CETSA beyond Soluble Targets: a Broad Application to Multipass Transmembrane Proteins

CETSA beyond Soluble Targets: a Broad Application to Multipass

Transmembrane Proteins Aarti Kawatkar,*,† Michelle Schefter,† Nils-Olov Hermansson,‡ Arjan Snijder,‡ Niek Dekker,‡

Dean G. Brown,† Thomas Lundbäck,‡ Andrew X. Zhang,*,† and M. Paola Castaldi*,†

†Discovery Sciences, BioPharmaceutical R&D, AstraZeneca, Boston, United States

‡Discovery Sciences, BioPharmaceutical R&D, AstraZeneca, Pepparedsleden 1, Gothenburg, Sweden

* S Supporting Information ABSTRACT: Demonstration of target binding is a key requirement for under- standing the mode of action of new therapeutics. The cellular thermal shift assay (CETSA) has been introduced as a powerful label-free method to assess target engagement in physiological environments. Here, we present the application of live- cell CETSA to different classes of integral multipass transmembrane proteins using three case studies, the first showing a large and robust stabilization of the outer mitochondrial five-pass transmembrane protein TSPO, the second being a modest stabilization of SERCA2, and the last describing an atypical compound-driven stabilization of the GPCR PAR2. Our data demonstrated that using modified protocols with detergent extraction after the heating step, CETSA can reliably be applied to several membrane proteins of different complexity. By showing examples with distinct CETSA behaviors, we aim to provide the scientific community with an overview of different scenarios to expect during CETSA experiments, especially for challenging, membrane bound targets.

S mall molecule−protein target engagement is a critical step for understanding the mechanism of action of drugs and the biology of disease-relevant targets. While biochemical and cellular reporter assays are widely used for hit identification owing to their robustness and throughput, these assays do not probe target engagement under disease relevant settings.1

Methodologies that can bridge the translational gap between screening models and disease relevant systems through evidence of cellular target engagement are thus urgently needed.

Ligand binding modulates conformational and thermal stability of proteins, and this is exploited by multiple technologies for the assessment of target engagement, including the use of reduced proteolytic digestion and resistance to thermally induced denaturation.2,3 The recently developed CEllular Thermal Shift Assay (CETSA) capitalizes on this latter biophysical principle, allowing for determination of drug target engagement in biologically relevant settings such as live cells and tissues.4,5 The CETSA methodology is based on sequential thermal denaturation and irreversible aggregation of target protein, a process that can be altered by the presence of a ligand. Separation of remaining soluble protein and irreversible aggregates is achieved either prior to detection through centrifugation or filtration4 or by choosing a readout that distinguishes between these states.5 Given the practical requirement for irreversible aggregation, CETSA has primarily been applied to soluble proteins,6,7 which readily aggregate when they denature and expose hydrophobic surfaces in live cells.4,8 CETSA has also been successfully applied to single- pass membrane proteins and, in isolated cases, to membrane- associated proteins using protocols that require detergents to extract solubilized protein, while leaving denatured and aggregated material.9,10 While the quantitative interpretation of these responses can be more complex, given that some of these proteins may take longer to form irreversible aggregates after thermal denaturation, few instances of CETSA on membrane proteins have been explored. This concept warrants an in-depth study to additional classes of membrane-bound proteins, especially multipass transmembrane proteins.11,12

Many sought after drug targets are complex multipass transmembrane receptors, e.g., G-protein-coupled receptors (GPCR) and ligand gated ion channels.13,14 For this reason, we were interested in exploring their CETSAbility; i.e., we wanted to understand to what extent denatured multipass membrane proteins in heated live cells also form nonextract- able aggregates and whether they can be practically distinguished from their native counterparts. While integral membrane targets have been studied previously using Tm shift assays,15−17 these studies are commonly based on heating of purified, detergent solubilized proteins, such that the working protocols more closely follow a CETSA workflow applied to cellular lysates. While in lysate CETSA is able to identify target engagement events, the process of lysis and membrane extraction prior to heating means that the target is not

Received:

May 19, 2019 Accepted:

July 22, 2019 Published: July 22, 2019 Articles pubs.acs.org/acschemicalbiology

Cite This: ACS Chem. Biol. 2019, 14, 1913−1920 © 2019 American Chemical Society

1913 DOI: 10.1021/acschembio.9b00399 ACS Chem. Biol. 2019, 14, 1913−1920

Downloaded via UNIV OF GOTHENBURG on October 22, 2019 at 10:26:31 (UTC).

