- Top of page
- Materials and methods
Adeno-associated viruses (AAV) are parvoviruses that have not been associated with any pathogenicity in humans 1. Twelve serotypes have now been identified 2–4 and these show tropism for many different cell types 2. Studies using AAV serotype 2 (AAV2) have demonstrated that the vector has considerable promise for gene transfer to the liver, muscle and eye, and it has been used to treat animal models of several inherited conditions, such as muscular dystrophy 5, haemophilia 6 and several types of inherited retinal degeneration 7. These pre-clinical studies have demonstrated that effective treatment of these disorders might be possible 8–10. On the whole, however, these trials have not demonstrated the level of efficacy observed in pre-clinical animal studies, and it appears that this might be due to host immune responses to either the gene transfer vector or the transgene product 11. Because the vast majority of the population are exposed to wild-type AAV in the first decade of life, patients may have pre-existing immunity to AAV. It is estimated that up to 80% of the population have circulating antibodies against AAV2 12, whereas 30–50% have antibodies with neutralizing activity 13, 14. This pre-existing immunity may inhibit transgene expression, and any boost to an anti-vector immune response may render future re-administration of the vector ineffective. Another potential barrier to efficacious long-term therapy is the development of immune responses against the transgene product.
Several detailed studies on the immune responses occurring following AAV-mediated gene delivery in animal models have been carried out, but have generated some conflicting reports and remain inconclusive, with immune responses appearing to depend on the route of administration, vector dose and species differences 15. Immune responses against AAV vectors leading to a reduction of transgene expression have been observed in dogs following intramuscular administration 16. Here, a T-cell response against AAV2 and AAV6 vector capsid proteins was observed, regardless of the transgene expressed, the muscle type injected or the cellular specificity of the promoter. By contrast, another canine muscular dystrophy model showed that T-cell infiltration was dependent on the transgene product; intramuscular administration of AAV2 vectors expressing β-galactosidase resulted in intense T-cell infiltration, whereas AAV2 expressing no transgene resulted in no cellular infiltration 17, despite similar doses being used in each study. Studies in mice, however, have demonstrated that hepatic gene transfer can achieve transgene-specific tolerance through the induction of regulatory T cells, even at high doses of vector 18, whereas lower doses can result in long-term transgene expression following intramuscular administration with no destruction of transduced cells 19.
The high level of pre-exposure to wild-type AAV in the human population has led to problems in clinical trials of AAV gene transfer 11. However, when AAV has been administered to an immune privileged site such as the central nervous system (CNS), a much more limited response against the vector has been observed systemically, with only a minority of patients developing neutralizing antibodies (NAbs) and no anti-vector antibodies being detected in the CNS itself 10. The eye is also well known to be an immune privileged site and therefore intraocular administration of AAV may result in a different pattern of immune responses compared to other routes. The mechanism of this immune privilege is multifaceted and is the result of several mechanisms, including physical barriers such as the blood–retinal barrier, immunological ignorance and peripheral tolerance of eye-derived antigens 20. An intraocular immunosuppressive microenvironment utilizes mechanisms such as FasL expression on the corneal endothelium 21 and the retinal pigment epithelium (RPE) 22 and immune deviation in both the anterior chamber 23 and the vitreous cavity 24. Immune privilege might afford the eye a level of protection, to a certain degree, against damaging inflammatory responses, therefore protecting and preserving the cells crucial to vision that are unable to repair damage or regenerate in adult differentiated tissue. Long-term AAV-mediated gene expression in the mouse eye has previously been demonstrated in several studies 25, 26, suggesting that the vector is well tolerated.
There have been extensive investigations into immune responses against AAV following administration to the liver or muscle. By contrast to the prevalence of NAbs in humans, the frequency of existing memory T-cells is very low in the normal healthy population 14. However, T-cell responses against AAV2 have been observed transiently in a clinical trial following the intrahepatic delivery of AAV2, where an expanded capsid-specific CD8 + effector population was identified. This led to the destruction of transduced hepatocytes in one patient and a substantial increase in neutralizing antibody levels was observed in all patients 9. However, to date, there does not appear to have been any detailed analysis of cellular and humoural immune responses following intraocular administration of AAV, and what effect these responses may have on repeat administration of vector; this information is critical because ocular gene therapy clinical trials are now underway to treat Leber's congenital amaurosis, a form of severe early onset retinal degeneration caused by mutations in RPE6527, 28. RPE65 encodes a 65-kDa protein that is a vital component in the visual cycle. Clinical trials have demonstrated that a single subretinal injection of AAV2 encoding human RPE65 can mediate a significant improvement in visual function, even in patients with advanced loss of vision. One study used a hybrid constitutive cytomegalovirus (CMV)/chicken β-actin promoter to drive the expression of hRPE65 and administered a dose of 1.5 × 1010 vector genomes 28. In this trial, patients were transiently immunosuppressed with corticosteroids and no cellular responses against AAV2 or RPE65, or anti-RPE65 humoural responses were detected, although one patient did show an increase in NAbs against AAV2. The other trial administered a dose of 1 × 1011 vector genomes of an AAV2 vector containing the endogenous hRPE65 promoter to drive hRPE65 cDNA expression 27, thereby restricting expression of the transgene to RPE cells only. Patients were also transiently immunosuppressed with corticosteroids in this trial. No cellular responses against the AAV2 capsid were detected, and no changes in humoural responses against RPE65 or AAV2 compared to pre-gene therapy levels were detected in any patients. Therefore, in both studies, the vector appears to be well tolerated in patients with missense mutations who are transiently immunosuppressed, with evidence of minimal inflammation or immune responses against the vector or transgene product. However, it has not yet been determined what effect previous subretinal exposure to the AAV2 vector has on the level of transgene expression following repeated subretinal injection or the effect that vector re-administration has on immune responses against the transgene product or the vector itself, particularly in the absence of immunosuppression. In the present study, we undertook a thorough evaluation of the cellular and humoural response in non-immunosuppressed mice following administration of AAV2 encoding an endogenous retinal protein (hRPE65) or a reporter gene encoding an exogenous reporter (green fluorescent protein, GFP), and we examined in detail the effects of these immune responses on transgene expression or the repeated administration of vector.
