Research Article Received: 14 February 2014

Revised: 14 April 2014

Accepted: 16 April 2014

Published online in Wiley Online Library

Rapid Commun. Mass Spectrom. 2014, 28, 1561–1568 (wileyonlinelibrary.com) DOI: 10.1002/rcm.6925

Application of the Exactive Plus EMR for automated protein–ligand screening by non-covalent mass spectrometry Hannah J. Maple1, Olaf Scheibner2, Mark Baumert3, Mark Allen3, Richard J. Taylor1, Rachel A. Garlish1, Maciej Bromirski2 and Rebecca J. Burnley1* 1

UCB Celltech, 216 Bath Road, Slough SL1 4EN, UK Thermo Fisher Scientific, Hanna-Kunath-Str. 11, 28199 Bremen, Germany 3 Advion Ltd, Kao Hockham Building, Edinburgh Way, Harlow CM20 2NQ, UK 2

RATIONALE: Non-covalent mass spectrometry (MS) offers considerable potential for protein-ligand screening in drug discovery programmes. However, there are some limitations with the time-of-flight (TOF) instrumentation typically employed that restrict the application of non-covalent MS in industrial laboratories. METHODS: An Exactive Plus EMR mass spectrometer was investigated for its ability to characterise non-covalent protein–small molecule interactions. Nano-electrospray ionisation (nanoESI) infusion was achieved with a TriVersa NanoMate. The transport multipole and ion lens voltages, dissociation energies and pressure in the Orbitrap™ were optimised. Native MS was performed, with ligand titrations to judge retention of protein-ligand interactions, serial dilutions of native proteins as an indication of sensitivity, and a heterogeneous protein analysed for spectral resolution. RESULTS: Interactions between native proteins and ligands are preserved during analysis on the Exactive Plus EMR, with the binding affinities determined in good agreement with expected values. High spectral resolution allows baseline separation of adduct ions, which should improve the accuracy and limit of detection for measuring ligand interactions. Data are also presented showing baseline resolution of glycoforms of a highly glycosylated protein, allowing binding of a fragment molecule to be detected. CONCLUSIONS: The high sensitivity and spectral resolution achievable with the Orbitrap technology confer significant advantages over TOF mass spectrometers, and offer a solution to current limitations regarding throughput, data analysis and sample requirements. A further benefit of improved spectral resolution is the possibility of using heterogeneous protein samples such as glycoproteins for fragment screening. This would significantly expand the scope of applicability of non-covalent MS in the pharmaceutical and other industries. Copyright © 2014 John Wiley & Sons, Ltd.

Non-covalent mass spectrometry has frequently been presented as an attractive screening technique for drug discovery programmes.[1–5] The advent of fragment screening[6] (i.e. screening small chemical compounds for binding to a target protein in vitro), in particular, has boosted the demand for novel biophysical screens capable of detecting and quantifying a range of interaction strengths, including weak binding.[7] The challenge has been to find a sufficient number of orthogonal screens that offer a reasonable balance between throughput, sample consumption and information content.[7,8] The main advantages of non-covalent or ’native’ MS (MS of proteins in a native-like state, maintaining non-covalent interactions) as a screening tool have been succinctly described as: speed, sensitivity and stoichiometry.[9,10] Despite these attractive facets, native MS has not seen widespread uptake as a screening tool within the pharmaceutical industry, and the reason for this perhaps also lies within the ’speed, sensitivity and stoichiometry’ paradigm.

Rapid Commun. Mass Spectrom. 2014, 28, 1561–1568

Copyright © 2014 John Wiley & Sons, Ltd.

