Cell-Mediated Immunity to Target the Persistent Human Immunodeficiency Virus Reservoir

Abstract

Effective clearance of virally infected cells requires the sequential activity of innate and adaptive immunity effectors. In human immunodeficiency virus (HIV) infection, naturally induced cell-mediated immune responses rarely eradicate infection. However, optimized immune responses could potentially be leveraged in HIV cure efforts if epitope escape and lack of sustained effector memory responses were to be addressed. Here we review leading HIV cure strategies that harness cell-mediated control against HIV in stably suppressed antiretroviral-treated subjects. We focus on strategies that may maximize target recognition and eradication by the sequential activation of a reconstituted immune system, together with delivery of optimal T-cell responses that can eliminate the reservoir and serve as means to maintain control of HIV spread in the absence of antiretroviral therapy (ART). As evidenced by the evolution of ART, we argue that a combination of immune-based strategies will be a superior path to cell-mediated HIV control and eradication. Available data from several human pilot trials already identify target strategies that may maximize antiviral pressure by joining innate and engineered T cell responses toward testing for sustained HIV remission and/or cure.

The absence of a cure for human immunodeficiency virus type 1 (HIV-1) after antiretroviral therapy (ART) imparts multiple burdens to the infected individual (stigma, criminalization, life-long therapy), society (continued infections, internal healthcare costs, PEPFAR [US President’s Emergency Plan for AIDS Relief]), industry (pricing vs access, capacity for worldwide drug production), and global resource governance (UNAIDS [Joint United Nations Programme on HIV/AIDS], AIDS orphans, Global Fund, effects of declining economy on country/region, HIV therapy programs, etc), justifying a focused investment in identifying novel strategies to achieve a cure and/or stable remission after HIV infection. It is now accepted that a cell-based therapy has been the only path to date that has resulted in the cure of a person with HIV-1 infection [1]. The paradigm-shifting, “N = 1” experimental treatment conducted by Hütter and colleagues has established the principle that curative approaches are possible. However, several factors may have contributed to this outcome, singly or in combination, to eradicate the reservoir, including allogeneic effects of graft-vs-host disease (GVHD), antithymocyte globulin, chemotherapy, CCR5 ablation, immunosuppression, and total body irradiation. What is clear is that allogeneic experiments will not be routinely effective, as this approach was a failure in several other patients treated with a similar clinical protocol [2, 3]. There have been some impressive efforts to try to deconvolute the Berlin patient to better understand the crucial elements that enable HIV-1 cure. Investigators at Brigham and Women’s Hospital in Boston performed a stem cell transplant using CCR5 wild-type donors in 2 individuals as part of a cancer-treating regime to explore the role GVHD plays in HIV-1 cure [4]. Both subjects appeared to be free of HIV-1 for several months in the absence of ART, which was notable (1 patient did not have a detectable viral load for 32 weeks), but eventually HIV plasma virus levels returned, with each patient having a high viral load before ART was resumed [5]. A separate outcome of long-term remission in association with a reduced reservoir has been reported for an infant infected at birth, who received ART 30 hours after birth and ART treatment continued for 18 months until unmonitored ART discontinuation [6]. For 27 months, the child’s viral load remained before the level of detection but unfortunately, like the Boston patients, detectable viral load was eventually observed, and the number of cured HIV-1–infected individuals remains at 1. There are several lessons that can be learned from these pioneering studies that currently inform HIV cure studies: (1) It is possible to reduce the latent reservoir; (2) HIV-1 reservoirs appear to be very difficult to eliminate or prevent; (3) prolonged absence of viral load does not mean cure; and (4) factors that mediate long-term remission of HIV-1 in the absence of ART are unclear.

A successful strategy to maintain stable remission after ART interruption (ie, functional cure) or viral eradication would be expected to (1) achieve sizable reductions of HIV reservoirs and (2) provide for the development of a successful immune-mediated response able to contain and then eradicate HIV upon any residual viral reactivation. Initial strategies to reduce HIV reservoirs on ART have primarily focused on reactivation of latent reservoirs or attempts to infuse resistant CD4 T cells in the presence of ART. Developing approaches that can also activate cell-mediated responses, whether innate (natural killer [NK] responses) or by HIV-specific T-cell effectors, remains an attractive concept if factors that allow HIV to evade such responses can be addressed. Here, we will review data and rationale supporting the concept that both innate and HIV-specific T-cell–mediated effector mechanisms can be applied toward HIV cure strategies.

