Francesco Fiorentino 1,⇑, Dante Rotili 1,⇑, Antonello Mai 1,2
1 Department of Drug Chemistry and Technologies, Sapienza University of Rome,
Piazzale Aldo Moro 5, 00185 Rome, Italy 2 Pasteur Institute, Cenci-Bolognetti Foundation, Sapienza University of Rome,
Piazzale Aldo Moro 5, 00185 Rome, Italy Native mass spectrometry (nMS) is a biophysical method for studying protein complexes and can provide insights into subunit stoichiometry and composition, protein–ligand, and protein–protein interactions (PPIs). These analyses are made possible by preserving non-covalent interactions in the gas phase, thereby allowing the analysis of proteins in their native state. Consequently, nMS has been increasingly applied in early drug discovery campaigns for the characterization of protein– drug interactions and the evaluation of PPI modulators. Here, we discuss recent developments in nMS-directed drug discovery and provide a timely perspective on the possible applications of this technology in drug discovery.
Keywords: native mass spectrometry; protein–ligand interactions; protein–protein interactions; PROTAC; membrane proteins
Introduction Proteins are a fundamental part of living organisms, given their involvement in all cellular pathways. They control all aspects of life, including, but not limited to, regulation of gene expression, catalysis of biochemical reactions, signaling, and transportation of molecules essential for cellular metabolism.1 Therefore, non- covalent interactions between proteins and their partners (including other proteins, nucleic acids, lipids, carbohydrates, and small molecules) are at the core of myriad biological processes, such as transcriptional regulation, cell differentiation, immune response, cell adhesion, and inflammation. These interactions are governed by a variety of forces, including hydrogen bonds, electrostatic interactions, Van der Waals, and hydrophobic forces.2 Importantly, the characterization of protein–ligand inter- actions is at the core of the drug discovery process. In fact, elucidation of protein structures and dynamics and the quantification of the kinetic and thermodynamic
Francesco Fiorentino graduated in medici- nal chemistry from Sapienza University of
Rome in 2016. He received his PhD in bio- physical chemistry from the University of
Oxford in 2020 under the supervision of Dame Carol Robinson, working on the elu- cidation of the structure and regulation of membrane proteins using mass spectrom- etry. Following a 1-year postdoc in the same lab, he joined the Mai group at
Sapienza University of Rome as a postdoc- toral researcher. His research focuses on the application of native mass spectrometry and other biophysical techniques to investigate the protein complexes involved in the epigenetic regulation of cellular homeostasis and in bacterial membrane biogenesis.
Dante Rotili graduated in medicinal chem- istry from Sapienza University of Rome in
2003. He received his PhD in pharmaceuti- cal sciences from the same University in
2007. From 2009 to 2010, he was a research associate in the Department of Chemistry,
University of Oxford, where he worked in collaboration with Chris Schofield in the development of chemoproteomic probes for the characterization of 2-oxoglutarate- dependent enzymes.
In 2020, he was appointed an associate professor of medicinal chemistry at
Sapienza University of Rome. His research focuses mainly on the development of modulators of epigenetic enzymes with potential applications in cancer, neurodegenerative, metabolic, and infec- tious diseases.
Antonello Mai graduated in pharmacy from Sapienza University of Rome in 1984, from where he received his PhD in 1992 in pharmaceutical sciences under the super- vision of
M.
Artico.
In 1998, he was appointed associate professor of medicinal chemistry and, in 2011, full professor of medicinal chemistry at Sapienza University of Rome. His research interests include the synthesis and biological evaluation of new bioactive small-molecule compounds, in particular modulators of epigenetic targets, for use as chemotherapeutic agents against cancer, metabolic disorders, neurodegenerative diseases, and parasitic infections. In addition, he works in the fields of antibacterial/antimycobacterial, antiviral, and central nervous system agents.
Native mass spectrometry-directed drug discovery: Recent advances in investigating protein function and modulation
⇑Corresponding authors: Fiorentino, F. (f.fiorentino@uniroma1.it), Rotili, D. (dante.rotili@uniroma1.it).
Drug Discovery Today d Volume 28, Number 5 d May 2023
REVIEWS KEYNOTE (GREEN) 1359-6446/ 2023 The Authors. Published by Elsevier Ltd. https://doi.org/10.1016/j.drudis.2023.103548This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). www.drugdiscoverytoday.com
1 parameters of protein–ligand binding constitute the earliest steps in the development of drugs that could be used to treat various pathologies.3
To date, a large toolbox of biophysical technologies is used during preclinical drug development to characterize protein–li- gand interactions in vitro, each with its own set of strengths and shortcomings. These include isothermal titration calorime- try (ITC),4 surface plasmon resonance (SPR),5 nuclear magnetic resonance (NMR) spectroscopy,6 and X-ray crystallography.7
Despite the tremendous advances obtained by using these tech- niques, significant technological problems in drug development persist. Indeed, some protein species, such as dynamic protein oligomers and membrane protein complexes, remain difficult to target despite their enormous therapeutic potential. The tran- sient nature of certain protein complexes, the instability of mem- brane proteins, and the possible lack of any enzymatic activity make these targets extremely challenging to investigate for drug development. Compounds that influence the function of these complicated targets cannot be easily evaluated by typical bio- physical methods, which provide information on the average solution properties of the system and might not inform on single binding events or ligand influence on oligomeric states. More- over, these approaches often require protein labeling or immobi- lization, which may alter the subtle equilibria being evaluated.
Hence, complementary methodologies that increase our under- standing of structural, thermodynamic, and kinetic features of drug action could accelerate preclinical drug development. nMS is a promising methodology to investigate protein–li- gand interactions that also allows the assessment of more chal- lenging targets. It is a technique that enables the preservation of non-covalent interactions and quaternary protein structure in the gas phase. Moreover, unlike conventional denaturing
MS, which often includes organic solvents and a low pH, nMS uses volatile, aqueous buffers at near-physiological pH and gen- tler voltages and ion transfer tube temperatures (usually
<200 C) compared with classic MS techniques. These differences with respect to other MS approaches contribute to the preserva- tion of the native or native-like structures of the studied protein complexes.8 This enables the quantification of dissociation con- stants for PPIs and protein–ligand interactions, the assessment of small-molecule mode of action, and the study of protein com- plex stoichiometry and subunit architecture and dynamics (Figure 1).9 nMS takes advantage of the milder nature of electrospray ion- ization (ESI) compared with other ionization methods [e.g., matrix-assisted laser desorption/ionization (MALDI), atmo- spheric pressure chemical ionization (APCI), or electron ioniza- tion (EI)] to transfer protein complexes from volatile buffered aqueous solutions into the gas phase and maintain the non- covalent interactions during this process.10–12 The most common ionization method used in nMS is nanoESI (nESI), a subtype of
ESI characterized by a smaller diameter of the capillary needle (usually 1–5 lm in nESI, whereas it reaches up to 0.5 mm in
ESI) orifice and very slow flow rates (10–50 nl/min),13 which
Buffer exchange + Ligand Sample loading (1-3 µL) nESI
Mass spectrometer 100 3000 3500 4000 4500 m/z 0 % 9+
8+ 10+ 11+ 2500 3000 3500 4000 4500 m/z 100 0 % 100
0 % 2000 3000 4000 5000 m/z 6000 9+ 6+ 13+ 12+ 11+
8+ 10+ 7+ 14+ 6+ 7+ 20 µM 0 µM 10 µM Protein–ligand interaction evaluation
Assessment of protein–protein interaction modulators
Assessment of PROTAC efficacy 20 µM 0 µM Drug Discovery Today
FIGURE 1 Typical workflow for the analysis of ligand binding via native mass spectrometry (nMS). The protein solution is first buffer exchanged into a water solution containing a volatile salt (usually ammonium acetate) using size exclusion chromatography (SEC) or dialysis. This step can be avoided when using nanoscale emitters. The protein is then incubated with increasing concentrations of the ligand of interest, and the mixture is injected into gold-coated emitters for mass spectrometric analysis. The resulting spectra show two peaks, one for the apo protein and one for the ligand-bound complex in the case of protein–ligand binding analysis. In the case of protein–protein interaction (PPI) modulators, the spectra will inform on the influence of the compounds on the assembly of multiprotein complexes. In the case of proteolysis-targeting chimeras (PROTACs), the spectra will show the formation of the protein–PROTAC–E3 ligase ternary complex.
