Lunch + poster session

Crosslinking mass spectrometry

  1. Christian Schäfer – Benchmarking XL-MS for the detection of protein interaction interfaces
  2. Banerjee Swati – DSSBU, a novel MS-cleavable cross-linker for protein structural studies
  3. Alexander Röhl – CLAUDIO: An automated cross-linking data analysis and validation workflow
  4. Georgina Charlton – Using Crosslinking Mass Spectrometry to Understand the Structure of Proteins Implicated in Huntington’s Disease
  5. Gabriella Gellen – Cross-linking mass spectrometry on P-glycoprotein
  6. Paul Gershon – Combination of XLMS with deep learning-based protein structure prediction: Protein structure, dynamics, processing and higher order assembly in Vaccinia virus
  7. Hugo Gizardin-Fredon – Going against the grain: development of denaturing mass photometry and its application for cross-linking reaction optimization
  8. Nicolas Francés – Enhancing crosslinking reactions in (photosynthetic) membrane systems
  10. Mostafa Kalhor – Transfer learning to tune Prosit to predict intensities for cleavable cross-linked peptides
  11. Eugen Netz – Cleavable and non-cleavable cross-linking identification and quantification in the versatile OpenMS software ecosystem
  12. Tereza Kadavá – Spider-like complement inhibitor: structural insight into the C4b-binding protein
  13. Zdenek Kukacka – Quantitative Cross-linking Mass Spectrometry Using Data-Independent Acquisition

Integrative studies

  1. Di Tang – A human monoclonal antibody bivalently binding two different epitopes in streptococcal M protein mediates immune function
  2. Sean Burnap – Mass photometry reveals SARS-CoV-2 spike stabilisation to impede ACE2 binding through altered conformational dynamics
  3. Rosi Fassler – Defining the binding pattern of the disordered Hsp33 chaperone using an integrative structural approach
  4. Francesca Sacco – Integration of NMR spectroscopy and mass spectrometry for a new analytical workflow to characterize the oxidative stress in mAbs
  5. Lin Xi – Receptor kinase signaling of BRI1 and SIRK1 is tightly balanced by their interactomes as revealed from domain-swap chimera in AE-MS
  6. Tal Oppenheim – Characterizing the full length Npl4-Ufd1 complex and interaction with Cdc48 through an interface residue switch
  7. Joost Snijder – Simultaneous antibody sequencing and epitope mapping by integrated cryo electron microscopy and mass spectrometry
  8. Jonas Schröder – Mutational analysis of SARS-CoV-2 N-protein by structural mass spectrometry


Benchmarking XL-MS for the detection of protein interaction interfaces

Christian Schäfer1, Jiaxuan Chen1, Magdalena Schachtl-Rieß1, Carles Pons2, Mareen Welzel1, Chop Yan Lee1, Patrick Aloy2, Katja Luck1

1 Institute of Molecular Biology, Mainz, Germany
2 Institute for Research in Biomedicine Barcelona, Spain

Introduction: Recently efforts were made to uncover the human protein interactome resulting in large interaction networks. While containing large amounts of information on which proteins interact, the information on how proteins bind each other is mostly missing. Cross-linking mass spectrometry (XL-MS) emerged as a new technique to obtain structural information on individual proteins and protein complexes by chemically linking amino acids from the same or different proteins together followed by detection via MS. Although promising results have been obtained for the mapping of protein interactions and the structure determination of proteins, it remains unclear how well XL-MS is suited for the identification of protein interaction interfaces and whether their detection is biased towards certain kinds of interface types or protein architectures.

Methods: We implemented a mammalian cell culture-based system for transient expression of two interacting proteins of interest. Using bioluminescence resonance energy transfer (BRET), the existence of the protein interaction in our system is first demonstrated prior to subjecting crosslinkers either onto intact cells, cell lysate, or on protein samples after co-immunoprecipitation. We are also testing different types of cross-linking reagents that vary in their length or reactivity. The Protein Data Bank was searched for interacting protein pairs with a known interface to assemble a reference set of protein pairs with interfaces. These protein pairs were further filtered based on criteria such as interface area or number of reactive groups in the interface that might influence crosslinking efficiency.

Results: We identified over 3000 structurally resolved protein interactions in the PDB that are in principle suitable to serve as reference set for XL-MS benchmarking. Of those, we selected 374 protein pairs for further testing. Of the over 50 features available for each interaction we selected 7 as important to judge the performance of crosslinking: Interface area & number of contacts, lysine content & number of lysine contacts, polar content & number of polar contacts, protein length, and fraction of disorder. A first set of 30 interactions was cloned into expression vectors and around 20 could be verified by the BRET assay. With the first interactions being confirmed in our system we performed initial tests on the crosslinking conditions. Crosslinking was done with DSS for in vivo crosslinking and BS3 for crosslinking in lysate as well as CDI in both living cells and lysate. While no results were obtained for crosslinking in vivo, crosslinked protein could be detected by Western Blot with both CDI and BS3 when the crosslinking was performed in cell lysate. When subjecting these samples to mass spectrometric analysis, crosslinked peptides were so far only obtained for samples crosslinked with BS3. We are in the process of further optimizing crosslink detection by MS. Mapping the obtained crosslinks for the protein pair RBM8A-MAGOHB on the known structure showed good overlap with the previously identified interface but also revealed crosslinked peptides that were not part of the crystallized protein fragments. Modelling the interface between both full length proteins using AlphaFold we were able to map the crosslinks and identified a putative secondary interface between both proteins. We aim to extend crosslink analysis soon to more protein pairs.

Novel aspect: A systematic analysis is performed to test the performance of XL-MS for the detection of protein interaction interfaces.


DSSBU, a novel MS-cleavable cross-linker for protein structural studies

Banerjee Swati1, Sýs Jakub1,2, Machara Aleš1, Junková Petra1, Hubálek Martin1

1 Institute of Organic Chemistry and Biochemistry of the Czech Academy of Sciences, Flemingovo náměstí 542/2, 160 00 Praha 6, Czech Republic
2 Faculty of food and biochemical technology, University of Chemistry and Technology, Prague, Technická 5, 166 28 Praha 6, Czech Republic

Introduction: Cross-linking – mass spectrometry technique is being used to conduct 3D structural studies of protein complexes along with few challenges including effective detection of cross-linked peptide from complex mixtures. Introduction of MS-cleavable cross-linker includes a higher confidence of cross-link assignment by reducing search space with the help of characteristics signature fragments over MS-non-cleavable cross-linkers. In our study, we present new in-house produced water-soluble, MS-cleavable cross-linker, disulfodisuccinimidyl dibutyric urea (DSSBU) and displayed its potential for identifying protein-protein interaction interface. We compared the efficiency of MS-cleavable DSSBU to BS3 (MS-non-cleavable, water-soluble), and DSBU (MS-Cleavable, water insoluble) to DSS (MS-non-cleavable, water-insoluble).

Methods: Bovine serum albumin (10μM) as the model system reacted with 50 and 100-fold molar excess of BS3, DSS, DSBU and DSSBU in triplicate sets for qualitative and quantitative comparison purpose. Cross-linkers were dissolved in water, DMSO and in 2% water in DMSO to test the effect of diluent. Reaction mixtures were incubated for 30 minutes at 25°C, quenched by Tris Buffer and digested by trypsin after reduction and alkylation. Peptides were desalted, prepared for LC-MS/MS measurement in Orbitrap Fusion Lumos Tribrid mass spectrometer followed by data processing in Merox. In qualitative approach the individual linkages by 4 cross-linkers were compared, while in quantitative approach the peak areas of individual cross-linked peptides were compared.

