Modeling pulmonary fibrosis by AAV-mediated TGFβ1 Expression : a proof of concept study for AAV-based disease modeling and riboswitch-controlled vector production

Abstract

Adeno-associated virus (AAV) vectors have gained considerable attention as tools for the genetic manipulation of various cell types, tissues and organs in vitro and in vivo, which is facilitated by their non-pathogenicity, weak immunogenicity, low biosafety requirements (biosafety level S1) and comparatively easy recombinant production. However, despite the identification of AAV6.2, a capsid protein VP1 F129L point-mutated variant of AAV6, which showed superior transduction efficiency to other serotypes in the lung of mice, no studies have utilized AAVs for pulmonary disease modeling so far. To investigate AAV6.2’s suitability for this purpose, first, its pulmonary transduction efficiency and cellular tropism, transgene expression stability and immunogenicity were studied. Our results demonstrated that AAV6.2 enables broad transduction of the murine lung with stable transgene expression observed in bronchial epithelial and type II alveolar epithelial cells over the full tested time period of four months. Notably, in contrast to a commonly used Adenovirus-5 vector, AAV6.2 did not induce any measureable acute inflammation upon intratracheal administration, thereby lowering the risk of altering relevant readouts by vector-mediated immune responses. Moreover, AAV6.2-mediated overexpression of TGFβ1 dose-dependently induced pulmonary fibrosis in mice, which mirrored several key features of the human pathology, thereby establishing proof-of-concept for AAV-mediated disease modeling. A time course experiment in direct comparison to Bleomycin-induced lung fibrosis – the most widely used model of pulmonary fibrosis – identified key differences and commonalities in disease onset and progression. We found that the most distinct difference between both models lays in an acute injury/inflammation response in the Bleomycin model prior to fibrosis development, as opposed to a phase of relatively simultaneously occurring fibrosis and inflammation in the AAV model. By next generation sequencing, mRNA and microRNA expression changes were tracked and used to identify common disease signatures during fibrosis onset and maintenance. Utilizing this approach and correlation computation using lung functional changes, protein and miRNA candidates were identified, which might present attractive candidates for fibrosis drug discovery research, among them lysyl oxidase-like 2 (LOXL2), hyaluronan synthase 2 (HAS2), fibromodulin (FMOD) as well as miR-181a-5p, miR-676-3p and miR-192-5p. Furthermore, suitability of the AAV6.2-TGFβ1 model for compound testing was also established by pharmacological intervention using a type I TGFβ receptor inhibitor. Finally, to facilitate production of AAV vectors independent of the transgene used, we developed a novel gene regulation system based on an AAV-integrated artificial, self-cleaving, guanine-responsive riboswitch. This approach enabled efficient production of high-titer AAV vectors with close-to-normal in vivo bioactivity, and further enabled dynamic transgene expression control after AAV transduction in vitro. Following further engineering, such vectors might be ultimately used as gene expression control devices in AAV gene therapy.