Skip Ribbon Commands
Skip to main content

Kay: Identifying rAAV Vectors with Enhanced Muscle Transduction via Capsid Evolution


The goal of this project is to develop a recombinant adeno-associated virus (rAAV) vector that is both more robust at transducing human muscle cells in vivo, and more resilient to neutralization by pre-existing human antibodies. To do this, we will use a viral evolution strategy. Replicating rAAV libraries containing 10e6 to 10e7 unique capsid sequences will be used to serially infect a mouse model containing chimeric human-mouse muscle fibers. After multiple rounds of infection, the most enriched capsid genomes (determined by DNA sequencing) are used to construct novel AAV treatment vectors. The relative transduction of the new capsid serotypes will be compared with standard AAV vector serotypes currently used in, or being considered for, clinical trials. The ultimate goal is to derive a clinical AAV vector candidate that can efficiently transduce human muscle at a level sufficient to express therapeutic levels of cloned human antibodies with broad spectrum protection against HIV.

The premise for these proposed studies is based on data from our recent l​iver work demonstrating that observed serotype-dependent AAV transduction properties in human trials more closely resembles preclinical studies in a humanized-liver mouse model than non-human primates. Moreover, using this chimeric humanized-liver mouse model and performing AAV capsid library screens, we were able to evolve a human selective AAV capsid that is currently being studied as a clinical candidate (Lisowski et al., in review). Based on these successes, we believe that humanized mouse models may represent a better approach for both selecting and evaluating clinically relevant AAV serotypes for gene therapeutic applications.

Our group was one of the first to employ multi-capsid shuffling approaches for directed evolution (1). Even though rAAV library screening for unique properties has become widespread, our approach is novel in that it utilizes replicating AAVs throughout the entire selection and evolution process. Selection approaches that use replicating AAVs have the potential to lessen problems of cross-packaging (a phenomenon whereby capsids package gene sequences that are unrelated to the capsid utilized for transduction) after multiple rounds of passaging. Perhaps even more importantly, it has become clear that capsid-mediated cellular AAV binding and uptake are not the only parameters that influence the efficiency, cell-type and species-specific transduction properties. A good example is the difference between AAV2 and AAV8, wherein equal numbers of virions will enter mouse hepatocytes in the liver but the transduction efficiency and transgene expression can differ by 1020 fold (2). Our previous work established that little, if any, wild type AAV replication occurs in mouse tissues whether a mouse or human adenovirus is used as the helper virus (1), thus preventing the cross-selection for mouse cell transduction by AAV capsids.

Tom Rando’s laboratory has developed a human muscle xenograft mouse model. The immunodeficient recipient mice are transplanted with muscle progenitors from human muscle biopsies prepared after a brief culture period. The human muscle fibers fuse with those from the mouse making a chimeric fiber composed of both human and mouse nuclei and proteins. His laboratory is working on a publication but will collaborate with our group on the project (see Rando letter). We know that these cells are amenable to adenovirus transduction as his lab commonly uses this vector to mark the human cells prior to transplant. 


1. Establish an efficient route to transduce the chimeric muscle fibers

2. Library selection of novel AAV capsids

3. Test best candidates in culture and xenograft model and establish human neutralization


To date we have created a new AAV capsid library and performed serial passages into three different types of human derived muscle cells as outlined above. In two of the three studies we have obtained strong selection for specific AAV capsid variants. We are currently making vectors from the top 10 selected capsids and planning to test their ability to transduce the human chimeric muscle tissue in vivo. ​


Grant at a Glance

Principal Investigator

Mark Kay, PhD

Grantee Institution

Stanford University, CA, USA

Project Title

Identifying rAAV Vectors with Enhanced Muscle Transduction via Capsid Evolution

Grant Award

Up to $0.6 million​, awarded October 2013

For general questions about the CAVD or the content of the these pages, please contact
For technical issues with the website, please contact