- Top of page
- Materials and methods
In the present study, we investigated the immune response to AAV2 vectors following subretinal delivery in immunocompetent mice. The main aim of the study was to evaluate the humoural and cellular responses to the vector or transgene after vector administration and to determine whether this impacted on repeated vector delivery. To our knowledge, this is the first study to investigate both cellular and humoural responses following AAV2 delivered to the subretinal space. We show functional rescue of the ERG in the rd12−/− mouse with AAV2.hRPE65.hRPE65 was observed following injection of vector into both eyes 3 weeks apart. This demonstrates that the principle that vector of the same serotype and expressing the same transgene can be re-administered effectively.
Over recent years, immune responses against gene delivery vectors have become one of the most important issues in the field of gene therapy. It has been established previously that, although adenoviral vectors induce strong immune responses that severely inhibit the efficacy of repeated vector administration, AAV vectors are much less immunogenic 38. However, more recent studies have shown that AAV is capable of stimulating an immune response that can be detrimental to gene delivery 9, 16, although the route of administration and vector dose appear to be key elements in determining the degree of anti-AAV immunity that is generated.
Our data show that subretinal administration of a lower dose (2 × 2 µl of 1 × 1011 vg/ml) AAV2 elicits minimal immune responses. NAbs were only detected in the serum of mice receiving the higher dose (2 × 2 µl of 5 × 1011 vg/ml) of AAV but not at the lower dose, suggesting that the development of NAbs are dependent on the dose of vector, which is in agreement with other studies 19, 39. Furthermore, no NAbs were detected in the ocular fluids of mice receiving the higher dose. Importantly, no cellular infiltrate was observed in any eyes following AAV injection, even in those mice with circulating NAbs. This is in agreement with previous studies investigating ocular gene delivery 40, but contrasts with other routes of administration such as intramuscular 16 and portal vein injection 41. Other studies, however, show that inflammation is not due the AAV vector, but rather is dependent on the transgene 17, 42.
By contrast to the low-dose re-administration, when a higher dose of vector was injected into the right and then the left eye 3 weeks apart, boosted NAb titres inhibited the transgene expression in a proportion of the eyes that received the second injection. The transgene expression from the eye that had received the first injection remained high in all animals. This suggests that the dose of vector delivered is a critical factor in the development of anti-vector immune responses, which has a substantial impact of the efficacy of subsequent administration of vector. The source of the variation in transgene expression remains to be established. This has important implications for the development of clinical gene delivery protocols.
In the future, it may be advantageous to engineer vectors that lack immunogenic motifs. The heparin sulphate proteoglycan (HSPG) motif on the VP3 capsid protein of AAV2 has been shown to be responsible for uptake into dendritic cells, leading to the activation of capsid specific CTLs. When this sequence was mutated to ablate HSPG binding, the immune response was attenuated, yet the tropism of the vector was unchanged. Furthermore, AAV serotypes that do not express HSPG binding motifs appeared inherently less immunogenic 43. Other studies have shown that a non-HSPG-binding mutant of AAV2 showed detargeting from the spleen and liver compared to wild-type AAV2, leading to higher levels of the non-HSPG-binding AAV2 mutant remaining circulating in the blood following intravenous injection 44.
Several studies have shown that NAbs against AAV2 do not exhibit cross reactivity against other serotypes, allowing the possibility of re-administration with other AAV serotypes; this ‘cross-administration’ approach has shown efficacy following intramuscular gene delivery 15, 19. Several different AAV serotypes are able to transduce the retina, so it may be possible to utilize this approach for effective re-administration in the eye. Greater immune responses have been observed in large animal models than in mice that have received the equivalent vector dose. Dogs appear to be particularly susceptible, but transient immunosuppression has been used to reduce immune responses, permitting long-term transgene expression 45, 46.
In conclusion, our data show that subretinal delivery of AAV2 in mice is well tolerated, even in the absence of immunosuppression. No cellular immune response was observed to the vector in any mice. Mice receiving the lower dose did not develop NAbs, whereas a proportion of the mice receiving the single high-dose of vector developed NAbs, and those receiving repeated administration of high-dose vector all developed the highest titres of NAbs, suggesting that the induction of NAbs is dependent on: (i) the initial dose of vector and (ii) NAb titres are boosted by repeat exposure to vector. Crucially, in the present study, we showed that lower doses vector could be re-administered and that repeat injections of vector were successful and demonstrated therapeutic efficacy. This is important because, if both eyes of a patient are to receive gene therapy, they are unlikely to be injected at the same time. Therefore, the demonstration that the same type of vector can be administered subretinally at a later time-point is vital to the development of a protocol requiring a second administration of vector. Although the data obtained in the present study suggest that immunosuppression may not be necessary for lower vector doses, it may allow increased doses of vector to be delivered and thus reduce the risk of immune responses still further.