STATE OF THE HIV-SPECIFIC T-CELL RESPONSE DURING SUCCESSFUL ART THERAPY

HIV infection induces a robust T-cell response [7]. However, in the great majority of HIV-infected individuals, the immune response fails to maintain control of HIV-1 replication, and a chronic infection ensues. HIV’s tremendous capacity to escape immunodominant HIV-specific immune responses because of an error-prone reverse transcriptase step and heightened immune cell activation that promotes high viral replication allows HIV to outrun the endogenous immune response. Before effective ART was available, HIV infection would functionally burn out most the HIV-specific CD8 T cells and eliminate the vast majority of CD4 T cells, leading to profound immunodeficiency and eventual death [8–11]. In individuals treated with effective ART after the infection has become well established, the loss of HIV antigen via the introduction of ART leads to a rapid decline in the total number of HIV-specific T cells. Unfortunately, the HIV-specific T-cell dysfunction is not fully reversed after viral replication is controlled by ART in chronic infection, as reflected by a lack of viral control after treatment interruption [12–14]. However, improvements in overall T-cell function are observed after prolonged ART [15]. With the recent focus of starting ART as soon as possible after HIV transmission, some studies have addressed how early ART can counter the detrimental effects of HIV replication on the efficacy of HIV-specific CD8 T-cell responses. As expected, early ART treatment also blunts the HIV-specific CD8 T-cell response so that there are far fewer total HIV-specific T cells [16]. Even if blunted, a more important question is whether early treatment preserves the ability of T cells to control HIV replication. Early-treated adults where ART was subsequently interrupted after years of suppression suggest that in some subjects, suppression is achieved yet its relation to T-cell responses is lacking. The Viro-Immunological Sustained CONtrol after Treatment Interruption (VISCONTI) cohort is a subset of 14 individuals who are controlling plasma viremia without ART yet whose history revealed that they were treated with ART shortly (Fiebig stage III–V) after viral transmission and remained on ART for many years [17]. However, unlike elite controllers who were never treated with ART and whose HIV-specific CD8 T cells are interpreted to contribute toward control of HIV replication, there is no evidence that CD8 T cells are contributing to the control of viremia in the VISCONTI individuals. CD8 T cells from these posttreatment controllers had limited ability to prevent HIV spread in an in vitro suppression assay relative to elite controllers and viremic individuals. The latter suggests that early treatment does not preferentially preserve highly suppressive CD8 T-cell responses and that other factors are responsible for the ability of these individuals to control HIV replication in the absence of ART. In fact, HIV-specific CD8 T cells from cohorts of individuals who were treated with ART soon after infection do not differ remarkably from HIV-specific T cells from individuals who were treated much later after the onset of HIV infection [17]. It should be noted that an analysis of a small number of VISCONTI individuals suggested that the HIV-specific CD4 T cells do appear as functional as those from elite controllers [18]. Other studies also suggest that T-cell dysfunction sets in quickly after HIV infection in most individuals. Examination of hyperacute individuals (Fiebig stage 1) shows that T-cell dysfunction is apparent prior to the establishment of the viral setpoint [7]. It is possible that ART treatment of hyperacute individuals may preserve CD8 T-cell function for use in HIV cure approaches, but clearly this approach would be limited to a small number of HIV-infected individuals. Studies directly evaluating relationships between CD8 T-cell HIV-specific responses and viremia after treatment interruption, or in viremic controllers off ART, have suggested a contribution of CD8 T-cell control when present alongside of functional innate dendritic cell function [19, 20]. Together, these data indicate that the HIV-specific CD8 T-cell response that remains after successful ART treatment is not sufficient to control HIV replication once ART is removed and will need to be optimized considerably to be part of a HIV cure strategy.

IMMUNE CHECKPOINT AS A MEANS TO REJUVENATE THE HIV-SPECIFIC IMMUNE RESPONSE TO CURE OR DURABLY CONTROL HIV REPLICATION

Pioneering studies from Rafi Ahmed’s group using murine lymphocytic choriomeningitis virus (LCMV) infection model described a T-cell differentiation process called exhaustion by which T cells chronically exposed to antigen lose T-cell effector functions [21]. Exhausted T cells express high levels of PD-1, which is a negative regulator of T-cell activation. Importantly, administration of agents that interfere with PD-1 binding with its ligands is able to partially restore T-cell function by exhausted T cells, giving hope that administration of these agents could have therapeutic value [22]. A series of clinical trials targeting the PD-1 pathway in melanoma patients validated checkpoint blockade as a promising new therapy by demonstrating pronounced and durable remissions in approximately 30% of individuals [23]. Individuals who had high levels of PD-1 and/or a related immune checkpoint molecule, CTLA-4, expressed on their tumor-infiltrating lymphocytes (TILs) were the most likely to benefit from PD-1 antibody (Ab) therapy, suggesting that tumor regression was in part the result of restoration of T-cell function to these TILs [24]. HIV-specific CD8 T cells share many characteristics with exhausted tumor-specific T cells [25], and in vitro studies have demonstrated that blocking PD-1 augmented the HIV-specific immune response [26–28]. The National Institute of Health’s AIDS Clinical Trials Group (ACTG) investigators recently asked whether anti–PD-1 therapy could also restore activity in HIV-specific T cells during chronic infection to help reduce the viral reservoir and maintain immune control of HIV replication. Preliminary results were presented by the ACTG A5326 Study Team [29]. Eight individuals successfully being treated with ART were treated with anti–PD-L1 Ab (BMS-936559) at 0.3 mg/kg, yet the amount of measurable virus did not change. Despite no change in viral measures, ex vivo measures of T-cell responses against HIV-1 Gag did strongly improve in 2 of the 8 HIV-infected individuals, consistent with the expected activity of anti–PD-L1 in augmenting T-cell function. This dose was chosen because only 1 cancer patient out of 3 treated had a mild adverse event at this dose level [30]. Cancer patients treated using higher doses of BMS-936559 had significantly more severe adverse events. At this low, 0.3-mg/kg dose, 1 HIV-infected individual was diagnosed with pituitary insufficiency that was likely caused by the therapy, and retinal toxicity was observed in an analogous primate trial. This study may signal a potential important difference between the expectations for potential adverse events due to immunotherapy between ART-suppressed, HIV-infected subjects and cancer subjects. Namely, the immune system in the cancer patients is largely suppressed and thus adverse events such as autoimmune reactions may be more prevalent in ART-controlled, HIV-infected individuals. Moreover, acceptable adverse events in cancer patients often receiving experimental treatments after failing standard of care options are often not acceptable adverse events in HIV-infected individuals who can be successfully treated on ART. While other strategies are likely to emerge that may also seek to reactivate otherwise exhaustion-prone CD8 T-cell responses, the inherent potential of these new or reactivated memory responses to lose specificity by epitope escape may still limit their long-term potential to mediate remission.

