Efficient Ligand Discovery Using Sulfur(VI) Fluoride Reactive Fragments

Sulfur(VI) fluorides (SFs) have emerged as valuable electrophiles for the design of “beyond-cysteine” covalent inhibitors and offer potential for expansion of the liganded proteome. Since SFs target a broad range of nucleophilic amino acids, they deliver an approach for the covalent modification of proteins without requirement for a proximal cysteine residue. Further to this, libraries of reactive fragments present an innovative approach for the discovery of ligands and tools for proteins of interest by leveraging a breadth of mass spectrometry analytical approaches. Herein, we report a screening approach that exploits the unique properties of SFs for this purpose. Libraries of SF-containing reactive fragments were synthesized, and a direct-to-biology workflow was taken to efficiently identify hit compounds for CAII and BCL6. The most promising hits were further characterized to establish the site(s) of covalent modification, modification kinetics, and target engagement in cells. Crystallography was used to gain a detailed molecular understanding of how these reactive fragments bind to their target. It is anticipated that this screening protocol can be used for the accelerated discovery of “beyond-cysteine” covalent inhibitors.


S7
Figure S10.Two-step binding mechanism for irreversible electrophilic protein modifiers and equations ( 1) and ( 2) for the calculation of kinetic parameters.displaying k inact , K I , and hence k inact /K I -a parameter to describe the overall modification efficiencies of SFs 2f-i.

2.1.
Solvents; reagents; consumables; materials Solvents were anhydrous and reagents purchased from commercial suppliers were used as received.
LC-MS with acidic modifier (low pH method): Solvent A (0.1% v/v solution of formic acid in water) and solvent B (0.1% v/v solution of formic acid in acetonitrile).Sample was eluted with a flow rate of 1.0 mL/min using the following gradient: Table S1.Low pH gradient for small molecule LC-MS analysis Time / min Solvent A (%) Solvent B (%) 0 97 3 1. 5  5 95 1. 9  5 95 2.0 97 3 LC-MS with basic modifier (high pH method): Solvent A (0.1% v/v 10 mM ammonium bicarbonate in water adjusted to pH 10 with ammonia solution) and solvent B (0.1% v/v ammonia in acetonitrile).Sample was eluted with a flow rate of 1.0 mL/min using the following gradient:  MDAP with basic modifier (high pH method): High pH: XSelect C18 column (100 mm × 19.0 mm, 5.00 µm packing diameter, 20.0 mL/min flow rate) using a gradient elution at ambient temperature using mobile phases of water with 0.1% 10 mM ammonium bicarbonate by volume (v/v) adjusted to pH 10 with ammonia solution and 0.1% v/v ammonia in acetonitrile.Flow rate: 40 mL/min.

S11
The gradient of acetonitrile required to elute product was determined by the LC-MS retention time of the desired material.The following methods were selected dependent on the retention time of desired material: NMR spectroscopy was carried out at ambient temperature using standard pulse methods on a Bruker AVII600 (1H = 400 and 600 MHz) in DMSO-d 6 and referenced to residual undeuterated solvent.Multiplicity is reported as follows: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet etc.All spin-spin coupling constants (J) are reported in hertz to the nearest 0.1 Hz.

Centrifuge
Plates were centrifuged using a Sorvall Legend RT (401198833) model.

