New Chemoproteomic Method Pinpoints Where Small Molecules Bind — With Unprecedented Precision
A UCLA-led research team has developed a powerful new platform for mapping exactly where small molecules interact with proteins across the entire proteome — a capability that could meaningfully accelerate drug discovery and improve understanding of how approved drugs behave inside living cells. The method, called SEE-CITE, addresses one of the most persistent technical bottlenecks in modern chemical biology and opens new possibilities for target identification, lead optimisation, off-target profiling, and mechanistic characterisation of approved therapeutics.
Published in Nature Chemistry in April 2026, the study introduces SEE-CITE (Silyl Ether Enables Chemoproteomic Interaction and Target Engagement) — a chemoproteomics platform that overcomes long-standing analytical challenges in photoaffinity labelling, one of the field’s most established techniques for identifying drug targets and mechanisms of action. In doing so, it brings the industry measurably closer to a proteome-wide, binding-site-resolved understanding of how small molecules engage their protein partners.
The Persistent Bottleneck in Photoaffinity Labelling
To appreciate the significance of SEE-CITE, it is worth understanding the problem it was designed to solve — and why that problem has remained unsolved for so long.
Photoaffinity labelling has been a cornerstone of target identification for over a decade. The technique attaches a light-activated chemical handle — most commonly a diazirine group — to a molecule of interest. When the probe is introduced to cells or cell lysates and exposed to UV light, it generates a highly reactive intermediate that crosslinks irreversibly to nearby proteins. Those proteins are then identified by mass spectrometry, revealing which targets the molecule engages inside a biological system.
The approach has been applied to an extraordinary range of molecular interactions: protein-lipid, protein-metabolite, protein-natural product, protein-glycan, and protein-drug binding have all been studied using photoaffinity labelling. It has become particularly valued in fragment-based drug discovery, where it enables target identification directly in cells rather than requiring purified proteins.
But photoaffinity labelling has always carried a fundamental limitation: while it can tell you which protein a molecule touches, determining precisely where on that protein the interaction occurs has been far more difficult. When a probe crosslinks to an amino acid residue, it creates a covalent adduct — a modified peptide with a large, chemically complex mass addition attached to it. During tandem mass spectrometry analysis, these adducts fragment in ways that are difficult to interpret, making it hard to confidently assign the labelling to a specific residue rather than a general region of the protein sequence.
Various partial solutions have been attempted. Researchers have used smaller, more minimalist probe designs to reduce the size of the adduct. They have developed improved computational tools to better interpret the resulting mass spectra. Progress has been made, but coverage has remained modest, localisation confidence has been inconsistent, and the ability to make quantitative, head-to-head comparisons of how different compounds engage specific binding sites has been largely out of reach. For drug discovery teams, this has meant that one of the most powerful target identification technologies available has delivered only a fraction of its potential insight.
SEE-CITE was built specifically to change that.
The SEE-CITE Solution: A Cleavable Handle That Simplifies the Analytical Problem
The core innovation in SEE-CITE is deceptively elegant. Rather than trying to better interpret the complex adducts generated by conventional photoaffinity labelling, the UCLA team engineered a probe design that removes the problem at source.
The key is a chemically cleavable silyl ether linkage incorporated into the photoaffinity probe. After UV crosslinking, and before mass spectrometry analysis, this silyl ether bond can be selectively cleaved under mild acidic conditions. This releases the bulky, drug-like portion of the probe from the crosslinked peptide, leaving behind only a small, well-defined mass tag at the labelled residue. The resulting modified peptides are structurally far simpler, fragment more predictably during tandem mass spectrometry, and yield mass spectra that are substantially easier to interpret.
The practical consequences are significant. Binding site localisation becomes more reliable. Coverage — the number of labelled sites that can be confidently identified across the proteome — increases substantially. And because both light- and heavy-isotope versions of the cleavable tag can be used in parallel, the platform enables quantitative, head-to-head comparisons of how different compounds engage specific binding sites. Two structurally distinct small molecules can be applied to the same sample, their relative engagement at individual amino acid residues measured directly, and the resulting data used to map which binding sites each compound prefers.
The name SEE-CITE reflects both the mechanism and the capability: Silyl Ether Enables Chemoproteomic Interaction and Target Engagement.
Building the Computational Infrastructure to Match
Developing the chemical platform was only part of the challenge. Extracting reliable binding site assignments from the resulting mass spectrometry data requires computational tools capable of accurately localising modifications to specific residues — a non-trivial problem, particularly when the fragmentation patterns produced by modified peptides differ from those of their unmodified counterparts.
To address this, the research team extended the MSFragger algorithm within the widely used FragPipe computational platform. MSFragger is one of the field’s leading tools for peptide identification in mass spectrometry-based proteomics, and the team built a dedicated PAL workflow — released publicly as part of FragPipe 22 — that introduces a new set of localisation scoring metrics (hyperscores) specifically designed for photoaffinity labelling and SEE-CITE data.
