Lentivirus as a Potent and Mechanistically Distinct Vector For Genetic Immunization

Abstract

T cell mediated immunity is critical for the prevention and control a broad range of infectious diseases and human malignancies. Genetic immunization is a promising approach for the elicitation of T-cell immunity. Recombinant lentivectors are now being developed and evaluated as antigen delivery platforms for genetic immunization and immune engineering. Early results are promising. Third generation lentivectors have been engineered to improve biosafety and reduce anti-vector immune responses. The ability of third generation lentivectors to efficiently transduce non-dividing cells, including dendritic cells, suggests important advantages compared to other antigen delivery platforms. Recent studies suggest that immunization with lentivectors induces remarkably potent and durable primary and memory T-cell immunity. The combination of skin targeted immunization and potentially unique mechanisms of immune induction likely contribute to the potent immunogenicity observed. Taken together, this accumulating evidence supports the ongoing development and clinical translation of lentivector-based genetic immunization strategies.

Introduction

T cell mediated immunity is critical for the prevention and control a broad range of infectious diseases and human malignancies (1,2). The induction of effective T cell immunity remains a daunting challenge. Genetic immunization is a promising approach for the elicitation of T-cell immunity, and genetic immunization strategies are now being developed in animal models and evaluated in clinical trials (1). Non-viral gene delivery strategies offer substantial theoretical advantages over vector mediated approaches. Currently however, viral vectors, and specifically adenovirus based vectors, are generally considered to be the most potent gene delivery vehicles for inducing T cell immune responses (3). While effective gene expression is clearly an important factor for immunogenicity, the clinical development of adenoviral vectors for immunization is limited by several other factors, including vector immunogenicity. The development of dominant immune responses against viral antigens is a major limiting factor for the generation of transgenic antigen specific immune responses. This appears to be a problem both in the context of widespread pre-existing antivector immunity in targeted populations, and the de novo generation of immunodominant T cell immunity against vector antigens in response to vector delivery. In this context, very recent studies suggest that lentivectors are promising viral vectors for eliciting antigen specific T-cell immunity (4–8). Immunization with lentivectors has been observed to induce remarkably potent and durable T-cell immunity. This is likely related to their capacity to transduce non-dividing cells, including dendritic cells in the target tissues, and to enable persistent antigen presentation through high level expression of transgenes and low interfering anti-vector immune responses. Here we review recent progress in the development of lentivector based genetic immunization with emphasis on the immunobiology of lentivectors and unique mechanistic features of lentivector-mediated immunization that may contribute to their observed potency and promising potential for clinical application.

Lentivector mediated genetic immunization induces potent and durable antigen specific immunity

Lentivectors have been under development for use in gene therapy since the mid-1990s. Substantial effort has led to the development of the “third” generation lentivectors now in use. These vectors incorporate several features designed to improve biosafety, and have been engineered to be essentially devoid of sequences encoding viral proteins (9,10). These modifications have greatly increased the feasibility of clinical translation. In addition, the deletion of genes encoding viral proteins has had important collateral immunological consequences as these vectors appear less likely to stimulate anti-vector immunity (11–13). Current efforts to develop lentivector-based immunization are primarily focused on these third generation vectors.

The development of lentivector-based genetic immunization was stimulated by the discovery that lentivectors could effectively transduce non-dividing cells, including dendritic cells, in vitro. Several groups independently are finding that lentivectors are remarkably effective for transducing human monocyte derived DCs in vitro. Reported efficiencies of DC transduction have varied from 30–40% in some studies (14,15), to as high as 70–100% in others (8,16–18). Similar efficiencies were observed in studies focused on the transduction of murine DCs, with transduction efficiencies reported from 40–50% to more than 90% (4,7,16,19). The observed capacity of lentivectors to transduce non-dividing cells suggests that both mature and immature human DCs could be transduced by lentivector. This has generally been shown to be the case, but under some conditions higher transduction efficiencies can be obtained with immature DCs (14,16). It is not clear why such wide variations in transduction efficiency have been observed, but it is likely that these differences relate to technical differences between investigators and systems, rather than biologic variables. Key factors that vary between studies include; 1) methods of lentivector preparation and/or titer determination, 2) differences in m.o.i. used for transduction, and 3) differences in promoters used to drive transgene expression.