See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. approached in the most biologically relevant form. Here, we present case studies on the application of live cell CETSA to three multipass transmembrane proteins, including ion channels and a GPCR, and highlight their unique challenges and, at times, unpredictable thermal denaturation behaviors upon engaging with small molecule ligands.

■RESULTS We began our efforts by applying CETSA to membrane targets with precedented tool compounds. In each case, optimization steps included testing different membrane extraction con- ditions to allow extraction of the target with the minimal amount of detergent. All experimental procedures reported here represent the optimized conditions. As model systems, we chose the translocator protein (TSPO), sarco/endoplasmic reticulum Ca2+-ATPase (SERCA2), and protease-activated receptor 2 (PAR2) to represent different protein classes, molecular weights, localizations, and functions.

TSPO (18 kDa) is a five-pass transmembrane domain protein localized primarily in the outer mitochondrial membrane and is expressed predominantly in steroid- synthesizing tissues, including the brain.18 One of the most characterized functions of TSPO is its role in the translocation of cholesterol from the outer to the inner mitochondrial membrane, and modulation of TSPO has been shown to affect steroid biosynthesis.19−21

In our optimizations using a range of 0.1−1% v/v of NP-40,

DDM, CHAPS, Digitonin, and CHAPSO detergents, we found that NP-40 gave us the most consistent and robust results (Supporting Information Figure 1 shows part of the optimization process), and we used these conditions to isolate

TSPO after treatment and heat shock with actives as well as an inactive compound. Thermal profiling of TSPO in HEK293 cells with the known modulator Alpidem22 (7.9 nM Ki) yielded a sizable stabilization with an overall thermal shift of up to 10

°C (Figure 1A), and the dose dependency of this effect was confirmed at 70 °C (Supporting Information Figure 1E). The benzodiazepine Ro5−4864 tool compound,23 which selectively binds to TSPO with nanomolar affinity (1.7 nM Ki), also produced a robust stabilization of TSPO of 5 °C degrees (Figure 1B).

Differences in stabilization amplitude could be due to differences in the structures and binding modes of the compounds. For example, Alpidem and Ro5 differ in their selectivity between benzodiazepine receptors such as TSPO,20 and Ro5 is known to have decreased affinity toward benzodiazepine receptors at higher temperatures.24 To confirm that our compound effect was due to specific target modulation

Figure 1. CETSA for TSPO inhibitors in HEK293 intact cells (N > 4). (A) Chemical structure of Alpidem (left). The results of immunoblotting of

TSPO thermal aggregation curves showing the effects of cellularly active Alpidem at 50 μM compared to DMSO control sample (middle).

Illustration of the thermal aggregation curves following quantification of the Western blots (right). (B) Chemical structure of Ro5−4864 (left). The results of immunoblotting of TSPO thermal aggregation curves showing the effects of cellularly active Ro5−4864 at 50 μM compared to DMSO control sample (middle). Illustration of the thermal aggregation curves following quantification of the Western blots (right). (C) Chemical structure of TSPO inactive compound (left). The results of immunoblotting of TSPO thermal aggregation curves comparing cellularly inactive

TSPO at 50 μM with DMSO control sample (middle). Illustration of the thermal aggregation curves following quantification of the Western blots (right). Each curve value is normalized to a 62 °C data point. Probing with loading controls like anti-Vinculin, anti GAPDH, and SOD1 showed no general stabilizing effect.

ACS Chemical Biology Articles DOI: 10.1021/acschembio.9b00399

ACS Chem. Biol. 2019, 14, 1913−1920 1914 and not a general effect on the membrane, we monitored

TSPO thermal stabilization under CETSA conditions with a structurally related but inactive compound. For this purpose, we chose AZ3451, a compound from the AstraZeneca collections, shown to be active against the GPCR PAR2 enzyme but inactive versus TSPO (data not reported). As expected, TSPO stabilization was not observed upon treatment with AZ3451 (Figure 1C), which generated a thermal aggregation profile of TSPO similar to that of the DMSO treated samples. In all these cases, SOD1 was used as a loading control, along with GAPDH and Vinculin, owing to their high melting temperature (Supporting Information Figures 2 and

3).

