Evaluation of an antibody to α4β7 in the control of SIVmac239-nef-stop infection

See allHide authors and affiliations

Science  06 Sep 2019:
Vol. 365, Issue 6457, pp. 1025-1029
DOI: 10.1126/science.aav6695

An antibody is not the antidote

An HIV therapeutic that would give long-term remission without sustained antiretroviral therapy (ART) is a long-term goal. Byrareddy et al. [Science354, 197 (2016)] reported that treating simian immunodeficiency virus (SIV)–positive macaques with an antibody against integrin α4β7 during and after ART results in sustained virologic control after stopping all treatment. Three studies in this issue question the reproducibility of that result. Di Mascio et al. sequenced the virus used in the 2016 study and found that it was a variant with a stop codon in the nef gene rather than a wild-type virus. Abbink et al. used the same antibody for α4β7 as before but tested control of a more commonly used pathogenic virus. Iwamato et al. used the same nef-stop virus as in the earlier paper but combined the antibody against the integrin with an antibody against the SIV envelope glycoprotein, which also blocks viral binding of the integrin. None of these three new studies found that treating with the antibody had any effect on virologic control after stopping ART treatment.

Science, this issue p. 1025, p. 1029, p. 1033


Treatment of SIV-infected rhesus macaques with short-term antiretroviral therapy (ART) and partially overlapping infusions of antibody to integrin α4β7 was reported to induce durable posttreatment viral suppression. In an attempt to replicate those observations, we treated macaques infected with the same virus and with the same ART and monoclonal antibody (mAb) regimens (anti-α4β7 versus control mAb). Sequencing demonstrated that the virus used was actually SIVmac239-nef-stop, not wild-type SIVmac239. A positive correlation was found at 2 weeks after infection between the frequency of repair of attenuated Nef-STOP virus to pathogenic Nef-OPEN and plasma SIV RNA levels. Levels of plasma viremia before the first antibody infusion and preinfection levels of α4β7hi CD4+ T cells, but not treatment with antibody to α4β7 , correlated with levels of viral replication upon discontinuation of all treatments. Follow-up plasma viremia, peripheral blood CD4+ T cell counts, and lymph node and rectal tissue viral load were not significantly different between anti-α4β7 and control mAb groups.

A goal in HIV treatment is to replace a lifetime of daily antiretroviral therapy (ART) with immunotherapeutic interventions that induce durable ART-free remissions (1). One proposed strategy involves the administration of monoclonal antibodies (mAbs) to α4β7, an integrin expressed at high levels on approximately 10 to 30% of circulating CD4+ T cells—the main target cells of HIV. Integrin α4β7+ lymphocytes traffic to gut-associated lymphoid tissue (GALT) by interacting with the integrin’s natural ligand, the mucosal vascular addressin cell adhesion molecule 1 (MADCAM-1), which is expressed on high endothelial venules within the gut. Functional α4β7 is also incorporated in SIV or HIV-1 virions, with high levels associated with sustained viral replication in the gut during the early phase of infection (2). The gut is reported to play a major role in the pathogenesis of SIV/HIV disease (3). Hence, a number of strategies have evolved that aim at mitigating the virus-induced damage to the GALT. This could potentially be accomplished by blocking the migration of vulnerable CD4+ T cells to the gut or by interfering with interactions of the virus with CD4+ T cells, particularly within the gut compartment. Although expression of α4β7 on the surface of CD4+ T cells is not required for infection, conformational structure studies of α4β7 on the surface of CD4+ T cells revealed a binding site for the HIV-1 envelope glycoprotein gp120 (4), and additional studies have suggested that CD4+ T cells expressing high levels of α4β7 (CD4α4β7 hi) may be preferential targets for HIV-1 infection (5).

Retrospective analyses have also provided evidence that high levels of α4β7 expressed on CD4+ T cells are predictive of increased susceptibility to AIDS virus infection and disease progression in both nonhuman primates (NHPs) (6, 7) and humans (8). Macaques with fewer than 10% CD4 α4β7 hi cells before rectal challenge with the SIVmac239, a virus commonly used in NHP studies, exhibited slower viral dissemination (6).