Compelling evidence suggests that lentivectors have features that will be advantageous for genetic immunization compared to other vectors now under development. In direct comparisons, lentivectors have been reported to be 2–10 times more effective in transducing human DCs and mouse DCs than adenovectors (20). In fact, these studies suggest that to achieve similar transduction efficiencies, 10–100 times more adenoviral vector is required (20). More importantly, unlike adenoviral mediated transduction, the transduction of human and mouse DCs with lentivectors generally does not alter the capacity of the DCs to respond to immune modifying stimuli, or to efficiently stimulate antigen specific T cell immunity (4, 21). Several studies now demonstrate that both human and mouse DCs transduced with lentivector can efficiently stimulate antigen-specific CTL activity in vitro (7,14,16,19). Further, adoptive transfer immunization with ex vivo transduced DCs has been shown to induce potent T cell immunity against lentivector encoded antigens in mice (4,7,16,20,22). This antigen specific T cell immunity prevented the development of tumors in mice challenged with lethal doses of tumor cells. In direct comparative studies, immunization with lentivector transduced DCs stimulated substantially stronger and longer-lasting CD8+ T cell responses than immunizations with peptide/protein pulsed DCs as determined quantitatively by in vivo killing assays and IFN-γ secretion (4). Further, immunization with lentivector transduced DCs elicited stronger CD8+ T cell immunity than immunizations with adenoviral vector transduced DCs (20). Functionally, the potent T cell immune responses elicited by lentivector transduced DCs have been shown to protect mice from viral infections (23) and significantly slow tumor growth (4).

The observed transduction efficiencies and immunogenicity of lentivector transduced DCs clearly supports the further development of these ex vivo DC engineering approaches. However, existing technologies for generating and maintaining DCs ex vivo are laborious and expensive. This remains a considerable barrier to widespread application for clinical immunotherapy, and severely limits the feasibility of this approach for population based preventive immunization. On the other hand, the ability of lentivectors to efficiently transduce DCs presents an attractive theoretical advantage for their development as vectors for direct immunization. Efforts to develop and evaluate strategies to induce immunity by direct in vivo injection of antigen encoding lentivectors are underway. Consistent with ex vivo results, recent studies show that direct administration of lentivector via intravenous or subcutaneous injection results in transduction of antigen presenting cells (APCs) including DCs in the spleen (24), and in the draining lymph nodes (6,25,26). Importantly, direct injection induces potent CTL activity that is similar to or higher than that induced by delivery of ex vivo transduced DCs (6,27). Induction of T cell immunity by direct immunization with lentivector has been demonstrated in a number of antigen systems including TRP2 (26), Her2-neu(26), NY-ESO (28), Melan-A and HLA-Cw (6,29), and for the model antigen OVA (25,27,30).

Accumulating evidence suggests that immune responses induced by lentivectors may have unique features compared with those stimulated using more conventional immunization strategies. Interestingly, immunization by direct injection with lentivector appears to be less dependent on CD4⁺ helper T cells for the generation of primary and memory CTL responses (27). Importantly, CD8⁺ T cell immunity induced by lentivector immunization can persist for extended periods of time compared to that induced by vaccinia vectors (25). These long-lasting antigen specific CD8⁺ T cells behave like true memory CD8⁺ T cells in that rapid antigen-specific proliferative responses occur in response to a second exposure to the same antigen (27). It has also been reported that more CD127⁺ antigen specific CD8⁺ T cells are detected at the peak of primary responses after lentivector immunization compared to peptide based immunization, supporting the conclusion that lentivector immunization favors the development of memory T cells (29). Importantly, it has also been shown that CD8⁺ T cell immunity can be recalled by repeated immunization with the same lentivector, suggesting that lentivector induces little if any interfering antivector immunity compared to other viral vectors. In an in vivo immunization model, direct comparisons demonstrated that while repeated injection of adenovector induced antivector immunity that limited immunity against transgenic antigen, effective immune responses against transgenic antigens could be induced even after exposure to injected lentivector (25). These experiments suggest that dominant anti-vector immune responses that limit the effectiveness viral vectors like adenovectors may be obviated by lentivector-mediated antigen delivery. Collectively these results suggest that direct immunization with recombinant lentivector not only stimulates potent effector CD8⁺ T cells, but also long-lasting memory T cells that can be recalled with boosting immunizations.

