An HIV-based lentiviral vector as HIV vaccine candidate: Immunogenic characterization

Vaccine 28 (2010) 1952–1961

Contents lists available at ScienceDirect

Vaccine journal homepage: www.elsevier.com/locate/vaccine

An HIV-based lentiviral vector as HIV vaccine candidate: Immunogenic characterization Franck Lemiale ∗ , Benyam Asefa, Delia Ye, Christopher Chen, Nikolay Korokhov, Laurent Humeau VIRxSYS Corporation, 200 Perry Parkway, Gaithersburg, MD 20877, United States

a r t i c l e

i n f o

Keywords: Lentivirus HIV Vaccine

a b s t r a c t Recently developed viral-vectored HIV vaccine candidates, despite achieving high levels transgene expression and inducing high magnitude immune responses to HIV, have faced limitations related to antivector immunity. In contrast, lentiviral vectors (LV) have been shown to be less sensitive to anti-vector neutralizing activity, while displaying desirable characteristics, such as transduction of non-dividing cells, including antigen-presenting cells, and long-term transgene expression. We have developed VRX1023, an HIV-based LV expressing HIV Gag, Pol and Rev under the control of the native HIV LTR. In mice, this vector induced significant mucosal and systemic cellular and humoral responses against HIV after sub-cutaneous injection. Similarly to other viral vectors, this LV candidate can be effectively used in DNA prime, LV boost strategies, where it elicited as high as 21% HIV Gag-specific CD8 responses as measured by intracellular cytokine staining. Moreover, anti-vector immunity is not an obstacle to repeated LV administrations, as shown by improved anti-HIV responses compared to single LV immunization. In head to head comparisons with Ad5 vectors expressing the same vaccine payload, VRX1023 elicited higher and more persistent cellular and antibody responses to HIV than its adenoviral counterpart. In preparation for clinical use, manufacturing scale-up of a highly purified VRX1023 vector lot following cGMP was successfully achieved without altering the robust immunogenicity observed with the research-grade vector. VRX1023, in addition to competing favorably with existing vectors such as Ad5 for anti-HIV immune responses, demonstrates unique features likely to address some of the pitfalls of current vector-based HIV vaccine strategies. © 2009 Elsevier Ltd. All rights reserved.

1. Introduction Challenges in developing an effective HIV vaccine are numerous. Over the recent years, with the advent of DNA vaccines and viral vectors, efforts have focused on cytotoxic T lymphocytes (CTL)-based vaccines. Indeed, CD8 T-cells appear to be key for in vivo control of HIV replication [1] and high frequencies of HIVspecific CD4 and CD8 T-cells are seen in HIV-1 infected subjects with non-progressive disease [2]. Some promising data have been generated by various groups [3–5] focusing on T-cell vaccine strategies aiming at controlling HIV replication after infection, as it is unlikely that they will prevent infection [6]. Particularly attractive are several viral vectors, including the adenovirus-based, which have been developed and used either alone or in combination with a DNA prime, in order to elicit focused anti-HIV cellular immune responses. Unfortunately, none of these strategies has yet been

∗ Corresponding author. Tel.: +1 301 987 0480x284. E-mail address: [email protected] (F. Lemiale). 0264-410X/$ – see front matter © 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.vaccine.2009.10.089

successful at developing a truly convincing vaccine candidate in humans, as illustrated recently by the interruption or cancellation of high-profile large clinical trials of adenoviral-based HIV vaccine candidates (Merck’s STEP and National Institutes of Health, Vaccine Research Center’s (NIH/VRC) Partnership for AIDS Vaccine Evaluation 100). However, recent studies exploring Ad5-based vaccines have revealed that there is no fundamental flaw in T-cell-based vaccine strategies, but rather showed that the Merck trial failure was specific to the vaccine [7,8]. Specifically, one of these studies showed that a T-cell-based vaccine should be able to control HIV infection, even if the adenovirus serotype 5 (Ad5) is not necessarily the ideal vector to achieve this. These setbacks highlight the need for exploration of innovative strategies and new vaccine vectors [9,10]. Among the most recent alternate viral vectors developed are HIV-based lentiviral vectors (LV). Some of them have been studied in humans over the past 5 years in phases 1 and 2 clinical studies evaluating the safety and tolerability of vector-modified T-cells in HIV-positive subjects. The findings from these trials, showing no report of serious adverse event related to the product [11], provide

F. Lemiale et al. / Vaccine 28 (2010) 1952–1961

1953

Fig. 1. Schematic representation of plasmids and lentiviral vectors. (A) The VRX1023 lentiviral vector is obtained by co-transfection of 293T cells with pVRX1023, containing the HIV-1 Gag, Pol and Rev transgenes, and with a packaging plasmid. (B) Early immunogenicity experiments used pVRX577 as the packaging plasmid, as LV production was later improved by the use of the pVRX092 packaging construct. pVRX092 differs from pVRX577 by the removal of the gag–pol sequence, as these are provided in trans by pVRX1023 at time of co-transfection. (C) For DNA prime, LV boost immunizations, the plasmid pVRX1053, containing the Gag, Pol and Rev antigens matching pVRX1023’s, was constructed and used as the priming agent. LTR: long terminal repeat; SD: splice donor; SA: splice acceptor; RRE: Rev-responsive element; IRES: internal ribosome entry site; GFP: green fluorescent protein; CMV: cytomegalovirus promoter; EHP: elongation factor-1␣/human T-cell leukemia virus chimeric promoter; pA: polyadenylation site; VSV-G: vesicular stomatitis-G protein.