VACCINATION APPROACHES TO MANUFACTURE T CELLS FOR HIV CURE STRATEGIES

Developing vaccination approaches to prevent HIV infection have been considered a priority since HIV was recognized as the etiological agent of AIDS. Given that preexisting HIV-specific T-cell responses are unable to control HIV spread after ART cessation, vaccination strategies to bolster and supplement the HIV-specific immune response may be a promising way to eliminate the HIV latent reservoir during an HIV cure approach. At the present time, there is no clear choice for the best vaccine to use as part of a HIV cure strategy, but it is important to keep in mind that goal of a vaccine in a HIV cure strategy is different from the goal of a vaccine that is used to prevent infection. For example, the MRKAd5 HIV gag/pol/nef vaccine was safe to administer and elicited broad HIV-specific T-cell responses [31, 32]; however, this vaccine did not protect individuals from HIV infection [33] and arguably promoted acquisition of infection in some individuals. The inability of this vaccination strategy to protect individuals from infection does not necessarily preclude it and other T-cell vaccines from being used in HIV cure-based approaches, as the goals of each are different. For a preventive vaccine to work, it must contain and eliminate 100% of the infectious HIV that is transmitted to be effective. For use in HIV cure strategies, significant but not necessarily absolute elimination of replication-competent HIV would be therapeutically useful as these approaches can be done multiple times to help ensure a durable HIV cure. Moreover, whereas Ad5-mediated immune stimulation may augment viral transmission and thus be counterproductive for a preventive vaccine [34], one can argue that Ad5-mediated activation in the presence of ART may help to reawaken latent reservoirs and be advantageous in a HIV cure strategy. A recent study in which an Ad26/MVA vaccine coupled to a Toll-like receptor 7 agonist was given to ART-controlled simian immunodeficiency virus (SIV)–infected rhesus monkeys highlights the potential of vaccines that induce potent T-cell responses to be used in HIV cure studies. This study demonstrated that time to viral rebound after ART was removed correlated with the strength of the T-cell response, and several animals could control SIV replication below the level of detection in the absence of ART for significant periods of time [35]. It will be challenging to find vaccine epitopes that consistently generate protective T cells across a wide array of MHC haplotypes. A similar concern may be present with the use of vaccines in cure strategies where epitope distribution in archived reservoirs (whether reactivated on ART or emerging after ART interruption) may not match otherwise robust therapeutic vaccine responses. For example, the use of vaccination as part of a “shock and kill” strategy was recently tested in a phase 1b/2a clinical trial. Here, a vaccine consisting of 4 synthetic peptides representing highly conserved regions within the p24 subunit of HIV Gag combined with granulocyte macrophage colony-stimulating factor (Vacc-4x) was used to induce HIV-specific T-cell responses [36] prior to the administration of romidepsin, a histone deacetylase inhibitor that has been shown to induce expression of HIV proteins from latently infected T cells [37]. Data from 6 evaluable individuals shows that there was a modest 38% (95% confidence interval, –67% to –8%; P = .019) reduction in infectious units per million cells compared to samples taken before and after treatment [38]. As the authors of this study acknowledge, there is much room for improvement as the reservoir may need to be reduced by 10 000-fold to prevent viral rebound [39].

Several questions remain to be addressed before vaccination approaches to restore T-cell responses are routinely used to target the reservoir. For one, can vaccine approaches generate enough HIV-specific T cells to efficiently survey and kill infected targets throughout the entire body after a “shocking” latency reversal agent is added? Additionally, will vaccination approaches be broad enough to recognize the entire latent reservoir? Because it will not be feasible to design patient specific vaccines, it is unclear whether a highly potent HIV-based vaccine will ever be able to generate a T-cell response that can recognize all replication-competent HIV within a given individual. Moreover, even though individuals undergoing “shock and kill” approaches will still be receiving ART, there is some concern that newly generated HIV-specific CD4 T cells will be preferentially targeted by emerging viruses during shock and kill therapies, which might augment the HIV latent reservoir. As already noted, HIV is rather adept in escaping the natural immune response. Vaccine approaches may preferentially amplify preexisting responses that already failed to control the virus the first time. While it can be argued that ART dramatically limits the ability of HIV to spread, evolve, and disable the HIV-specific immune response, it remains to be shown whether vaccine-generated T-cell responses will be potent enough to significantly reduce the reservoir. A novel approach under investigation includes the expression of SIV antigens using cytomegalovirus-based vector to generate nontraditional T-cell responses able to mediate long-term control after infection in a number of animals [40, 41]. If these vectors are proven safe and efficacious in humans, it will be important to establish if similar immune responses can contribute to viral control after ART interruption in humans.

