Dynamics of Protein Kinases and Pseudokinases by HDX-MS

Abstract

Dynamics of the protein kinase fold are deeply intertwined with its structure. The past three decades of kinase biophysical studies revealed the key dynamic features of the kinase domain and, more recently, how these features may endow catalytically impaired kinases—or pseudokinases—with signaling properties. Hydrogen-deuterium exchange coupled with mass spectrometry (HDX-MS) is proving to be a valuable approach for detailed studies of kinase and pseudokinase dynamics. Here we briefly discuss the methods that have provided insights into protein kinase dynamics, describe how HDX-MS is being used to answer questions in the kinase/pseudokinase field, and provide a detailed protocol for collecting an HDX-MS dataset to study the impacts of small molecule binding to a pseudokinase domain. As more pseudokinase conformational disrupting small molecules are discovered, HDX-MS is likely to be a powerful approach for exploring drug-induced changes in pseudokinase dynamics.

1. INTRODUCTION

In eukaryotes, the protein kinase fold is remarkably well conserved and yet wonderfully diverse (Manning et al., 2002). From vastly different substrate specificities (Miller & Turk, 2018), to distinct regulatory mechanisms (Endicott et al., 2012), to non-catalytic signaling modes (Kung & Jura, 2016), the kinase domain is a molecular marvel of signal transduction. Since the first structural elucidation of a protein kinase domain three decades ago (Knighton et al., 1991), a key question has been: how does the fold achieve such broad biological functions? To understand how this common architecture is repurposed in cell signaling with altered specificity and functionality, the field has turned to a variety of biophysical techniques (Endicott et al., 2012; Lorenzen & Pawson, 2014; Xiao et al., 2015). The answer appears to partially lie in the dynamic nature of the domain.

The first hints of kinase conformational plasticity came from static snapshots of kinase domains (Huse & Kuriyan, 2002; Johnson et al., 1996). X-ray crystal structures of the cAMP-dependent protein kinase, or PKA, catalytic subunit revealed a canonical protein kinase domain that is bi-lobal, comprising a primarily β-stranded N-lobe and a largely α-helical C-lobe (Knighton et al., 1991). This architecture allows ATP, divalent metal ions, and peptide substrates (containing serine, threonine, or tyrosine) to dock precisely between the two lobes in an orientation suitable for the phosphotransferase reaction (Adams, 2001). To allow for efficient association and dissociation of substrates, the N-lobe is relatively dynamic (Kornev & Taylor, 2015). Comparing the crystal structure of a cyclin dependent kinase 2 (CDK2) monomer with a breakthrough structure of cyclin A-bound CDK2 first demonstrated that a single helix within the N-lobe—helix αC—can adopt a range of conformations, forming a critical salt bridge in the inward, active state (De Bondt et al., 1993; Jeffrey et al., 1995). In addition, flexibility of the glycine-rich loop above the ATP-binding site is crucial for facilitating nucleotide binding and catalysis, as was visualized in early structures of PKA and of phosphorylase kinase bound to different nucleotides (Narayana et al., 1997; Owen et al., 1995). A third important feature found in nearly all protein kinases is the ‘activation loop’ region, the conformational landscape of which is frequently controlled by post-translational modification (such as phosphorylation). This was seen vividly in structural comparisons of the inactive (unphosphorylated) and active (phosphorylated) forms of the insulin receptor tyrosine kinase domain (IRK) (Hubbard, 1997; Hubbard et al., 1994). The dynamics of these three regions, highlighted in Figure 1, work in concert with the rest of the domain to facilitate catalysis, but they are also exploited by numerous regulatory mechanisms (Huse & Kuriyan, 2002). Many kinases are controlled via the αC helix, which can be allosterically toggled inward and outward by intra- and/or intermolecular interactions (e.g., cyclin binding to CDKs) and/or the flexible autoinhibitory activation loop, which is often phosphorylated to relieve pseudosubstrate autoinhibition (e.g., IRK).

Figure 1.

Dynamic features of the eukaryotic protein kinase domain revealed by early crystal structures. A) Overlay of inactive, unliganded CDK2 (dark blue) and active, cyclin A- and ATP-bound CDK2 (gold) illustrates movement of the αC helix (see red arrow), as well as the activation- and glycine-rich loops, upon allosteric activation of CDK2 by cyclin A. B) Overlay of the PKA catalytic domain in its unliganded state (green) and ADP-bound state (red), revealing closure of the glycine-rich loop upon nucleotide binding (as depicted by the red arrow). C) Overlay of inactive (unphosphorylated) and unliganded IRK (purple) and active (phosphorylated) AMPPNP-bound IRK (cyan), illustrating rearrangement of the activation loop (A-loop) as depicted by the red arrow, as well as closure of the αC helix and glycine-rich loops upon phosphorylation and nucleotide binding. PDB IDs: 1HCL (Schulze-Gahmen et al., 1996), 1FIN (Jeffrey et al., 1995), 1CMK (Zheng et al., 1993), 1BKX (Narayana et al., 1997), 1IRK (Hubbard et al., 1994), 1IR3 (Hubbard, 1997).

Although crystal structures have provided invaluable clues into kinase mechanisms, many questions of how protein motions achieve catalysis still remained (Xiao et al., 2015). After the initial set of protein kinase structures in the 1990s, a battery of in-solution biophysical methods advanced our understanding of kinase dynamics in the decade that followed. Studies using nuclear magnetic resonance (NMR) or hydrogen-deuterium exchange coupled to mass spectrometry (HDX-MS) provided global insights into kinase dynamics (Lorenzen & Pawson, 2014; Xiao et al., 2015). The use of two-dimensional NMR to study PKA, for instance, revealed that the binding of nucleotide to this kinase domain results in large perturbations that propagate to the peptide-binding site of the enzyme (Masterson et al., 2008). Early HDX-MS studies of extracellular regulated protein kinase 2 (ERK2) shed light on the changes in conformational dynamics that occur upon phosphorylation of the activation loop, demonstrating how activation loop movements allosterically impact other key regions (Hoofnagle et al., 2001). In more recent years, advances in NMR instrumentation have allowed detailed observations of protein kinase energy landscapes and inactive/active state transitions (Hadzipasic et al., 2020; Pitsawong et al., 2018; Xie et al., 2020). Progress with HDX-MS is also providing key insights into dynamics of ligand binding and complex formation (described below). In addition, spectroscopic methods, such as infrared (IR) spectroscopy and electron paramagnetic resonance (EPR) spectroscopy, employed for a more focused analysis of particular regions within the kinase have more detailed views of the dynamics of the activation loop of kinases like Aurora A (Ruff et al., 2018). Given all of these approaches, along with advances in computational modeling, members of the kinase community now have many powerful tools at their fingertips to investigate and understand dynamics.

We focus here on HDX-MS as an approach that is well-equipped to address a variety of questions that are particularly crucial for pseudokinases – where phosphotransfer has been dispensed with, but dynamic conformational transitions provide the primary window on function (Kung & Jura, 2019). HDX-MS reveals information on intrinsic dynamics as well as local conformational changes induced upon perturbation, making it complementary to other in-solution methods, as well as to structure determination approaches like crystallography and cryo-electron microscopy (Engen, 2009; Engen & Komives, 2020; Hoofnagle et al., 2003). Although NMR is currently more capable of providing residue-specific information on conformation states at fast time scales, the information is generally limited to residues that can be labeled and detected in the NMR experiment. HDX-MS, by contrast, can measure protein movements on a wide range of time scales and is amenable to large proteins and protein complexes without the need for isotopic labeling during cell culture or access to strong, expensive magnets (Peacock & Komives, 2021). Below, we present an overview of HDX-MS, followed by some discussion of how the technique has provided key insights into the dynamics of the kinase fold and its regulation. We then review how HDX-MS has helped resolve questions about how the catalytically impaired pseudokinases may function by virtue of dynamic conformational changes related to those that normally control the activation state of a bona fide protein kinase.

