An efficient proteome-wide strategy for discovery and characterization of cellular nucleotide-protein interactions

Abstract

Metabolite-protein interactions define the output of metabolic pathways and regulate many cellular processes. Although diseases are often characterized by distortions in metabolic processes, efficient means to discover and study such interactions directly in cells have been lacking. A stringent implementation of proteome-wide Cellular Thermal Shift Assay (CETSA) was developed and applied to key cellular nucleotides, where previously experimentally confirmed protein-nucleotide interactions were well recaptured. Many predicted, but never experimentally confirmed, as well as novel protein-nucleotide interactions were discovered. Interactions included a range of different protein families where nucleotides serve as substrates, products, co-factors or regulators. In cells exposed to thymidine, a limiting precursor for DNA synthesis, both dose- and time-dependence of the intracellular binding events for sequentially generated thymidine metabolites were revealed. Interactions included known cancer targets in deoxyribonucleotide metabolism as well as novel interacting proteins. This stringent CETSA based strategy will be applicable for a wide range of metabolites and will therefore greatly facilitate the discovery and studies of interactions and specificities of the many metabolites in human cells that remain uncharacterized.

Fig 1. Optimized CETSA workflow for profiling protein-nucleotide interactions in whole cell and lysate conditions, and relative affinities and specificities of the eight nucleotides for their identified hit proteins in lysate-ITDR experiments.

(A) ITDR and melt curves workflows were applied to lysate experiments and in-cell ITDR and ITTR to whole cell experiments as a proxy for tracking the uptake and conversion of metabolite precursors in vitro. (B) Colour scale represents the minimal dose threshold (MDT) (μM) of the isothermal dose response at 52°C, with the lowest MDT (therefore higher relative affinity) in blue and the highest MDT (therefore lower relative affinity) in yellow. Hit proteins are grouped according to their associated biochemical pathways and function (black lines). Asterisks (*) indicate destabilizing hits. Black dotted boxes indicate proteins with similar ITDR responses that did not meet the hit selection criteria.

Fig 2. CETSA shifts and ITDR_CETSA with cAMP and cGMP in K562 lysates.

(A) CETSA shifts (left) and ITDR_CETSA (right) for subunits of cAMP-dependent protein kinase A (PKA) (regulatory: PRKAR1A, PRKAR2A, PRKAR1B, PRKAR2B and catalytic PRKACA), with cAMP (blue), cGMP (red) and control untreated CETSA curves (grey) measured in K562 cell lysates. (B) ITDR_CETSA hit plot with AUC (area under the curve) versus R² plot of ITDR_CETSA curves from cAMP/cGMP-treated conditions. Each dot represents the average AUC (relative thermal shift) and R² ITDR_CETSA (dose response trend) measurements per protein from two replicates. The size of the dot negatively correlates with the MDT. (C) ITDR_CETSA for small membrane A-kinase anchor protein (SMAKA) with cAMP and cGMP. Colors as in Fig 2A. (D) ITDR_CETSA for homologs of phosphofructokinase (PFK)—PFKM (muscle), PFKL (liver), PFKP (platelet) with cAMP. Data is presented as two individual technical replicates for each protein. Cells were stimulated for the indicated durations (0, 15, 60 mins and overnight (ON)) with 0.5 mM of forskolin (For), 2 mM of Db-cAMP (Db) or Br-cAMP (Br), the samples were used to measure (E) cAMP level in lysates, (F) target engagement using CETSA-western blot for PFKM, RIA, RIIB and actin, and (G) lactate secretion, glucose uptake, intracellular lactate and glucose levels. RLU: relative luminescence units. * <0.05, **, <0.01, two-tailed t-test.

Fig 3. Protein hits from lysate CETSA with deoxynucleotides.

