Inhibition of antigen presentation during AAV gene therapy using virus peptides

Abstract

The clinical trial using adeno-associated virus (AAV) vector delivery of mini-dystrophin in patients with Duchenne Muscular Dystrophy (DMD) demonstrated a cytotoxic lymphocyte (CTL) response targeting the transgene product. These mini-dystrophin-specific T-cells have the potential to clear all transduced muscle, presenting the general gene therapy concern of overcoming the CTL response to foreign proteins that provide therapeutic benefit. In this study, we exploited a natural immunosuppression strategy employed by some viruses that results in CTL evasion only in transduced cells. After transfection of the plasmids encoding viral peptides and ovalbumin, which includes the immune-domain epitope SIINFEKL, several viral small peptides (ICP47 and US6) inhibited the SIINFEKL peptide presentation. A single AAV vector genome that consisted of either transgene AAT fused with SIINFEKL epitope and, separately, ICP47 expressed from different promoters or a single fusion protein with ICP47 linked by a furin cleavage peptide (AATOVA-ICP47) decreased antigen presentation. Compared with AAV/AATOVA in which decreased AAT expression was observed at late time points, persistent transgene expression was obtained after systemic administration of AAV/AATOVA-ICP47 vectors in mice. We extended this strategy to DMD gene therapy. After administration of AAV vector encoding human mini-dystrophin fusion protein with ICP47 into mdx mice, a lower mini-dystrophin-specific CTL response was induced. Importantly, the ICP47 fusion to mini-dystrophin inhibited CTLs mediated cytotoxicity. Although demonstrated herein using AAT and mini-dystrophin transgenes in an AAV context, the collective results have implications for all gene therapy applications resulting in foreign peptides by immune suppression in only genetically modified cells.

Introduction

Duchenne muscular dystrophy (DMD) is the most common form of muscular dystrophy during childhood. DMD affects 1/3500 males and one third of these cases are caused by spontaneous mutations or deletions of the dystrophin gene. The progressive muscle degeneration restricts patients to a wheelchair in their early teens and leads to death after age 18 due to respiratory infection, complicated by heart failure. Dystrophin is a large (427 kDa) cytoskeletal protein in both skeletal and cardiac muscle. Currently, there is no cure for patients with DMD and prednisone was a routine treatment (1).

Gene therapy represents a promising approach to cure this disease via delivering (using either viral or non-viral vectors) a functional copy of a gene or by repairing the mutated locus. Among gene delivery vehicles, adeno-associated virus (AAV) vectors have been extensively studied for DMD gene therapy (2–29). AAV is a non-pathogenic single-strand DNA parvovirus whose replication relies on helper functions provided by the co-infection of particular viruses (i.e. adenovirus or herpes virus). AAV transduces both dividing and non-dividing cells and has broad tissue tropism from many serotypes and variants. Importantly, recombinant AAV (rAAV), in which transgenic DNA substitutes all viral open reading frames, induces long-term episomal transgene expression with no to rare integration events in host chromosomes (30). To date, over 150 Phase I clinical trials with rAAV have been carried out without acute adverse events attributable to the vector.

AAV vectors have demonstrated therapeutic effects for the treatment of DMD in animal models, including mice, rats, and dogs. Additionally, the characterization of isolated AAV serotypes expedites their application for DMD therapy. For instance, injection of AAV8 into neonatal mice results in transduction of every muscle in the body, including the heart and diaphragm (31). The systemic application of AAV9 induces extensive transduction in the heart while AAV6 muscular injection induces strong transgene expression in mdx mice (6,31–33). Due to the rAAV packaging limitation (<5kb) and the size of dystrophin cDNA (about 14kb), a panel of ‘mini’- and ‘micro’-dystrophin genes have been developed. These cassettes are able to be entirely packaged into AAV virions and have demonstrated success in mdx mice and dogs with DMD (15,17). Based on extensive studies in animal models, Phase I clinical trials have been initiated by delivery of mini-dystrophin into muscles via AAV vectors in patients with DMD. However, after direct muscular administration of AAV vectors encoding mini-dystrophin (minidys), we noted a CTL response to the dystrophin, which may relate to revertant fiber development induced prior to gene therapy (34). Unfortunately, about 50% of DMD patients have revertant fibers with Dystrophin expression (35–38). Revertant fibers are also observed in animal models (murine and canine) (39–42). These results highlight a major challenge for the gene therapy community in general: avoiding a CTL immune response against the therapeutic transgene product. General immunosuppression regimens using chemotherapy agents and antibodies have been proposed to prevent CTL immunity (14,28). This strategy will affect the whole body immune response which may not be necessary for gene therapy. For CTL mediated elimination of transduced cells after gene therapy, the optimal approach should be to only impact the gene therapy vector transduced cells. This notion led to the design of an effective strategy to evade CTL mediated killing by interfering with antigen presentation only in transduced cells. Viruses found in nature rely on strategies to evade the human immune response including the production of small viral proteins that interfere with antigen presentation (termed VIPRs). VIPRs can interfere with the MHC class I presentation pathway at essentially every step, from antigen degradation to trafficking via the Golgi-secretory pathway to the cell surface (43–57). In this study, we demonstrated that several VIPRs inhibited antigen presentation from endogenous expression proteins, and fusion of the VIPRs with mini-dystrophin decreased antigen presentation and blocked dystrophin-specific CTL response. These dystrophin-specific CTLs did not eliminate target cells with expression of mini-dystrophin fusion protein with VIPR efficiently.