2. OVERVIEW OF HDX-MS

HDX-MS has emerged in recent years as a core method in structural biology. The principle behind the method is the propensity of labile protons bound to heteroatoms within a polypeptide chain to exchange with protons present in aqueous solvent (Englander et al., 1997; Hvidt & Linderstrom-Lang, 1954, 1955). Protons within simple, unstructured peptides exchange rapidly, while amide protons engaged in hydrogen bonds (or otherwise shielded from solvent) do so at a measurably slower rate. Biophysicists can take advantage of this property to measure relative exchange rates for local regions within a protein, thereby providing insight into the global conformational dynamics of a given protein or protein complex in a region-specific manner (Engen, 2009). In practice, this is achieved by replacing the ¹H₂O-based solvent with a ²H₂O (deuterium oxide, herein referred to as D₂O)-based solvent via dilution, allowing for isotopic exchange of protium atoms with deuterium. The presence of these deuterated amide protons is then monitored over time using a sensitive method like mass spectrometry.

To prevent deuterated amide protons from undergoing back-exchange when in contact with excess H₂O-based buffer (such as during chromatographic separation), the reaction must be quenched by lowering the pH and temperature. Rosa and Richards demonstrated that acidic conditions are optimal for minimizing the rate of amide hydrogen exchange (pH_min~2.5), and thus are often used, along with low temperatures (e.g., close to 0°C), as quench conditions in standard HDX-MS workflows today (Jensen & Rand, 2016; Rosa & Richards, 1979). Although hydrogens bound to heteroatoms in peptide side chains or at N- and C-termini also undergo rapid exchange, their much higher pH_min values result in much higher back exchange rates compared to amide protons, allowing standard HDX-MS to specifically probe protein backbone exchange dynamics. That is, with the exception of proline residues, which lack an amide proton. For more details on fundamental hydrogen exchange theory, we refer the reader to a helpful review (Jensen & Rand, 2016).

Whereas unstructured peptides exchange at millisecond-second time scales, peptides engaged in higher order structure can exchange on significantly slower time scales due to hydrogen bonding and solvent inaccessibility. A general HDX reaction mechanism, derived from the Linderstrøm-Lang model of peptide exchange kinetics, can be expressed:

$$N-H_{cl}\begin{array}{c}{k}_{op}\\ \rightleftharpoons \\ k_{cl}\end{array}N-H_{op}\begin{array}{c}{D}_{2}O\\ \to \\ k_{int}\end{array}N-D_{op}\begin{array}{c}{k}_{cl}\\ \rightleftharpoons \\ k_{op}\end{array}N-D_{cl}$$

where the observed rate of exchange is dependent upon the intrinsic exchange rate (k_int), or the rate at which an amide proton N–H is converted into N–D, and the equilibrium of the proton between open (H_op) and closed (H_cl) states, given that it must be in the open state to undergo exchange (Hoofnagle et al., 2003; Jensen & Rand, 2016). As protein refolding is much faster than deuterium exchange (i.e. k_cl >> k_int) under most conditions for structured proteins, a so-called EX2 kinetic exchange regime with a single transition from the undeuterated state to the fully deuterated state is observed, where the observed exchange rate (k_ex) can be expressed:

$$k_{ex}=\frac{k_{op}}{k_{cl}}*k_{int}$$

An alternative regime, called EX1, is possible for cases with multiple conformation states present when amide proton exchange is faster than refolding (i.e. k_cl << k_int), and thus k_ex = k_op. For globular proteins like protein kinase domains in their native state, peptides largely follow the EX2 kinetic regime, making HDX-MS a useful measure of differential solvent accessibility within local regions of the protein.

A typical experimental workflow for HDX-MS is quite simple (Figure 2), although high-resolution LC-MS instrumentation (resolution at least 15,000 full width at half maximum) is generally preferred. The experiment begins with a purified protein sample; for example, we generally aim for a protein concentration ranging from 0.3 mg/ml to 0.6 mg/ml (thus injecting 1–10 μg protein) with >95% estimated purity by Coomassie-stained SDS-PAGE, although the amount of required protein will depend on instrumentation. The sample is diluted ten- to twenty-fold into D₂O-based buffer (25°C, pH 7.4) for a given amount of time, followed by 1:1 dilution into cold H₂O-based quench buffer at low pH (4°C, pH 2.3) to minimize amide proton back-exchange. Typically, protein buffers with minimal temperature sensitivity are preferred, such as HEPES or sodium phosphate buffers, and buffer concentration should be at least 10 times lower than that of quench buffer so that pH will be lowered to the acidic pH upon quenching. Samples are then rapidly digested using an acid-stable protease such as pepsin, followed by a short (~ 7 min) linear acetonitrile gradient for peptide separation and subsequent mass spectrometry analysis for peptide detection and identification. The deuterium uptake profiles of individual peptides are then compared across various labeling time points to elucidate local exchange behavior. Datasets can be presented in a number of ways to connect region-specific deuterium uptake values to protein structure. In a later section, we provide a detailed protocol for collecting an HDX-MS dataset, as well as suggestions for presenting HDX-MS data.

Figure 2.

Overview of a typical HDX-MS workflow. The protein sample is first purified and its purity is assessed using standard approaches such as gel filtration chromatography and SDS-PAGE (discussed in section 5.1). Hydrogen-deuterium exchange samples are then prepared by diluting the protein sample into D₂O-based buffer for a given amount of time, t, and the reaction is quenched at low pH and low temperature (discussed in section 5.2). During this reaction, amide protons (shown in yellow) exchange with deuterium (shown in purple) at different rates that depend on their structural context. The deuterated samples are then digested using an acid-stable protease such as pepsin, and the resulting peptides are separated and analyzed by liquid chromatography-mass spectrometry (LC-MS) (discussed in section 5.3). The deuterium uptake levels for a given peptide can then be presented as a function of time in a classic uptake plot shown at bottom right, or by other useful data representations that include structural context (discussed in section 6).

3. HDX-MS TO PROBE DYNAMICS OF THE KINASE FOLD

HDX-MS had an early impact on the kinase field (Lorenzen & Pawson, 2014). The laboratory of Natalie Ahn paved the way with HDX-MS studies of ERK2 that shed light on the effects of phosphorylation (as mentioned above) and nucleotide binding (Hoofnagle et al., 2001; Lee et al., 2005). While binding of the nucleotide probe AMP-PNP impacted the N-lobe of ERK2 similarly whether the activation loop was phosphorylated or not, a key regulatory region known as the DFG motif was specifically protected from exchange upon AMP-PNP binding to the phosphorylated form of ERK2 (Lee et al., 2005). This result suggests that ERK2 activation loop phosphorylation and nucleotide binding cooperate to properly orient the active site (which contains the DFG motif) for catalysis. In a separate study, this group further utilized HDX-MS to determine the location within the ERK2 kinase domain of docking sites that bind regions of substrates distal to the site of phosphorylation (Lee et al., 2004). These docking sites are key determinants of substrate specificity in mitogen-activated protein kinases (Miller & Turk, 2018).

More recent studies have benefitted from the advances in both LC-MS instrumentation and data analysis software to enable greater peptide coverage and more complex ligand binding studies. For example, a recent set of studies utilized HDX-MS to uncover regulatory inter-domain interactions within the kinase AKT1. AKT1 has an N-terminal pleckstrin homology (PH) domain that is known to bind phosphatidylinositol-3,4,5-trisphosphate (PtdIns(3,4,5)P₃) in the plasma membrane to promote AKT1 kinase activity (Siess & Leonard, 2019). HDX-MS experiments of full length AKT1 revealed that its binding to liposomes containing PtdIns(3,4,5)P₃ retained protection of the PH domain from exchange, but increased exchange in the C-terminal kinase domain, supporting a model of autoinhibitory intramolecular PH domain/kinase domain interactions that are disrupted when the PH domain binds phosphoinositides (Lucic et al., 2018). Subsequent crystallographic studies showed that the PH domain docks onto the C-lobe of the kinase domain in a manner reminiscent of its engagement with PtdIns(3,4,5)P₃ (Truebestein et al., 2021). These studies nicely illustrate both the capability of HDX-MS to address or refine structural hypotheses and the power of combining HDX-MS with complementary approaches.