support for further use of these vectors in broader types of applications, such as direct injection, of particular interest for vaccine applications. The rationale for focusing on LV is based on some key characteristics: these vectors can efficiently transduce non-dividing cells, including antigen-presenting cells (APC) [12,13], and they have been shown to support long-term antigen expression [14]. In contrast to Ad5, LV are not endemic in the human population and therefore overcome the pre-existing immunity issue, a major obstacle for Ad5-based vaccine strategies. This report describes the development and preclinical immunogenicity testing of a research- and cGMP-grade LV-based HIV vaccine candidate and its comparison to Ad5-based immunogens, when these vectors are used either alone or in combination with a DNA prime. 2. Materials and methods 2.1. Vector constructs The pVRX1023 plasmid for lentiviral vector VRX1023 was based on pUC18 backbone and was constructed using distinct elements of the HIV NL4-3 molecular clone. HIV-derived elements include the 5 and 3 long terminal repeat (LTR), unmodified gag–pol sequence, splice donor and acceptor sites (SD/SA), Rev response element (RRE), and Rev sequence. Engineered elements include the insertion of an green fluorescent protein (GFP) gene, which expression is mediated by the internal ribosome entry site (IRES) fused to the 3

end of the rev gene. Rev protein plays a dual role in this vector configuration. First, it supports an efficient transport of full-length vector RNA essential for Gag–Pol protein expression; second, it serves as an antigen. The vector design provides efficient Gag–Pol and Rev expression accompanied by moderate expression of GFP (Fig. 1A). Two generations of packaging constructs were used for LV production. Early immunogenicity studies were performed with LV produced using pVRX577, a packaging construct expressing HIV Gag/Pol, Rev, and Tat under the control of the cytomegalovirus (CMV) promoter (Fig. 1B) [15]. Translation of Tat and Rev is separated by an IRES sequence. HIV gag/pol is flanked by a splice donor/acceptor site, between which the RRE is located for RNA transport. In this construct, the Vesicular Stomatitis G protein (VSVG) was placed under the control of the elongation factor-1␣/human T-cell leukemia virus chimeric promoter, which is separated from activation by the upstream CMV promoter by the bovine growth hormone and human ␣2-globin polyadenylation sites and the human ␣2-globin transcriptional pause site. The SV40 polyadenylation signal is located downstream of VSV-G. Regions of potential recombination were degenerated in the packaging construct to decrease the risk of recombination. The packaging plasmid pVRX092 was developed to exclude the gag–pol gene from the packaging plasmid, as these are provided in trans by the VRX1023 plasmid [15]. Exclusion of the gag–pol sequence preserves the expression of VSV-G and HIV-1 tat and rev genes essential for VRX1023 vector production. pVRX1053 was used as a DNA plasmid, with transgenes matching VRX1023 LV’s, for DNA priming. This plasmid is based on the

1954

F. Lemiale et al. / Vaccine 28 (2010) 1952–1961

pMG-A vector from InVivogen (San Diego, CA, USA). It contains the HIV-1 gag–pol and rev genes driven by the upstream CMV promoter. An expression cassette containing the HIV-1 major splice donor (SD), gag–pol genes, and RRE from HIV-2 was cloned between the CMV promoter and the splice acceptor site upstream of the rev gene, to enable expression of rev from the spliced message (Fig. 1C). For lentivector production at a research-grade, viral vector stocks pseudotyped with VSV-G were prepared by transient transfection of the human embryonic kidney (HEK) 293T cell line as described previously [15]. Briefly, a packaging plasmid, either pVRX577 or pVRX092, was co-transfected with pVRX1023 vector plasmid by calcium-phosphate precipitation onto HEK293T17 cells (ATCC). Supernatant from transfected cells was collected at 36 h after transfection, passed through a 0.22 ␮m filter and then lentiviral vectors were concentrated by high-speed centrifugation at 10,000 × g for 16–18 h at 4 ◦ C. The vector pellet was resuspended in storage buffer (60 mM NaCl, 25 mM HEPES, pH 7.2) and frozen at −80 ◦ C. LV manufactured at cGMP-grade was initially prepared as above by transient transfection. The crude vector supernatant was then harvested from cell factories (Nunc) for downstream clarification using disposable filters of decreasing pore sizes. The clarified harvest was concentrated approximately 100-fold by tangential flow filtration, followed by diafiltration. The concentrated harvest was treated with an endonuclease to reduce the size of residual free nucleic acids. The treated harvest was efficiently purified by size exclusion chromatography (SEC). The SEC also serves as a buffer exchange step for the purified vector as the column is equilibrated with a vector storage buffer. The purified and formulated buffer was rendered sterile by filtration through a 0.22 ␮m capsule filter. The final material was divided into aliquots and stored at −80 ◦ C [16]. Quality control testing was performed on the filled vector before being released for further use. The biological titers of concentrated vectors are expressed as the number of transducing units per ml (TU/ml) and were determined by transducing HeLa-tat cells in limiting dilutions in the presence of 8 ␮g/ml of polybrene. After passaging the transduced cells twice over 7 days, total genomic DNA was extracted and vector-specific real-time PCR analysis was carried out to determine the number of integrated viral genomes. Concentration of HIV-1 Gag p24 in vector preparations was measured by p24 ELISA. Measurement of p24 content, which is proportional to the number of vector particles (VP), versus the number of LV genomes (TU) is used to characterize the quality of vector preparations. To perform direct head to head comparisons with our LV candidate, other immunogens were obtained from outside sources: Dr. Ioana Stanescu (FIT Biotech, Tampere, Finland), kindly provided the GTU-MultiHIV-B DNA and Ad5-MultiHIV. GTU-MultiHIV-B is a plasmid DNA candidate containing the MultiHIV payload, consisting of an antigenic fusion protein (HIV Rev, Nef, Tat and Gag) and CTL epitopes from HIV Pol and Env [17]. The Ad5-MultiHIV is an adenoviral type 5 vector containing the matching payload. In addition, the VRC5409, an Ad5 vector containing the HIV GagPol (dMyr, dRT, dIN and PR), was obtained from Dr. Gary Nabel (Vaccine Research Center, NIAID, NIH, Bethesda) [18]. 2.2. Animals and immunizations Six-week-old female BALB/c mice (Jackson Laboratory, Bar Harbor, ME) were housed at BIOQUAL, Inc. (Rockville, MD) in filtertopped cages on standard rodent diet and allowed to acclimate for at least 1 week prior to the study. Five mice per group were immunized sub-cutaneously (SC) with 100 ␮l of lentiviral vectors (108 particles) diluted in PBS at the base of tail. In the case of prime/boost regimens, mice were primed weekly for 3 weeks with 100 ␮g of pVRX1053: mice were