Results

Viral peptides inhibit antigen presentation from endogenous expressed protein

Particular viruses can persist within a host for the lifetime of the organism with the help of viral proteins interfering with antigen presentation (VIPRs). To determine whether VIPRs can inhibit antigen presentation, we co-transfected a plasmid encoding VIPR US6 (from cytomegalovirus, 164 amino acids) or ICP47 (from herpes simplex virus, 88 amino acids) into 293 cells along with plasmids encoding ovalbumin, which contains the potent ovalbumin SIINFEKL peptide (OVA) sequence, and its cognate MHC molecule (H-2k^b) to allow SIINFEKL/H-2k^b complex surface presentation. Next, these cells were co-incubated with OT-1 spleen cells and stained for CD8 and CD69 (early activation markers) or intracellular IFN-γ using flow cytometry. When analyzing the percentage of CD8⁺CD69⁺ (activated T-cells) cells, both of the VIPR treated groups (ICP47 or US6) inhibited OVA epitope presentation up to 95% compared with the positive control group (Fig. 1A). Similar results demonstrated the ability of ICP47 or US6 to inhibit T-cell activation >10-fold when co-staining for CD8 and intracellular IFN-γ (Fig. 1B).

Figure 1.

Inhibition of OVA epitope presentation by VIPRs. 293 cells were transfected with 1 μg of pCBA- H2k^b and 1 μg of pCMV-OVA as well as 1 μg of pCMV-ICP47 or pCMV- pUS6 in a 6-well plate. 24h later, OT-1 spleen cells were added and incubated. After co-culture overnight, the cells were collected for CD69 expression (A) and intracellular IFN-γ staining (B) of CD8⁺ OT-1 spleen cells by flow cytometry. The result was the representative of three independent experiments.

The ability of VIPRs to inhibit antigen presentation was also confirmed in the cell lines from different species including: (i) murine B16, (ii) rat C6, (iii) canine MDCK, (iv) monkey COS-1, and (v) human 293s as our positive control. To determine the relative potency of ICP47 and US6 in these contexts, various ratios of OVA antigen (fixed to the amount of H-2k^b) and VIPR plasmids were co-transfected. In this system, a ratio of 1: 1 of VIPR to OVA indicates that the same concentration of VIPR and OVA plasmid was used for transfection; while 1: 10 and 1: 100 ratios indicate that a 10-fold and 100-fold lower concentration of VIPR DNA was used compared with that the OVA plasmid, respectively. As shown in Figure 2, ICP47 potently inhibited antigen presentation in human, monkey, and dog cells. Similar results were obtained with US6. However, stronger inhibition of antigen presentation from US6 was achieved in murine B16 cells compared with ICP47 (Fig. 2). Both ICP47 and US6 only had modest effects on inhibition of antigen presentation in rat C6 cells. These results indicate that the function of VIPRs is, at least in part, species dependent.

Figure 2.

VIPRs inhibition of antigen presentation in cells of different species. VIPR plasmids (pCMV-ICP47 or pCMV-US6) and pCMV-OVA at different ratios, as well as pH2k^b, were co-transfected into different cells. After incubation with OT-1 spleen cells, T-cell activation was detected by staining CD8⁺CD69⁺ cells (% inhibition y-axis). (A) ICP47, (B) US6. The result was the mean and standard deviation from three independent experiments.

To study the inhibition by a VIPR on antigen presentation in vivo, we used a tumor xenograft mouse model. Due to its smaller size, compared with US6, which is 164 amino acids, VIPR ICP47 (88 amino acids) was used in following experiments to increase utility for AAV therapeutic applications. 293 cells were transfected with plasmids expressing OVA and the MHC molecule H-2k^b, as well as either an ICP47 or GFP (control). Then, 1×10⁷ 293 cells in matrigel were subcutaneously injected into the left flank of SCID/NOD mice. Immediately thereafter, in vitro activated OT-1 spleen cells were transfused via the tail vein and tumor size was measured every 2–3 days. After 20 days, no tumors were found in mice transplanted with 293 cells transfected with GFP demonstrating the ability of the activated OT-1 cells to kill OVA presenting 293 cells in vivo (Table 1). In stark contrast, all mice receiving 293 cells transfected with ICP47 developed tumors in the presence of activated OT-1 cells (Table 1). As expected, control mice receiving GFP transfected cells without OT-1 infusion had tumors in similar size to the ICP47 treatment group (Table 1). These results demonstrate the capacity of VIRP ICP47 for immune evasion via inhibition of transgenic antigen presentation in vivo.

Table 1.

