Dendritic cell-based human immunodeficiency virus vaccine

Abstract

Dendritic cells (DC) have profound abilities to induce and coordinate T-cell immunity. This makes them ideal biological agents for use in immunotherapeutic strategies to augment T-cell immunity to HIV infection. Current clinical trials are administering DC-HIV antigen preparations carried out ex vivo as proof of principle that DC immunotherapy is safe and efficacious in HIV-infected patients. These trials are largely dependent on preclinical studies that will provide knowledge and guidance about the types of DC, form of HIV antigen, method of DC maturation, route of DC administration, measures of anti-HIV immune function and ultimately control of HIV replication. Additionally, promising immunotherapy approaches are being developed based on targeting of DC with HIV antigens in vivo. The objective is to define a safe and effective strategy for enhancing control of HIV infection in patients undergoing antiretroviral therapy.

Rationale for immunotherapy in HIV infection

Control of HIV infection by combination antiretroviral therapy (ART) has greatly decreased the morbidity and deaths due to AIDS. However, although ART results in an increase in CD4⁺ T-cell count to more than 200 cells per mm³ in 95% of people, recovery of CD4⁺ T cells is limited in a significant portion of individuals due to incomplete control of HIV replication [1]. HIV persists in reservoirs in blood and gut-associated lymphoid tissue (GALT) during ART despite undetectable HIV RNA (<50 copies per mL) in blood [2]. Ironically, prolonged suppression of viral load and antigen burden by ART leads to decreases in anti-HIV CD4⁺ and CD8⁺ T-cell frequencies [3–5]. This T-cell immunity cannot eliminate virus replication, which can recrudesce due to antiviral drug resistance or to stopping ART secondary to toxicity. Moreover, ART is costly, can have severe adverse effects and requires life-long usage. Therefore, based on the classic paradigm that antimicrobial drugs work in concert with host immunity to control infection, immunotherapies (aka immunovaccines) are being developed to enhance anti-HIV T-cell immune responses during ART.

The basic principle of immunotherapy for HIV infection is that `more is better', i.e. a greater magnitude of T-cell immunity and breadth of targeted antigens will improve control of HIV replication, and in the best of all worlds, allow for discontinuing ART. Unfortunately, there is no single measure of T-cell immunity or constellation of immune parameters that stand as criteria for host control of HIV infection. Delineating such `correlates of immune protection' in HIV infection has been equivalent to the search for the Holy Grail amongst HIV immunologists. A plethora of studies have shown that several T-cell phenotypic and functional characteristics are associated with, and in some cases correlated with, long-term survival or slow progression of HIV infection [6]. Of note is that chronic immune activation and inflammation in HIV infection has been linked to translocation of microbial products from the gut to the blood [7]. This provides fertile ground for HIV replication and disease progression as well as leading to exhaustion of the regenerative capacity of T cells, termed replicative senescence [8]. The most desirable immunotherapy for HIV infection would therefore limit its T-cell activation properties to those that enhance T-cell control of the infection. Indeed, the breadth of interferon γ (IFNγ) -producing CD8⁺T cells targeting multiple Gag consensus peptides correlates with lower viral loads and higher CD4⁺ T-cell numbers in blood [9]. A second immune parameter recently linked to slow progression of HIV infection is polyfunctional CD8⁺ and CD4⁺ T-cell activity, defined as production of more than one immune mediator by the same cell, i.e. cytokines [IFNγ, interleukin 2 (IL-2) and tumour necrosis factor α (TNF α)], a chemokine [macrophage inflammatory protein 1β] and the cytotoxic degranulation marker, CD107a [10]. This is being combined with a new paradigm of immune correlates of protection in `elite controllers', i.e. individuals who suppress their HIV infection without drug therapy. These individuals have higher CD8⁺ T-cell polyfunction and proliferation, and IFNγ and IL-2 production by CD4⁺ T cells, than patients with progressive HIV infection [6]. Thus, the efficacy of an immunotherapeutic vaccine for HIV infection should at least in part depend on its capacity to limit immune activation and viral replication through induction of polyfunctional CD8⁺ and CD4⁺ T cells specific for a broad range of Gag epitopes.

In defining immune parameters of HIV infection, a basic principle of immunology has largely been ignored, i.e. the role of professional antigen presenting cells (APC) – dendritic cells (DC), monocytes/macrophages and B lymphocytes. Since the first recognition of a loss of anti-HIV CD8⁺ and CD4⁺ T-cell reactivity in HIV infection in the 1980s, most studies have not considered the fundamental principle that T-cell cognate responses require presentation of antigen by major histocompatability complex (MHC) class I or II molecules on APC. This is complicated by the requirement of CD4⁺ T-cell help plus host membrane and soluble factors produced by APC, for optimal efficiency of CD8⁺ T cells. Once activated and armed to kill, the CD8⁺ cytotoxic T lymphocytes (CTL) seek out, identify and lyse infected cells by MHC class I-dependent mechanisms.

DC for immunotherapy of HIV infection

The most potent of the professional APC are DC of myeloid origin. These DCs are geographically positioned as sentinels, detecting `danger signals' and linking innate and adaptive immune responses [11]. Because of the exceptional ability of myeloid DC to activate T-cell immunity in response to microbial pathogens, these cells have been exploited as ex vivo and in vivo tools for immunotherapy of HIV infection [12]. In 1996, long before translation into HIV clinical trials, myeloid DC were first used in immunotherapy of cancer [13]. The main breakthrough that allowed ex vivo immunotherapeutic approaches in humans was growth of myeloid DC from blood monocytes by culturing in IL-4 and granulocyte-macrophage colony-stimulating factor (GM-CSF) [14]. This technique for generation of monocyte-derived DC (MDDC) has been scaled up to obtain large quantities of commercial grade MDDC for multiple injections [15].