(A) ITDR_CETSA hit plot with AUC versus R² plot of ITDR_CETSA curves after lysate treatment with AMP, dAMP, dCMP, dGMP, dTMP, and dUMP. Each dot represents the average AUC (relative thermal shift) and R² ITDR_CETSA (dose response trend) measurements per protein from two replicates. The size of the dot negatively correlates with the MDT. (B) Schematic of the proteins in the core nucleotide metabolism responding to the five deoxynucleotides and AMP. SAMHD1 and RRM1 were hits in the in-cell ITDR_CETSA experiment (Fig 5). (C) ITDR_CETSA (top) and CETSA curves (bottom) with the dNMPs for enzymes involved in the deoxypyrimidine salvage and synthesis pathways; adenylate kinase 4 (AK4), deoxycytidine kinase (DCK), deoxyguanosine kinase (DGK), thymidine kinase (TK1), thymidylate kinase (TYMK) and thymidylate synthase (TYMS). Data is presented as two individual technical replicates for each condition from one representative experiment. (D) Other protein hits from dNMPs-ITDR_CETSA involved in the nucleotide metabolism: 7-methylguanosine specific 5’-nucleotidase (NT5C3B), adenine phosphoribosyltransferase (APRT) and UMP synthase (UMPS). Data is presented as two individual technical replicates for each condition from one representative experiment.

Fig 4. Protein hits from NAD(P)(H) ITDR_CETSA.

(A) ITDR hit plots with AUC versus R² plot of ITDR_CETSA curves from lysates treated with NADPH, NADP, NADH, and NAD. Each dot represents the average AUC (relative thermal shift) and R² ITDR_CETSA (dose response trend) measurements per protein from two replicates. The size of the dot negatively correlates with the MDT. (B) Relative affinities and specificities of the four nicotinamide-based nucleotides for their identified hit proteins. Colour scale represents the minimal dose threshold (MDT in μM) of the isothermal dose response at 52°C, with the lowest MDT (therefore higher relative affinity) in blue and the highest MDT (therefore lower relative affinity) in yellow. Hit proteins are grouped according to their associated biochemical pathways and function (black lines), and overlaps indicate their association with more than one pathway/function. Asterisks (*) indicate destabilizing hits. Black dotted boxes indicate proteins with similar ITDR responses that did not meet the hit selection criteria. (C) Protein domain analysis in the NAD(P)(H) hit proteins. Hit proteins are clustered in 5 broad Interpro domain clusters (1–5), with 5 subclusters (A-E) under cluster 1. NAD annotated protein (black circular outline), NADP (red circular outline), NAD and NADP annotated (grey circular outline). Unannotated proteins are not outlined. Annotations were taken from Uniprot KB. (D) Thermal precipitation assay with recombinant CORO1C and increasing doses of NADPH at ITDR_CETSA 52°C.

Fig 5. Hit proteins from the in-cell ITDR_CETSA experiments with thymidine.

(A) In-cell MS-ITDR_CETSA curves of enzymes involved in the thymidine metabolism cascade. (B) Lysate-western blot (WB)-ITDR_CETSA validation with dTXP. Lysates were treated with 10, 2.5, 0.6, 0.1 and 0 (control) mM of dT, TMP, TDP or TTP and probed for RRM1 and ABCF1 with actin as loading control. (C) In-cell MS-ITDR_CETSA curves of ABCF1 and ABCF2. Data is presented as two individual technical replicates for each condition from 2 biological replicates (biological replicate 1 (dT_b1, blue) and biological replicate 2 (dT_b2, red)). (D) Thermal stabilization kinetics between TYMS (red) and SAMHD1 (blue) from the in-cell ITTR_CETSA experiments with thymidine. Data is presented as two biological replicates at 37°C (TYMS_37°C, SAMHD1_37°C) and 52°C (TYMS_52°C, SAMHD1_52°C), with focus on the incubation duration of 0 to 30 mins. (E) Summary of hits identified by: in-cell MS-ITDR_CETSA (dotted lines); lysate MS-ITDR_CETSA (solid lines); validated by WB-ITDR_CETSA (grey dotted lines). Common hits from the in-cell and lysate MS-ITDR_CETSA (yellow circle); metabolites (grey circle); known metabolite-protein interactions (grey solid lines).