacturing practices (cGMP) product. B.63521Δ11mutC has an 11-amino-acid (aa) truncation at the N terminus of gp120 that enhances antigenicity of the V2 and C1 epitopes that are targets for antibody-dependent cell-mediated cytotoxicity (ADCC) (7,8) and prevents gp120 dimerization (8), thus optimizing manufacture. B.63521Δ11mutC also has a mutated V3 loop to prevent clipping of gp120 when expressed in CHO cells for manufacturing (S. M. Alam, H. X. Liao, and B. F. Haynes, unpublished data). Alum (Alhydrogel) was chosen as the adjuvant for the B.63521Δ11mutC protein boost because of its use in the RV144 HIV-1 vaccine phase 2b trial that demonstrated an estimated 31.2% vaccine efficacy (9). ADA gp140 was used for the MVA-expressed gp140 boost because this gp140 shows unusual stability as a secreted gp140 (B. Moss, personal communication).
The regimen for vaccination was based on the regimen used in RV144 and on our clinical experience with DNA priming and MVA boosting (1,(9)(10)(11). The RV144 regimen codelivered gp120 protein boosts with the ALVAC canarypox vaccine used for both priming and boosting. ALVAC was delivered to one arm and gp120 protein to the opposite arm. In this study, we delivered the gp120 protein or MVAgp140 boosts to the contralateral arm from the MVA62B boosts and used our most successful clinical regimen for immunizations: priming with DNA at months 0 and 2 and boosting with MVA at months 4, 6, and 10 (11).
Following the 3rd boost, the groups with gp120 and MVAgp140 in boosts had comparable increases in gp120 and V1V2 binding antibody responses, whereas the plus-gp120 group had enhanced CD4 ϩ T cells over the MVA-only group. Given the comparable levels of elicited antibodies, and the potential for vaccine-elicited CD4 ϩ T cells both to protect from infection (12) and to serve as targets for infection (13,14), MVAgp140 and the more generally used gp120 boost immunogens merit further evaluation for advancement to clinical studies.

RESULTS
Serum binding antibody response. Binding antibody responses against a consensus group M gp120 (Con6 gp120) and MN gp41 were measured using enzyme-linked immunosorbent assays (ELISAs) and longitudinal serum samples. gp120-and gp41specific IgG responses developed 2 weeks after the first MVA boost (week 20), peaked after the second boost (week 28) declined with time, and were boosted again after the 3rd boost ( Fig. 1A and B). At 2 weeks after the 2nd boost (week 28), the group treated with MVA plus gp120 (MVAϩgp120 group) elicited the highest levels of Con6 gp120specific binding response, with a median titer 10-fold higher than that of the MVA-only group and 2.5-fold higher than that of the MVAϩMVAgp140 group (Fig. 1A). Following the 3rd boost, the MVAϩgp120 and MVAϩMVAgp140 groups had similar median magnitudes of antibody responses, whereas the titer elicited by MVA only was 4-fold lower. However, the differences among groups were not significant (Fig. 1A). Binding magnitudes for MN gp41 were overall comparable among groups at all time points tested (Fig. 1B).
Next, we measured IgG responses to the vaccine-matched B.63521 gp120 protein. The MVAϩgp120 group showed significantly higher responses following the 2nd boost, postcontraction at the time of the 3rd boost, and after the 3rd boost than did both the MVA-only and MVAϩMVAgp140 groups (Fig. 1C). The differences between the MVAϩgp120 group and the MVAϩMVAgp140 group were 13.8-fold and 5.2-fold at weeks 28 and 42, respectively (Fig. 1D).
Breadth of binding antibody response. Following the 3rd boost, the MVAϩgp120 and MVAϩMVAgp140 groups had similar magnitudes and breadths of binding antibodies for reference panels of gp120 and gp140 antigens ( Fig. 1E and F and Table 1). However, transient differences in magnitudes and breadths were seen after the 2nd boost for gp120 and gp140, with the group median binding mean fluorescence intensity (MFI) of the MVAϩgp120 group being significantly higher for the gp120 antigen panel (P values that are not controlled for multiple comparison [raw_p] ϭ 0.03 for each) ( Fig. 1E and Tables 1 and 2) and trending higher for the gp140 panel (raw_p Ͼ 0.05) ( Fig. 1F and Tables 1 and 2) than the other two groups.