Notably, there is a profound plasticity in the properties and functions of myeloid DC in vivo that can be recapitulated in vitro [16]. Using a simplified model, myeloid DC develop from CD34⁺ stem cell precursors in the bone marrow and migrate to tissues where they, in an immature or steady-state, attain distinct phenotypic and functional characteristics. The two basic subsets of tissue DC are Langerhans cells and interstitial DC, each containing different DC subsets within them [17]. Most myeloid DC lack expression of T cell, B cell, monocyte and natural killer cell markers, and express CD11c in blood. There is no single distinct phenotypic marker for myeloid DC. The phenotypic and functional properties of myeloid DC are determined in large part by their inflammatory cytokine milieu that is triggered by infection. This has been reproduced in vitro to derive DC of increased functional potency. Thus, whilst GM-CSF and IL-4-generated MDDC are the current choice for immunotherapy, other cytokine milieus can differentiate MDDC with potent immunostimulatory properties, e.g. IFNα, TNFβ and IL-15 [18].

A series of issues in addition to the type of DC are critical in the use of DC for immunotherapy of HIV infection (Table 1). Initial recognition of invading pathogens by DC is a critical event in antigen processing, and predominately takes place via pattern-recognition receptors (PRR), which recognize pathogen-associated molecular patterns expressed by various microorganisms [19]. These PRR are primarily Toll-like receptors (TLR) and C type lectins. DC differentially expresses both of these molecules. For example, myeloid DC in the blood express predominantly TLR2 (recognizing peptidoglycan) and TLR3 (recognizing dsRNA), enabling them to induce T-cell responses to pathogens. The C-type lectins Langerinand DC-specific intercellular adhesion molecule-3-grabbing nonintegrin (DC-SIGN) are expressed by Langerhans cells and interstitial DC, respectively. Ligands for these PRR could be powerful adjuvants in immunotherapy of HIV infection.

Table 1.

DC immunotherapy for HIV infection: major treatment issues

Type of DC	Myeloid DC, plasmacytoid DC
	Langerhans cells, interstitial DC
DC maturation, processing and migration	Toll-like receptor agonists
	CD40L,
	Cytokines IFNγ, TNFβ, IFNα, IL-1β
	PGE2
	C-type lectins (DEC 205, DC-SIGN)
Route of delivery	IV, SC, ID, intranodal
DC lifespan	Anti-apoptotic mediators (Bcl-2, Bcl-xL, Akt1, CpG, TRANCE)
T-cell suppression	Treg (TGF, IL-10), PD1, CTLA-4
HIV infection	ART
	Viral immune escape variants
HIV antigen	Consensus versus autologous strains
	Whole inactivated virus
	Infected apoptotic cells
	Exosomes
	Recombinant proteins
	Peptides
	RNA
	DNA
	Microbial vectors
	VLPs
Immunological end-points	Increase in polyfunctional T cells
	Increase in breadth and magnitude of CTL to Gag, other HIV proteins
	Normalize T-cell numbers and activation levels
Virological end-points	Lower viral set-point

DC, dendritic cells; IFN, interferon; TNF, tumour necrosis factor; IL, interleukin; PGE2, prostaglandin E2; TGF, transforming growth factor; PD1, programmed death-1; CTLA-4, cytotoxic T lymphocytes antigen 4; DC-SIGN, DC-specific ICAM 3-grabbing nonintegrin; ART, antiretroviral therapy; VLP, virus-like particles; IV, intravenous; SC, subcutaneous; ID, intradermal; Treg, T regulatory cells.

An additional type of DC is termed plasmacytoid in that it appears to be of lymphoid lineage [20]. These CD11c-negative DC are major producers of type I IFN [21], and also have antigen processing and presentation capabilities [22]. Because plasmacytoid DC are only available in small quantities and are difficult to culture or selectively target in vivo, they have not been used in immunotherapy of HIV infection. An intriguing finding is that HIV infection of plasmacytoid DC activates TNF-related apoptosis-inducing ligand (TRAIL) and thereby confers cytotoxic properties to these DC [23]. Similar TRAIL-dependent `killer' DC can be activated by IFNγ or IFNα treatment of blood CD11c⁺ myeloid DC, and mediate apoptotic death of tumour cells [24]. This new form of DC is being considered in treatment of cancer and could be applied to HIV immunotherapy.

In sum, the use of DC for HIV immunotherapy exploits the natural pathways of antigen recognition and processing in host immune control. The basic rationale for DC immunotherapy is to control HIV infection more effectively and lower the viral `set-point' after removal of ART, i.e. when viral load remains relatively stable (Fig. 1). This is a tall order, given the complex parameters in HIV infection that underlie the exhaustion and premature ageing of HIV-specific CD4⁺ and CD8⁺ T cells. Notably, antibodies mediating neutralization, cytotoxicity, complement-dependent lysis and other antiviral reactivities are also regulated by DC [25] and are important in HIV infection [26, 27]. Research on this aspect of DC immunotherapy for HIV infection is currently lacking, and is not addressed in this review.

Fig. 1.

Anti-HIV T-cell responses in chronic infection: Can DC immunotherapy control residual HIV infection? HIV, initial infection with HIV; set point, when viral load remains relatively stable; DC Rx, DC immunotherapy; ATI, analytical treatment interruption; ART, antiretroviral therapy; CTL, cytotoxic T lymphocytes.