The effect of VIPR on antigen presentation in human cells in vivo

	293/GFP	293/GFP + OT-1	293/ICP47	293/ICP47 + OT-1
Tumor formation	6/6	0/6	4/4	4/4
Tumor size (cm3)	0.545± 0.139	0	0.612±0.198	0.483±0.157

Inhibition of antigen presentation from a therapeutic cassette using VIPR linked by a furin cleavage site

To tailor VIPRs for AAV gene therapy, we initially made the construct pTR/CBA-AATOVA/CMV-ICP47 in which two different transgenes, alpha-1 antitripsin (AAT) and ICP47, were expressed from different promoters, CBA or CMV, respectively (Supplementary Material, Fig. S1). After transfection of the construct pTR/CBA-AATOVA/CMV-ICP and H2k^b plasmid into 293 cells, OVA antigen presentation was analyzed. The construct including ICP47 did not induce any antigen presentation when compared with the construct without ICP47 (Supplementary Material, Fig. S1). Although, this genetic arrangement works demonstrates VIPR-mediated inhibition of antigen presentation, the construct of pTR/CBA-AATOVA/CMV-ICP is too large to be packaged (>5kb) as an intact genome in AAV capsid. Therefore, we explored whether two transgenes expressed from a single promoter yet fused by a furin cleavage site (single ORF) would also inhibit antigen presentation. As such, we have made the construct pTR-CBA-AATOVA-ICP47, in which human AAT cDNA was fused with the OVA peptide containing SIINFEKL sequence and linked to ICP47 by a furin cleavage site as depicted in Figure 3. After transfection of pTR-CBA-AATOVA-ICP47 into 293 cells with H-2k^b expression plasmid, 293 cells then were incubated with OT-1 spleen cells. The activation of OT-1 spleen cells was analyzed for CD69 expression by flow cytometry. As shown in Figure 3, no OT-1 CD8 cell activation was observed from the AAT expression cassette (negative control) while AATOVA induced OT-1 activation to 16.35%. Antigen presentation in cells transfected with AAVOVA linked to ICP47 decreased by 75% (4.1% of OT-1 T cell activation) suggesting immune evasion using the furin linked context. To investigate this, AAV2 vector preparations packaged with these cassettes were tested in vivo using a murine model following systemic administration. At different time points post-injection, serum was collected for transgene product analysis. At 6 weeks post-injection, AAV2/CBA/AAT-OVA-ICP47 resulted in stable AAT expression out 16 weeks (Fig. 4). This was in contrast to a gradual decrease in AAT over the time in mice treated with AAV2/CBA/AAT-OVA which doesn’t contain the VIRP sequence (Fig. 4). This result suggests that ICP47 blocks CTL mediated cytotoxicity of target cells within 16 week observation in vivo. It was noted that the AAT expression in mice receiving AAT-OVA-ICP47 vectors was lower than that in mice with AAT vector at week 2 post AAV administration (P < 0.05) and then plateaued thereafter (Fig. 4), which was different from that of the AAT vector (peaked at week 6 post AAV vector injection). This result may due to the low efficiency of ICP47 to interfere with murine TAP function and the balance between antigen presentation from the transgene and ICP47 interference (49). The weak CTL immune response induced at the early time point from AAT-OVA-ICP47 was still able to eliminate some AAV transduced cells and then resulted in lower AAT expression. It is unlikely that low antigen presentation was attributed to the low AAT expression from AAT-OVA-ICP47 vectors since similar transduction was obtained at week 1 after AAV vector administration in mice with the AAV vector. Furthermore AAT expression in mice receiving the AAT-OVA vector continually decreased to lower levels than in mice administered with AAT-OVA-ICP vectors at week 16 post injection (Fig. 4, P < 0.05). These results indicate that similar antigen (AAT) was produced in every transduced cell among different vectors.

Figure 3.

The inhibitory effect of ICP47 on antigen presentation from a transgene linked to ICP47 by furin. 293 cells were co-transfected with 1 μg of pCMV-H2k^b and 1 μg of pTR/CBA-AATOVA-ICP47 or pTR/CBA-AAT or pTR/CBA-AATOVA in a 6 well plate for 24 hr. OT-1 spleen cells were added and incubated overnight. CD69 expression on CD8 cells was detected by flow cytometry. The result is representative of three independent experiments.

Figure 4.

The kinetics of transgene expression after administration of AAV2/CBA-AATOVA-ICP47 vectors. 1 × 10¹⁰ particles of AAV2/AAT or AAV2/AATOVA, or AAV2/AATOVA-ICP47 vectors were administered via retro-orbital injection into C57BL mice. At different time point, the blood was collected and AAT level in the plasma was detected by ELISA. The data represent the average and standard deviation from 4 mice. *** indicated the P < 0.001, ns indicated the P > 0.05, when AAT expression at week 16 was compared with AAT expression at week 2 for the same group.

To gain additional evidence that the target cells transduced by an AAV vector encoding a VIPR fusion protein can evade transgene-specific CTL-mediated elimination, we carried out the following experiment. Two weeks after systemic administration of AAV2 vectors, activated OT-1 spleen cells were infused by tail vein injections since at that time no obvious AAT expression was inhibited as shown in Figure 4. Two days after OT-1 cell infusion, the blood was collected for AAT analysis. As shown in Figure 5, significant decrease of AAT level in blood was only observed in mice treated with AAV2/CBA/AAT-OVA vectors, but not in mice receiving AAV2/CBA/AAT or AAV2/CBA-/AAT-OVA-ICP47. The collective results implicate that introduction of VIPR into a transgene cassette by a linker furin cleavage site inhibits transgene antigen presentation and therefore immune evasion of transgene-specific CTLs in AAV transduced target cells.

Figure 5.

Evasion of OVA-specific CTLs after administration of AAV2/CBA-AATOVA-ICP47 vectors. At week 2 after administration of 1 × 10¹⁰ particles of AAV vectors into C57BL mice, activated OT-1 cells were infused. Two days later, the AAT level in the blood was detected by ELISA. The data represent the average and standard deviation from 4 mice. ** indicated the P < 0.01, ns indicated the P > 0.05.

