If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corresponding authors at: Newcastle University, Faculty of Medical Sciences, Biosciences Institute, Framlington Place, Newcastle Upon Tyne NE2 4HH, United Kingdom
Corresponding authors at: Newcastle University, Faculty of Medical Sciences, Biosciences Institute, Framlington Place, Newcastle Upon Tyne NE2 4HH, United Kingdom
MALDI-TOF MS is a powerful analytical technique that provides a fast and label-free readout for in vitro assays in the high-throughput screening (HTS) environment. Here, we describe the development of a novel, HTS compatible, MALDI-TOF MS-based drug discovery assay for the endoplasmic reticulum aminopeptidase 1 (ERAP1), an important target in immuno-oncology and auto-immune diseases. A MALDI-TOF MS assay was developed beginning with an already established ERAP1 RapidFire MS (RF MS) assay, where the peptide YTAFTIPSI is trimmed into the product TAFTIPSI. We noted low ionisation efficiency of these peptides in MALDI-TOF MS and hence incorporated arginine residues into the peptide sequences to improve ionisation. The optimal assay conditions were established with these new basic assay peptides on the MALDI-TOF MS platform and validated with known ERAP1 inhibitors. Assay stability, reproducibility and robustness was demonstrated on the MALDI-TOF MS platform. From a set of 699 confirmed ERAP1 binders, identified in a prior affinity selection mass spectrometry (ASMS) screen, active compounds were determined at single concentration and in a dose-response format with the new MALDI-TOF MS setup. Furthermore, to allow for platform performance comparison, the same compound set was tested on the established RF MS setup, as the new basic peptides showed fragmentation in ESI-MS. The two platforms showed a comparable performance, but the MALDI-TOF MS platform had several advantages, such as shorter sample cycle times, reduced reagent consumption, and a lower tight-binding limit.
In the endoplasmic reticulum, peptides originating from proteasomal degradation of ubiquitylated proteins are trimmed to a length of eight to twelve amino acids by the endoplasmic reticulum aminopeptidase 1 (ERAP1) [
]. These short peptides subsequently form a stable complex with human leukocyte antigen (HLA) molecules and are presented on the cell surface for immune cell recognition [
]. It has an internal sequence preference for hydrophobic and positively charged residues, which is caused by its large negative electrostatic potential internal cavity [
]. The trimming of peptides to the correct length is proposed to be mediated by these structural features that shape the active and regulatory binding sites [
]. The extent to which ERAP1 shapes the immunopeptidome is still not completely understood, however, its dysregulation has been linked to cancer and several auto-immune and auto-inflammatory diseases [
High-throughput in vitro enzyme assays are routinely used in the early stages of the drug discovery process to screen the activity of millions of compounds against a specific target [
]. These techniques are used as they are fast, sensitive and generally robust but they are also susceptible to artifacts, leading to large initial false positive rates, and therefore require substantial subsequent follow up [
]. Moreover, current fluorescent and chemiluminescence HTS assays for ERAP1 need the inner-filter effect correction term to adjust product fluorescence intensity in the presence of high substrate concentrations [
]. Mass spectrometry (MS) is a label-free technique that enables direct quantification of assay analytes such as substrates and products from in vitro assays. In the case of ERAP1, screening with an unlabelled substrate is particularly preferrable as the enzyme activity is mediated by the substrate's structural features [
]. This results in 40 minutes acquisition times per 384 well plate which is not fast enough for full-deck screens of millions of compounds. There are, however, solutions to increase the throughput on this platform, for instance by using the BLAZE mode which is a direct injection without cartridge, resulting in cycle times of 2.5 seconds [
During the last decade, instrument advancements in another soft ionisation technique, matrix assisted laser desorption/ionisation (MALDI) time-of-flight (TOF) MS, have enabled laser scanning speeds of less than one second per sample, significantly increasing throughput of this technique [
]. Furthermore, MALDI-TOF MS assays can be performed with very low sample volumes, unlike the RF MS assays, and are therefore amenable to miniaturisation. These qualities combined with fully automated workflows have made MALDI-TOF MS assays highly attractive for HTS [
]. In a typical MALDI-TOF MS HTS workflow (Fig. 1), an in vitro enzymatic reaction is performed for a set time and then quenched using an acidic solution that contains an internal standard. The sample is then mixed with matrix, which aides analyte desorption and ionisation. High-throughput capable assays using this platform have been successfully established for several targets, including kinases,[
Differential analyte derivatization enables unbiased MALDI-TOF-based high-throughput screening: a proof-of-concept study for the discovery of catechol-o-methyltransferase inhibitors.