Intact protein LC-MS
Intact protein masses were recorded by LC-MS using an Agilent G230B time-of-flight (ToF) Accurate Mass Series mass spectrometer, interfaced with an Agilent 1290 infinity II series column oven (G7116B) and an Agilent 1290 infinity II series liquid chromatography high speed binary pump (G7120A).Protein samples were injected using an Agilent 1290 infinity II series multisampler with dual needles (G7167B) with a 0.5-10 µL injection volume and at a temperature of 4-20 °C.Chromatography was carried out on an Agilent Bio-HPLC PLRP-S (1000 Å, 5 µm × 50 mm × 1.0 mm, PL1312-1502) reverse phase HPLC column at 70 °C.The sample was eluted at 0.5 mL/min using a gradient system from Solvent A (water, 0.2% (v/v) formic acid) to Solvent B (acetonitrile, 0.2% (v/v) formic acid) according to the following conditions:  21  20   The eluent was injected directly into an Agilent ToF mass spectrometer (G6230B) using a dual AJS ESI source and scanning between 600-3200 Da with a scan rate of 1.2 s in positive mode.The following MS parameters were used: capillary voltage limit: 4000 V; desolvation temperature: 350 °C; drying gas flow: 10 L/min.Data acquisition was carried out in 2 GHz Extended Dynamic range mode.Spectra were processed using Mass Hunter Qualitative Analysis™ B06.00 (Agilent) with the Maximum Entropy method employed.The total ion chromatograms (TIC) were extracted (region containing protein) and the summed scans were deconvoluted (using a maximum entropy algorithm) over a m/z range with an expected mass range dependent on the protein: The deconvoluted spectra were exported as csv files and analyzed using R Studio (Version 3.6.3)software to generate PDF files of the spectra. 1 The median of the protein only controls were subtracted from the sample spectra to remove baseline signal.The peak height for unmodified protein and protein modified by a reactive fragment were recorded and used to calculate percentage modification using equation ( 3): % = ((intensity of modified protein)/(intensity of protein only+(intensity of modified protein)))*100 (3)

Selection of 352 amine functionalized fragments
Selection was based on the molecular clustering approach used for the PhABit libraries. 2,3A diverse set of compounds containing only one aliphatic amine group was considered.The starting set for the selection was the solution sample store of the GSK compound collection.However, compounds were also checked for the availability in eMolecules and Enamine building block databases.The selection was then carried out using Pipeline Pilot (Version 20.1.0.2208) software. 4Initial selection criteria were applied: GSK compounds having at least one sample with a concentration of 10 mM in DMSO or above, and with at least 150 μL available.Unstable compounds and compounds with other liabilities were removed using proprietary GSK filters.Further selection criteria were then applied: aromatic ring count≤2; HBDs/HBAs≤4; heavy atoms≤15; 150<M W ≤250; BioByte cLogP<9; 1 aliphatic amine group. 5The structures were tagged based on 6 aliphatic amine types, differentiating based on primary/secondary, hindered/non-hindered and cyclic/non-cyclic.The amines were also tagged by the calculated pK a using ChemAxon pK a . 6All amines having acidic pK a <7 and secondary hindered aliphatic amines were excluded.A ChemAxon LibMCS clustering was performed on the remaining molecules. 7ince the aim of the current selection was to choose compounds which can have analogues, all compounds were excluded which did not belong to a cluster with at least 4 members.A diverse selection was then performed on the remaining molecules using the Diverse Molecules component from Pipeline Pilot and ECFP4 fingerprinting. 8A 352-membered library was then selected from the remaining molecules.The selected amines (10 mM, 5 µL per well/screen) were ordered from GSK's solution stores in Labcyte 384LDV plates:

HTC-D2B protocol
To a Labcyte 384LDV plate containing 352 amine functionalized fragments (10 mM) in DMSO (5 µL per well) was added a stock solution of OSu 2a (10 mM) and NEM (30 mM) in DMSO (5 µL per well).The plate was sealed, centrifuged (1 min, 1000 rpm), and allowed to sit at room temperature for 1 h.After the reaction (assumed product concentration: 5 mM), a Labcyte ECHO © 555 Liquid Handler was used to transfer the library of SFs to a Greiner 384 white low volume plate.Purified protein diluted from original stock into a buffer made up of HEPES (pH 7.5, 25 mM) and NaCl (50 mM) in distilled MS-grade water (15 µL per well) was subsequently added across the plate.The plate was sealed, centrifuged (1 min, 1000 rpm), incubated for 24 h, and subsequently analyzed by intact protein LC-MS.
An analogous protocol was also carried out using OSu 3a for the initial screen against CAII, and the screening conditions for the panel of purified proteins were as follows: Table S7.Screening conditions used for the HTC-D2B protocol against a range of purified proteins. .Samples were digested with trypsin (Promega) 1:10 E:S (37 °C, 16 h) in 100 mM ammonium bicarbonate.After removal of the supernatant, peptides were extracted using acetonitrile.Combined supernatants were concentrated in a SpeedVac centrifuge and acidified (0.1% formic acid, 0.05% trifluoroacetic acid) prior to injection into the LC-MS/MS system.

