Distinguishing Allosteric Effects from Orthosteric Binding in Protein-Ligand Interactions

Summary

Application of amide hydrogen-deuterium exchange mass spectrometry to map interactions of low affinity fragment and ligands is demonstrated. This protocol describes a method for distinguishing orthosteric binding from allosteric changes accompanying high affinity ligand and low-affinity fragment binding to target protein, Hsp90, and finds important applications in fragment-based drug design.

Abstract

A fundamental challenge in deciphering protein-ligand interactions is distinguishing binding changes at orthosteric sites from the associated allosteric changes at distal sites, as structural data does not always reveal allostery. Ligands mediate both orthosteric and allosteric effects on target proteins and hence, in the context of screening low affinity fragments, it is important to describe fragment efficacy in terms of both direct binding and long-range allosteric responses. This presents a significant problem especially for low affinity ligands. Amide Hydrogen Deuterium Exchange Mass Spectrometry (HDXMS) is a robust method that can provide structural insights and information on conformational dynamics for both high affinity and transient protein-ligand interactions. Here, we describe the use of HDXMS on the ATPase domain of Hsp90, to parse orthosteric and allosteric effects mediated by two high affinity ligands and two low affinity fragment compounds. A comparison of deuterium exchange in ligand-bound-Hsp90 versus apo-Hsp90 was used to describe composite changes that combine both orthosteric effects and allosteric changes. Allostery can be discerned by correlating HDXMS results with structural information about orthosteric binding from crystallographic structures of protein-ligand interactions. Results from this approach indicated that fragments and ligands both mediate interactions at overlapping orthosteric sites but elicit distinct allosteric effects. However, orthosteric interactions of Hsp90 with fragments are inherently weaker due to faster dissociation rates (k_off). This approach finds important applications in fragment screening, ranking, and lead compound design in fragment-based drug discovery.

Introduction

Drug development necessitates a complete understanding of the interaction of natural ligands with their target proteins, and utilizes this information to find alternate inhibitors or activators. Traditional drug development pipelines involve a high throughput screening (HTS) strategy to identify lead compounds¹. An alternative strategy is to use fragments as building blocks for lead compound generation. These have multiple advantages over traditional HTS strategies, including but not limited to, being intellectual property-free, optimizable, and modular². Fragments are defined as small chemical compounds (<300 Da) which mediate fewer than three H-bonding contacts with their target proteins³. Fragments are essentially the active moieties of drug molecules. Characterization of fragment-protein interactions poses unique challenges to current structural biology methods due to their low-binding affinities. Another important drawback of structural biology tools, such as X-ray crystallography and cryo-EM, is that they provide insights into kinetically constrained endpoint states which primarily provide information on orthosteric binding contacts between ligands and proteins. This is especially relevant in structures of protein-ligand interactions obtained by soaking ligands with protein crystals, where large-scale conformational movements in solution upon ligand binding, are likely to be undetected. X-ray crystallography also requires extensive optimization for crystallization and only provides a static structure of proteins. However, proteins in solution are dynamic molecules and this dynamics is important for their function⁴. In addition, monitoring proteins in solution offers the additional advantage of capturing transient intermediate changes. Hence, in order to comprehensively map binding effects of ligands to proteins, we need a dynamic overview in addition to the structural information available⁵. Nuclear magnetic resonance (NMR) spectroscopy can provide dynamic structural information but is limited by its analyte size and suffers from sensitivity issues. Additional techniques such as surface plasmon resonance (SPR)⁶ and bio-layer interferometry (BLI)⁷ can sensitively detect structural changes and capture the binding kinetics of protein-ligand interactions, but do not provide any local structural information. Consequently, capturing dynamic changes in both orthosteric binding sites and allosteric sites, with local structural information and binding kinetics, is critical to provide a systemic model for protein-ligand interactions⁸.