High-throughput matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry-based deubiquitylating enzyme assay for drug discovery.
]. Guidance has been published, highlighting the need to carefully design the assay to avoid signal interference from solvents commonly used in biochemical in vitro assays and other investigations have focussed on the evaluation of assay quality [
Fig. 1Automated MALDI-TOF MS based high-throughput screening workflow for in vitro enzyme assays. Liquid handling devices are used for (1) compound, (2) enzyme, (3) substrate (S), and (4) internal standard (iSt) in quench solution delivery which is interrupted by sensible incubation times (indicated above arrows) to allow compound enzyme alignment and product (P) formation over time. An isotopically labelled internal standard is spiked into the samples to allow quantification. (5) The sample is mixed with matrix and delivered onto the target plate which is acquired by MALDI-TOF MS. (6) Upon laser ionisation, analyte ions are separated in a flight tube according to their mass-to-charge (m/z) ratio before they hit a detector. (7) In the mass spectrum, active and inactive compounds can be distinguished based on the observed product formation.
In this work, a MALDI-TOF MS in vitro assay for ERAP1 was developed based on an established RF MS assay. To improve the ionisation efficiency of the original peptide substrates, we incorporated arginine residues into the peptide sequence. In vitro assay optimisation and validation with known ERAP1 inhibitors was conducted with these novel basic peptides. To assess the assay stability, robustness and reproducibility, DMSO plates, a robustness set, and a ∼10k compound validation set were screened, respectively. On both MS platforms, 699 compounds that were identified as ERAP1 binders in an affinity selection mass spectrometry (ASMS) experiment were screened at single concentration and in a dose-response to validate the identified hits. Comparable MS performance was observed in these tests which indicated successful establishment of a HTS compatible MALDI-TOF MS platform for ERAP1 screening.
2. Experimental
2.1 Materials
4-(2- hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), α-cyano-4-hydroxycinnamic acid (CHCA), ammonium dihydrogen phosphate (NH4H2PO4), ammonium hydroxide, bovine serum albumin (BSA), 3-((3-cholamidopropyl) dimethylammonio)-1-propanesulfonate (CHAPS) hydrate, dimethyl sulfoxide (DMSO), formic acid LC-MS grade, L-Leucinethiol (Leu-SH), sodium chloride (NaCl), tris(2-carboxyethyl)phosphine (TCEP), trifluoroacetic acid (TFA), Tween-20, and water (H2O) MS grade were purchased from Sigma-Aldrich. Acetonitrile (MeCN), ethanol (EtOH), and water (H2O) were purchased from Fisher Scientific in MS grade. MS grade acetonitrile was also purchased from VWR Chemicals. Bovine serum albumin (BSA), glycerol, and trifluoroacetic acid (TFA) were purchased from Fisher Scientific.
The C-terminal 6-His tagged ERAP1 full-length protein (haplotype 2) was generated according to Giastas et al. 2019 with the modifications described by Hutchinson et al. 2021, and stored in 50 mM HEPES, 100 mM NaCl pH 7.0, 10% glycerol, 0.5 mM TCEP. Ten mM stocks of the non-basic assay peptides (NBP) YTAFTIPSI (monoisotopic mass: 1011.5), acetylated YTAFTIPSI (Ac-YTAFTIPSI) (monoisotopic mass: 1053.5) and TAFTIPSI (monoisotopic mass: 848.5), as well as the basic assay peptides (BP) YTAFRIRSI (monoisotopic mass: 1125.6) and TAFRIRSI (non-heavy (monoisotopic mass: 962.6) and heavy labelled (Heavy-labelled: isoleucine position eight 13C15N; monoisotopic mass: 969.6)) (Cambridge Research Biochemicals) were prepared in DMSO. An assay buffer with 5 mM HEPES, 100 mM NaCl pH 7.0, 0.01% BSA, supplemented with either 0.002% Tween-20 or 0.1 mM CHAPS was used for all further dilutions of the NBP or BP, respectively. The known ERAP1 inhibitor DG013A was synthesised in house at GSK according to the methods described in Zervoudi et al. 2013. In addition, compounds for single-concentration screenings of a 699 and ∼9,600 compound validation screen, as well as a robustness set, were obtained from the GSK internal compound library. The compounds were dispensed into assay plates by the Echo acoustic dispenser (Labcyte) to give final assay concentrations of 0.1-1% DMSO and 10 µM compound for single concentration screens. Plates for dose-response curves were recorded with final assay compound concentrations between 3×10−3 nM and 1×104 nM (known ERAP1 inhibitors for assay validation) and between 1.7 nM and 1×105 nM (699 compound set screening hits). Up to 1 mM TCEP final assay concentration was supplemented to the assay for the full curve recordings of Leu-SH.