KEYNOTE (GREEN) KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
2 www.drugdiscoverytoday.com make nESI more amenable for studying protein complexes. In fact, given the smaller orifice, the initial droplet size is at least one order of magnitude smaller, thus reducing the energy and the number of fission events necessary to yield gas-phase ions.8
Decades of advances in mass analyzers and detectors made it pos- sible to transmit and detect intact proteins. Further develop- ments, including manipulation of pressure gradients within the instruments, enable the preservation of non-covalent interac- tions in the gas phase within the instrument.14,15 Moreover, the integration of nMS with multistage tandem MS (MSn) approaches enables the identification of unknown ligands bound to the target protein. In tandem MS experiments, the precursor ions are mass selected (e.g., in a quadrupole or ion trap) and then subjected to a specific type of activation [e.g., via collision- induced dissociation (CID) or electron-capture dissociation (ECD)]. Following dissociation, the ions are then analyzed according to their m/z. Depending on the used instrument, tan- dem MS experiments can be performed multiple times, thereby allowing for MSn experiments, which can then be used to iden- tify ligands dissociated from their protein partners based on their fragmentation pattern. In this regard, ground-breaking work by
Klassen et al. paved the way for both the evaluation of protein– ligand binding via nMS and the use of MSn experiments in drug discovery.16–19
The main advantages of nMS compared with other biophysi- cal techniques include the absence of any label or protein immo- bilization, low sample consumption (picomoles of protein are used for each measurement), direct measurement of interactions, and the capacity to distinguish distinct protein species within a heterogeneous population.9,20 Given these features, nMS is rec- ognized as a highly informative technology for early drug discov- ery campaigns. In particular, nMS provides assistance for NMR,
X-ray crystallography, and cryogenic electronic microscopy (cryo-EM) studies because it allows evaluation of the purity, homogeneity, and integrity of protein samples, as well as the measurement of complicated stoichiometries.9,21–23
Similar to other techniques, nMS has not only advantageous properties but also limitations, such as the presence of possible false negatives (gas-phase breakdown of hydrophobic interac- tions) and false positives (nonspecific binding). These constraints can be solved by optimizing sample preparation or nESI approaches24–26 and by using appropriate statistical methods to account for nonspecific interactions developing during the nESI process.16,27
Overall, nMS complements the existing portfolio of in vitro characterization methods and is considered a viable method for identifying, validating, and characterizing hit/lead compounds.
This application has been bolstered by advances in instrument technology, which facilitate the examination of bigger, more complicated protein complexes. Indeed, technological advances have resulted in ultra-high mass range spectrometers, which allow for the investigation of complexes formed between small molecules and large proteins, hence expanding the dynamic range of nMS.28–33 Moreover, the optimization of solubilization methods of membrane proteins is vital for the observation of intact membrane protein complexes in a mass spectrometer,34–
36 and has allowed the analysis of interactions between small molecules and membrane proteins,29,37–39 including G-protein- coupled receptors (GPCRs).24,40–42 Finally, nMS coupled with variable temperature nESI enabled the determination of thermo- dynamic parameters [Gibbs free energy upon binding (DG), enthalpy (DH), and entropy (-TDS)] of protein–ligand interac- tions and PPIs, which are important parameters to consider dur- ing drug development.43
In this review, we discuss the most recent applications of nMS for the analysis of protein–ligand interactions, the assessment of
PPI modulators, and the evaluation of the so-called ‘proteolysis- targeting chimeras’ (PROTACs) (Figure 1). We describe the wide applicability of nMS in early drug discovery and highlight the new advancements, future opportunities, as well as challenges of nMS for investigating protein function and modulation (Table 1).
Applications of native MS in drug discovery Protein–ligand interaction evaluation nMS coupled to nESI has been widely used to detect and quantify protein–ligand interactions, including small molecules, peptides, and, particularly in the case of membrane proteins, lipids.22,41,44–
48 Nonetheless, care should be taken when analyzing native mass spectra displaying non-covalent protein–ligand interactions and appropriate data analysis and statistical approaches should be used to account for nonspecific interactions arising during the
ESI process.27 For a complete overview of the accurate determina- TABLE 1
Summary of the advantages and challenges for the different applications of nMS in drug discovery.
Application Advantages Challenges Protein–ligand interaction evaluation
Calculation of KD values with minimal sample consumption
Observation of single binding events and quick identification of allosteric mechanisms
Assessment of protein–ligand interactions for multiple compounds in a single experiment
Experiments still relatively low throughput without use of chip-based nESI platforms
Possible experimental false negatives (gas-phase breakdown of hydrophobic interactions) or false positives (nonspecific binding)
Assessment of PPI modulators and PROTACs Observation of protein complex stoichiometry and compound- mediated disruption/stabilization with no need for labeling or immobilization
Evaluation of allosteric effects and cooperativity in context of protein oligomerization or multiprotein complex formation
Minimal sample consumption Experiments still relatively low throughput without use of chip-based nESI platforms
Multiprotein complexes in equilibrium with their subcomplexes or subunits can yield complex spectra that are difficult to interpret
KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
KEYNOTE (GREEN) www.drugdiscoverytoday.com 3 tion of protein–ligand interactions using nMS, we refer the reader to recent reviews by Bennett and colleagues49 and Gavriili- dou et al., with the latter focusing primarily on high-throughput nMS for drug screening.50 nMS can successfully evaluate protein–ligand interactions in a reasonably medium-throughput way (100 ligands per hour) using a procedure known as bioaffinity MS. This approach involves the incubation of a ligand or combination of ligands with a protein of interest, with the resultant solution being examined directly by nMS. The identification of bound small molecules may be determined easily by measuring the difference in the mass-to-charge ratio (m/z) between the ligand-bound and apo protein peaks (Figure 1). The use of nMS for compound library screening is gaining increasing interest, as exemplified by the studies by Vu et al.51 and Nguyen et al.,52 which we discuss below. Hence, chip-based automated nESI platforms, such as the
NanoMate,53 are being increasingly used because they provide constant nESI conditions and greater reproducibility, along with augmented throughput. An example of this approach was pro- vided by Vu et al., who applied nMS to screen a library of natural product fragments against 62 Plasmodium falciparum proteins selected as potential malaria drug targets. Each protein was incu- bated with fragments at ligand:protein molar ratios varying from
5:1 to 20:1. This approach enabled the identification of 96 poten- tial binders, which formed complexes with 32 proteins. The 96 natural product fragments had different chemotypes and were different from the known antimalarial aminoquinolines, quino- lones, or diamidines. Interestingly, the technique managed to distinguish between promiscuous and nonpromiscuous pan- assay interference compounds (PAINS), such as polyphenols, epoxides, Michael acceptors, and b-lactams, thereby confirming the ability of nMS to capture specific interactions.51
More recently, Nguyen and coworkers integrated nMS and untargeted metabolomics to develop a method to identify natu- ral products that interact with potential drug targets.52 Specifi- cally, they incubated crude natural product extracts containing thousands of small molecules with the target proteins and applied a quick, low-volume gel filtration step using 0.5-ml desalting columns to remove unbound ligands. This was fol- lowed by nMS analysis to capture and quantify protein–ligand interactions (Figure 2a) and by the identification of the bound hits through metabolomics. To do so, they used nanoscale nESI emitters (internal diameter of 250 nm), which, differently from the widely used microscale emitters, are tolerant of the presence of salt and enable the analysis of complex mixtures, such as those containing thousands of natural products.25 Through this approach, the authors identified novel and known binders of human carbonic anhydrases (hCA), a family of metalloenzymes that catalyze CO2 hydration,54 which is attracting attention because it is implicated in the initiation and progression of sev- eral pathologies, including cancer.55 To determine the identity of the bound ligands, they performed MSn analysis with a linear ion trap (Figure 2a). This approach, already exploited for the identification of ligands bound to either soluble or membrane proteins,28,56 involved the isolation of the ligand-bound protein followed by dissociation and fragmentation of the ligand to gain a fragmentation spectrum (Figure 2a). This led to identification of 1,2,3,4,6-penta-O-galloyl-b-D-glucose (Figure 2b) as a new ligand of hCAI and the binding was later confirmed in nMS- based titration experiments, yielding a KD value of 27 lM.25
Agasid et al. also used nanoscale nESI emitters (internal diam- eter of 100 nm) for the analysis of GPCRs in buffers containing high concentrations of sodium ions.24 This allowed for the detec- tion of adenosine 2A receptor (A2AR) in lauryl maltose neopentyl glycol (LMNG) detergent micelles bound to up to seven sodium ions, as well as investigation of the effects of A2AR agonists (NECA and CGS21680) and antagonists (XAC and ZM241385) on sodium binding. Indeed, whereas both agonists abolished protein–sodium ion interaction, this was instead retained in the presence of both antagonists, in line with solution-based studies in which only inactive conformations maintain sodium ions in the allosteric binding pockets.57 The glucagon receptor was successfully observed in mixed micelles containing the recently developed first-generation, dendritic oligoglycerol deter- gent (G1)35,42 and cholesteryl hemisuccinate. The authors cap- tured glucagon binding, whereas no sodium adducts were observed this time. They also observed the binding of the nega- tive allosteric modulator NNC0666, which had been added dur- ing purification to stabilize the protein and the identity of which was confirmed via MSn on a Orbitrap-Ion Trap mass spectrometer.
Another interesting method used for the quantification of protein–ligand affinities was described by Ren and coworkers, who developed an online
SEC-nMS platform (Figure 2c).