DC maturation is a requisite for activating anti-HIV T-cell responses

To achieve optimal T-cell activation against HIV by DC immunotherapy, it is critical to regulate the state of maturation of the DC. A key issue is that immature DC are poor stimulators of T cells and can be tolerogenic. As used here, DC maturation is defined as an effector function [28], with increased immunogenic capacity for activation of antigen-specific, T-cell immune responses. This process includes enhanced expression of surface molecules such as MHC class I and II and T-cell co-stimulatory molecules CD40, CD80, CD86 and CD83, and a decrease in antigen uptake capacity [29]. As DC mature, they migrate to lymphoid organs and present antigen to naïve T cells in the context of MHC class I and II molecules, provoking antigen-specific T-cell differentiation [30]. The DC secrete cytokines that promote CD4⁺ T-cell differentiation into T helper 1 (Th1), Th2 or T regulatory cells (Treg) or Th17 cells [31]. IL-12p70 is essential for generation of Th1 responses, whilst low IL-12 secretion and high IL-10 production is critical for generating Th2 responses [32]. Sequential triggers of antigen-specific T-cell responses by mature DC result in adaptive T-cell responses that involve TLR and cytokine signalling [25]. Thus, treatment of DC with combinations of TLR agonists, such as poly I:C (TLR3 agonist), imidazoquinolines (TLR7/8 agonists) and CpG oligodeoxynucleotides (TLR9 agonists) [33–35], and inflammatory cytokines, such as IL-1β, TNFα, IFNγ and IFNα [36], induce potent, antigen-specific T-cell responses. We have found that maturation of DC with recombinant CD40L, a member of the TNF superfamily expressed on activated CD4⁺ T cells, stimulates maximum anti-HIV CD8⁺ T-cell reactivity [37]. CD40L primes DC for production of high levels of IL-12 and other cytokines that preferentially elicit Th1 responses leading to induction of CD8⁺ CTL [38]. A trimeric, recombinant CD40L (Amgen, Seattle, WA, USA) is the most potent maturation factor for DC, and together with IFNγ results in extraordinarily high levels of IL-12 in vitro [39]. Combination of IFNγ with CD40L also upregulates IL-27 production and Th1 cell recruitment, and decreases IL-10 production [40].

Interestingly, we have recently found that blocking of IL-12 in CD40L-matured DC cultures by anti-IL12 monoclonal antibodies does not alter the ability of the DC to activate HIV-specific, memory CD8⁺ T-cell responses [37]. This suggests that DC can directly activate anti-HIV memory CTL without requiring IL-12. Nevertheless, the capacity to produce large quantities of IL-12 by DC is probably a requisite for effective immunotherapy of HIV infection, given the important role of IL-12 in priming of Th1 cells, which in turn enhance activation of CD8⁺ T cells [36].

The efficacy of DC modality in immunotherapy of HIV infection is dependent on whether these DC are not overstimulated in vitro. Such overstimulation with maturation factors in vitro could exhaust their capacity to enhance T-cell immunity after re-administration in vivo. The optimal DC preparation for immunotherapy would allow for augmentation of CD4⁺ and CD8⁺ T cells immediately after infusion [41]. In this model, in vitro maturation would activate DC for IL-12 production upon subsequent interaction with Th1 cells in vivo. The IL-12 produced by the infused DC should subsequently polarize CD4⁺ Th1 cells, which preferentially produce IFNγ that activates CD8⁺ CTL. Thus, by activating DC in vitro to enhance CD8⁺ and Th1 over Th2 responses in vivo, the deficiency in T-cell reactivity during HIV infection could potentially be reversed.

There are few comparative in vitro or in vivo data on factors for maturation of DC. We have employed a cocktail of pro-inflammatory cytokines, IL-1β, IL-6 and TNFα [42] to mature DC for immunotherapy of HIV [43]. This combination plus prostaglandin E₂ has proven to be highly immunogenic in vitro [44] and is widely used for DC immunotherapy [45]. An alternative is to mature DC with a combination of inflammatory cytokines, TNFα, IL-2, IFNγ and IFNα, and dsRNA poly I:C, termed αDC1 [36]. This cocktail can enhance the in vitro production of IFNγ by HIV antigen-stimulated, CD8⁺ T cells from HIV-infected patients on ART [37]. The αDC1 also respond to subsequent maturation with CD40L, supporting that this could work as an immunotherapy in vivo.

The in vivo lifespan of the DC is important for their use in the immunotherapy [46]. Studies in mice indicate that activated antigen-bearing DC are short-lived in draining lymph nodes [47]. Moreover, uptake of such dead DC by immature DC can lead to tolerance [48]. Strategies to counter this shortened lifespan include transfection of DC with anti-apoptotic Bcl-2 and Bcl-xL [49], the Akt1 protein kinase B family protein [50], CpG [51] and tumour necrosis factor-related activation-induced cytokine (TRANCE) [52]. This shortened lifespan of DC has important implications for the efficiency of DC immunotherapy of HIV infection.

Effect of HIV infection on DC

There is the potential for direct infection and subsequent dysregulation of the DC by HIV. Indeed, both CXCR4- and CCR5-tropic strains HIV can infect myeloid and plasmacytoid DC via these co-receptors and CD4 [53]. HIV replication, particularly by the X4 strain, is more efficient in immature than mature DC [54]. A second pathway of infection of DC is via C type lectins, in particular DC-SIGN [55]. HIV binds by high mannose moieties on gp120 to C type lectins [56], syndecan-3 [57] and DC immunoreceptor [58], and enters the cell by endocytosis. Production of infectious virions ensues, and the viral progeny can be transferred to other DC or CD4⁺ T cells for further rounds of replication [59, 60]. Such trans-infection preferentially targets activated CD4⁺ T cells [61, 62] and memory T-cell subsets [63].

This suggests that immature DC used in ex vivo therapy protocols or targeted in vivo as an immunotherapy could serve as sites of replication of HIV. The logical question is whether HIV infection in vivo during immunotherapy would adversely affect DC numbers and function. This is supported by lower numbers and abnormal function of myeloid DC in untreated HIV infection [64, 65]. Indeed, myeloid DC from subjects with untreated HIV infection are in an abnormal, semi-mature state [66], and induce Treg that can suppress antiviral CD8⁺ T-cell proliferation and function [67]. Additional defects are decreased expression of chemokine receptors CXCR2 and CXCR4, and spontaneous production of inflammatory cytokines [68, 69].