2.2 Peptide detection
To compare the ionisation efficiency of the different assay peptides (i.e., NBP vs. BP), equimolar concentrations (1 µM) of the NBP and BP were diluted in a 1:1 mixture of the CHAPS assay buffer and quench solution (0.75% TFA) in an 8 µL total volume. The limit of detection (LOD) was determined using the peptide concentration where at least three out of five technical replicates showed a peak with a signal-to-noise ratio (S/N) above 3. To determine signal linearity under assay conditions, increasing product concentration detection was tested while decreasing substrate concentrations (2 µM overall peptide concentration) in 8 µL volume. Substrate and product peptides were diluted in the assay buffer and a constant internal standard concentration of 1 µM diluted in the quench solution was used for limit and linearity of detection experiments.
2.3 ERAP1 in vitro assay
The above-mentioned assay buffers and a TFA quench solution resulting in 0.38% TFA after the quench were used. Different assay parameters, such as incubation times, substrate concentration, and enzyme concentrations were tested and optimised. For the optimised in vitro assay, 2 µM BP substrate and 0.25 nM ERAP1 were used in 4 µL assay volume. After 60 min incubation at room temperature, 4 µL quench solution with BP internal standard at 1 µM final concentration were added. For the NBP MALDI-TOF-MS assay, concentrations of 5 µM NBP substrate, 3 nM ERAP1 and 2.5 µM NBP internal standard were used. Compounds were pre-incubated with the enzyme for 30 min at room temperature before addition of the substrate. No enzyme was present in the negative control columns. The in vitro assay and the MALDI target plate preparation and acquisition (see subsequent section) were carried out at two different locations, Newcastle University and GSK Stevenage, due to the different HTS capabilities available on site. The Xrd-384 reagent dispenser (fluidX) and Mosquito LV (SPT Labtech) were used for semi-automated liquid handling at Newcastle University. The assay volumes were adapted to 5 µL dispenses with a Multidrop Combi dispenser (Thermo Scientific) to allow compatibility of the assay with the full automation workflow at GSK. The final concentrations of assay components and reaction incubation times were consistent at these two locations.
2.4 MALDI-TOF MS target preparation and acquisition
At Newcastle University, a Mosquito liquid handling robot was used to mix equal volumes of sample and matrix (5.6 mg/ml CHCA in 85% MeCN, 0.1% TFA, 1 mM NH4H2PO4) and to spot 0.5 μL of the mixture onto a stainless steel MTP384 MALDI target plate (Bruker Daltonics) before the plates were manually loaded into the mass spectrometer. The fully automated target plate spotting and acquisition at GSK took place on a customised platform. A robotic arm (Analytic Jena) transferred plates from their storage compartments onto a rotating disk on which the following pipetting steps were executed with the immobile CyBio Well vario (Analytic Jena) pipetting head: assay transfer into an assay-matrix-mixing plate; matrix (6.25 mg/mL CHCA dissolved in 70% MeCN, 0.1% TFA) transfer into the assay-matrix-mixing plate; mixing of sample and matrix in the assay-matrix-mixing plate; spotting of mixture onto the MALDI target plate; all with optional tip cleaning cycles in between. After heated vacuum drying, the MALDI target plate was transferred into the rapifleX PharmaPulse MALDI TOF/TOF mass spectrometer (Bruker Daltonics); this custom-made setup allowed spotting of the next MALDI target plate while the other plate was acquired.