It was also reported that the administration of a primatized mAb to α4β7 before and during acute infection after repeated low-dose intravaginal challenges with SIVmac251 (9) was able to protect a significant proportion of animals from infection. In those animals treated with antibody to α4β7 that did become infected, the GALT was relatively protected from infection, indicating that even when infection occurred, blocking α4β7 had a positive effect on the outcome of infection (9).

These observations suggest the possible induction of a durable positive impact on the course of infection even after the antibody was cleared from the system, although any underlying mechanism remains unclear.

Durable control of established SIV infection was described in a follow-up study (10) in which animals treated with anti-α4β7 mAb administered after the acute phase of infection, in a regimen partially overlapping with administration of ART, showed dramatic control of plasma viremia and maintained high levels of peripheral blood (PB) CD4+ T cells after discontinuation of all forms of treatment. By contrast, animals that received an isotype-matched-control mAb experienced viral rebound and decline of CD4+ T cells. Given the potential clinical importance of these results to HIV infection and the availability of a humanized anti-α4β7 mAb [Vedolizumab (11)] that allowed initiation of testing of this approach in HIV-1–infected humans (NCT02788175), we undertook the present NHP study in an attempt to confirm the preclinical observation of durable posttreatment virologic control and potentially to gain new insights into the mechanisms of this effect. We sought to replicate the experimental protocol of the prior study (10), using the same viral stock, dose, route of inoculation, and the same antibody preparation and ART regimen (10). We included lymph node and rectal biopsies to compare the viral dynamics in these sites with the dynamics in PB.

Twenty-two juvenile to adult Indian rhesus macaques) negative for Mamu-A001, -B008 and -B017 alleles were infected intravenously with a 200 median tissue culture infectious dose (200 TCID50) of the viral stock. A sample size of 22 was chosen to achieve greater than 80% power to detect at least a 100-fold difference in virologic set points between the experimental (active) and control arms. In characterizing the virus stock, single-genome amplification (SGA) and sequence analysis revealed that the virus challenge stock was not wild-type SIVmac239 as reported (10); rather, 100% of viral genomes sequenced had a STOP codon at residue 93 of Nef (SIVmac239-nef-stop). This previously characterized variation results in a truncated Nef protein and a Nef-negative phenotype. Although in vivo selection in macaques infected with SIVmac239-nef-stop eventually results in repair to an open reading frame and reversion to a Nef-positive phenotype, the kinetics and specific pathways of repair can vary between animals, resulting in variable dynamics of viral replication and host immune responses (12).

The study design recapitulated the design of the prior study (10) and included five phases (Fig. 1A). After 5 weeks of SIV infection, all 22 animals began a 13-week course of ART with the reverse transcriptase inhibitors tenofovir and emtricitabine (provided by Gilead) and the integrase strand transfer inhibitor L-870812 (provided by Merck). Animals were assigned to balanced groups according to sex, weight, and area under the curve (AUC) of plasma viremia from the day of infection to the day of start of ART to receive intravenous infusions of either the primatized anti-α4β7 mAb (n = 12 animals, active group) or an isotype-matched-control mAb (n = 10 animals, control group) once every 3 weeks for a total of eight infusions beginning at week 9. ART was discontinued at week 18, and infusions of the anti-α4β7 mAb or the control antibody were continued to week 30. One out of the 12 anti-α4β7 mAb–treated animals (OTI) developed antibodies against the anti-α4β7 mAb in association with lower serum levels of the mAb, and data from that animal were excluded from further analysis (figs. S1 and S2, A and B). The remaining 11 animals treated with antibody to α4β7 demonstrated therapeutic levels (> 18 μg/ml) (13) of the mAb in the serum and lack of evidence of clinically relevant antidrug antibody formation (figs. S1 and S2A).