In its current state, HDX-MS is most useful for qualitative biophysical inquiry, making it suitable for comparative studies across different peptide-, ligand-, drug-, or lipid-bound conditions or across differently mutated protein variants. In addition, HDX-MS is exceptionally well suited for identifying intrinsically disordered regions within proteins. This can be a particularly valuable feature in the design of constructs for protein crystallization, for example, employing HDX-MS to identify unstructured regions that would be a barrier to crystal packing (Fowler et al., 2016). In the coming years, we predict that the application of HDX-MS will increasingly impact our understanding of drug binding on kinase domain dynamics. Although HDX-MS is already an important component in drug discovery (Masson et al., 2017), it will be a key technique for understanding how current FDA-approved kinase inhibitors (such as the numerous and structurally diverse inhibitors of the epidermal growth factor receptor, or EGFR) bias kinase conformational dynamics in a way that could modulate kinase non-catalytic functions (Kung & Jura, 2016, 2019). Indeed, the Ahn lab recently used HDX-MS to reveal differences in activation loop dynamics upon binding to ERK2 of two inhibitors that induce distinct conformations (Pegram et al., 2019). As the chromatography, mass spectrometry, and HDX-MS data processing systems improve, we also predict more insight into allostery within larger macromolecular complexes. For example, the binding site of the stress response kinase GCN2 (general control nonderepressible 2) within purified ribosomes was mapped to the ribosomal P-stalk region, and GCN2 was shown to be allosterically activated by this interaction (Inglis et al., 2019) in a tour de force study enabled by HDX-MS. Despite these examples, studies of the protein kinases within higher order complexes have largely lagged behind those of the lipid kinases, which includes impressive work by the Burke, Hurley, and Williams laboratories, and from which useful insights give us a glimpse of what the protein kinase field can expect in the future (Burke et al., 2011; Dornan et al., 2020; Rostislavleva et al., 2015; Vadas & Burke, 2015; Young et al., 2016).

4. HDX-MS AS A TOOL TO STUDY PSEUDOKINASE DYNAMICS

HDX-MS has only recently been applied to studies of pseudokinases. Lacking key residues thought necessary for catalytic activity, pseudokinases were initially considered remnants of evolution that may serve primarily as scaffolds for assembly of signaling complexes (Zeqiraj & van Aalten, 2010). As the field has progressed, more dynamic roles for pseudokinases have been discovered, including roles as signaling switches and allosteric activators (Boudeau et al., 2006; Mace & Murphy, 2021). Key to this shift in perspective has been the appreciation that pseudokinases are not simply rigid molecules as the term scaffold implies, but are just as dynamic as their catalytically active relatives. Two key studies highlight the power of HDX-MS for investigating the conformational dynamics of pseudokinase domains and for helping to settle the debate over rigidity versus dynamics in pseudokinases.

In one study on the highly divergent mixed lineage kinase domain-like (MLKL) pseudokinase, Petrie et al. used HDX-MS to understand its ‘molecular switch’-like function (Petrie et al., 2018). MLKL is the terminal effector protein in the necroptosis pathway that mediates pro-inflammatory cell death (Murphy, 2020). MLKL contains an N-terminal four-helix bundle (4HB) domain that is thought to engage lipids and permeabilize the plasma membrane to execute the final step in necroptosis (Petrie et al., 2019). Release/exposure of the 4HB domain and MLKL oligomerization (also crucial) are tightly regulated by the C-terminal pseudokinase domain of MLKL. In several structural and functional studies, Murphy and colleagues have discovered key steps in this process, including the importance of conformational switching of the pseudokinase domain (Murphy, 2020; Murphy et al., 2013; Petrie et al., 2018). Although MLKL lacks kinase activity, its conserved ability to bind ATP appears to be a crucial feature for MLKL tetramerization, which is required for necroptosis. In an effort to understand allosteric communication between the 4HB and pseudokinase domains of MLKL, Petrie et al. turned to HDX-MS. One key experiment examined the impact of ATP binding to a monomeric variant of MLKL and revealed allosteric effects on both helix αC and the adjacent region of the 4HB domain—both of which increased in hydrogen exchange in the presence of ATP, suggesting inter-domain communication resulting from ATP-site ligation. HDX-MS studies like those of Petrie et al. can help illuminate how pseudokinase domains contribute in larger signaling complexes and how conformational dynamics, likely influenced by nucleotide binding, direct and control pseudokinase domain signaling.

Not all pseudokinases have retained the ability to bind ATP (Murphy et al., 2014). In our own recent study, we leveraged HDX-MS to address a longstanding question of whether pseudokinases incapable of ATP binding are rigid or dynamic molecules (Sheetz et al., 2020). An unusual group of receptor tyrosine kinases (RTKs) with intracellular pseudokinase domains, namely ROR1, ROR2, RYK, and PTK7, all play important roles in WNT signaling. Through X-ray crystallography, we showed that the pseudokinase domains of these receptors all adopt an inactive conformation highly reminiscent of IRK, with their ATP-sites fully occluded by side-chains (Artim et al., 2012; Mendrola et al., 2013; Sheetz et al., 2020). Given suggestions in the field that pseudokinases that do not bind ATP may be ‘locked’ in a more rigid conformation state (Kornev & Taylor, 2009; Patel et al., 2017; Scheeff et al., 2009) to function as rigid scaffolds, we turned to HDX-MS to compare the intrinsic dynamics of these four pseudokinases with those of the bona fide kinase IRK. We found that the intrinsic dynamics of the RTK pseudokinase domains are largely similar to those of IRK, particularly in functionally dynamic regions such as the activation loop and the signature αC helix (Figure 3A). To understand whether the conformation of the pseudokinase domains can be modulated though binding of small molecules – which might be useful in exploiting them as therapeutic targets – we screened for compounds that bind these four pseudokinases. HDX-MS served as a valuable tool in determining that small molecules that bind one of these pseudokinases (ROR1) can disrupt the conformation, impacting the dynamics of the ROR1 pseudokinase in a manner consistent with a transition towards a more active-like state. In the next section, we present a detailed protocol for collecting an HDX-MS dataset to study drug-induced changes in pseudokinase dynamics, using our ROR1 experiment as an example.

Figure 3.

A) HDX-MS comparisons of IRK with the pseudokinase domains of ROR1, ROR2, RYK, and PTK7 demonstrate that IRK-like dynamics are largely conserved in pseudokinases that do not bind nucleotide (Sheetz et al., 2020). HDX data (mean ± SD for three independent labeling experiments) at 10 s and 10 min are shown for the PTK7 (slate blue), ROR2 (magenta), RYK (green), and ROR1 (cyan) pseudokinase domains. Data for IRK (gray) are depicted as the range for each point. The x axis represents IRK-equivalent median residue number. The % exchange value for the pseudokinase domains rises and falls across the sequence with a similar pattern to that seen in IRK, suggesting similar conformational dynamics (Sheetz et al., 2020).