Data analysis:
Uninterpreted tandem MS spectra were searched for peptide matches against the sequence for CAII using the Mascot (Version 2.6.0)software with a 5 ppm mass tolerance for peptide precursors and 20 mDa mass tolerance for fragment ions. 9Raw files were searched using trypsin as the enzyme with up to 2 missed cleavages and the variable modifications carbamidomethylation on cysteine and oxidation on methionine were allowed.Masses corresponding to [SF−HF] were allowed as variable modification(s) on cysteine, histidine, lysine, tyrosine, serine, and threonine as well as the protein N-terminus.MS/MS spectra were manually validated and annotated.

Virtual docking of hit compounds
Virtual molecular docking of SF 2e (or 2d) was carried out in Molecular Operating Environment (MOE) (Version 2019.0101). 10PDB file 3CAJ (human CAII complexed with ethoxzolamide) was used.The docking tool was used to map SF 2e (or 2d) in the Zn 2+ binding site

S23
using ethoxzolamide as a template.The methods used were Placement: Triangle Matcher; Score: London dG and Refinement: Rigid Receptor; Score: GBVI/WSA dG.

5.1.
Protocol for kinetic analyzes (CAII) The (resynthesized and purified) SF hits (2b-e) were plated into seven Greiner 384 white low volume plate from a 10 mM source in DMSO.Differing volumes were added such that the final concentrations for each compound would be: 100, 50, 20, 10, or 5 µM after adding 15 µL of protein stock solution per well.Volumes of DMSO were made to be consistent for every well.Purified CAII diluted from original stock into a buffer made up of HEPES (pH 7.5, 25 mM) and NaCl (50 mM) in distilled MS-grade water (15 µL per well) was subsequently added across the plates.Final concentrations: 0.5 µM protein; 100, 50, 20, 10, or 5 µM SF.The plates were sealed, centrifuged (1 min, 1000 rpm), and the first six plates were immediately queued for analysis by intact protein LC-MS at a temperature of 20 °C with the final plate paused for a 24 h timepoint.The resulting modification yields were plotted in GraphPad Prism (Version 5.0.4) software against the time of sampling. 11Non-linear regression analyzes were conducted using a 'one-phase association' model, with the y=0 value constrained to '0' and plateau value constrained to '100'.The observed rate constants (k obs , value 'K' in GraphPad) were extracted for each compound at each concentration, then replotted against the corresponding compound concentration.Further non-linear regression analyzes were conducted using a 'Michaelis-Menten' model, from which k inact values ('Vmax' in GraphPad) and K I values ('Km') were extracted.

Protocol for iterative screen against CAII
Amine selection: Amines which were close analogues of initial hit SFs 2b-e were selected.Initial selection criteria were applied: GSK compounds having at least one sample with concentration 10 mM in DMSO or above and at least 50 μL sample size.Unstable compounds and compounds with other liabilities were removed using proprietary GSK filters.Further selection criteria were then applied: 150<M W <350, BioByte cLogP<9 and only 1 aliphatic amine group. 596 Amine-functionalized fragments were selected from the GSK compound collection based on Tanimoto similarity to original hits, and these comprised 79 primary sulfonamides and 17 secondary/tertiary sulfonamides or aryl sulfones.

HTC-D2B iterative screen against CAII:
To a Labcyte 384LDV plate containing 96 amine functionalized fragments (10 mM) in DMSO (5 µL per well) was added a stock solution of OSu 2a (10 mM) and NEM (30 mM) in DMSO (5 µL per well).The plate was sealed, centrifuged (1 min, 1000 rpm), and allowed to sit at room temperature for 1 h.After the reaction (assumed product concentration: 5 mM), a Labcyte ECHO © 555 Liquid Handler was used to transfer the library of SFs to a Greiner 384 white low volume plate.Purified BCL6 diluted from original stock into a buffer made up of HEPES (pH 7.5, 25 mM) and NaCl (50 mM) in distilled MS-grade water (15 µL per well) was subsequently added across the plate.Final concentrations: 0.5 µM protein; 20 µM SF.The plate was sealed, centrifuged (1 min, 1000 rpm), incubated at 20 °C for 24 h, and subsequently analyzed by intact protein LC-MS.