Fig. 1 Viral and CD4 T cell count dynamics.

(A) Study design. (B and D) Plasma viral loads (left) and (C and E) CD4+ T cell counts from 12 monkeys receiving ART + anti-α4β7 mAb and 10 monkeys receiving ART + control immunoglobulin G (IgG). (F) The log of geometric means for plasma viral load and (G) geometric means for CD4 T cell counts for monkeys treated with ART + anti-α4β7 mAb (blue, excluding OTI) or IgG (red).

During the acute phase of infection and before any intervention, peak plasma viremia averaged ~106 copies/ml with a wide range (1.9 × 105 to 4.3 × 107 copies/ml), with peak values achieved between 2 and 5 weeks after infection (Fig. 1 and fig. S3). These viral dynamics are more typical of attenuated NefSIV variants than wild-type SIVmac239 (14, 15). Similarly, PB CD4+ T cell counts in the entire group of animals (Fig. 1 and fig. S3) did not show the marked decline at week 2 after infection that was observed in the prior study (10) or in SIVmac251 infection of A001-, B008-, or B017-Indian rhesus macaques, as reported in the accompanying manuscript Abbink et al. (mean percent CD4+ T cell drop at week 2 after infection, 35%; P < 0.001 versus 8%, not statistically significant in the present study) (16). Single-genome sequencing of the nef region of plasma virus at week 2 after infection (25 to 60 independent SGA clones per animal) showed a striking variation in the frequency of Nef-STOP remaining at that time point, ranging from 0 to 91% (average 26%) (Fig. 2E). There was a positive correlation between the frequency of viral sequences with a repaired Nef-OPEN sequence at week 2 after infection and the level of plasma viremia [Spearman rank correlation (ρ) = +0.66, P < 0.001] (Fig. 2A), underscoring the important role played by Nef in viral replication and dissemination throughout the body (14). By week 5 after infection, at the time of initiation of ART, all viruses had repaired the Nef-STOP codon to Nef-OPEN sequences, although the codons and amino acids used to restore the Nef-OPEN sequence varied both within and between animals, with potential phenotypic consequences (Fig. 2E) (12). These observations are overall consistent with the dynamics of Nefrepair reported in the accompanying manuscript by Iwamoto et al. (17), in which faster reversion to Nef-OPEN were observed in the settings of higher levels of viral replication and greater drops in CD4+ T cells at week 2 after infection.

Fig. 2 Dynamics of Nef-stop repair and predictors of viral set-point.

(A and B) Correlations between the percentage of virus in plasma restored to Nef-OPEN at week 2 after infection and (A) plasma viral load at week 2and (B) plasma viral load at week 5. (C and D) Correlation between plasma viral load set point (average of week 45 and week 48 after infection) and AUC of log plasma viremia (C) during the first 5 weeks of infection, before ART initiation, or (D) during the first months of ART, prior to antibody infusion. Red dots, control animals; blue dots, α4β7-treated animals. (E) Pie charts summary of the Nef sequencing data. Each slice corresponds to the proportion of viruses in plasmas of animals at weeks 2 and 5 after infection with specific amino acid substitutions at position 97 of Nef, where black indicates Nef-STOP. Animals were ranked from top to bottom in each group on the basis of the amount of Nef-STOP virus. Plasma viral loads throughout the course of the study reveal association between early and long-term viral replication, independent of the treatment received. (F) The log of the geometric means of animals achieving >5 log viremia at set point (black) and of animals achieving <4 log viremia at set point (green) are shown along with the viral kinetics of each subject (solid lines indicate animals from the active group, and dashed lines indicate animals from the control group).

After initiation of ART, plasma viremia decreased in two phases, as previously modeled in HIV-1 infection (18). There were no significant differences between the two treatment groups (Fig. 1). During the administration of the antibodies, when plasma viral load was dramatically reduced as a result of ART, a transient nonstatistically significant increase in PB CD4+ T cell counts was observed in the α4β7-treated group (Fig. 1), potentially because of alterations in CD4+ T cell trafficking with blocking of migration of cells to the gut caused by mAb blockade of α4β7. A recent study demonstrated dramatic increases in both CD4+ T cell and CD8+ T cell counts in the PB after the administration of the rhesus recombinant anti-α4β7 mAb in uninfected Indian rhesus macaques (19).