Through this procedure, they calculated the KD values for the interaction of small-molecule inhibitors with the catabolic enzyme indoleamine 2,3-dioxygenase 1 (IDO1)58 and these val- ues were in line with chip-based53 nESI-nMS measurements. To do so, they incubated IDO1 with increasing concentrations (0.5–100 lM) of different ligands, followed by a 3-min SEC run coupled to a nESI-nMS system (Figure 2d, upper-left panel). Fur- thermore, one-shot competition experiments by mixing IDO1 with an equimolar mixture of two potential binders (called com- pounds 1 and 2 in the paper) indicated a higher abundance of
IDO1-1 adducts (Figure 2d, upper-right panel), in line with KD calculations. In addition, the authors analyzed protein–ligand interactions in the gas phase by performing in-source CID exper- iments. These are executed by increasing the collision voltage until full complex dissociation, allowing calculation of the
VC50 parameter (i.e., the collision voltage required to dissociate the protein–ligand complex by 50%), which was found to be independent from the protein:ligand ratio. Moreover, a higher
VC50 value was obtained for the more tightly bound ligand (com- pound 1), thus being in line with the measured KD values (Fig- ure 2d, lower panels). Overall, this method enabled accurate measurements in this system, with the main advantages relying on the automation and rapidity of the analysis and the mini- mization of the residence time in ammonium acetate. By con- trast, assessment of weak interactions might be hindered in cases in which the dissociation rate is fast enough to occur dur- ing the chromatographic separation. nMS has also been applied in conjunction with fluorescence thermal shift assays for the identification of new ligands of the serine/threonine kinase c-Jun N-terminal kinase 3 (JNK3).59 Fol- lowing identification of seven potential hits through fluores- cence thermal shift (FTS) assays, nMS was used to calculate the
KEYNOTE (GREEN) KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
4 www.drugdiscoverytoday.com binding affinities of these compounds, which have a diverse sub- set of scaffolds, including N-(1H-pyrazol-4-yl)thieno[2,3-d]pyri midin-4-amine (four compounds), 5-(phenylamino)-1H-1,2,3- triazole-4-carboxamide (one compound), triazolmethypiperidine (one compound), and piperidinecarboxamide (one compound). nMS successfully captured protein–ligand interactions for four of these compounds, with 5, bearing a 5-(phenylamino)-1H- 1,2,3-triazole-4-carboxamide core, being the most tightly bound ligand, with a KD value of 21 lM.
Recently, D’Amico et al. described an ion mobility-nMS (IM- MS) method coupled to droplet microfluidics for analyzing pro- tein–ligand interactions by assessing the influence of ligand
Drug Discovery Today FIGURE 2 Evaluation of protein-ligand interactions. (a) Schematic indicating the key steps necessary for the identification of small molecules bound to the target protein via native mass spectrometry (nMS) following incubation with extracts containing thousands of natural products. (b) Structure of 1,2,3,4,6-penta-O- galloyl-b-D-glucose. (c) Schematic of workflow used for size exclusion chromatography (SEC)-nMS. (d) Evaluation of the binding affinity between IDO1 and two small molecules via SEC-nMS (upper panels) along with measurement of the VC50 constant and demonstration that the VC50 value is independent from the protein:ligand ratio (lower panels). (e) Structure of darobactin A. (f) Mass spectra of the b-barrel assembly machinery (BAM) complexes in the presence of both cardiolipin (CL) and darobactin (DB) (left panel). Central panel: expansion of the 23+ charge state to focus on CL, darobactin, and CL + darobactin binding to BAM. Right panel: quantification of the darobactin-bound peak intensities relative to the corresponding darobactin-free species. Adapted, with permission, from 52 (a), 58 (d), and 66 (f).
KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
KEYNOTE (GREEN) www.drugdiscoverytoday.com 5 binding on protein stability in the gas phase.60 Before mass spec- trometric analysis, IM enables the separation of protein ions based on their size and shape within a drift tube filled with neu- tral gas under the influence of a weak electric field. The time that it takes for an ion to travel through the drift tube is proportional to its collision cross-section (CCS), which refers to the rotation- ally averaged projection area of the ion.61 Following optimization of the microfluidics system for sample introduction in the mass spectrometer, the authors performed collision-induced unfolding (CIU) assays on the protein lysine deacylase SIRT5, a promising target for the development of modulators that could serve as potential drugs in many pathologies.62–64 CIU experiments, com- prising analysis of the change in drift time in the IM cell as a func- tion of collision voltage, enable the detection of subtle changes in protein structure and stability.65 In the case of SIRT5, CIU finger- prints indicated the presence of three different states and two transition regions, corresponding to a compact native-like struc- ture and two unfolded states accessed during the CIU experiment.
Hence, CIU experiments were used to screen a small library of 96 small molecules against SIRT5. These assays are based on the notion that compound binding to the target protein increases its stability and, thus, higher voltages would be necessary to unfold the protein. This screen identified 24 compounds capable of increasing the median collision voltage necessary to reach both
CIU transitions (CIU50s), indicating that these molecules exert a stabilizing effect on SIRT5. Five of these molecules were also tested for their inhibitory potency, with the compounds exhibit- ing the greatest SIRT5 inhibitory activity also causing the largest shifts in CIU50 for the first transition, whereas the opposite trend was observed for the second transition. Based on the assumption that these compounds are
SIRT5-competitive inhibitors (although no data were reported in the paper), the authors sug- gested that the first transition represents the unfolding of the sub- strate binding domain, while the second transition may represent the unfolding of SIRT5 Zn2+-binding domain. However, further research will be required to corroborate these claims. Although this technology decreases data acquisition time and sample con- sumption, a significant decrease in sensitivity was found owing to the increased flow rates and emitter sizes necessary for interfac- ing with the microfluidic platform. Overall, these authors pre- sented a novel method for screening small molecules against protein targets, and it would be interesting to extend its applica- tion to other protein systems.
An intriguing example of the application of nMS to mem- brane protein–ligand binding as provided by Kaur et al., who cap- tured the interaction between the b-barrel assembly machinery (BAM) complex and the recently discovered peptide-based antibiotic darobactin A (Figure 2e).66 The BAM complex, which comprises five subunits, including the outer membrane (OM) protein BamA and the lipoproteins BamB–E, promotes the fold- ing and insertion of OM proteins in the OM of Gram-negative bacteria. nMS experiments indicated that the BAM complex pref- erentially interacts with negatively charged lipids, such as phos- phatidylglycerol (PG) and cardiolipin (CL). Interestingly, CL binding to the BAM complex was shown to enhance its affinity for darobactin A, as demonstrated by the spectra indicating that darobactin preferentially binds to Bam-CL compared with the apo BAM complex (Figure 2f).
Finally, nMS has also been applied for the analysis of protein– ligand interactions directly from cell lysates and enabled evalua- tion of overexpressed proteins directly from crude samples with- out purification.67
Rogawski and colleagues applied this approach to the kinase domain of Bruton’s tyrosine kinase (BTK_KD) overexpressed in HEK293T cells. They first incubated
BTK_KD-overexpressing cells with either the covalent drug ibru- tinib or its non-covalent counterpart, ibrutinib-NH2, then resus- pended the cell pellets in 150–300 mM ammonium acetate and lysed the resulting suspension, which was centrifuged and fur- ther diluted before nMS analysis. Through this method, the authors captured the interaction between BTK_KD and both ibrutinib and ibrutinib-NH2. The same approach was then trans- lated to other BTK ligands (pluripotin, LY2409881, vemurafenib, and PP-121) and enabled the authors to rank their binding affini- ties, with pluripotin being the tightest binder. Overall, this method has the potential to be used to validate hit/lead com- pound cellular target engagement, although it still requires pro- tein overexpression, which could alter the physiological state of the studied cellular system and impair off-target binding iden- tification. Nonetheless, it is a promising orthogonal approach to conventional assays, such as in-cell NMR and cellular thermal shift assays (CETSA), because, unlike these assays, it allows for direct observation of protein–ligand binding and could be used to evaluate drug combinations. Along the same line, Olinares et al.68 previously described a workflow that enables the analysis of protein assemblies directly from cell lysates, without the requirement for overexpression, by combining affinity isolation with antibody-conjugated beads and nMS. Therefore, it would be intriguing to combine the methods described by Rogawski et al. and by Olinares et al. to explore the influence of small mole- cules on protein complexes.
Assessment of protein–protein interaction modulators nMS has also been extensively used for evaluation of both dis- ruptors and stabilizers of PPIs. To this end, the Robinson group applied nMS to both soluble and membrane proteins to clarify the mode of action of known drugs and to support the develop- ment of novel potential PPI disruptors.69–71 For instance, this approach was applied by Fiorentino and coworkers to the Pseu- domonas aeruginosa lipopolysaccharide (LPS) transport (Lpt) pro- tein LptH to gain insights into its monomer/dimer equilibrium and PPI disruption via small molecules.69 The Lpt system is a multiprotein complex formed by seven different proteins respon- sible for LPS transport from the inner membrane to the OM.72
Among them, the periplasmic protein LptA forms an oligomeric bridge connecting the two membranes and is pivotal for the translocation pathway. In the multidrug-resistant opportunistic pathogen P. aeruginosa, the ortholog of LptA, LptH, forms dimers in solution.73 Using nMS, the authors quantified the monomer– dimer equilibrium of LptH and assessed the potency and efficacy of the antimicrobial peptide thanatin and small-molecule disrup- tors, obtaining information on their structure–activity relation- ships (SARs). Specifically, nMS experiments indicated that LptH exists mainly as a dimer, which is disrupted by thanatin as well as the known LPS transport inhibitor IMB-881.69 Based on its structure, the authors evaluated a library of
5-carboxy-8- hydroxyquinoline derivatives bearing a small acyl side chain.