Hence, it is essential to work with a more holistic model of protein-ligand interactions, which includes both orthosteric and allosteric changes^9,10. The large body of available structural information on protein-ligand complexes is limited to details of binding interactions at orthosteric sites. This lack of information on changes at non-orthosteric regions upon ligand binding necessitates a complete description of the changes across the protein in solution. Protein dynamics has been shown to play an important role in distal allosteric communication and modulation, and hence capturing changes in conformational dynamics is crucial to develop a systemic model for ligand binding^11,12 that can be extended to fragment protein interactions. Amide hydrogen-deuterium exchange mass spectrometry (HDXMS) provides a map of the protein dynamics in solution at peptide-resolution, by measuring rates of deuterium uptake at peptide reporters across the protein. HDXMS measures changes in H-bonding and solvent accessibility in backbone amide hydrogens (H-bonding plays a major role in determining deuterium uptake rates) in protein-drug interactions¹³. Since H-bonds play an important role in protein-ligand interactions, HDXMS is uniquely poised to monitor ligand binding¹⁴ and has recently emerged as an important tool for biopharmaceutical discovery and development^15,16,17. It offers significant advantages in studying protein-ligand complexes, which include no limitations on target protein size, ability to analyze proteins in physiological solution states without the need for concentrated protein samples, two advantages which eliminate artefacts due to aggregation and crowding.

A comparative analysis of deuterium exchange across multiple peptide reporters in the presence and absence of a ligand provides a protein-wide map of the changes in solution dynamics due to ligand binding^18,19. This offers a read-out of protein dynamics from seconds to longer timescales, determined by the deuterium labelling time^20,21. HDXMS of protein-ligand complexes reports on both orthosteric changes at the binding site and long-range conformational changes at allosteric sites, in response to ligand binding^22,23. Overlaying information on protein dynamics with structural data from orthosteric sites enables us to describe long-range conformational changes distal from binding sites. A complete description of both these changes has important applications in describing the interactions of low-affinity fragments with proteins. An approach to map these composite changes involves an initial dynamic description of the natural inhibitors or tight-binding ligands, which provides a standard reference to compare the binding effects of fragments. This initial interaction map of natural ligands acts as the reference fingerprint to compare different fragments to test their binding interactions. The reference fingerprint includes information on the peptide-reporters and magnitude of deuterium uptake changes and kinetics of these changes.

Here, we apply HDXMS to describe orthosteric and allosteric changes in response to the binding of fragments and high-affinity ligands to the N-terminal ATPase domain of Hsp90^{24,25,26,27,28}. The protocol focuses initially on Hsp90 and its interactions with two of its high-affinity ligands purified from natural sources: Radicicol²⁹ and 7-N-Allylamino-17-demethoxygeldanamycin (17-AAG)³⁰. The composite changes are differentiated into orthosteric and allosteric changes based on crystallographic structures to identify the regions and HDXMS-specific peptide-reporters that correspond to these respective changes. This information can then be expanded to map the effects of two low affinity fragments, the phenolic compounds, Methyl 3,5-Dihydroxyphenylacetate (Fragment 1) and 2,4 Dihydroxypropiophenone (Fragment 2)^24,31 with dissociation constants of ~500 µM. Further, a workflow is described for the application of this approach to fragment screening for generating a ranking system to sort fragments based on the magnitude of changes in protein dynamics at various loci.

The main advantage of this approach is its wide-applicability to any protein or multi-protein complex. HDXMS studies of proteins have been carried out in various environments, for instance membrane proteins can be characterized in membrane-mimetic nano-discs, detergents, and as assembled macromolecular complexes such as viruses. These highlight the robustness of the approach in describing the dynamics of a wide range of protein targets. HDXMS analyses of the binding of peptide-inhibitors to their target protein offer additional insights into complementary interface residues on the inhibitor end. Since HDXMS involves no disruptive labels, dyes, or specific osmolyte conditions, protein-ligand interactions can be monitored in solution at physiologically similar conditions. These also offer the possibility of studying these interactions with different physical perturbants such as temperature, osmolyte, pH, and other perturbants such as lipids, nucleic acids, and other proteins.