Single shot screen of chemoproteomic probes against CAII
The (synthesized and purified) active probes 2j, and 2k, as well as negative control 2l, were plated into a Greiner 384 white low volume plate from a 10 mM source in DMSO.Purified CAII diluted from original stock into a buffer made up of HEPES (pH 7.5, 25 mM) and NaCl (50 mM) in distilled MS-grade water (15 µL per well) was subsequently added across the plate.Final concentrations: 1 µM protein; 50 µM SF.The plate was sealed, centrifuged (1 min, 1000 rpm), incubated at 4 °C for 24 h, and subsequently analyzed by intact protein LC-MS.Modification yields were measured as 78% for 2j and 95% for 2k.Negligible modification was observed for 2l.

Chemoproteomics workflow
Acetylation of NeutrAvidin agarose slurry: NeutrAvidin agarose slurry (10 mL) (Thermo Fisher Scientific, 29204) was centrifuged (2 min, 2000 rpm) prior to removal of the supernatant.The beads were washed three times with PBS.PBS (9 mL) and Sulfo-NHS-acetate (482 μL, 400 mM in DMSO) (Thermo Fisher Scientific, 26777) were added, the beads were incubated at room temperature for 30 min on a falcon tube roller, and then centrifuged (2 min, 2000 rpm).Supernatant was removed and the incubation step with freshly made NHS-acetate was repeated.The reaction was quenched by adding Tris (2 mL, 1 M, pH 7.5).The beads were washed once with PBS and twice with 20% EtOH.20% EtOH (5 mL) was added, and the beads were stored at 4 °C.
Chemoproteomics workflow: HEK293T cell line (female human origin, The Francis Crick Institute cell service) was used for this study.The cells were maintained at 37 °C with 5% CO 2 in DMEM media supplemented with 10% fetal bovine serum and 1% L-Glutamine-Penicillin-Streptomycin solution (200 mM L-glutamine, 10,000 U/mL penicillin and 10 mg/mL streptomycin).
Proteins were precipitated using ice-cold acetone and the resulting pellets were washed twice with ice-cold 80% acetone.The air-dried pellets were dissolved in 0.2% SDS (400 µL) in HEPES (50 mM, pH 8.0) by vortexing and sonicating.
The peptide samples were cleaned-up using C18 96-well plate (BioPureSPE Macro 96-Well, 100 mg PROTO C18, The Nest Group).The wells were conditioned with acetonitrile (300 μL) (centrifuged for 1 min at 50 g) and equilibrated twice with 0.1% TFA (300 μL) (centrifuged for 1 min at 150 g).0.1% TFA (100 μL) was added to digest samples, and these were subsequently loaded on the plate (centrifuged for 1 min at 150 g).The samples were washed twice with 0.1% TFA (200 μL) (centrifuged for 1 min at 150 g).The plate was centrifuged once more at 200 g for 1 min.The peptides were eluted in two steps with 0.1% TFA (150 μL) in 50% acetonitrile (centrifuged for 1 min at 200 g) into a collection plate.The plate was frozen, and the samples were dried in Labconco CentriVap Benchtop Vacuum Concentrator at 35 °C.

LC-MS/MS analysis
Peptides were redissolved in 0.1% formic acid in water and samples were loaded with iRT standard (Biognosys AG) onto Evotips (as prepared according to manufacturer's instructions, EV2001) followed by loading onto the Evosep One LC system in front of the Orbitrap Fusion Lumos (Thermo Fisher Scientific).The Evosep One was fitted with a 15 cm column (EV1113) and the predefined method for a 44 min run was employed.Data for all samples was acquired in Data Independent Acquisition mode (DIA

Data analysis
The DDA and DIA data was searched using Pulsar search engine inside Spectronaut (Version 14.10.201222.47784). 12A spectral library was first generated by searching the DDA and DIA data against the Homo Sapiens (August 2019), common contaminants and avidin fasta files. 12BGS factory settings (default) were used, except no fixed modifications were selected.The library contained 12860 precursors that correspond to 10695 peptides from 2163 protein groups.
The DIA data was then searched against the generated library using BGS factory settings (default).The data was normalized using global average normalization strategy with automatic row selection.Run wise imputation (Q-value = 30%) was applied to the data set.The sample t-test was carried out in Spectronaut to assess differential abundances (probe/DMSO, probe/competition and probe/negative control). 12ta was exported and the following thresholds were used for statistical significance: Data visualization of exported candidates table was carried out in GraphPad Prism (Version 5.0.4) software. 116.5.