Results: from our study, considering three types of diluents (water, DMSO, 2% water in DMSO) for each cross-linker (DSSBU, BS3, DSBU, DSS) show that, the number of linkages provided by four cross-linkers were greatly enhanced when 2% water in DMSO was used as the cross-linker’s diluent. Especially, water soluble BS3 and DSSBU provided almost twice more linkages in this diluent. Total linkages obtained by all four cross-linkers also increased by 17%. Hence, subsequent comparative study of four cross-linkers were carried out using 2% water in DMSO as a diluent. The comparative study of four cross-linkers demonstrated that majority of the identified linkages are common for almost all four cross-linkers. However, there are also some unique linkage, which are very specific to each cross-linker. The usage of water soluble DSSBU enabled us to detect higher number of unique linkages when compared with BS3. It also provided higher sequence coverage of both cross-linked α and β peptides, by offering a greater number of y and b ions in MS/MS spectra compared to BS3. The same observation was noticed for water-insoluble DSBU, when it was compared to DSS. Comparison between commercial DSBU and our in-house cross-linker DSSBU reflected that both of them provided numerous common linkages. Although, higher number of unique linkages were obtained using DSBU, DSSBU alone provided us a few unique linkages which were not detected when DSBU was used. Moreover, the label free quantitative approach, where peak area of common cross-linked peptides was used to assess peptide’s abundance for each cross-linker, exhibits that majority of common cross-linked peptides of BSA were most abundant when DSSBU was used. Hence, both qualitative and quantitative aspect of the study reveals DSSBU as a potent MS-cleavable cross-linker.

Novel aspect: DSSBU, a novel water-soluble MS-cleavable cross-linker is introduced to show its potential for application in structural proteomic studies.


CLAUDIO: An automated cross-linking data analysis and validation workflow

Alexander Röhl1,2, Eugen Netz1,2, Oliver Kohlbacher1,2,4, and Hadeer Elhabashy1,2,3

1 Applied Bioinformatics, Department of Computer Science, University of Tübingen, Sand 14, 72076 Tübingen, Germany
2 Institute for Bioinformatics and Medical Informatics, University of Tübingen, Sand 14, 72076 Tübingen, Germany
3 Department of Protein Evolution, Max Planck Institute for Biology, Max-Planck-Ring 5, 72076 Tübingen, Germany
4 Institute for Translational Bioinformatics, University Hospital Tübingen, Hoppe-Seyler-Str. 9, 72076 Tübingen, Germany

Introduction: Cross-linking mass spectrometry is a high-throughput technique for the characterization of protein structures and interactions. Cross-linking data usually consists of pairs of cross-linked residues within a protein or between proximate protein subunits that reflect known and novel information about protein structure and interactions. Those cross-links often contain untapped potential concerning multimeric protein complexes. Cross-links in homo-oligomers, for example, pose a challenge as the linked proteins possess identical sequences, making the distinction between intra-protein links and links between multiple identical subunits difficult. Here we introduce CLAUDIO, an open-source pipeline for structural analysis and validation of protein cross-linking data, and detection of homo-oligomerization signals.

Methods: CLAUDIO follows two main approaches, a structure analysis, where cross-links are validated by mapping them on their corresponding structures, if they exist, and comparing their topological distances to linker ranges, and a peptide sequence overlap analysis, which points to homo-oligomers, for intra-protein links.

Preliminary data: CLAUDIO’s input includes the UniProt IDs of the proteins, the sequences of the cross-linked peptides, and the positions of the cross-linked residues. CLAUDIO  implements a workflow that utilizes these data and evaluates them based on structures from The Protein Data Bank and AlphaFold Protein Structure Database using TopoLink. For intra-protein links, CLAUDIO can extract signals of overlapping peptide sequences and map them by homology to homo-oligomers from SWISS-MODEL. CLAUDIO’s output consists of the initial data extended by the results of distance evaluation and cross-link type re-assignments. We applied CLAUDIO to a cross-linking dataset of ~4186 cross-links from murine mitochondria, derived from two studies. It was able to structurally validate 82.7% of the intra-links and 71,7 % of the inter-links, for which structures could be found, within ~100 minutes. It was able to reassign about 37.3% of the intra-protein links as potential signals of homo-oligomers, of which it validated 65,9% by homology. The remaining 34,1% may be interesting leads for homo-oligomers currently unidentified by SWISS-MODEL.

Novel aspect: CLAUDIO can automatically validate protein cross-links from large datasets and find new signals of homo-oligomers.


Using Crosslinking Mass Spectrometry to Understand the Structure of Proteins Implicated in Huntington’s Disease

Georgina Charlton1, Kostas Thalassinos1

1. University College London

Introduction: Huntington’s disease (HD) is an inherited neurodegenerative illness which causes the breakdown of nerve cells in the brain. HD is caused by CAG repeat expansion in the huntingtin (HTT) gene. Expression of the protein FAN1 is closely linked to a delay in age of onset and slower progression of HD. FAN1 is known to interact with the mismatch repair protein MLH1. MLH1 is essential for somatic expansion of the HTT CAG repeat, a process strongly linked to disease progression. FAN1 binding prevents MLH1 from interacting with the DNA repair protein complex MSH3-MSH2, leading to the stabilisation of CAG repeat. Large regions of FAN1 and MLH1 are unstructured and cannot be crystalised making structural determination by traditional methods difficult.

Methods: Interactome studies using FAN1 pull downs have given a better understanding of proteins which interact with FAN1. Crosslinking mass spectrometry was used to obtain a better understanding of FAN1 and MLH1’s structures. Using crosslinking mass spectrometry and Alpha Fold predictions we aim to get a better understanding of FAN1 and its binding sites to MLH1 and the protein complex formed between FAN1, MLH1 and PMS2.

Preliminary data: Using cell cultures of U20S cells interactome studies were performed, by enriching for FAN1 in FAN1 overexpressing cells to pull down FAN1 and its interactions and quantify them using label free quantification. FAN1 is thought to bind with MLH1 via the SPYF motif and MIM box in the N-terminal region of FAN1. Mutating the SPYF motif and MIM box, in the N-terminal region of FAN1, showed a decrease in the binding of FAN1 to MLH1 compared to the wild type. The Mutation of the S126 to D in the SPYF motif of FAN1, mimicking the effect of phosphorylation decreased the binding to MLH1, showed a decrease in the binding of FAN1 to MLH1, showing that the binding of FAN1 and MLH1 is phosphorylation regulated. Using on bead crosslinking mass spectrometry, we showed sites of interest on FAN1 and MLH1 which could directly interact with each other. The crosslinking data was then used in conjunction with Alpha Fold predictions to create structural models of the FAN1-MLH1 complex. We were able to show that MLH1 directly interacts with FAN1, and sites which could be important for this interaction. Further studies in this research will study the binding of MLH1 and MSH3, which promotes CAG repeat expansion leading to the progression of Huntington’s disease. A better understanding of how MLH1 binds to FAN1 and to MSH3 will give a better understanding of how Huntington’s disease progresses and allow the creation of an MLH1 mutant which binds to FAN1 but not MSH3 in the hopes of slowing down Huntington’s disease progression. The combination of XL-MS with AlphaFold allows us to obtain high-quality structural models for a large, flexible protein complex.

Novel aspect: On-bead crosslinking analysis of proteins associated with HD to better understand their structure.


Cross-linking mass spectrometry on P-glycoprotein

Gabriella Gellen1, Eva Klement2, Zsolt Bacso, Gitta Schlosser1

1 MTA-ELTE Lendület Ion Mobility Mass Spectrometry Research Group, ELTE Eötvös Loránd University, Institute of Chemistry
2 Biological Research Centre Szeged, Centre of Excellence of the European Union, Szeged, Hungary
3 Department of Biophysics and Cell Biology, University of Debrecen, Hungary

The ABC transporter P-glycoprotein (Pgp) has been found to be involved in multidrug resistance in tumor cells. Conformations of Pgp are extensively investigated, but these studies do not take into account the effect of lipids and cholesterol on Pgp, although they were shown to play important role in its conformations. In order to maintain the lipid environment, cross-linking mass spectrometry technology was applied to map Pgp’s structure. Experiments were carried out using different cross-linkers on living cells. After membrane protein extraction, complexes were enriched by means of monoclonal anti Pgp antibodies on magnetic beads, followed by on-bead enzymatic digestion. LC-MS/MS measurements were performed on an Orbitrap Fusion™ Lumos™ Tribrid™ Mass Spectrometer. Data were processed with Proteome Discoverer and Protein Prospector software tools. Results revealed protein-protein interactions of Pgp and proteins that had been known to be in proximity with Pgp. Identified monolinks hold information about solvent accessibility on Pgp while intraprotein cross-links complement 3D structure and aid detection of antibody binding sites.