Conventional modification searches assign modifications to the first plausible residue in a peptide sequence when the evidence is ambiguous. The new hyperscore approach instead evaluates the quality and pattern of both shifted (modification-bearing) and unshifted fragment ions across every candidate residue, providing a principled, evidence-based localisation call rather than a default assignment. For peptide-spectrum matches where shifted ions are absent — which occurs in roughly twenty percent of cases, likely due to complete loss of the PAL adduct during fragmentation — the method can still infer the most probable labelling region from the pattern of unshifted ions.
The practical result is a marked improvement in localisation confidence. Case studies from the paper illustrate cases where conventional variable modification searches placed the modification on the wrong residue — an issue that the new workflow resolves by correctly identifying the labelled site based on ion-level evidence. For drug discovery applications, where the difference between a surface-exposed residue and an active site residue can determine whether a binding site is therapeutically relevant, this improvement in accuracy is not a minor technical detail. It is operationally significant.
Validated Against Two FDA-Approved Leukaemia Drugs
The SEE-CITE platform was put to work on a carefully chosen proof-of-concept system: two FDA-approved drugs used in the treatment of chronic myelogenous leukaemia (CML) that share the same primary target — the BCR-ABL1 oncogenic kinase — but operate through fundamentally different mechanisms.
Dasatinib is a type I ATP-competitive kinase inhibitor that binds in the active site of ABL1. Its off-target profile has been extensively characterised by chemoproteomics and other methods, and it is known to engage multiple Src family kinases in addition to its primary target. Asciminib, approved by the FDA in 2021, works through a completely different mechanism — it binds allosterically to the myristoyl pocket of ABL1, a site entirely distinct from the ATP-binding site. Because asciminib is a more recent approval, its off-target profile is considerably less well understood.
This pairing made the two drugs an ideal system for testing SEE-CITE. Using isotopically differentiated probes derived from each drug — conventional PAL probes as benchmarks alongside SEE-CITE probes — the team applied quantitative chemoproteomic analysis to both recombinant ABL1 and intact cancer cell lines.
On recombinant ABL1, the results were unambiguous. SEE-CITE correctly identified the known binding sites of both drugs on the kinase — dasatinib’s probe labelling residues in and around the ATP-binding pocket, asciminib’s probe labelling residues associated with the myristoyl pocket — with quantitative ratios clearly distinguishing the two binding modes. The method also identified labelling at residues in close proximity to the binding sites but not directly within them, likely reflecting the dynamics of probe binding, the flexible linker in the SEE-CITE design, and some remaining ambiguity in residue-level assignments. Despite these nuances, the ABL1 data provided a clear, structurally interpretable validation of the platform.
The cell-based protein-level profiling data told an equally informative story about the two drugs’ selectivity profiles. Consistent with dasatinib’s active-site mode of inhibition, its probe showed markedly broader kinase engagement than the asciminib probes, capturing previously characterised off-targets including BTK, CSK, LYN, and RIPK2. The asciminib probes displayed a substantially narrower kinase profile — consistent with the expectation that allosteric, non-ATP-competitive inhibition would minimise kinase off-targets — though a handful of kinases, including AGK, PKN1, and PDPK1, showed evidence of asciminib probe binding.
Uncovering Previously Unknown Binding Sites
The most discovery-relevant findings from the study came from applying SEE-CITE to characterise asciminib’s off-target interactions at the binding site level — territory that had not previously been mapped for this drug.
Two proteins emerged as particularly notable: RTN4 (reticulon-4) and COX5A (cytochrome c oxidase subunit 5A).
RTN4 is an endoplasmic reticulum membrane protein with multiple biological roles, including regulation of axonal growth and ER morphology. Its best-characterised functional domain, Nogo-66, interacts with the Nogo receptor (RTNR) and plays a role in inhibiting neuronal regeneration. SEE-CITE identified a labelling site on RTN4 at a cysteine residue (C1101), and mutational analysis confirmed that this site is relevant to the protein’s biological function: a C1101S mutant showed reduced labelling, and functional assays demonstrated that asciminib treatment inhibits the Nogo-66–RTNR interaction. The significance of this finding extends beyond asciminib specifically — it demonstrates that SEE-CITE can identify functionally relevant binding sites that would not have been discovered through conventional target identification approaches, including on proteins not typically considered part of the kinase-focused drug discovery landscape.
The COX5A findings were equally striking. COX5A is a subunit of Complex IV of the mitochondrial electron transport chain — a protein with no obvious structural relationship to the myristoyl pocket of ABL1 that asciminib was designed to target. Yet multiple complementary experiments — protein-level AfBPP profiling, quantitative SEE-CITE site mapping, competition studies with excess asciminib, and dose-dependent labelling — all pointed consistently to COX5A as a specific asciminib binder. Molecular docking placed asciminib in a plausible binding region near residues T79–Y82 on COX5A, and functional studies confirmed the biological consequence: overexpression of COX5A partially rescues mitochondrial respiration (state 3 respiration with pyruvate/malate as substrates) in cells treated with asciminib, while cells overexpressing a mutant COX5A lacking the binding site do not show this effect.