KEYNOTE (GREEN) KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
6 www.drugdiscoverytoday.com nMS experiments revealed that short (up to five carbons) acyl chains had higher disruption activity, and that protecting car- boxyl and hydroxyl groups via methylation and methoxymethy- lation, respectively, increased compound activity (Figure 3a).69
This study presented a new method for evaluating LptA/H PPI disruptors and led to the identification of new quinoline-based hit compounds, which represent the basis for the development of novel LPS transport inhibitors.
A similar approach was used to evaluate the effects of allos- teric inhibitors on the dimerization of the severe acute respira- tory syndrome-coronavirus
2 (SARS-CoV-2) main protease (Mpro). Mpro is the enzyme responsible for cleaving along the two viral polypeptides to release the nonstructural proteins nec- essary for replication. Through nMS, El-Baba et al. showed that
Mpro is preferentially a dimer in solution and then analyzed the effect of four fragments74 on the monomer–dimer equilibrium.70
Notably, one compound (x1187), which had been indicated to bind to the dimer interface,74 was shown to disrupt the dimeric assembly (Figure 3b). Conversely, the three remaining com- pounds, which had previously been shown to interact with a solvent-exposed surface,74 had no effect on Mpro dimerization.
Further nMS-based kinetic experiments indicated that the frag- Drug Discovery Today
FIGURE 3 Assessment of protein-protein interaction (PPI) disruptors. (a) Mass spectra showing the effect of the quinoline derivative 2e on LptH monomer/dimer equilibrium and relative quantification (right panel). (b) Native mass spectra of Mpro (5 lM) in the presence of increasing equivalents of x1187. (c) Upper panel: native mass spectrum of Mpro (5 lM) in the presence of the 11-mer substrate (50 lM) at t = 30 s. Peaks labeled ‘TSAVLQ’ and ‘+substrate’ indicate acyl– enzyme complex and the non-covalent enzyme–substrate complex, respectively. Central panel: mass spectra of the 15+ charge state at three representative times showing the substrate cleavage reaction. Inset: plot of the relative abundance of the enzyme-substrate complex as a function of time. Lower panel: bar chart indicating the half-lives of the enzyme-substrate complex in the presence of the tested small molecules. (d) Structure of ARS-1620 (upper panel). Mass spectra recorded following addition of ARS-1620 (10 lM) to preincubated mixtures of 1 lM SOScat with 3 lM of KRasG12C–GDP (central panel) or KRasG12C–
GTP (lower panel). (e) Structure of Kobe0065 (upper panel). Mass spectra recorded following addition of Kobe0065 (2.5 lM) to preincubated mixtures of
1 lM SOScat with 3 lM KRas–GTP (central panel) or KRasG13D–GTP (lower panel). (f) Structure of BAY-293 (upper panel). Mass spectra recorded following addition of BAY-293 (2.5 lM) to preincubated mixtures of 1 lM SOScat with 3 lM KRas–GTP (central panel) or KRasG13D–GTP (lower panel). (g) Structure of BI- 3406 (upper panel). Mass spectra recorded following addition of BI-3406 (2.5 lM) to preincubated mixtures of 1 lM SOScat with 3 lM KRas–GTP (central panel) or KRasG13D–GTP (lower panel). Peaks corresponding to KRas, SOScat, binary, and ternary complexes are colored in purple, chartreuse, cyan, and orange, respectively. Adapted, with permission, from 70 (c) and 76 (f).
KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
KEYNOTE (GREEN) www.drugdiscoverytoday.com 7 ments could increase the lifetime of the enzyme–substrate com- plex, thereby slowing the rate of substrate processing by up to
40% (Figure 3c).70 This work highlights the versatility of nMS, which was used to acquire thermodynamic and kinetic informa- tion on the effects of small molecules on the monomer–dimer equilibrium and enzymatic processing. More recently, Zhu et al. used a nMS-based bioaffinity selection method to screen for small molecules obtained from crude herbal extracts against
Mpro.75 They identified the three flavonoids (baicalein, scutel- larein, and ganhuangenin) as Mpro ligands with KD values in the low micromolar range. These compounds did not alter the dimeric assembly of Mpro, but could still inhibit its enzymatic activity, with IC50 values in line with the measured KDs.
Moghadamchargari and coworkers recently used nMS to inves- tigate interactions between the oncoprotein KRas and the catalytic domain of the guanine nucleotide exchange factor Son of Seven- less (SOScat), which is responsible for reloading KRas with GTP.76
The KRas isoform investigated was human KRas4B (residues 1–
169), referred to as KRas for the sake of simplicity. The authors investigated the binding of SOScat to wild-type KRas, as well as three oncogenic mutants (G12C, G13D, and Q61H). In this con- text, they examined the effect of the covalent KRasG12C inhibitor
ARS-162077 and three known small-molecule KRas–SOS disruptors (Kobe0065,78 BAY-293,79 and BI-340680) on the interaction between SOScat and different KRas variants (Figure 3d–g). The authors demonstrated that ARS-1620 could react with GDP- loaded KRasG12C, but showed limited reactivity in the presence of GTP (Figure 3d). In line with this, ARS-1620 could disrupt the
KRasG12C–GDP/SOScat complex, but was ineffective against the
KRasG12C–GTP/SOScatassociation(Figure3d).Kobe0065(IC50=20- lM), reported as a KRas-GTP ligand,78 was unable to disrupt the
KRas–GTP or KRasG12D–GTP interactions with SOScat (Figure 3e) and could not abolish SOScat binding to KRasG12C. Different from
Kobe0065, BAY-293 (IC50 = 21 nM) and BI-3406 (IC50 = 5 nM) have been reported to bind directly to SOScat.79,80 BAY-293 could not disrupt the SOScat interaction with either KRas–GTP (Figure 3f, middle panel), KRasG12C–GTP, or KRasG12D–GTP (Figure 3f, lower panel), whereas it was effective for unloaded KRas and KRasG12C.
Similarly, BI-3406 (tested at 2.5 lM) acted as a KRas/SOScat disrup- tor (Figure 3g, middle panel) but could not disrupt the KRasG12D/
SOScat interaction (Figure 3g, lower panel) and was effective only at 20 lM, 4000 its reported IC50. This study not only advanced our understanding of KRas/SOS interactions, but also provided key insights into the modes of action of known KRas/SOS disrup- tors, none of which were effective in the presence of GTP-loaded
KRas. Overall, this report exemplifies the power of nMS in clarify- ing drug modes of action, thereby providing a foundation for the development of optimized modulators.
Bolla et al. elucidated the influence of chlorhexidine, a com- mon antiseptic, on the oligomerization of the chlorhexidine efflux pump AceI from Acinetobacter baumannii and on its tran- scriptional regulator AceR.71 nMS experiments demonstrated that AceI exists in a pH-dependent monomer–dimer equilibrium, with the functional form being the dimeric AceI. Interestingly, this equilibrium is altered by chlorhexidine, which increases AceI dimer formation, thereby facilitating the functional form of the efflux pump. Further nMS experiments revealed that the tran- scriptional regulator AceR exists mostly as a dimer, although a small fraction of tetramers is also present in solution. nMS data also showed that A. baumannii RNA polymerase constitutively binds the promoter region upstream of AceI. When AceR was introduced into the system, its dimeric form competitively inter- acted with the DNA fragment, inhibiting the RNA polymerase promoter binding, thus impairing AceI transcription. Interest- ingly, chlorhexidine addition increased the amount of tetrameric
AceR, which is unable to interact with DNA, thereby allowing
RNA polymerase to bind the promoter and start transcription.71
Overall, by assessing the influence of small molecules on pro- tein–protein and protein–nucleic acid interactions, nMS revealed crucial insights into the mechanism of antibiotic drug resistance.
This approach could be expanded to the identification of small molecules acting either as AceI dimerization disruptors or as antagonists of the AceR–chlorhexidine interaction, therefore contributing to the discovery of novel antibacterial agents for the treatment of chlorhexidine-resistant bacteria infections.