It is not expected that such immunosuppressive conditions will be highly operative during ART, where most protocols for DC immunotherapy will be applied. Blood DC are not significant reservoirs for HIV infection during ART [70]. Although HIV can productively infect monocytes in vivo, large numbers of functional DC can be obtained from monocytes of HIV-infected individuals on ART [43]. These DC have relatively normal antigen processing machinery [71] and respond well to maturation factors, CD40L and cytokine/TLR ligand cocktails, as indicated by stimulation of HIV-specific CTL activity and IFN-γ production in CD8⁺ [37, 72, 73] and CD4⁺ [74] T cells. Although DC from HIV-infected patients on ART produce less than normal amounts of IL-12 in response to CD40L, combining CD40L and IFN-γ or using the αDC1 cocktail induces high amounts of bioactive IL-12p70 [37, 39]. Finally, HIV infection of MDDC is susceptible to protease and fusion inhibitors, which do not block DC maturation or activation of antiviral CD8⁺ or CD4⁺ T cells [75, 76]. Thus, cumulative evidence supports that DC derived from monocytes of patients on ART maintain functional integrity necessary for use in HIV immunotherapy. Nevertheless, the natural function of DC in patients on ART that would be targeted by in vivo immunotherapy remains to be addressed.

Type of HIV antigen used in DC-based immunotherapy

Dendritic cells have the capacity to process proteins through MHC class I cross-presentation pathways for stimulation of CD8⁺ T cells, and MHC class II pathways for activation of CD4⁺ T cells. Even with this versatility, the most challenging aspect in choice of HIV antigen for DC immunotherapy is the exceptionally large degree of genetic diversity of the virus. Indeed, use of autologous HIV sequences has revealed T-cell specificities not observed with consensus virus antigen [77, 78]. Similarly, use of antigen based on autologous HIV sequences revealed that polyfunctional T-cell reactivity is less evident in response to immune escape variants [79]. Thus, the most effective immunotherapeutic vaccines would utilize antigen based on autologous HIV, i.e. the quasi-species of virus unique to each host. Indeed, such personalized vaccines are being developed for DC immunotherapy of cancer [80, 81].

The most impressive results in anti-HIV immunotherapy trials to date have used DC loaded with whole, inactivated HIV virions derived from the patients' autologous virus [82]. This antigen was aldrithiol-2 (AT-2)-inactivated virus, which retains its virion proteins in a native conformational state [83]. Prior to translation of this DC-virus model to the clinic, studies showed that DC loaded with the AT-2-inactivated virus stimulated CD8⁺ T-cell reactivity in vitro [84, 85], with processing of the HIV proteins through an MHC class I pathway [86]. The strength of this form of autologous HIV antigen includes its methodological simplicity and cost-effectiveness. We have opted for an alternative approach based on the pioneering work of Albert et al. [87], where DC are loaded with apoptotic preparations of HIV-infected T cells [85, 88–90]. This antigenic model could be superior to whole virus preparations by containing a natural array of all viral proteins, i.e. virion-associated and nonassociated, that could be processed by both MHC class I and II pathways for T-cell stimulation [91]. This also can be tailored to the specific patient by use of autologous HIV. Moreover, DC loaded with apoptotic bodies induce greater anti-HIV CD8⁺ and CD4⁺ T-cell activity than DC loaded with necrotic cell or heat-inactivated viral preparations [88]. Live cells can also serve as sources of antigen for DC [89, 92]. Potential pitfalls with this HIV antigen model include its methodological complexity, which requires infectious, autologous HIV that is difficult to culture from patients on virus-suppressive ART, and autologous CD4⁺ T cells for superinfection with autologous virus, which are then driven into apoptosis by UV light. Moreover, self-reactive CD8⁺ T cells have been generated in response to DC expressing caspase-cleaved fragments of HIV-infected apoptotic cells [93].

A modification of this cell antigen model is exosomes (aka microvesicles) derived from HIV-infected cells [94]. Exosomes are nanoparticles that are the flotsam and jetsam of cells, being secreted as single membrane organelles approximately the size of retroviruses (100 nm diameter). They result from outward budding of endosomes, eventually being released into the extracellular environment. Exosomes released by HIV-infected cells contain infectious virus and Gag [95–97].

Recombinant viral proteins theoretically are a relatively safe and simple form of antigen for immunotherapy that could cover the whole proteome of HIV. They are, however, very expensive in clinical grade quantities, and are preferentially shunted through the MHC class II pathway. The latter limitation can be overcome by complexing the proteins with liposomes, which diverts much of the protein to the MHC class I pathway [98, 99]. Alternatively, proteins can be transduced across cell membranes to drive entry into MHC class I pathways [100]. Addition of biologically active Tat protein can enhance proteolysis of other, recombinant HIV proteins by DC, which activate T cells specific for subdominant MHC class I epitopes [101]. Taking lessons from our oncology colleagues, most clinical trials of DC vaccination for treatment of cancer have used protein subunits of MHC class I binding, 9–11mer peptides representing tumour-associated antigens [102]. This has resulted in T cells specific for the tumour antigens, although usually of relatively low frequency [45]. This approach is safe and relatively economical, but requires matching of the peptides to their appropriate MHC class I haplotype and only targets a tiny region of the proteome. Peptides complexed with liposomes can induce CD8⁺ T-cell responses in vitro [103, 104], but again are limited to their specific MHC class I alleles. This MHC restriction problem can be addressed by using MHC class I `supertype' peptides that bind to a variety of human leucocyte antigen (HLA)-A and -B alleles [105]. We have recently used DC loaded with HIV immunodominant supertype peptides in a phase I clinical trial, and found this to be safe and immunogenic [43]. Moreover, T-cell reactivity to single peptides could prime or restimulate a broader array of anti-HIV T cells by epitope or determinant spreading, as has been shown in cancer patients treated with peptide-loaded DC [106, 107]. This is theoretically due to release of antigens of different specificities by cells that have been recognized and lysed by the vaccine peptide-specific T-cells.