ADOPTIVE T-CELL APPROACHES

Instead of relying on the remnants of an exhausted T-cell response that have repeatedly failed to control HIV or the inability of creating a universal HIV immunogen that can guarantee an effective T-cell response, engineering the T-cell response via gene therapy allows investigators to generate HIV-specific T-cell responses that simply cannot be manufactured by the endogenous immune system. Designing an HIV-specific T-cell response to target the latent reservoir via gene therapy to be an effective part of an HIV cure strategy creates the opportunity to determine a priori the specificity (ie, chimeric antigen receptors [CARs]) and infectivity of activated cells (ie, CCR5 deletion) in order to ensure maximal target recognition and long-term retention. The history of cell and gene therapy to treat HIV infection has recently been reviewed [42], but it is important to emphasize that most, if not all, of the pioneering “first in human” adoptive T-cell therapy trials were performed in either HIV-infected or cancer-bearing individuals [43]. This codevelopment of cell therapy in both HIV and cancer has accelerated progress in both, and the lessons learned from each field are likely to propel each further for the foreseeable future. From >2 decades of studies, 3 important questions have emerged: What is the best way to manufacture the T cells? What is the best way to redirect engineered T cells to recognize and destroy HIV-infected cells? And what is the best way to protect infused T cells from deletion?

WHAT IS THE BEST WAY TO MANUFACTURE T CELLS FOR HIV CURE STRATEGIES?

There is no consensus regarding how many T-cell subsets exist, nor is there a consensus as to which T-cell subsets are best suited for adoptive T-cell therapy [44–47]. For applications that require robust T-cell activity initially, but not long-term persistence [48–51], a T-effector memory (T_EM) subset is ideal. However, because these cells are terminally differentiated and do not persist, they are not well suited to generate durable control of HIV replication in vivo. Thus, the prevailing view is to focus on naive (T_N), central memory (T_CM), and stem cell memory (T_SCM) subsets. T_N have long telomeres and the potential to differentiate to other T-cell subsets once infused. While freshly isolated naive T cells have few effector functions, these cells do acquire polyfunctionality when ex vivo expanded for 7–10 days. Importantly, the Riddell group has shown that these cells engraft at high levels without the need for lymphoid depletion [52], which is commonly used to increase engraftment in cancer patients, but will need to be carefully done in HIV-infected individuals due to concerns for toxicity. Second, in tumor models T_CM have shown superior in vivo expansion and persistence compared to T_EM, but T_SCM have demonstrated the most robust antitumor activities showing equivalent or better activity when administered at a 10-fold less cell dose [53, 54]. Importantly, most studies on using T-cell subsets for adoptive therapy have focused on treating tumors, while little, if any, work has been reported for treating HIV infection.

Feasibility, cost, and required dose of T cells should also be discussed when considering how to manufacture T cells for HIV cure studies. Currently, there is not a robust way to purify T cells into T_N, T_CM, and T_SCM subsets using a Good Manufacturing Practice (GMP)–compliant protocol, although it is expected that in the next several years such a strategy will be available [55]. Even when this technology is available, one would have to consider whether the added expense and additional culture time that will be required enables or hinders a HIV cure strategy. Any purification strategy is likely to have a significant impact on the number of T cells that can be initially used in a T-cell engineering process. The optimal dose of engineered T cells remains an outstanding question, and will most definitely depend on the application. For instance, in a cancer patient in which only approximately 10 million T cells could be manufactured to express a CD19-specific CAR, a complete and durable remission was observed due to the tremendous ability of T cells to expand in vivo [56, 57]. However, for HIV cure strategies in which an agent is infused to shock HIV out of latency, so that engineered T cells can recognize and eliminate cells harboring this latent virus, allowing for a high number of T cells that can infiltrate all the areas that HIV may be hiding would appear to be advantageous. Moreover, because T-cell subsets have unique trafficking patterns, one can theoretically argue that a mixture of T-cell subsets would provide greater coverage throughout the body during a latent reservoir targeting schema [58]. Unlike cancer, where there is likely an abundance of antigen targeted by T-cell therapy, an HIV-infected individual successfully treated by ART has low HIV antigen levels and thus we would not expect a robust in vivo expansion of engineered cells. Last, while T cells can be expanded ex vivo for extended periods of time and absurdly large numbers of T cells can be manufactured if enough bioreactor resources are allocated [59, 60], this extended culture is associated with lower T-cell engraftment, persistence, and function [61]. For the reasons listed above, unpurified T cells and keeping these T cells in ex vivo culture for a minimal amount of time may be the best strategy to manufacture T cells for HIV cure studies. “How many T cells should be infused?” is a question often asked during a discussion of strategies to use engineered T cells to cure HIV. However, perhaps a better question to ask is how many T cells need to persist to cure HIV. Thus, it is not how many cells are infused, but how many cells survive and persist that is important, as engineered T-cell persistence correlates with durable control of cancer [62]. T-cell persistence is driven by many factors including T-cell type, method of expansion, state of host lymph system, and presence of cognate antigen. Preconditioning patients with a drug such as cyclophosphamide that temporarily reduces T-cell numbers to provide “space” for the infused T cells is currently being tested in the clinic (ClinicalTrials.gov identifiers NCT02388594 and NCT02225665). Moreover, treatment with interleukin (IL) 7 and/or IL-15 have been considered to improve T-cell engraftment and persistence [63]. It is unclear now whether these approaches will be necessary and/or tolerated by HIV-infected individuals [64]. Efforts to improve T-cell engraftment though improved T-cell culturing methods rather than potential toxic preconditioning methods will improve the feasibility of engineered T cells to aid in HIV cure strategies.