Protocol

1. Preparing D ₂ O buffer, Quench and Hsp90 Protein Solutions

1. Prepare 100 µM Hsp90 protein solution (expressed in E. coli ²²) in Hsp90 aqueous buffer (20 mM HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid), 300 mM NaCl, 10% (v/v) glycerol, 0.5 mM TCEP (tris(2-carboxyethyl)phosphine), pH 7.5).
2. Prepare Hsp90 D₂O buffer by vacuum evaporation of the Hsp90 aqueous buffer. Dry the Hsp90 aqueous buffer by vacuum evaporation to remove H₂O. Subsequently, reconstitute the dried buffer constituents with an equivalent volume of D₂O to make the Hsp90 D₂O buffer.
   1. pH stability is critical. Ensure that the pH of buffers is stable and measured accurately. Control pH and pH_read carefully at each step of the process.
   2. Prepare sufficient aliquots of Hsp90 D₂O buffer for the entire set of experiments from a single buffer solution batch to minimize variability from factors such as pH, temperature, concentrations, and buffer constituents.
   3. Optimize and determine specific temperature, concentrations, and buffer constituents for each protein-ligand system. Refer to^32,33 for in-depth methodological support.
3. Determine deuterium exchange reaction ratios (Ratio of Hsp90 protein solution: ligands in dimethyl-sulfoxide (DMSO): Hsp90 D₂O buffer: Hsp90 quench solution) to achieve the highest deuterium concentration by maximizing the ratio of D₂O buffer.
   1. For Hsp90 deuterium exchange reactions, mix 1 µL of 100 µM Hsp90 protein solution (from step 1.1) with 2 µL of ligand in DMSO/water (for ligand-bound experiments) or 2 µL of DMSO/water-blank (for ligand-free condition) and 27 µL of Hsp90 D₂O buffer (from step 1.2) to achieve a final D₂O concentration of 90% and a final Hsp90 protein concentration of 3.3 µM in a 30 µL deuterium exchange reaction volume.
   2. If needed, concentrate protein solutions prior to initiating the deuterium exchange reaction. For lower concentrations of proteins, use 2 - 4 µL of protein and adjust the deuterium exchange reaction volumes accordingly to obtain the highest D₂O concentration achievable; e.g., increase total reaction volume to 100 - 250 µL.
4. Prepare ligand solutions in water (for high-affinity inhibitors) or DMSO (for low-affinity fragments).
   NOTE: The concentrations of ligand solutions are determined to ensure that the ligand concentration saturates the binding site under the final deuterium exchange reaction conditions (step 1.3.1).
   1. Subsequently, for mapping high-affinity inhibitor binding of radicicol and 17-AAG, prepare 300 µM radicicol and 17-AAG inhibitor solutions in DMSO to maintain final deuterium exchange reaction concentrations (step 1.3.1) of radicicol (KD = 19 nM) and 17-AAG (KD = 33 nM)³³ at 20 µM with a final Hsp90 protein concentration of 3.3 µM (6:1 ligand to target protein ratio) to saturate the Hsp90 binding site with inhibitors.
   2. For the low affinity fragments, prepare 75 mM solutions of Fragment 1 and Fragment 2 in DMSO to maintain final deuterium exchange reaction concentrations of Fragment 1 (KD = 490 µM) and Fragment 2 (KD = 570 µM) at 5 mM with a Hsp90 protein concentration of 3.3 µM (~ 1500:1 ligand to target protein ratio) to saturate the Hsp90 binding site with fragments.
5. Prepare the Hsp90 quench solution by adding 2% trifluoroacetic acid (TFA) to water such that addition of 20 µL of Hsp90 quench solution to 30 µL of deuterium exchange reaction (step 1.3.1) reduces the pH of the final quenched exchange reaction to 2.5.
6. Prepare fresh Hsp90 quench solutions before each set of experiments and test their ability to reduce the pH of the final deuterium exchange reaction to 2.5 before every experimental set.
7. Alternatively, use a phosphate buffered quench of pH 2.5 to reduce the pH of the final quenched deuterium exchange reaction to pH 2.5. Ensure that addition of high molarity phosphate buffers does not cause salting out in capillaries.
   NOTE: TFA is preferred as a quench solution for its relative inertness. Estimated time required for preparation of D₂O buffers and Hsp90 quench solution is 3 - 4 hours. Prepared buffers can be stored at 4 °C.