Mini-dystrophin/VIPR inhibits an antigen-specific T cell response

A previous clinical trial for treatment of DMD suggested that patient T cells specific to the dystrophin protein (possibly from revertant fibers), eliminated cells transduced with AAV-mini-dystrophin. To investigate the effect of VIPRs on antigen presentation following mini-dystrophin transgene expression, we first cloned the mini-dystrophin transgene fused to a VIPR (ICP47) by a furin cleavage sequence. This mini-dystrophin-ICP47 fusion is driven by CMV promoter (pTR-CMV-dys/ICP). Next, we determined whether mini-dystrophin-ICP47 was able to inhibit antigen processing and presentation compared with mini-dystrophin without the VIRP. At 24 h after co-transfection of pTR-CMV-dys/ICP or pTR-CMV-dys or pTR/CMV-ICP47 in the presence of OVA and H-2k^b expression plasmids (pCMV-OVA and pCBA-H2k^b, respectively) in 293 cells, spleen cells from OT-1 mice were added. After overnight incubation, OT-1 spleen cells were collected and analyzed for OVA-specific CD8 T cell activation by measurement of the early activation marker CD69. The results demonstrate that ICP47 peptide directly driven by the CMV promoter (positive control) induced inhibition of OVA antigen presentation by 97.2% when compared with the no VIPR control. When ICP47 was fused to mini-dystrophin, OVA antigen presentation was still inhibited albeit to a lesser extent (67.2% inhibition compared with the no VIPR control, Supplementary Material, Fig. S2). The above results demonstrate two important outcomes: (i) the mini-dystrophin/ICP47 fusion inhibits OVA antigen presentation in trans due to the fusion to ICP47, and (ii) mini-dystrophin/ICP47 does not result in the full extent of antigen inhibition as demonstrated by expression of only ICP47 (Supplementary Material, Fig. S2). However, the data from the plasmid transfection do indicate that ICP47 from mini-dystrophin fusion protein executes the function to inhibit antigen presentation.

Since fusion of ICP47 to the mini-dys cassette increases AAV vector size, next we examined whether ICP47 fused to the mini-dys expression cassette compromised the transgene dystrophin expression. We made AAV2/CMV-dys and AAV2/CMV-dys-ICP47. After transduction in 293 cells using the same vector genome particles in the presence of adenovirus, at day 3, the cell lysate was collected for western blot. It was found that a little lower (20% decrease) dystrophin expression from CMV-dys-ICP47 vectors was observed than that from CMV-dys vectors. This result may result from the contamination of some fragment vectors in CMV-dys-ICP47 vector production due to the larger size of CMV-dys-ICP47 (58) (Supplementary Material, Fig. S3).

To study the antigen presentation from AAV-mini-dystrophin transduction in vivo, AAV rh32/33 was chosen since it has been reported that rh32/33 administration is able to elicit much stronger CTL response to the transgene product than other AAV serotypes (59). We have demonstrated that muscular injection of an AAV/rh32/33vector encoding AAT with integration of the OVA SIINFEKL epitope induced an OVA-specific CTL response which killed SIINFEKL peptide pulsed target cells (Supplementary Material, Fig. S4). To study whether ICP47 inhibits antigen presentation from human mini-dystrophin in mdx mice, we packaged CMV-mini-dystrophin or CMV-mini-dystrophin-ICP cassette into AAV capsid Rh32/33. First, we determined whether a mini-dystrophin-specific CTL response would be elicited from AAV mediated mini-dystrophin delivery. At week 6 after direct muscular injection of AAV rh32.33 encoding mini-dystrophin, the spleen cells were harvested for mini-dystrophin CTL analysis by detection of intracellular IFN-γ staining. As shown in Figure 6, the spleen cells were stimulated with different pools of peptides (DP) derived from mini-dystrophin and PBS as a negative control. The strongest intracellular IFN-γ expression was achieved in CD8 lymphocytes with peptide pool 3 (DP3). This result indicates that a mini-dystrophin-specific CTL response was induced after AAV mediated delivery of mini-dystrophin in mdx mice and, therefore, we chose to use mini-dystrophin peptide pool DP3 for following studies.

Figure 6.

Mini-Dystrophin-specific CTL response was elicited in mice receiving AAV/CMV-mini-dys vector administration. 5 × 10¹⁰ particles of AAV rh32/33/mini-dys vectors were administered via muscular injection in mdx male mice. At week 6, mice were euthanized and spleen cells were harvested. The spleen cells were treated with different dystrophin peptide pools (DP1, DP2, DP3, DP4, and DP5) for 2 days. The intracellular IFN-γ expression on CD8 spleen cells was detected by flow cytometry. A, the representative results were shown. B, the data represent the average and standard deviation from 4 mice. ** indicated the P < 0.01.

Next, we administered AAV rh32/33 vectors encoding mini-dystrophin fused with ICP47 into mdx mice via muscular injection. At week 6 after AAV injection, spleen cells were incubated with mini-dystrophin peptide pool DP3 and then intracellular IFN-γ staining was performed. As shown in Figure 7, a mini-dystrophin-specific CTL response was elicited against peptide pool DP3 in mice treated with AAV rh32/33 encoding mini-dystrophin. The results demonstrate that the mini-dystrophin-specific CTL response was significantly lower in mice treated with AAV rh32/33/mini-dys-ICP47 than in mice with AAV rh32/33/mini-dys vectors (Fig. 7). These results indicate that ICP47 may potentially inhibit the mini-dystrophin-specific CTL response after AAV delivery of the fusion protein. In addition, it is impossible to rule out the possibility that the lower dystrophin-specific CTL response was contributed from low mini-dystrophin expression in mdx mice receiving AAV rh32/33/mini-dys-ICP47.