Data availability
The raw mass spectrometry proteomics files and database search results will be deposited at the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org)via the PRIDE partner repository.

Data analysis:
Uninterpreted tandem MS spectra were searched for peptide matches against the sequence for BCL6 using the Mascot (Version 2.6.0)software with a 5 ppm mass tolerance for peptide precursors and 20 mDa mass tolerance for fragment ions. 9Raw files were searched using trypsin as the enzyme with up to 2 missed cleavages and the variable modifications carbamidomethylation on cysteine and oxidation on methionine were allowed.Masses corresponding to [SF−HF] were allowed as variable modification(s) on cysteine, histidine, lysine, tyrosine, serine, and threonine as well as the protein N-terminus.MS/MS spectra were manually validated and annotated.

Protocol for iterative screen against BCL6
Amine selection: Amines were selected to include those that had molecular similarity to initial hit SFs 2f-i.The similarity searching was performed using a Pipeline Pilot (Version 20.1.0.2208) software protocol. 4Initial selection criteria were applied: GSK compounds having at least one sample with concentration 10 mM in DMSO or above and at least 50 μL sample size.Unstable compounds and compounds with other liabilities were removed using proprietary GSK filters.Further selection criteria were then applied: 150<M W <350, BioByte cLogP<9 and only 1 aliphatic amine group. 5Substructure filters were used to create a pool of available molecules.SmallWorld clustering was used with the original query compound(s) and the pooled compounds, and the compounds were tagged as to whether they were generated with BioDig molecule generator from the original query compound.A 352-membered library was selected from the remaining molecules.The selected amines (10 mM, 5 µL per well/screen) were ordered from GSK's solution stores in Labcyte 384LDV plates.

HTC-D2B iterative screen against BCL6:
To a Labcyte 384LDV plate containing the selected 352 amine functionalized fragments (10 mM) in DMSO (5 µL per well) was added a stock solution of OSu 2a (10 mM) and NEM (30 mM) in DMSO (5 µL per well).The plate was sealed, centrifuged (1 min, 1000 rpm), and allowed to sit at room temperature for 1 h.After the reaction (assumed product concentration: 5 mM), a Labcyte ECHO © 555 Liquid Handler was used to transfer the library of SFs to a Greiner 384 white low volume plate.Purified BCL6 diluted from original stock into a buffer made up of HEPES (pH 7.5, 25 mM) and NaCl (50 mM) in distilled MS-grade water (15 µL per well) was subsequently added across the plate.Final concentrations: 1 µM protein; 50 µM SF.The plate was sealed, centrifuged (1 min, 1000 rpm), incubated at 4 °C for 24 h, and subsequently analyzed by intact protein LC-MS.
Displacement study with GSK137 The (resynthesized and purified) SF hits (2f-i, 2p, 2q, 2s, and 2t) were plated into a Greiner 384 white low volume plate from a 10 mM source in DMSO, or DMSO-d 6 only as a control.GSK137 (11) was also added from a 10 mM source in DMSO.Purified BCL6 diluted from original stock into a buffer made up of HEPES (pH 7.5, 25 mM) and NaCl (50 mM) in distilled MS-grade water (15 µL per well) was subsequently added across the plate.Final concentrations: 1 µM protein; 50 µM SF; 50 µM GSK137.The plate was sealed, centrifuged (1 min, 1000 rpm), incubated at 20 °C for 24 h, and subsequently analyzed by intact protein LC-MS.This was performed in duplicate and average covalent modification yields were used for analysis.Each solution was aliquoted (3 × 20 µL) into wells of an opaque 384-well PCR microplate (Axygen PCR-384-LC480-W) and coated with silicone oil DC 200 (5 µL; Fluka Chemicals 85413).The microplate was centrifuged (1 min, 200 g) prior to insertion into a Fluodia T70 spectrophotometer (Photal Instruments).The fluorescence intensity was detected (emission wavelength: 486 nm; excitation wavelength: 610 nm) as the plate instrument interior was heated from 27-75 °C using a temperature stepping mode (waiting time before measurement: 60 s; fluorescence measurement interval: 45 s).The approximate temperature gradient was 0.5 °C min -1 .
The fluorescence intensity data from the individual replicates of each sample type were normalized and identically truncated within GraphPad Prism 5.0.4. 11These values were subsequently fitted to equation ( 4) corresponding to a Boltzmann sigmoid in order to obtain the melting temperature ("V50"; T m ) of the protein: Y=BOTTOM+((TOP+BOTTOM)/(1+exp((V50-X)/SLOPE))) ( 4) The three obtained T m values were averaged to give a final T m value in each case.The normalized fluorescence data for the three replicates of each sample were averaged to yield a single dataset for each case, which were then plotted in GraphPad Prism 5.0.4 to give the plots shown in the manuscript. 11ble S9.Thermal melting temperatures of unmodified and covalently modified BCL6 BTB/POZ domain as determined by differential scanning fluorimetry measurements.
The arithmetic mean and standard deviation from three replicates are shown.The (resynthesized and purified) SF hits (2f-i) were plated into seven Greiner 384 white low volume plate from a 10 mM source in DMSO.Differing volumes were added such that the final concentrations for each compound would be: 100, 50, 20, 10, or 5 µM after adding 15 µL of protein stock solution per well.Volumes of DMSO were made to be consistent for every well.Purified BCL6 diluted from original stock into a buffer made up of HEPES (pH 7.5, 25 mM) and NaCl (50 mM) in distilled MS-grade water (15 µL per well) was subsequently added across the plates.Final concentrations: 0.5 µM protein; 100, 50, 20, 10, or 5 µM SF.The plates were sealed, centrifuged (1 min, 1000 rpm), and the first six plates were immediately queued for analysis by intact protein LC-MS at a temperature of 4 °C with the final plate paused for a 24 h timepoint.The resulting modification yields were plotted in GraphPad Prism (Version 5.0.4) software against the time of sampling. 11Non-linear regression analyzes were conducted using a 'one-phase association' model, with the y=0 value constrained to '0'.The observed rate constants (k obs , value 'K' in GraphPad) were extracted for each compound at each concentration, then replotted against the corresponding compound concentration.Further non-linear regression analyzes were conducted using a 'Michaelis-Menten' model, from which k inact values ('Vmax' in GraphPad) and K I values ('Km') were extracted.