A promising antigen approach derived from cancer models in DC immunotherapy is plasmid RNA [108, 109]. Myeloid DC transfected with mRNA coding for HIV proteins can stimulate antigen-specific CD8⁺ and CD4⁺ T-cell reactivity in vitro [110–112]. Previous problems with mRNA lability have been overcome by transfection of DC with mRNA by nucleofection, resulting in high expression of HIV or simian immunodeficiency virus (SIV) RNA [113, 114]. This RNA lability also can be limited by targeting monocytes in vitro before they are transformed into DC [115]. Lysosomal targeting of the mRNA-coded antigen can be enhanced by mRNA expression with lysosome-associated membrane protein-1, which redirects antigen along MHC class II pathways [111, 114, 116]. Stimulation of T cells from HIV-infected subjects with gag mRNA-electroporated DC expands both CD4⁺ and CD8⁺ T cells with broad recognition of Gag epitopes [117]. This has been adapted for mRNA derived from autologous HIV [117, 118] and SIV [114]. The efficacy of mRNA-DC immunotherapy can also be enhanced by co-expression of viral and TLR-coding mRNA [119], and cytokines such as TNF-family proteins [120] and IL-12 [121]. Stimulation of DC through TLR is of special interest as it can stabilize MHC class II molecule expression on the cell surface [122], which is a hallmark of the immunostimulatory capacity of DC [123]. Interestingly, mRNA can induce maturation of DC through TLR3 interactions with dsRNA secondary structures [124, 125], which has been of interest in cancer immunotherapy [45].

HIV DNA-based plasmids coding for HIV antigens have been expressed in DC [126], and have been proposed for immunotherapy [127]. Problems with DC transfected with plasmid DNA include low nucleic acid expression and cell viability [113, 128, 129]. However, new in vivo delivery systems using electroporation in mice show improved immunogenicity [130]. Both the adjuvanicity and delivery efficiency of HIV DNA to DC have been improved by coupling of DNA to microparticulates, such as biodegradable, pH sensitive, poly-beta amino esters [131]. The strong immediate-early DNA promoter of human cytomegalovirus, together with CD40 signalling, also can result in highly efficient expression and presentation of DNA-based antigens in DC [132]. Finally, cell toxicity due to DNA transfection can be blocked by cotransfecting the DC with survival gene, Bcl-xl [46], and co-expression of cell viability factors such as survivin [133] and TNF-α [134].

Microbial vectors including adenoviruses (Ad), poxviruses, Semliki Forest virus, lentiviruses, yeast and bacteria are being used to deliver HIV RNA and DNA to DC. HIV-expressing Ad vectors are particularly efficient in transducing DC and activating T cells [135–137]. However, the Merck STEP vaccine trial using Ad5 gag, pol and nef vectors as a prophylactic vaccine was discontinued due to lack of efficacy and possible enhancement of HIV transmission [138]. Concerns with this approach include potent antivector responses and enhanced T-cell activation, the latter possibly resulting in greater HIV replication. A promising alternative to Ad vectors is lentivirus vectors, which can transduce DC and maintain sustained long-term expression of HIV transgenes [139, 140]. Concerns about risks of generating replication-competent helper virus have been addressed by new replication-defective and self-inactivating lentiviral vectors [141]. Finally, DC loaded with virus-like particles (VLP) are immunogenic and safe, and hold promise for DC immunotherapy [142, 143]. Advantages of VLP include that these small particles act as potent immunogens, they can incorporate and express several HIV antigens and immunostimulatory molecules and they do not replicate [144].

Route of DC administration

The major interactions of DC with T cells occur in lymph nodes, subsequent to migration of the antigen-loaded DC from the periphery [11]. Moreover, regardless of the initial route of infection, HIV targets T cells in the lymph nodes [145] and GALT [146]. Immunotherapy protocols for HIV have attempted to recapitulate this process by injecting DC intradermally or intravenously. Because the virus continues to target lymphatic CD4⁺ T cells during ART, an effective therapeutic vaccine must generate a durable, HIV-specific response systemically. However, injection of DC intradermally in cancer patients only results in migration of 2–4% to local lymph nodes, with lower levels of migration after subcutaneous injection [147, 148]. Also, migrating DC may not always maintain their functional capacity. Using the paradigm of a second step in DC maturation after injection [36, 37], antigen-loaded DC migrating to the lymph node must respond to CD40L on activated CD4⁺ T cells and secrete large amounts of IL-12 and other cytokines such as IL-23 [149]. Recently, DC have been injected directly into lymph nodes to increase the number of DC in this critical site and to enhance the longevity of the DC [147]. One problem has been inaccurate delivery to perinodal fat tissue instead of lymph nodes [150]. As in cancer immunotherapy, a combination of different routes of DC injection could be effective in HIV infection [151].

Evaluating immunological efficacy of DC immunotherapy for HIV infection

The most impressive T-cell correlates for efficacy of DC immunotherapy are increasing the breadth of Gag-specific CD8⁺ T-cell IFNγ production, and the ability of HIV-specific CD8⁺ and CD4⁺ T cells to produce multiple cytokines, chemokines and cytotoxicity factors. Such polyfunctional CD8⁺ T cells emerge after long-term ART [152]. A key issue that has not been addressed in DC immunotherapy of HIV infection is the role of viral diversity in T-cell dysfunction. The predominant quasi-species of virus in chronically infected individuals are highly divergent from the original infecting founder strain [153]. This divergence has been linked to antiviral pressure from CD8⁺ T-cell immunity [154] coupled with the extraordinary high level of HIV replication and mutation, eventually resulting in `immune escape variants' of HIV. These variants are by definition poorly recognized by memory CD8⁺ T cells circulating in the host due to different host and viral mechanisms [155].