Though most studies thus far have focused on the analysis of CD8⁺ T cell responses, immunization with lentiviral vectors has also been shown to induce potent CD4⁺ T-cell mediated immunity and humoral responses. Two recent studies demonstrate that lentivectors encoding secreted antigens or fusion antigens targeted to the MHC Class II presentation pathway induce effective CD4⁺ T cell responses (27,30). Further, it has been shown that immunization with a very low dose of recombinant lentivector expressing envelope E-glycoprotein of west Nile virus induces strong and long-lasting antibody responses, and effectively protects mice from infection by virulent West Nile viruses for a prolonged period of time (31).

Collectively, rapidly accumulating data strongly suggests that lentivector–based immunizations stimulate potent humoral and cellular immune responses. This has been shown for immunization both by adoptive transfer of genetically engineered DCs, and by direct in vivo injection of lentivectors. Importantly, this immunity is long lasting and leads to the development of long-term memory responses. Functionally, these immune responses have been shown to inhibit tumor growth and prevent infections. This potent immunogenicity, along with improved safety profiles of the vectors, the general lack of pre-existing antivector immunity in targeted populations, and the relative absence of antivector immunity following immunization, are encouraging for the clinical development of lentivector-based immunotherapies and vaccines.

Unique mechanistic features of lentivector-mediated immunization

Despite extensive efforts to develop recombinant viral vectors for genetic immunization, the mechanisms by which immune responses are primed and maintained have yet to be defined and remain a subject of considerable interest. When antigens are delivered by recombinant viral vectors, which cells in the target tissues are transduced? Are DC subsets targeted? What is the source of antigen for antigen presentation? Which DCs present antigen to T-cells? Is the duration of gene expression an important factor for the induction or maintenance of antigen specific immune responses? What are the relative contributions of direct presentation and cross-presentation to T-cell activation, and how do these pathways affect the nature and durability of the resulting immune responses? A better understanding of these mechanisms will be essential for the rational design of effective immunization approaches (32). The answers to these questions are likely to be complex. Mechanisms of immune induction may vary depending on both vector-specific features, and the routes and circumstances of vector delivery. Several new studies are beginning to address these critical issues. Comparisons between lentivectors and other more extensively studied recombinant viral vectors suggest important differences in mechanisms of immune induction. These differences may help explain the remarkable potency of lentivectors observed thus far and help us better understand the mechanistic factors that shape the nature of the immune response during priming.

Collaborative and integrative models of antigen presentation

CD11c+ DCs are essential APCs for T cell priming, but the CD11c+ DC phenotype encompasses a heterogeneous population of cells (33). In mice, there are at least 7 different DC subsets identified by patterns of cell surface molecule expression (34). These subsets can be grouped collectively into categories of “tissue derived” or “blood derived” DCs subsets. The relative contributions of DC subsets to T-cell induction remain to be determined.

Traditionally, the mechanism of T cell priming in response to cutaneous immunization has largely been explained by the classical paradigm of skin immune function (35,36). In this paradigm skin DCs, including dermal DCs and epidermal Langerhans cells (LCs), are thought to acquire antigen in the skin, and in response to environmental signals they mature and migrate to draining lymph nodes. DCs are functionally plastic and present antigens in the context of immune skewing signals that are responsive to signals from the environment. In the draining lymph nodes they present antigens obtained in the periphery to lymph node T cells in the context of immune skewing signals that reflect conditions in the periphery. In this way, skin-derived DCs “integrate” antigen specificity and environmentally responsive immunomodulation.

