Immunomodulation in Administration of rAAV: Preclinical and Clinical Adjuvant Pharmacotherapies

Recombinant adeno-associated virus (rAAV) has attracted a significant research focus for delivering genetic therapies to target cells. This non-enveloped virus has been trialed in many clinical-stage therapeutic strategies but important obstacle in clinical translation is the activation of both innate and adaptive immune response to the protein capsid, vector genome and transgene product. In addition, the normal population has pre-existing neutralizing antibodies against wild-type AAV, and cross-reactivity is observed between different rAAV serotypes. While extent of response can be influenced by dosing, administration route and target organ(s), these pose concerns over reduction or complete loss of efficacy, options for re-administration, and other unwanted immunological sequalae such as local tissue damage. To reduce said immunological risks, patients are excluded if they harbor anti-AAV antibodies or have received gene therapy previously. Studies have incorporated immunomodulating or suppressive regimens to block cellular and humoral immune responses such as systemic corticosteroids pre- and post-administration of Luxturna® and Zolgensma®, the two rAAV products with licensed regulatory approval in Europe and the United States. In this review, we will introduce the current pharmacological strategies to immunosuppress or immunomodulate the host immune response to rAAV gene therapy.


INTRODUCTION
Adeno-associated virus (AAV) is a 26nm, non-enveloped virus of Parvoviridae family. It is 4.7kb single-stranded DNA genome containing 4 open reading frames (ORFs) (rep, cap, aap, and MAAP) flanked by inverted terminal repeats (ITRs) (1,2). In therapeutic gene delivery, the viral ORFs are replaced by the desired transgene expression cassette and referred as recombinant AAV (rAAV). It has emerged as a leading vector to deliver genetic therapies due to its ability to transduce diverse cell types and safety profile.
A significant obstacle in clinical delivery of rAAV is host immune response triggered by rAAV capsid, genome, and therapeutic protein produced (3). Although AAV infection is non-pathogenic in humans, initial exposure induces humoral and cellular anti-capsid response that are reactive to rAAV due to capsid similarity (4,5). Pre-existing neutralizing antibody (NAb) can effectively block rAAV transduction even at low levels (1:5) (6). Most rAAV clinical trials exclude seropositive patients; given the high seroprevalence (60% for AAV2), limiting patients suitable for rAAV therapy (7,8). Furthermore ex vivo studies have shown predominantly pre-existing memory phenotype cytotoxic T lymphocytes (CTL), following exposure to rAAV can undergo expansion and potentially lead to elimination of transduced cells (9,10).
After rAAV administration, capsid-derived epitopes can be presented by professional antigen presenting cells (APC) via major histocompatibility complex (MHC) class I pathway and activate CTL (11). The activation of CTL can result in targeted destruction of transduced cells, as observed in rAAV2 hemophilia B clinical trial (12). Despite initial stable therapeutic factor IX (FIX) expression (>10% activity) for 4 weeks, FIX levels gradually declined to baseline (<1%). This was associated with asymptomatic, self-limiting transaminitis, and corresponding changes in capsid-specific CTL population (5). In the subsequent study using AAV8, administration of steroids was able to negate this response and maintain therapeutic FIX levels albeit a 50-70% decline from peak levels (13). Moreover, transgene protein product-specific CTL was observed in human rAAV trials for Duchenne's Muscular Dystrophy (14) and a-1-antitrypsin (15). Regulatory T cells (Treg) modulate immune tolerance towards transgene product and capsid that are vital to durable expression of therapeutic protein (16,17). Although the full clinical significance of innate response to rAAV is unclear (18), unmethylated CpG motifs in rAAV vector genome interact with toll-like receptor (TLR) 9 present in plasmacytoid dendritic cells and Kupffer cells, releasing type I interferons activating cellular and humoral responses in mouse models (19,20), and has been suggested as the cause of loss of expression in a rAAV8 hemophilia B trial (21). Furthermore, rAAV capsid-targeting TLR2, various DNA sensors, and complement activation may also play a role (22). Different pharmacotherapies have been used to modulate immune responses in current in vivo rAAV studies. Here, with a particular focus on licensed agents, we discuss the pharmacology of each drug (Figure 1), and their applications in enabling safe and long-term expression of rAAV gene therapies ( Table 1).