Bellamy-Carter and coworkers recently used nMS to investi- gate the influence of different small molecules acting as stabiliz- ers of the interactions between the eukaryotic regulatory protein
14-3-3r and three of its binding partners: the tumor suppressor p53, the leucine-rich repeat kinase 2 (LRRK2), and the estrogen receptor a (ERa).81 nMS experiments demonstrated that 14-3- 3r is a dimer with the highest affinity for ERa followed by LRRK2 and p53 (Figure 4a–c, upper panels). The fungal diterpenoid gly- coside fusicoccin A (Figure 4d), known to stabilize the interac- tions between 14-3-3r and its partners to a different extent,82–
84 was shown to further increase the abundance of the 14-3- 3r/ERa complex. while having little or no effect in the case of p53 and LRRK2, respectively (Figure 4a–c). Interestingly, the nMS spectra revealed that the fusicoccin A effect increased in the presence of a higher ERa concentration, thereby suggesting a cooperative binding mode, whereby fusicoccin A preferentially interacts with the 14-3-3r/ERa complex rather than with the sin- gle subunits. Finally, the authors screened a drug cocktail of potential 14-3-3r PPI stabilizers. Incubation of 14-3-3r/ERa with a cocktail of seven different molecules indicated that the only compound able to stabilize the interaction is fusicoccin A and, to a lesser extent, its biosynthetic deacetoxy precursor fusicoccin
J. This work highlights the utility of nMS as a screening approach for PPI stabilizers because it can not only provide a measurement of their potency, but also enable elucidation of subtle aspects, such as cooperative binding.
Recent work by Yen et al., resulting from a collaboration between Oxford University and OMass Therapeutics scientists, exemplified the ability of nMS to capture the effects of ligand binding to the turkey b1-adrenergic receptor (tb1AR), which influ- ence its coupling to different G proteins.85 Initial measurements performed at a ligand:tb1AR 50:1 molar ratio enabled the deriva- tion of significant SARs by demonstrating that isoprenaline could induce 100% complex formation between tb1AR and an engi- neered mini-G stimulatory (mini-Gs) protein, whereas com- pounds lacking the essential moieties for receptor binding were less effective (Figure 4e, upper panel). Among these derivatives, orciprenaline, in which the para-hydroxy group of isoprenaline is moved to the meta position, displayed a 60% reduction in mini-Gs coupling. Similarly, 1-phenyl-2-(2-propylamino)ethan- 1-ol, lacking both the catechol hydroxy groups, abolished 90%
KEYNOTE (GREEN) KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
8 www.drugdiscoverytoday.com (g) (e) (b) (c) (f) (a)
Fusicoccin A (d) Drug Discovery Today FIGURE 4 Assessment of protein-protein interaction (PPI) stabilizers. (a) Structure of fusicoccin A. (b–d) Deconvoluted mass spectra showing the interaction between
14-3-3r (monomer concentration 5 lM) and each interacting partner [p53 (b), leucine-rich repeat kinase 2 (LRRK2) (c), and estrogen receptor alpha (Era) (d),
25 lM] in the absence or in the presence of 5 lM fusicoccin A. Bar charts show the stoichiometry of the PPIs detected by native mass spectrometry (nMS). (e)
Mass spectra showing the interaction between tb1AR (5 lM) and mini-Gs (6 lM) in the presence of isoprenaline, orciprenaline, 1-phenyl-2-(2- propylamino)ethan-1-ol, and isopropyldopamine (250 lM). (f–g) Mass spectra showing the interaction between tb1AR (5 lM) and mini-Gs (f) or mini-Gi/s (g) (6 lM) in the presence of the full agonists isoprenaline, norepinephrine, carmoterol, and the partial agonist dobutamine (25 lM). The peaks assigned to tb1AR-mini-Gs complex, tb1AR, mini-Gs, and tb1AR-mini-Gi/s complex are depicted in orange, blue, gray, and magenta respectively. The structures of each compound are indicated alongside each spectrum. Adapted, with permission, from 81 (b–d) and 85 (e–g).
KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
KEYNOTE (GREEN) www.drugdiscoverytoday.com 9 mini-Gs coupling (Figure 4e, central panels). Moreover, isopropy- ldopamine, lacking the b-hydroxy group, completely abolished mini-Gs coupling (Figure 4e, lower panel). Further nMS experi- ments performed at a ligand:tb1AR 5:1 molar ratio recapitulated the effects of full agonists (isoprenaline, norepinephrine, and carmoterol), partial agonists (dobutamine and salbutamol), and antagonists (cyanopindolol, carazolol, and carvedilol) by show- ing that full agonists are capable of inducing 100% tb1AR-mini- Gs complex formation, whereas partial agonists were less effec- tive, and antagonists could not enable mini-Gs coupling (Fig- ure 4f). In addition, experiments in the presence of mini-Gi/s protein [in which the helix 5 motif of mini-Gs was replaced with the sequence of the G inhibitory (Gi) protein] enabled the assess- ment of agonist-biased signaling. nMS data showed that isopre- naline could induce mini-Gi/s coupling, whereas carmoterol was less effective and norepinephrine could not stimulate mini-Gi/s coupling at all (Figure 4g). Competition experiments in the presence of both mini-Gs and mini-Gi/s further demon- strated that isoprenaline preferentially induces mini-Gs coupling.
Interestingly, the authors also showed that Zn2+ has a key role in stabilizing the tb1AR-mini-Gs complex formation. In line with this, treatment with the divalent cation chelator ethylenedi- aminetetraacetic acid (EDTA) abolished tb1AR–mini-Gs interac- tions, which were reinstated by treatment with
ZnCl2.
Drug Discovery Today FIGURE 5 Evaluation of proteolysis-targeting chimeras (PROTACs). (a) Schematic of the protein degradation mechanism promoted by PROTACs. By interacting with an
E3 ligase and the protein of interest (POI), PROTACs physically bring the two proteins into proximity, thereby promoting POI ubiquitination and consequent proteasomal degradation. (b) Structures of the VHL-recruiting PROTACs AT1 and MZ1. (c) Mass spectra showing the interaction between the VCB complex (5 lM) and AT1 (10 lM) (upper panel); Brd4BD2 (5 lM) and AT1 (10 lM) (central panel); and VHL/elongin-B/elongin-C (VCB) (5 lM), AT1 (10 lM), and Brd4BD2 (5 lM) (lower panel). Bar charts provide the quantification of the relative abundance of each species. (d) Mass spectra showing the interaction between VCB (5 lM) and AT1 (10 lM) (upper panel); Brd4BD1 (5 lM) and AT1 (10 lM) (central panel); and VCB (5 lM), AT1 (10 lM), and Brd4BD1 (5 lM) (lower panel). Bar charts provide the quantification of the relative abundance of each species. Adapted from 87 (c,d).
KEYNOTE (GREEN) KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
10 www.drugdiscoverytoday.com Molecular dynamics (MD) suggested that Zn2+ ions promote the structural transition during the formation of complexes between
G-proteins and receptors, and subsequent site-directed mutagen- esis studies showed that tb1AR mutants with decreased Zn2+ binding were also characterized by reduced mini-Gs coupling.
In this study, no protein–drug interactions were detected because the voltages used to enable the liberation of tb1AR from deter- gent micelles led to dissociation of the small molecules.
Nonetheless, this study demonstrates the ability of nMS to guide drug design by monitoring the effect of potential drugs on speci- fic pharmacological pathways (e.g., GPCR–G-protein coupling) rather than directly detecting protein–ligand interactions.