Recent studies have questioned this classical paradigm of direct T cell priming. In a series of experiments, Heath and colleagues demonstrated that although migrating skin derived DCs are found in the draining lymph nodes after cutaneous HSV infection, they did not appear to be the antigen presenting cells (APCs) that primed naïve T cells (37,38). Rather, their studies demonstrated that a CD8+ DC population that was not skin derived, but instead resident in the draining lymph nodes, was the primary DC population capable of presenting antigen to CD8+ T-cells. Parallel observations have subsequently been made in a number of circumstances including; 1) skin infection by HSV, influenza (IAV), and vaccinia virus (VV) (39,40), 2) lung infection by HSV and IAV (41), and 3) after parenteral infection with LCMV virus or Listeria monocytogenes bacteria (42). These results led to a proposed revision of the classical paradigm. Specifically it was proposed that skin DCs acquire antigen in the periphery, and then deliver antigen to lymph node resident DCs. These lymph node resident DCs then cross present antigen to naïve T cells, in a “collaborative” form of antigen transfer and presentation between tissue resident and lymph node resident DC populations (43–45). This “collaborative” cross priming model offers the alluring possibility that antigen transfer between DC populations could efficiently distribute antigen to a large network of lymph node resident DCs for presentation. It would however, require an efficient and as yet undiscovered mechanism for the functional transfer of small quantities of antigen to multiple DCs. Further, a collaborative model would seemingly dissociate the “sentinel” function of skin DCs, which enables them to utilize their functional plasticity to skew immune responses based on environmental conditions, from their antigen presentation function. The collaborative model would require additional mechanisms to faithfully transfer and translate inflammatory or tolerogenic signals from the periphery into appropriate T-cell responses.

Indirect evidence, either supporting or challenging the “collaborative” model, can be found in a variety of studies. For example, some studies suggest that direct contact with pathogens or TLR ligands is necessary to license DCs to fully activate the effector function of T-cells (46,47). This raises the question of how the lymph node resident DCs are licensed to prime naïve T cells if there is no direct contact with cutaneous pathogens. On the other hand, it has been shown that DCs can be activated through ligation of TLR 3 by engulfed dsRNA derived from infected cells (48), and that TLR4 ligands phagocytosed by DCs influence processing and presentation of associated Antigens (49). These studies indirectly suggest that LN resident DCs could be “cross-licensed” provided that sufficient TLR ligands already present in vesicles of arriving DCs could be efficiently transferred into recipient lymph node DCs.

Why is it that DCs migrating from the sites of skin infection apparently fail to prime naïve T cells in these studies? DCs migrating from sites of “danger” typically are mature and licensed, and capable of priming T-cell responses (35). Interestingly, the replication competent viruses studied thus far are generally either cytopathic (HSV, VV, IAV) or, like LCMV, have well-described mechanisms for immune evasion that directly or indirectly altered the antigen presentation function of infected DCs (32,50–54). It is likely that in many cases viral infection compromises the antigen presenting function of infected DCs, and that cross-presentation by other uninfected DCs including lymph node resident CD8+ DCs (55), becomes an alternative compensatory mechanism for immune induction. This interpretation predicts that in vivo transduction of peripheral DCs, such as skin DCs, by recombinant vectors that do not alter the antigen presentation function of transduced cells would enable these DCs to directly prime naïve T-cells. This would enable migrating peripheral DC populations to maintain the “integration” of immunomodulation and antigen presentation functions. Under these circumstances, readily accessible skin DCs could be targeted to induce and control the nature of systemic immune responses, making cutaneous genetic engineering an attractive strategy for systemic immune modulation.

Targeting the skin immune system

The skin contains a readily accessible network of DCs and other cells, including keratinocytes and mast cells, with innate immune function that collectively play an important role in initiating immune responses and in immune regulation (36). The high density, accessibility, and functional plasticity of skin DCs contributes to their function as “sentinels” for detecting and responding to “danger” and makes them attractive targets for vaccine delivery. Historically, the empiric immunogenicity of skin has made it a preferred target for vaccine delivery (36). Recent studies of the immunogenicity of the current influenza vaccine in human subjects demonstrate that intradermal injection of antigen results in the induction of more potent immune responses with lower doses of antigen than other routes of immunization (56–58). These studies support the ongoing development of skin targeting immunization strategies