In contrast to the gp120 and gp140 responses, levels of binding to the V1V2 panel were similar for sera from the MVAϩgp120 and MVAϩMVAgp140 groups after both the 2nd and 3rd boosts, with both being approximately 2-fold higher than that for the MVA-only group ( Fig. 1G and Table 1).
The durability and boostability of binding antibody responses were evaluated as the fold decline of the antibody response over the 12 weeks following the 2nd boost (week 28/week 40) and the fold increase of the antibody response following the 3rd boost over that following the 2nd boost (week 42/week 40) ( Fig. 2 and Table 1). The MVAϩgp120-boosted group underwent the largest declines ( Fig. 2A and Table 1). By the 3rd boost, the MVAϩMVAgp140 group showed the strongest boost for all 3 antigen panels (raw_p values of 0.032 to 0.008 against the MVAϩgp120 group) ( Fig. 2B and Table 2).
Specificity of binding antibodies for linear epitopes. IgG from the vaccinated animals targeted linear epitopes in the C1, V2, C2, V3, C4, and C5 regions of gp120 (Fig.  3A and C and Table 3). In gp41, antibody responses were targeted to the C terminus of the N-heptad repeat, the N terminus of the gp41 immunodominant region (NHR_ID), the gp41 immunodominant region (gp41_ID), 3S (an epitope C terminal to gp41_ID that contains 3 serine residues), C-heptad repeat (CHR), and KE (Kennedy epitope) (  Table 3).
The contribution of each epitope to the total Env linear epitope binding after the 2nd and 3rd boosts was evaluated by determining the percentage of the binding intensity for each epitope within the total Env linear epitope binding intensity. Among the 3 groups, the MVAϩgp120 group was the most focused on V3, with 81.1% and 70.8% of total linear epitope binding (group median) targeting V3 at weeks 28 and 42, respectively, compared to 34.2% to 49.9% for other groups (Fig. 3A). The MVA-only group showed a trend for higher proportions of binding responses targeting KE, with 23.2% and 17.1% of total linear epitope binding intensities targeting KE at weeks 28 and 42, respectively, compared to 0.8% to 6.1% for other groups (Fig. 3A). The  MVAϩMVAgp140 group showed a trend for higher proportions of binding responses targeting C5 and C2 (epitope C2.2) than those of the other groups (Fig. 3A). Significant differences (raw_p Ͻ 0.05, exact Wilcoxon rank sum test) were observed among groups for binding to epitopes in C2, C4, and C5 following either the 2nd (week 28) or 3rd (week 42) boost (Fig. 3C). Highly significant differences (raw P Ͻ 0.01) were found after the 3rd boost for C2.2 (MVAϩMVAgp140 and MVAϩgp120 Ͼ MVA only) and after the 2nd boost for C5.1 (MVAϩMVAgp140 Ͼ MVA only and MVAϩgp120) and C5.2 (MVAϩMVAgp140 Ͼ MVA only) ( Fig. 3C and Table 2). Due to the large number of samples that were saturated for binding to V3.1, especially for the MVAϩgp120 group, differences among groups for V3.1 binding were not evaluated.
Binding responses to peptides representing different clades/strains were generally similar among vaccine groups and at the two tested time points (following the 2nd and 3rd boosts) (Fig. 3B). All groups bound most strongly to clade B sequences. The MVAϩgp120 group had slightly higher proportions of binding to clade C sequences than the other 2 groups, with binding to clade C sequences (including consensus C, C.ZM651, C.TV1, and C.1086) accounting for 25.8% and 29.9% of total linear binding at weeks 28 and 42, respectively, compared to 17.2% to 25.3% for other groups. The MVAϩMVAgp140 groups showed the most "balanced" coverage of different clades, with 12 strains (out of 13 total) each contributing 3% or more to total Env binding for both time points, compared to 7 strains and 9 strains for MVA-only and MVAϩgp120 groups, respectively (Fig. 3B). In particular, binding for clade AE strains (including consensus AE, AE.A244, and AE.TH023) contributed 9.1% and 11.6% to total binding for the MVAϩMVAgp140 group at weeks 28 and 42, compared to 4.4% to 5.5% for the other groups.