5. PROTOCOL: BOTTOM-UP HDX-MS TO PROBE LIGAND-INDUCED CONFORMATIONAL DYNAMICS OF A PSEUDOKINASE DOMAIN

5.1. Purification of the ROR1 pseudokinase domain from insect cells

5.1.1. Equipment

1. Standard static incubator for insect cell culture, maintained at 27°C (e.g. Fisherbrand Isotemp BOD Refrigerated Incubator, 566 L, Porcelain Steel, Fisher Scientific, catalog # 3720A)

2. Standard shaking incubators for insect cell culture, maintained at 27°C (e.g. Multitron, INFORS-HT)

3. Standard shaking bacterial incubator set to 37°C (e.g. Eppendorf 1–24 Incubating Shaker, catalog #14-285-852)

4. Standard refrigerated tabletop centrifuge (e.g. Sorvall ST 40R, ThermoFisher Scientific, catalog # 75-004-525)

5. Standard refrigerated benchtop microcentrifuge (e.g. Eppendorf, catalog # 5406000640)

6. Standard refrigerated high speed floor centrifuge for large volumes (e.g. Avanti JXN-26, Beckman Coulter, catalog # B38619, with a 6 × 1 l fixed-angle rotor JLA-8.1000, Beckman Coulter, catalog # 969328)

7. Standard refrigerated high speed floor centrifuge for medium volumes (e.g. Sorvall Lynx 6000, catalog # 75006590, with a T29-8 x 50 fixed angle rotor, catalog # 75003009)

8. Water bath maintained at 42°C

9. Sterile standard 1.6 ml microcentrifuge tubes (e.g. VWR, catalog # 490004–436)

10. Sterile standard 6-well tissue culture plates (e.g. Corning, catalog # 3471)

11. Sterile standard T25 and T75 (e.g. Corning, catalog # 430639 and 3290)

12. Sterile standard glass unbaffled 125 and 300 ml DeLong flasks for insect cell propagation (Bellco Glass Inc., catalog # 2510-00125 and 2510-00300)

13. Sterile standard unbaffled 2.8-liter Fernbach flasks (e.g. Chemglass Life Sciences, catalog # CLS-2020-11)

14. Sterile Falcon 15 ml conical centrifuge tubes (ThermoFisher Scientific, catalog # 14-959-53A)

15. Nanodrop Microvolume UV-Vis Spectrophotometer (e.g. Nanodrop One, ThermoFisher Scientific, catalog #ND-ONE-W)

16. QSonica Q125 ultrasonic processor with a 1/8-inch mini-probe (catalog # 1204P07)

17. Cole-Parmer Ultrasonic Processor with a ¼-inch (6.5-mm) tapered microtip (catalog # EW-04711–70)

18. Tricorn 5/50 column (Cytiva, catalog # 28406409) packed with Fractogel trimethylammoniumethyl (TMAE) anion exchange resin (Millipore Sigma, catalog # 1168810100)

19. Tricorn 5/50 column (Cytiva, catalog # 28406409) packed with Fractogel Sulfoisobutyl (SO₃⁻) cation exchange resin (Millipore Sigma, catalog # 1.16890.0100)

20. Centrifugal spin concentrating unit (e.g., Amicon Ultra, 10,000 Dalton molecular weight cutoff, Millipore Sigma, catalog # UFC801024)

21. Fast protein liquid chromatography (FPLC) system (e.g., AKTA Pure, Cytiva)

5.1.2. Reagents

1. pFastbac1 6xHis-ROR1^457−752 plasmid DNA (or pseudokinase of choice)

2. Antibiotic stocks: 50 mg/ml kanamycin, 7 mg/ml gentamicin, 10 mg/ml tetracycline

3. Bluo-Gal (e.g., Invitrogen, catalog # 15519028)

4. Isopropyl-β-D-thiogalactoside (IPTG; Sigma-Aldrich catalog # 367-93-1)

5. Ethanol, molecular biology grade

6. Isopropanol, molecular biology grade

7. Phosphate-buffered saline (PBS)

8. RNase A from bovine pancreas (ThermoFisher Scientific, catalog # J61996.MC)

9. Tris(2-carboxyethyl)phosphine (TCEP; Hampton Research, catalog # HR2–801)

10. Luria Broth (LB) media: 1% w/v tryptone, 0.5% w/v yeast extract, 1% w/v NaCl

11. Super Optimal broth with Catabolite repression (SOC) media: 2% w/v tryptone, 0.5% w/v yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl₂, 10 mM MgSO₄, and 20 mM glucose

12. MAX Efficiency DH10Bac Competent Cells (ThermoFisher Scientific, catalog # 10361012)

13. Solution I: 15 mM Tris-HCl pH 8, 10 mM ethylenediaminetetraacetic acid (EDTA)

14. Solution II: 0.2 M NaOH, 1% Sodium dodecyl sulfate (SDS)

15. Solution III: 3 M potassium acetate, pH 5.5

16. Spodoptera frugiperda Sf9 cells (Expression Systems, catalog # 94–001S)

17. Serum-free Sf9 medium such as ESF921 (Expression Systems catalog # 96-001-01)

18. Cationic lipid transfection reagent (e.g., Gibco Cellfectin II, catalog #10362100)

19. TE Buffer: 10 mM Tris, pH 8.0, 1 mM EDTA

20. Lysis buffer: 50 mM Tris-HCl, pH 8.5, 300 mM NaCl, 10 mM 2-mercaptoethanol, 10 mM imidazole, 1 mM phenylmethylsulfonyl fluoride (PMSF)

21. cOmplete protease inhibitor cocktail, EDTA-free (Millipore Sigma, catalog # 11873580001)

22. Benzonase nuclease (Millipore Sigma, catalog # 70746–4)

23. Nickel nitrilotriacetic acid (NiNTA) agarose beads (Qiagen, catalog # 30210)

24. Mouse monoclonal anti-his tag antibody, clone 6AT18 (Sigma-Aldrich, catalog # SAB1305538)

25. AEX Buffer: 50 mM Tris-HCl, pH 8.0, 2 mM dithiothreitol (DTT)

26. CEX Buffer: 50 mM MES, pH 6.0, 2 mM DTT

27. SEC Buffer: 20 mM HEPES, pH 7.0, 150 mM NaCl, and 100 μM TCEP

28. Laemmli Sample Buffer (4X): 125 mM Tris, pH 6.8, 4% SDS, 40% glycerol, 20% 2-mercaptoethanol, 0.02% Bromophenol Blue

29. Tris-Glycine SDS-PAGE Running Buffer (10X): 250 mM Tris, 1.92 M glycine, 1% SDS, pH 8.3

30. Coomassie Stain: 50% methanol, 10 glacial acetic acid, 0.1% Coomassie Brilliant Blue R-250 (Bio-Rad, catalog # 1610400)

31. Coomassie Destain: 40% methanol, 10 % glacial acetic acid

5.1.3. Expression and purification of the ROR1 pseudokinase domain

5.1.3.1. Generation of baculovirus

1. Prior to start, ensure that the pseudokinase or kinase domain of interest (in this case, ROR1 pseudokinase domain, residues 457–752) is cloned into the pFastbac1 vector (or similar). This protocol uses nickel ion affinity purification, so just upstream of the pseudokinase open reading frame the construct contains an N-terminal hexahistidine tag. Verify cloning by Sanger sequencing.

2. Prepare fresh LB/agar bacterial plates for the generation and isolation of recombinant bacmid DNA. Plates should contain 50 μg/ml kanamycin, 7 μg/ml gentamicin 10 μg/ml tetracycline, 100 μg/ml Bluo-gal, 40 μg/ml IPTG. Air dry plates overnight, then store at 4°C wrapped in both plastic and aluminum foil for a maximum of six weeks.

3. Transform 20–50 μl of DH10Bac chemically component cells by adding 0.5 μg purified pFastbac1 6xHis-ROR1^457−752, incubating for 30 min on ice, heat shocking at 42°C for 30 s in a pre-warmed water bath, adding 500 μl SOC media, and incubating for 4 h at 37°C with shaking at 225 RPM. Plate 50 μl and 150 μl onto two separate plates and incubate for 48 h at 37°C.

4. Check plates after 48 h. Pick a white colony from a bacmid plate and re-streak onto a fresh bacmid plate, and incubate for another 48 h at 37°C.

5. Using a sterile pipet tip, pick a single white colony from a bacmid plate, and inoculate a 5 ml LB culture containing 50 μg/ml kanamycin, 7 μg/ml gentamicin, and 10 μg/ml tetracycline. Incubate with shaking at 37°C for 12–24 h. It is recommended to also pick a blue colony as a negative control for bacmid PCR described below.

