A global atlas of substrate specificities for the human serine/threonine kinome
Jared Lee JohnsoniD, Tomer M. YaroniD, Emily M. HuntsmaniD, Alexander Kerelsky, Junho SongiD, Amit RegeviD, Ting-Yu LiniD, Katarina M. LiberatoreiD, Daniel M. Cizin, Benjamin M. CoheniD, Neil VasaniD, Yilun MaiD, Konstantin KrismeriD, Jaylissa Torres RoblesiD, Bert van de KooijiD, Anne E. van VlimmereniD, Nicole Andrée-BuschiD, Norbert F. Käufer, Maxim V. DorovkoviD, Alexey G. Ryazanov, Yuichiro TakagiiD, Edward R. KastenhuberiD, Marcus DaSilva GoncalvesiD, Benjamin David HopkinsiD, Olivier ElementoiD, , Lewis C. CantleyiD
Abstract
Protein phosphorylation is one of the most widespread post-translational modifications in biology. With the advent of mass spectrometry-based phosphoproteomics, more than 200,000 sites of serine and threonine phosphorylation have been reported, of which several thousand have been associated with human diseases and biological processes. For the vast majority of phosphorylation events, it is not yet known which of the more than 300 protein Ser/Thr kinases encoded in the human genome is responsible. Here, we utilize synthetic peptide libraries to profile the substrate sequence specificity of nearly every functional human Ser/Thr kinase. Viewed in its entirety, the substrate specificity of the kinome was substantially more diverse than expected and was driven extensively by negative selectivity. Our kinome-wide dataset was used to computationally annotate and identify the most likely protein kinases for every reported phosphorylation site in the human Ser/Thr phosphoproteome. For the small minority of phosphosites where the protein kinases involved have been previously identified, our predictions were in excellent agreement. When this approach was applied to examine the signaling response of tissues and cell lines to hormones, growth factors, targeted inhibitors, and environmental or genetic perturbations, it revealed unexpected insights into pathway complexity and compensation. Overall, these studies reveal the full extent of substrate specificity of the human Ser/Thr kinome, illuminate cellular signaling responses, and provide a rich resource to link unannotated phosphorylation events to biological pathways.
02 /Specification & Classification
Click any tag to browse matching papers- Citations
- 731
- FWCI
- 67.20
- Field percentile
- top 1%
- Organism
- Assay
- Research Focus
- Approach
In-depth summary
Claude Opus 5.5 · xhigh effort · 5 Nov 2026 · from the full text of version 2 (23 May 2022) · CC BY-NC 4.0
Positional scanning peptide arrays profile substrate sequence specificities across 303 human serine/threonine kinases, enabling computational kinase-substrate annotation and signaling pathway inference across the human phosphoproteome.
Over 200,000 human serine and threonine phosphorylation sites have been identified by mass spectrometry, yet the upstream kinases are known for fewer than 4% of sites. Profiling nearly the entire functional human Ser/Thr kinome reveals widespread negative selectivity and allows sequence-based ranking of candidate kinases for any reported phosphosite.
Key results
ShownCombinatorial peptide library screening defines optimal phosphorylation motifs for 303 Ser/Thr kinases, reorganizing the kinome into 17 motif clusters shaped by both positive and negative residue selection and phospho-priming.
Profiling 303 kinases (286 newly profiled human motifs, 1 mouse CDKL5, 1 louse PINK1, and 15 published human motifs, including 86 dark kinases) across 198 peptide mixtures (22 amino acids x 9 positions from -5 to +4) showed that ~60% fall into 3 major motif classes while ~40% span Clusters 4-17, and 129 of 303 kinases preferred phospho-Thr or phospho-Tyr in at least one position. · Phosphorylation site substrate specificity of…; DISCUSSION; Peptide library arrays; Figure legends
ShownScoring ~50,000 human phosphorylation sites against the 303 kinase motifs accurately recovers known kinase-substrate relationships from PhosphoSitePlus.
Over 98% of ~50,000 phosphorylation sites scored in the top 10% for at least one profiled kinase; across 969 sites verified by at least 3 independent reports, established substrates achieved a median 93% percentile score for their reported kinase, and back-mapping all 303 kinases yielded a median 92% percentile rank for the reported kinase. · Phosphorylation motifs for the entire human…; Substrate scoring; Known kinase-substrate pairs; Figure legends
ShownMotif enrichment analysis of perturbation phosphoproteomics datasets identifies primary kinase targets, secondary compensatory signaling, and temporal signaling dynamics.