Global Effects
Corticosteroids Corticosteroids (CCS; methylprednisolone, prednisolone and prodrug prednisone) bind to glucocorticoid receptors modifying diverse downstream transcriptional signaling. This includes annex I, MAPK phosphatase 1, and NF-kB resulting in anti-inflammatory and immunosuppressive properties (63). They have broad inhibitory effects on both innate and adaptive immune cells by reducing pro-inflammatory cytokine and chemokines, T-and to a lesser extent, B-cells production (64). CCS are used short-term in conjunction with systemically delivered gene therapies to negate transaminitis and associated CTL-induced injury transgene loss (30, 65), and reduce T-cell infiltrates in muscular fibers in non-human primates (NHP) (33). They are also adopted in approved gene therapies for inherited retinal dystrophy (25) and spinal muscle atrophy (SMA) (26).
Subsequently increasing doses of systemic rAAV have been delivered in preclinical and clinical studies with significant hepatic sequelae. High dose intravenous AAV9 (2×10 14 vector genomes (vg)/kg) in NHP resulted in marked transaminitis and acute liver failure (66), posing concerns over dosage related hepatotoxicity (67). Furthermore, clinical phase II trial for X-linked myotubular myopathy delivered intravenous rAAV8.AT132 (NCT03199469) 3×10 14 vg/kg in high dosage group, with 16-weeks of prednisolone commencing 1 day prior to dosing. Three patients with pre-existing intrahepatic cholestasis (68) experienced severe hepatobiliary complications culminating in death. The exact mechanisms of the hepatotoxicity remain to be elucidated. These studies however build evidence that short-course CCS alone is likely to be insufficient to inhibit formation of capsid-reactive T cells (13) and rAAV-mediated immune response with systemic high dosages. Therefore, the addition of other immunosuppressive agents maybe beneficial. In a AAVrh10-microRNA study delivering 4.2×10 14 vg intrathecally into two adult patients, the first developed meningoradiculitis after intrathecal infusion despite corticosteroids (IV methylprednisolone on day 0 and oral prednisone tapered over 4 weeks). In the second patient, the addition of rituximab and rapamycin to the regimen resulted in a lower increase of NAb and T-cell response (29) and these drugs are further discussed.

Rapamycin (Sirolimus)
Rapamycin is a macrolide immunosuppressant that binds to the same intracellular target (immunophilin) as tacrolimus; however, rapamycin/FKPB12 complex inhibits a crucial cell-cycle kinase known as mammalian target of rapamycin (mTOR). Beneficial downstream effects include Treg generation, suppressing CTL and T helper (T H ) activation and at higher doses, B-cell proliferation and differentiation (69)(70)(71).
Rapamycin has beneficial effects on circumventing existing antibodies and studied in current hemophilia gene therapy trials. Hemophilia patients develop inhibitors (antibodies) to clotting factor replacement and another cause for exclusion in gene therapy trials. In a murine hemophilia A model, rapamycin (4mg/kg three times a week) was given in addition to B-cell depleting anti-CD20 antibodies to suppress T H and Treg response suppressing inhibitor development (37). Intraperitoneal prednisolone with rapamycin was shown to inhibit B-cell activation in murine spleen and bone marrow, reducing preexisting anti-capsid immunoglobulin G (IgG) by up to 93% after 8 weeks (72). Additionally, co-administrating AAV vectors with rapamycin encapsulated in synthetic vaccine particles (SVP [Rapa]) enabled re-dosing of AAV8 at 4 × 10 12 vg/kg in mice and NHP (39). SVP [Rapa] provided sufficient reduction of B and T cell activation in an antigen-selective manner, inhibited CTL liver infiltration, and efficiently blocked memory T cell response. Potential of intramuscular rAAV9 re-administration is currently investigated for Pompe disease (NCT02240407) (73), by attenuating T and B cell response with rapamycin and rituximab respectively. Preliminary results were successful in preventing formation of anti-capsid and anti-transgene antibodies (38), with aims to enable rAAV re-administration and maintain effectiveness in different underlying mutations.