Assessment of PROTAC efficacy PROTACs are heterobifunctional molecules constituted by an E3 ligase-recruiting moiety connected via a linker to a warhead that binds the protein of interest (POI). PROTACs bring an E3 ligase and a
POI into proximity, thus promoting POI poly- ubiquitination and proteasomal-mediated degradation, and are then released and recycled to induce the degradation of a new
POI (Figure 5a), displaying a catalytic mode of action.86 Com- pared with classical inhibitors, they usually show higher potency and selectivity, prolonged pharmacodynamic effects, and the ability to evade most resistance mechanisms based on target mutations. The most targeted E3 ligases are Cereblon, usually tar- geted by thalidomide and derivatives, and the Von Hippel–Lin- dau protein (VHL), usually targeted by the small-molecule
VH032 and related compounds.86 Over the years, specialized protocols in biochemical and bio- physical methods have been devised to characterize the forma- tion of the POI–PROTAC–E3 ligase ternary complex to assess in vitro the potential of PROTACs. These include ITC and SPR, which provide key information on the thermodynamics and kinetics of PROTAC binding. More recently, nMS has been indi- cated as a powerful tool to capture POI–PROTAC–E3 ligase tern- ary complex formation. Through nMS, Beveridge et al. captured the interaction between the E3 ligase complex VHL/elongin-B/ elongin-C (VCB), a bromodomain and extra-terminal domain (BET) protein (Brd3BD2, Brd4BD1, or Brd4BD2) and the PROTACs
AT1 and MZ1 [the POI ligand of which is the pan-BET inhibitor (+)-JQ1] (Figure 5b).87 Native mass spectra indicated that both
AT1 and MZ1 preferentially form a ternary complex with Brd4BD2 (Figure 5c,d show the AT1-induced VCB-Brd4BD2 and VCB- Brd4BD1 complex formation, respectively), with AT1 being more selective than MZ1, in line with published data.88 Optimal bind- ing was reached when the E3 ligase:PROTAC:POI molar ratio was
1:2:1 (5 lM:10 lM:5 lM), whereas a 20 lM PROTAC concentra- tion led to the so-called ‘Hook effect’, comprising a decrease in ternary complex formation in the presence of high PROTAC con- centrations, which promote binary interactions. Moreover, by comparing the ternary complex formation with a highly cooper- ative POI (Brd4BD2, a = 4.7) and a less cooperative one (Brd4BD1, a = 0.2), nMS experiments captured cooperative PROTAC bind- ing (i.e., binding of the PROTAC to the first protein complex enhances the affinity for the second one, thereby facilitating ternary complex formation over the binary ones; compare Fig- ure 5c and d). Specifically, when AT1 (10 lM) was mixed with
5 lM VCB, the ternary complex fraction was 0.2, whereas it was 0.5 for both Brd4BD2 and Brd4BD1 at the same concentra- tions. Notably, when the three components were incubated together at a ligase:PROTAC:POI 1:2:1 molar ratio, the ternary complex was observed to a much higher extent with Brd4BD2 (0.82) than with Brd4BD1 (0.65), thus confirming what had previ- ously been observed in other experiments (Figure 5c,d).88 Finally, competition experiments in the presence of three (Brd3BD2,
Brd4BD1, and Brd4BD2) or five (Brd3BD2, Brd4BD1, Brd4BD2,
Brd2BD2, and BrdT) BET family members simultaneously, con- firmed both the preferential binding of AT1 and MZ1 toward
Brd4BD2 and the higher selectivity of AT1.87 More recently, Sternicki and colleagues used nMS to assess the complex formation between VCB and either Brd4BD1 or Brd4BD2 induced by GNE-987,89 a recently developed VHL-recruiting
Brd4-targeting PROTAC.90,91 The authors showed that GNE-987 targets Brd4BD1 preferentially, with a complex ratio of 0.70 when
GNE-987 (7.81 lM) was mixed with equimolar amounts of VCB and Brd4BD1 (9 lM). At the same concentrations, the complex ratio for Brd4BD2 was 0.34, with maxima reached when GNE- 987 was increased up to 15.625 and 31.25 lM (0.44 and 0.45, respectively). Conversely, when GNE-987 was tested at these concentrations in the presence of Brd4BD1, the ‘hook effect’ was apparent, and the ternary complex ratio decreased to
0.55.89 These data are in accordance with SPR measurements, which indicate a much higher ternary complex half-life (t1/2) for Brd4BD1, although the magnitude of the difference between the amount of ternary complex formed for Brd4BD1 compared with Brd4BD2 as measured by nMS is not as large as the window of difference in SPR t1/2 measurements, in which Brd4BD1 t1/2 is
100-fold longer than that of Brd4BD2.91 Nevertheless, nMS pro- vided a steady-state equilibrium measurement, whereas SPR mea- sured a real-time kinetic event; thus, each measurement is unique and influenced by different factors.
A subsequent study by Song and colleagues described nMS experiments on the VCB–MZ1–Brd4BD2 complex on a Fourier- transform ion cyclotron resonance (FT-ICR) mass spectrometer and demonstrated that the VCB–MZ1–Brd4BD2 ternary complex dissociates following application of increasing CID voltages, which cause ejection of MZ1, whereas the VCB–Brd4BD2 com- plex, not observed in solution in the absence of MZ1, is retained.
Further increases in CID voltages induced the release of the peripheral subunit Brd4BD2.92 IM-MS experiments performed on a quadrupole/ion mobility separation/time-of-flight (Q-IMS- ToF) instrument suggested that the 13+ and 12+ charge states observed in the initial experiments assume a more compact con- formation, compared with the 15+ charge state, which exists as an extended conformer, whereas the 14+ charge state is a mix- ture of the two. In line with this, the application of increasing collision voltages caused the preferential dissociation of MZ1 from the more compact conformers and the preferential ejection of BRD4BD2 from the more extended ones. These data demon- strate that higher charge states are characterized by greater
Coulombic repulsions and PPIs, whereas the low charge states assume a native-like conformation. To this end, the evidence that VCB and Brd4BD2 maintain their interactions even after
MZ1 ejection supports the presence of specific intermolecular interactions between VCB and Brd4BD2, as described in the corre- sponding crystal structure.88 The specific non-covalent interac- KEYNOTE (GREEN)
Drug Discovery Today d Volume 28, Number 5 d May 2023
KEYNOTE (GREEN) www.drugdiscoverytoday.com 11 tions between the two proteins during the CID experiments might vary from those found in the gas phase. Therefore, although these studies provide a valuable starting point, they need to be validated with orthogonal experiments. Nonetheless, these three studies demonstrate that PROTAC efficacy can be suc- cessfully probed via nMS, and the integration of different approaches and instruments could provide insights into both binding affinities and structural features.
Concluding remarks The progress that has been achieved in the technology of mass spectrometers has paved the way for nMS to evolve into a remarkably flexible tool for analyzing protein–protein and pro- tein–small molecule interactions. Its great sensitivity, ease of use, speed, broad dynamic range, and minimal sample consump- tion make it an essential component of the biophysical toolbox widely used for early drug screening campaigns. As the numbers of research groups and studies using nMS for drug discovery applications expand and, consequently, more researchers are trained in this area, the investigation of protein–ligand interac- tions by nMS will become increasingly regular in drug discovery. nMS will become progressively integrated with other methods, such as ITC, SPR, and structural techniques, such as X-ray crystal- lography and cryo-EM, offering unparalleled insights into pro- tein–ligand binding and PPI modulation, which will have a substantial influence on future drug development processes.
The increased use of nMS in drug discovery should take place with an understanding of the underlying origins of possible experimental false negatives (gas-phase breakdown of hydropho- bic interactions) and false positives (nonspecific binding). To this end, future developments in sample preparation, ionization techniques, and equipment should help to successfully over- come these limitations. Nonetheless, nMS analysis is unique in capturing the effects that ligand binding has in solution, such as impacts on protein complex formation,85 oligomerization,69 or enzyme–substrate/co-substrate binding.70 Moreover, by either monitoring direct ligand binding or the effect of protein–ligand interactions, nMS-based competitive-binding assays can also easily define the specificity of a potential drug for a given binding site. Furthermore, binding sites can be distinguished using the right experimental parameters, showing complicated allosteric processes.70,76,85,93 In addition, the use of nonvolatile buffers or salts, buffer additives such as charge-reducing agents, and the development of innovative detergents and membrane mimetics in the case of membrane proteins, should help to preserve labile native-like protein–small molecule complexes for nMS detection.
In line with this, recent studies demonstrated the feasibility of analyzing both soluble and membrane proteins directly from cell lysates,67,94 or membrane vesicles,95,96 respectively, thus prevent- ing potential artifacts that might arise from sample preparation.
Given that processing the complex spectra generated by these techniques remains time-consuming and difficult to interpret, advances in data analysis are crucial in this context for enabling the full development of nMS analysis of cell lysates and mem- brane vesicles. To this end, initial steps have been taken toward the development of software for analyzing intricate mass spec- tra.97,98 The integration of currently available software platforms with machine learning and artificial intelligence approaches will accelerate nMS spectral interpretation and allow the pharmaceu- tical sector to adopt nMS as a standard technique.
Given its great potential in capturing PPIs, we anticipate that nMS will have an ever-increasing role in assessing the potency and mode of action of small molecules functioning as PPI mod- ulators, including PROTACs. In some circumstances, measuring the interaction between two (or more) proteins remains difficult and requires the presence of specific tags or immobilization. By contrast, nMS permits the direct observation of these interac- tions, quantification of complex formation, analysis of coopera- tive binding, and accurate assessment of the stoichiometry of multiprotein complexes or oligomers. Therefore, a single experi- ment yields a plethora of information on the influence of a small molecule on the stability of protein complexes. As seen in the case studies included herein, several groups are already using nMS for the evaluation of PPI modulators, and we anticipate that the pharmaceutical industry will adopt nMS for these applica- tions. To this end, chip-based automated nESI platforms, such as the previously mentioned NanoMate,53 are pivotal for increas- ing the throughput of nMS experiments and enabling its applica- tion to small-molecule libraries.
Authors’ contributions All authors contributed to this article.
Declaration of interests The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Data availability No data was used for the research described in the article.
Acknowledgments This work was supported by Italian Ministry of University
FISR2019_00374 MeDyCa (A.M.), ‘Sapienza’ Ateneo Project
2021 n. RM12117A61C811CE (D.R.), Regione Lazio Progetti di
Gruppi di Ricerca 2020 – A0375-2020-36597 (D.R.). F.F. is sup- ported by the EU’s Horizon Europe program under the Marie
Skłodowska-Curie grant agreement EpiPolyPharma 101062363.
References 1. Eisenberg D, Marcotte EM, Xenarios I, Yeates TO. Protein function in the post- genomic era. Nature. 2000;405:823–826.
2. Yamauchi O.
Noncovalent interactions in biocomplexes.
Phys Sci Rev.
2016;1:20160001.
3. Zhou P, Huang J, Tian F. Specific noncovalent interactions at protein-ligand interface: implications for rational drug design. Curr Med Chem. 2012;19:226–238.
4. Su H, Xu Y. Application of ITC-based characterization of thermodynamic and kinetic association of ligands with proteins in drug design. Mini Review. Front
Pharmacol. 2018;9:1133.