To specifically evaluate the role of skin DCs in immune induction, we immunized mouse skin with recombinant lentivectors, and then evaluated transgene expression and antigen presentation in specific DC subsets isolated from the draining lymph nodes. Using either RT-PCR or detection of luciferase expression to measure transduction, we found transgene expression predominantly in skin derived DCs (phenotypically defined as CD11c⁺B220⁻CD8⁻CD11b⁺DEC205⁺) (25). Importantly, these skin derived DCs were the predominant DC population capable of priming naïve CD8⁺ T cells (25). In the same experiments, and consistent with previous reports, skin DCs from mice immunized with recombinant vaccinia vector did not prime naïve T-cells (39). In contrast, in these animals only the CD8⁺ lymph node resident DC subset primed naïve CD8⁺ T cells, even though transgene expression was found in other DC subsets as well (25). Other recent reports are consistent with this observation, and suggested that dermal DCs likely play an important role in priming T cell immune responses (59–61). Taken together these results suggest that cutaneous delivery of lentivector enables transduction of skin DCs, and that skin DCs can effectively present transgenic antigens to prime naïve T cells. These results support the “integrative” model of DC function whereby skin DCs both present antigen and deliver environmentally programmed signals that modulate or skew immune responses. Integration of these functions in the same cell, in this case a skin DC, could effectively link antigen-specificity with environmentally responsive effector function, enabling the induction and control of physiologically appropriate antigen-specific immune responses.

In aggregate, the currently available evidence suggests that in the case of vector mediated skin immunization, the mechanisms of T cell priming are largely dependent on the effect of the vector on the infected DCs. If the vector is lytic, or significantly disables antigen presentation function, cross-presentation by uninfected DCs is likely to be the dominant antigen presentation mechanism. This pathway may be particularly effective in the case of lytic vectors, as several studies suggest that antigen can be efficiently transferred from stressed or dying cells to DCs for cross-presentation (55,62). Transduced, dying DCs could serve as lymph node resident antigen factories capable of producing and delivering, or “spreading”, antigen to surrounding dendritic cells for cross-presentation. Alternatively, if the vector enables infected DCs to survive and maintain their antigen presentation function, viable antigen expressing DCs could directly present endogenously synthesized antigens to T-cells. Under these circumstances, the DCs presenting antigen would also be “experienced” in that they would have been exposed to environmental conditions in the periphery, and would be “licensed” to induce appropriate immune responses. It is important to note that these two mechanisms are not mutually exclusive and relative contributions of either mechanism to T-cell priming could vary over time. It has been observed that at early time points, T-cells cluster around vaccinia infected DCs, suggesting that in early infection these infected DCs are viable and capable of directly presenting antigen, even though CD8⁺ LN resident DCs become the dominant antigen presenting cells 48h after vaccinia infection (63). It is also likely that transduced non-APCs, such as keratinocytes in the case of skin immunization or myocytes in the case of intramuscular immunization, can provide an important source of antigen for cross-presentation by either infected or uninfected DCs. Lentivectors appear to be relatively unique among the viral vectors studied to date in that DC transduction appears to result in high levels of transgene expression without altering the viability or antigen presentation function of the infected DCs. The observation that comparatively low titers of lentivector injected into skin can stimulate potent and persistent immune responses suggests that the combination of lentivector gene delivery and skin targeted immunization is a promising approach for clinical development.

Persistent antigen presentation

Several recent studies suggest that the duration of antigen presentation may be an important determinant in the potency and persistence of vaccine elicited T cell immunity. Although in vitro experiments suggest that a few hours exposure to antigen is sufficient to drive antigen specific T cell proliferation, it is general accepted that prolonged antigen presentation in vivo is required to stimulate “fit” CD8+ T cell responses and T-cell memory (64,65). The enhancement of DNA vaccine immunogenicity by co-delivery of DNA encoding antigen and an anti-apoptotic protein supports the theory that prolonged antigen presentation is beneficial for inducing potent and persistent T cell immunity (66). Prolonged presentation of antigen has also been shown to promote memory CD4⁺ T cells specific for influenza antigens, and prolonged low-dose antigen exposure has been shown to be critical for the generation of long-lived memory CD8⁺ T cell responses (67). The issue remains controversial however, as results vary between experimental systems. In the setting of Listeria Monocytogenes infection, the duration of antigen presentation in vivo was shown to be limited to 3 days, and inversely correlated with the appearance of CTLs, suggesting that antigen-specific CTLs may participate in the clearance of the corresponding antigen presenting cells (68). On the other hand, after HSV infection, antigen presentation appears to persist even after CTL activity reaches its maximum (69) possibly as a result of protection of APCs by CD4⁺ T cells (70).