Mycophenolate Mofetil
Inosine monophosphate dehydrogenase (IMPDH) is the ratelimiting enzyme for guanosine nucleotide synthesis, and type II IMPDH is upregulated in activated lymphocytes. Mycophenolate mofetil (MMF), prodrug of mycophenolic acid, preferentially inhibits type II IMPDH, suppressing T and B cells proliferation (74). In mice MMF reduced rAAV transduction efficiency by depleting guanosine triphosphate required for vector genome second strand synthesis (75), but this was not observed in higher animals. No difference in AAV8-hFIX transgene expression was observed when administered with tacrolimus in NHP (6), highlighting the difficulties of recapitulating human immune system in mouse models.

Calcineurin Inhibitors
Ciclosporin and tacrolimus are immunosuppressants that inhibit calcineurin, a key signaling phosphatase, by binding to their respective immunophilins -cyclophilin and FKBP12 (76). A major downstream effect is suppression of interleukin (IL)-2 transcription, thereby inhibiting T cells differentiation, survival, and subsequent antibody production and CTL activities via effector T H cells. Daily systemic administration of tacrolimus (0.06mg/kg/day) has been shown to prolong rAAV8 and rAAV9 expression in NHP skeletal muscle, up to 42 weeks from 8 and 16 weeks respectively (47). No generalized toxicity was reported but T-cell and macrophages infiltrations were observed. FIGURE 1 | Mechanisms of action of approved pharmacotherapies for immunomodulation with rAAV gene therapy. Pre-existing NAb can inhibit receptor-mediated endocytosis thus transduction of rAAV (A). TLR9 recognizes CpG motifs, and TLR2 on cell surface or endosomal membrane recognizes vector capsid, both of which lead to release of pro-inflammatory cytokines (B). Recent evidence shows that ITRs facilitate bidirectional transcription to form dsRNA, which triggers cytosolic MDA5 and downstream type I interferon response (C). Upon endosomal escape, rAAV can be degraded by proteasome and loaded on MHC class I by the endoplasmic reticulum (D). Recognition by memory CTL (E) leads to expansion and differentiation into CTL, and both can commence effector functions leading to loss of transgene expression (F). On the other hand, rAAV can also transduce APC, for instance dendritic cells, and transgene protein product can be phagocytosed (G). They are processed by proteasomes and endosomes respectively and the antigens can be presented on MHC class II molecules (H), leading to downstream activation of T H and B-cells; among other actions, B cells would differentiate into plasma cells and produce antigen-specific antibodies (I). Created with BioRender. com. APC, antigen presenting cells; ATO, arsenic trioxide; CCS, corticosteroids; Chemo, chemotherapeutics; CIs, calcineurin inhibitors; CTL, cytotoxic T lymphocytes; dsRNA, double-stranded ribonucleic acid; HCQ, hydroxychloroquine; IFN, interferon; IL, interleukin; ITR, inverted terminal repeats; MHC, major histocompatibility complex; MMF, mycophenolate mofetil; NAbs, neutralizing antibodies; NF-kB, nuclear factor kappa B; PIs, proteasome inhibitors; RAPA, rapamycin; rATG, rabbit anti-thymocyte globulin; RTX, rituximab; TH, T helper cells; TNF, tumor necrosis factor; TLR, toll-like receptor. The first approved gene therapy in Europe, alipogene tiparvovec (Glybera), incorporated 12-week immunosuppression regimen with ciclosporin (3mg/kg/day) and MMF (2g/day) (32). In the initial regimen, 9/14 subjects showed humoral and cellular response against rAAV1 (77). Subsequent study (AMT-011-02) modified the regimen to commence ciclosporin and MMF from day -3 with additional methylprednisolone on day 0 resulting in transient cellular responses without clinical sequalae (78).
Ciclosporin and tacrolimus were found to inhibit Treg proliferation and activity in vitro (79), and similar effects were observed in tacrolimus-treated allograft patients ex vivo (80); this could be detrimental in inhibiting the development of peripheral tolerance following rAAV administration. However, preclinical delivery of ciclosporin and non-depleting CD4 receptor antibody (NDCD4) have been shown to induce antigen-specific Treg, enabling AAV intravenous re-administration after 3 months (48).