2. Setting Up Deuterium Exchange-Liquid Chromatography Coupled to Mass Spectrometry System (LC/MS)

1. Prepare the trapping stage of the LC/MS by adding an online pepsin column and a C18-trap column in a commercial HDX specific module. Maintain the immobilized pepsin at 12 °C and trap column at 4 °C to reduce deuterium back-exchange with solvent.
   NOTE: A simple setup in the absence of a commercial HDX-manager module involves maintaining the trap and C18 columns at 0 to 4 °C (ice-bath or a refrigerator). The quenched sample is then automatically eluted through the pepsin column under high pressure with solvent water at pH 2.5 (by adding formic acid) and the resultant peptides are trapped in a trap column. Another alternative to an inline pepsin column is use of immobilized pepsin-beads for proteolytic cleavage before injection into the trap column.
2. Prepare the analytical stage of the LC/MS by attaching a reverse phase C18 LC column downstream of the trap column such that the trapped peptides are eluted onto the reverse-phase C18 column by a gradient of water:acetonitrile (pH 2.5 by addition of formic acid).
3. Set an LC gradient with three distinct steps: an initial 92:8 water:acetonitrile ratio to remove non-specific peptides, a linear gradient from 92:8 to 15:85 water:acetonitrile ratio, and 15:85 water:acetonitrile ratio to remove any residual peptides or un-cleaved products.
4. Connect the LC outlet to the source of the mass spectrometer.
5. Calibrate the mass spectrometer with reference compounds before data collection (e.g., Glu-fibrinogen peptide or leucine-enkaphalin solutions). Add the reference compound solutions to the mass spectrometer and select 'continuous calibration' mode during data collection. Collect mass spectrometry data in MS/MS mode for data independent analysis (DIA).
   NOTE: Data collection automatically collects precursor ion and fragment ion spectra along with the retention time of individual precursor ions.
6. Modify trapping time for pepsin proteolysis, LC gradient, and elution time to improve sequence coverage, if required.
   NOTE: The experimental setup, peptide identification, and deuterium exchange have also been previously described^22,32. This representative LC/MS setup described for Hsp90 can be directly applicable for non-aggregating homogenous protein samples. The estimated time required for preparation of solutions, columns, and the calibration of the mass spectrometer (MS) is around 2 - 3 hours.

3. Determining Peptide-List from Undeuterated Hsp90 LC/MS Experiments

1. Prepare undeuterated reactions of ligand-free Hsp90 by mixing 1 µL of Hsp90 protein solution (step 1.1) + 2 µL DMSO-blank + 27 µL of Hsp90 aqueous buffer. Add 20 µL of Hsp90 quench solution (step 1.5) to reduce the pH of the solution to 2.5.
2. Inject this sample into the HDX-manager fitted with pepsin, trap, and C18 columns with outlets to a mass spectrometer (steps 2.1 to 2.5). Press the 'start' button in the HDX-manager to start pepsin-proteolysis followed by mass spectrometry data collection during LC gradient.
3. Obtain the peptide-database (theoretical list of all possible peptides) using the protein primary sequence and the proteolytic enzyme used for cleavage (pepsin)^34,35 using the vendor provided^34,35 or other integrated analysis software^37,38.
4. Search for peptides identifiable in the sample against the peptide-database using the MS/MS mass spectrometry data from the undeuterated Hsp90 protein experiments (step 3.1).
5. Identify and collect a list of peptides along with their LC retention times from precursor and fragmentation profiles obtained simultaneously throughout the LC gradient ³⁶.
6. Filter the peptide-list to remove peptides with low intensity, poor fragmentation profiles, and high error. Typically use cut-offs, in validated or vendor-provided software^37,38,39, to select peptides with minimum peak intensity of 2,000 Arbitrary Units (AU), maximum MH+ error of 10 ppm and a minimum of one fragment.
7. Filter the peptide list to ensure that peptides that were eluted only during the gradient (step 2.3) are selected, to maintain reproducibility in obtaining a final peptide list.