Combination of XLMS with deep learning-based protein structure prediction: Protein structure, dynamics, processing and higher order assembly in Vaccinia virus

Paul Gershon1, Yeva Mirzakhanyan1, Andris Jankevics2, Richard A. Scheltema2

1. UC-Irvine;
2. Utrecht University, Utrecht, Netherlands

Introduction: How can we synthesize XLMS with protein structure prediction methods? Vaccinia is a large DNA virus packaging ~75 distinct gene products covering >100 fold abundance range, along with a phospholipid envelope and a 190,000 base pair double-stranded DNA genome. With Mr = ~3.26 GDa, it contains ~500 million atoms. However, its molecular architecture is essentially unknown. Being enveloped, asymmetric and polymorphic, Vaccinia is not amenable to X-ray crystallography. CryoEM has shown limited success. We have generated an XLMS dataset comprising ~135,000 CSMs via 10 distinct chemical crosslinkers. The dataset arose from >3,000 individual searches using 10 distinct search engines. With stringent thresholding the dataset condensed to ~22,000 unique crosslinked peptide pairs (with regard to accession, crosslink position and peptide Mr).

Methods: Virus was amplified and purified to proteomic purity. Trypsinization methods were found that yielded quantitative digestion. Crosslinking experiments were performed with 10 distinct crosslinkers after determining optimum crosslinking stoichiometry. Peptides were fractionated. Data were acquired on various spectrometers. Spectral data were used as input to 10 search engines with stringent thresholding of results. Search data were consolidated using in-house code. The ~75 packaged Vaccinia proteins were subjected to structure prediction using deep learning methods. Solvent accessible surface distances between crosslinked residues were calculated using TopoLink. Crosslinked protein structural models and assemblies thereof were visualized in ChimeraX.

Preliminary data: The XLMS dataset alone revealed some very clear protein networks. However, interpretation of the majority of these required comprehensive three-dimensional structural models in order to generate placeholders for crosslinked proteins and thereby transition from protein networks to molecular architecture. We considered deep learning structure prediction methods. Since many packaged Vaccinia proteins have no structural or sequence homologs outside of the poxviridae, we did not anticipate predictions likely to be highly accurate. However, after extensive statistical benchmarking of Vaccinia proteins we concluded that many structural prediction models likely showed sub-2 Angstrom accuracy. In the context of a structural model, XLMS data helped us elucidate the biology of a key virion structural protein, uncovering a concerted pathway of processing and conformational rearrangement that accompanies proteolytic processing, and virus assembly and maturation through several levels of structural hierarchy: Key processing steps appear to accompany transition from monomer to homotrimer, to hexamer-of-trimers to a regular tessellated mesh. The pathway appears to be accompanied by dynamic disulfide rearrangement. XLMS data helped validate predicted protein structures as well as informing the selection of alternatives at each step of the assembly pathway.

Novel aspect: Combination of deep learning and XLMS for the elucidation of protein molecular architecture and dynamics within a large DNA virus.


Going against the grain: development of denaturing mass photometry and its application for cross-linking reaction optimization

Hugo Gizardin-Fredon1,2, Oscar Hernandez-Alba1,2, Sarah Cianférani1,2

1 Laboratoire de Spectrométrie de Masse BioOrganique, IPHC UMR 7178, Université de Strasbourg, CNRS, 67000 Strasbourg, France
2 Infrastructure Nationale de Protéomique ProFI – FR2048, 67087 Strasbourg, France

Introduction: Mass photometry (MP) is a biophysical technique that recently gained interest in structural biology. It is a versatile, fast and low sample-consuming technique allowing to measure proteins mass distributions in native buffers. Despite its routine use to study native protein-protein interactions or multiprotein complexes oligmerization, MP workflows are still not adapted for the characterization of covalent assemblies. We report here on the development of a suitable and efficient MP protocol in denaturing conditions (called dMP). After evaluating its efficiency on reference multimeric proteins (BSA, ADH, GLDH), we applied dMP to fine-tune protein cross-linking (XL) reactions (impact of XL reagents, concentrations, buffers etc.) before MS analysis, and further benchmarked dMP against conventional 1D SDS-PAGE gel analysis.

Methods: Native stock solutions of proteins (bovine serum albumin BSA, Baker’s yeast alcohol dehydrogenase ADH and bovine heart L-Glutamate dehydrogenase GLDH, Sigma) were prepared at 1mg/mL in PBS. Protein samples for cross-linking were reacted with DSBU/DSAU (CF Plus Chemicals) and  PhoX (Bruker) at 25, 100, 400 molar excesses, and migrated on 1D SDS-PAGE gels. MP measurements were done in triplicates on a TWOMP (Refeyn Ltd) using AcquireMP software. Native MP was done by diluting samples to 10-40nM in an 18µL PBS droplet before analysis4. dMP measurements were done by first incubating the samples in urea 5.4M or guanidine 6M. Denatured samples were finally diluted to 10-40nM in an 18µL PBS droplet before MP analysis. Data were processed using DiscoverMP software.

Preliminary data: We first evaluated the impact of denaturing agents (types, concentrations) on the MP droplet stability and measurements quality by analyzing protein-free droplets. Results allowed selection of urea (5.4M) and guanidine (6M) as MP-compatible denaturing agents. We demonstrated that droplet concentrations of <0.8M urea or guanidine allowed efficient focusing and MP measurements of similar quality as those in PBS. To further optimize our dMP protocol, we next used reference proteins (BSA, ADH, GLDH): BSA served to assess mass precision and accuracy in dMP conditions while ADH and GLDH were used to further optimize denaturation conditions (denaturation duration and potential refolding assessment). Altogether, our results allowed obtaining a robust and efficient 2-step dMP protocol that unsure 95% of irreversible denaturation: 1) denaturation (5min, in 5.4M urea) followed by 2) addition of 2uL in an 18uL PBS droplet and immediate MP analysis. We finally investigated potentialities of dMP to monitor and optimize XL reactions in the frame of XL-MS workflows. The XL reaction is classically monitored using 1D SDS-PAGE gel, which has a series of limitation as being time-consuming, with low mass resolution, limited mass range, and biases in relative quantification of low abundant cross-linked species. When optimizing DSBU XL conditions for GLDH, dMP demonstrated that increasing DSBU molar excess from 25 to 400 expands from 25 to 77% the proportion of hexamer covalently stabilized. Dodecamer presence did not exceed 6% with no visible aggregation, giving a very similar picture to the native non-cross-linked GLDH (~70% hexamer). Moreover, we could quickly screen different XL reagents (PhoX, DSAU, DSBU), highlighting that their size (spacer length/flexibility) affects the oligomerization states that are stabilized upon XL reaction. Interestingly, dMP affords rapid and straightforward relative quantification of covalent versus non-covalent oligomers, along with simultaneous control of the formation of unwanted non-specific XL aggregation products. 

Novel aspect: Development of a straightforward MP protocol in denaturing conditions that outperforms 1D SDS-PAGE analysis for monitoring of chemical cross-linking reaction


Enhancing crosslinking reactions in (photosynthetic) membrane systems

Nicolas Francés(1,2), Sylvie Kieffer-Jaquinod (1), Anne-Marie Hesse (1), Gilles Curien (2), Myriam Ferro (1), Giovanni Finazzi (2), Pascal Albanese (1,2)

(1) BGE-EDyP, U13 / FR2048 UGA-CEA-Inserm-CNRS. CEA,17 Avenue des Martyrs, 38054 Grenoble, France. (2) PCV-LPM, UMR5168/UMR1417 UGA-CNRS-CEA-INRAe. CEA,17 Avenue des Martyrs, 38054 Grenoble, France.