Fig. 2. LV immunization elicits sustained anti-HIV immunity. Three groups of five mice were immunized twice at days 0 and 15 with 8 × 107 transducing units (TU) of VRX1023 injected sub-cutaneously (SC). Each group was sacrificed at 10 days, 3 weeks and 2 months post-last immunization, respectively (panel A). T-cell responses to HIV Gag (dot plots and left y-axis) were measured by intracellular cytokine staining (ICS). Splenocytes were stimulated in vitro with Gag peptides as described in Section 2. Panel B shows the percentage of cytokine positive cells (IFN␥, TNF␣ and IL2 combined on the same channel) out of the CD3+ CD8+ T-cell subset. Percentage of cytokine producing cells are shown for each individual animal (dots). Bars represent geometric means. Serum samples collected at time of sacrifice in each group was analyzed by ELISA for anti-HIV IgG detection (line graph and right y-axis). Data are shown, for each timepoint, as geometric means, along with standard deviations.

injected intramuscularly (IM) with 200 ␮l divided equally into both quadriceps. LV immunizations were performed 2 weeks after the third DNA immunization. Blood and mucosal samples were collected, animals sacrificed, and spleens harvested for cellular immunity assessment, at 2 weeks post-LV immunization unless specified otherwise. Salivary secretions were collected by injecting mice intraperitoneally with 50 ␮l of pilocarpine (0.5% [w/v] in PBS) (Sigma, St. Louis, MO) to induce salivation. Approximate volumes of 100 ␮l of saliva were collected and stored at −20 ◦ C. 2.3. HIV-specific immunoglobin G (IgG) and IgA analysis The anti-HIV-1 ELISA was performed as previously described [19]. Microwell plates coated with HIV-1 antigens obtained from viral lysates (Genetic Systems rLAV EIA; Bio-Rad Laboratories, Redmond, Wash) were used. Mouse sera samples from both pre- and post-immunization were spun down at 15,000 rpm for 10 min to separate the red and white blood cells from the serum. Mouse serum and saliva were diluted in PBS, 1% fetal bovine serum, 0.02% Tween 20. Sera and saliva were diluted 1:100 and 1:2 respectively. Diluted samples were transferred at a volume of 100 ␮l per well onto ELISA plates. Following an overnight incubation, plates were washed five times in PBS-0.2% Tween 20 and incubated with 100 ␮l horseradish peroxidase-conjugated goat anti-mouse IgG (1:5000; Chemicon International Inc., Temecula, CA) for 1 h at room temperature. Plates were washed five times, and 100 ␮l of OPD peroxidase substrate (Sigma) was added to each well. The reaction was stopped after 30 min by addition of 50 ␮l of 2N H2 SO4 . The plates were read on an ELISA plate reader at 490 nm. 2.4. Cytokine secretion analysis Intracellular cytokine staining (ICS) was performed as previously described [20]. After collection, washes and red blood cell lysis, 106 splenocytes were stimulated using HIV-1 peptide pools (15-mer peptides covering the whole Gag (123 peptides) protein and the Pol protein (249 peptides) from HIV-1 subtype B; 11

F. Lemiale et al. / Vaccine 28 (2010) 1952–1961

1955

Fig. 3. Sub-cutaneous (SC) LV immunization induces both systemic and mucosal humoral and cellular immunity. Mice were immunized twice at days 0 and 15 with 2 × 107 TU of VRX1023 injected SC, then sacrificed at day 25 (panel A). The negative control group received 2 × 107 TU of a lentiviral vector carrying GFP in place of the gag, pol and rev transgenes. Anti-HIV antibodies were detected at the systemic (serum, panel B) and mucosal level (saliva, panel C) by ELISA as described in Section 2. CD8+ T-cells expressing cytokines (IFN␥, TNF␣ and IL2 combined on the same channel) after HIV Gag peptides stimulation were detected by ICS (panel D). Results are shown as individual values (dots) and geometric mean in each group (bars).

residues overlap between each peptide (2.5-␮g/ml of each peptide). All HIV peptides were obtained through the AIDS Research and Reference Program, Division of AIDS, NIAID, NIH. Stimulation was performed as previously described [20]. For cytokine staining, cells were incubated with anti-mouse CD8 FITC, anti-Interferon (IFN)␥ PE, anti-tumor necrosis factor (TNF)-␣ PE, anti-interleukin (IL) 2 PE, anti-mouse CD4 PerCP and anti-mouse CD3 APC-Cy7 antibodies for 25 min at 4 ◦ C, then washed and analyzed by flow cytometry for intracellular cytokines on a BD-Cytek 5-color FACSCalibur. For later experiments assessing individual cytokines and cell polyfunctionality, cells were stained with anti-mouse CD8 FITC, anti-IFN␥ PE, anti-TNF␣ PE-Cy7, anti-IL2 APC, anti-CD4 PerCPCy5.5 and anti-CD3 APC-Cy7 antibodies, then analyzed on an 8-color LSR II flow cytometer (BD). All flow cytometry data were analyzed using the FlowJo software (Treestar). 2.5. Statistical analysis Comparison between vaccination groups was carried out using paired Student’s t-test analysis with a two-tailed distribution. A value of p < 0.05 was considered statistically significant. 3. Results

anti-HIV Pol T-cell response, although at a lower magnitude (data not shown). Antibody responses shown in Fig. 2B also confirm that anti-HIV responses are sustained over the course of the study. 3.2. Sub-cutaneous administration induces both systemic and mucosal immune responses After establishing the kinetic of the responses induced, and to further define VRX1023’s immunogenicity, a variety of anti-HIV responses were assessed at different sites. Mice were sacrificed 10 days after receiving two SC administrations of VRX1023 (Fig. 3A) and samples were collected. Transgene expression from the vector elicited anti-HIV IgG and IgA in the serum and saliva respectively, as demonstrated by ELISA responses significantly higher than those elicited by injection of a negative control LV expressing no transgene (Fig. 3B and C). Moreover, anti-HIV Gag CD8+ T-cells were detected at the systemic level after VRX1023 immunization, while minimal responses were elicited by an LV carrying no HIV transgene (Fig. 3D). Of note, SC appeared to be the most immunogenic of all routes tested in the context of VRX1023 immunizations, as it induced higher systemic responses than the systemic (IV, IM and IP) routes tested, while eliciting anti-HIV IgA mucosal responses as high as those elicited using mucosal routes such as IN (data not shown).