4. Additional Optimization

1. Add additional post-quench reagents such as denaturants in quench solution (urea and guanidine hydrochloride) and repeat steps 3.1 to 3.7 to increase the number of well-resolved peptides. Modify the chromatographic gradient and time to improve separation and resolution of pepsin fragment peptides.
2. Add reducing agents such as tris(2-carboxyethyl)phosphine (TCEP) or dithiothreitol (DTT) in quench solutions for better proteolytic cleavage of proteins that have disulphide bonds.
3. Incubate quenched samples with denaturants such as urea/guanidium-HCl (which helps to unfold the protein for optimal proteolytic cleavage) to improve primary sequence coverage.
4. Improve data resolution by optimizing the protocol to generate multiple overlapping and nested peptides.
   NOTE: Estimated time required for a single LC/MS experiment is 20 min and is dependent on the LC gradient time. Four deuterium labelling time points and an undeuterated sample experiment are together estimated to take 3 - 4 h per individual experimental replicate, per ligand, or fragment-bound or ligand-free condition. LC/MS data collection time for three individual replicates for each fragment or ligand binding condition is expected to be 9 - 12 h.

5. Deuterium Exchange Reaction of High-Affinity Ligand-Protein Interaction to Identify Peptide-Reporters

1. Prepare deuterium exchange reactions of ligand-free Hsp90 by addition of 1 µL of 100 µM Hsp90 protein solution (step 1.1) + 2 µL DMSO-blank + 27 µL of Hsp90 D₂O buffer (step 1.2) resulting in a final labelling D₂O concentration of 90% and Hsp90 concentration of 3.3 µM.
2. Prepare similar deuterium exchange reactions of high-affinity inhibitor-bound Hsp90 protein by adding 1 µL of Hsp90 protein solution (step 1.1) + 2 µL high-affinity inhibitor solution (radicicol and 17-AAG from step 1.4.1) + 27 µL of Hsp90 D₂O buffer (step 1.2) resulting in a final labelling D₂O concentration of 90% and Hsp90 concentration of 3.3 µM.
3. Perform deuterium exchange reactions by incubating deuterium exchange reactions (steps 5.1 and 5.2) for specific deuterium labelling time-points (0.5 min, 1 min, 2 min, 5 min and 10 min).
4. Include additional deuterium exchange time-points such as 100 min and 24 h, if needed. Include millisecond deuterium exchange reactions with a stopped-flow instrument, if required.
   NOTE: A deuterium exchange reaction for Hsp90 was carried out for the following deuterium labelling times: t= 0.5, 1, 2, 5, and 10 min. Although, this time series represents a shorter labeling time series for deuterium exchange, this was the optimal labeling-time-window where the largest changes were observed.
5. Add the prepared Hsp90 quench solution (step 1.5) to quench the deuterium exchange reaction by reducing the pH to 2.5. Inject quenched samples to liquid chromatography coupled to a mass spectrometer (LC/MS) setup (step 3.2).
6. Analyze the peptides in the peptide-list (step 3.7) and determine deuterium uptake values for each peptide in all the experimental conditions: ligand-free Hsp90 (step 5.1) and high-affinity inhibitor-bound Hsp90 (step 5.2). Calculate deuterium uptake values for each peptide at each of the deuterium labelling time-points (step 5.3).
7. Compare deuterium uptake values for peptides and calculate differences in deuterium uptake for each peptide, between ligand-free Hsp90 and high-affinity-bound Hsp90.
8. Filter peptides that show significant differences in deuterium uptake above the significance threshold of 0.5 Da.
9. Identify Hsp90 residues involved in ligand binding by analyzing ligand-bound structures of Hsp90 from PDB.
   1. Determine residues within 4 Å distance from the ligand using a structure visualization tool such as PyMOL and classify them as orthosteric residues. Load the protein-ligand complex structure in PyMOL using the PDB identifier 4EGK for the Hsp90-Radicicol structure. Click and select the ligand radicicol (RDC) in the sequence and use the action menu to modify the selection to include residues within 4 Å distance. Classify these amino-acids as orthosteric residues.
   2. Additionally, include residues that have been annotated as binding sites for the ligand (in either PDB or literature)²².
   3. Determine the list of peptides in Hsp90 that show significant differences (>0.5 Da) in deuterium uptake between the ligand-free and ligand-bound states.
10. Classify peptides that show significant differences from step 5.9 and span one or more orthosteric residues as orthosteric reporter-peptides.
11. Classify peptides that show significant differences from step 5.9 but do not include any orthosteric residues as allosteric reporter-peptides.
    NOTE: Changes at these peptides represent long-range allosteric changes in response to ligand binding at orthosteric sites.