Introduction: Membrane proteins play a vital role in maintaining cell homeostasis, and some are involved in crucial reactions in the energy transducing membranes of chloroplasts and mitochondria. The required steps of solubilization and purification make structural studies challenging, and crosslinking (XL) mass spectrometry (MS) has emerged as an alternative tool for studying their structure. PhoX˙¹ is an innovative reagent used to enrich XL-peptides with a phosphonic moiety, however, the yield is generally low due to negative repulsion with the membrane/protein surface (1). In order to address this problem, we propose that Trimethylphenylammonium chloride (TMPAC) could accumulate at the membrane interface (2), mask negative charges (3), and serve as an enhancer to bring PhoX closer to the membrane surface, thereby enabling it to react with the present lysine residues (4).

Methods: Experiments were conducted using Spinach (Spinacia oleracea) thylakoid membranes. The effects of TMPAC on the integrity of the membrane and the generation of crosslinks were evaluated using photosystem II (PSII) structure and biokinetic assays. The compound was shown to be a competent enhancer, increasing the number of crosslinks identified by 81%, without modifying the protocol for XL-peptides enrichment. These findings demonstrate the potential of TMPAC in combination with PhoX as a powerful tool to elucidate the conformation of membrane proteins.

Preliminary data: Our research revealed that the combination of TMPAC and PhoX enhances the yield of identified crosslinks by nearly one-fold, enabling greater accessibility of the crosslinking reagent without altering the physiological state of the system. Our 3D modeling analysis suggests that TMPAC induces novel crosslinks in protein regions with negative surface charges. Moreover, it was shown to have critical importance considering loose terminal regions when modeling complete 3D structures. Overall, TMPAC is a promising tool for advancing our understanding of membrane-bound proteins. It is commercially available, cost-effective, and highly efficient, facilitating crosslinking reactions in native membranes without generating artifacts and disruptions. Also, it shows a great display in combination with low crosslinking reagent concentrations. Nonetheless, further experimentation is required to explore its full potential for in vivo studies and optimal use in diverse experimental setups.



Roman Tuzhilkin1, Vladimír Ondruška1, Miroslav Šulc1

1 Department of Biochemistry, Faculty of Science, Charles University; Hlavova 2030/8, 128 00 Prague, Czech Republic

Introduction: Azurin is a small blue copper protein participating in redox reactions during anaerobic respiration in Pseudomonas aeruginosa and is also commonly used as a model for Electron Transfer (ET) processes or coordination sphere of metal ion in metalloproteins. Azurin naturally contains Cu(II/I) as central ion and is redox active for single electron ET. Moreover, azurin with no central ion (apo-azurin) is capable to bind other metal cofactors – e.g. Zn(II) – forming redox inactive Zn-form (main contamination of protein samples during their recombinant expression and purification). Existence of different azurin mutants, prepared by site directed mutagenesis, their metal-forms, and introduction of non-canonical amino acids or metal complexes to study ET emphasizes importance of more universal methods of purification and characterization.

Methods: Our study focuses: (a) on developing and optimizing methods of recombinant expression (e.g. introduction of non-canonical methionine analogues) and designing general purification protocol useful for more azurin variants, (b) finding approaches to characterize azurin samples with different central ion (mainly Zn content employing previously unreported application of basic native polyacrylamide gel electrophoresis (nPAGE)), and (c) oligomerization (employing chemical and photo induced cross-linking (XL) with mass spectrometry (MS) detection.

Results: Three member of non-natural methionine analogues were successfully introduced employing recombinant expression in E. coli and incorporation rate of each of them within azurin sequence was characterized by MS. The purification process was modified and protein acidic precipitation after cell disruption and metal titration with following SEC resulted in higher azurin yield and more flexibility and robustness. Apo/Cu(I/II)/Zn(II)-azurin ratio was successfully characterized utilizing nPAGE thanks to their differing pI value. To summarize this part, preparation of highly pure and unified azurin protein with metal-cofactor content characterization was achieved which in case of azurin ET studies is highly important because even minor contamination with Zn can be confusing and misguide received ET data interpretation. Moreover, azurin homooligomeric protein-protein interaction was studied employing DSS and DSG (amino-reactive homobifunctional XL) and the role of Lys 122 was confirmed by site-directed mutagenesis and MS analysis coupled with liquid chromatography. Azurin oligomerization was also studied employing methionine diazirine analogue and photo induced XL followed by MS analysis. Previously described role of C-terminal part within oligomerization surface was corroborated and azurin structure of oligomeric interaction surface was mapped on molecular resolution.

Novel aspect: Uncommon methodology approaches can impact research of commonly used azurin protein as ET model in another dimension.


Transfer learning to tune Prosit to predict intensities for cleavable cross-linked peptides

Mostafa Kalhor1, Dr. Mathias Wilhelm1

1. Technical University of Munich, Germany

Introduction: Chemical cross-linking mass spectrometry (XLMS) is an effective tool for analyzing protein structure and protein-protein interactions. However, due to the large search space, identifying cross-linked peptides remains a challenging task. Prior research has shown that incorporating fragment intensity information into the matching process can circumvent this problem. Here, we extend Prosit to predict the fragment ion intensities for cleavable cross-linked peptides using minimal data.

Methods: The main challenge is the limited number of training data. For this reason, we systematically evaluated methods to fine-tune a pre-trained model of Prosit to allow the prediction for unknown modifications using the concept of transfer learning. 

Results: To provide training and test datasets, XL datasets were analyzed using plink2 and XlinkX.  Until now, 170k MS2 and 37k MS3 spectra for cleavable cross-linked peptides were collected.Next, Prosit was fine-tuned with the available MS3 spectra covering cleavable cross-linked peptides. The final model achieves a very high accuracy (mean spectral angle of 0.81(R>0.9)). After that, we extended the structure of Prosit and fine-tuned it by CMS2 spectra (only cross-linked peptides with charge +4). Currently, our model achieves high accuracy (mean spectral angle of ~ 0.8 (R>0.9)) for prediction of the intensity values of b+, b++, y+, and y++ ions (normal and XL fragments. Next step is to fine-tune the Prosit with collected MS2 spectra of non-cleavable crosslinked peptides. We expect that integrating XL-Prosit into available XLMS search engines will improve the sensitivity and specificity of cross-linked peptide identifications.


Cleavable and non-cleavable cross-linking identification and quantification in the versatile OpenMS software ecosystem

Eugen Netz1,2, Ruben Grünberg1,2, Tjeerd M.H. Dijkstra3,4, Oliver Kohlbacher1,2,3

1 Applied Bioinformatics, Department of Computer Science, University of Tübingen, Sand 14, 72076 Tübingen,Germany
2 Institute for Bioinformatics and Medical Informatics, University of Tübingen, Sand 14, 72076 Tübingen, Germany3 Institute for Translational Bioinformatics, University Hospital Tübingen, Schaffhausenstraße 77, 72072 Tübingen,
4 Dept. for Women’s Health, University Clinic Tübingen, Calwerstraße 7, 72076 Tübingen, Germany

Introduction: Cross-Linking Mass Spectrometry (XL-MS) datasets are becoming larger and denser, as MS instruments continue to improve and more complex pipelines are being used. Especially MS cleavable cross-linkers have led to XL-MS studies on whole organelles or even entire cells, increasing search spaces in XL-MS identification. More complex samples also require more fractionation and result in larger amounts of raw MS data. To analyze all this data effectively, software pipelines with efficient code and versatile deployment possibilities from desktop computers to high performance computing cloud infrastructures are needed.

Methods: The OpenPepXL and OpenPepXLCleavable algorithms were developed as part of the OpenMS software ecosystem. OpenMS is an open source algorithm library and a set of tools for the analysis of proteomics and metabolomics mass spectrometry data. The previously described OpenPepXL algorithm for the identification of non-cleavable cross-linkers in mass spectrometry data was extended for the usage of MS-cleavable crosslinkers like DSBU and DSSO with tandem mass spectrometry.