B-Cell Specific
Rituximab Rituximab (RTX) is a chimeric mouse/human monoclonal antibody targeting CD20 present in pre-B and mature B cells except plasma cells. It depletes B cells by inducing apoptosis, antibody dependent cell-mediated cytotoxicity and complement dependent cytotoxicity, thereby limiting antibody production and epitope presentation via MHC class II to T H cells (81).
A preclinical model for hemophilia B showed RTX with ciclosporin dampened NAb response to human FIX and capsid without affecting Treg (51). As ciclosporin inhibits T H cell, this further improves B-cell inhibition profile. Variable responses have been observed in RTX's effect on reducing pre-existing AAV NAb. A small group of patients with rheumatoid arthritis were treated with combination of methotrexate and RTX, lowering anti-AAV2 and anti-AAV5 NAb in a subset of patients with variable magnitudes (50). For AAV2, 9/28 patients showed at least a half-log reduction, and inferred individuals with NAb titer ≤1:1000 were more likely to respond to RTX but the contribution of methotrexate is unknown. Considering the supportive evidence from previous AAVrh10-microRNA with RTX (29), further study in RTX application is warranted.

IgG-Degrading Cysteine Proteinase
Imlifidase (Idefirix, Hansa Biopharma) is a IgG-degrading cysteine protease derived from Streptococcus pyogenes (IdeS), which specifically cleaves opsonizing IgG at the lower hinge region of the heavy chains, resulting in a F(ab') 2 and a non-functioning dimeric Fc fragment (82). It could potentially overcome a limitation of RTX and cleave existing capsid-specific IgG. Using a laboratory version of IdeS with rAAV8, significant reductions in anti-AAV8 IgG and NAb levels, with enhanced liver transduction and transgene expression and observed in passively immunized murine models and naturally immunized NHP (53). Notably, the study also explored rAAV re-administration with IdeS pre-treatment in NHPs. In the first study (n=1), no induction of anti-capsid IgG and NAb, along with lower IgM and increased transgene level was observed for 21 days after second rAAV8-hFIX administration. However, this was not replicated in a larger cohort (n=5) immunized with rAAV-LK03, that developed anti-capsid IgM and IgG. Further studies are required as the IdeS dosing regimen differed between studies, and two rAAV-LK03 vectors (expressing GAA and hFVIII) were used in the latter study.

Proteasome Inhibitors
Proteasome inhibitors (PIs) are licensed for multiple myeloma. Second-generation carfilzomib is irreversible and more specifically inhibits chymotrypsin-like activity than bortezomib, the reversible first-generation inhibitor, which also inhibits lysosomal and calcium-activated cellular proteases (54,83). After endosomal escape, rAAV particles either enter the nucleus for transgene expression, or become ubiquitylated then degraded by proteasome (84). The latter pathway results in unsuccessful transduction, and capsid-derived peptides are presented to CTL by MHC class I molecules, provoking elimination of transduced cells and loss of transgene expression (85). In addition, these inhibitors may have immunomodulatory role in suppressing dendritic cells function and downstream T-cell stimulation (86).
PIs have been investigated in preclinical models for their ability to increase rAAV availability and reduce CTL responses. Bortezomib has been shown to dose-dependently decrease cell surface MHC class I antigen presentation and inhibit CTLmediated lysis after rAAV administration in vitro (87). Moreover, a single bortezomib dose given with rAAV8 dosing enhanced transgene expression by >50% for one year (compared to~10%) in hemophilia A mice, and longer in-range clotting time for at least 10 months in hemophilia A dogs (57). Both bortezomib and carfilzomib enhance rAAV2 transduction in vitro, but bortezomib is more efficacious than carfilzomib in vivo when administered by retro-orbital injection with rAAV2 (56). Although no toxicity was found in the animal models, peripheral neuropathy and myelosuppression are adverse effects observed in humans (54). Emerging evidence showing variations in PI effectiveness across cell types and AAV serotypes (88), which warrants further study.