A mixture of engineered CD4 and CD8 T cells are also expected to be required to enable an effective HIV cure strategy. The importance of virus-specific CD4 T cells in resolving viral infection cannot be overstated. Seminal studies using the LCMV model demonstrated that CD4 T-cell help is critical for maintaining CD8 T-cell function during chronic infection. Mice that were transiently depleted of CD4 T cells before infection with strains that induce chronic LCMV infection exhibited more pronounced CD8 T-cell exhaustion and higher viral burden compared with mice having an intact CD4 T-cell compartment [65–67]. Similarly, loss of CD4 T-cell help in response to murine gammaherpesvirus infection resulted in the failure to control viral replication long term [68]. CD4 T cells have a variety of mechanisms that aid other immune cells [69], including expression of CD40-L [70] and production of IL-2 [71] and IL-21 [72], as well as an underappreciated ability to directly limit viral replication via interferon (IFN)–γ production [73] and cytotoxic activity [74]. In adoptive T-cell therapy, CD4 T-cell help has been shown to be important for the maintenance and survival of transferred virus-specific CD8 T cells after bone marrow transplantation [75, 76]. Additionally, expanded autologous CD4 T cells as effectors have shown promising results against metastatic melanoma [77]. In an in vitro model, CD4 T cells isolated during acute HIV infection were able to restore proliferative capacity to exhausted CD8 T cells from patients with progressive HIV infection [11]. These results suggest that strategies with both antigen-specific CD4 and CD8 T cells may be optimal to maximize the T-cell–mediated response potential against HIV infection.

WHAT IS THE BEST WAY TO REDIRECT ENGINEERED T CELLS TO RECOGNIZE AND DESTROY HIV-INFECTED CELLS?

Transducing T cells with an HIV-specific major histocompatibility complex (MHC) class I restricted T-cell receptor (TCR) could be considered as means to deliver a large number of highly functional HIV-specific T cells as part of an HIV cure strategy [78, 79]. The use of natural or engineered high-affinity TCRs would enable T cells to recognize a minute number of HIV-generated peptides [80, 81], which may be of great importance as “shock” approaches may only be able to induce limited amounts of HIV-derived protein. Moreover, the entire HIV coding sequence can be targeted as MHC class I presents intracellular epitopes. Additionally, the use of a high-affinity receptor allows CD4 T cells to recognize MHC class I–presented peptides [82, 83], allowing this approach to also contribute to restore CD4 T-cell help. However, there are significant challenges to the TCR approach including human leukocyte antigen (HLA) restriction, which would force the development of many unique therapies to treat most of the HIV-infected individuals; on-target, off-tissue targeting, which can have disastrous toxicities [84, 85]; although there are epitopes that correlate with control, no one has described a universal, escape-proof epitope, suggesting that multiple TCRs for each HLA allele would be required to effectively target the reservoir; and last, in general, HIV has learned how to escape the natural T-cell response via mutation and downregulation of MHC molecules. For these reasons, the use of CARs expressing T cells would avoid most of these limitations in HIV cure strategies. However CAR-based approaches are limited to targeting HIV Env as it is the only viral protein expressed on the cell surface.

CARs endow T cells MHC-independent recognition of HIV in both CD4 and CD8 T cells. The concept of the CAR was initially developed Zelig Eshhar in the 1980s [86]. In 1991, Art Weiss developed CD4-zeta as a tool for studying T-cell activation [87], and a nearby biotech company, Cell Genesys, proposed that it could be effective for treating HIV. CD4-zeta was the first CAR tested in human. This CAR used CD4 as the HIV Env binding agent and linked this to the CD3 zeta chain so that when CD4 bound Env, the engineered T cell would activate its killing machinery to eliminate the infected cell. Although it had limited clinical success, it paved the way for the development of cancer-based CARs that are now delivering highly efficacious, groundbreaking clinical results in patients with advanced tumors. In the process of developing CART19 (a CD19-specific CAR T cell that earned “breakthrough therapy” designation from the US Food and Drug Administration), a number of notable improvements were made to CAR-based therapy that may allow for second-generation CD4 CARs to be more likely to be effective in the clinic: (1) lentiviral vectors are superior to previously used murine retroviral vectors in terms of safety, transduction efficiencies, vector silencing, and ability to drive expression of a transgene; (2) the incorporation of the costimulatory domains to modify TCR zeta signaling improves both the function and durability of the engineered T-cell response; and (3) advances in ex vivo T-cell expansion technology and patient conditioning prior to cell infusion have improved the number of cells and function of those cells that engraft after infusion [88].

The antigen-binding function of a CAR is usually accomplished by the inclusion of an antigen-specific single-chain variable fragment (scFv) antibody, containing the V_H and V_L chains joined by a peptide linker. This is joined to an intracellular T-cell signaling domain such as 4-1BB followed by the addition of the TCRζ signaling chain that drives T-cell activation following binding of the scFv moiety to its cognate antigen. While there is no shortage of HIV_ENV-specific Abs [89–91], the use of CD4 as our HIV_ENV binding moiety may provide the most universal binding molecule for the detection of infected cells otherwise expressing Env. The added advantage of proposing to use CD4 to retarget CD8 T cells to HIVENV-expressing cells is that it has been tested, albeit unsuccessfully, and the safety data collected in these clinical trials may enable future development of this future generations of this approach [92–94]. Indeed, these studies show the long-lasting engraftment of these cells with a half-life measured >16 years, with no transformation events induced by the integrating vector after >500 patient-years of follow-up [95]. Table 1 summarizes the advantages of a CD4-based CAR. Last, gene therapy approaches allows one to generate HIV-specific T cells from largely functionally competent T cells, effectively resetting the exhaustion clock on HIV-specific T cells within the treated individual. However, once these cells are infused and ART is removed, these engineered T cells may become susceptible to the functional defects described above. Thus, it might be necessary to further engineer these cells to resist these functional defects to ensure long-term control of HIV replication.