Avidity of antibody responses. The avidity of the antibody elicited for three linear epitopes-clade B V3, clade C V3, and gp41-ID-was measured by an avidity index assay (11,15). For all animals/epitopes where the avidity indices were quantifiable, avidity indices were higher after the 3rd than after the 2nd boost (Fig. 4).
Neutralizing antibody responses. Sera from the 3 groups neutralized tier 1A, tier 1B, and some tier 2 viruses. Animals from all vaccine groups developed neutralizing antibodies against tier 1 MN.3 (clade B) and MW965.26 (clade C) viruses, with 50% inhibitory dose (ID 50 ) titers up to 5,274 ( Fig. 5A to D). The MVAϩgp120 group showed the highest titers after both the 2nd and 3rd boosts, and the difference was significant between MVAϩgp120 and MVA-only groups after the 3rd boost for MW965.26 (raw_p ϭ 0.03, exact Wilcoxon rank sum test; Fig. 5D and Table 2).  Table 2). All differences are insignificant once adjusted for multiple comparison ( Table 2).

FIG 3
Proportions of total linear binding to each identified epitope (A) or to each strain included in the epitope mapping peptide library (B) and magnitude of binding to each epitope identified (C). Each sector in panel A represents a group median magnitude to an epitope, which was calculated as the highest binding to a single peptide within each epitope region as defined in Table 3. Each sector in panel B represents a group median total linear epitope binding for a strain, which was calculated as the sum of binding intensities to all linear peptides for that strain. Values in panel C are the highest binding to a single peptide within each epitope region. Bars represent group medians and whiskers represent interquartile ranges. *, raw_p Ͻ 0.05; **, raw_p Ͻ 0.01 (Table 2). gp120 and MVAgp140 Boosts of a DNA/MVA HIV Vaccine Journal of Virology A subset of vaccinated animals developed neutralizing antibodies against tier 1B viruses, with titers up to 122 at week 28 ( Fig. 5E and F). The MVAϩgp120 group trended higher for magnitudes of neutralizing responses against both clade C and non-clade C tier 1B viruses than the other 2 groups. Except for Bal, the response rates and titers were lower for the tier 1B viruses after the 3rd than after the 2nd MVA boost ( Fig. 5G and H). This contrasted with the neutralizing activity for the tier 1A viruses, which remained at similar positivity rates and titers at weeks 28 and 42 for all 3 groups. This decline in activity was particularly evident for the 2 tested clade C tier 1B viruses. After the 2nd MVA boost, response rates to both clade C tier 1B viruses were 60% or greater and several titers were above 1:80. After the 3rd MVA boost, the highest response rate was 20% and titers were below, or close to, background (1:20). The MVAϩMVAgp140 group maintained low-level responses slightly better than the other 2 groups (Fig. 5G and H   declined for both positivity rates and serum titers, especially for the MVA-only and MVAϩgp120 groups (Fig. 5E to H). Notably, vaccinated animals developed low levels of neutralizing antibodies (ID 50 titers up to 61) against tier 2 viruses after the 2nd boost, with positivity rates ranging  Table 2). All differences are insignificant once adjusted for multiple comparison (  Fig. 5E and F). The MVAϩgp120 group showed the highest positivity rate and titer among the 3 groups, with titers significantly higher than for the other 2 groups for isolate 25710-2.43 (raw_p Ͻ 0.05, exact Wilcoxon rank sum test [ Table 2]). Neutralization against tier 2 viruses was undetectable following the 3rd boost for all but one animal in the MVAϩMVAgp140 group ( Fig. 5G and H). ADCC responses. In assays using B.63521Δ11 gp120-coated target cells, ADCC activity was detected in 100% of animals in both the MVAϩgp120 and MVAϩMVAgp140 groups ( Fig. 5I and J). Consistent with its high titers of elicited binding response for B.63521Δ11, the MVAϩgp120 group showed higher ADCC titers and peak granzyme B activity than the other 2 groups (raw_p Ͻ 0.05 at week 28 and P Ͻ 0.01 at week 42, exact Wilcoxon rank sum test) (Fig. 5I and J and Table 2). Group median titers were 1:85,000 and 1:800 for the MVAϩgp120 and MVAϩMVAgp140 groups, respectively, after the 2nd boost, and 1:40,000 and 1:3,000, respectively, after the 3rd boost.