After discontinuation of ART at 18 weeks after infection, we did not observe differences between active and control groups in either plasma SIV RNA levels or PB CD4+ T cell counts (Fig. 1), nor in PB CD8+ T cell or B cell counts or percent of proliferating cells (figs. S5 and S6). Similarly, the levels of SIV-DNA and SIV-RNA in lymph node and rectal biopsies were similar between the active and control groups (Fig. 3 and fig. S4). This lack of effect of the antibody to α4β7 on plasma SIV RNA, PB CD4+ T cells, and tissue levels of SIV RNA and DNA is in contrast to the prior study (10) but consistent with the findings in the accompanying manuscripts (16, 17).

Fig. 3 Cell-associated-viral-load.

SIV-DNA and SIV-RNA log of geometric mean levels from (A and C) lymph nodes and (B and D) rectal biopsies for monkeys treated with ART + anti-α4β7 mAb (blue, excluding the OTI) or ART + IgG (red).

Another clear difference between Byrareddy et al. (10) and the present study is the rate of PB CD4+ T cell count decline between the control groups of the two studies, not only during the acute phase of infection as described above but also during the chronic phase of infection. At 7 months after discontinuation of ART, both the antibody to α4β7– and control antibody–treated groups in the present study showed higher PB CD4+ T cells counts (~500 cells/μl) and lower plasma SIV-RNA levels (~1 to 2 log SIV-RNA copies/ml) than those reported in the prior study (10) or typically seen during chronic infection with SIVmac239 (20) or SIVmac251 [(21) and accompanying manuscript (16)]. Consistent with these observations, tissue SIV-DNA levels in our control group were also lower than those of control animals in the prior study (10).

Plasma RNA measured at 7 months after discontinuation of ART showed a significant association with plasma viremia levels measured before assignment to anti-α4β7 or control antibody groups (ρ = 0.38, P = 0.04 for plasma viremia AUC before ART and ρ = 0.62, P = 0.001 for viral levels during the first month of ART) (Fig. 2, C and D) and regardless of treatment group (Fig. 2F). This suggests a direct association between early levels of viral replication and the level of the ultimate viral setpoint regardless of attempted immunotherapeutic intervention. Because there was a positive correlation between the frequency of viral sequences with a Nef-STOP repaired to Nef-OPEN sequence early in infection (week 2 after infection) and the level of plasma viremia, it is likely that virus in those animals in which the Nef-STOP was not quickly repaired did less initial damage to the immune system and thus allowed for better immune control of virus, as reflected by the lower viral setpoint observed after discontinuation of all intervention, even after the virus ultimately reverted to phenotypically functional Nef-OPEN. However, similar low levels of viral replication and high CD4+ T cell counts at set point were also observed in a group of intrarectally challenged SIVmac251-infected animals that also were started on ART at 5 weeks of infection [(16), the acute group]. These data point to a potential role of early ART administration in establishing a lower set point after ART discontinuation. This could have important implications in understanding the recently observed wide variation in the duration of posttreatment control of HIV-1 viremia in individuals undergoing empirical interruption of ART (22).

The reasons for the overall differences in the present study from the results reported in the prior study (10) are unclear; however, they could be due to differences in the rate of repair of the STOP codon in the SIVmac239-nef-stop virus used in the two studies.