6. Deuterium Exchange Reactions of Fragment-Protein Interactions to Determine Orthosteric and Allosteric Effects due to Fragment-Binding to Hsp90

1. Prepare deuterium exchange reactions of fragment-bound Hsp90 by addition of 1 µL of 100 µM Hsp90 protein solution (step 1.1) + 2 µL of fragment solutions in DMSO (Fragment 1 and 2 from step 1.4.2) + 27 µL of Hsp90 D₂O buffer (step 1.2) resulting in a final labelling D₂O concentration of 90% and Hsp90 concentration of 3.3 µM and fragment concentration of 20 mM.
2. Select a small-set of suitable deuterium labelling time-points (30 s and 5 min) which show the highest changes in high-affinity-ligand-protein interactions.
3. Preferably include shorter deuterium labelling time-points (30 s), since differences in deuterium uptake upon fragment-binding are readily apparent at these short deuterium labelling time-points due to the weak binding affinities of fragment compounds.
4. Perform deuterium exchange reactions for fragment-Hsp90 interactions by incubating the deuterium exchange reactions (step 6.1) for the specific time points (30 s and 5 min), followed by addition of Hsp90 quench solution to reduce the pH to 2.5.
5. Determine deuterium uptake values (similar to step 5.7) for orthosteric and allosteric reporter-peptides identified in steps 5.10.3 and 5.11.
6. Analyze fragment-protein interaction data at these reporter-peptides to qualitatively determine the orthosteric and allosteric effects of fragment-binding based on the number of peptides or regions that show changes upon each fragment binding.
7. Quantitate these changes by measuring the differences in deuterium uptake at each of these reporter-peptides. The differences indicate the amount of protection at these reporter-peptides due to fragment-binding and indirectly report the strength of the interaction¹⁰.
8. Identify reporter-peptides that show significant differences in deuterium uptake above the significance threshold of 0.5 Da (similar to 5.10). Determine the number of orthosteric and allosteric peptides that show significant differences upon fragment-binding for each fragment (Fragment 1 and 2, similar to 5.11 and 5.12).