Figure 7.

Antigen presentation inhibition from mini-dystrophin fusion with ICP47. 5 × 10¹⁰ particles of AAV rh32/33/mini-dys-ICP47 or AAV rh32/33/mini-dys vectors were administered via muscular injection in mdx male mice. At week 6, mice were euthanized and spleen cells were harvested. The spleen cells were treated with dystrophin peptide pool 3 for two days, the intracellular IFN-γ expression on CD8 cells was detected by flow cytometry. The data represent the mean and standard deviation of five mice. ** indicated the P < 0.01, * indicated the P < 0.05, when compared with CMV-dys group.

Cells expressing mini-dystrophin-ICP47 evade mini-dystrophin-specific T cell eradication

The above study demonstrated that administration of AAV vector encoding ICP47 fusion protein inhibited the transgene-specific CTL response. Next, we investigated whether the target cells with mini-dystrophin expression were able to evade elimination mediated by dystrophin- specific CTLs. A lot of studies have demonstrated that no human dystrophin-specific CTL response was elicited in mdx mice after direct intra-muscular injection of AAV vectors, and Treg or anergy/exhausted T lymphocyte can be induced after intramuscular injection of AAV vectors (60,61). Additionally, ICP47 doesn’t function very efficiently in mouse tissues (49). Therefore, we performed this experiment in an ex vivo system with specific mini-dystrophin CTLs obtained from AAV/mini-dystrophin treated mice as effector cells, and 293/H2k^b cells, which can express H-2k^b stably, with mini-dystrophin expression by transfection of plasmid TR/CMV-mini-dys or plasmid TR/CMV-mini-dys-ICP as target cells. The spleen cells from mice with AAVrh32/33/CMV-mini-dystrophin vector administration were collected and incubated with 293/H2k^b cells transfected with plasmids pTR/CMV-GFP or pTR/CMV-mini-dys or pTR/CMV-mini-dys-ICP at a ratio of effector to target 100: 1 for 4 hrs. Then, the supernatant was harvested for an LDH release assay. As shown in Figure 8, compared with target 293/H2k^b cells transfected with pTR/CMV-GFP, much more 293/H2k^b cells transfected with pTR/CMV-mini-dystrophin were destroyed by mini-dystrophin-specific T cells as indicated by the high release of LDH in supernatant. Importantly, 293/H2k^b cells transfected with pTR/CMV-mini-dys-ICP47 demonstrated a significant decrease in LDH release when compared with the no VIPR control (Fig. 8). This result indicates that ICP47 prevents mini-dystrophin-specific CTL eradication of cells expressing mini-dystrophin.

Figure 8.

ICP47 expressing cells evade elimination mediated by mini-dystrophin-specific CTLs. Spleen cells were collected from mdx mice at week 4 after muscular administration of 5 × 10¹⁰ particles of AAV rh32/33/CMV-dys. 293 cells transfected with pTR/CMV-dys or pTR/CMV-dysOVA for 24 hrs were harvested. 1 × 10⁶ spleen cells were incubated with 1 × 10⁴ 293 cells with dystrophin expression for 4 hrs and supernatant was collected for LDH detection. The cytotoxicity was calculated. The results were the average and standard deviation from 4 mice. ** indicated the P < 0.01, when compared with dys group.

Discussion

Duchenne muscular dystrophy (DMD) usually results from the lack of a functional dystrophin protein in muscle fibers due to various mutations in the dystrophin gene. In our Phase I clinical trial, a dystrophin-specific CTL response was elicited after AAV mediated delivery of mini-dystrophin. This clinical result brings a concern to all gene therapy communities employing a strategy in which foreign peptides will be generated. In this study, we characterized a novel approach to decrease antigen-specific CTL induction using two therapeutically relevant transgene products by blocking antigen presentation and therefore CTL mediated elimination, in only AAV transduced cells.

In the clinical study with AAV mediated delivery of mini-dystrophin, a dystrophin-specific T cell immune response was related to dystrophin expression from revertant fibers that might serve as antigens to elicit T cell immunity (34). Revertant fibers could result from spontaneous mutations that restore the open reading frame in a small number of myofibers (62,63). The revertant fibers are found in up to 47% of patients with DMD (35,37,62–64). In a recent survey, approximate 30% of DMD patients have T cell immunity against dystrophin, and the development of dystrophin-specific T cell response increases in aged patients and decreases in patients with corticosteroid therapy. Several facts suggest that the dystrophin CTL response after gene therapy may be more severe in DMD patients including the high expression of HLA class I molecules on regenerating muscle fibers, more efficient antigen presentation in dystrophic fibers, and the high amount of antigen presentation cells residing in dystrophic muscles (65–68).