1561

* Correspondence to: R. J. Burnley, UCB Celltech, 216 Bath Road, Slough SL1 4EN, UK. E-mail: [email protected]

Mass spectrometry data can be acquired relatively rapidly (typically 30 seconds to 1 minute per sample), representing a ’medium-throughput’ screen[5] in comparison with ’highthroughput’ techniques such as surface plasmon resonance (SPR), which can deal with entire fragment libraries of several thousand compounds in a rapid, automated approach. In order to increase the ’speed’ and thus the capacity of an MS-based screen, the data acquisition time would need to be reduced and ideally compounds would be screened as mixtures rather than individually. Mixing several compounds prior to analysis to increase throughput is currently challenging due to the tailing peaks and adducting often associated with native mass spectra, particularly for larger proteins. Sample preparation is also a bottleneck, as formation of adducts from buffer components can impose strict requirements for buffer exchange and desalting. Mass spectrometry is an inherently sensitive technique, and offers benefits in terms of sample consumption over other techniques such as saturation-transfer difference- (STD-) NMR and isothermal titration calorimetry (ITC).[5,11] It does not, however, rival SPR in terms of sample consumption. The concentration of protein and ligand used in ESI-MS analysis is directly related to the presence of non-specific

H. J. Maple et al. gas-phase binding,[12] thus, a decrease in the limit of detection would allow this artefact to be minimised by using lower protein concentrations. Although direct determination of the stoichiometry of the ligand–protein complex is highly valuable, extraction of these data is limited by the spectral resolution in mass-to-charge ratio (m/z) attainable for the multiply charged species. The limiting factor for resolving apo/holo protein species is not the inherent instrumental resolution, but rather the ability to decluster and desolvate the protein ions effectively. Here, we differentiate this as ’spectral resolution’, referring to the observed peak width of individual protein charge states including unresolved adducts. In particular, proteins containing complex modifications (e.g. glycosylation or heterogeneous labelling) are not currently amenable to screening by native MS due to poor spectral resolution of multiple species with similar molecular weights (MWs). Recently, it has been shown that by modifying an Exactive Plus Orbitrap mass spectrometer (Thermo Fisher Scientific, Bremen, Germany), it is possible to perform native MS of large proteins and protein complexes from intact IgG antibodies (150 kDa) to GroEL (800 kDa).[13,14] The resulting spectra offered substantially higher spectral resolution and sensitivity than those obtained by time-of-flight (TOF) instrumentation. Complete or near-complete desolvation, plus the ability to resolve adducts (e.g. Na+) completely from the main peaks, improved both the spectral resolution and the mass accuracy achievable. These factors make the instrument potentially very attractive for protein-ligand binding applications. Effective desolvation and transmission of very large protein ions, however, require the use of higher pressures and accelerating voltages throughout the mass spectrometer.[13,15,16] Such conditions are often detrimental to the analysis of smaller native proteins, and protein–ligand complexes, due to the increased likelihood of collisional activation leading to gas-phase unfolding and ligand dissociation. In this work we investigate whether instrumental parameters on the Exactive Plus EMR (coupled to a TriVersa NanoMate™; Advion, Harlow, UK) can be adjusted for the analysis of weak protein–small molecule interactions, and evaluate the scope and potential for this instrument as a tool in drug discovery programmes. Since TOF mass analysers are typically, although not exclusively, used for these types of analyses, comparisons are made with TOF data throughout.

EXPERIMENTAL Details of sample preparation, binding measurements and sensitivity measurements are provided in the Supporting Information. Exactive Plus EMR MS analysis

1562

The Exactive Plus EMR Orbitrap™ mass spectrometer (based on the modified Exactive Plus described previously[13,14,17]) was used in conjunction with a TriVersa NanoMate, a chip-based system for automated static nanoESI. A volume of 2 μL of sample was introduced using an Advion spray chip with 5 μm diameter nozzles, which deliver a sample flow rate of approx. 100 nL/min. The NanoMate was operated in