Preliminary data: Both OpenPepXL and OpenPepXLCleavable can be combined with many tools in the OpenMS ecosystem and deployed in versatile ways. KNIME can be used to connect tools into workflows via visual programming by drag-and-dropping tools. This enables straightforward combination of processing steps tailored to a specific pipeline. Precursor monoisotopic mass correction, quantification and statistical analysis of results can be combined with standard protein and cross-link search algorithms. Python bindings for entire tools and internal algorithms enable fast and interactive prototyping for the development of new ideas. Compatibility with workflow languages like nextflow and containerization enables the development of reproducible and highly scalable workflows for high performance computing infrastructures. In the identification of MS-cleavable cross-links OpenPepXLCleavable performs at similar sensitivity and specificity as other state-of-the-art tools like MeroX. It shows fast runtimes and a small memory footprint. Different search modes can be used to enable fast runtimes even on weaker machines, with a small loss of sensitivity. To analyze a BSA dataset of 31,229 MS2 spectra against a complex entrapment database of 10000 protein sequences, OpenPepXLCleavable needed 40 min in its most sensitive mode and 4 min in its fastest mode on a single modern CPU core, while using less than 12 GB of memory (MeroX RiseUp needed 160 min and 40 GB memory and took 20 min in its proteome mode). It is roughly equivalent with MeroX on DSBU data with 223 BSA links and 6 incorrect links, compared to MeroX with 232 BSA links and 25 incorrect links on this dataset at a 5% FDR. OpenPepXLCleavable was more effective on DSSO data with 161 correct and 8 incorrect links, compared to MeroX with 150 correct and 36 incorrect links at a 5% FDR.

Novel aspect: OpenMS contains state-of-the-art XL-MS identification tools that can be deployed in versatile ways and combined with additional data analysis steps.


Spider-like complement inhibitor: structural insight into the C4b-binding protein

T. Kadavá1,2, J. F. Hevler1,2 , A. J. R. Heck1,2

1 Biomolecular Mass Spectrometry and Proteomics, Bijvoet Center for Biomolecular Research and Utrecht Institute for Pharmaceutical Sciences, University of Utrecht, Padualaan 8, 3584 CH Utrecht, the Netherlands
2 Netherlands Proteomics Center, Padualaan 8, 3584 CH Utrecht, the Netherlands

Introduction: The complement system is a highly conserved proteolytic cascade crucial for an innate immune response. Its overactivation leads to autoimmune diseases, and therefore it must be tightly regulated. The C4b‑binding protein (C4BP) is a key fluid phase complement inhibitor that interacts with C4b and C3b, mediating their inactivation. Human C4BP forms a higher-order structure consisting of three protein chains, C4BP alpha, C4BP beta, and Vitamin K-dependent protein S (ProS). These are present in the circulation in various ratios, forming three different stoichiometric variants (7α, 7α1β+ProS, 6α1β+ProS).

Methods: An integrative mass spectrometric (MS) approach was used to obtain a better understanding of the C4b‑binding protein dynamic and heterogeneous assembly. Specifically, cross-linking mass spectrometry (XL‑MS) with structural modeling was used to gain structural insight into the C4BP and the C4BP-C4b complex. Glycoproteomics combined with mass photometry and charge detection MS (CD-MS) was utilized to characterize the composition of the assembly.

Preliminary data: Here, we use an integrative MS approach to address various aspects of the C4b-binding protein. Combining XL-MS and structural modeling, we characterize the higher-order structure of C4BP. Specifically, our data provide insight into a) the oligomerization core assembled from the C-termini of C4BP alpha and beta subunits, b) the C4BP beta-ProS interaction mediated by complement control protein (CCP) domain I and Laminin-G like domain 2, c) the C4BP alpha interactions with C4b mediated by CCPI-IV. Using glycoproteomics, CD-MS, and mass photometry, we explored the composition of this heterogeneous spider-like protein complex.

Novel aspect: Providing structural and compositional information on prominent complement inhibitor C4b-binding protein and the C4BP-C4b complex.


Quantitative Cross-linking Mass Spectrometry Using Data-Independent Acquisition

Valerie Prochazkova1,2; Zdenek Kukacka1; Petr Novak1,2

1 Institute of Microbiology, the Czech Academy of Sciences, Prague, Czech Republic,
2 Faculty of Science, Charles University, Prague, Czech Republic

Introduction: Chemical cross-linking in combination with mass spectrometry (CXMS) has been developed into a powerful tool for mapping interaction networks and three-dimensional structures of proteins and their complexes. However, proteins are intrinsically dynamic, and they can form different conformations. Adding quantitative information to CXMS offers a unique opportunity to study flexibility and structural rearrangement of proteins. In this study, we report the benefits of utilizing data-independent acquisition and novel urea-based isotopically labeled cross‑linkers.

Methods: First, we used different mixtures (9:1, 1:1 and 1:9) of MS-cleavable DSPU and its isotopically labeled analogue to modify model proteins (BSA, BCA) testing quantitative potential of our strategy. Subsequently holo and apo forms of calmodulin and myoglobin were modified by isotopically labeled and non-labeled reagents to quantify the structural rearrangement upon calcium or heme binding, respectively. Both, the cross-link formation and quantification were performed in single data-independent experiment where tryptic peptides were measured in the broad-band mode with mass accuracy below 1 ppm and subsequently fragmented without isolating precursor ions at fixed collision energy resulting in sub-ppm accuracy for fragment ions as well. The raw data were converted into hybrid mgf file which was interpreted by MEROX search engine.

Preliminary Data: Acquired data on model proteins clearly demonstrate the potential of our quantitative cross‑linking strategy. The high mass accuracy in MS1 and MS2 modes enables unambiguous identification of cross-linked peptides. Moreover, the quantitative information is not derived from the MS1 experiment only. Due to the presence of isotopically labelled reporter ions in MS/MS spectra, it is possible to improve the qualitative and quantitative aspects of quantitative cross-linking experiments. Observed changes nicely overlap with high resolution structural models and previously published data. Our results lead to an assumption that presented data-independent acquisition method can be utilized for quantitative cross-linking experiments studying structure and dynamics of proteins and protein assemblies in solution.

Novel Aspect: A novel strategy for quantitative cross-linking mass spectrometry using novel isotopically labeled cross-linkers and data-independent acquisition


A human monoclonal antibody bivalently binding two different epitopes in streptococcal M protein mediates immune function

Wael Bahnan1, Lotta Happonen1, Hamed Khakzad2,3,†, Vibha Kumra Ahnlide1, Therese de Neergaard1, Sebastian Wrighton1, Oscar André1, Eleni Bratanis1, Di Tang1, Thomas Hellmark4, Lars Björck1, Oonagh Shannon1, Lars Malmström1, Johan Malmström1, Pontus Nordenfelt1,*

1 Division of Infection Medicine, Department of Clinical Sciences Lund, Faculty of Medicine, Lund University, Lund, Sweden
2 Equipe Signalisation Calcique et Infections Microbiennes, École Normale Supérieure Paris-Saclay, Gif-sur-Yvette, France
3 Institut National de la Santé et de la Recherche Médicale (INSERM) U1282, Gif-sur-Yvette, France
4 Department of Clinical Sciences Lund, Division of Nephrology, Lund University, Lund, Sweden

Introduction: Antibodies are essential components of immune system to combat intruding pathogens, like Group A streptococcus causing significant morbidity and mortality in human population. The streptococcal M protein, as a major virulence factor, counteracts the human immune response via various pathomechanisms, leading to huge challenge in the targeted antibody discovery and development. Here, we have generated anti-M protein antibodies derived from a healthy donor recently recovering from a GAS infection. A new type of interaction was later found in one of the monoclonal antibodies, designated as dual-Fab cis binding where two identical Fabs simultaneously bind to two distinct epitopes. This unique binding pattern turns out to be of importance by promoting a number of protective immune functions. 