Chemotherapeutics
Second strand synthesis after capsid uncoating in nucleus is long-recognized as the rate-limiting step of rAAV transduction (89); an improvement in such efficacy could allow rAAV administration at lower dose. As traditional chemotherapeutics directly or indirectly induce DNA damage, thereby initiating DNA damage response (DDR) to repair lesions (90), it has been postulated that these repair mechanisms could increase conversion of rAAV genome into dsDNA (91), or divert DDR proteins that would otherwise impede dsDNA production (92). Several chemotherapy agents were evaluated previously (91,93) and a high throughput screening study identified teniposide, a type II topoisomerase inhibitor pharmacologically similar to etoposide, as a potent transduction enhancer (94). Tail vein injection of rAAV2-Luc with teniposide (at doses of 1×10 11 vg and 20mg/kg respectively) resulted in bioluminescence 2-log higher 48 hours post-administration without hepatotoxicity. This difference reduced to~1 log at 8 days post-administration (study endpoint). Further study is required to determine whether the effect is sustained, and evaluate potential long-term effects of non-tissue-selective chemotherapy.

Anti-Malarials
Hydroxychloroquine is an anti-malarial that interferes with TLRs and cyclic GMP-AMP synthase (cGAS), dampening downstream pro-inflammatory cytokine and type I IFN production (60). A study injected hydroxychloroquine subretinally (18.75mM) with rAAV2, resulting in 5.9-fold improvement in photoreceptor transgene expression (61). However, endosomal acidification is essential for rAAV escape (84), and hydroxychloroquine increases endosomal and lysosomal pH (60), this effect may not be replicated or consistent with systemic application.

Triple T-Cell Directed Therapy
This study highlights importance of pharmacotherapy choice. rAAV2-hFIX (8×10 12 vg/kg) was delivered intrahepatically to NHP alongside 2-drug regimen of MMF and rapamycin compared to 3-drug adding Daclizumab (40). The addition of daclizumab resulted in decreased CD4 + CD25 + FoxP3 + Treg and consistent formation of inhibitory antibodies to hFIX; this was not observed in the 2-drug group. Daclizumab is a humanized monoclonal antibody targeting CD25 present on interleukin-2 receptor commonly found in activated T cells and CD4 + CD25 + FoxP3 + (96). This indicates careful selection of immunosuppressive agents is necessary as Treg play a critical role in regulating immune response to rAAV products, particularly observed in liver and muscle gene transfer (97).

Triple T-Cell Directed Therapy: Delayed rATG
Timing of T cell immunosuppressant regimen was evaluated with liver-directed rAAV2-hFIX, at 7.5×10 12 vg/kg via hepatic artery in NHP (41). Rabbit anti-thymocyte globulin (rATG), a rabbit polyclonal IgG, causes T-cell and plasma cell depletion and modulation of other immune effectors (98). Used with MMF (25 mg/kg) and rapamycin (4mg/kg, then 2mg/kg), a 35-day delay in rATG administration prevented formation of antitransgene humoral response compared to commencing immunosuppression on day 0 (41). Neither group had cellular response to capsid or transgene, and 2 of 3 NHP in the delayed rATG group did not develop anti-capsid antibodies. It is possible by postponing rATG lowers the Th17/Treg ratio, allowing peripheral tolerance to the transgene product (41).