Protocol for crystallography
Methodology BCL6 BTB domain (BCL6 5-129(C8Q,C67R,C84N)-FLAG , 10 µM) was incubated with each compound (500 µM, eq.0.25% v/v DMSO) in 20 mM Tris pH 8.5 and 150 mM NaCl (2 mL) at 20 °C for 6 h.To remove unreacted compound, the protein-fragment complex was buffer exchanged into 20 mM Tris pH 8.5, 150 mM NaCl and 5% v/v glycerol by centrifugation.The protein-fragment complex was concentrated to >6 mg/mL, as determined by spectrophotometry.Sitting drop crystallization was performed at 20 °C by combining the protein-fragment complex solution, precipitant solution (11% or 12% w/v polyethylene glycol (PEG) 3350, 16.8% v/v glycerol and 0.1 M bis-tris propane pH 7.2-7.3)and apo BCL6 seeds (in 17% w/v PEG 3350, 20% v/v glycerol and 0.1 M bis-tris propane pH 7.3) in a 10:8:2 ratio, respectively.Crystals were typically obtained within a few days.Precipitant solution was added to the crystallization drops to provide cryoprotection.Single crystals were directly flash frozen in liquid nitrogen from the drops for diffraction data collection at 100 K on the ID-30A beamline at the European Synchrotron Radiation Facility.5][16][17] In all instances, including for twinned data the correct symmetry was established to be P3 2 21.The structures were solved using an in-house search model with one protein molecule in the asymmetric unit of the P3 2 21 space group.Difference density was found near Tyr57 which could be modeled as the covalent ligand.Whilst the covalent bond was clear in all instances, the density for the rest of the ligand was variable in quality between the structures.The ligand was initially built with Grade. 180][21] Electron density maps were created in CCP4MG within the CCP4 suite, and further structural figures in the manuscript were prepared in MOE (Version 2019.0101). 10,20ta deposition: The atomic coordinates and structure factors will be deposited in the Protein Data Bank, www.rcsb.org.The Amine-functionalized fragments were obtained from the GSK compound collection.
Compounds listed in the order in which they were discussed in the manuscript:
Figure S1.Full structures of reactive moieties and amine-functionalized fragments used to trial the HTC conditions.