3.1. LV immunization induces sustained anti-HIV immunity In order to characterize the kinetic of LV-induced anti-HIV immune responses, T-cell and antibody responses were monitored in mice at several timepoints post-immunization. After two R&Dgrade VRX1023 SC injections at 2 weeks interval, groups of mice were sacrificed at various timepoints and responses to HIV were monitored by ICS (T-cell responses) and ELISA (IgG) (Fig. 2A). While a significant anti-HIV Gag CD8+ T-cell response could be readily detected as early as 10 days post-immunization (p = 0.005), it significantly increased and peaked at 3 weeks post-immunization (p = 0.0003 at 3 weeks compared to 10 days), and was then sustained over the rest of the study (p = 0.13 between 3 weeks and 2 months) (Fig. 2B). Similar findings were also obtained with the

3.3. VRX1023 immunogenicity can be significantly improved in DNA prime, vector boost settings and by vector manufacturing optimizations To evaluate whether DNA vaccination could efficiently prime the immune system prior to LV-based vaccination, the pVRX1053 plasmid was constructed, with transgenes matching those contained in VRX1023 (Fig. 1A and C). To determine the respective immunogenicity of the DNA and LV candidates, groups of mice were immunized with either of them, or a combination of the two in prime/boost regimens (Fig. 4A). As shown in Fig. 4B, the pVRX1053 DNA candidate did not induce any significant anti-HIV T-cell response, when injected IM three

1956

F. Lemiale et al. / Vaccine 28 (2010) 1952–1961

Fig. 4. VRX1023 immunogenicity can be significantly improved in DNA prime, vector boost settings and by vector manufacturing optimizations. Groups of five mice were immunized with either PBS, VRX1053 DNA, VRX1023 LV, or a combination of these as described in panel A. When immunized with DNA, mice received 100 ug of pVRX1053 IM. For LV immunization, mice received 109 TU of VRX1023 SC. (Panel B) Anti-HIV CD8+ T-cell responses were assessed by ICS after stimulation of splenocytes with Gag and Pol peptide pools. Bars indicate, for each animal, the percentage of cells staining positively both for cytokines (IFN␥, TNF␣ and IL2 combined on the same channel) and for the activation marker CD69, within the CD8+ subset of CD3+ lymphocytes. (Panel C) Examples of cytokine secretions as an illustration of immunogenicity improvement provided by LV manufacturing optimization. Each set of Gag-induced cytokine secretion is the result of independent experiments. (Panel D) Summary of T-cell responses, measured by ICS, in mice receiving various vector preparations.

times weekly, with no vector boost. The VRX1023 LV used alone, however, did elicit significant anti-HIV CD8+ T-cell responses, which could be further boosted by a second LV administration. The highest T-cell responses to HIV in this experiment were obtained in mice receiving a combination of three DNA primes, followed by one LV boost. This prime (3×)/boost (1×) regimen elicited responses in the range of up to 8–9% cytokine positive CD8+ T-cells, as assessed by ICS. LV immunization also elicited anti-HIV Pol responses, which could be significantly improved by the DNA prime, LV boost regimen, although these responses remained at a lower magnitude compared to the anti-HIV Gag response. To demonstrate that immunogenicity improvements could be obtained by the optimization of not only the immunization regimen (DNA prime, LV boost) but also by vector manufacturing, immunogenicity characterization of various LV preparations was performed. Fig. 4C and D, illustrating the cytokine responses elicited in mice immunized with each vector preparation, show that the vector quality can significantly impact its immunogenicity. Improved vector purification resulted in increased anti-HIV Gag T-cell responses. Additionally, the use of an alternate packag-

ing construct further improved the response, inducing significantly higher T-cell responses to Gag while the dose used was lower. These results also highlight a characteristic of the immunogenicity provided by LV: anti-Gag responses are elicited by both the transgene and the Gag contained in the lentiviral particle. In the case of the last vector preparation, the lower number of transducing units injected was balanced by a higher p24 content in the vector preparation, resulting in improved Gag response. As responses to HIV Pol are driven solely by antigen expression, and not elicited by the vector particles, these remain stable when similar TU are used and are not improved by injection of more non-transducing particles. These results indicate that the quality of the vector preparations, assessed at time of manufacturing using the VP/TU ratio, can inform on the type of immune response to be induced by each vector lot. 3.4. At equal doses, LV induces higher anti-HIV immunity than Ad5 As it was determined that DNA prime, LV boost regimens were optimal for LV to induce anti-HIV immunity, this combination was

F. Lemiale et al. / Vaccine 28 (2010) 1952–1961

1957

Fig. 5. At equal doses, LV induces higher anti-HIV immunity than Ad5. Groups of five mice were immunized with three DNA primes (100 ␮g DNA IM/animal), followed by a vector boost (108 TU/animal, SC) according to the schedule described in panel A. Head to head comparisons were performed between LV and Ad5 candidates from two sources. In panel B, mice were primed with a plasmid encoding the multiHIV payload, then boosted with equal doses of either the LV VRX-multiHIV or with Ad5-multiHIV. Animals were sacrificed 1 month post-immunization and samples collected for analysis. T-cell responses (bars, left y-axis) are measured by ICS to HIV Gag and are shown as geometric means, along with standard deviations (percentages of cytokine positive cells (IFN␥, TNF␣ and IL2 combined on the same channel) within the CD8+ lymphocyte subset). Anti-HIV IgG were detected in the sera at 1:100 by ELISA (line graph; right y-axis). Results shown represent the ratio of the antibody response at time of sacrifice to the pre-immunization baselines’. In panels C and D, a second Ad5 candidate, VRC-Ad5, containing the same gag–pol transgene as its lentiviral counterpart, VRX1023, was evaluated. All animals were primed with the same pVRX1053 plasmid before boosting with equal doses of either vector. Panel C illustrates anti-HIV cellular responses obtained at 2 months post-vector boost with each candidate, and represents the cumulative geometric means and standard deviations within each group. Values are percentages of cells secreting all three cytokines simultaneously among CD4 T-cells (polyfunctional T-cells). Anti-HIV IgG responses in sera were assessed by ELISA at 10 days, 1 and 2 months post-vector boost (panel D). Lines represent geometric means in each group, along with standard deviations. Pre-immunization baseline values, similar among all animals, were averaged and showed as a dotted line.