This is a direct demonstration that SEE-CITE can not only identify off-target binding sites but also provide a mechanistic link from the binding event to a functional consequence — precisely the kind of causal insight that drug development teams need to interpret unexpected clinical observations, understand drug tolerability, or evaluate whether an off-target interaction might represent a liability or an opportunity.
Proteome-Wide Coverage: Scale and Scope of the Platform
Beyond these specific case studies, the SEE-CITE paper reports proteome-wide binding site data of considerable scope. Across the comparative experiments using asciminib and dasatinib probes, the platform identified labelled binding sites across 731 proteins labelled by the asciminib probe, with 49 kinases among the captured targets.
The method showed relatively modest bias towards highly abundant proteins — an important characteristic for a discovery platform, since many clinically relevant targets are expressed at relatively low levels. Low-abundance proteins including STARD7 and NDUFA13 were successfully captured, while some medium-abundance proteins such as STING and ABL1 itself were absent from certain datasets, likely reflecting a combination of expression levels, peptide detectability, and probe labelling efficiency at those targets.
The paper also demonstrates that SEE-CITE is compatible with heterologous protein overexpression, extending the platform’s reach to targets that may be missed at endogenous expression levels. Using overexpressed STING as a test case, the method identified two cysteine labelling sites — C91 (a known palmitoylation site) and C148 (a residue involved in STING polymerisation) — that were not detectable by conventional gel-based analysis. This heterologous overexpression compatibility is a practically important feature for discovery campaigns targeting proteins with limited endogenous abundance.
The distribution of labelled residues across the dataset was also informative: approximately 55% of localised modifications mapped to glutamate or aspartate residues, with a further 25% at tyrosine, histidine, or cysteine — a profile consistent with diazirine reactivity preferences reported in previous studies and providing a useful frame of reference for interpreting SEE-CITE data in future applications.
Implications for Drug Discovery Strategy
Taken together, the capabilities demonstrated in the SEE-CITE paper have implications across several stages of the drug discovery and development process.
In early discovery, the ability to map where fragments and early-stage small molecules bind across the proteome — at residue-level resolution, quantitatively — provides a richer picture of a molecule’s binding preferences than conventional target identification can offer. Fragment-based programmes in particular could benefit from knowing not just which proteins a fragment touches, but which specific pockets it occupies, enabling more informed decisions about which fragments to elaborate and in which direction.
In lead optimisation, quantitative binding site comparison between structural analogues could support selectivity optimisation with greater precision than is currently achievable. Understanding whether a structural modification shifts engagement between binding sites, reduces off-target pocket occupancy, or affects allosteric versus active site binding are all questions that SEE-CITE is designed to answer.
For approved assets and late-stage candidates, the method offers a systematic route to off-target characterisation — one that goes beyond protein-level identification to provide mechanistically interpretable binding site data. The asciminib–COX5A interaction discovered in this study is a compelling example: a binding event on a mitochondrial protein that affects cellular respiration, identified and mechanistically characterised through a single chemoproteomic platform. For organisations seeking to understand the full pharmacological profile of their clinical assets, this kind of capability has real value.
More broadly, the growth of interest in allosteric drug discovery, targeted protein degradation, and covalent drug development all point towards a future in which binding site-level knowledge is increasingly central to medicinal chemistry strategy. SEE-CITE is well positioned as a platform technology for generating that knowledge systematically and at scale.
Availability and Accessibility
The computational infrastructure required to run SEE-CITE analyses is not restricted to the Backus laboratory. The dedicated PAL workflow within FragPipe 22 — including the new hyperscore localisation tools — has been made publicly available, meaning research teams with existing mass spectrometry capabilities can begin applying the method to their own discovery programmes. The synthetic routes to SEE-CITE probes are described in detail in the paper’s supplementary material, providing a practical starting point for teams wishing to develop probes for their own molecules of interest.
The availability of both the chemical methodology and the computational infrastructure as open tools is a significant factor in the likely adoption of SEE-CITE across the field. Methods that require highly specialised equipment or proprietary software face structural barriers to widespread use; SEE-CITE has been deliberately built to minimise those barriers.
Looking Ahead
SEE-CITE is one of several methodological advances in recent years that collectively point towards a step change in the resolution and scale at which small molecule-protein interactions can be characterised. Paired with advances in structural biology, machine learning-guided structure prediction, and increasingly sophisticated mass spectrometry instrumentation, platforms like SEE-CITE are part of an emerging toolkit that promises to make the drug discovery process more systematic, more informed, and ultimately more efficient.
For the senior leaders shaping drug discovery strategy across biotech and pharma, the arrival of binding-site-resolved, proteome-wide chemoproteomic platforms represents a meaningful expansion of what is technically achievable — and a signal that the gap between target identification and mechanistic understanding is narrowing. The SEE-CITE paper is a rigorous, well-validated demonstration of what that looks like in practice.
Source: Ngo, Takechi, Sivakumar et al., “Small-molecule binding-site discovery using silyl ether-enabled chemoproteomics.” Nature Chemistry, April 2026. DOI: 10.1038/s41557-026-02127-4