With thanks to Thomas Powell, Martin Ebner and Andrew Creese, Immunocore Ltd., Oxford, UK

Summary

• This application note describes the use of different cross-linking mass spectrometry strategies to probe structural information in TCR:pHLA complexes..
• This offers significant advantages in decreasing the time burden required for data analysis using traditional crystallographic approaches.
• The Byosphere workflow simplifies data analysis by merging samples prepared with various enzymes or linkers into a unified, easily interpretable output file.

Introduction

TCR-based therapies are at the forefront of innovative treatment approaches, harnessing the inherent ability of TCRs to identify and bind to specific peptide antigens (typically 8 – 20 amino acids in length). These are displayed by major histocompatibility complex (MHC) molecules on the surfaces of infected or cancerous cells. In humans, these molecules are known as human leukocyte antigens (HLAs).

ImmTAX technology, developed by Immunocore Ltd., uses affinity-enhanced T cell receptors (TCRs) to target specific peptide-HLA complexes (pHLAs). These TCRs are fused with an anti-CD3 effector domain that recruits and activates T cells to destroy the target cells.

TCR Binding Orientation

For a TCR based therapeutic candidate to progress through the pipeline, the binding orientation of the TCR relative to the pHLA is understood to be fundamentally important. For most candidates, the TCR should bind in a canonical orientation, with the α and β chains of the TCR aligning diagonally across the peptide-binding groove of the HLA molecule (see Figure 1). However, in some rare cases, non-canonical or reverse binding occurs. In these instances, the TCR α and β chains interact with the opposing helices of the HLA. Such binding often leads to suboptimal activation of T cells and potentially immunogenic reactions. These binders are not typically considered ideal biotherapeutic candidates.

Figure 1. Binding orientations of TCR-peptide-HLA complexes:TCRs binding in a canonical orientation (left) make ideal biotherapeutic candidates, whilst those binding in a non-canonical orientation do not. 

Crosslinking-MS

Cross-linking mass spectrometry (XL-MS) is a powerful technique that uses chemical crosslinkers to create covalent bonds between nearby protein residues. Following digestion, the resulting cross-linked peptides are identified using liquid chromatography-mass spectrometry (LC-MS). This approach generates spatial restraint information by pinpointing cross-linked residue pairs.

Despite its utility, analyzing XL-MS data remains computationally challenging due to the complexity of identifying cross-linked peptides, highlighting the need for advanced software solutions.

Here we present three case studies demonstrating how XL-MS in combination with advanced Byosphere data processing tools, was applied to investigate the 3D structure and conformation of TCR molecules and TCR-pHLA complexes. Notably, in the case of TCR-pHLA complexes, XL-MS was able to differentiate TCRs bound in canonical and non-canonical orientations.

Experimental

Sample Preparation

Each TCR or TCR:pHLA complex was crosslinked with sulfo-SDA, BS3 or a DMTMM/ adipic acid mix using a range of protein-to-crosslinker molar ratios (specified in text). Samples were enzymatically digested and analysed by LC-MS. All mass spectrometry experiments were performed on a Q Exactive Plus (Thermo Fisher Scientific).

Data Processing

Raw data files were processed in Byosphere using the cross-link workflow. As Byosphere is vendor neutral software, data can be taken from any LC-MS systems. XL (Cross-link) searches can be customized to accommodate for mixed linkages between different amino acids.

Results

Case study 1 – Identification of intramolecular Cross-links within an TCR

Heterobifunctional cross-linkers are the most used format in XL-MS experiments. These reagents possess two different reactive groups, allowing them to link functional groups on biomolecules.

Among these, BS3 cross-linkers are particularly popular. This molecule primarily targets lysine residues due to its reactivity with primary amines - but may also react with a protein N-terminal as well as serine, threonine, tyrosine or asparagine, albeit to a lesser extent.

In the Byonic (MS2 identification) node, the ‘Crosslink Custom’ option was enabled and configured as shown in Figure 2. The search was configured to look for BS3 cross-links between K, S, T, Y and N residues as well as the Protein N-terminals.

Figure 2. S-S, Cross-link parameters located withing the Byonic node: Cross-link searches can be customized to accommodate for mixed linkages between different amino acids. 

Combining digests from multiple enzymes can enhance sequence coverage and provide complementary MS2 peptide data for validating cross-link locations. To analyse mixed sample sets generated using different enzymes, additional MS/MS identification (Byonic) nodes can be incorporated into the workflow and configured to search only specific subsets of samples.

In this case study, 5 samples of a model TCR were incubated with different concentrations of BS3 and digested seperately with either chymotrypsin or a trypsin/Lys-C mix. The Byonic node was duplicated, with each set to use the same cross-linking criteria but tailored to the corresponding enzyme specificity. The ‘samples’ field was used to assign each node to the appropriate sample group, based on a specific text string included in the sample names. Details of this set-up are outlined in Figure 3.

Figure 3. A- Samples tab set-up showing sample name text for trypsin and chymotrypsin samples. B- Processing node set-up showing Samples field and Digestion specificity for the two searches. 

The Byosphere project inspection view is organized into multiple panels that streamline the review of all identified peptides. It provides metrics such as the overall score and Xlink-specific score, along with visual tools to assess MS2 assignment quality and theoretical MS1 isotope patterns, all of which support result validation.