7. Additional Interpretation

1. Analyze the deuterium uptake kinetics over time to predict the relative k_off rates for ligands (or fragments) with similar dissociation constants (K_D). Measure decreases in observed deuterium uptake differences (ligand-bound protein versus ligand-free protein) with increases in deuterium labelling time.
2. Compare dissociation rates (k_off) of ligands and rank them based on faster observed decreases in deuterium uptake differences with increases in deuterium labelling times at orthosteric reporter-peptides.
3. Identify ligands with faster dissociation rates by comparing those that show earlier decreases in deuterium uptake differences with increases in deuterium labelling time at orthosteric reporter-peptides.
   NOTE: Estimated time for data analysis can be from 2 - 10 days and is dependent on the number of fragments or ligands analyzed.

Results

In order to identify the reporter peptides that represent changes in Hsp90 upon ligand binding, changes in deuterium uptake were quantified for Hsp90 in the presence and absence of the high affinity ligands. Differences in deuterium uptake were determined at pepsin-proteolyzed peptides between the high-affinity-ligand bound-Hsp90 and ligand-free-Hsp90 and reporter peptides that showed significant differences in deuterium uptake (>= 0.5 Da) were identified. The error in a single deuter...

Discussion

Critical steps in the protocol: It is essential that the pH of solutions, including protein buffers and LC-solutions, are all maintained at a pH of 2.5 to minimize loss of deuterium labelling. It is also critical that deuterium exchange experiments be carried out at saturating concentrations of ligands to maintain a homogenous population of ligand-bound protein. This can be estimated from the ligands' or fragments' dissociation constants and need to be consistent among all fragment-protein deuter...

Materials

Name	Company	Catalog Number	Comments
Deuterium Oxide	Cambridge Isotope, Tewksbury, MA	DLM-6-1000
HEPES Buffer	Sigma Aldrich, St. Louis, MO	H3375 SIGMA
Glycerol	Sigma Aldrich, St. Louis, MO	G5516
NaCl	Sigma Aldrich, St. Louis, MO	S7653
TCEP	Sigma Aldrich, St. Louis, MO	C4706
DMSO	Sigma Aldrich, St. Louis, MO	D8418
TFA	Sigma Aldrich, St. Louis, MO	302031
FA	Fisher Scientific, Singapore	A117-50
Glu-fibrinogen	Sigma Aldrich, St. Louis, MO	F3261
Leucine-Enkaphalin	Waters, Milford, MA	186006013
ACN	Sigma Aldrich, St. Louis, MO	34851
pNIC28-Bsa4 vector	Addgene, Cambridge, MA	26103
BL21(DE3) E. coli strain	Merck-millipore Novagen, Singapore	69450
Terrific Broth medium	Sigma Aldrich, St. Louis, MO	T9179
Kanamycin	Sigma Aldrich, St. Louis, MO	60615
chloramphenicol	Sigma Aldrich, St. Louis, MO	C0378
isopropyl β-D-thiogalactopyranoside (IPTG)	Sigma Aldrich, St. Louis, MO	IPTG-RO ROCHE
imidazole	Sigma Aldrich, St. Louis, MO	I5513
Protease Inhibitor Mixture Set III, EDTA free	Merck-Millipore, Singapore	539134
Vibra-Cell processor	Sonics & Materials Inc., Newtown, CT	VC 505 / VC 750
nickel-nitrilotriacetic acid Superflow resin	Qiagen Inc., Valencia, CA	30410
HiLoad 16/60 Superdex-200 column	GE Healthcare, Waukesha, WI	28989335
Vivaspin 20 filter concentrators	GE Healthcare, Waukesha, WI	28932360
Poroszyme Immobilized Pepsin Cartridge, 2.1 mm x 30 mm	Thermofischer, Sigapore	2313100
ACQUITY UPLC BEH C18 Column	Waters, Milford, MA	186002350
ACQUITY UPLC BEH C18 VanGuard Pre-column	Waters, Milford, MA	186003975
nanoAcquity HDX sample manager	Waters, Milford, MA
Synapt G1 ESI mass spectrometer	Waters, Milford, MA
nanoAcquity Auxillary Solvent Manager	Waters, Milford, MA
nanoAcquity Binary Solvent manager	Waters, Milford, MA