Correlations between antibody specificities and antiviral functions. Correlative relationships between antibody specificities and antibody function can provide insights on specificities that may mediate antiviral functions. We examined the association between linear antibody specificities and neutralization and ADCC responses measured at week 28 using data for all three groups. ID 50 neutralizing titers for tier 1A and tier 1B clade B viruses were associated with binding magnitudes to peptides in the C1, V3, and the gp41_ID regions, with Spearman r values of Ͼ0.6 and raw_p values of Ͻ0.05 ( Fig.  6 and Table 4). The highest Spearman r value was for the correlation between neutralizing activity for MN.3 and linear V3.1 binding (Spearman r ϭ 0.85; false discovery rate [FDR]-adjusted P value [FDR_p] Ͻ 0.05). Interestingly, the highly conserved gp41_ID was among epitopes with the most scores above a Spearman r of 0.5 (raw_p Ͻ 0.05) for tier 1B and tier 2 viruses. The ADCC titers measured with B.63521Δ11 gp120-coated cells associated most strongly with two C1 epitopes (Spearman r value of 0.59 and 0.48 and raw_p of 0.027 and 0.08, respectively). Further examination of scatter plots revealed that correlations between binding magnitudes and neutralization titers for most pairs with r values of Ͼ0.6 were driven by data from all three groups, with the exception of MW965, for which correlations were driven by higher neutralization titers for the MVAϩgp120 group (Fig. 6B).
T cell responses. At 1 week following the last boost, lymphocytes were tested for effects of the boosts on Env-specific CD4 ϩ T cells in blood and lymph nodes and on Env-specific CD8 ϩ T cells in blood (Fig. 7). These assays, which tested overlapping peptide pools for stimulation of gamma interferon (IFN-␥), interleukin 2 (IL-2), tumor necrosis factor alpha (TNF-␣), CD40L, IL-4, and IL-21, revealed higher proportions of elicited CD4 ϩ T cells expressing IFN-␥, IL-2, TNF-␣, and IL-4 in peripheral blood mononuclear cells (PBMC) than in lymph nodes but more similar levels of elicited CD40L-expressing cells in these two compartments ( Fig. 7A and C). The MVAϩgp120 group showed higher frequencies of IL-2-and TNF-␣-expressing CD4 ϩ T cells in PBMC than the other groups (P Ͻ 0.05, exact Wilcoxon rank sum) ( Fig. 7A and Table 2). No significant difference was observed among groups for Env-specific CD8 ϩ T cell response in PBMC (Fig. 7B).

DISCUSSION
The current study compared the magnitude, durability, specificity, breadth, and functional activity of antibodies elicited by MVAϩgp120 protein boosts, MVAϩMVAgp140 boosts, or MVA-only boosts of a DNA-primed response. Our results reveal that the addition of either gp120 protein or MVA-expressed secreted gp140 to boosts elicits higher, broader, and more functional antibody responses than the MVA-only boosts. They also reveal that protein, but not MVAgp140, boosts enhanced CD4 ϩ T cell responses. Below we discuss these results and how they might relate to protective vaccine responses.
