Mutagenesis of hexon “FX” hepatic tropism

Occam's razor is the principle of parsimony. Paraphrased, it means “if several hypotheses are possible, it is likely the simplest that is correct.” What does the philosophy of a 14th century logician and Franciscan friar have to do with Alba et al and 21st century hematology?

This story starts approximately 25 years ago when adenoviruses were initially developed as gene transfer vectors. Adenoviridae are double-stranded nonenveloped DNA viruses with icosahedral symmetry, and a homotrimeric fiber projecting from each vertex of the approximately 90-nm capsid (see figure). Early on, it was found that human adenovirus type 5–derived vectors injected intravenously into mice efficiently transduced hepatocytes. CAR (Coxsackie and adenovirus receptor), the principle in vitro receptor for almost half of the human adenovirus serotypes, was identified in 1997.^1,2  When it was quickly shown that many tissues, including liver, expressed high levels of CAR, most if not all of the labs working in this domain made the Occamian leap and assumed that the CAR-tropic Ads (most serotypes from species A, C, D, E, and F) were using this cell-adhesion molecule to infect hepatocytes.

The fly in the ointment came when one tried to retarget Ad vectors to other tissues.³  The CAR-binding residues in the fiber knob were initially identified by Roelvink and colleagues via cocrystallization of CAR and the fiber knob.⁴  Mutations were introduced into the knob to generate recombinant proteins that no longer bound CAR, and then “CAR-ablated” vectors were created. Although the CAR-ablated vectors could not use CAR to infect cells in vitro, the vectors still efficiently transduced liver cells, to the surprise of many. It wasn't until 2008, when Waddington et al untangled the mountain of conflicting data, that it was found that hexon, the major protein of the adenovirus capsid, was binding coagulation factor X (FX) when adenovirus vectors were injected intravenously into mice.⁵  This hexon-FX interaction formed a bridge to heparin sulfates on hepatocytes that, in turn, mediated transduction.

In this issue of Blood, Andy Baker's latest study identified key residues in the hexon hypervariable regions (the 7 protruding loops that harbor antigenic epitopes that, in most cases, define the adenovirus serotype) that are involved in FX binding.⁶  Alba et al use a collection of approaches based on modeling, cryoEM, swaps and substitutions of amino acids in hexon hypervariable regions 5 and 7, and intravenous gene transfer in mice to not only pinpoint the amino acids involved, but also to formally demonstrate that Ad vectors could be engineered to bypass liver transduction.

Where do we go from here? While these paradigm-shifting results remind us that William of Occam never worked with an adenovirus, we have quite a way to go before we can harness the power of adenovirus vectors for clinical gene transfer via the systemic circulation. Even in children, one finds cross-reacting anti-adenovirus Abs that can opsonize or neutralize vectors. Add to this the fact that erythrocytes have CAR on their membranes,^7,8  platelets bind some adenovirus types, and that the fenestra size in human liver may be near the limit to allow the approximately 90-nm adenovirus particle to enter. We see that we still have work to do before one can master the fate of adenovirus vectors in the circulation. Nonetheless, the field now has one less hurdle to cross.