In the case of lentivector based immunization, antigen presentation is prolonged in vivo. Following skin injection of lentivector, antigen presentation can be detected in the draining lymph node of mice 3 weeks after immunization (25). In contrast, antigen presentation after vaccinia vector skin immunization decreased rapidly and was barely detectable after 5 days (25). In fact, the magnitude of antigen presentation 3 weeks after lentivector immunization was substantially greater than that observed 5 days after vaccinia vector immunization. It is likely that non-cytopathic vectors including lentivector will enable prolonged antigen presentation compared to cytopathic vectors such as vaccinia vector. Importantly, in these studies prolonged in vivo antigen presentation correlated with more potent and prolonged CTL effector function in lentivector immunized mice, suggesting that the prolonged in vivo antigen presentation induced by lentivectors may be another unique factor that contributes to the observed potency and persistence of CD8+ T cell immunity.

Clinical feasibility of lentivector mediated genetic immunization

Compared to other recombinant viral vectors, the use of recombinant lentivectors is relatively new in the field of genetic immunization, but the studies we have reviewed suggest that these vectors have important immunological advantages including the induction of immune responses with high potency and durability, and low interfering vector specific responses. Very recent efforts suggest that these vectors can be engineered to be even more effective through modifications that enable cell-specific targeting and/or inducible gene expression (71,72). However, the use of retroviruses in humans raises serious safety concerns. Among these is the possibility of the generation of replication-competent lentivectors, and the potential for oncogensis. Recent results suggest that lentivector insertion into the host genome has relatively low oncogenic potential, partially alleviating this concern (73). Further, third generation lentivectors incorporate the self-inactivating LTR promoter that reduces the likelihood of generating replication competent lentivirus, and this complication can be carefully monitored in vivo (74). Now, non-integrating lentivectors are being developed that may further improve the safety profile of these vectors. These non-integrating lentivectors contain a mutated form of integrase in the viral particles that prevents the integration of the lentivector genome into host chromosomes. Importantly, the level of gene expression driven by the non-integrating lentivectors remains relatively high and long-lasting, as long as the transduced cells are not dividing (75,76). This appears to be compatible with effective gene expression in DCs, and suggests that immunization with non-integrating lentivector may result in similarly potent immune responses as those induced by integrating viruses. This possibility has not yet been addressed in animal models, and is a critical issue for future development. Finally, production of clinical grade lentivector in sufficient quantities for clinical applications is currently problematic. Efforts are underway to establish improved lentivector packaging cell lines, an example of which has recently been described (77). Similarly, ongoing efforts to standardize large scale production of lentivectors should facilitate broader applications in clinical trials (78).

With these and other developing modifications, the risk-benefit balance is beginning to favor the further development of lentivectors for clinical applications. Very recently, results from a small clinical trial evaluating the safety a conditionally replicating HIV derived vector have been reported (79). In this study autologous T-cells modified to express an antisense gene against the HIV envelope were adoptively transfer into chronically infected HIV patients. After 21–36 months of observation the investigators found sustained gene transfer and no evidence for insertional mutagenesis or the generation of replication competent lentivector. Immune function improved in four of the five subjects. With the initiation of additional clinical trials the potential risks associated with the clinical application of lentivectors will be further evaluated. The combination of evolving vector modifications, a deeper understanding of the mechanisms of lentivector mediated genetic immunization, and the availability of additional clinical trial results is likely to drive the ongoing development of clinically applicable genetic immunization strategies for the prevention and therapy of cancer and infectious diseases.