Methods: Single B cell purification, bating, and isolation / B cell staining, baiting, and sorting / Reverse transcription, family identification, and cloning / General cell culture and transfection / Transfection, expression, and purification / Bacterial strains, growth, and transformation / NGS of bacterial genomes / Antibody screening and flow cytometry / Tissue microarray immunohistochemistry / Agglutination assays / SIM imaging / Binding curves / Cross-linking of antibody F(ab’)2 fragments to the M1 protein / Sample preparation for mass spectrometry (MS) / Liquid chromatography tandem mass spectrometry (LC–MS/MS) / Computational modeling / Fluorescent Xolair competition experiments / Imaging-based binding assays / Phagocytosis assay / NF-κB activity luciferase assay / Animal model

Preliminary data: Three human anti-M monoclonal antibodies were generated and characterized through our integrated workflow and extensive functional assays. Among these, only one antibody (Ab25) is found to mediate immune effector function. We further discovered and validated that Ab25 displays an unexpected mode of antigen–antibody interaction, which we termed as dual-Fab cis binding. Dual-Fab cis-binding antibodies engage by binding two nonidentical epitopes on its target antigen, which is the M protein in this study. Strikingly, compared to a single-Fab-binding antibody (Ab49) which shares one of Ab25’s two recognized epitopes, only the dual-Fab anti-body exhibits strong immune function. Except for offering an important perspective in the future development of antistreptococcal therapeutics, this discovery could give rise to a series of follow-up research questions related to evolution, mechanism, and functionality in the field of antibody-antigen interaction. 

Novel aspect: This study reveals dual-Fab binding existence and its influential role in mediating immune function against GAS infection.


Mass photometry reveals SARS-CoV-2 spike stabilisation to impede ACE2 binding through altered conformational dynamics

Burnap SAa, b & Struwe WBa, b

a Kavli Institute for Nanoscience Discovery, Dorothy Crowfoot Hodgkin Building, University of Oxford, Oxford, UK.
b Physical and Theoretical Chemistry Laboratory, Department of Chemistry, University of Oxford, Oxford, UK.

Introduction: The severe acute respiratory syndrome (SARS)-coronavirus (CoV)-2 is decorated with trimeric spike glycoproteins that mediate binding to host cells. The structural study of these spike complexes is hindered by proteolytic cleavage and inherent instability, resulting in the transition from a prefusion to the more stable postfusion state. The introduction of several proline substitutions within spike have enabled a dramatic stabilisation of the prefusion state, increasing protein yield, to act as a superior immunogen. Furthermore, the glycosylation state of spike greatly influences antibody recognition and is therefore of paramount importance for vaccine design. However, it remains unclear how stabilisation modulates glycan processing, and the extent to which these alterations influence spike function. 

Methods: A combination of mass photometry, glycomics and glycoproteomics were utilised to determine the influence of SARS-CoV-2 spike stabilisation upon glycan processing and function in the context of ACE2 binding capacity. Glycoengineering of in vitro produced spike proteins enabled the functional determination of specific glycan classes, alongside the comparison of the effect of stabilisation across SARS-CoV-2 variants.

Results: Utilising mass photometry we reveal how proline stabilisation of the SARS-CoV-2 spike protein directly alters ACE2 binding stoichiometry across both Wuhan-hu-1 and Omicron-B.1.1.529 strains. Differences in ACE2 binding were not driven by changes in glycan processing and receptor binding domain dynamics via glycan gating may depend solely on the presence/absence of glycans (macroheterogeneity) at these sites rather than changes in structure (microheterogeneity). It is worthy to note that these N-glycosylation sites are preserved among SARS-CoV-2 variants.  Our data quantitatively shows that 2P spikes sample open conformations to a greater extent than HexaPro and an increase in ACE2 binding via greater receptor binding domain accessibility, as confirmed through antibody binding. Researchers using stabilised spikes to study function or structure should consider that these alterations may not represent native spike dynamics found on the surface of the virus, as evidenced through altered ACE2 binding in our study.

Novelty: Our study highlights the power of mass photometry in discerning complex, oligomeric interactions between viral spike glycoproteins and host receptors. 


Defining the binding pattern of the disordered Hsp33 chaperone using an integrative structural approach

Rosi Fassler1, Oded Rimon1, Nurit Meyer1, Dina Schneidman2 Dana Reichmann1

Department of Biological Chemistry, Silberman Institute of Life Sciences, The Hebrew University of Jerusalem, Safra Campus Givat Ram, Jerusalem, Israel
School of Computer Science and Engineering, The Hebrew University of Jerusalem,
Jerusalem 9190401, Israel

Introduction: Cells constantly cope with environmental challenges, including oxidative stress, which has potentially fatal consequences on protein structure, function, and stability.  Therefore, it is not surprising that evolution armed cells with a system of stress-regulated chaperones which maintain proteome health during stress conditions. ATP-independent chaperones serve as the first line of defense against protein denaturation and aggregation by utilizing their structural plasticity induced by specific stress conditions. One such protein is the redox regulated-intrinsically disordered chaperone, Hsp33. When oxidized, Hsp33 undergoes redox-dependent unfolding essential for anti-aggregation activity. Unlike other proteins, Hsp33 must lose its structure to gain function. Here, we are focusing on defining the role of this unfolding for substrate binding.

Methods: Here we developed an integrative platform combining structural mass spectrometry techniques HDX-MS and XL-MS with computational modeling to define substrate promiscuity of Hsp33. We mapped the conformational changes in the Hsp33-client complex using HDX-MS in the Hsp33 wild type and variants harboring non-native sequences. HDX-MS allowed us to map conformational changes on the substrate and the chaperones induced by binding, involving structural and intrinsically disordered domains. As an alternative approach, XL-MS was applied using reagents of various lengths (DSBU, BS3, and DMTMM) and specificity, which allowed us to cover a wide range of protein conformations and spatial distance restraints. These constraints were used to derive a structural ensemble of chaperone-substrate complexes and investigate the chaperone plasticity rooted in its sequence.

Preliminary data: To understand the function, structure, and change, in the binding interface of Hsp33, we used the Hsp33-REV, Hsp33-STIL, and Hsp33-PGBD variants. Despite the large sequential modification, all Hsp33 variants were shown to function as well as the original in terms of binding, affinity, and recognition of clients. Most of the binding between Hsp33 and its substrate is concentrated around the very conserved and tightly folded N-terminal domain, with few single contact points at the end of the unfolded linker region and the disordered C-terminal domain. The Hsp33 variants, despite their extensive sequence modification, have a very similar binding pattern. However, they shift the contact points towards the non-native linker and disordered C-terminus while still conserving three binding hotspots and sustaining tight interaction with client protein. The variants also establish a unique binding hotspot, not found in wild-type interaction, on the C-terminal domain of the substrate citrate synthase, indicating flexibility not only in the chaperone’s biding pattern but also in the clients. The results display the high flexibility of Hsp33, which allows flawless redistributions of contact points that adopt more stable conformations despite the non-native sequence modification. This study maps the redox-regulated cascade of structural rearrangements and multiple states of Hsp33, which define the role of protein plasticity in substrate promiscuity required for chaperone activity and recognition of multiple targets during oxidative stress.

Novel aspect: We established a novel integrative methodology platform, which allows high-resolution mapping of protein dynamics and binding sites in structurally challenging proteins.


Integration of NMR spectroscopy and mass spectrometry for a new analytical workflow to characterize the oxidative stress in mAbs

Linda Cerofolini1,2, Enrico Ravera1,2,3,4, Christian Fischer7, Andrea Trovato5, Francesca Sacco1, Gabriella Angiuoni5, Marco Fragai1,2,3, Fabio Baroni5

1. Magnetic Resonance Centre (CERM), University of Florence, Via L. Sacconi 6, 50019 Sesto Fiorentino, Italy.
2. Consorzio Interuniversitario Risonanze Magnetiche di Metalloproteine (CIRMMP) Via L. Sacconi 6, 50019 Sesto Fiorentino, Italy.
3. Department of Chemistry “Ugo Schiff”, University of Florence, Via della Lastruccia 3, 50019, Sesto Fiorentino, Italy.
4. Florence Data Science, University of Florence, Italy
5. Analytical Development Biotech Department, Merck Serono S.p.a, an affiliate of Merck KGaA, Darmstadt, Germany. Via Luigi Einaudi 11, 00012 Guidonia Montecelio (RM), Italy.
6. Bruker BioSpin GmbH, Rudolf-Planck-Str. 23, 76275 Ettlingen, Germany

Introduction: The assessment of the Higher Order Structure (HOS) is a powerful methodology to characterize the structural features of biologics. Forced oxidative stress studies are routinely performed to investigate stability profile and monitor the critical quality attributes during the development of pharmaceutical formulations. To this aim, a multi-analytical approach combining NMR spectroscopy, mass spectrometry, differential scanning calorimetry, surface plasmon resonance, computational tools and bioassays has been employed to characterize the effects of forced oxidative stress on a monoclonal antibody. This integrated strategy has provided qualitative and semi-quantitative characterization of the samples and information at residue-level of the effects that oxidation has on the HOS of the tested mAb, correlating them to the loss of the biological activity.