The goal for adding boosts to the DNA/MVA vaccine was to enhance antibody responses to gp120, especially those that target the V1V2 region of gp120. Both B.63521 gp120 and MVAgp140 boosts enhanced these antibody responses. Following the 3rd and last boost, mean magnitudes of binding to the gp120 antigen panel and the V1V2 panel were 4.3-and 3.0-fold higher and 2.1-and 2.5-fold higher for the MVAϩgp120 and MVAϩMVAgp140 groups, respectively ( Table 1). The gp120 vaccine immunogen boosts strongly enhanced the binding magnitude for the matched B.63521Δ11 protein (i.e., 17.9-fold by the gp120 protein boosts but only 3.5-fold by the heterologous MVAgp140 boosts [ Fig. 1C and D]). The enhanced binding antibody   response elicited by gp120 was more V3 focused, constituting 81% and 71% of total linear gp120 binding at the 2 time points tested, compared to 34 to 50% V3-specific IgG responses for the other 2 groups. No significant enhancement of V1V2 IgG responses with these immunogens was observed in the study. Further studies with other envelope immunogens are needed to determine the impact of immunogen sequence and antigenicity on the induction of V1V2 IgG responses. In addition to the magnitude of binding antibody responses, the boost immunogens affected the durability of antibody responses and the performance of repeated boosts. The MVAϩMVAgp140 group showed a trend for a slower decline and a more effective 3rd boost of responses than for the MVAϩgp120 group. This lower contraction and higher boost with MVAgp140 suggest that MVAgp140 may elicit more durable and replenishable responses than the gp120 protein boost. Although lower levels of preboost antibody responses likely contributed to the efficiency of the 3rd MVAgp140 boost, there is precedent for late boosts, with MVA being highly effective in both preclinical and clinical settings (16)(17)(18).
In contrast to antibody responses, which were enhanced by the addition of both the gp120 protein and the MVA-expressed gp140 to boosts, only the gp120 boost increased the Env-specific CD4 ϩ T cell response over those elicited by the MVA-only boost vaccine. The MVA-specific T cell response was not measured. However, we consider it unlikely that adding 1 ϫ 10 8 50% tissue culture infective doses (TCID 50 ) of MVA gp140 to the 1 ϫ 10 8 TCID 50 of MVA62B in the MVA gp140 group significantly enhanced CD4 ϩ responses because of the very shallow dose-response curve for MVA boosts of a constant DNA prime (19). In these studies, 100-fold increases in MVA dose were associated with only a 2-to 3-fold increase in elicited CD4 ϩ T cells. In the canarypox prime/protein boost RV144 trial, the elicitation of five-function T cells expressing CD40L, IL-2, IL-4, IFN-␥, and TNF-␣ or three-function T cells expressing CD40L, IL-2, and IL-4 correlated with a reduced risk of infection (12). However, the elicitation of higher magnitudes of CD4 ϩ T cells by the DNA prime/MVAϩgp120 FIG 7 Env-specific T-cell responses. ICS was used to evaluate Env-specific CD4 ϩ (A) and CD8 ϩ (B) T cells in PBMC and Env-specific CD4 ϩ T cells in axillary lymph nodes (C). Tests were done at 1 week following the 3rd boost. *, P Ͻ 0.05 for pairwise comparison (exact Wilcoxon rank sum test). All differences are insignificant once adjusted for multiple comparison (Table 2). gp120 and MVAgp140 Boosts of a DNA/MVA HIV Vaccine Journal of Virology protein boost vaccine could also be a concern for HIV vaccine development because responding CD4 ϩ T cells have been hypothesized to enhance infections (14,20,21). Successful prevention of virus acquisition may require a balance in antibody and CD4 ϩ T cell responses in which changes in levels of virus-specific CCR5 ϩ CD4 ϩ T cells, preferential targets for HIV-1 infection (13), do not mute the effect of protective antibodies. Indeed, in 4 preclinical trials in which gp120 or gp140 protein boosts have been used with DNA/MVA prototypes of the GOVX-B11 vaccine, the protein boosts did not enhance protection but rather showed a trend for reduced prevention of virus acquisition or postinfection viremia control (22, 23; R. Amara, personal communication). In 3 of these trials, infection and/or poor control of viremia directly correlated with the induction of vaccine-specific IFN-␥ ϩ CCR5 ϩ CD4 ϩ T cells in blood (22; Amara, personal communication). Thus, the ability of the MVAgp140 boosts to enhance antibody responses without boosting CD4 ϩ T cell responses may be favorable for achieving protection by DNA/MVA vaccines and merits further investigation. The gp120 and MVAgp140 boosts enhanced functional as well as binding antibody responses. The gp120 boosts enhanced both tier 1A neutralizing and ADCC responses, whereas MVAgp140 enhanced ADCC responses. Elicitation of higher neutralizing titers for tier 1A viruses by the protein boost likely reflected the strong ability of the gp120 protein to boost V3 responses.
