A structure-based approach to accelerated recombinase engineering - Project Summary/Abstract
HIV and HTLV are both retroviruses that cause life-threatening, incurable disease in millions of
individuals throughout the world. These retroviruses insert their DNA into the genomes of the cells
they infect. A fraction of the the infected cells do not produce virus. There is little, other than the
proviral DNA, that distinguishes these latently-infected cells from those that have not been infected. In
contrast to current therapeutics, which target viral proteins (i.e. HIV protease inhibitors or reverse
transcriptase inhibitors), we are developing a new class of retroviral therapeutics that target the
proviral DNA itself. Our approach uses engineered recombinases that specifically act on the Long
Terminal Repeats (LTRs) that flank the integrated provirus. We have already demonstrated that such
recombinases can efficiently remove HIV provirus from patient-derived cells when delivered using a
lentiviral vector. We have also shown that there are no deleterious effects when the engineered, anti-
HIV recombinase is constitutively expressed in transgenic mice. In contrast to CRISPR/Cas9 and
other nuclease-based approaches to targeting proviral DNA, recombinases act with single-nucleotide
precision thereby making them intrinsically better for this purpose than nucleases. (The unpredictable
indels associated with anti-HIV CRISPR/Cas9 appear to enhance the rate of viral escape.)
This proposal targets the primary limitation of the engineered recombinase approach, namely
that altering recombinase specificity is extremely time consuming. To date, engineering of new
recombinases has relied on random mutagenesis as a source of genetic diversity. This is highly
inefficient since the overwhelming majority of mutations screened occur in regions of the protein that
are not associated with DNA recognition. Moreover, given the vast number of potential mutations
across the entire protein, random screening, necessarily, samples the available sequence-space very
sparsely. In this application, we propose to use a new, structure-guided approach to identifying
recombinases with altered specificity. We will again screen millions of clones for the desired activity,
but the genetic diversity in our screens will be limited to key regions of the protein, and only a sensible
subset of the amino acids at these key positions will be part of the library. (To clarify precisely which
amino acids should be varied, we will determine crystal structures of anti-HIV and anti-HTLV
recombinases in complex with their LTR targets and probe the functional role of specific mutation sites
within the recombinases.) We anticipate that dense-sampling of the most relevant regions of
sequence-space will dramatically streamline the recombinase design process. This will allow us to
quickly target HIV and HTLV strains that our current-best recombinases cannot excise.