Figure S2 .
Figure S2.Summary of LC-MS analyzes across 12 wells (reactions with amines S1-S12) used to trial the high-throughput chemistry protocol.DMSO condition shown in red; DMSO:water (9:1) condition shown in blue.

Figure S4 .Figure S5 .
Figure S4.LC-MS analysis of six wells selected at random to assess the conversion of the HTC protocol in a plate-based format.

Figure S6 .
Figure S6.MS/MS spectrum of peptide 59 ILNNGH*AFNVEFDDSQDKAVLK 80 modified by SF 2d confirming His64 as the site of covalent modification.

Figure
Figure S7.X-ray crystal structure of CAII (PDB: 3CAJ), and virtual docking showing SF 2e covalently bound to His64 in the CAII pocket.

Figure S11 .
Figure S11.Summary of iterative screen against CAII with reactive moiety 2 and library of sulfonamide-containing fragments.Dashed line shows hit threshold at 70% (mean + 2 SDs); hits colored green; non-hits colored blue.

Figure S12 .
Figure S12.MS/MS spectra for identification of the sites of covalent modification for BCL6 hits 2f-i.a) MS/MS spectrum of peptide 47 TVLMACSGLFY*SIFTDQLKR 66 modified by SF 2f indicating Tyr57 as the site of covalent modification.For compound 2f, a strong neutral loss of 45.05 was observed following MS/MS fragmentation (ions noted by the asterisk).The nature of this loss was not further investigated.b) MS/MS spectrum of peptide 47 TVLMACSGLFY*SIFTDQLKR 66 modified by SF 2g indicating Tyr57 as the site of covalent modification.c) MS/MS spectrum of peptide 98 EGNIMAVMATAMYLQMEH*VVDTCR 121 modified by SF 2h indicating His115 as the site of covalent modification.For compound 2h, minor modification was observed at Y110; this was not investigated further.d) MS/MS spectrum of peptide 98 EGNIMAVMATAMYLQMEH*VVDTCR 121 modified by SF 2i indicating His115 as the site of covalent modification.For compound 2i, minor modification was observed at Y110; this was not investigated further.

Figure S13 .
Figure S13.DSF traces for BCL6 BTB/POZ domain modified by SF fragments or binding of GSK137.a) The modification reaction of three selected meta-substituted benzamide fragments thermally stabilized the protein.b) The modification reaction of two azetidinyl fragments gave different effects on the thermal stabilization of BCL6.c) The modification reaction of three piperidinyl fragments thermally destabilized the protein.d) The binding of GSK137 thermally stabilized the protein.

Figure S14 .
Figure S14.Kinetic analyzes for BCL6 hits.a) Time courses (various concentrations plotted against time and fitted to a single exponential function to determine k obs ) showing concentration-dependent modification of SFs 2f-i with BCL6.b) k obs measurements plotted against the measured concentrations of SFs 2f-i to determine k inact and K I .c) Table displaying k inact , K I , and hence k inact /K I -a parameter to describe the overall modification efficiencies of SFs 2f-i.
BCL6 BTB monomer is shown in a blue ribbon representation.Compounds are shown in ball-and-stick representation with green carbon atoms.BCL6 BTB domain + compound SF 2f 2F o -F c map in the active site contoured at +1.0σ (blue); +0.7σ (cyan) BCL6 BTB domain + compound SF 2s F o -F c map in the active site contoured at +1.0σ (blue) + compound SF 2f F o -F c map in the active site contoured at ±3.0σ (blue/red)) fluoride carboxylic acid precursors were supplied by Enamine.

Table S2 .
High pH gradient for small molecule LC-MS analysis FCC was conducted using Teledyne ISCO CombiFlash ® R f + apparatus with RediSep ® silica cartridges.Solvent systems for FCC are reported in solvent:solvent ratios.