Interestingly, for all three groups, the 2nd MVA boost elicited higher responses to tier 1B and tier 2 viruses than the 3rd boost. This phenomenon may reflect multiple protein boosts driving responses away from "neutralizing" conformational epitopes and toward "nonneutralizing" linear epitopes. It also could reflect immunoglobulin editing for autoreactive antibody responses, a phenomenon that can occur for tier 2 broadly neutralizing antibodies that are frequently autoreactive (24).
Both boost immunogens enhanced ADCC responses for bound B.63521Δ11 protein. This likely reflects both boosts enhancing the binding antibody response to B.63521Δ11. The higher ADCC responses elicited by MVAϩgp120 than MVAϩMVAgp140 likely reflected the very high titers of binding activity for B.63521Δ11mutC elicited by the matched gp120 boost.
To determine if any correlations existed between functional responses and responses to linear peptides, week 28 data from all groups were pooled and analyzed for correlations between neutralizing antibody and ADCC activity with binding to specific peptides. Neutralizing titers were found to associate with magnitude of binding to linear epitopes C1.1, C1.2, V3.1, V3.2, and gp41_ID (Spearman r Ͼ 0.6; raw_p Ͻ 0.05). These correlations were strongest and most frequent for the two tested tier 1 viruses. The finding of correlations with the gp41_ID region, albeit modest, are of interest because these correlations, with a highly conserved region in Env, included 4 of 9 tested tier 1B and tier 2 viruses. Monoclonal antibodies (MAbs) to the gp41_IDR capture virions (25,26) and mediate ADCC (27) but have not been reported to have neutralizing activity. The anti-gp41_ID antibodies could also be biomarkers for a true mechanism that contributes to the neutralizing activities observed.
ADCC activity was found to be moderately associated with C1.1 and C1.2 binding. In RV144 subjects, a conformational C1 MAb (A32) blocked ADCC activity in over 90% of subjects (7). A32 did not bind to linear peptides in the microarray linear epitope mapping (data not shown). Whether the linear C1-binding antibodies, like the ones seen in this study, can mediate ADCC activity warrants further investigation. The higher linear C1-binding responses also may correlate with conformational C1 responses that were not quantified in this study.
In conclusion, gp120 protein and MVAgp140 were similar in their abilities to increase levels of antibody responses for gp120 and the V1V2 region of gp120 over the MVA-only boosts in the GOVX-B11 DNA/MVA vaccine regimen but differed in their abilities to enhance CD4 ϩ T cell responses. These data demonstrate that both recombinant gp120 protein and MVA-expressed secreted gp140 are effective strategies for enhancing potential protective nonneutralizing antibody responses to HIV.
The study has had a direct impact on the protein boost protocol being advanced

MATERIALS AND METHODS
Immunogens. The GOVX-B11 vaccine consists of the pGA2/JS7 DNA and MVA/HIV62B (MVA62B) immunogens. The DNA prime expresses VLPs displaying the ADA gp160 Env. The DNA used in this trial also coexpressed rhesus granulocyte-macrophage colony-stimulating factor (GM-CSF) (5,28). The DNA was produced, purified, and suspended in a proprietary buffer at 3 mg/ml at Aldevron (Fargo, ND). MVA62B expresses VLPs displaying ADA gp150 Env (29,30). MVA62B (lot 001-03-12) was produced in chicken embryo fibroblasts (CEF) at IDT Biologica, purified by tangential-flow filtration, and suspended at Linear epitope mapping. Serum epitope mapping of heterologous strains was performed as previously described (32)(33)(34), with minor modifications. Briefly, array slides were provided by JPT Peptide Technologies GmbH (Germany) by printing a library designed by B. Korber, Los Alamos National Laboratory, onto epoxy glass slides (PolyAn GmbH, Germany). The library contains overlapping peptides (15-mers overlapping by 12) covering 7 full-length gp160 consensus sequences (clades A, B, C, and D, group M, CRF1, and CRF2) and gp120 sequences of 6 vaccine strains (MN, A244, Th023, TV-1, ZM641, and 1086C). Three identical subarrays, each containing the full peptide library, were printed on each slide. All array slides were blocked with phosphate-buffered saline (PBS) containing 1% powdered milk, 5% normal goat serum, and 0.05% Tween for 1 h, followed by a 2-h incubation with 1:50 diluted test sera and a subsequent 45-min incubation with goat anti-human IgG conjugated with AF647 (Jackson ImmunoResearch, PA). Array slides were scanned at a wavelength of 635 nm with an InnoScan 710 AL scanner (Innopsys, Carbonne, France) using XDR mode. Images were analyzed using MagPix 8.0 software to obtain binding intensity values for all peptides. Binding of postimmunization serum to each peptide was subtracted from its baseline value, which was defined as the median signal intensity of the triplicates of the peptide for the matched prebleed serum plus 3 times the standard error of the triplicates. The magnitude of binding to each identified epitope was defined as the highest binding by a single peptide within the epitope region.