6. Isolate the bacmid DNA. Spin down 1.5 ml culture at 14,000 × g for 1 min to pellet cells. Remove supernatant and resuspend pellet in 0.3 ml of Solution I (15 mM Tris-HCl pH 8, 10 mM EDTA, with freshly added 100 μg/ml RNase A). Add 0.3 ml of freshly made Solution II (0.2 M NaOH, 1% SDS) and gently mix. Incubate for 5 min at room temperature. Add 0.4 ml of Solution III dropwise (3 M potassium acetate, pH 5.5), adding slowly, and mixing gently during addition by flicking tube. A thick white precipitate (containing protein and genomic DNA) should form. Incubate sample on ice for 5–10 minutes. Centrifuge for 10 min at 14,000 × g at 4°C. Transfer supernatant to a fresh tube containing 0.8 ml of 100% isopropanol. Avoid transferring white precipitate. Mix gently by inverting, and incubate on ice for 5–10 min. Centrifuge sample for 15 min at 14,000 × g at room temperature. Remove supernatant and add 0.5 ml of 70% ethanol. Invert to wash pellet. Centrifuge 5 minutes at 14,000 × g at room temperature. Carefully remove supernatant with a pipette. Allow pellet to air-dry in an open tube for ~30 min at room temperature. Add 40 μl of TE buffer (10 mM Tris, pH 8.0, 1 mM EDTA) to the pellet and incubate at room temperature for 10 min. Gently tap the tube to mix and check the DNA concentration using a Nanodrop spectrophotometer; typical yield is 10–100 μg bacmid DNA. Check for proper insertion of ROR1 (or other pseudokinase/kinase domain insert of interest) by PCR and DNA gel electrophoresis. Use bacmid DNA for transfection as soon as possible. For long-term storage, make aliquots and store at −20°C to avoid freeze/thaw cycles.

7. Transfect insect cells with recombinant bacmid DNA. For ROR1 production, we use Sf9 cells from the fall armyworm Spodoptera frugiperda. Healthy Sf9 cells are typically maintained in ESF921 serum-free medium from Expression Systems at a density ranging from 0.7 to 8.0 × 10⁶ cells/ml by shaking at 110 RPM at 27°C, and cells are split every 3–4 days. On the day prior to transfection, 35 ml propagation culture is seeded at a final density of 1×10⁶ cells/ml in an unbaffled 125 ml glass DeLong shaker flask. On the day of transfection, seed 2 ml of cells at a final density of 0.8 × 10⁶ into wells of a 6-well plate. Prepare a transfection mixture by adding 5 μl of bacmid (at 400 ng/μl for a total of 2 μg bacmid DNA) + 100 μl cell culture media to 8 μl Cellfectin II + 100 μl medium and incubate at room temperature for 30 min. Add this entire mixture to one of the wells containing 2 ml seeded cells. Include a negative control condition by transfecting cells with bacmid lacking the ROR1 insert for comparison during test expressions. After incubation at 27°C for 4–5 days in a static incubator, agitate the cells by pipetting, collect the entire contents, and centrifuge at 1000 × g for 10 min. Store the supernatant in 2 ml tubes at 4°C.

8. Scale up the P0 transfection virus to a P1 virus stock. In a T75 flask, seed 15 ml of Sf9 cells at a final density of 0.6 × 10⁶ cells/ml and allow them to adhere for 30 min. Gently add 1 ml P0 virus to the flask and allow for viral amplification over 4 days in a static incubator maintained at 27 °C. Collect the P1 supernatant by transferring to a Falcon 15 ml conical tube followed by gentle centrifugation at 1000 × g for 10 min. Store the viral stock at 4°C.

5.1.3.2. Small-scale protein test expression

1. Before scaling up for large-scale expression, it is prudent to ensure efficient viral production and soluble protein expression with a smaller scale test expression. Begin by seeding 15 ml of healthy Sf9 cells into a T75 flask at a density of 1.5–2.0 × 10⁶ cells/ml. After allowing cells to adhere for 30 min, infect the cells with 1 ml of P1 virus. Alongside, prepare control conditions of uninfected cells and cells infected with a non-ROR1 control. Incubate cells at 27°C in a static incubator for 3 days.

2. Harvest the cells by transferring cells and supernatant into Falcon 15 ml conical tubes and spinning down at 1000 × g for 10 minutes. Pour off supernatant and gently resuspend cells in ice cold 1X PBS buffer, and repeat the centrifugation step. Again, pour off supernatant and resuspend cells in 1 ml ice cold lysis buffer (50 mM Tris-HCl, pH 8.5, 300 mM NaCl, 10 mM 2-mercaptoethanol, 10 mM Imidazole, 1 mM PMSF, protease inhibitor cocktail) and transfer to a 1.6 ml microcentrifuge tube.

3. Sonicate to lyse cells using a mini-sonicator probe. For 1 ml volumes, we typically use a QSonica Q125 ultrasonic processor with a 1/8-inch mini-probe. We sonicate test expressions using two rounds of 30 s sonication at 50% amplitude with 1s-1s on-off cycles and brief end-over-end mixing of lysate between rounds. Sonication is performed at in a 4°C cold room with cell suspensions kept on ice during sonication.

4. Spin down the lysate at 4°C in a refrigerated microcentrifuge at 14,000 × g for 30 min. Transfer the supernatant to a fresh 1.6 ml microcentrifuge tube and add 50 μl of pre-equilibrated 50% NiNTA-agarose slurry. Rock samples at 4°C for 30 min. Wash the beads three times with lysis buffer containing 40 mM imidazole using a gentle centrifugation speed of 500 × g.

5. Remove final wash supernatant and elute protein by adding 25 μl of lysis buffer + 5 mM EDTA and 15 μl of 4X Laemmli sample buffer.

6. Assess expression levels by running 10 μl of sample onto two SDS-polyacrylamide gel electrophoresis (PAGE) gels containing 12% acrylamide (37.5:1 acrylamide:bis-acrylamide), staining one gel with Coomassie blue stain and using the other for a Western blot to probe for the 6xHis tag (we use a mouse anti-His antibody from Sigma-Aldrich, clone 6AT18). Compare ROR1 expression to control conditions and to a molecular weight standard, such as Precision Plus Dual Color Protein Standard (Bio-Rad). If little-to-no protein is detectable by Western blotting, add an additional viral amplification step by repeating step 8 in the previous section using the P1 stock to generate a P2 stock, and repeat the test expression. Ensure that cells are >98% viable and growing at a healthy rate (~24–40 hours doubling time).

5.1.3.3. Large-scale protein expression

1. Once expression of ROR1 in a small-scale test is confirmed, one can confidently proceed with large-scale protein expression. Begin by generating fresh baculovirus for large-scale expression by seeding 100 ml of healthy Sf9 cells into a 250 ml flask at a density of 1.0 × 10⁶ cells/ml and incubate at 27°C shaking at 110 RPM. The next day, cells should have a density of ~1.5–2.0 × 10⁶ cells/ml. Add 1–2 ml of P1 virus stock and incubate for 4–5 days at 27°C shaking at 110 RPM. Harvest the P2 virus by spinning down the cells at 1000 × g for 10 min. Transfer the supernatant to fresh tubes and store at 4°C until ready for large-scale infection.

2. Prior to large-scale protein purification, we recommend performing the following procedures on a small scale (15–50 ml media) to ensure that protein expression conditions are optimal.

3. To proceed with large-scale expression, scale healthy cells up to 4 l in four 2.8-liter unbaffled Fernbach flasks (1 liter culture / flask) from initial starting culture using standard conditions: 27°C shaking at 110 RPM.

4. Once cells reach 4 l at 1.5–2.0 × 10⁶ cells/ml cell density, infect the cells with fresh P2 virus generated in step one. Use ~24 ml virus per liter of culture. Allow cells to incubate at standard shaking conditions for three days.