A major mechanism for HIV evasion of T-cell immunity could be a decrease in the binding of mutated viral peptide epitopes to MHC class I on APC, and consequent weak binding to T-cell receptors. We are therefore facing a daunting task in arming DC with either consensus HIV that codes for T-cell epitopes highly divergent from autologous virus, or with autologous HIV that is predominantly populated with immune escape variants. Even though memory CD8⁺ (or CD4⁺) T cells have recovered much of their immunological competence during ART, they could still be unable to recognize HIV immune escape epitopes in the injected or in vivo targeted DC. One approach to overcome this potential problem would be to engineer the DC to prime naive CD8⁺ and CD4⁺ T cells to these immune escape variants. Although seemingly counterintuitive, a significant portion of HIV variants could retain enough capacity to bind to MHC class I and II molecules in myeloid DC to elicit a de novo, primary T-cell response in naïve T cells that have been regenerated during ART. We are testing this hypothesis using cryopreserved peripheral blood mononuclear cells (PBMC) and plasma from HIV-negative volunteers who became infected with HIV in the Multicenter AIDS Cohort Study (Fig. 2). These cryopreserved samples are available before infection, after infection and after ART. T cells from these samples are being primed with MDDC derived from the same patients after they have suppressed viral infection during ART, similar to the aim of DC immunotherapy. The MDDC are loaded with HIV peptides representing the range of sequential, genetic divergence from the founder virus. With this unique model, we can compare the primary reactivity of naïve T cells to memory T cell, recall responses, using autologous virus sequences and PBMC obtained before infection and at various times after infection, and after ART. Preliminary results indicate that DC loaded with HIV immune escape variants for Gag, Env and Nef can prime CD8⁺ T cells in vitro (B. Colleton, X. Huang, N. Melhem, Z. Fan, R. Shankarappa, J. Mullins and C.R. Rinaldo, unpublished results). This could be adapted to infectious virus isolates as antigen and possibly priming of sorted, naïve T cells obtained after HIV infection using phenotypic markers such as CD31 (platelet endothelial cell adhesion molecule) and CD103 (αE integrin).

Fig. 2.

Model for assessing primary (naive) and recall (memory) T-cell responses to the autologous HIV-1 founder strain and immune escape variants in the MACS. Tn, naïve T cells obtained prior to HIV infection; Tm1, Tm2, etc., memory T cells obtained at sequential times after HIV infection; MD DC, monocyte-derived DC generated during ART. HIV VR1, HIV VR2, etc., autologous HIV or HIV peptides at various times after infection; Naive, priming of the naïve T cells with autologous virus or viral peptides; Memory, stimulation of recall responses in memory T cells. Thus, primary T-cell responses will be examined for all HIV isolates, whilst memory T-cell reactivity will be examined for HIV present on or before particular Tm cells. MACS, Multicenter AIDS Cohort Study; DC, dendritic cells; ART, antiretroviral therapy.

T-cell suppressor activity

There are three predominant forms of regulation of T cell immunity pertinent to DC immunotherapy of HIV infection: thymus-derived CD25⁺ FoxP3⁺ CD4⁺ Treg, programmed death-1 (PD-1) programmed death-1 and 2 ligands (PD-L1/L2) and CTL antigen-4 (CTLA-4)-CD80/86 pathways [156]. Consideration of these factors is becoming part of designing DC immunotherapy for cancer [45, 81, 157], but little has been carried out in this regard for HIV infection. One reason is confusion about the role of these regulatory mechanisms in HIV infection. Treg suppress T-cell activity by direct cell-to-cell contact, which can involve CTLA-4 signalling. Treg could suppress anti-HIV T cells and thereby enhance HIV infection [158, 159] or in contrast could decrease immune activation and effectively inhibit HIV disease progression [160–162]. To further complicate matters, other forms of Treg, i.e. Tr1 and Th3, can be generated in the periphery in response to antigen and produce high levels of IL10 and transforming growth factor-β [163]. These cells suppress antigen-specific immune responses and can be generated from naïve T cells in the presence of antigen and IL-10 [164]. Tr1 cells do not express IL-2Rα or FoxP3, in contrast to thymus-derived Treg [165, 166]. Notably, Tr1 cell differentiation is increased by IL-10 and IFN-α [167] and immature DC [168].

Programmed death-1 is upregulated on dysfunctional T cells during chronic HIV infection [169–171]. Blockade of this pathway reconstitutes the T cells, allowing them to expand and produce effector cytokines [172]. The ligand for PD-1, PD-L1, is expressed by DC, and when engaged with PD-1, will impair T-cell function [173]. TLR7/8 ligands can enhance PD-Ll expression on myeloid DC from patients with HIV infection [174]. This sounds a note of caution in using such TLR ligands as adjuvants in DC immunotherapy. Another negative regulator of T cells, CTLA-4, is a homolog of CD28. CTLA-4 competes with CD28 through binding to CD80 and CD86 on DC, and thereby decreases DC-mediated activation of T cells. Although there are no data on CTLA-4 regulation of anti-HIV T cells, CTLA-4-expressing T cells have been associated with inhibition of anti-tumour T-cell responses [175]. Consequently, monoclonal antibodies specific for CTLA-4 are being used in DC immunotherapy trials to enhance tumour-specific T-cell reactivity. Similar inhibition of regulatory cells and suppressor pathways could be necessary in DC immunotherapy of HIV infection.

Clinical trials of DC immunotherapy for HIV infection

An ideal immunotherapy strategy will allow for interruption of antiviral drug therapy after vaccination and lower the viral load `set-point' for risk of disease progression (Fig. 1). The safety of such analytical treatment interruption (ATI) for the purposes of evaluation of therapeutic modalities remains in question. Indeed, the early termination of the SMART study due to an approximate twofold risk of HIV disease progression or death in patients on intermittent ART than on continuous ART has resulted in increased scrutiny of safety during ATI [176]. Notably, the patient population and the monitoring of participants in SMART and those enrolled in DC immunotherapeutic trials should be quite different. SMART subjects had a median nadir CD4⁺ T-cell count of approximately 250 cells per mm³, and nearly 25% had a prior AIDS-defining condition. The participants were only monitored every 2 months. Finally, there was no increased risk for disease progression in the earliest months of ATI in SMART. Thus, ATI use in DC immunotherapy trials should be limited to shorter periods such as 12 weeks to differ substantially from the SMART strategy. Moreover, the patients in these studies should be free of prior AIDS-defining conditions and have a minimum CD4⁺ T-cell count of 500 cells per mm³ prior to the ATI. The patients should be monitored very closely during the ATI (every 1–2 weeks) and restarted on ART with CD4⁺ T-cell counts below 350 cells per mm³.