Commonly, immunosuppression, such as chemotherapy drugs (cyclosporine, mycophenolate mofetil) and antibodies (anti-thymocyte globulin), has been used to eliminate effective CTLs or block the interaction of CTLs and treated cells (14,28). However, this systemic approach comes with a long list of adverse effects and, when considering a CTL response to a long-term production of foreign therapeutic product, would have no clear end to the immunosuppression regimen. The ideal approach to inhibit specific CTL induction or block pre-existing CTL-mediated elimination of AAV transduced cells is to express specific molecules able to interfere with antigen presentation pathway from the AAV transgene cassette. This approach ensures that all cells that express mini-dystrophin, or other therapeutic proteins, undergo immune evasion. Numerous viruses have evolved VIPRs to interfere with the immune response, specifically in transduced cells (43–57). VIPRs have evolved to inhibit the MHC class I presentation pathway, specifically in transduced cells, at essentially every step including: i) inhibiting transcription of nearly all components of the MHC class I pathway, including MHC class I heavy chain (HC), β₂ microglobulin (β₂-m), TAP1, TAP2, and tapasin, ii) blocking peptide degradation by proteasomes, iii) inhibiting TAP and tapasin function, iv) redirecting the MHC class I molecule into the cytosol for proteasome-mediated degradation, v) retaining MHC class I molecules in the ER/Golgi to prevent surface presentation, vi) mis-leading MHC class I molecules to the lysosome for degradation, and vii) enhancing endocytosis of MHC class I molecules on the cell surface into endosomes for degradation (43–46). US6 is one of four VIPRs encoded by human cytomegalovirus, and functions to inhibit the transporter for antigen processing (TAP), thereby blocking the import of peptides into the endoplasmic reticulum (ER) for loading onto MHC class I complexes (48,50,52). Herpes simplex virus (HSV) is an α-herpesvirus that establishes lifelong infection in neuronal cells from which it periodically reactivates. HSV inhibits antigen presentation on MHC class I by the evolved gene product ICP47 (49,53–56). ICP47 directly targets MHC class I antigen presentation by binding to the TAP complex, preventing transport of peptides from the cytosol to the ER where peptides are loaded into the nascent MHC class I heavy-chain β₂ microglobulin (β₂m) complex. Consistent with previous studies, co-expression of US6 or ICP47 inhibited antigen presentation and the efficiency was noted as species-dependent. Both US6 and ICP47 inhibited antigen presentation in human cells but not in rat cells (Fig. 2). Generally, high amount of US6 or ICP47 were required to effectively block antigen presentation in murine cells (Fig. 2). It has been demonstrated that ICP47 poorly inhibits TAP in mouse cells (57). This is due to ∼100-fold decreased binding of ICP47 to murine TAP as compared with human TAP (49). However, this study still demonstrated the effective inhibition of antigen presentation from ICP47 in mice after administration of AAV vector encoding therapeutic transgene linked with ICP47. It is plausible that stronger VIPRs in mice may exert a more profound antigen presentation inhibition if these VIPRs are fused to therapeutic transgene.

One general concern of VIPR utilization is that down-regulation of MHC I expression on the cell surface may increase natural killer (NK) cell-mediated eradication of target cells (69–71). Although our preliminary data in vivo suggested that NK cells may not clear AAV-VIPR transduced target cells after administration of AAV/AATOVA-ICP47, the possibility still exists, and the sensitivity of NK cells on MHC class I deficient cells may be different between human and mouse. If this is the case, alternate strategies may be explored to block NK cell recognition of AAV transduced cells in conjunction with VIPR production. For example, integration of the shRNA miR-UL112 from CMV into the transgene cassette with VIPR fusion may allow evasion from NK cell-mediated elimination while also allowing escape from the CTL effect (72,73). It has been demonstrated that miR-UL112 reduced the target cell killing by NK cells via the down-regulation MICB expression. MICB is a stress-induced ligand of the NK cell activating receptor NKG2D and is critical for NK cell killing of virus infected cells and tumor cells (72,73).

Another potential concern is that increasing the size of the AAV vector genome will comprise the AAV packaging/production capacity and inclusion of a full length VIPR (ICP47 is 88 amino acids or US6 164 amino acids) increases the transgenic mini-dystrophin cassette size to 5.5kb. Studies have demonstrated that a short fragment of ICP47, 32 amino acids (amino acids 3–34), was found to be the minimal region harboring the ability to inhibit peptide-binding to TAP (74) thereby increasing the ‘therapeutic’ gene packaging capacity. In other studies, the C-terminal 20 residues from the luminal domain of US6 are demonstrated to be essential for the inhibition of TAP (75). Future studies will investigate whether these minimal VIPR elements fused to therapeutic transgenes can elicit similar levels of immune evasion compared with the larger sequences investigated herein.

The third concern for AAV vector mediated gene delivery is the capsid-specific CTLs that can eliminate AAV transduced target cells (76–78). Since capsid antigen cross-presentation from AAV transduction is proteasome dependent (79), it is highly possible that VIPR fused to transgene may inhibit the antigen presentation from AAV capsid proteins. This experiment is worth to perform in future studies.

Recently, several studies from pre-clinical and clinical trials have demonstrated that AAV gene delivery also can induce tolerance to the AAV vector mediated by anergy/exhausted T cells and regulatory T cells after AAV vector injection (60,61,80–85). The mechanisms of Treg generation after AAV administration remain unknown. The following factors may contribute to Treg induction: (i) long-term transgene induces T cell exhaustion and/or Treg initiation by the persistent antigen exposure, (ii) the administration route, (iii) AAV capsids from different serotypes, (iv) the antigen from different transgenes and (v) AAV doses. Development of effective strategies to induce transgene-specific Tregs may represent alternative approach to overcome CTL mediated elimination of AAV transduced target cells in clinical trials.

