YTHDF2 – An unprecedented epitranscriptomic reader protein

YTHDF2 – A unprecedented epitranscriptomic reader protein

Challenge
The epitranscriptomic YTHDF2 reader is a member of a family of proteins that selectively binds to a modified base (m6A) in mRNA.  Our client tasked us with setting up a complete “gene to lead” fragment-based drug discovery campaign. Since the entire family is unprecedented, there were no “tool compounds” typically required to set up assays. As we progressed, further challenges presented themselves.

Solution
Once we determined that we could immobilize YTHDF2 while retaining its selective m6A binding, we could use our proprietary TINS fragment screening technology, which does not require tool compounds to setup. TINS generated a list of primary hits from the ZoBio fragment library. Using a nearly identical immobilization approach, we observed binding of many of these raw hits in SPR. The binding response could be used to optimize the assay for fragment screening.

We subsequently confirmed that hits from both TINS and SPR targeted the desired binding site on YTHDF2 using protein observed NMR experiments. This information rich assay revealed that buffer components were also selectively binding to the desired site and effectively competing against the fragment hits. With this knowledge we could further optimize our assays and structural biology approaches.

Inverse SPR assay
The classic “forward” SPR assay identified compounds that selectively bind to YTHDF2 with ideal behavior. However, this assay does not provide insight into the biological activity of compounds, that is whether or not they prevent binding of m6A containing RNA. In order to do so, we developed an inverse SPR assay in which RNA carrying the relevant chemical modification, is immobilized. This allows us to directly probe the ability of test compounds to interfere with the binding of YTHDF2 to the immobilized RNA, in other words, a biophysics-based biological assay. We further refined the assay to quantify the level of competition based on the affinity of the ligand to discriminate and prioritize chemotypes.

The assay cascade
We efficiently guided medicinal chemistry efforts by focusing on functionally active chemotypes, as determined by iSPR. These compounds were further characterized by biochemical assays (trFRET). Chemotypes exhibiting good correlation in the different assays were prioritized. From the start of the project, NMR played a crucial role in obtaining structural information about the way selected chemotypes bound the target. Initially we could rapidly map the binding sites of 10’s of compounds at low resolution. Subsequently, when it was not possible to obtain liganded structures do to crystal packing issues, we elucidated the solution structure of ligand-protein complexes that were then used to guide medicinal chemistry.

Achievements
By capitalizing on our unique expertise in developing customized biophysical assays, we created a biologically relevant compound screening cascade in the absence of tool compounds. The cascade delivered chemically diverse sets of validated fragments that we successfully elaborated towards lead compounds using structural information that could only be gleaned through NMR.

Mettl3/Mettl14 – An epitranscriptomic writer complex

Mettl3/Mettl14 – An epitranscriptomic writer complex

Challenge
Mettl3/Mettl14 is a large (90 kDa) heterodimeric “writer” complex that uses S-adenosyl methionine (SAM) as a cofactor.  Identifying low affinity, validated fragment hits in the large amount of electron density derived from X-ray diffraction data was extremely difficult.

Solution
X-ray crystallography provides the necessary high resolution structural data needed to efficiently elaborate hits from fragment or HTS campaigns. However, even highly validated fragment hits typically exhibit low affinity (100s to 1000s of µM). As a result, the occupancy of ligand binding sites in the crystal may be substantially less than 100%, which degrades the quality of the electron density of the fragment. Moreover, fragments are small and overall somewhat symmetric, further reducing identifiable features of their electron density.

We set up a solution competition binding experiment using NMR as a readout to focus on fragments that bind at the SAM site. We quickly found a fragment containing a CF3 moiety that bound at the SAM site. We could then use 19F NMR studies as a clean, sensitive and fast approach to reveal other ligands that bound at the same site. Once we were certain of the binding site, we were readily able to identify electron density at the SAM site and confidently assign it to the ligand.

Achievements
We developed a highly efficient structure pipeline capable of determining up to 10 sub 2Å structures per month which has been the basis for elaborating fragment hits to cellular activity and beyond. Multiple chemotypes are promising, allowing for a robust hit-to-lead campaign.