Table S4 .
Elution gradient used for intact protein LC-MS

Table S5 .
Deconvolution conditions for proteins

. CAII -site(s) of binding studies
alkylated with 10 mM iodoacetamide(r.t., 30 mins, dark) ).Data for one replicate of each condition was also acquired by Data Dependent Acquisition mode (DDA).DIA Lumos settings were as follows: Transfer capillary set to 300 C and 2.2 kV applied to the nanospray needle (Evosep).MS1 data acquired in the Orbitrap with a resolution of 120k, max injection time of 20 ms, AGC target of 1e6, in positive ion mode, in profile mode, over the mass range 393-907 m/z.DIA segments over this mass range (20 m/z wide/1 Da overlap/27 in total) were acquired in the Orbitrap following fragmentation in the HCD cell (32%), with 30k resolution over the mass range 200-2000 m/z and with a max injection time of 54 ms (dynamic) and AGC target of 1e6.DDA data used the same source settings with the following MS method changes: MS1 resolution = 60k, charge state inclusion 2-6 + , MIPS mode (Peptide), dynamic exclusion of 15 s, intensity threshold of 5e4, DDA carried out with quadrupole isolation of 1.4 Da, HCD energy of 32%, MS2 acquired in the Orbitrap with 15k resolution, max injection time of 22 ms, AGC target of 1e6 in centroid mode. 13

7. BCL6 -site of binding studies and iterative screen
Greiner 384 PP F-bottom plate from a 10 mM source in DMSO.Purified BCL6 diluted from original stock into a buffer made up of HEPES (pH 7.5, 25 mM) and NaCl (50 mM) in distilled MS-grade water (50 µL per well) was subsequently added across the plate.Final concentrations: 2 µM protein; 100 µM SF.The plate was sealed, centrifuged (1 min, 1000 rpm), incubated at 4 °C for 24 h, and 15 µL aliquots were subsequently removed and analyzed by intact protein LC-MS.The remaining samples (1 μg) were separated by SDS-PAGE to remove excess unbound compound.Gels were stained with colloidal Comassie InstantBlue and bands corresponding to BCL6 were excised, reduced with 10 mM TCEP (65 °C, 30 mins), and alkylated with 10 mM iodoacetamide (r.t., 30 mins, dark).Samples were digested with trypsin (Promega) 1:10 E:S (37 °C, 16 h) in 100 mM ammonium bicarbonate.After removal of the supernatant, peptides were extracted using acetonitrile.Combined supernatants were concentrated in a SpeedVac centrifuge and acidified (0.1% formic acid, 0.05% trifluoroacetic acid) prior to injection into the LC-MS/MS system.LC-MS/MS analysis:Digested samples were injected on an Easy-nLC 1200 UHPLC system (Thermo Scientific).The nanoLC was interfaced to an Orbitrap Fusion Lumos Mass Spectrometer (Thermo Scientific).Tryptic peptides were loaded on a 2 cm × 75 µm Acclaim PepMap 100 C18 trapping column (Thermo Scientific) and separated on a 25 cm × 75 μm, 2 µm particles, PepMap C18, 2 μm particle column (Thermo Scientific) using a 50 min gradient of 2-38% acetonitrile, 0.2% formic acid and a flow of 300 nL/min.
LC-MS/MS based peptide sequencing was performed by data dependent analysis (DDA).Full MS 350-1600 Da at 120,000 resolution, MS AGC target 1e6, MS Maximum IT 50 ms, followed by MS/MS using a 3 sec.cycle time, stepped collision energy mode at 25, 29 and 32 CE%, Isolation window 1.6 m/z, fixed first mass 145 m/z, 15,000 orbitrap resolution, MS/MS AGC target 1e5 and MS/MS Maximum IT 200 ms.Alternatively, samples were analyzed by Parallel Reaction Monitoring (PRM) targeting precursor ions relative to the modified peptides.Full MS 380-1500 Da at 120,000 resolution, MS AGC target 1e6, MS Maximum IT 50 ms, followed by PRM scans using a fixed collision energy of 30 CE%, Isolation window 1.2 m/z, scan range of 140-2000 m/z, 30,000 orbitrap resolution, MS/MS AGC target 1e5 and MS/MS Maximum IT 100 ms.

Table S8 .
Quantitative output from intact LC-MS analysis of BCL6 BTB/POZ domain reacted with SFs after incubation of DMSO-d 6 or GSK137.Extent of displacement measured as (DModification/Modification(DMSO-d 6 treatment)).

Table S10 .
Data collection and refinement statistics.Values in parentheses are for the highest resolution shell.