Table 1.

Key Features of Chimeric Antigen Receptors that Utilize CD4 to Target HIV-Infected Cells

Feature
Builds upon an existing CAR backbone
Successfully used safely in 4 human clinical trials and resulted in long-term engraftment and control of tumors
Does not require MHC restriction to detect target
Unlikely to be immunogenic as only human sequences are employed
Does not bind well to MHC class II, thus minimizing off-target effects (as noted in trials where CD4 CARs have been used) and trafficking issues
As CD4 binding is an invariant feature of all HIV-1, escape from a CD4-based CAR would likely result in a severe fitness cost to the virus

Abbreviations: CAR, chimeric antigen receptor; HIV-1, human immunodeficiency virus type 1; MHC, major histocompatibility complex.

PROTECTING ENGINEERED HIV-SPECIFIC T CELLS FROM HIV INFECTION

The challenge of infusing genetically modified CD4 T cells in HIV-infected patients is that these cells are likely be targeted and destroyed by the virus, limiting their durability and effectiveness. Likewise, infusing CD8 T cells expressing CD4-based CARs makes these cells targets of HIV infection upon activation (ie, CD8 cells express CCR5). Moreover, infusing engineered CD8 T cells that bind and concentrate HIV near the cell membrane may also lead to infection of these cells due to transient expression of CD4 T cells on CD8 T cells after activation [96–98]. Thus, there is an argument to protect all T cells engineered to target HIV. There are several ways to protect T cells from HIV infection [42, 99]. We will focus on 2 approaches that prevent HIV entry and enable engineered T cells to selectively expand in the presence of HIV infection.

CCR5 ZINC FINGER NUCLEASES

Based on the results of the Berlin patient, investigators at Sangamo Biosciences and the University of Pennsylvania developed a zinc finger nuclease (ZFN) approach that could allow infusion of autologous CCR5-deficient T cells [100]. ZFN-mediated gene disruption is a molecular process for editing the sequence of the human genome at a specific location by (1) creating a double strand break in the genome of a living cell at a predetermined target site and (2) allowing the natural mechanisms of DNA repair to “heal” this break, which often leads to frameshift mutations [100]. The first completed clinical trial using infusion of 10 billion CCR5 ZFN–treated T cells in HIV-infected individuals primarily showed safety [101]. The effect on HIV viremia in individuals treated with CCR5 ZFN–modified T cells, while not achieving complete cure, was encouraging. In the most extreme example of viral control, 1 patient returned to an undetectable viral load 12 weeks after treatment interruption and an initial period of viral rebound. This patient was not an elite controller, as his historic viral load set point was 165 000 RNA copies/mL of blood. However, this patient was a Δ32 CCR5 heterozygous individual, and thus CCR5 disruption by ZFN treatment was more likely to result in full CCR5 deficiency in the T cells of this subject. In follow-up studies, our collaborators at Sangamo Biosciences were able to recruit 7 other Δ32-heterozygous individuals to receive autologous CCR5 ZFN–modified T cells. Remarkably, 2 of the 7 repeated what was observed with the University of Pennsylvania patient, with 1 patient remaining suppressed for up to 48 weeks. However, as with the Boston patients, the virus eventually reemerged in both patients, leading to a restart of ART [102]. Thus, use of CCR5 ZFNs is a safe and feasible way to render T cells resistant to HIV infection. We hypothesize that the antiviral effect we observed in our study was due to a small number of HIV-specific CD4 T cells that became HIV resistant. We further hypothesize that transducing CCR5 ZFN–treated T cells with a CD4 CAR construct will be a powerful combination that could lead to long-term control of HIV infection.

FUSION INHIBITORS

HIV forms a 6-helix bundle that acts like a harpoon to enter cells [103]. Through molecular mimicry, peptides from the heptad repeat 2 domain (HR2) of gp41 can be used to block the formation of this complex, thereby blocking HIV entry. These observations provided the rationale for the antiviral drug enfuvirtide [104]. Von Laer and colleagues fused HR2 peptides of 36 or 46 amino acids (C36 or C46) to a nonspecific carrier protein (the low-affinity nerve growth factor receptor or CD34) as a means to engineer T cells to be resistant to infection [105, 106]. As with the CCR5 ZFN approach, targeting the entry and fusion step of the HIV life cycle has obvious advantages. Inhibiting these steps in primary CD4 T cells would not only protect these cells from infection, but could also inhibit viral spread in an infected host and possibly promote antiviral immune responses and more generalized immune reconstitution. Furthermore, targeting highly conserved regions of the HIV envelope that are required for fusion would decrease the chances of viral escape. Recently, the von Laer and Johnson groups collaborated to study a C46-based construct in a humanized mouse model of HIV infection [107]. While a temporary enrichment of the C46-expressing cells was observed, there were no differences in plasma viral loads between mice receiving the C46 inhibitory vector or the Green Fluorescent Protein (GFP)-expressing control vector. Investigators at the University of Pennsylvania and Sangamo Biosciences speculated that this lack of in vivo protection was the result of limited amounts of the C46 construct at the cell surface, as the inhibitory activity of this protein is likely dependent upon with surface expression levels [108]. They further hypothesized that if an HR2-based inhibitory peptide could be delivered to the precise site of viral entry, its effectiveness could be enhanced. By linking C34 to CXCR4 or CCR5, potent HIV resistance was observed. T cells expressing either C34-CCR5 or C34-CXCR4 selectively enriched in the presence of HIV infection, going from 25% of the T-cell population and ending up >60% after 7–10 days of additional culture. This enrichment was observed against a wide of array of HIV strains, suggesting that this approach will be highly effective in the vast majority of individuals. Importantly, C34-CXCR4, and to a much lesser extent C34-CCR5, constructs protected T cells in an in vivo mouse model [109]. The advantages and disadvantages of using CCR5 ZFNs or C34-CXCR4 approaches to protect engineered T cells from HIV infections are summarized in Table 2.

