AAV Vector Dose Determines TLR9 Dependence of CD8+ T Cell Response to Transgene Product

Adeno-associated viral (AAV) vectors are currently evaluated in multiple Phase III clinical trial for the treatment of hemophilia and neuromuscular disorders. A major concern is the potential for immune responses. Viral vectors are initially sensed by the innate immune system, which shapes subsequent adaptive immune responses. Particularly, toll-like receptors (TLRs) have been reported as major sensors of pathogens during innate immune response. TLRs recognize pathogen-associated molecular patterns (PAMPs). Our previous studies found that cross-priming of AAV capsid-specific CD8⁺ T cells depended on TLR9-MyD88 pathway. TLR9 is an endosomal DNA receptor that responds most potently to unmethylated CpG motifs as found in bacterial and viral DNA. Similarly, others documented TLR9-dependent CD8⁺ T cell responses against non-secreted transgene products such as LacZ and hemagglutinin upon muscle-directed AAV gene transfer. Similarly, we published that CD8⁺ T cell responses to a secreted ovalbumin (ova) transgene product were substantially reduced (although not entirely eliminated) upon muscle gene transfer in TLR9-deficient mice [J Innate Immun. 7:302-14]. For those studies, we had used a self-complementary scAAV genomes, which we found to more strongly activate TLR9 than conventional single-stranded ssAAV vectors. Here, we performed intramuscular injections of 3 doses of ssAAV1-CMV-ova vector (2X10¹⁰, 2X10¹¹ and 1X10¹² vg) in wild-type (WT), TLR9-/-, or MYD88-/- C57BL/6 mice. Using MHC tetramer (H2-K^b –SIINFEKL), ova-specific CD8⁺ T cell frequencies were monitored in peripheral blood for up to 6 weeks. As expected from prior studies, TLR9-/- mice showed a substantially reduced response (1.2% tetramer⁺ of CD8) at the low dose when compared to WT (12% tetramer⁺ of CD8) animals (p<0.0001, n=5/group). To our surprise, CD8⁺ T cell responses were similar in TLR9-/- and WT mice at the 2 higher doses. TLR9-/- mice displayed 16% and 3.3% tetramer⁺ of CD8 frequencies at the median and the high doses, respectively; which was comparable to WT mice, where 15% and 4.8% tetramer⁺ of CD8 frequencies were observed (n=5/group). Therefore, sensing of the AAV genome by TLR9 is more critical for the CD8⁺ T cell response to the secreted transgene product at lower vector doses (possibly related to the lower levels of transgene expression). Interestingly, transgene product-specific CD8⁺ T cell responses were much reduced in MyD88-/- mice, in which 0.2% and 1.7% tetramer⁺ of CD8 frequencies were found for low and median doses. Therefore, an alternative signaling pathway that includes the MyD88 adaptor molecule likely exists that is more critical than TLR9 above a certain level of expression. The reduced strength of the CD8⁺ T cell response seen at the highest vector dose compared to the medium dose may be explained by a transient increase in FoxP3⁺ Treg and in PD-1⁺ T cells that we observed 1 week after gene transfer and that was significantly greater at the highest vector dose. In related experiments, we performed intramuscular gene transfer using a ssAAV1-EF1a-FIX vector in hemophilia B mice (C3H/HeJ F9-/-, 1x10¹¹ vg/mouse). Here, we used either a vector with native sequences or with an expression cassette that was entirely devoid of CpG motifs (and there stimulates TLR9 less effectively). CpG depletion did not have substantial effects on antibody formation against human FIX or the viral capsid. However, CD8⁺ T cell infiltrates in skeletal muscle were markedly reduced but not entirely eliminated when tissue sections were examined 1 month after gene transfer. In conclusion, TLR9 signaling is one important factor in the activation of transgene product-specific CD8⁺ T cells in AAV gene transfer, but other pathways exist that may be more critical depending on vector dose or levels of expression.