Spectra were acquired in positive ion, reflector mode from m/z 920 - 1220. In flexControl (version 4.0), 5,000 shots at a 10 kHz frequency per spot, random walk pattern (complete sample), M5 Thin-layer laser, 50 μm x 50 μm scan range, 2000 μm spot diameter, and 200 shots per raster position were selected for data acquisition. Mass suppression up to m/z 635 was applied. At GSK, these parameters had to be slightly adjusted to 10,000 shots at a 5 kHz frequency per spot, custom M5 flat laser, 50 shots per raster position, and mass suppression up to m/z 736. The laser power was individually adjusted for each measurement by aiming for an overall spectrum intensity ∼105 arbitrary units (a.u.). Before acquiring the data, the method was calibrated with Bruker Daltonics peptide calibration standard II. Full plate data acquisition and processing was executed through the MALDI PharmaPulse software (version 2.2). Different peak detection algorithms were used for the different assay peptides (NBP: SNAP2 and BP: centroid). A signal to noise ratio (S/N) of three was used as cut-off. The S/N was increased to five for reaction velocity measurements to increase signal reproducibility [
High-throughput matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry-based deubiquitylating enzyme assay for drug discovery.
] (assay volume 25 µL) used 1 nM ERAP1 and 5 µM of the substrate NBP in the Tween assay buffer while using the same incubation time and temperature as the MALDI-TOF MS assay. To stop the reaction, 25 µL of the quench solution containing the acetylated internal standard (Ac-YTAFTIPSI) was added. Assay plates were transferred onto the RapidFire365 system (Agilent Technologies) coupled to an API4000 triple quadrupole mass spectrometer (SCIEX). Samples were aspirated under vacuum for 250 ms before loading onto a C18 solid-phase extraction cartridge, using solvent A (0.1% (v/v) formic acid) at a flow rate of 1.5 mL/min for 2000 ms. Elution with solvent B (80% LC-MS grade acetonitrile, 0.1% (v/v) formic acid) at 1 mL/min for 2500 ms followed. The cartridge was re-equilibrated with solvent A for 500 ms at 1.5 mL/min before the next injection. The total cycle time was ∼7 s per well. The mass spectrometer was operated in positive electrospray multiple reaction monitoring mode, with transitions (Q1/Q3) for each species as follows: YTAFTIPSI 1012.6/697.4; TAFTIPSI 849.4/534.3; and Ac-YTAFTIPSI 1054.6/739.5. A dwell time of 50 ms was used for all transitions. A spray voltage of 5000 V and a source temperature of 600°C were applied. All peaks were integrated and processed using the RapidFire peak integration software (version 4.0.1). Subsequent, data analysis was carried out in house on a customised software platform using the product and substrate signals for normalisation (Supplementary table 1) [
Data analysis was conducted using Microsoft Excel 2016 and GraphPad Prism (version 5.0.4 and 9.0.0). To reduce spot-to-spot variability in MALDI-TOF MS, the product signal was normalised to the internal standard signal according to equation 1. For this, the area of the protonated and sodiated adduct peaks was summed to obtain the peptide signal.
(1)
Using the normalised product area, the absolute product concentration was calculated (equation 2) to enable subsequent calculation of enzyme kinetic parameters, such as KM. A standard curve with known substrate and product concentrations was produced to obtain the correction factor (linear regression curve slope; Fig. 2) for equation 2.
(2)
Fig. 2Optimised basic peptides in MALDI-TOF MS. (A) Exemplary spectrum obtained by MALDI-TOF MS as readout for the enzymatic reaction, showing detection of the product (orange), internal standard (red), and substrate (purple) as [M+H]+, [M+Na]+ and [M+2Na-H]+. (B) Linearity of detection of the product signal after normalisation against the internal standard signal when plotted against the product concentration. Shown are mean and standard deviation from three analytical replicates with each five technical replicates with the linear regression analysis.
To compare the product generation between the assay setups, the % product formed was calculated. For the NBP in MALDI-TOF MS, the % product was calculated based on product and substrate peak areas (Supplementary table 1) [
In all other cases, the % enzyme activity was determined per assay plate and calculated according to equation 3 where the normalised product area (norm) from sample, as well as positive and negative controls were used:
(3)
The Z’ calculation (equation 4) was used to determine assay stability and quality. This equation is based on the standard deviation of positive () and negative () controls, as well as on their mean values ():
(4)
GraphPad Prism was used to determine the linear regression, column statistics (CV), correlation analysis, enzyme kinetic parameters (Michaelis-Menten analysis) and pIC50 values (four-parameter logistical curve fit) for known ERAP1 inhibitors. The data analysis of the full curve screening with the hit compounds from the 699 compound screen was conducted with a GSK in house software platform that calculates robust cut-off values and then performs the four-parameter logistical curve fit. For visualisation, Affinity designer (version 1.7.2.471), ChemDraw 20.0, and Microsoft PowerPoint 2016 were used.