5.1.3.4. Purification of the ROR1 pseudokinase domain

1. To harvest cells for purification, pour each liter culture into a 1 l centrifuge bottle and spin down the cells at a slow speed (so as not to lyse cells) of 2000 × g for 10 min at 4°C. We use an Avanti JXN-26 refrigerated floor centrifuge with a 6 × 1 liter fixed-angle rotor (JLA-8.1000 from Beckman Coulter) to perform this step. Slowly pour off media and gently wash the pellets with 0.5–1 l ice-cold PBS and repeat centrifugation step.

2. Slowly pour off PBS and gently resuspend pellets in 40 ml of ice-cold lysis buffer (50 mM Tris-HCl, pH 8.5, 300 mM NaCl, 10 mM 2-mercaptoethanol, 10 mM Imidazole, 1 mM PMSF, protease inhibitor cocktail) and transfer to 50 ml conical tubes. Add 250 units benzonase per resuspended pellet.

3. Sonicate to lyse cells. Though conditions may vary depending on sonicator and probe, we use a Cole-Parmer 500-Watt Ultrasonic Processor with a ¼-inch (6.5-mm) tapered microtip to lyse 40 ml of cell suspension at a time. We use a gentle sonication program of two rounds of 30 s sonication at 35% amplitude with a 1s-1s on-off cycles and brief end-over-end mixing of lysate between rounds. Sonication is performed in a 4°C cold room with cell suspensions kept on ice during sonication. Clarify lysate by centrifugation for 30 min at 30,000 × g at 4°C.

4. Load clarified lysate onto a pre-washed Kontes column containing 3 ml pre-equilibrated NiNTA resin. Wash beads with 50 column volumes of lysis buffer.

5. Elute hexahistidine-tagged ROR1 with step-wise elutions containing increasing concentrations of imidazole up to 400 mM, taking one fraction per column volume. We typically elute with 3 column volumes (CVs) of lysis buffer containing 40 mM, 100 mM, 200 mM, and finally 400 mM imidazole. Assess purity and yield at this stage by SDS-PAGE using Tris-Glycine based running buffer and Coomassie staining. We typically prepare SDS-PAGE samples by adding 9 μl of each fraction to 3 μl of 4X Laemmli sample buffer and run each sample side-by-side in a 15-well, reducing SDS-PAGE gel containing 12% acrylamide (37.5:1 acrylamide/bis-acrylamide), running the gel for ~1 hour at a constant 150 V, or until dye front has reached the bottom of the gel. We run samples alongside a Precision Plus Protein Dual Color Standard (Bio-Rad), expecting the ROR1^457−752 protein product to run at a molecular weight of approximately 35 kDa.

6. Combine desired fractions and dilute with freshly made, ice-cold anion exchange buffer AEX (50 mM Tris-HCl, pH 8.0, 2 mM DTT) until the final NaCl concentration is 20 mM. Using a FPLC instrument such as an AKTA Pure (Cytiva) maintained at 4°C load protein onto a TMAE anion-exchange column pre-equilibrated with buffer AEX + 20 mM NaCl. We typically use Tricorn 5/50 columns (Cytiva) packed in-house with Fractogel TMAE strong anion exchanger resin (Millipore Sigma) with a standard flow rate of 3 ml/min. Elute ROR1 protein with a linear NaCl gradient up to 1 M. We typically elute over 60 CVs (60 ml for a 1 ml column) and collect the eluate in 2 ml fractions. Assess purity and yield by SDS-PAGE and Coomassie staining using the same gel and running conditions as above. For this step, we typically run 15 μl fraction sample + 5 μl (4X) Laemmli sample buffer per well and compare samples with Precision Plus molecular weight standards as before. If necessary, repeat this step using cation exchange in freshly made, ice-cold buffer CEX (50 mM MES, pH 6.0, 2 mM DTT) with an NaCl gradient of 50 mM to 1 M on an SO₃⁻ cation exchange column.

7. Combine desired fractions and concentrate using a pre-washed centrifugal spin concentrating unit with a 10,000 Dalton molecular weight cutoff at 3000 × g until the sample is concentration to 0.5–1 ml. Using an FPLC, inject up to 0.5 ml concentrated protein onto a Superdex 75 10/300 gel filtration column pre-equilibrated with freshly made, ice-cold 20 mM HEPES, pH 7.0, 150 mM NaCl, and 100 μM TCEP. Following gel filtration of ROR1, run fresh gel filtration standards (Bio-Rad) to directly compare with elution conditions of ROR1. ROR1^457−752 should elute as a globular monomer of approximately 35 kDa, and a single symmetrical peak should indicate high purity. Further assess purity and yield by SDS-PAGE and Coomassie staining similarly to step 6, looking for the absence of non-ROR1^457−752 bands. Additionally, protein purity can be assessed using intact mass spectrometry.

5.2. Hydrogen/deuterium exchange reactions of ROR1 +/− ponatinib

5.2.1. Equipment

1. Standard set of single-channel and/or multi-channel benchtop pipettes

2. 2.0 ml dolphin microcentrifuge tubes, not autoclaved (e.g. VWR, catalog # 490010–618)

3. Ice bucket or cold room

4. Liquid nitrogen

5.2.2. Reagents

1. Approximately 0.5 mg/ml purified ROR1 protein from section 5.1 above

2. Deuterium oxide (99.9%), low paramagnetic (Cambridge Isotope Laboratories, Inc. catalog # DLM-11)

3. Exchange buffer, H₂O-based: 20 mM HEPES, pH 7.4, 200 mM NaCl, 100 μM TCEP

4. Exchange buffer, D₂O-based: 20 mM HEPES, pD 7.4, 200 mM NaCl, 100 μM TCEP (Note: pD = pH + 0.4. Buffer should be adjusted to a desirable pD by adding DCl (Cambridge Isotope Laboratories, Inc. catalog # DLM-2) or NaOD (Cambridge Isotope Laboratories, Inc. catalog # DLM-45).

5. 1 mM Ponatinib (Selleckchem, catalog #S1490), or another drug/ligand of interest in 100% DMSO

6. Quench buffer: 100 mM sodium phosphate buffer, pH 2.4, 2 M optical-grade guanidine hydrochloride (ThermoFisher), 1% formic acid (FA)

7. Urea-D4 (Cambridge Isotope Laboratories, Inc. catalog # DLM-1269) dissolved in the D₂O exchange buffer, 7 M solution at pD 7.4

5.2.3. HDX sample preparation

1. Preparation of the HDX samples can be done well in advance of data collection, as samples can be stored at −80°C for up to one month. Alternatively, robotic sample preparation systems can be utilized to improve reproducibility and increase throughput. Robotic HDX sample preparation typically takes longer than 8–10 h for an entire labeling time course series, including blank wash steps between runs. Thus, for protein samples that may be unstable over the course of several hours at room temperature (while awaiting processing by the robot), manual sample preparation should be used instead. The use of multi-channel pipettors for manually preparing manual HDX-MS samples with technical replicates can save significant amount of time.

2. For manual sample preparation, begin with a ~0.5 mg/ml stock solution of purified ROR1 pseudokinase domain in size exclusion buffer. To prepare undeuterated standards, pipette 5 μl protein into a fresh non-autoclaved 2.0 ml dolphin microcentrifuge tube. Add 95 μl of H₂O-based buffer containing 20 mM HEPES, pH 7.4, 200 mM NaCl, 100 μM TCEP. Perform a mock quench step by transferring 90 μl of diluted ROR1 into a fresh tube on ice containing 90 μl of ice-cold quench buffer (100 mM sodium phosphate buffer, pH 2.4, 2 M guanidine hydrochloride, 1% FA). Immediately snap-freeze in liquid nitrogen and store at −80°C.

3. Repeat step 2 at least three times for technical replicates, or use a multi-channel pipettor to make technical replicates. We recommend preparing several additional undeuterated samples if LC-MS optimization is required.