5. O’Connell N. Protein ligand interactions using surface plasmon resonance.
Methods Mol Biol. 2021;2365:3–20.
KEYNOTE (GREEN) KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
12 www.drugdiscoverytoday.com 6. Becker W, Bhattiprolu KC, Gubensäk N, Zangger K. Investigating protein–ligand interactions by solution nuclear magnetic resonance spectroscopy.
ChemPhysChem. 2018;19:895–906.
7. Turnbull AP, Emsley P. Studying protein-ligand interactions using X-ray crystallography. Methods Mol Biol. 2013;1008:457–477.
8. Benesch JL, Ruotolo BT, Simmons DA, Robinson CV. Protein complexes in the gas phase: technology for structural genomics and proteomics. Chem Rev.
2007;107:3544–3567.
9. Liko I, Allison TM, Hopper JT, Robinson CV. Mass spectrometry guided structural biology. Curr Opin Struct Biol. 2016;40:136–144.
10. Laganowsky A, Reading E, Hopper JT, Robinson CV. Mass spectrometry of intact membrane protein complexes. Nat Protoc. 2013;8:639.
11. Mehmood S, Allison TM, Robinson CV.
Mass spectrometry of protein complexes: from origins to applications. Annu Rev Phys Chem. 2015;66:453–474.
12. Boeri Erba E, Signor L, Petosa C. Exploring the structure and dynamics of macromolecular complexes by native mass spectrometry. J Proteomics. 2020;222
103799.
13. Wilm M, Mann M. Analytical properties of the nanoelectrospray ion source.
Anal Chem. 1996;68:1–8.
14. Katta V, Chait BT. Observation of the heme-globin complex in native myoglobin by electrospray–ionization mass spectrometry. J Am Chem Soc. 1991;113:8534.
15. Ganem B, Li YT, Henion JD. Detection of noncovalent receptor–ligand complexes by mass spectrometry. J Am Chem Soc. 1991;113:6294.
16. Kitova EN, El-Hawiet A, Schnier PD, Klassen JS. Reliable determinations of protein-ligand interactions by direct ESI-MS measurements. Are we there yet? J
Am Soc Mass Spectrom. 2012;23:431–441.
17. Wang W, Kitova EN, Klassen JS. Influence of solution and gas phase processes on protein–carbohydrate binding affinities determined by nanoelectrospray Fourier transform ion cyclotron resonance mass spectrometry.
Anal Chem.
2003;75:4945.
18. El-Hawiet A, Shoemaker GK, Daneshfar R, Kitova EN, Klassen JS. Applications of a catch and release electrospray ionization mass spectrometry assay for carbohydrate library screening. Anal Chem. 2012;84:50–58.
19. Sun J, Kitova EN, Klassen JS. Method for stabilizing proteinligand complexes in nanoelectrospray ionization mass spectrometry. Anal Chem. 2007;79:416–425.
20. Allison TM, Bechara C. Structural mass spectrometry comes of age: new insight into protein structure, function and interactions.
Biochem Soc Trans.
2019;47:317–327.
21. Fiorentino F, Bolla JR. Mass spectrometry analysis of dynamics and interactions of the LPS translocon LptDE. Methods Mol Biol. 2022;2548:109–128.
22. McDowell MA et al.. Structural basis of tail-anchored membrane protein biogenesis by the GET insertase complex. Mol Cell. 2020;80:72–86.
23. Olinares PDB et al.. Native mass spectrometry-based screening for optimal sample preparation in single-particle cryo-EM. Structure. 2021;29:186.
24. Agasid MT, Sørensen L, Urner LH, Yan J, Robinson CV. The effects of sodium ions on ligand binding and conformational states of G protein-coupled receptors—insights from mass spectrometry.
J Am Chem Soc.
2021;143:4085–4089.
25. Nguyen GTH, Tran TN, Podgorski MN, Bell SG, Supuran CT, Donald WA.
Nanoscale ion emitters in native mass spectrometry for measuring ligand- protein binding affinities. ACS Cent Sci. 2019;5:308–318.
26. Townsend JA, Keener JE, Miller ZM, Prell JS, Marty MT. Imidazole derivatives improve charge reduction and stabilization for native mass spectrometry. Anal
Chem. 2019;91:14765–14772.
27. Shimon L, Sharon M, Horovitz A. A method for removing effects of nonspecific binding on the distribution of binding stoichiometries: application to mass spectroscopy data. Biophys J. 2010;99:1645–1649.
28. Gault J et al.. Combining native and ‘omics’ mass spectrometry to identify endogenous ligands bound to membrane proteins.
Nat Methods.
2020;17:505–508.
29. Gault J et al.. High-resolution mass spectrometry of small molecules bound to membrane proteins. Nat Methods. 2016;13:333–336.
30. Sobott F, Hernández H, McCammon MG, Tito MA, Robinson CV. A tandem mass spectrometer for improved transmission and analysis of large macromolecular assemblies. Anal Chem. 2002;74:1402.
31. Laganowsky A et al.. Membrane proteins bind lipids selectively to modulate their structure and function. Nature. 2014;510:172–175.
32. Xie Y, Zhang J, Yin S, Loo JA. Top-down ESI-ECD-FT-ICR mass spectrometry localizes noncovalent protein-ligand binding sites.
J Am Chem Soc.
2006;128:14432.
33. Li H, Nguyen HH, Ogorzalek Loo RR, Campuzano IDG, Loo JA. An integrated native mass spectrometry and top-down proteomics method that connects sequence to structure and function of macromolecular complexes. Nat Chem.
2018;10:139–148.
34. Barrera NP, Di Bartolo N, Booth PJ, Robinson CV. Micelles protect membrane complexes from solution to vacuum. Science. 2008;321:243–246.
35. Urner LH et al.. Anionic dendritic polyglycerol for protein purification and delipidation. ACS Appl Polym Mater. 2021;3:5903–5911.
36. Bolla JR, Agasid MT, Mehmood S, Robinson CV. Membrane protein-lipid interactions probed using mass spectrometry.
Annu Rev Biochem.
2019;88:85–111.
37. Bolla JR, Fiorentino F, Robinson CV. Mass spectrometry informs the structure and dynamics of membrane proteins involved in lipid and drug transport. Curr
Opin Struct Biol. 2021;70:53–60.
38. Fiorentino F et al.. Dynamics of an LPS translocon induced by substrate and an antimicrobial peptide. Nat Chem Biol. 2021;17:187–195.
39. Fiorentino F, Mai A, Rotili D. Lysine acetyltransferase inhibitors: structure– activity relationships and potential therapeutic implications. Future Med Chem.
2018;10:1067–1091.
40. Yen HY et al.. Ligand binding to a G protein-coupled receptor captured in a mass spectrometer. Sci Adv. 2017;3:e1701016.
41. Yen HY et al.. PtdIns(4,5)P2 stabilizes active states of GPCRs and enhances selectivity of G-protein coupling. Nature. 2018;559:423–427.
42. Urner LH et al.. Modular detergents tailor the purification and structural analysis of membrane proteins including G-protein coupled receptors. Nat Commun.
2020;11:564.
43. Laganowsky A, Clemmer DE, Russell DH. Variable-temperature native mass spectrometry for studies of protein folding, stabilities, assembly, and molecular interactions. Annu Rev Biophys. 2022;51:63–77.
44. Marcoux J et al.. Mass spectrometry reveals synergistic effects of nucleotides, lipids, and drugs binding to a multidrug resistance efflux pump. Proc Natl Acad
Sci U S A. 2013;110:9704–9709.
45. Mehmood S et al.. Mass spectrometry captures off-target drug binding and provides mechanistic insights into the human metalloprotease ZMPSTE24. Nat
Chem. 2016;8:1152–1158.
46. Liko I et al.. Lipid binding attenuates channel closure of the outer membrane protein OmpF. Proc Natl Acad Sci U S A. 2018;115:6691–6696.
47. Bolla JR, Sauer JB, Wu D, Mehmood S, Allison TM, Robinson CV. Direct observation of the influence of cardiolipin and antibiotics on lipid II binding to
MurJ. Nat Chem. 2018;10:363–371.
48. Fiorentino F, Bolla JR, Mehmood S, Robinson CV. The different effects of substrates and nucleotides on the complex formation of ABC transporters.
Structure. 2019;27:651–659.
49. Bennett JL, Nguyen GTH, Donald WA. Protein-small molecule interactions in native mass spectrometry. Chem Rev. 2022;122:7327–7385.
50. Gavriilidou AFM, Sokratous K, Yen HY, De Colibus L. High-throughput native mass spectrometry screening in drug discovery. Front Mol Biosci. 2022;9 837901.
51. Vu H et al.. Fragment-based screening of a natural product library against 62 potential malaria drug targets employing native mass spectrometry. ACS Infect
Dis. 2018;4:431–444.
52. Nguyen GTH et al.. Multiplexed screening of thousands of natural products for protein-ligand binding in native mass spectrometry.
J Am Chem Soc.
2021;143:21379–21387.