3. Results and discussion
3.1 Optimisation of the assay substrate and internal standard peptides
In previous work, Liddle et al. developed a RF MS based assay for ERAP1, utilising the peptide YTAFTIPSI, a 9-amino acid antigenic epitope from Gag-Pol polyprotein from human immunodeficiency virus 1, [
]. In this assay, N-terminal trimming by ERAP1 resulted in the 8mer product TAFTIPSI.
In MALDI-TOF MS experiments, these peptides had relatively poor limits of detection (LOD YTAFTIPSI: 8 fmol, LOD TAFTIPSI: 31 fmol) (Supplementary table 2) and were represented by sodium adducts as the dominant peaks (Supplementary Fig. 1) rather than the protonated form. Moreover, when evaluating detection performance over a whole assay plate, reduced assay stability which was indicated by an increased data scatter (CV = 16), and consequently a lower assay quality (Z’ = 0.5) was observed with these peptides in comparison to the RF MS setup (CV = 3, Z’ = 0.8) (Supplementary table 1).
To mitigate this issue, we incorporated two arginine residues in the peptide sequences YTAFRIRSI and TAFRIRSI (BP); as peptides with arginine have an increased ionisation efficiency due to their improved gas-phase basicity [
Improved matrix-assisted laser desorption/ionization mass spectrometric analysis of tryptic hydrolysates of proteins following guanidination of lysine-containing peptides.
]. When comparing equimolar amounts of the NBP with the BP, for product and substrate, respectively, a higher signal intensity was always recorded for the BP (Supplementary Fig. 1). The LOD was improved for the BP to 1 fmol and 0.5 fmol on target for the product and substrate peptides, respectively (Table 1). Furthermore, the primary detected species was the protonated form of the BP (Fig. 2A). In addition, a new internal standard peptide was designed that was the product peptide with a C-terminal, heavy labelled, isoleucine. In MALDI-TOF MS, the internal standard is used to enable relative quantification [
]. The range of linear detection with this normalisation of the product against the internal standard showed no major ionisation differences between these peptides (Fig. 2B). Altogether, this suggested that the BP were better suitable for detection by MALDI-TOF MS.
Table 1Sensitivity of basic peptide (BP) detection in MALDI-TOF MS. The expected m/z of the basic assay peptides protonated and sodiated adduct peaks is displayed alongside the limit of detection which is the on-target concentration at which the peptide peaks can be detected with S/N ≧ 3.
As the BP showed favourable ionisation properties in MALDI-TOF MS, the enzymatic reaction conditions were optimised and tested for these novel peptides. The modified YTAFRIRSI peptide was a substrate of ERAP1, as successful cleavage of the N-terminal tyrosine was observed (Fig. 2A). To optimise the assay, the reaction velocity was measured with different substrate concentrations to determine KM, the substrate concentration at which half of the maximum velocity was reached. In the assay with the BP, the KM is 2 µM (Fig. 3A). This is generally consistent but slightly lower compared with the KM reported by Liddle et al. in the assay setup using the NBP (KM = 5 µM) [
] which could indicate a slightly increased binding affinity by inclusion of the two arginine residues into the substrate sequence. The enzyme concentration was titrated with a fixed substrate concentration of 2 µM and an optimal concentration of 0.25 nM ERAP1 was chosen (Fig. 3B). With this enzyme concentration, linear reaction progression was observed. The enzyme concentration was lower compared to the assay described by Liddle et al. [
Fig. 3Optimisation and validation of the ERAP1 assay on a MALDI-TOF MS platform. (A) Reaction velocity observed at different substrate concentrations to obtain enzymatic constant (KM = 2 µM) by Michaelis-Menten analysis. Shown are mean and standard deviation from four analytical replicates, each with five technical replicates. (B) Enzyme titration with 2 µM substrate concentration to determine linear reaction progression. Shown are mean and standard deviation from one analytical replicate with five technical replicates with the linear regression fit. (C) Dose-response curves recorded for two known ERAP1 active site inhibitors (Leu-SH and DG013A) to determine pIC50 values for assay validation by four-parameter logistical curve fit. Shown are mean and standard deviation from three analytical replicates, each with four technical replicates and R2 values.
