Phenomics-Based Discovery of Novel Orthosteric Choline Kinase Inhibitors

Abstract

Choline kinase alpha (CHKA) is a central mediator of cell metabolism linked to cancer and immune regulation. Cellular and clinical evaluation of CHKA has been hampered by challenges in the development of drug-like choline kinase inhibitors. Here, we identify CHKA as an unexpected off-target of histone methyltransferase inhibitors using an integrated phenomic approach. We confirm CHKA as a direct protein target of the aminoquinazolines UNC0638 and UNC0737 using a combination of chemoproteomic, biochemical, cellular, and metabolic profiling assays, possibly explaining the previously reported discrepancies observed for different G9a/GLP inhibitor scaffolds in cellular assays. Using primary human cell model systems, we discover that CHKA modulation impairs IgG secretion and B-cell maturation consistent with the notion that choline metabolism plays an important role in immune signalling. Co-crystal structures of UNC0638 and UNC0737 with CHKA unravel an unexpected binding mode and suggest the inhibitors as attractive starting points for the development of selective chemical tools to further explore the biological role of CHKA in cancer and immune metabolism.

Keywords: cell painting; chemical probes; chemical proteomics; metabolomics; phenotypic.

Figure 1

Identification of CHKA as an off‐target of UNC0638 and UNC0737 by phenotypic and chemoproteomic profiling. A) Chemical structures of UNC0638, UNC0737 and A‐366. B) Cell painting results showing high similarity for the methyltransferase‐active UNC0638 as well as the negative control UNC0737 in U2OS cells. Non‐LCH: Non‐lysosomotropic features. C) Affinity‐based chemoproteomics workflow. D) Competitive chemoproteomic data reveal CHKA as a shared off‐target for both UNC0638 and UNC0737 in MCF7 lysates. Dashed lines represent log₂ competition<‐1 (p‐values from two‐sided two sample t‐test, n=2). 1,473 proteins were identified. E) Protein‐protein interaction network of EHMT1/2 complex partners (adapted from string‐db.org). F) Comparison of mean LFQ intensities. As expected, the G9a/GLP‐inactive UNC0737 does not compete with EHMT1/2 binding in contrast to the active compound UNC0638.

Figure 2

Thermal profiling by PISA and metabolomics confirm intracellular CHKA target engagement. A) UNC0638 and UNC0737 engage and stabilise CHKA in intact HepG2 cells. Dashed lines represent amplitude>0.05,<‐ 0.05 and log₁₀(adjusted p‐values)>2 (two‐sided two sample t‐test, Benjamini‐Hochberg corrected, n=3). 5,635 proteins were identified. B) Comparison of significantly changed proteins between affinity‐based chemoproteomics (strongly competed proteins with FC<‐1) in MCF7 cells and thermal profiling (significantly changing proteins log₁₀(p‐value)>2) of intact HepG2 cells. C) Effects of UNC0638 and UNC0737 on choline metabolism in MCF7 cells. NMR analysis of intracellular choline and phosphocholine (pCho) levels relative to vehicle control following incubation with indicated compounds. No change (100 %) is marked with a dashed line. Data are shown as mean±SD from three replicates (n=3).

Figure 3

Confirmation of CHKA as a direct target of UNC0638 and UNC0737. A) ITC results for UNC0638 (K _D=270±51 nM) and UNC0737 (K _D=59.3±14 nM) using recombinant human CHKA. B) UNC0638 and UNC0737 inhibit CHKA enzymatic activity in a dose‐dependent manner. Data are shown as mean±SD of three replicates. Graph is representative of two independent biological replicates (n=2). C) Co‐crystal structures of UNC0638 (PDB: 8BI6) and UNC0737 (PDB: 8BI5) bound to CHKA. UNC0737 and UNC0638 (coloured in yellow) both occupy the HC3 binding site (PDB: 3F2R). HC3 is coloured in dark grey. D) Stick representation with residues contributing to compound binding.

Figure 4

Effect of CHKA inhibitors on immune signalling. A) Network analysis of BioMAP profiling using indicated inhibitors over all 12 tested systems. CHKA inhibitors cluster together over multiple concentrations indicating mechanistic similarity in contrast to the CHKA‐inactive A‐366 (internal BioMAP control). Line indicates Pearson's correlation of full BioMAP profiles >0.7. Circle size represents concentration of compounds. B) BioMAP B cell profiling results indicating a common response via reduction of secreted IgG and sIL‐17A. The Y‐axis represents a log‐transformed ratio of the biomarker readouts for the drug‐treated sample (n=1) over vehicle controls (n ≥6). The grey region around the y‐axis represents the 95 % significance envelope generated from BioMAP historical vehicle controls. C) UNC0638 and UNC0737 reduce stimulated PBMC secretion of IgG, expressed as % of vehicle control in cell cultures from healthy blood donors. Data are shown as mean±SD calculated from seven donors (n=7). D) UNC0638 and UNC0737 reduce stimulation‐induced B cell maturation. The percent of CD27⁺IgD⁻ memory B cells of live CD19⁺ B cells is depicted for three different healthy donors, analysed after 6 days of culture in absence or presence of stimulation. E) Cell viability results for UNC0638 and UNC0737. The percent of live CD19⁺ B cells is depicted for three different healthy donors, analysed after 6 days of culture in absence or presence of stimulation. F) Correlation of competitive chemoproteomic data of UNC0638 or UNC0737 in PBMC lysates using UNC0965 affinity beads.