Potential unappreciated differences in the macaques studied could also have contributed to the divergent results. Arguing against this is that there were no differences in a variety of known restriction factors between the Indian origin rhesus macaques used in the present study and those in the prior study (tables S1 and S2) (10). Differential levels of expression of α4β7 on the CD4+ T cells of the two different cohorts of macaques also could have played a role in altering viral kinetics, as could other host factors affecting the pharmacodynamics of the immunotherapeutic intervention. In our cohort, baseline levels of percent α4β7hi in PB CD4+ T cells (15.6% ± 1.6% mean ± SE) positively correlated with early levels of viral replication (Log plasma viremia at start of ART, ρ = 0.48, P = 0.01) as well as with levels of viral replication upon ART discontinuation [log SIV-RNA/106 cells in lymph nodes at week 30 after infection, ρ = 0.42, P = 0.04 (fig. S7E); log SIV-RNA/106 cells in rectal tissue at week 30 after infection, ρ = 0.4, P = 0.03 (fig. S7F); log plasma viremia at setpoint, ρ = 0.36, P = 0.05 (fig. S7B)]. Lack of precisely corresponding analyses from the prior study preclude direct comparisons.

Last, whereas multiple studies have clearly demonstrated a potential role for α4β7 in the pathogenesis of HIV/SIV infection (11), the contrast of the present results and the results reported in the accompanying manuscripts (16, 17) with those reported in the prior study suggest that achieving any potential therapeutic benefit by blocking α4β7 in HIV infection is unlikely to be straightforward. We observed no evidence that treatment with antibody to α4β7 leads to enhanced virologic control after ART cessation.

Supplementary Materials

Materials and Methods

Figs. S1 to S7

Tables S1 and S2

References (2329)

References and Notes

Acknowledgments: We thank the animal care staff and technicians at the NIH Animal Facility in Poolesville (Maryland, USA) for their care and handling of the animals, and we thank A. Tobery, J. Kaplan, A. Bonvillian, R. Byrum and M. St. Claire from the Division of Clinical Research for veterinary support and technical assistance throughout the course of the study. We thank the Quantitative Molecular Diagnostics Core and the Viral Evolution Core of the AIDS and Cancer Virus Program of the Frederick National Laboratory for expert performance of viral load assays and viral sequence analysis, respectively. We also thank R. Wiseman and D. O’Connor from the Genetics Services Unit of the Wisconsin National Primate Research Center for performing Genotyping assays. We thank J. Nkolola (Harvard Medical School) for performing enzyme-linked immunosorbent assays to measure anti-α4β7 mAb concentration in the plasma of treated animals. We thank P. Lusso for in vitro tests performed on the anti-α4β7 mAb administered in this study. We thank R. Turnier, L. Krymskaya, and M. Davies from Leidos Biomedical Research for flow cytometry support. We also thank Merck for providing L-870812 and Gilead for providing tenofovir and emtricitabine. Last, we thank F. Villinger from Emory University/Yerkes National Primate Research Center for providing the viral stock. Funding: Genotyping assays were performed with support from P51OD011106. Antibody to α4β7 and control antibodies were provided by the NIH NHP Reagent Resource supported by R24OD010976 and U24AI126684. This work was supported in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract HSN261200800001E. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of any trade names, commercial products, or organizations imply endorsement by the U.S. government. Authors contributions: M.D.M., J.D.L., S.S., I.K., B.F.K., K.A.R., Y.W., H.C.L., and A.S.F. designed the research; M.D.M., J.D.L., S.S., I.K., P.D., B.F.K., A.J.B., K.A.R., and Y.W. performed the research; M.D.M. and S.S. made the figures and tables; all authors analyzed and interpreted the data and critically reviewed the manuscript. M.D.M. wrote the manuscript. H.C.L., J.D.L., and A.S.F. provided oversight and help with editing of the manuscript. Competing interests: The data presented in this manuscript are tabulated in the main paper and in the supplementary materials. A.S.F. is a co-inventor on patent no. 20160075786 held by the National Institute of Allergy and Infectious Diseases (NIAID). This patent covers the use of antagonists of the interaction between HIV gp120 and a4b7 integrin. The authors declare no competing financial interest. Data and materials availability: The data presented in this manuscript are tabulated in the main paper and in the supplementary materials.
View Abstract

Navigate This Article