Figure 4 shows the generated dataset, creating an easy-to-navigate table of peptide identifications. An example of a typical HCD MS/MS spectra for a cross-linked tryptic peptide (connected via the TCR β-chain N-terminus and β-chain lysine 194) from within the ImmTAX molecule is shown in Figure 4e. The high degree of sequence coverage enables confirmation of the connected peptides, ruling out potential misassignments.

Figure 4: Cross-link inspection view layout: A- Project table containing information about the sample set. B- Protein Coverage displaying coverage map. C- Peptide table lists all identified and filtered peptides. Selecting a peptide row in the table displays the corresponding XIC chromatograms, MS2 plots and isotope plots. D- Extracted Ion Chromatogram (XIC) Plots show the XIC of the selected peptide.  E- MS2 Plot shows the MS2 spectrum of the Cross-linked peptide pair. F - Isotope plot shows the MS1 isotope spread of the peptide. Green bars indicate the theoretical isotope distribution for comparison and validation purposes. 

After validating the results and eliminating false positive assignments, a report was generated using the disulfide template in Byosphere. The template was customized to present a detailed coverage map that included tryptic and chymotryptic peptides, all crosslinked base and partner peptides, and the exact positions of the crosslink sites on each peptide. [TP1] 

The identified cross-links may subsequently be mapped onto a crystal structure. For example, Figure 5 shows three cross links that are observed from ImmTAC α-chain lysine 26. Each cross link is within the expected linker length threshold.

Figure 5. Zoomed crystal structure of a model TCR with selected intramolecular cross-links shown. 

Case study 2 – Identification of Reverse Binders in TCR–peptide–HLA Complexes

As well as determining intra-protein interactions, XL-MS may also be used to probe inter-protein interactions. Recently, it has been shown that XL-MS may also be used to determine the broad binding orientation of TCR based therapeutics relative to their peptide-HLA targets.¹

Immunocore Ltd. was able to use XL-MS as a method to more rapidly differentiate candidates that bind in functional canonical orientations than those that do not. Here, two model TCR-pHLA complexes were cross-linked, tryptically digested and analysed by LC-MS/MS. Cross-links were identified using Byosphere.

In this work, a different heterobifunctional cross-linker, sulfo-SDA was utilized. One arm of this reagent consists of an NHS ester, which similarly to BS3, predominantly reacts with lysine residues, as well as the protein N-terminus and serine, threonine and tyrosine residues. The second functional group contains a diazirine moeity, which upon UV activation has the capacity to react with either the side chain or backbone carbon of any other amino acid. Non-specific cross-linkers may easily be incorporated into the Byosphere workflow, by listing every relevant amino acid as a potential cross-linking site (see Figure 6).

Figure 6. Cross-link searches customized to accommodate intermolecular crosslinks using a non-specific cross-linker. 

Multiple cross-links were identified between the TCR and pHLA in each sample. The MS/MS data was of good quality and the binding residues were correctly assigned in each case (see Figure 7). By understanding which domain of the TCR was in close proximity to a particular residue of the HLA, we could rapidly discern one TCR as a canonical binder and the second as a non-canonical binder. These results were subsequently validated by plotting the intermolecular cross-links on crystal structures when available (see Figure 8).

Figure 7. MS2 plot shows the annotated raw data and confirms the Xlink sites. 

Figure 8. Intermolecular cross-links identified between TCRs and pHLA for canonical and noncanonical binders. 

Case study 3 – Use of Multiple Cross-link Reagents (with different specificities) in the Same Project

To garner more structural information about a molecule or complex, multiple cross-linkers may be used in tandem with different specificities or spacer lengths.

Here, we probed similar TCR-pHLA interfaces to the previous experiment. Here, BS3 was again employed as a cross-linking reagent. In a separate reaction on the same complex, a DMTMM/ adipic acid dihydrazide mix was used. Two types of cross-links may form using this mix. Firstly, adipic acid dihydrazide crosslinks between acidic residues may be formed. Additionally, DMTMM crosslinks between acidic and basic residues may also be formed.

Each mixture was tryptically digested and analysed by LC-MS/MS. The Byonic node was triplicated, with each set tailored to the corresponding cross-link specificity. The ‘samples’ field was used to assign each node to the appropriate sample group, based on a specific text string included in the sample names. Figure 9 details this set up as well as the Byosphere output table which lists the each of each intermolecular cross-link binding position and type. The identified crosslinks strongly suggest this TCR candidate was a non-canonical binder. To demonstrate this, Figure 10 shows these crosslinks plotted on a historical crystal structure of a non-canonical binder.

Figure 9.A-Byosphere processing node for incorporating datasets from different cross-link reactions into a single output. B-Output reporting table listing each intermolecular cross-link. 

Figure 10. Intermolecular cross-links identified between TCR and pHLA using BS3 (blue), DMTMM (light blue) and adipic acid dihydrazide (yellow). 

Conclusion

This article presents a straightforward and practical method for analysing XL-MS data. By utilizing dedicated software tools, both intra- and inter-molecular cross-links can be efficiently identified and validated, with intuitive access to the raw data. The software’s capabilities streamline the analysis process by enabling the integration of samples prepared with different enzymes or linkers into a single, easily interpretable results file.

References

[1] Powell T, Karuppiah V, Shaikh SA, et al. Determining T-cell receptor binding orientation and peptide-HLA interactions using cross-linking mass spectrometry. J Biol Chem 2025; 31: 108445