Scoring published MS datasets identified FAM20C motif depletion in FAM20C-deleted HepG2 cells, PKA motif enrichment in isoproterenol-stimulated myotubes (30 min), PLK1 downregulation with ATM/ATR upregulation in BI-2536-treated HeLa cells, 35 upregulated ATM target sites in irradiated A549 cells (6 Gy), and time-resolved PI3K/MAPK/IKK dynamics in insulin-treated 3T3-L1 adipocytes and LPS-stimulated dendritic cells. · Global motif analysis reveals how kinase…; DISCUSSION; Kinase enrichment analysis; Figure legends
Methods, data and reproducibility
Expressed and purified recombinant kinases in bacteria, insect, or mammalian cells and profiled their substrate specificities using positional scanning peptide arrays (198 peptide mixtures across positions -5 to +4 around a 1:1 Ser/Thr acceptor). Constructed position-specific scoring matrices (PSSMs) to cluster the kinome, rank ~50,000 PhosphoSitePlus sites, and test motif enrichment in perturbed phosphoproteomes.
- Recombinant Ser/Thr kinases profiled by positional scanning peptide arrays (286 human, 1 mouse CDKL5, 1 Pediculus humanus corporis PINK1, plus 15 previously published human motifs) · multiple species (human, mouse, Pediculus humanus corporis) · n = 303 Ser/Thr kinases
- High-confidence human Ser/Thr phosphorylation sites and experimentally validated kinase-substrate pairs from PhosphoSitePlus (July 2021) · human · n = ~50,000 phosphorylation sites; 969 verified sites (>=3 reports)
- Aligned kinase domain sequences and protein structures from PDB and AlphaFold · human · n = 290 kinase domains · PMID: 31875044
- Published perturbation phosphoproteomics datasets (FAM20C-deleted HepG2, isoproterenol myotubes, BI-2536 HeLa, 6 Gy irradiated A549, insulin 3T3-L1 adipocytes, LPS C57BL/6J dendritic cells) · multiple species (human, mouse) · n = 6 datasets
- Methods & tools
- GeneSmart, ImageQuant, SciPy, FigTree, logomaker, PhosphoSitePlus, Coral, PYMOL, BioRender, Chemdraw
- Reagents & environment
- Python packages SciPy (v1.7.1) and logomaker; FigTree (v1.4.4) (S20, S23).
- Reproducibility gaps
- No public repository URL or accession is provided in the text for the custom substrate scoring/enrichment scripts or raw densitometry files, though PSSMs and annotations are referenced in Supplementary Tables S1-S3.
Limitations (2)
- Kinase predictions rely solely on the linear amino acid sequence surrounding the phosphorylation site and do not incorporate tissue specificity, scaffolding interactions, or subcellular localization (S4, S6). · stated by the authors
- Kinase motif enrichment analysis of phosphoproteomics datasets only scored singly phosphorylated sites and excluded multi-phosphorylated peptides as well as kinases significant for both upregulation and downregulation (S22). · stated by the authors
Who should read:Read this if you analyze mass spectrometry phosphoproteomics data and want to infer upstream serine/threonine kinases or pathway activity from substrate sequence motifs.
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Accurate, high-coverage assignment of in vivo protein kinases to phosphosites from in vitro phosphoproteomic specificity data
Brandon M. Invergo·PLoS Computational BiologyPeer-reviewed·May 13, 2022
- Comprehensive substrate specificity profiling of the human nek kinome reveals unexpected signaling outputs
B. van de Kooij, P. Creixell, … M. B. Yaffe·bioRxiv·Jan 8, 2019
- Chemical proteomics decrypts the kinases that shape the dynamic human phosphoproteome
Florian P Bayer, Julian Müller, … Bernhard Küster·bioRxiv·Nov 18, 2025
- Prediction of Kinase-Substrate Associations Using The Functional Landscape of Kinases and Phosphorylation Sites
Marzieh Ayati, Serhan Yılmaz, … Mehmet Koyutürk·bioRxiv·Apr 22, 2022
- High-throughput mapping of kinase-substrate interactions with protein language models and phospho-PCA
Changhua Yu, Edward Pao, David Van Valen·bioRxiv·Oct 28, 2026
- High-throughput profiling of sequence recognition by tyrosine kinases and SH2 domains using bacterial peptide display
A. Li, R. Voleti, … N. H. Shah·bioRxiv·Aug 1, 2022