Methods: Oxidative stress: Formulated antibody incubated at 0.1%H2O2 for 1h at room temperature. NMR spectroscopy: 1D-1H and 2D-1H13C methyl ALSOFAST HMQC with SIERRA filter on Bruker AVANCE-NEO spectrometer operating at 900 MHz equipped with a TCI cryo-probe. Statistical analysis performed using MestReNova (MestreLab). Reducing Peptide Mapping: LC-MS on a Orbitrap Fusion Lumos system in DDA mode. Data processing by Expressionist MS Refiner (Genedata). Differential Scanning Calorimetry: Nano differential scanning calorimeter (TA Instruments). Data elaboration using the two-state-scaled unfolding model (software NanoAnalyze). Cell-based Potency Assays: Luminometer (Infinite 200, Tecan). Data elaboration by the 4PL algorithm using Graphpad prism software. Binding affinity to FcRn: Biacore T200 system with NTA sensor chip (Cytiva). KD data elaborated using a “Two-state reaction” fitting model.

Preliminary Data: Upon oxidative stress, only the biological activity of the mAb’s Fc portion (investigated by Surface Plasmon Resonance in terms of binding to the FcRn receptor) was altered, showing a strong decrease of the affinity for the target receptor. No alteration of the Fab’s biological activity was observed. The NMR fingerprint of the methyl-groups region, coupled with a statistical analysis (PROFILE and ECHOS methods), indicated the preservation of the global folding of the mAb after oxidation, but in agreement with the results from reducing peptide mapping by LC-MS/MS, suggested a series of local, structural modifications responsible for the decreased activity. The highest oxidation levels were indeed observed in case of the four methionines located in the CH2 and CH3 domain (Fc portion) of the mAb: Met-252, Met-397, Met-428 and Met-458. A structural model (generated with Alpha-fold) allowed to observe that Met-252 and Met-428 are close to the interface between the CH2 and CH3 domains of the Fc portion, that is crucial for the interaction with the FcRn receptor, while Met-397 and Met-458 are away from the FcRn binding site and are not involved in the stabilization of the CH2-CH3 interface. As a consequence, oxidation of this two methionines does not appear to impact the binding activity. Obtained results were in agreement with Nano DSC data: only the melting temperatures corresponding to the CH2 and CH3 domain of the molecule were modified after oxidative stress.

Novel Aspect: NMR analysis integrated in a traditional analytical workflow grants a detailed characterization of minor stress-related structural changes affecting biological functions.


Receptor kinase signaling of BRI1 and SIRK1 is tightly balanced by their interactomes as revealed from domain-swap chimera in AE-MS

Lin Xi 1, Xu Na Wu 1,2, Jiahui Wang 1, Zhaoxia Zhang 1, Mingjie He 1,3, Zeeshan Zeeshan 1, Thorsten Stefan 1, Waltraud X Schulze 1*

1. Department of Plant Systems Biology, University of Hohenheim, Stuttgart, Germany
2. State Key Laboratory of Conservation and Utilization of Bio-Resources in Yunnan and Center for Life Science, School of Life Sciences, Yunnan University, Kunming, China
3. Department of Biology, Duke University, Durham, NC, USA

Introduction: Plants response to different external and internal stimulation by activation of signal transduction pathways. Receptor kinases are localized at plasma membrane and were involved in perception of signals at the cell surface. Here we use affinity enrichment mass spectrometry acquisition (AE-MS) of the LRR receptor kinases BRI1 and SIRK1 to study the stimulus-dependent interactomes in response to brassinolide(BL) and/or sucrose. Our results reveal the different recruitment ability of BRI1 and SIRK1 under BL and sucrose treatment. By using domain-swap chimera, along with machine learning approach, we attribute structural features of the receptors to the interaction with their co-receptors or substrate transporters. Our work reveals a tightly controlled balance of signaling cascade activation dependent on the internal status of the plant.

Methods: The target receptor like kinases (RLKs) were fused with GFP and stable transformed into Arabidopsis. The seedlings were cultured hydroponically, and supplied sucrose or BL before the root-sample harvesting. The microsomal fractions were extracted for AE-MS (Keilhauer et al. 2015, PMID: 25363814). For each bait protein, sucrose starvation condition was used as control. The treatments-inducible interactome was built for each bait protein. To establish data analysis workflow, Maxquant and Persus were used. Domain-swap RLKs were further constructed and carried out the same AE-MS procedure. Machine learning methods were applied and recruitment features of certain domains could be annotated. Finally, in order to test the effect on growth by chimera RLKs, the phenotype complimentary experiment was performed.

Preliminary data: 1) The workflow for treatments-inducible interactomes have been built; 2) Co-receptors recruitment behavior have been collected for SIRK1 and BRI1; 3) Domain-swap between SIRK1 and BRI1 were constructed and their AE-MS have been performed; 4) Using machine learning, the recruitment feature for each domain was annotated to SIRK1 and BRI1.

Novel aspect: in vivo pull-down mass-spectrometry for studying plants receptor like kinases; Using mathematic models to assist in annotating features of domains.


Characterizing the full length Npl4-Ufd1 complex and interaction with Cdc48 through an interface residue switch

Tal Oppenheim1, Meytal Radzinski1, Merav Braitbard1, Eliya Goldberger1, Tommer Ravid1, Dina Schneidman-Duhovny2, Dana Reichmann

1. The Hebrew University of Jerusalem, Department of Biological Chemistry
2. The Hebrew University of Jerusalem, School of Computer Science and Engineering

Introduction:A core element of protein quality control in eukaryotes is Cdc48 (VCP/p97), a highly conserved ~600Da AAA-ATPase. Cdc48 together with its co-factors, Npl4 and Ufd1 (UN) recognizes, extracts, and delivers ubiquitinated proteins to the proteasome for final degradation. Mutations in the Cdc48 homologue are associated with many human pathologies. While well-established in-vivo studies have defined the interaction between Npl4-Ufd1 and Cdc48 as essential for cellular viability, the molecular basis and full-length structures of Cdc48-Npl4-Ufd1 is not yet characterized. This is mainly due to the significant flexibility of the main interaction domain of Cdc48 and Npl4, as well as the native disorder of Ufd1. Here we introduced an integrative approach to defining the structural ensemble and dynamics of the Cdc48-UN complex.

Methods: Here we introduce a powerful integrative approach that combined XL-MS and computational modeling to define the structural ensemble of the interaction between highly flexible regions of Cdc48 and the Npl4-Ufd1 complex. We utilized zero-length DMTMM or cleavable DSBUcrosslinkers followed by stringent crosslinking analysis using MeroX and manual peptide validation. For integrative modeling, a multimolecular assembly was used, via a pairwise docking approach, through XlinkAssembler (based on the combinatorial docking algorithm CombDock, modified to support XL-MS). We succeeded in deriving an extensive inter-residue map to compile an atomic model of the Npl4-Ufd1 and Cdc48-Npl4-Ufd1 complexes, showing both the dynamics and stabilization of the Npl4-Ufd1 complex upon binding to Cdc48, and identifying a binding switch residue of the Cdc48-Npl4 interaction. 