used to assess the comparative immunogenicity of LV and adenoviral vectors (Ad5). A first Ad5 vector, kindly provided by Dr. Ioana Stanescu (FIT Biotech, Tampere, Finland), was tested head to head with our LV platform. As the Ad5 candidate contained the MultiHIV payload [17], a matching LV was constructed by replacing the gag–pol and nef transgenes from VRX1023 by the MultiHIV transgenes. In order for the DNA primes to match the transgenes from the boost, all mice were primed using the FIT Biotech’s GTU-MultiHIV plasmid, then boosted with equal doses of either vector, LV or Ad5, both containing the same MultiHIV payload (Fig. 5A). At time of sacrifice, 1 month post-boost, T-cell responses to HIV Gag in the LV-immunized group were significantly higher than in the Ad5immunized group (p = 0.018; Fig. 5B). Consistent with the T-cell responses findings, antibody responses to HIV also indicated a higher humoral immunogenicity for the LV candidate compared to the Ad5 immunogen. To extrapolate to other Ad5 candidates our findings that the VRX1023 LV was more immunogenic than Ad5, a LV versus Ad5 comparison was repeated by using another Ad5-based immunogen provided by the Vaccine Research Center, NIAID, NIH. As both the Ad5 and the LV candidates express HIV-1 clade B Gag and Pol, all groups of mice were primed with the same pVRX1053 DNA plasmid, then boosted with equal doses of either the Ad5 candidate VRC5409 or with the R&D-grade LV VRX1023. Results demonstrated that the two vector platforms overall elicited comparable T-cell responses to HIV, although, as shown in Fig. 5C, LV induced a

higher degree of polyfunctional responses to HIV antigens in CD4+ T-cells at 2 months. All other aspects of the T-cell immunity (TNF␣, IFN␥ and IL2 cytokines considered independently, CD8+ T-cells responses) showed only a moderate advantage provided by the LV platform that did not reach statistical significance (data not shown). When measuring the anti-HIV IgG responses in the serum, VRX1023 appeared to induce higher antibody levels from the early timepoints (p = 0.01 at day 10; Fig. 5D). Interestingly, this set of data also confirmed the persistent immunogenicity observed earlier with VRX1023, as the Ad5-induced anti-HIV antibodies declined over time, while the LV-induced antibodies kept on increasing over time throughout the course of the study (p = 0.003 at month 2), likely reflecting different expression patterns between these two vectors. 3.5. LV manufacturing can be scaled-up to GMP-grade while retaining its immunogenicity As lentiviral vector manufacturing was scaled-up ahead of clinical use, we wanted to demonstrate that cGMP-grade LV preparation could be performed, while retaining the immunogenicity of the research-grade candidate previously tested above. VRX1023 LV candidates produced either using the R&D-grade process, or using the cGMP-clinical grade process were compared head to head in mice, either alone or in a DNA prime, LV boost setting. Results shown in Fig. 6 demonstrate that mice vaccinated with equal dose of each immunogen responded with similar level of anti-HIV Gag T-cell immunity overall. Detailed analysis

1958

F. Lemiale et al. / Vaccine 28 (2010) 1952–1961

Fig. 6. LV manufacturing can be scaled-up to cGMP-grade while retaining its immunogenicity. Comparison between the two manufacturing processes was performed by immunizing mice either with lentiviral vector alone (panel A), or with DNA prime/lentiviral vector boost (panel B). Groups of five Balb/c mice were immunized SC twice at a 2-week-interval with 3 × 107 TU of VRX1023 produced using either manufacturing process (A), or were primed three times with 100 ␮g of pVRX1053 IM at weeks 0, 1 and 2, then boosted SC at week 4 with 3 × 107 TU of VRX1023 (panel B). All animals were sacrificed at 17 days post-last immunization and T-cell responses to HIV Gag were assessed by ICS. Results shown illustrate various HIV-specific cytokine secretions detected in CD8+ T-cell subsets following HIV Gag peptide stimulation. Polyfunctionality is defined as cells expressing all three cytokines (IFN␥, TNF␣ and IL2) simultaneously. Results are shown as individual animal responses, along with geometric mean in each group. None of the functions analyzed revealed any statistical difference between the two groups immunized with VRX1023, except for the IFN␥ response in panel B.

of the immunogenicity of each vector, however, revealed some qualitative differences. With no DNA prime, LV-induced T-cell responses to HIV Gag were characterized by an absence of IL2 secretion, but high levels of IFN␥ and TNF␣, similar between the two groups, and significant polyfunctional CD8 responses. When LV was combined with a DNA prime however, Gag-specific CD8 Tcells from both groups secreted high levels of TNF␣ and IL2, and show similar polyfunctionality, but cGMP-grade vectors induced significant IFN␥ responses that were absent in the R&D-grade group. 4. Discussion While the HIV vaccine field is still struggling with the interpretation of the outcome of high-profile Ad5-based HIV vaccine trials

and its exact implication for future trials, alternate vaccine strategies are dearly needed. Specifically, alternate viral vector platforms that could overcome the pitfalls of adenovirus-based strategies are in great need. With the observation that, in the non-human primate model, live-attenuated simian immunodeficiency viruses (SIV) have so far been the most efficient immunogens for providing protection from SIV challenge [21], but that this approach was rendered non viable for human use because of safety issues [21–23], we have attempted to develop a vaccine vector system that is safer than attenuated virus, and yet more potent than viral-like particles (VLPs). We have constructed and tested several HIV-1-based vectors. VRX1023, our selected and optimized lentiviral vector, is pseudotyped with vesicular stomatitis virus glycoprotein (VSV-G) envelope and provides LTR-driven expression of unmodified gag–pol and rev genes.

