Comprehensive Characterization of Missense Mutants in Factor IX - SUMMARY
Deficiencies in coagulation factor IX (FIX) cause the bleeding disorder hemophilia B, while high levels of FIX
pose a risk for thrombosis. Thus, genetic variation in the F9 gene encoding FIX can impact bleeding by
decreasing FIX expression or activity, or can impact thrombosis by increasing FIX expression or activity.
However, when a new DNA variant is discovered in the F9 gene, we typically lack the evidence needed to
confidently determine if the variant alters the function of the encoded FIX protein and, if so, how severely. For
example, in the national hemophilia genotyping project MyLifeOurFuture, the majority of F9 missense variants
(i.e. DNA changes predicted to change a FIX amino acid) discovered in patients with hemophilia B had
insufficient evidence to be classified as pathogenic. Functional studies, where a FIX variant’s stability, activity or
other properties are evaluated in vitro, can provide strong evidence to inform interpretation of F9 variants.
However, traditional functional studies are time- and resource-intensive, so testing the hundreds of F9 variant
we have observed so far would be impractical. Evaluating the thousands of possible FIX missense variants we
could observe as more individuals are sequenced would be impossible. Instead, we propose a new approach to
express and characterize nearly every possible missense variant in the FIX protein to advance our understanding
of FIX biology, improve the interpretation of genetic variation in the F9 gene, and advance hemophilia care and
treatments.
To accomplish this goal, we will employ deep mutational scanning, a method we developed for measuring the
effects of massive numbers of missense variants of a protein simultaneously. Here, we will display a library of
nearly all possible FIX missense variants tethered to the surface of cultured human cells. We propose to exploit
this FIX surface display library in two aims: 1) Quantifying the effect of nearly every possible F9 missense variant
on FIX expression and secretion, and 2) Quantifying the effect of nearly every possible F9 missense variant on
specific FIX functions including Gla-domain gamma-carboxylation and activation of FIX by factor XIa (FXIa).
These aims will reveal how nearly all possible missense variants in FIX impact expression, secretion, gamma-
carboxylation, and activation by FXIa. In aim 3, we will use these large-scale functional data to dissect the
mechanism by which F9 pathogenic variants disrupt FIX function. We will also use machine learning, leveraging
the functional data along with other features to predict pathogenicity for each missense variant in FIX. Taken
together, the functional data we generate, the analyses we propose, and tools we build will transform the
characterization of F9 variants. They will also serve as a resource to better understand FIX biology, improve the
clinical translation of F9 genetic information, and inform new treatments.