Adeno-associated viruses: Monkey see, monkey do

Like many other gene therapy vectors, effectively transfecting target cells without provoking an antibody response that limits readministration has often proved to be a bridge too far for adeno-associated viruses (AAVs). Now James Wilson's group at the University of Pennsylvania has isolated two new AAV serotypes from chimps that may provide the solution to this tricky problem.¹

AAVs seemed to be the ideal solution for problems that adenoviral vectors have in delivering long-term expression of transgenes in the target region. These problems result from the immune and inflammatory response adenoviral vectors provoke, which usually leads to rapid elimination of the transduced cells.^2,3,4

Unlike adenoviruses, AAVs are well tolerated and do not cause a strong innate end response or cytotoxic T cell response. In addition, transgenes delivered with AAVs tend to be expressed for longer than adenoviral-delivered genes. However, AAVs still do provoke an antibody response, and transfection efficiencies using AAVs are often low.

Antibody responses to one AAV serotype tend not to affect another, so to some extent the problem of readministration of the transgene could be overcome with judicious use of the six AAV serotypes previously available. However, for AAVs to be a really effective gene delivery system, an expanded set of AAVs with higher transfection efficiencies was needed. Now Wilson's group has isolated two new AAV serotypes from the chimpanzee and in doing so, it seems, has effectively achieved this goal.

The approach the authors used to isolate the new serotypes was unique and inventive. Genomes of latent AAVs (ie lacking helper viruses) were amplified using PCR primers designed to a conserved region of sequence in the gene encoding the capsid and flanking the variable regions of this gene. This method is more sensitive and more widely applicable than previous methods that required in vivo rescue of AAV by helper adenovirus. In future, this approach should allow quick and efficient isolation of new AAV variants and identification of previously isolated serotypes.

The targeting of chimpanzee AAVs was also a clever, and ultimately successful, strategy. The chimp AAVs are sufficiently similar to human AAVs that they should be able to deliver genes to human target cells and be propagated with human adenovirus. Conversely, the new AAVs (AAVs 7 and 8) are sufficiently different in the variable capsid area to AAVs 1–6 that there was no serological crossreactivity with these other types, or other human AAVs under development. Thus, these two new serotypes provide exciting new delivery options for gene therapists, which will be unhampered by any prior therapeutic attempts using human-derived AAVs.

There is no reason why further AAV serotypes cannot be isolated in the same way. Therefore, this new work paves the way to the isolation of an even larger set of AAVs that should circumvent a problem of repeated administration and production of neutralizing antibodies against a given AAV serotype.

Perhaps the most striking endorsement of the potential of these new serotypes as gene delivery vectors comes from the amazing efficiency that AAV8 demonstrates in transferring genes into liver cells. The Wilson group showed that this serotype was one or two orders of magnitude better than any other serotype previously used!

It will be important to determine why AAV 8 is expressed at high and sustained levels in the liver. One possibility might be that expression of the α-1 antitrypsin (A1AT) transgene under the control of the thyroid binding globulin (TBG) is optimal for long-term expression in the liver. This enabled long-term expression in the liver and in an almost unattenuated fashion for at least 50 days.

Another possibility is that AAVTBGA1AT elicits a low immune response in the liver, especially after delivery by the portal vein.

The third, and most likely, explanation is that modifications of the AAV 8 capsid protein enabled higher affinity binding and entry of the virus into the cell. Previously identified receptors include heparin sulfate proteoglycan,⁵ alphaVbeta5 integrin⁶ and fibroblast growth factor receptor 1 (FGFR1).⁷ In addition to expression of receptors and coreceptors, impaired intracellular trafficking and escape from the endocytic pathway of AAV prior to processing in the nucleus is a rate-limiting step for AAV⁸ (Figure 1).

Figure 1

AAV binds to cells by receptors and coreceptors including heparin sulfate glycoprotein, alphaVbeta5 integrin and fibroblast growth factor receptor. The AAVs are endocytosed by invagination of the plasma membrane. The AAVs must escape from this endocytic vesicle to travel to the nucleus where second-strand DNA synthesis occurs enabling expression of the transgene.

Enabling binding, entry and final transcription of the AAV transgene, all without eliciting an immune response, is a challenging goal. The exact role that capsid proteins play in this process (including viral entry, endosomal escape and nuclear transport (Figure 1)) is not known.⁸ Nevertheless, several investigators are investigating ways to engineer known AAV capsids and genomes in an attempt to achieve this goal for different cell types.

An equally attractive alternative to ‘building your own’ is to ‘browse the catalog’ of naturally available AAVs, in other words identifying AAVs that exhibit desired properties of high-affinity receptor binding, intracellular transportation, expression and evasion of the natural immune response.

The new results from Wilson's group demonstrate an effective method for identifying novel AAVs with different capsid variations. New AAVs potentially have different intracellular properties and provoke different types of immune responses. Thus, there is now real hope that naturally occurring AAVs can be identified that will transfer genes to target cells and allow them to be expressed for long enough and at high enough levels to be an effective genetic treatment.