]. Therefore, two known ERAP1 inhibitors that target the active side of the enzyme were used to validate the optimised assay with the BP. L-Leucinethiol (Leu-SH) is a broad-spectrum metallopeptidase inhibitor and a moderate inhibitor of ERAP1 in cellular assays [
]. In contrast, DG013A is a pseudopeptidic compound whose binding affinity is based on interactions between inhibitor side chains and active site specificity pockets of the enzyme rather than the active site zinc ion [
]. The pIC50 for both compounds were successfully identified (DG013A pIC50 = 8.0; Leu-SH pIC50 = 7.7) (Fig. 3C). In the RF MS setup with the NBP, a pIC50 of 7.0 was reported for DG013A (Supplementary Fig. 2) which was comparable to the MALDI-TOF MS and literature data (pIC50 = 7.3 – 7.5), considering that different enzymatic assays were used to evaluate inhibitor performance [
]. This already indicates that differences in the absolute pIC50 values can arise from the different assays.
3.3 Automation of the MALDI-TOF MS assay
To perform effective HTS campaigns, full automation of the assay is required. The assay stability over multiple wells, plates, days, and the reactivity to known mechanisms of interference needs to be assessed. To test well-to-well variability, three DMSO plates were screened on the automated MALDI-TOF MS platform. No boundary or time artifacts were observed in the single plate view, indicating intra-plate stability (Fig. 4A). In HTS, the coefficient of variation (CV) is used to evaluate assay sensitivity while the Z’ value monitors robustness. The Z’ is calculated with the mean and deviation of positive and negative control data and should be ideally Z’ >0.5 for large HTS campaigns [
]. The data scatter from three technical replicates demonstrated good inter-day reproducibility of the MALDI-TOF MS assay due to low CV (2.67 – 3.04%) and excellent Z’ values (0.89 - 0.91) (Fig. 4B). The assay stability was now comparable to the established RF MS platform (Supplementary Fig. 3).
Fig. 4Automation of MALDI-TOF MS based ERAP1 assay for screening of DMSO, pilot test and ∼9,600 compound validation sets. (A) Exemplary plate view showing enzyme activity across an entire DMSO plate, including a negative control column. (B) Violin plots from three analytical DMSO plate replicates (excluding negative control columns) with CV and Z’ to evaluate assay stability. (C) Enzyme activity data scatter from pilot test set (1403 compounds) with indicated 75% and 125% hit cut-off (red dashed line). (D) Enzyme activity data scatter from ∼9600 compound set using one technical replicate with indicated 75% and 125% hit cut-off (red dashed line). (E) Enzyme activity correlation of overlapping hits identified in replicate runs of the ∼9600 compound screen with the linear regression. For all spots, one technical replicate was used.
Next, the assay suitability for single concentration screening was evaluated using standardised internal compound test sets. The pilot test set contained approximately 1400 compounds with diverse drug-like structures and common pan-assay interference compounds (PAINS), such as chelators, redox active, and highly conjugated compounds that can interfere with the enzyme activity [
Hit discovery for public target programs in the european lead factory: experiences and output from assay development and ultra-high-throughput screening.
SLAS DISCOVERY: Advancing the Science of Drug Discovery.2020; 26: 192-204
]. In this set, two compounds dropped the enzyme activity below a 75% cut-off and we did not observe assay interference from the known PAINS (Fig. 4C). In addition, this set was also used to further interrogate assay reproducibility and stability. Across the 4 plates, Z’ values ranged from 0.7 to 0.9. Together, this gave us confidence that our MALDI-TOF-MS assay was robust and suitable for single concentration screening. Thus, the assay was taken forward to a larger ∼9,600 compound set screen to evaluate the HTS capabilities.
This ‘10k compound set’ is representative of the full HTS collection based on a broad range of chemical diversity. Stable assay performance was observed across the 56 assay plates with an average Z’ of 0.8. Evaluation of the set showed that a 75% enzyme activity cut-off was sufficient to exclude background activity and to reveal on average 0.9% inhibitory hit compounds in two replicates, which was in the range of the anticipated hit rate (Fig. 4D) [
]. Correlation of the enzyme activity of the identified hits from the replicate screens showed high reproducibility and robustness of the assay (r = 0.90, R2 = 0.80) (Fig. 4E), supporting the use of the assay in the HTS environment. The next step was to test the platform with a compound set of interest for ERAP1.
3.4 Screening of 699 ERAP1 binders from an ASMS screen
A set of 699 compounds, previously identified as binders of ERAP1 in an ASMS screen, were assayed for functional activity on the MALDI-TOF MS and RF MS platforms at a single concentration, with the BP and NBP, respectively. Subsequent dose-response assays were conducted to confirm these hits. Moreover, this allowed for performance comparison between the two different platforms.