4. To prepare deuterium-labeled samples, repeat step 2 but replace H₂O-based buffer with D₂O-based buffer containing 20 mM HEPES, pD 7.4, 200 mM NaCl, 100 μM TCEP at 25°C. Incubate the samples at 25°C for the desired amount of time and quench using ice-cold (4°C) quench buffer. Immediately following the cold quench step, sample tubes should be immersed in liquid nitrogen to snap freeze and then stored at −80°C until data collection. For kinase/pseudokinase samples such as ROR1, we generally recommend the following time points as a starting point: 10 s, 1 min, 10 min, 1 h, and 2 h. This range is a “broad sweep” found to be appropriate for monitoring uptake kinetics of peptides within typical protein kinase domains and was selected based on scouting experiments.

5. Again, repeat step 4 for a minimum of three technical replicates and for the desired number of time points. Samples should be stored at −80°C prior to data collection. With adequate LC-MS systems, it should not be necessary to collect replicate data in series and can be carried out over multiple days if preferred (Houde et al., 2011).

6. In order to assess the impact of small molecules on the dynamics of a pseudokinase (or kinase) domain, this procedure can be performed in the presence of the compound for comparison with data obtained with the compound. For example, to measure changes in ROR1 dynamics when bound to ponatinib, the initial protein stock should be prepared by diluting 1 mM ponatinib (in 100% DMSO) 20-fold into ~0.5 mg/ml ROR1 in size exclusion buffer, to give a final 50 μM ponatinib concentration. Additionally, the D₂O-based HDX buffer should also include 50 μM ponatinib (and 5% DMSO) final concentration to match. The remaining sample preparation conditions are unchanged. Here, it is important to add sufficient drug so that you have at least ~85% drug-bound population after 10- to 20-fold D₂O dilution (Kochert et al., 2018), so knowledge of the affinity of the compound for your pseudokinase is important.

7. Finally, fully deuterated standards should be prepared to correct the deuterium-labeled dataset for back exchange. This can be done by repeating step 2, but replacing the H₂O-based buffer with a D₂O-based denaturant such as 7 M urea. We typically incubate protein samples in 7M-urea-D₂O (pD 7.4) for 1–10 min to exchange all amide hydrogen atoms in the protein.

5.3. In-line pepsin digestion, reversed-phase chromatography, and mass spectrometry

5.3.1. Equipment

1. UPLC (e.g., Waters ACQUITY UPLC M-Class System with HDX Technology)

2. Mass spectrometer (e.g., Waters Synapt G2-Si HDMS Q-TOF Mass Spectrometer)

3. UPLC C18 peptide trap column (e.g., Waters ACQUITY UPLC BEH C18 VanGuard Pre-column, 2.1 mm × 5 mm, catalog # 186003975)

4. UPLC C18 separation column (e.g., Waters AQUITY UPLC BEH C18 column, 1 mm × 100 mm, catalog # 186002346)

5. Pepsin column (e.g., Waters Enzymate BEH pepsin column catalog #186007233)

6. 250 μl Hamilton glass syringe, or similar, for sample injection (e.g., Waters, catalog # 430000865)

5.3.2. Reagents

1. Mobile phase A: mass spectrometry grade water + 0.1% Formic Acid (FA) (e.g., Burdick and Jackson catalog # 39253 and # 56302, respectively)

2. Mobile phase B: mass spectrometry grade acetonitrile (ACN) + 0.1% FA (e.g., Burdick and Jackson catalog # LC015)

5.3.3. LC-MS data acquisition

1. For data collection, the UPLC-MS system must be fitted with the appropriate components for HDX-MS. We use a Waters HDX nanoAcquity UPLC-M with in-line protein digestion capability equipped with a Waters Enzymate BEH pepsin column a Waters ACQUITY UPLC BEH C18 VanGuard pre-column peptide trap, and a Waters ACQUITY UPLC BEH C18 separation column. All columns are housed in a Waters HDX Manager at 2°C. Upon pepsin digestion, desalting, and peptide separation, the separated peptides are analyzed using a Waters Synapt G2-Si mass spectrometer for determining the mass of each peptic peptide by a time-of-flight (TOF) analyzer. We utilize data-independent analysis (DIA) involving MS^E data collection performed with a 15 V to 30 V ramp collision-induced dissociation (CID) for determining the amino acid sequence of each peptic peptide.

2. Begin data collection by determining the optimal LC-MS conditions for the pseudokinase/kinase domain being studied. To do this, we recommend running undeuterated protein samples with the following starting LC conditions: a 3 min desalting step at 5 % acetonitrile, followed by a 7 min 5–35% mobile phase B linear gradient.

3. We generally aim for a sequence coverage above 90%. Sub-optimal LC separation of peptides often results in poor sequence coverage, so altering the acetonitrile gradient or using an alternative reversed-phase column (such as a charge-surface-modified one, e.g., Waters ACQUITY UPLC CSH C18 separation column) for the protein under investigation should be considered when necessary. Additionally, mass spectrometers capable of ion mobility mode (such as the Waters Synapt G2-Si HDMS Q-TOF mass spectrometer) may provide a more optimal peptide coverage than the standard TOF mode alone, especially when studying large protein complexes.

4. Once optimal LC-MS conditions are determined, frozen HDX samples may then be individually thawed and immediately (to minimize back-exchange) injected onto the pepsin column for MS^E data acquisition. Wash syringe in between injections with mass spectrometry grade water + 0.1% FA. Follow each injection with a blank run by injecting water + 0.1% FA to ensure sample carryover between runs is not an issue.

5. In order to ensure reproducibility and to obtain sufficient information for statistical analysis, a minimum of three technical replications per biological sample for each labeling timepoint should be obtained. When conclusions are to be drawn about specific structural elements, we recommend a minimum of three biological repeats be performed, using separate protein preparations. Biological repeats are especially important when complex protein sample preparation procedures are used, such as the purification of post-translationally modified proteins. Based on our experience, inconsistent deuterium uptake for the same protein sample occurs when a protein is not freshly purified. Avoid using protein samples that have repeatedly been frozen and thawed, since these will give variable results as the protein integrity is compromised. The quality of the protein sample is often the most important element for reproducibility and interpretability of results. In particular, if the protein under study has any tendency to aggregate, this can be manifest as increased protection and the results can be misleading as well as irreproducible.

6. DATA ANALYSIS AND PRESENTATION

6.1. Analyzing HDX-MS data

6.1.1. Initial peptide identification

To initially assess the efficiency of the LC-MS protocol and to ultimately analyze complete HDX-MS datasets, the analysis must start with peptide identification from the MS^E spectra of peptic peptides using data obtained from the undeuterated protein sample. An MS^E experiment involves linking MS with LC information for each peptide: the parent ion m/z of each peptic peptide is determined, followed by CID for detecting the b- and y ion-series of each peptic peptide – and this information is linked with the retention time of each parent ion. The analysis software uses this mass information to determine the amino acid sequence of each peptic peptide. Several software packages are available for peptide identification; in our studies we use the ProteinLynx Global SERVER 3.0.3 (PGLS) from Waters Corporation. The undeuterated MS^E dataset is used to search specifically against the sequence of the recombinant protein of interest (e.g., 6xHis-ROR1^457−752) as well as porcine pepsin to monitor the health of the pepsin column. This initial search generates ion accounting files (IA files) that include information on the amino acid sequence, the monoisotopic as well as centroid mass, the retention time, the charge state, and the peptide intensity count of each peptic peptide.