The first DC-HIV trial used immature, autologous blood DC, or allogeneic DC, from HLA-identical HIV-seronegative siblings of six HLA-A*0201 HIV-infected patients not on ART, that were loaded with recombinant HIV-1 MN gp160 or HLA-A*0201-restricted Gag or Pol epitopes [177] (Table 2). The patients were infused intravenously six to nine times at monthly intervals. Three of the six DC recipients had increases in HIV-specific T-cell reactivity. No clinically significant adverse effects were noted. Follow-up studies showed that DC persisted for at least 1 week after infusion [178].

Table 2.

Results of published DC immunotherapy trials for HIV infection

DC type	DC maturation	HIV antigens	Route	No. subjects	Immunological responsesa	Virological responses	References
Blood DC	None	rHIVMN gp 160, Gag and pol peptides	IV	6	3	No change in plasma HIV RNA	[177, 178]
IL-4/GM-CSF MDDC	1L-1β, IL-6, TNF α	AT-2-inactivated autologous HIV	SC	18	18	Decrease in plasma HIV RNA (viral set-point) at 112 days; prolonged reduction in viral load <90% in eight patients	[82]
IL-4/GM-CSF MDDC	Monocyte-conditioned medium	Heat-inactivated autologous HIV	SC	12	0	After ATI, decrease in HIV RNA of ≥0.5 log10 copies per mL and an increase in HIV doubling time in four patients	[180]
IL-4/GM-CSF MDDC	TNF α	Gag, Env and Nef peptides	SC	7	2	No significant changes in viral load after ATI	[181]
IL-4/GM-CSF MDDC	IL-1β, IL-6 and TNF α	Gag, Env and Pol peptides	IV, SC	18	16	No ATI; no change in suppressed viral load	[43]

rHIVMN, X4 tropic strain; DC, dendritic cells; TNF, tumour necrosis factor; IL, interleukin; IV, intravenous; SC, subcutaneous; GM-CSF, granulocyte-macrophage colony-stimulating factor; MDDC, monocyte-derived DC; AT-2, aldrithiol-2; ATI, analytical treatment interruption.

^aImmunological response is defined as any evidence of HIV-specific T-cell response after DC vaccination.

More recent trials have used MDDC generated ex vivo with GM-CSF and IL-4. A schematic representation of this approach is given in Fig. 2. In the most extensive study, 18 HIV-infected, untreated subjects with high HIV loads received three subcutaneous doses of DC loaded with AT-2 inactivated, autologous HIV, and matured with a cocktail of IL-1β, IL-6 and TNFα [82]. This trial was based on in vitro studies [84] and DC therapy of SIV infection in the monkey model [179]. A decrease in median plasma HIV RNA (viral set-point) was noted at 112 days, with a prolonged reduction in viral load of over 90% in 8 of 18 subjects. The decrease in viral load correlated with detection of HIV-specific, IL-2 and IFN-γ expressing CD4⁺ T cells and Gag-specific, perforin-expressing CD8⁺ T cells, total baseline CD4⁺ T cells, HIV-specific, IL-2-producing CD4⁺ T cells and an increased total CD4⁺ T-cell count. This study shows efficacy of immunotherapy with inactivated, autologous virus antigen and cytokine-matured DC in patients with chronic HIV infection in the absence of ART.

A third trial subcutaneously injected 12 HIV-infected individuals on virus-suppressive ART with autologous DC that had been matured with monocyte-conditioned medium and loaded with heat-inactivated, autologous HIV [180]. After ATI, a decrease in HIV RNA of 0.5 log₁₀ copies per mL or more and a slower increase in HIV RNA levels over time was observed in 4 of 12 patients. Serum neutralizing antibodies to HIV did not change. The decrease in viral load set-point correlated with an increase in proliferative CD4⁺ T-cell responses to p24 antigen in responding patients and an increase in the frequency of CTL (CD8⁺ granzyme B⁺ cells) in tonsillar biopsies. However, the magnitude and the breadth of the total HIV-specific CD8⁺ T cell responses decreased progressively during the course of the study. The decrease in the magnitude of the CD8⁺ T cell responses was not due to virological changes, as it was similar in the four patients who had lower viral loads as in the eight patients who had no change in viral load. This study supports the use of DC loaded with whole, inactivated, autologous HIV antigen as a safe and partially immunogenic and virus-suppressive vaccine in patients on ART, but was limited by an imprecise reagent for DC maturation.

A fourth trial administered six doses of autologous DC matured with TNFα and loaded with seven Gag, Env and Nef, HLA-A*2402-restricted immunodominant peptides to four HLA-A*2402 patients on virus-suppressive ART, who subsequently underwent ATI [181]. Vaccination was safe, with two of four patients demonstrating significant increases in numbers of IFNγ-producing CD8⁺ T cells to one or two of the seven HIV peptides. No significant changes were seen in viral load or CD4⁺ T-cell counts. This study was limited by a relatively small number of patients, and HIV antigen preparations and DC maturation methods of minimal potency.

Our group studied 18 HLA-A*0201, HIV-infected individuals on virus-suppressive ART. The subjects were vaccinated with ex vivo matured, synthetic HIV peptide loaded, autologous MDDC, administered either intravenously or subcutaneously and followed for 48 weeks for safety analysis [43]. The MDDC were matured ex vivo with a cocktail of IL-1β, IL-6 and TNFα and loaded with three HLA-A*0201 Gag, Env and Pol peptides, as well as an influenza A virus matrix peptide, that were immunodominant, MHC class I supertype antigens [105]. Immunological responses were measured by IFN-γ enzyme-linked immunospot at weeks 2, 4 and 6 following completion of a two vaccine series. This ex vivo peptide loaded, autologous DC vaccination was safe, and induced a significantly increased frequency of HIV peptide-specific IFN-γ-positive cells, 2 weeks following the second vaccine, with three individuals responding to all four peptides. The T-cell responses did not differ with the route of vaccine administration. Interestingly, preliminary results suggest that there was an increase in the frequency of CD25^high- FoxP3⁺ Treg by 2–4 weeks that could have prevented a more robust, HIV-specific T-cell response to the therapy (B. Macatangay, S. Riddler, N. Connolly, C.R. Rinaldo and T. Whiteside, unpublished results). This emphasizes the need for more complete analysis of activation and suppression properties of T cells in DC immunotherapy trials.