F. Lemiale et al. / Vaccine 28 (2010) 1952–1961

In the present study, in addition to demonstrating the immunogenicity of the VRX1023 candidate, route optimization has showed that SC injection of LV induces as good mucosal antibody responses to HIV as the IN route, known to be optimal for induction of mucosal immunity [19,24,25]. In addition, SC produces better systemic responses to the transgene than IN immunization (Fig. 3), while also outperforming all other systemic routes tested (data not shown). Indeed, it has been shown by others that delivery of antigens to the skin results in the induction of more potent immune responses with lower doses of antigen than other routes of immunization [26–28]. Recent results suggest that cutaneous delivery of LV enables transduction of dermal DCs, a cell type that can effectively present transgenic antigens to prime naïve T-cells. In fact, a study by He et al. [14] showed that dermal DCs are targeted through sub-cutaneous vaccination with lentiviral vectors in mice. The same study showed that cell-sorted dermal DCs form the most potent antigen-presenting subset after SC LV immunization. Earlier studies [29] also reported that after sub-cutaneous injection of LV, the majority of transduced cells in the draining lymph nodes (LN) have a DC phenotype. The presence of transduced cells in the LN is most likely a consequence of transduction at the site of injection followed by their migration to the draining LNs [30]. The viability of LV-transduced DCs is not altered and these cells can process and present transgenic antigens through both the class I and class II restricted processing pathways to prime naïve CD8+ and CD4+ T-cells [31]. VSV-G, the viral envelope used to pseudotype our vectors, is likely critical to VRX1023 immunogenicity following SC administration, as VSV-G was found to be the most efficient viral glycoprotein for cutaneous transduction [32]. Due to the very broad tropism of VSV-G pseudotyped viral particles, other cell types, such as macrophages, fibroblasts and muscle cells, are also transduced at the injection site. It is likely that transduced non-antigenpresenting cells such as keratinocytes can provide an important source of antigen for cross-presentation [33] and for the induction of T helper and B cell responses. The observation that LV injected into skin can stimulate significant levels of immune responses suggests that the combination of LV gene delivery and skin-targeted immunization is a promising approach. Superiority of the SC route has also been described by Lopes et al. [34], who have shown that SC immunization is superior compared to IV immunization. One of the hallmarks of successful vaccination is the induction of strong and persistent memory T-cell responses, which depends on antigen availability and the duration of antigen exposure during priming [35]. Antigen persistence for sufficient time and above a minimal quantitative threshold is necessary for successful priming [36]. Most vaccine strategies use non-persistent vectors, either alone or in various prime/boost regimens (DNA prime/modified vaccinia virus Ankara boost, DNA prime/Ad5 boost). Although these vectors can stimulate very strong CTL responses, these are short lived, as antigens are produced for a limited time [37–39]. As antigen exposure decreases, the memory T-cell responses they elicit become predominantly central memory T-cells. As these cells have limited effector function and require antigen-induced expansion, they may be inappropriate for preventing initial systemic dissemination and extensive early replication of HIV [40]. In contrast, vectors that provide persistent level of antigen expression maintain functionally differentiated effector memory T-cells that are more likely to control the early steps of HIV infection. This hypothesis has recently been confirmed by results in non-human primates, where effector memory T-cells elicited by persistent antigen expression, in absence of neutralizing antibody response, prevented the establishment of progressive SIV infection after mucosal challenge [40]. Our present results suggest that LV-mediated immunization may be able to stimulate similar type of responses, as we have showed here that LV injection induces anti-HIV cellular and humoral immune

1959

responses that persist over a period of weeks and months. The induced immunogenicity was significantly more sustained than the one induced by the Ad5-based vaccine candidates we tested using identical immunization conditions. A possible mechanism for our observations may be found in the observations by He et al. [14], who have observed that, following skin injection of LV, antigen presentation can be detected in the draining lymph nodes of mice 3 weeks after immunization. In contrast, antigen presentation after vaccinia vector skin-targeted immunization decreased rapidly and was barely detectable after 5 days [14]. While the superiority of LV to other viral vectors in terms of persistence of antigen expression is a key attractive feature, another no less important parameter to consider for viral vector-based vaccination is the pre-existing immunity to the vector and its sensitivity to antibody neutralization. While anti-vector immunity is a potential issue for any immunization regimen based on multiple booster injections, it is also a potential problem for any live recombinant viral or bacterial vaccine, because of their relative lack of efficacy in individuals previously exposed to the vector. It has been shown to be particularly important in the case of adenovirus-based vectors. These vectors, while considered to be among the most potent gene delivery vehicles for inducing T-cell immune responses, have their clinical development limited by vector immunogenicity. This appears to be a problem both in the context of widespread preexisting anti-vector immunity in targeted populations, and also with de novo generation of immunodominant T-cell immunity against vector antigens in response to vector delivery. These limitations have been demonstrated in the Merck STEP, and NIAID’s PAVE 100 studies of adenovirus 5 (Ad5)-based HIV vaccine candidates [41]. The recent disappointing results using the Ad5 platform [9,10] are a reminder that anti-vector immunity may play a significant role for discriminating between promising viral vector platforms. Contrary to Ad5, with high prevalence of neutralizing antibodies (NAbs) in most of the human population [42–45], there are no preexisting NAbs to envelopes used for pseudotyping LV, such as VSV-G [46]. Moreover, as observed in the past by various other groups, LV do not appear to induce significant levels of anti-vector immune responses. He et al. [14] have found that pre-exposure to immunizing doses of lentivector does not inhibit an immune response against an antigen delivered by a second injection of the same vector, indicating that vector-specific immunity does not limit the immunogenicity of lentivector vaccination or boosting at the tested doses. The same study has also shown that CD8+ T-cell immunity can be recalled by repeated immunizations with the same LV, suggesting that LV immunization induces little if any interfering anti-vector immunity compared to other viral vectors. This obviates a major limitation of adenovirus and other viral vectors that become ineffective for boosting or re-immunization due to the presence of anti-vector immune responses. Our own findings corroborate these previously published results, as we have shown here that LV can be repeatedly administered and still successfully boost the HIV-specific immune responses (Fig. 4B). Complementary to the present study, we have demonstrated in side-by-side comparison of LV to Ad5 that minimal anti-vector neutralizing activity is mounted in LV-immunized animals, compared to Ad5-immunized mice (B. Asefa et al., unpublished data). To fully characterize LV-mediated immunogenicity, we have shown here that LV immunization can significantly benefit from DNA priming. Similarly to the promising results generated by the combination of various other viral vectors to DNA prime, LV can indeed be successfully used in the same setting. In the present study, levels of anti-HIV cellular responses reached in DNA primed, LV-boosted animals were of extremely high magnitude, with some immunized mice having as high as 21% of CD8 T lymphocytes secreting cytokines to Gag stimulation only, 3 weeks post-LV injection. Interestingly, these high anti-HIV responses were obtained