Table 2.

Comparison of CCR5 Zinc Finger Nuclease– and C34-CXCR4–Mediated HIV-1 Resistance of CD4 T Cells

	CCR5 ZFN	C34-CXCR4
Potency	100% effective CCR5-dependent stains; 0% effective against strains that use co-receptors other than CCR5	Highly effective against all stains tested but not 100% effective against any HIV-1 strain.
No. of HIV-1 resistant T cells that can be delivered	Limited; currently ~30%–50% of the CCR5 alleles can be modified. This equates to 10%–15% of the cells being CCR5 deficient. ~10%–15% of the cells are CCR5 heterozygotes	With optimized lentiviral technology, 70% transduction efficiencies are attainable
Most likely viral evolution pathway	Although not observed yet, we would predict that the virus would evolve to use co-receptors other than CCR5	HIV-1 escaped from membrane-bound C34 by diminishing the time HIV-1 needed to fuse with the membrane. It is unclear whether HIV-1 will escape the same way or differentially to C34-CXCR4
Ability to study engineered T cells in vivo	Difficult; CCR5 is too dim to detect on the surface of peripheral T cells to clearly call which cells are CCR5 positive and which are CCR5 negative. PCR-based assays must be used to determine the number of engineered T cells in a given sample. T-cell functional assay of engineered T cells recovered from patients is not possible	Straightforward; antibodies exist that recognizes C34. This facilitates flow cytometry–based assays that can be used to study the engineered T cells at a single cell level. This greatly improves our ability to study the differentiation, persistence, trafficking, and function of the engineered T cells after they are infused back into the patient.
Immunogenicity	Absence of CCR5 from the cell surface does not generate an immune response	C34 should be immunogenic, although clinical trials to date using membrane-bound C34 have not observed this. If antibodies are generated against C34-CXCR4, then these antibodies have the potential to be very effective against HIV-1. This can only be determined via a phase 1 clinical trial
Genotoxicity and durability	Limited; CCR5 ZFN disruption is a hit-and-run therapy in which the off-target effects are unlikely to be deleterious. Once CCR5 ZFN–mediated disruption occurs, it is a permanent change and nothing needs to be maintained	Lentiviral vectors integrate preferentially into the coding region of active genes. No vector-mediated transformations have been observed in cancer patients treated with CARs, which are expressed by similar lentiviral vectors. Additionally, C34-CXCR4 must continually be expressed to be effective. Our vector design is optimized to continually enable long term expression, and expression of transgenes expressed from vector are still high 5 years and counting from the date of infusion in our cancer patient clinical trials. Nonetheless, vector silencing is a concern with this approach
Ability to tether additional antiviral mechanisms	Limited; we have demonstrated that dual disruption of CCR5 and CXCR4 is possible, but without further improvements in disruption frequency, it is difficult to imagine how this would be clinically useful	Lentiviral engineering is very supple as long as the total payload is <7–8 kb. Multiple genes can be delivered using picornaviral 2A sequences; interfering RNA approaches are possible using Pol III promoters; on/off switches and suicide genes can also be added
Effect of the gene therapy on T-cell function	Loss of CCR5 expression in T cells appears to be very well tolerated in vivo as Δ32 homozygotes have highly functional immune systems; however, the ability of CCR5- deficient T cells to compete with CCR5 WT cells trafficked to sites of infection in vivo has not been carefully studied	CCR5 ZFN and C34 transduced T cells function well with no detectable differences relative to untransduced T cells using in vitro studies. It is unclear how constitutively expressed C34-CXCR4 will affect T-cell function and trafficking in vivo

Abbreviations: CAR, chimeric antigen receptor; HIV-1, human immunodeficiency virus type 1; PCR, polymerase chain reaction; WT, wildtype ; ZFN, zinc finger nuclease.

INSIGHTS INTO HOW CAR T CELL-MEDIATED RESPONSES MAY BE OPTIMIZED INTO COMBINATION IMMUNOTHERAPY STRATEGIES TO IMPLEMENT CLINICALLY

The expectation that combination approaches that may enact pressure on HIV from which escape is more difficult is well founded in the development of ART. It is also known that T-cell adaptive responses are best served by a robust innate response that can act to limit pathogen load and increase the likelihood of success for emerging adaptive responses. For example, adoptive transfer of LCMV-specific CD4 T cells into mice infected with a chronic strain of LCMV were able to restore function to exhausted CD8 T cells and, when combined with anti–PD-1 Ab therapy, led to a striking reduction in viral load [110]. Based on these 2 principles (distinct mechanisms for suppression and complementarity of innate responses to adaptive), it is proposed that human implementation of CAR T cells against viral infection may be best considered as part of a multiattack and phased strategy that may best leverage the chances of durable success and expansion of CAR T cells.