In summary, fusion of VIPRs to therapeutic transgenes inhibits the induction of antigen-specific CTL response from AAV gene therapy thereby blocking the specific CTL mediated elimination of AAV transduced cells. The approach proposed herein for evading CTL-mediated killing of transduced cells extends beyond the treatment of AAT or DMD and are applicable to other transgenes and vector applications. Thus, this study has the potential to impact all gene therapy communities relying on strategies that require foreign epitope expression in hopes of avoiding the clinical transgene-associated immune complications observed in several trials to date.

Materials and Methods

Cells and virus

Human embryonic kidney 293 cells, murine melanoma B16 cells, rat glioma C6 cells, monkey kidney fibroblast-like Cos-1 cells and canine kidney MDCK cells were purchased from ATCC and maintained in DMEM medium with 10% FBS and 1% penicillin–streptomycin.

AAV vector production was previously described using the triple plasmid transfection method (86). The virus titer was determined by Q-PCR.

Mice

C57BL/6 mice and mdx mice as well as NOD/Scid mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA). The OT-1 mouse is transgenic for a T cell receptor (TCR) that recognizes the ovalbumin (OVA) SIINFEKL peptide bound by H-2K^b (Taconic Farms, Germantown, NY). All mice were maintained in a specific pathogen-free facility at the University of North Carolina at Chapel Hill. The University of North Carolina Institutional Animal Care and Use Committee approved all procedures.

Construction of recombinant plasmids

To make the H2k^b expression plasmid, the fragment from H2k^b plasmid (generously provided by Dr. Flavell, Yale University) cut with EcoRI was cloned into the EcoRI site of pTR/CBA-F9/OVA which was then cut with Nsi and SacI, blunted, and ligated to form pTR/CBA-H2k^b. To generate pTR/CMV-OVA, the fragment from pAC-neo-OVA (generously provided from Dr. M. Bevan, the University of Washington, Seattle) with EcoRI cut, blunted, and was cloned into the site of pTR/CMV-EGFP with AgeI and SalI digestion and blunt ligation. To make pTR/CMV-ICP, p7.5k131A/ICP47rc was digested with EcoRI (Blunt) and SalI and then inserted into the site of pTR/CMV-GFP cut with AgeI (blunt) and SalI. pCMV-US6 was provided by Dr. Hengel (50,87). To make the construct pTR/CB-AAT/OVA, PCR was performed on the OVA plasmid pAC-neo-OVA using primer pair OVA-F and OVA-R (Supplementary Material, Table S1). The PCR product was digested with BsaXI and cloned into pTR/CBA-AAT to form pTR/CBA-AAT/OVA-1. Next, pTR/CBA-AAT/OVA-1 was used as a template for PCR with AAT-F1 and OVA-R1 primers. After digestion of PCR product with EcoRI and NotI, the fragment was cloned into pTR/CBA-AAT to generate pTR/CBA-AAT/OVA. The fragments from PCR (AAT-F1/OVA-R2 and ICP-F/ICP-R) with the templates pTR/CBA-AAT/OVA and ICP47 plasmid were digested with EcoRI or NotI, respectively, and then cloned into pTR/CBA-AAT/OVA for generation of pTR/CBA-AAT/OVA-ICP. To make pTR/CMV-mini-dys-ICP, we used the following primer pairs dys-F1/dy-R1 and ICP-F2.1/ICP-R2 (Supplementary Material, Table S1) to amplify the fragments from pTR/CMV-mini-dys and ICP plasmid, respectively. The mini-dystrophin and ICP47 PCR products were digested with ApaI and XhoI, respectively, and then cloned into pTR/CMV-mini-dys by partial digestion with ApaI and XhoI to generate pTR/CMV-mini-dys. All constructs were verified by sequencing.

Peptides

Peptides from mini-dystrophin were synthesized at Microprotein Sequencing & Peptide Synthesis Facility in the University of North Carolina at Chapel Hill and were >80% pure. Peptides were dissolved in DMSO at a concentration of 20 mg/ml and stored at −20 °C.

Antigen presentation assay in vitro

VIPR function analysis in vitro was performed by co- transfection of VIPR expression plasmid and OVA expression plasmid (either VIPR and OVA in separated plasmids or they were linked in one cassette), along with an H-2k^b plasmid (Mouse MHC-class I molecules) into 293 cells. Two days later, spleen cells from OT-1 mice were incubated with 293 cells overnight. OT-1 spleen cell activation was detected by immunostaining of CD8 and CD69 expression on the cell surface or IFN-γ expression intracellularly. A total of 10⁶ activated CD8+ T cells were incubated with fluorescein isothiocyanate (FITC)-labeled rabbit anti-mouse CD69 (clone H1.2F3; BD Biosciences, San Jose, CA) and a Phycoerythrin (PE)-labeled anti-CD8 (clone 53–6.7; BD Biosciences) for 1h at RT. Cells were then washed and analyzed by flow cytometry (UNC Chapel Hill Flow Cytometry Core Facility).