Preliminary data: Understanding the full-length architecture of the Cdc48-Ufd1-Npl4 complex is challenging due to the conformational changes in the ternary complex. Cdc48 has a partially disordered N-terminal domain, which moves relative to the ATPase domains following ATP binding. On the other hand, Npl4 includes a flexible N-terminal domain and Ufd1 is intrinsically disordered. The XL-MS interaction map shows that Npl4 and Ufd1 have multiple interaction sites, located throughout the length of both proteins, enabling to derive the structural ensemble of the Npl4-Ufd1 complex. Specifically, we found cross-reactivity of well-conserved Npl4 and Ufd1 domains (e.g., Ufd1-N with UBXL and Zn-finger domains of Npl4; Ufd1-C with SHP domains of Npl4). Importantly, we found that Cdc48 significantly changes the dynamics and inter-residue interactions of the Npl4-Ufd1 complex, through interaction between Cdc48 and mainly Npl4. The structural model of the Cdc48-Npl4-Ufd1 complex reveals that Npl4 recognizes Cdc48 via its UBXL domain as well as the zinc finger domain. An unstructured region of Npl4 as defined by AlphaFold and FIDPnn could serve as a flexible arm, allowing the UBXL domain to interact differently upon binding to different proteins, including Cdc48. This suggests that Npl4 might have multiple states of interactions with the partially disordered Ufd1 protein, enabling high-affinity interactions with other proteins, such as Cdc48. To validate the model, we identified a highly conserved cysteine at the Cdc48-Npl4 binding interface which is central to the stability of the Cdc48-Npl4-Ufd1 complex. Molecular dynamics validated the role of this residue on the binding interface architecture and stability. Mutation of this cysteine leads to a decrease in cellular growth and protein quality control in yeast.  Our results provide insight into the architecture of the Cdc48-Npl4-Ufd1 complex as well as the consequences of its site-directed inhibition in cells.

Novelty: Using the integrative structural mass spectrometry approach we propose a model for altering Cdc48 activity, which has a therapeutic potential.


Simultaneous antibody sequencing and epitope mapping by integrated cryo electron microscopy and mass spectrometry

Marta Šiborová, Douwe Schulte, Joost Snijder

Biomolecular Mass Spectrometry and Proteomics, Bijvoet Center for Biomolecular Research, Utrecht University, Padualaan 8, 3584CH Utrecht, The Netherlands.

Introduction: Antibody sequences are typically recovered by sequencing the coding mRNAs from the producing B cells. This approach is the state of the art for throughput and sequencing accuracy, but still provides a limited view of the antibody repertoire on the level of the functionally secreted polypeptide product in bodily fluid. Recent advances in de novo antibody sequencing by MS have made it possible to derive antibody sequence information directly from the secreted polypeptide product. Here we explore the additional use of novel de novo model building algorithms for cryoEM density that generate complementary sequence information of reconstructed antibodies on a level akin to top-down LC-MS/MS.

Methods: A benchmark of publicly available cryoEM density maps of viral antigen-antibody complexes (SARS-CoV-2, HIV-1, RSV, and ebolavirus) is used as input for ModelAngelo for de novo model building. The de novo sequences from ModelAngelo are used as input for the in-house developed template-based assembly software Stitch. The sequence accuracy is evaluated by computing the distance to the known sequences and evaluating assignment to the correct germline precursor of the variable segment. Experimental cryoEM maps of Herceptin-HER2 and CR3022-Spike complexes are integrated with experimental de novo antibody sequencing data by bottom-up mass spectrometry in Stitch. In addition, cryoEM maps of polyclonal antibody mixtures from Electron Microscopy based Polyclonal Epitope Mapping (EMPEM) experiments are analyzed in the same pipeline to evaluate if the map quality allows for antibody sequence profiling.

Preliminary data: Experimentally derived cryoEM maps of antibody-antigen complexes allow for de novo sequence modelling at accuracies on the order of 80-90% for maps up to ca. 4 Angstrom nominal resolution. Compared to bottom-up MS data, cryoEM-based sequence accuracy is lower, but the read-length far exceeds the average 20-30 amino acids observed in MS experiments, yielding accurate V-gene assignments despite low sequencing accuracy. We show that these long cryoEM derived sequence reads serve as useful templates for assembly of shorter MS-based de novo sequence reads. Finally, we demonstrate that de novo sequence reads of high quality can be extracted from cryoEM maps from EMPEM experiments.

Novel aspects: We developed a workflow that allows for simultaneous sequence profiling and epitope mapping of antigen specific antibodies.


Mutational analysis of SARS-CoV-2 N-protein by structural mass spectrometry

Jonas Schröder1,3, Timothy Soh1,2,3, Jens Bosse1,2,3,4, Charlotte Uetrecht1,3,5

1 CSSB Centre for Structural Systems Biology, Hamburg, Germany
2 Department of Virology, Hannover Medical School, Hannover,Germany
3 Leibniz Institute of Virology (LIV), Hamburg, Germany
4 Cluster of Excellence RESIST (EXC 2155), Hannover Medical School, Hannover, Germany
5 Deutsches Elektronen Synchrotron DESY, Hamburg & University of Siegen, Siegen, Germany

The multifunctional and highly abundant N-protein (Nucleocapsid) of SARS-CoV-2 is crucial for several processes during infection starting with protection and packaging of viral RNA but also replication, suppression of immune response or virus assembly. Although the mutation rate is lower compared to the spike protein, N also changes across variants indicating an adaption to the host under immune pressure. In this context, we want to identify host proteins that are involved in protein-protein-interactions (PPIs) with the N protein in a spatiotemporal manner and investigate possible changes in the N-interactome across the mutants. Here, distinct mutations may alter preferred binding partners and assembly states with an effect on the whole infection cycle.

Based on computational predictions of possible N-specific host protein complexes using AlphaFold-Multimer, we selected candidates binding to mutation-prone regions of N or occupying common binding sites for expression and purification. Taking into account different proteoforms including post-translational modifications (PTMs), we purify proteins from mammalian cells and confirm complex formation by native mass spectrometry.


MS SPIDOC: mass spectrometry meets single particle imaging

Steven Daly1, Thomas Kierspel2,3,4,5, Alan Kádek4,5,6, Kristina Lotenzen4, Jan Commandeur1, Charlotte Uetrecht2,3,4,5,7

1 MS Vision,Televisieweg 40, 1322 AM Almere, Netherlands
2 Deutsches Elektron-Synchrotron DESY, Notkestraße 85, 22607 Hamburg, Germany
3 Centre for Structural Systems Biology CSSB, Notkestraße 85, Building 15, 22607 Hamburg, Germany
4 European XFEL GmbH, Holzkoppel 4, 22869 Schenefeld, Germany
5 Leibniz Institute for Experimental Virology (LIV), Martinistraße 52, 20251 Hamburg, Germany
6 BIOCEV – Institute of Microbiology CAS, Průmyslová 595, 252 50 Vestec, Czech Republic
7 School of Life Sciences, University of Siegen, Adolf-Reichwein-Str. 2, 57068 Siegen

Native mass spectrometry (MS) enables the ionization and transfer of structurally intact non-covalent protein complexes into the gas-phase. This makes it a perfect tool to study protein assembly from single proteins, through intermediate structures to the full complex in a mass and conformation specific manner. However, the structural information that can be gained with techniques like top-down MS or ion mobility is limited. Accordingly, other experimental approaches such as X-ray diffractive imaging are necessary to get a full understanding of the proteins and their assemblies.

MS SPIDOC (Mass Spectrometry for Single-Particle Imaging of Dipole Oriented Protein Complexes) is a Horizon 2020 funded research and innovation program which aims to combine native mass spectrometry and x-ray diffractive imaging. In particular, well established methods from MS like m/z selection, ion trapping or ion mobility are adapted as part of the sample delivery system for X-ray diffraction.

In contrast to conventional diffractive imaging of crystallized proteins, the proteins here are delivered as single particle without the need for crystallization. Naturally, this leads to an increase in the required brightness of the X-ray source. Thus, single-particle Xray diffractive imaging (SPI) is only conducted at X-ray free electron lasers [1], the brightest X-ray sources in the world. This contribution will highlight the ongoing efforts of the MS SPIDOC consortium to develop this sample delivery system for the use at beamlines of the European XFEL. The current state of the designing and manufacturing of the instrument prototype will be presented as well as the results of the first testing of individual component modules.

Event Timeslots (1)

ictadmin ictadmin