We are conducting a follow-up phase I trial that is using autologous HIV-infected, apoptotic CD4⁺ T cells as the source of antigen. This is based on our in vitro DC T-cell model, where CD4⁺ T cells were infected with an X4 strains of HIV, inactivated by treatment with psoralen-UV light and used to load DC [88]. This preparation of DC induced both CD8⁺ and CD4⁺ T-cell reactivity in PBMC from HIV-infected subjects in vitro. In our trial, 16 subjects are being enrolled as ART naïve with >350 CD4⁺ T cells per mm³, and a plasma virus load of ≥5000 RNA copies per mL to assure the isolation of HIV from their blood for use as antigen. They then start ART and are eligible to receive the vaccine after they reach <50 copies of HIV RNA per mL of plasma. We are maturing the DC with TNFα, IL-2, IFNγ, IFNα and poly I:C [36] to maximize the maturation of these DC after injection. We are administering three doses of this DC vaccine subcutaneously at 2 week intervals, with an ATI at 6 weeks after the third vaccination. The subjects will receive a fourth vaccination 2 weeks after start of ATI, and be followed for adverse events, T-cell numbers and function, and viral load for 36 weeks thereafter. This study is powered to detect a 0.65 log reduction in plasma viral load during ATI.

Future directions: targeting DC in vivo

Current DC vaccines involve ex vivo manipulation of autologous DC as outlined in Fig. 3. These vaccines attempt to optimize DC maturation to direct immune responses toward either Th1 or Th2 profile. To streamline these complicated vaccination procedures, there are several strategies either in early phase clinical trials or currently undergoing animal model testing targeting DC in vivo. A topical vaccine, DermaVir (Genetic Immunity, McClean, VA, USA), is completing phase I testing in the AIDS Clinical Trials Group. DermaVir is a DNA plasmid vaccine encoding HIV genes coupled with polyethyleneimine mannose [182]. DermaVir targets DC in vivo by introducing the vaccine between the epidermal and dermal skin layers where high numbers of Langerhans cells and dermal DC reside. A minor exfoliation procedure designed to remove the epidermal layer and generate a small amount of inflammation to recruit DC to the area is used, after which the DermaVir product is applied topically where it is quickly taken up by DC and transported to local lymph nodes [183]. DermaVir has been shown to be safe and immunogenic in Rhesus macaques [184]. Future formulations could include a IL-15 plasmid that enhances HIV-specific central memory T-cell induction by DermaVir [185].

Fig. 3.

Steps in the preparation of DC ex vivo for use as an HIV immunotherapeutic vaccine. DC, dendritic cells, PBMC, peripheral blood mononuclear cells; GM-CSF, granulocyte-macrophage colony-stimulating factor; IL-4, interleukin 4.

A second strategy targets HIV antigens to specific surface receptors on DC in vivo by complexing them, e.g. antibodies to DC-SIGN [186] and liver/lymph node-specific ICAM-3-grabbing nonintegrin (L-SIGN) [187]. Such antibody-mediated targeting of antigen to C type lectins is more efficient than loading DC with antigen in vitro [186, 188]. Notably, DEC-205, an endocytic receptor on DC, has been targeted using anti-DEC-205 monoclonal antibodies fused with antigenic protein in mice to produce robust, ovalbumin-specific T-cell- and antibody responses [189–191]. This immunogenic process can be enhanced in mice by addition of the TLR3 ligand, poly I:C [192]. This has been translated in vitro to human cells as a model for HIV immunotherapy [193]. MDDC derived from the blood of 11 HIV-infected patients were treated with a low dose of chimeric anti-DEC-205 antibody complexed to Gag p24. After 1 week of expansion, CD8⁺ T cells recognized eight different Gag peptides. Recently, this model has been modified using single chain anti-DEC-205 antibody fused to Gag to enhance antigen presentation by DC in lymphoid tissues of mice at lower doses [194]. Such potent, dose-sparing, DC-targeted regimens hold promise for in vivo targeted, DC immunotherapy of HIV infection. A third strategy is delivery of biodegradable nanoparticles to immature DC ex vivo or in vivo [195, 196]. This technology requires further study focusing on the ability of nanoparticles to induce DC maturation, antigen presentation and T-cell responses, as well as their durability.

Additional forms of therapeutic immunizations aim to attenuate HIV infection by targeting DC with adjuvanted antigens. Targeting TLR on DC is being widely used as an adjuvant in DC immunotherapy of cancers [197] and is being promoted for use in HIV infection. Other adjuvant approaches include cytokines such as GM-CSF. Kran, et al. [198] studied HIV-infected patients previously immunized with p24-like-peptides targeting DC in vivo with a local adjuvant, GM-CSF. Ninety per cent of the 40 patients given the vaccine demonstrated cellular immune responses, and 62% remained off drug therapy 1.5 years after completing immunization. Alternatively, DC processing of viral antigens could be enhanced by fusing MHC class I trafficking signal to viral proteins, which increases expansion of antigen-specific, CD8⁺ and CD4⁺ T cells [199].

Concluding remarks

The current ex vivo and future in vivo-based DC immunotherapies hold promise for improving control of HIV infection. The former approach should provide critically important proof of principle, whilst the latter approach will be most applicable to the wide geographical and socio-economic range of people infected with the virus. It is evident from studies of DC therapy of cancer, however, that there could be setbacks before there are major advances. This is exemplified by the recent failure of a randomized phase III trial that was designed to demonstrate the superiority of autologous peptide-loaded, cytokine-matured DC vaccination over standard chemotherapy in stage IV melanoma patients [200]. No difference in survival was demonstrated for the DC therapy arm, leading to discontinuation of the trial. Various explanations have been posited for this failure based on the numerous factors that potentially affect DC immunotherapy [45]. Nevertheless, we should take heed of this experience in developing a DC-based immunotherapy for the fearsome enemy of cancer as a lesson directing our pursuit of a DC immunotherapeutic vaccine against the equally formidable foe of HIV.