Intracellular IFN-γ staining

For intracellular IFN-γ staining in OT-1 spleen cells, before spleen cells were collected, 5 μM/ml (10 μg/ml) Brefeldin A (BD pharmingen) was added for 5 h to enhance intracellular accumulation of IFN-γ. Cells were stained with a PE anti-CD8 antibody and fixed with Cytofix/Cytoperm (BD Pharmingen). Then, the cells were incubated with a FITC-conjugated anti-IFN-γ antibody. After washing, the cells were analyzed by flow cytometry. For intracellular IFN-γ staining in spleen cells from mdx mice immunized with mini-dystrophin, spleen cells were incubated with different pools of peptides derived from mini-dystrophin at the concentration of 1mg/ml. Three days later, brefeldin A was added and INF-γ staining was performed as described above.

For intracellular IFN-γ staining in mdx mice, after AAV/min-dys vector injection, spleen cells collected from mdx mice were incubated with 10 μg/ml five pools of mini-dystrophin synthetic peptides for 48h. The five pools are peptide poo1 1 (DP1, 1–265 amino acid), peptide poo1 2 (DP2, 266–530 amino acid), peptide poo1 3 (DP3, 531–795 amino acid), peptide poo1 4 (DP4, 796–1060 amino acid) and peptide poo1 5 (DP5, 1061–1325 amino acid). Five hours before spleen cells were collected, 5 μM/ml (10 μg/ml) Brefeldin A (BD pharmingen) was added. Cells were stained with a PE anti-CD8 antibody and fixed with Cytofix/Cytoperm (BD Pharmingen). Then, the cells were incubated with a FITC-conjugated anti-IFN-γ antibody. After washing, the cells were analyzed by flow cytometry.

In vivo tumor graft experiment

293 cells were transfected with the plasmid pTR/CBA-AATOVA or pTR/EGFP and H2kb expression plasmid with/without the plasmid pCMV-ICP47. One day later, 1×10⁷ 293/kb cells in 0.2 ml Matrigel were injected into female SCID/NOD mice (4–6 weeks old) subcutaneously, 1×10⁷ activated spleen cells from OT-1 mice were injected via tail vein simultaneously. The tumor size was measured at the indicated time points.

Preparation of activated OT-1 cells

As described by Qiao et al. (88). Briefly, naive OT-I cells were isolated from spleen and lymph nodes. After lysis of red blood cells, single cell suspension was cultured in the presence of 1 μg/ml SIINFEKL peptide. Cells were harvested at day 3 after activation and used for in vivo injection.

AAT expression

After systemic administration of AAV vector, blood was collected at indicated time points and AAT level was detected by ELISA.

In vivo CTL killing assay

As described before (89,90), a total of 10⁷ spleen cells from C57BL mice were incubated with either a high (5 μM) or low dose (0.5 μM) of carboxyfluorescein succinimidyl ester (CFSE) in PBS at RT for 12 min. Then FBS was added to stop CFSE labeling. After washing, CFSE^high cells were incubated with 10 μg/ml OVA SIINFEKL. After peptide pulsing, two populations (CFSE^high and CFSE^low) of target cells were washed and mixed together, and then 10⁷ cells of each population were injected into 5×10¹⁰ particles of AAV/AATOVA or AAV/AAT infected mice via tail vein. Spleen cells were collected 24 h later and analyzed by flow cytometry. The percent-specific lysis was determined by following formulas: the ratio of recovery of non-peptide-treated control spleen cells to peptide-sensitized spleen cells = (percentage of CFSE^low cells)/(percentage of CFSE^high cells). The percent-specific lysis (%) = 100 × [1 - (ratio of cells recovered from naive mice/ratio of cells recovered from infected mice)].

Western blot

293 cells were transduced with 5000 particles of AAV2/CMV-dys or AAV2/CMV-dys-ICP47 vectors per cell in the presents of adenovirus at a MOI of 5. Cells were harvest for western blot at day 3 post transduction. Protein was extracted by RIPA lysis buffer (Thermo Fisher Scientific). The protein concentration was determined by BCA Assay (Thermo Fisher Scientific). A total of 60 μg protein was loaded to 4–12% SDS-PAGE gel. After electrophoresis, proteins were transferred to NC membrane. The ∼160 KDa minidystrophin band was detected by NCL-DYS3 antibody (Leica Biosystems). The Band intensity was measured by Amersham™ Imager 600 (GE Healthcare Life Sciences).

In vitro CTL function assay

The CTL cytotoxicity assay was performed by detection of lactate dehydrogenase (LDH) release following the manufacturer’s instruction (Thermo Scientific). Briefly, mdx mice were intramuscularly administered with 5×10¹⁰ particles of AAV rh32/33/CMV-dys. Four weeks later, spleen cells were collected as the effector cells. 293/H2k^b cells were transfected with pTR/CMV-dys or pTR/CMV-dysOVA, and harvested 24 h later as the target cells. 1×10⁶ spleen cells were incubated with 1×10⁴ 293 cells with dystrophin expression for 4 h and supernatant was collected for LDH detection. The number of LU from only target cells without the addition of spleen cells was used as a maximum, while the number of LU from only spleen cells served as background. The cytotoxicity was calculated as follows: % Cytotoxicity = (Experimental value—Effector Cells Spontaneous Control—Target Cells Spontaneous Control)/(Target Cell Maximum Control – Target Cells Spontaneous Control) × 100.

Statistical analysis

Data were presented as the mean ± standard error. All statistical calculations were performed using the statistical software (GraphPad Prism 5.0 software). Differences between different groups, which were evaluated by the Student’s t test, were considered to be statistically significance when P values were <0.05.

Supplementary Material

Supplementary Material is available at HMG online.