Avidity. The affinity maturation of the vaccine-elicited antibody responses to the immunodominant region of gp41 (gp41_ID) and V3 linear epitopes (V3.B and V3.C) was measured by an avidity index assay using the binding antibody multiplex assay. Avidity index was measured by inclusion of a denaturation step (treating the samples with 0.1 M sodium citrate, pH 3.0) (15). Untreated wells contained PBS during the avidity incubation step. The avidity index is defined as the MFI in the treated well divided by MFI in the untreated well times 100. Monoclonal antibodies 7B2 (35), 447-52D (36), and DH151 (37) were included as positive controls for gp41_ID, V3.B, and V3.C, respectively.
ADCC-GranToxiLux assay. Antibody dependent cellular cytotoxic activity mediated by test animal plasma/serum was detected according to our modification of the previously described GranToxiLux cell-mediated cytotoxicity procedure (38). CEM.NKRCCR5 target cells were coated with recombinant B.63521Δ11mutC gp120 protein. The assay readout is the percentage of antigen-coated target cells taking up granzyme B. A positive response is present if the peak percent granzyme B activity across 6 dilutions is greater than or equal to 8%.
Neutralizing antibodies. Neutralizing antibodies were measured as a function of reductions in luciferase (Luc) reporter gene expression after a single round of infection in TZM-bl cells (39,40). TZM-bl cells (also called JC57BL-13) were obtained from the NIH AIDS Research and Reference Reagent Program, as contributed by John Kappes and Xiaoyun Wu. Briefly, a pretitrated dose of virus was incubated with serial 3-fold dilutions of test sample in duplicate in a total volume of 150 l for 1 h at 37°C in 96-well flat-bottom culture plates. Freshly trypsinized cells (10,000 cells in 100 l of growth medium containing 75 g/ml of DEAE dextran) were added to each well. One set of 8 control wells received cells plus virus (virus control) and another set received cells only (background control). After 48 h of incubation, 100 l of cells was transferred to a 96-well black solid plate (Costar) for measurements of luminescence using the Britelite luminescence reporter gene assay system (PerkinElmer Life Sciences). Assay stocks of molecularly cloned Env-pseudotyped viruses were prepared by transfection in 293T/17 cells (American Type Culture Collection) and titrated in TZM-bl cells as described previously (39). This assay has been formally optimized and validated (40) and was performed in compliance with good clinical laboratory practices, including participation in a formal proficiency testing program (41). Additional information on the assay and all supporting protocols may be found at http://www.hiv.lanl.gov/content/nab-reference -strains/html/home.htm.
Statistics. Antibody and T cell response measurements were first screened in an omnibus test. Differences between pairs of groups in the levels of measurements that showed a P value of Ͻ0.05 were further tested using Wilcoxon rank sum exact test. Correlations between binding antibody responses and neutralization and ADCC responses were tested using the Spearman correlation test. P values from the Wilcoxon rank sum test and Spearman correlation test were corrected for false discovery rate (FDR) using the Benjamini-Hochberg (BH) method (43). Statistical significance indicated on figures are based on P values that are not controlled for multiple comparison (raw_p). Both raw_p and FDR-adjusted P values (FDR_p) are shown in Tables 2 and 4.