As with peptide identification, several software packages have been developed for determining the relative deuterium uptake of each peptic peptide, including DynamX (Waters Corporation), ExMS (Kan et al., 2011; Kan et al., 2019), and HDExaminer (Sierra Analytics). Using DynamX, IA files are imported in order to determine the deuterium uptake of each peptic peptide relative to the corresponding undeuterated peptic peptide – based on the centroid mass difference of each peptide derived from the unlabeled- and labeled-protein. In our studies of pseudokinase dynamics, the following stringent MS^E data filtering are applied to minimize peptide misidentification and to increase the accuracy of peptide identification: a minimum sequence length of 5, a maximum peptide length of 20 residues, minimum products per amino acid of 0.3, and a maximum MH⁺ error of 5 ppm. Raw MS files for deuterated protein samples are uploaded for determining the centroid mass of each peptic peptide. DynamX then performs an automatic search for peptide peaks in MS spectra for deuterium-unlabeled and -labeled samples and generates relative deuterium uptake plots for each condition at different labeling timepoints, based on the centroid mass difference between the unlabeled and the labeled peptic peptide. Once the relative deuterium uptake for each peptic peptide is calculated by DynamX, it is imperative to inspect the assigned MS spectra manually in DynamX in order to discard any misassigned peptides. The deuterium uptake of a peptide at different charge states ought to be consistent; otherwise, a high standard deviation of deuterium uptake for the corresponding peptide is observed. Higher charge states tend to undergo greater back-exchange during LC-MS analysis compared to lower charge states of the same peptide. In addition, significant broadening of mass envelopes is often observed with higher charge state peptides. Therefore, it is acceptable to exclude +4 or higher charge states from the analysis, as long as all experiment conditions are processed equivalently.

It is also important to note that mass informatics software generally uses the monoisotopic mass (MH⁺) of peptides to determine their amino acid sequence. Longer peptides have a greater tendency to deviate from the expected MH⁺, which can result in peptide misidentifications. For this reason, in addition to greater back-exchange in longer peptides, we set the peptide maximum cutoff length of 20–25 residues in our analysis. The extent of back-exchange will vary depending on instrumentation, so we recommend that readers assess the extent of the back exchange under their conditions by first comparing the deuterium uptake of an unlabeled and fully-deuterated protein before proceeding to labeling time course experiments.

6.1.2. Downstream data analysis

Upon initial HDX-MS data processing, key information such as relative deuterium uptake ± standard deviation for each peptide at a given time point can be extracted for further processing and/or data visualization. We highly recommend correcting the relative deuterium uptake of each peptide for back exchange using the fully-deuterated standard sample. Calculating the relative deuterium uptake using an experimentally determined sample accounts for variability in back-exchange among different peptide lengths and sequence compositions. In order to obtain this normalized percentage of deuterium uptake (%D_t) at incubation time t for a given peptide, the following equation can be used:

$$\%D_t=100*\frac{\left(m_t-m_0\right)}{\left(m_f-m_0\right)}$$

where m_t is the centroid mass at time t, m₀ is the centroid mass of the undeuterated control, and m_f is the centroid mass of the fully deuterated control. When datasets are generated comparing two separate conditions of the same protein sample, for instance in the analysis of exchange differences upon drug binding, a percent deuteration difference value (Δ%D_exchange) can be calculated. This is expressed by the equation:

$$\Delta \%D_{\mathit{\text{exchange}}}=\%D_{x,t}\left(\mathit{\text{bound}}\right)-\%D_{x,t}\left(\mathit{\text{unbound}}\right)$$

where x is a given peptide, and t is the D₂O incubation time. Several post-processing software packages have been developed by HDX-MS users to facilitate analysis and visualization, such as DECA (Lumpkin & Komives, 2019), Deuteros (Lau et al., 2021; Lau et al., 2019), and HaDeX (Puchala et al., 2020).

6.2. Presenting HDX-MS data

As with most other methodologies, HDX-MS datasets can be presented a number of different ways. A few useful representations (visualized in Figure 5) that we find particularly informative for our studies include:

1. Individual peptide uptake plots that display the level of deuterium uptake for a particular peptide over time (Figure 5A). Such uptake plots are very useful for highlighting exchange differences (or similarities) in key regions of interest.

2. Structural heat maps that display uptake levels or differences in deuterium exchange for a given time point projected onto a structural model of the protein (Figure 5B). Such structure heat maps are invaluable for connecting dynamic HDX-MS information with structural information, for both monomeric proteins as well as protein complexes.

3. Peptide coverage heat maps that display both uptake information and the level of peptide coverage for the protein being investigated (Figure 5C). Generally, a linear diagram indicating secondary structural elements and/or domain boundaries is depicted horizontally at the top of the figure, with adjacent lines corresponding to peptide residue boundaries from peptides analyzed in the experiment. Each peptide can be colored by level of deuterium uptake or levels of differential exchange.

4. Whole dataset uptake plots that display deuterium uptake for all peptides for a given labelling experiment, with median residue number of the peptide on the x-axis (e.g., as shown in Figure 3). Datasets across timepoints, protein variants, or different experimental conditions can be combined with this representation to provide a comprehensive view of deuterium uptake differences or similarities.

Figure 5.

Example representations of pseudokinase domain HDX-MS data. A) Peptide uptake plots for differentially exchanging regions of ROR1 upon drug binding. B) Structural heat map for the difference in deuterium exchange upon ponatinib binding to the ROR1 pseudokinase domain. Data from the 1 min timepoint are projected onto the structure of ponatinib-bound ROR1, color coded as described in the associated legend. C) Coverage heat map for the difference in deuterium exchange upon ponatinib binding to the ROR1 pseudokinase domain. At the top of the panel, a secondary structure diagram for the ROR1 pseudokinase domain is shown in black. Below, short lines represent peptides from ROR1, colored according to data from the 1 min timepoint in the HDX-MS experiment. All data shown represent biological triplicates, each with technical triplicates as described (Sheetz et al., 2020). PDB ID: 6TU9 (Sheetz et al., 2020).

Finally, it is worth considering the following tips when presenting HDX-MS data at conferences or in publications:

1. Choose a visual representation that best fits your question. For example, structural heat maps are very useful for conveying how the presence of a drug alters structural dynamics in the context of the protein’s structure. When exchange differences in key regions of interest are to be highlighted, peptide uptake plots comparing different experimental conditions are often an excellent choice for conveying specific peptide information.

2. Show more data than a few selected uptake plots to avoid cherry-picking. In addition to presenting uptake plots, provide data for the entire polypeptide chain (for example using a peptide coverage heat map) so that the peptide of interest can be understood within the context of the larger protein being examined. When possible, uptake plots of multiple, overlapping peptides can provide greater confidence when interpreting exchange differences. The goal is both to provide a complete view of the dataset and to highlight key areas of interest.

3. Follow data reporting standards. Resulting from discussions at the first International Conference on Hydrogen-Deuterium Exchange in 2017, a set of community recommendations serves as useful guidelines for HDX-MS users. These guidelines provide specifications on data reporting and include useful supplemental data templates to facilitate the evaluation of HDX-MS datasets during and after peer review. We refer the reader to these important guidelines for more details (Masson et al., 2019).

7. SUMMARY

Like protein kinases, pseudokinase domains are dynamic and competent to adopt (and switch between) distinct conformation states. Although X-ray crystallography has led the way in providing high resolution structural information on kinases and pseudokinases, the resulting structural pictures are static, and the field also needs rigorous methods for assessing the dynamics of the kinase fold. HDX-MS is one technique that is well suited to probe pseudokinase dynamics in solution and to validate changes in conformation and in conformational exploration induced by binding of small molecule pseudokinase binders, protein binding partners, post-translational modifications, and mutations. In this chapter, we provide a detailed protocol for applying HDX-MS to a pseudokinase domain to assess intrinsic dynamics as well as drug-induced changes in conformational dynamics. Discovery of pseudokinase-targeted small molecules (and protein binding partners) is in its infancy, but we envision that HDX-MS will be a powerful methodology for connecting structure to mechanism in the coming years.

Figure 4.

Example LC-MS raw data.

A) Sample UPLC chromatogram of separated peptides prior to MS analysis. B) Sample stacked spectral plots from a ROR1 peptide that is protected from deuterium exchange in the presence of the drug ponatinib. Centroids of mass envelopes from each time point from these spectra are used to generate uptake plots. FDS, fully-deuterated standard.