Thus, we envision that best strategies will employ a 2-stage intervention toward HIV remission or cure (Figure 1). The first will involve a phase on ART that debulks as much of the latent reservoir as possible to limit the ability of infectious viruses that can rebound. Examples of first-phase complementary strategies that would provide for a period of suppression and reduce viral measures beyond ART may include IFN-α immunotherapy [111] to activate intrinsic resistance and innate effector responses; IL-15 immunotherapy to expand antiviral immune responses [112]; BCL-2 antagonist Casp8p41 or RIG-I pathway stimulant retinoic acid to reduce latent T-cell reservoirs [113, 114]; infusion of a combination of broadly neutralizing antibodies [115] to both neutralize and elicit antibody-dependent cell-mediated cytotoxicity (ADCC) responses; infusion of dual-affinity retargeting molecules that bind HIV envelope and recruit cytotoxic T cell (CTL) [116]; and infusion of alpha4beta7 integrin antibodies to inhibit HIV replication, activate innate responses, and lower reservoir levels in target tissues [117]. Such “priming” strategies will be expected to reduce the active or persistent reservoir on and off ART, increasing the likelihood that engineered T cells will be effective in destroying the remaining reservoir and/or limiting viral spread by providing an added “immune-mediated viral suppression window” after ART interruption for the CAR T cells to expand rather than get overwhelmed in the presence of a sudden rise in HIV antigen.

Figure 1.

A working model to exemplify combination immunotherapy strategies that may best support chimeric antigen receptor T-cell strategies toward durable control of HIV replication. Abbreviations: ART, antiretroviral therapy; bNAbs, broadly neutralizing antibodies; HIV-1, human immunodeficiency virus type 1; IFN, interferon; NK, natural killer.

While we expect that the list of antiviral strategies able to be considered as complementary to CAR T-cell strategies will continue to expand from those listed above, below we note 2 strategies that already have human safety and initial anti-HIV data to exemplify the fact that combination strategies can be presently pursued in human studies.

IFN-α immunotherapy. IFN-α belongs to a family of type I IFNs produced by leukocytes (and particularly plasmacytoid dendritic cells) as part of the host’s antiviral response able to activate intrinsic antiviral factors and NK cell–mediated responses [118]. The 2 dominant subtypes, IFN-α2a and IFN- α2b, differ by only a few amino acids and are both used clinically with strong antiviral effects [118, 119]. Results of human clinical trials in HIV-infected individuals support a predominantly direct anti-HIV effect of therapeutic doses of IFN-α2 [120–128], which include benefits of delayed disease progression and improved survival in the pre-ART era. In progressive HIV disease, use of type I IFNs in the absence of ART-mediated viral suppression results in a modest temporal decrease in viral load of approximately 0.5 log₁₀ copies HIV RNA per milliliter [127]. Importantly, the antiviral effects of IFN-stimulated gene expression likely differ from the effects of IFN immunotherapy when added to ART. HIV viremia will increase type I IFN receptor signaling well above that produced by IFN immunotherapy by also inducing the host’s IFN-β response. IFN-β has 100-fold greater IFN-αR affinity than IFN-α, and negatively regulates CD8 T-cell responses [129]. In the setting of CAR T-cell activity after ART interruption, we expect that IFN-α2 therapy will intensify HIV suppression by eliciting intrinsic and NK-mediated mechanisms without mediating negative effects on CAR-T-cell responses due to weaker interferon-alpha/beta receptor (IFNAR) interaction [129, 130].

Broadly neutralizing antibodies. The discovery of neutralizing Abs, such as the CD4 binding site Abs VRC01 and 3BNC117, has shown that potent cross-clade Abs can be generated by the human humoral immune response [131, 132]. A first-in-human dose-escalation phase 1 clinical trial of 3BNC117 has been completed, showing that the treatment was well tolerated, and a single 30 mg/kg(–1) infusion reduced the viral load in HIV-infected individuals by 0.8–2.5 log₁₀ copies/mL over a 28-day study period [133]. It is expected that a combination of broadly neutralizing antibodies (bNAbs) targeting different epitopes will maximize inhibition. A similar approach resulted in the rapid suppression of plasma viremia for 3–5 weeks in a subset of chronically simian/human immunodeficiency virus (SHIV)-infected macaques with low CD4 T-cell levels [134]. The combination of 2 or more bNAbs targeting different epitopes in the viral envelope would be expected to lower viral replication and thus complement CAR T-cell strategies [134].

In conclusion, it is expected that a HIV cure dependent on sustained cell-mediated responses may require the combination of added safe and available immunotherapy strategies, inclusive of all 3 arms of the immune response (innate, humoral, and T cells) if possible, to emulate the same efficient sequence of our own natural immune responses (yet maximized for efficiency by immunotherapy) able to eradicate viral infections.

Notes

Acknowledgments. We gratefully acknowledge Colby Maldini, Rachel Leibman, Julie Jadlowsky, and Livio Azzoni for helpful suggestions.

Disclaimer. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health (NIH).

Financial support. This work was supported by the NIH (AI104280, AI117950, AI07632, and UM1AI126620 to J. L. R. and AI094603, AI110434, and UM1AI126620 to L. J. M.). UM1AI126620 is cofunded by the National Institute of Allergy and Infectious Diseases, National Institute of Mental Health, National Institute of Neurological Disorders and Stroke, and National Institute on Drug Abuse.

Supplement sponsorship. This supplement was supported by grants from Merck & Co., Inc. and Gilead Sciences, Inc.

Potential conflicts of interest. J. L. R. has licensed HIV CAR technology to Tmunity Therapeutics, owns equity in Tmunity, and receives sponsored research from Tmunity. L. J. M. reports no potential conflicts. Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.