wileyonlinelibrary.com/journal/rcm

positive ion mode with a chip nozzle voltage of approx. 1.75 kV and a spray pressure of 0.5 psi. A schematic of the mass spectrometer, indicating the parameters optimised, is shown in Fig. S1 (Supporting Information). By tuning DC voltages on the ion lenses and transport multipoles between the source and C-trap, ions of higher m/z values can be transmitted preferentially. Increasing the pressure in the HCD cell (collision cell) and the Orbitrap analyser can improve the MS signal obtained for large or native proteins, and applying fragmentation energies increases desolvation. For larger protein ions, lower resolution settings (R = 8750 at m/z 200) are used to capture the signal but not to accumulate noise. For these analyses, the following parameters were optimised (see also Table 1). The capillary temperature used was 150°C. The HCD and in-source dissociation energies were turned off. The S-lens RF was decreased from 200 V to 100 V. The transfer lens RF voltages were altered; of these, the transfer multipole DC in particular affects the transmission of different protein ions. The pressure readback in the Orbitrap cell was 2.3 × 10 10 mbar. The resolution setting was varied between 8750 and 140 000 (FWHM at m/z 200), with R = 8750 giving better signal-to-noise, and isotopic resolution of the protein ions attainable at R >35 000. LCT Premier MS analysis For compound binding and sensitivity tests, a modified LCT Premier time-of-flight mass spectrometer (Waters Corp., Manchester, UK) was used in conjunction with a TriVersa NanoMate chip-based nanoESI source (spray parameters as above). The tuning parameters were carefully adjusted to minimise in-source dissociation. The most influential parameters were: aperture 1 voltage (25 V); ion guide 1 voltage (50 V), cone voltage (20 V) and desolvation temperature (20°C).

Table 1. Mass spectrometer parameters used for native analysis of lysozyme or Bcl-xL m/z range: Capillary temperature (°C): S-Lens RF level: S-Lens voltage (V): Skimmer voltage (V): Injection flatapole offset (V): Bent flatapole DC (V): Transfer multipole DC (V): HCD multipole DC (V): Inter flatapole DC (V): Flatapole exit DC (V): MP2 and MP3 RF (V): Gate lens voltage (V): C-Trap RF (V): In-source dissociation voltage (V): HCD energy (eV): UHV sensor (mbar): Ion injection time (ms): Microscan count: FT resolution:

400–15000 150 100 (200) 25 15 7 6 4 (6) 0 6 0 990 5 2950 0 (25) 0 2.31e 10 2 (20) 5 (10) 8000*

Numbers in parentheses indicate values used for protein X. *Resolution setting used for monitoring ligand binding; value varied for other analyses.

Copyright © 2014 John Wiley & Sons, Ltd.

Rapid Commun. Mass Spectrom. 2014, 28, 1561–1568

Protein–ligand screening by automated higher resolution MS A Speedivalve (Edwards Ltd, Crawley, UK) connected between the rotary pump and source pumping line allowed manual adjustment of instrument pressures,[18] with the source backing pressure typically being adjusted between 2.4 and 5 mbar. Synapt G2 MS analysis For the sensitivity tests, the Waters Synapt G2 TOF mass spectrometer was also used in conjunction with a TriVersa NanoMate chip-based nanoESI source (spray parameters as above). The instrument was operated in sensitivity TOF mode, with cone voltage 35 V, extraction cone voltage 1.2 V, desolvation temperature 150°C, source backing pressure 2.7 mbar, and m/z range 50–5000, using a quadrupole MS profile of: mass 1 = 1000, dwell time 20%, ramp time 20%; mass 2 = 2000, dwell time 20%, ramp time 40%, mass 3 = 4000.

RESULTS Automated native mass spectral analysis of protein-ligand complexes with an Exactive Plus EMR Mass spectrometry of proteins and protein complexes in a native-like state requires a ’gentle’ ionisation procedure, whereby the internal energy of the protein ions is minimised. This is critical for the application of native MS to fragment screening where weak binding interactions (KD = high μM–mM) are to be expected. A model system used in-house for calibrating these

instrumental tuning parameters is lysozyme:NAGn (NAG = N-acetylglucosamine, n = number of sugar units).[5] Three saccharide ligands to lysozyme are used, NAG3, NAG2 and NAG, which have binding affinities spanning three orders of magnitude (KD ca 10 μM to 1 mM).[5] The instrumental parameters were tuned as described in the Experimental section. The S-lens RF voltage was reduced and the transfer lens RF voltages altered, with the transfer multipole voltage in particular having an effect on the dissociation of the compound from the protein–ligand complex ion. The parameters used are given in Table 1. Calibration of the m/z scale was with CsI clusters, and a mass accuracy of

Application of the Exactive Plus EMR for automated protein-ligand screening by non-covalent mass spectrometry.

Non-covalent mass spectrometry (MS) offers considerable potential for protein-ligand screening in drug discovery programmes. However, there are some l...
852KB Sizes 0 Downloads 4 Views