B and T Cell-Directed Therapy
This intensive immunosuppressive therapy included T-cell-targeting ATG and tacrolimus, B-cell targeting rituximab, with MMF and methylprednisolone to deliver global immunosuppression (99). This 5-drug regimen with rAAV5-PBGD 1×10 13 vg/kg infusion resulted in reduced T-cell response in NHP, but did not prevent NAb emergence following regimen removal. This suggests that drug selection, initiation and duration of suppression, and role of global immunosuppression are important considerations.

DISCUSSION
AAV gene therapy has the potential to be durable and transformative treatment for previously incurable, life-limiting genetic diseases. However, human immune responses to the viral vector, transgene, and protein product determine the therapeutic efficacy and possibility of re-administration. Studies showed cross-reactive anti-capsid NAb present at 15 years (100), CTL and Treg infiltrates at injection site after 5 years (101); and in NHP adverse effects related to high-dosage (42,66). With the increasing applications of systemic rAAV at higher dosages in clinical trials, further understanding of innate and adaptive immune responses to rAAV gene therapies is essential to safe and efficacious treatment.
Multiple approaches are being developed to evade the host immune response such as evaluating effects of empty capsids (102), capsid engineering guided by antigenic footprints (103), and plasmapheresis (104). The use of existing licensed medications for their immunosuppression and immunomodulation properties offers the advantages of flexibility (by allowing variations of drug combinations, dose, and duration of immunosuppressive course), accessibility, and well-documented pharmacological and safety profiles. As summarized above, a range of pharmacological agents have been used in clinical and preclinical studies, and the timing of immunomodulation, duration, and drug regimen itself have all contributed to treatment efficacy. Corticosteroids are the most commonly used agents to resolve transaminitis, however, its relationship with resolution by corticosteroids and T-cell response are not always clear as observed in a hemophilia A trial (28, 105). Also, rAAV vectors and patients' characteristics must be thoroughly evaluated to optimize safe delivery of highdose systemic rAAV or re-dosing.
To better design immunomodulation regimens, thorough considerations of the underlying immunological mechanisms are essential. Peripheral tolerance mediated by Treg to counteract CTL responses in hepatic AAV studies remains an important area of development (106). Reports on Treg in liver and their persistence in muscle fibers after intermuscular delivery (17) further emphasizes the need for Treg-sparing therapies. Moreover, binding (non-neutralizing) antibodies in mice seemed to have a different biodistribution profile than NAb and higher efficacy in liver transduction (107). A proposed late-phase innate response triggered by ITRs' inherent promoter activity that generates dsRNA that activates cytosolic MDA5 sensors and releases type I interferons as demonstrated in mice xenografted with human hepatocytes (108), poses further questions as to the ideal immunosuppression regimen. Lastly, the lack of fully predictive animal models (3,109), and possibility of alternative, non-immune-mediated toxicity such as dorsal root ganglion toxicity with AAV9 (110), continue to represent challenges in safety and efficacy evaluation.
CRISPR-Cas9 is a promising therapeutic tool that allows genetic target-specific cleavage and editing (111). The first clinical trial is currently underway for Leber's congenital amaurosis 10 (NCT03872479), EDIT-101, consists of Staphylococcus aureus Cas9 (SaCas9) and two guide RNA packaged in AAV5 vector for subretinal redelivery. One concern is that the prevalence of anti-SaCas9 antibodies and T-cell in humans are reported to be 78% (111). Studies showed pre-existing SaCas9 immunity in mice resulted in increased CTL response leading to hepatocyte apoptosis and loss of transgene (112). Although no adaptive immune response towards SaCas9 was reported (113), the eye is a relatively immunoprivileged site, these data will not necessarily predict immune response in humans or systemic administration. By gaining a precise understanding of the immune mechanisms, drug repurposing (for instance JAK inhibitors for type I interferon signaling, anti-interleukin-6 human monoclonal antibodies), alongside with how and when to immunomodulate around rAAV dosing and required duration, will help to fully maximize gene therapy safety and efficacy.

AUTHOR CONTRIBUTIONS
WC researched on and prepared the draft. JN reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.