SHORT COMMUNICATION AAVrh.10 immunogenicity in mice and humans. Relevance of antibody cross-reactivity in human gene therapy R Thwaite 1 , G Pagès 1 , M Chillón 1,2 and A Bosch 1 Simian adeno-associated virus (AAV) serotype rh.10 is a promising gene therapy tool, achieving safe, sustained transgene expression in the nervous system, lung, liver and heart in animal models. To date, preexisting immunity in humans has not been confirmed, though exposure is unexpected. We compared the humoral immune response with serotypes AAVrh.10 and AAV9 in mice, and AAVrh.10, AAV9 and AAV2 in 100 healthy humans. Mice, injected-intravenously, raised significantly more anti-AAV9 than anti-AAVrh.10 IgG (immunoglobulins), and sera demonstrated greater neutralizing capacity, correspondingly. Antibody cross- binding studies in mice showed negligible cross-recognition between AAVrh.10, AAV9 and AAV2. In humans, IgG prevalence against the most common human serotype, AAV2, was 72%; AAV9, 47% and AAVrh.10, a surprising, 59%. Yet, neutralizing-antibody seroprevalences were 71% for AAV2, 18% for AAV9 and 21% for AAVrh.10. Thus, most anti-AAV9 and anti-AAVrh.10 IgG were nonneutralizing. Indeed, sera generally neutralized AAV2 more strongly than AAVrh.10. Further, all samples neutralizing AAVrh.10 or AAV9 also neutralized AAV2, suggesting antibody cross-recognition. This contrasts with the results in mice, and highlights the complexity of tailoring gene therapy to minimize the immune response in humans, when multiple-mixed infections during a lifetime evoke a broad repertoire of preexisting antibodies capable of cross reacting with non-human serotypes. Gene Therapy (2015) 22, 196–201; doi:10.1038/gt.2014.103; published online 20 November 2014 INTRODUCTION Adeno-associated virus (AAV) vectors do not provoke a strong innate immune response as dendritic cells are poorly transduced. Consequently, markers of innate immunity tend not to be upregulated, though evidence is growing, concerning their relevance to the outcome of AAV-mediated gene transfer. 1 More overtly problematic is the challenge posed by preexisting anti-AAV antibodies present in patients’ serum before therapy. These arise with prior exposure, given AAVs are widespread in humans (reviewed by Calcedo and Wilson 2 ). In addition, generation of antibodies and T-cell responses against the transgene product, particularly a foreign protein, may depend on the AAV serotype and its capacity to infect antigen-presenting cells. 3–5 Thus, selecting the appropriate viral capsid is critical when planning a therapeutic approach using AAV vectors, since it determines tropism 6 and has major implications regarding the host immune response. We focus on AAVrh.10, a rhesus macaque serotype, 7 hypo- thesizing that humans are less likely to be exposed, hence preexisting antibodies should be minimal. Recombinant AAVrh.10 shows promise for gene therapy. It stably-transduces neurons and to a lesser extent oligodendrocytes, after intracranial or intrathecal administration, demonstrating capacity to revert different central- and peripheral-nervous system pathologies in mouse models such as late-infantile neuronal ceroid lipofuscinosis, 8 metachromatic leukodystrophy, 9 diabetic neuropathy, 10 and amyotrophic lateral sclerosis 11 among other diseases. Recently rAAVrh.10 was approved for two clinical trials to treat late-infantile neuronal lipofuscinosis (NCT01161576) and Sanfilippo Type A syndrome. 12 Furthermore, intravenous administration of AAVrh.10 efficiently transduces liver, heart and dorsal root ganglia and reverses cardiomyopathy in a mouse model of Friedreich’s ataxia. 13 Several studies report the prevalence of neutralizing antibodies (NAbs) against AAV serotypes 1, 2, 5, 6, 7, 8, 9 and the capsid hybrid rh32.33. 14,15 However, no data on preexisting antibodies to AAVrh.10 in the general human population, nor on the immunogenicity of AAVrh.10 among serotypes have been reported. Here we show the humoral immune response to AAVrh.10 is significantly weaker than AAV9 in mice. Yet, in humans, immunoglobulin G (IgG) prevalence against AAVrh.10 is greater than against AAV9. However, these immunoglobulins are a mixed population of NAbs and non-NAbs. Further characterization suggests cross-reactivity with abundant-preexisting antibodies raised against AAV2. RESULTS AND DISCUSSION We previously demonstrated serotype-dependent differences in the titers of NAbs generated after intrasciatic administration in mice. Three weeks post injection of AAV1, 2 or 8, AAV8-transduced animals had the lowest titers. 16 In addition, using AAVrh.10, we have observed less NAbs were raised after intrathecal injection to mice compared with AAV serotypes 1, 8 and 9 (unpublished data). Here, we compared the immunogenicity of AAVrh.10 and AAV9 in mice first by quantifying the total anti-AAV IgG raised 3 weeks after intravenous injection of either virus, and then testing the sera’s capacity to neutralize the virus using a luciferase reporter. We chose AAV9 for comparison as it crosses the blood-brain barrier and is therefore a popular candidate for efficient central 1 Department of Biochemistry and Molecular Biology, Center of Animal Biotechnology and Gene Therapy (CBATEG), Universitat Autònoma de Barcelona, Bellaterra, Barcelona, Spain and 2 Institut Català de Recerca i Estudis Avançats (ICREA), Barcelona, Spain. Correspondence: Dr A Bosch, Department of Biochemistry and Molecular Biology, Center of Animal Biotechnology and Gene Therapy (CBATEG), Universitat Autònoma de Barcelona, Edifici H, Bellaterra 08193, Barcelona, Spain. E-mail: assumpcio.bosch@uab.es Received 5 June 2014; revised 7 October 2014; accepted 10 October 2014; published online 20 November 2014 Gene Therapy (2015) 22, 196 – 201 © 2015 Macmillan Publishers Limited All rights reserved 0969-7128/15 www.nature.com/gt