Hydroxylation of N-acetylneuraminic Acid Influences the in vivo Tropism of N-linked Sialic Acid-Binding Adeno-Associated Viruses AAV1, AAV5, and AAV6

Adeno-associated virus (AAV) vectors are promising candidates for gene therapy. However, a number of recent preclinical large animal studies failed to translate into the clinic. This illustrates the formidable challenge of choosing the animal models that promise the best chance of a successful translation into the clinic. Several of the most common AAV serotypes use sialic acid (SIA) as their primary receptor. However, in contrast to most mammals, humans lack the enzyme CMAH, which hydroxylates cytidine monophosphate-N-acetylneuraminic acid (CMP-Neu5Ac) into cytidine monophosphate-N-glycolylneuraminic acid (CMP-Neu5Gc). As a result, human glycans only contain Neu5Ac and not Neu5Gc. Here, we investigate the tropism of AAV1, 5, 6 and 9 in wild-type C57BL/6J (WT) and CMAH knock-out (CMAH(−/−)) mice. All N-linked SIA-binding serotypes (AAV1, 5 and 6) showed significantly lower transduction of the heart in CMAH(−/−) when compared to WT mice (5–5.8-fold) and, strikingly, skeletal muscle transduction by AAV5 was almost 30-fold higher in CMAH(−/−) compared to WT mice. Importantly, the AAV tropism or distribution of expression among different organs was also affected. For AAV1, AAV5 and AAV6, expression in the heart compared to the liver was 4.6–8-fold higher in WT than in CMAH(−/−) mice, and for AAV5 the expression in the heart compared to the skeletal muscle was 57.3-fold higher in WT than in CMAH(−/−) mice. These data thus strongly suggest that the relative abundance of Neu5Ac and Neu5Gc plays a role in AAV tropism, and that results obtained in commonly used animal models might not translate into the clinic.