A stable assay performance (Z’ = 0.8 – 0.9) was observed on the MALDI-TOF MS platform. The plate view showed that the compound set contained potent and weak inhibitors (Fig. 5A). Interestingly, one compound also appeared to show activating functionality. Correlation of the inhibitory hits from two replicate screens showed strong correlation (r = 0.96, R2 = 0.92), confirming high reproducibility of the single concentration screen (Fig. 5B). When the compounds were screened with the established RF MS setup with the NBP, comparable correlation was observed between the inhibitory hits from two replicate screens (r = 0.98, R2 = 0.95), confirming the high quality of both platforms (Fig. 5C).
Fig. 5Activity screening for 699 ERAP1 binders from an affinity selection mass spectrometry experiment on the MALDI-TOF MS (BP assay) and RF MS (NBP assay) platform. (A) Exemplary plate view showing enzyme activity on the MALDI-TOF MS platform across an assay plate with positive and negative controls to identify ERAP1 inhibitors and activators. (B) Correlation of enzymatic activity from two MALDI-TOF MS single concentration screen replicates; one acquisition per compound, linear regression red dashed line. (C) Correlation of enzymatic activity from two RF MS single concentration screen replicates; one acquisition per compound, linear regression red dashed line. (D) Correlation of the averaged (MALDI-TOF MS: three analytical replicates; RF MS: two analytical replicates) pIC50 determined in the dose-response screens on both platforms from the hit compounds of the single concentration screen.
The in vitro assay with the BP was aimed to be transferred onto the RF MS platform as slight enzymatic activity differences were reported earlier when comparing the two different assay substrates. Unexpectedly, upon acquisition of ‘substrate only’ wells by RF MS, the 9mer substrate, the 8mer product, and the 7mer precursor ions were observed (Supplementary Fig. 4A-B). Ion chromatograms were evaluated to identify if this was due to product contamination of the ‘substrate only’ wells. In those wells, product and substrate signal were detected at the same retention time (3.9 min) (Supplementary Fig. 4C). Acquisition of a ‘substrate/product mixture’ well showed that the product is detected at two different retention times, once at 3.7 min and the other one co-eluting with the substrate at 3.8 min (Supplementary Fig. 4C), demonstrating that product and substrate have different retention times. This shows that only one peptide is eluted from the column after substrate injection, indicating that in-source fragmentation is the source of product detection in substrate only samples; in-source fragmentation of analytes is a common phenomenon in ESI MS [
Although the RF-MS and MALDI-TOF-MS in vitro assays differ slightly, comparison shows that out of the 699 compounds, 105 were identified as hits and approximately one third was identified on both platforms (Fig. 6A). Dose-response screening was conducted to further investigate all hits that were identified in the single concentration screens; except for five compounds due to lacking compound availability in storage. A pIC50 >4.0 was assigned to most compounds, identifying them as real hits (Fig. 6B). The data from the MALDI-TOF MS single concentration screen shows that a large group of compounds that were real hits but only detected by one platform in the single concentration screen were clustered around the activity cut-off (Fig. 6C).
Fig. 6Hit comparison of the 699 compound screen on the MALDI-TOF MS and RF MS platform. (A) Venn diagram of hits identified in at least two analytical replicates on the MALDI-TOF MS or RF MS platform from the single concentration screen. (B) Venn diagrams displaying the number of single concentration screen hits that show a pIC50 >4.0 together with the overall number of subsequently analysed compounds in MALDI-TOF MS (left) or RF MS (right) dose-response screening based on the single concentration screen setup with which they were identified. (C) Enzyme activity data scatter from three averaged MALDI-TOF MS single concentration screens with one technical replicate each. The 75% enzyme activity cut-off is marked with the red dashed line. Hit compounds from the single concentration screen which were only detected on one assay platform but that show inhibitory activity in the MALDI-TOF MS dose-response-curve screen are indicated (blue square: MALDI-TOF MS single concentration hits, red square: RF MS single concentration hits) alongside the overlapping hits (purple triangle).