53. Zhang S, Van Pelt CK, Wilson DB. Quantitative determination of noncovalent binding interactions using automated nanoelectrospray mass spectrometry. Anal
Chem. 2003;75:3010–3018.
54. Supuran CT. Carbonic anhydrases: novel therapeutic applications for inhibitors and activators. Nat Rev Drug Discov. 2008;7:168–181.
55. Alterio V, Di Fiore A, D’Ambrosio K, Supuran CT, De Simone G. Multiple binding modes of inhibitors to carbonic anhydrases: how to design specific drugs targeting 15 different isoforms? Chem Rev. 2012;112:4421–4468.
56. Park H, Jung J, Rodrigues E, Kitova EN, Macauley MS, Klassen JS. Mass spectrometry-based shotgun glycomics for discovery of natural ligands of glycan-binding proteins. Anal Chem. 2020;92:14012–14020.
57. White KL et al.. Structural connection between activation microswitch and allosteric sodium site in GPCR signaling. Structure. 2018;26:259–269.
58. Ren C et al.. Quantitative determination of protein-ligand affinity by size exclusion chromatography directly coupled to high-resolution native mass spectrometry. Anal Chem. 2019;91:903–911.
59. Cheng C et al.. Identifying new ligands for JNK3 by fluorescence thermal shift assays and native mass spectrometry. ACS Omega. 2022;7:13925–13931.
60. D’Amico CI, Polasky DA, Steyer DJ, Ruotolo BT, Kennedy RT. Ion mobility-mass spectrometry coupled to droplet microfluidics for rapid protein structure analysis and drug discovery. Anal Chem. 2022;94:13084–13091.
KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
KEYNOTE (GREEN) www.drugdiscoverytoday.com 13 61. Benesch JL, Ruotolo BT. Mass spectrometry: come of age for structural and dynamical biology. Curr Opin Struct Biol. 2011;21:641–649.
62. Fiorentino F, Castiello C, Mai A, Rotili D. Therapeutic potential and activity modulation of the protein lysine deacylase sirtuin
5.
J Med Chem.
2022;65:9580–9606.
63. Fiorentino F, Mautone N, Menna M, D’Acunzo F, Mai A, Rotili D. Sirtuin modulators: past, present, and future perspectives.
Future Med Chem.
2022;14:915–939.
64. Taurone S et al.. Biochemical functions and clinical characterizations of the sirtuins in diabetes-induced retinal pathologies. Int J Mol Sci. 2022;23:4048.
65. Dixit SM, Polasky DA, Ruotolo BT. Collision induced unfolding of isolated proteins in the gas phase: past, present, and future. Curr Opin Chem Biol.
2018;42:93–100.
66. Kaur H et al.. The antibiotic darobactin mimics a beta-strand to inhibit outer membrane insertase. Nature. 2021;593:125–129.
67. Rogawski R et al.. Intracellular protein-drug interactions probed by direct mass spectrometry of cell lysates. Angew Chem Int Ed Engl. 2021;60:19637–19642.
68. Olinares PDB, Dunn AD, Padovan JC, Fernandez-Martinez J, Rout MP, Chait BT.
A robust workflow for native mass spectrometric analysis of affinity-isolated endogenous protein assemblies. Anal Chem. 2016;88:2799–2807.
69. Fiorentino F, Rotili D, Mai A, Bolla JR, Robinson CV. Mass spectrometry enables the discovery of inhibitors of an LPS transport assembly via disruption of protein–protein interactions. Chem Commun (Camb). 2021;57:10747–10750.
70. El-Baba TJ et al.. Allosteric inhibition of the SARS-CoV-2 main protease: insights from mass spectrometry based assays.
Angew Chem Int Ed Engl.
2020;59:23544–23548.
71. Bolla JR, Howes AC, Fiorentino F, Robinson CV. Assembly and regulation of the chlorhexidine-specific efflux pump
AceI.
Proc Natl Acad Sci U S A.
2020;117:17011–17018.
72. Lundstedt E, Kahne D, Ruiz N. Assembly and maintenance of lipids at the bacterial outer membrane. Chem Rev. 2020;121:5098–5123.
73. Bollati M et al.. Crystal structure of LptH, the periplasmic component of the lipopolysaccharide transport machinery from Pseudomonas aeruginosa. FEBS J.
2015;282:1980–1997.
74. Douangamath A et al.. Crystallographic and electrophilic fragment screening of the SARS-CoV-2 main protease. Nat Commun. 2020;11:5047.
75. Zhu D et al.. Efficient discovery of potential inhibitors for SARS-CoV-2 3C-like protease from herbal extracts using a native MS-based affinity-selection method.
J Pharm Biomed Anal. 2022;209 114538.
76. Moghadamchargari Z et al.. Molecular assemblies of the catalytic domain of SOS with KRas and oncogenic mutants. Proc Natl Acad Sci U S A. 2021;118. e2022403118.
77. Janes MR et al.. Targeting KRAS mutant cancers with a covalent G12C-specific inhibitor. Cell. 2018;172:578–589.
78. Shima F et al.. In silico discovery of small-molecule Ras inhibitors that display antitumor activity by blocking the Ras-effector interaction. Proc Natl Acad Sci U S
A. 2013;110:8182–8187.
79. Hillig RC et al.. Discovery of potent SOS1 inhibitors that block RAS activation via disruption of the
RAS-SOS1 interaction.
Proc Natl Acad Sci U S A.
2019;116:2551–2560.
80. Hofmann MH et al.. BI-3406, a potent and selective SOS1-KRAS interaction inhibitor, is effective in KRAS-driven cancers through combined MEK inhibition.
Cancer Discov. 2021;11:142–157.
81. Bellamy-Carter J et al.. Discovering protein-protein interaction stabilisers by native mass spectrometry. Chem Sci. 2021;12:10724–10731.
82. De Vries-van Leeuwen IJ et al.. Interaction of 14-3-3 proteins with the Estrogen
Receptor Alpha F domain provides a drug target interface. Proc Natl Acad Sci U S
A. 2013;110:8894–8899.
83. Stevers LM, De Vries RMJM, Doveston RG, Milroy LG, Brunsveld L, Ottmann C.
Structural interface between LRRK2 and 14-3-3 protein.
Biochem J.
2017;474:1273–1287.
84. Doveston RG et al.. Small-molecule stabilization of the p53–14-3-3 protein- protein interaction. FEBS Lett. 2017;591:2449–2457.
85. Yen HY et al.. Mass spectrometry captures biased signalling and allosteric modulation of a G-protein-coupled receptor. Nat Chem. 2022;14:1375–1382.
86. Tomaselli D, Mautone N, Mai A, Rotili D. Recent advances in epigenetic proteolysis targeting chimeras (epi-PROTACs). Eur J Med Chem. 2020;207
112750.
87. Beveridge R, Kessler D, Rumpel K, Ettmayer P, Meinhart A, Clausen T. Native mass spectrometry can effectively predict PROTAC efficacy. ACS Cent Sci.
2020;6:1223–1230.
88. Gadd MS et al.. Structural basis of PROTAC cooperative recognition for selective protein degradation. Nat Chem Biol. 2017;13:514–521.
89. Sternicki LM, Nonomiya J, Liu M, Mulvihill MM, Quinn RJ. Native mass spectrometry for the study of PROTAC GNE-987-containing ternary complexes.
ChemMedChem. 2021;16:2206–2210.
90. Fidanze SD et al..
Discovery and optimization of novel constrained pyrrolopyridone
BET family inhibitors.
Bioorg Med Chem Lett.
2018;28:1804–1810.
91. Pillow TH et al.. Antibody conjugation of a chimeric BET degrader enables in vivo activity. ChemMedChem. 2020;15:17–25.
92. Song JH et al.. Native mass spectrometry and gas-phase fragmentation provide rapid and in-depth topological characterization of a PROTAC ternary complex.
Cell Chem Biol. 2021;28:1528–1538.
93. Patrick JW et al.. Allostery revealed within lipid binding events to membrane proteins. Proc Natl Acad Sci U S A. 2018;115:2976–2981.
94. Vimer S, Ben-Nissan G, Sharon M. Mass spectrometry analysis of intact proteins from crude samples. Anal Chem. 2020;92:12741.
95. Chorev DS et al.. Protein assemblies ejected directly from native membranes yield complexes for mass spectrometry. Science. 2018;362:829.
96. Chen S et al.. Capturing a rhodopsin receptor signalling cascade across a native membrane. Nature. 2022;604:384–390.
97. Quetschlich D et al.. NaViA: a program for the visual analysis of complex mass spectra. Bioinformatics. 2021;37:4876–4878.
98. Marty MT, Baldwin AJ, Marklund EG, Hochberg GK, Benesch JL, Robinson CV.
Bayesian deconvolution of mass and ion mobility spectra: from binary interactions to polydisperse ensembles. Anal Chem. 2015;87:4370–4376.
KEYNOTE (GREEN) KEYNOTE (GREEN) Drug Discovery Today d Volume 28, Number 5 d May 2023
14 www.drugdiscoverytoday.com