1960

F. Lemiale et al. / Vaccine 28 (2010) 1952–1961

with the latest LV manufacturing improvements, which have significantly increased the immunogenicity of earlier research-grade vectors. Additionally, the transition to cGMP-grade manufacturing was performed without any loss of immunogenicity, while providing a much cleaner vector product, with some potential benefits regarding the tolerability of vaccination, as cell debris and other impurities, potentially inducers of inflammatory reactions, were removed. All the initial protocol optimization, route finding, and comparisons to Ad5 vectors described herein were performed using early generation, non-optimized preparations of LV. As our present findings have demonstrated a higher short- and long-term anti-HIV immunogenicity of LV compared to Ad5, the major immunogenicity improvements obtained with the latest LV generations would suggest an even greater superiority of LV compared to Ad5 candidates in terms of immunogenicity. As immunizations with different LV vector lots, containing various ratios of genome-encoding particles and empty VLPs, induced different magnitudes of responses to the various HIV antigens, our result demonstrate that LV immunogenicity is provided by both the transgene expression (Fig. 3) and by the antigen input provided by the vector particle itself (Fig. 4D). This is a rather unique characteristic, otherwise provided only by attenuated HIV/SIV vaccine formulations, which is capable of inducing a uniquely balanced response of humoral and cellular immunity. It combines the advantages of attenuated viruses and VLP-based immunogens, without the pitfalls of each platform. Lentiviral vectors are characterized by a range of unique features. Particularly when directed to the skin, their engineered tropism, kinetic of antigen expression and low sensitivity to vector neutralization bypass major pitfalls of current HIV vaccine strategies, while favorably competing, in terms of immunogenicity, with the most promising viral vectors, namely Ad5. As they have benefited from increased safety records after their clinical use in humans over the past few years [47], their development as HIV vaccine candidate should be considered, as alternate HIV vaccine strategies are currently dearly needed. Author disclosure statement F.L., B.A., D.Y., C.C., N.K. and L.H. are employees of VIRxSYS Corporation. Acknowledgements The authors are grateful to Drs. Gary McGarrity and David Weiner for critical review of the manuscript. References [1] Schmitz JE, Kuroda MJ, Santra S, Sasseville VG, Simon MA, Lifton MA, et al. Control of viremia in simian immunodeficiency virus infection by CD8+ lymphocytes. Science 1999;283:857–60. [2] Pantaleo G, Koup RA. Correlates of immune protection in HIV-1 infection: what we know, what we don’t know, what we should know. Nat Med 2004;10:806–10. [3] Shiver JW, Emini EA. Recent advances in the development of HIV-1 vaccines using replication-incompetent adenovirus vectors. Annu Rev Med 2004;55:355–72. [4] Mcmichael AJ. HIV vaccines. Annu Rev Immunol 2006;24:227–55. [5] Harari A, Bart PA, Stohr W, Tapia G, Garcia M, Medjitna-Rais E, et al. An HIV-1 clade C DNA prime, NYVAC boost vaccine regimen induces reliable, polyfunctional, and long-lasting T cell responses. J Exp Med 2008;205:63–77. [6] Johnston MI, Fauci AS. An HIV vaccine—evolving concepts. N Engl J Med 2007;356:2073–81. [7] Perreau M, Pantaleo G, Kremer EJ. Activation of a dendritic cell-T cell axis by Ad5 immune complexes creates an improved environment for replication of HIV in T cells. J Exp Med 2008;205(12):2717–25. [8] Liu J, O’brien KL, Lynch DM, Simmons NL, La Porte A, Riggs AM, et al. Immune control of an SIV challenge by a T-cell-based vaccine in rhesus monkeys. Nature 2009;457:87–91.