Nevertheless, good correlation between the pIC50 values obtained from the full curve screening on the MALDI-TOF MS and RF MS platform was observed (r = 0.76, R2 = 0.95) (Fig. 5D). This correlation is in the same range as other mass spectrometry platform comparisons that are based on the same in vitro assay [
]. The compounds exhibited a slightly higher potency on the MALDI-TOF MS platform; the difference was however relatively small and they still displayed the same rank order. As assay conditions like substrate and enzyme concentrations were adjusted between the two platforms, this could indicate differences in the mechanistic activities of ERAP1 with the BP as ERAP1 activity is mediated by the substrate sequence [
]. While the differences arise from the in vitro assay conditions, the MS setups showed a comparable performance.
4. Conclusion
In this study, a label-free, HTS compatible in vitro assay was developed to screen for inhibitors of ERAP1 on a MALDI-TOF MS platform. First, we investigated the substrate peptide design and discovered low ionisation efficiency and assay stability of the previously described RF MS peptides using the MALDI-TOF MS assay; we were able to increase the limit and linearity of detection of the peptides by introducing arginine residues into the substrate sequence. The assay conditions, including substrate and enzyme concentrations, were optimised with the new basic peptides and validated by testing known ERAP1 inhibitors. Screening of multiple analytical replicates of DMSO plates, pilot test and ∼9,600 compound sets demonstrated assay stability, robustness, hit frequency, and reproducibility, confirming suitability of the assay for HTS. Single concentration screening of 699 confirmed ERAP1 binders from an affinity selection mass spectrometry experiment reproducibly revealed potent and weak inhibitors. The confidence in hit identification was equal to the established RF MS assay. The pIC50 values, obtained from subsequent dose-response curve screening of the hits, correlated between the two platforms and showed activity differences arising from the enzymatic assays. This study showed that the novel MALDI-TOF MS platform performance was comparable to the already established RF MS setup by providing several advantages. MALDI-TOF MS uses lower sample volumes, allowing greater miniaturisation, and it also has a shorter sample acquisition time compared to RF MS. In addition, the assay with the modified peptides reduced the required reagent concentrations and thereby lowered the tight binding limit. Moreover, no in-source fragmentation of the basic peptides was observed in MALDI-TOF MS, in contrast to the RF MS setup. The findings described here show a robust comparison of two label-free MS platforms for biochemical screening. Moreover, we describe the first label-free HTS MALDI-TOF MS campaign to identify ERAP1 inhibitors, adding to the growing number of targets that can be investigated by MALDI-TOF MS, including kinases, phosphatases, methylases, trimethylamine-lyase, deubiquitylases, and ubiquitin E3-ligases. All of these first in line assays provide an insight into assay design and suitability. MALDI-TOF MS is also a potential tool to address difficult targets, as recently shown by Winter et al. who included post-reaction analyte derivatisation to analyse transferase activity.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: L.M. received a joined studentship from the EPSRC (EPSRC CDT MoSMed; EP/S022791/1) and GSK. R.S.A., A.K.B., J.P.H., M.V.L., S.P., R.E.P.-H., J.M.Q., J.E.R., C.L.T. completed everything as work-for-hire for the employer GSK. M.E.D. is a Marie Sklodowska Curie Fellow within the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 890296.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
We thank EPSRC MoSMed CDT for the studentship award to L.M. We thank the members of the GSK screening, profiling, and mechanistic biology (SPMB) team for the initial development of the RF MS based ERAP1 assay and their ongoing support, in particular Ben Allsop, Laurie Gordon, and Michelle Pemberton. Furthermore, we would like to thank everyone from GSK and associated companies involved in earlier R&D stages supportive of this work and the reagent supply.
Due to GSK proprietary ownership of the compound libraries used within this study, the supporting data cannot be made openly available.
Differential analyte derivatization enables unbiased MALDI-TOF-based high-throughput screening: a proof-of-concept study for the discovery of catechol-o-methyltransferase inhibitors.
High-throughput matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry-based deubiquitylating enzyme assay for drug discovery.
Improved matrix-assisted laser desorption/ionization mass spectrometric analysis of tryptic hydrolysates of proteins following guanidination of lysine-containing peptides.
Hit discovery for public target programs in the european lead factory: experiences and output from assay development and ultra-high-throughput screening.
SLAS DISCOVERY: Advancing the Science of Drug Discovery.2020; 26: 192-204
13707-X&id=gr1.jpg){kind=link}
13707-X&id=gr2.jpg){kind=link}
13707-X&id=gr3.jpg){kind=link}
13707-X&id=gr4.jpg){kind=link}
13707-X&id=gr5.jpg){kind=link}
13707-X&id=gr6.jpg){kind=link}