[9] Cohen J. AIDS research. Did Merck’s failed HIV vaccine cause harm? Science 2007;318:1048–9. [10] Sekaly RP. The failed HIV Merck vaccine study: a step back or a launching point for future vaccine development? J Exp Med 2008;205:7–12. [11] Levine BL, Humeau L, Boyer J, Macgregor R-R, Rebello T, Xiaobin L, et al. Gene transfer in humans using a conditionally replicating lentiviral vector. PNAS 2006;103:5. [12] Naldini L, Blomer U, Gallay P, Ory D, Mulligan R, Gage FH, et al. In vivo gene delivery and stable transduction of nondividing cells by a lentiviral vector. Science 1996;272:263–7. [13] Esslinger C, Romero P, Macdonald HR. Efficient transduction of dendritic cells and induction of a T-cell response by third-generation lentivectors. Hum Gene Ther 2002;13:1091–100. [14] He Y, Zhang J, Donahue C, Falo jr LD. Skin-derived dendritic cells induce potent CD8(+) T cell immunity in recombinant lentivector-mediated genetic immunization. Immunity 2006;24:643–56. [15] Lu X, Humeau L, Slepushkin V, Binder G, Yu Q, Slepushkina T, et al. Safe two-plasmid production for the first clinical lentivirus vector that achieves >99% transduction in primary cells using a one-step protocol. J Gene Med 2004;6:963–73. [16] Vladimir Slepushkin, Nancy Chang R, Cohen YG, Bing Jiang E, Deausen D, Berlinger G, et al. Large-scale purification of a lentiviral vector by size exclusion chromatography or mustang Q ion exchange capsule. J Bioprocess 2003:89–95. [17] Blazevic V, Mannik A, Malm M, Sikut R, Valtavaara M, Toots U, et al. Induction of human immunodeficiency virus type-1-specific immunity with a novel gene transport unit (GTU)-MultiHIV DNA vaccine. AIDS Res Hum Retroviruses 2006;22:667–77. [18] Huang Y, Kong WP, Nabel GJ. Human immunodeficiency virus type 1-specific immunity after genetic immunization is enhanced by modification of Gag and Pol expression. J Virol 2001;75:4947–51. [19] Lemiale F, Kong WP, Akyurek LM, Ling X, Huang Y, Chakrabarti BK, et al. Enhanced mucosal immunoglobulin A response of intranasal adenoviral vector human immunodeficiency virus vaccine and localization in the central nervous system. J Virol 2003;77:10078–87. [20] Lemiale F, Haddada H, Nabel GJ, Brough DE, King CR, Gall JG. Novel adenovirus vaccine vectors based on the enteric-tropic serotype 41. Vaccine 2007;25:2074–84. [21] Reynolds MR, Weiler AM, Weisgrau KL, Piaskowski SM, Furlott JR, Weinfurter JT, et al. Macaques vaccinated with live-attenuated SIV control replication of heterologous virus. J Exp Med 2008;205(11):2537–50. [22] Baba TW, Jeong YS, Pennick D, Bronson R, Greene MF, Ruprecht RM. Pathogenicity of live, attenuated SIV after mucosal infection of neonatal macaques. Science 1995;267:1820–5. [23] Baba TW, Liska V, Khimani AH, Ray NB, Dailey PJ, Penninck D, et al. Live attenuated, multiply deleted simian immunodeficiency virus causes AIDS in infant and adult macaques. Nat Med 1999;5:194–203. [24] Lubeck MD, Natuk RJ, Chengalvala M, Chanda PK, Murthy KK, Murthy S, et al. Immunogenicity of recombinant adenovirus-human immunodeficiency virus vaccines in chimpanzees following intranasal administration. AIDS Res Hum Retroviruses 1994;10:1443–9. [25] Gherardi MM, Perez-Jimenez E, Najera JL, Esteban M. Induction of HIV immunity in the genital tract after intranasal delivery of a MVA vector: enhanced immunogenicity after DNA prime-modified vaccinia virus Ankara boost immunization schedule. J Immunol 2004;172:6209–20. [26] Belshe RB, Newman FK, Cannon J, Duane C, Treanor J, Van Hoecke C, et al. Serum antibody responses after intradermal vaccination against influenza. N Engl J Med 2004;351:2286–94. [27] Kenney RT, Frech SA, Muenz LR, Villar CP, Glenn GM. Dose sparing with intradermal injection of influenza vaccine. N Engl J Med 2004;351:2295–301. [28] LA Montagne JR, Fauci AS. Intradermal influenza vaccination—can less be more? N Engl J Med 2004;351:2330–2. [29] Esslinger C, Chapatte L, Finke D, Miconnet I, Guillaume P, Levy F, et al. In vivo administration of a lentiviral vaccine targets DCs and induces efficient CD8(+) T cell responses. J Clin Invest 2003;111:1673–81. [30] Dullaers M, Thielemans K. From pathogen to medicine: HIV-1-derived lentiviral vectors as vehicles for dendritic cell based cancer immunotherapy. J Gene Med 2006;8:3–17. [31] He Y, Zhang JMIZ, Robbins P, Falo jr LD. Immunization with lentiviral vectortransduced dendritic cells induces strong and long-lasting T cell responses and therapeutic immunity. J Immunol 2005;174:3808–17. [32] Hachiya A, Sriwiriyanont P, Patel A, Saito N, Ohuchi A, Kitahara T, et al. Gene transfer in human skin with different pseudotyped HIV-based vectors. Gene Ther 2007;14:648–56. [33] He Y, Falo jr LD. Lentivirus as a potent and mechanistically distinct vector for genetic immunization. Curr Opin Mol Ther 2007;9:439–46. [34] Lopes L, Dewannieux M, Gileadi U, Bailey R, Ikeda Y, Whittaker C, et al. Immunization with a lentivector that targets tumor antigen expression to dendritic cells induces potent CD8+ and CD4+ T-cell responses. J Virol 2008;82:86–95. [35] Pulendran B, Ahmed R. Translating innate immunity into immunological memory: implications for vaccine development. Cell 2006;124:849–63. [36] Salerno-Goncalves R, Sztein MB. Cell-mediated immunity and the challenges for vaccine development. Trends Microbiol 2006;14:536–42. [37] Hanke T, Samuel RV, Blanchard TJ, Neumann VC, Allen TM, Boyson JE, et al. Effective induction of simian immunodeficiency virus-specific cytotoxic T lymphocytes in macaques by using a multiepitope gene and DNA prime-modified vaccinia virus Ankara boost vaccination regimen. J Virol 1999;73:7524–32.

F. Lemiale et al. / Vaccine 28 (2010) 1952–1961 [38] Amara RR, Villinger F, Altman JD, Lydy SL, O’neil SP, Staprans SI, et al. Control of a mucosal challenge and prevention of AIDS by a multiprotein DNA/MVA vaccine. Science 2001;292:69–74. [39] Robinson HL, Weinhold KJ. Phase 1 clinical trials of the National Institutes of Health Vaccine Research Center HIV/AIDS candidate vaccines. J Infect Dis 2006;194:1625–7. [40] Hansen SG, Vieville C, Whizin N, Coyne-Johnson L, Siess DC, Drummond DD, et al. Effector memory T cell responses are associated with protection of rhesus monkeys from mucosal simian immunodeficiency virus challenge. Nat Med 2009;15(3):293–9. [41] Fauci AS, Johnston MI, Dieffenbach CW, Burton DR, Hammer SM, Hoxie JA, et al. HIV vaccine research: the way forward. Science 2008;321:530–2. [42] Kostense S, Koudstaal W, Sprangers M, Weverling GJ, Penders G, Helmus N, et al. Adenovirus types 5 and 35 seroprevalence in AIDS risk groups supports type 35 as a vaccine vector. Aids 2004;18:1213–6.

1961

[43] Nwanegbo E, Vardas E, Gao W, Whittle H, Sun H, Rowe D, et al. Prevalence of neutralizing antibodies to adenoviral serotypes 5 and 35 in the adult populations of The Gambia, South Africa, and the United States. Clin Diagn Lab Immunol 2004;11:351–7. [44] Sumida SM, Truitt DM, Kishko MG, Arthur JC, Jackson SS, Gorgone DA, et al. Neutralizing antibodies and CD8+ T lymphocytes both contribute to immunity to adenovirus serotype 5 vaccine vectors. J Virol 2004;78:2666–73. [45] Sumida SM, Truitt DM, Lemckert AA, Vogels R, Custers JH, Addo MM, et al. Neutralizing antibodies to adenovirus serotype 5 vaccine vectors are directed primarily against the adenovirus hexon protein. J Immunol 2005;174:7179–85. [46] Palmowski MJ, Lopes L, Ikeda Y, Salio M, Cerundolo V, Collins MK. Intravenous injection of a lentiviral vector encoding NY-ESO-1 induces an effective CTL response. J Immunol 2004;172:1582–7. [47] Lemiale F, Korokhov N. Lentiviral vectors for HIV disease prevention and treatment. Vaccine 2009;27(25–26):3443–9.