Next, we used CETSA to probe the modulation of SERCA2a/b, a 10/11-pass transmembrane protein responsible for trafficking Ca2+ between the cytosol and the ER, thus regulating the levels of ER calcium.25 Thapsigargin and

CDN1163 (Figure 2A) are two reported SERCA2 modulators with different modes of action. Thapsigargin is an inhibitor of

SERCA2 activity with a reported Kd of 0.2 nM that has been shown to increase the cytosolic levels of calcium.26 Crystal structures have been reported for the complex of SERCA2 with this inhibitor. CDN1163 is a reported activator27 with

Figure 2. SERCA2 thermal aggregation curves in HeLa cells comparing cellular activity (N = 3). (A) Chemical structure of thapsigargin (left). The results of immunoblotting of SERCA2 thermal aggregation curves of thapsigargin at 10 μM compared to DMSO control sample (middle).

Illustration of the thermal aggregation curves following quantification of the Western blots (right). Signal was normalized to actin and 52 °C.

Difference at 53 °C showed statistical significance using an unpaired two tailed t test with p < 0.05. (B) Chemical structure of CDN1163 (left). The results of immunoblotting of SERCA2 thermal aggregation curves of thapsigargin at 10 μM compared to DMSO control sample (middle).

Illustration of the thermal aggregation curves following quantification of the Western blots (right). Signal was normalized to actin and 52 °C.

Figure 3. (A) Chemical structures of optimized PAR2 inhibitors from two different series. Reported Kd data were obtained using mutant-stabilized

PAR2.32 (B) The PAR2 crystal structure shows two distinct sites of interaction. (C) The thermal melting transitions were followed using fluorescent size exclusion chromatography (tFSEC) of mutationally stabilized PAR2 overexpressed and solubilized from HEK293 cells. The signals indicated by the arrows are integrated to quantify nonaggregated monodisperse PAR2.33 (D) Melting curves for the solubilized recombinant PAR2 receptor indicate a thermal unfolding temperature Tm of 43 °C, while stabilization to 49 and 53 °C was observed in the presence of 75 μM of

AZ8838 and AZ3451, respectively.

ACS Chemical Biology Articles DOI: 10.1021/acschembio.9b00399

ACS Chem. Biol. 2019, 14, 1913−1920 1915 relatively weak activity (low single digit μM EC50 in ER calcium modulation with saturated activity at 10 μM28) and no reported structural or binding data.

We applied CETSA as a potential way of differentiating between the behaviors of thapsigargin and CDN1163 in live cells. We detached and harvested HeLa cells in a medium without FBS, treated with the compound for 1 h, and then heated. After cooling, we added NP-40 (0.25% v/v final concentration) and lysed the cells with three freeze−thaw cycles in liquid nitrogen. In this case, we observed a reproducible thermal aggregation curve with a steep slope and a Tagg of approximately 53 °C (see Supporting Information

Figures 4 and 5 for all replicates), suggesting SERCA2 unfolds and aggregates fully. In addition, we observe a small but consistent shift for thapsigargin on the last high-temperature portion of the thermal aggregation curve, while no thermal stabilization could be observed for CDN1163 (Figure 2A,B).

We hence conclude that in such cases CETSA experiments require work in a narrow temperature interval (less than 1 °C) and with a high throughput detection methodology allowing for numerous replicates to support differentiation between ligands.

Having observed different CETSA profiles for two targets with known tool compounds, we next applied CETSA to a

Figure 4. (A) PAR2 thermal aggregation curves in live 1321N1hPAR2 cells showing increased levels of solubilized receptor at elevated temperatures in the presence of either AZ8838 or AZ3451 at 50 μM compared to DMSO control. Whereas the overall shift in aggregation temperature is limited, there is a prominent and temperature independent shift in the baseline levels of receptor post-thermal transition. Probing with anti-Vinculin antibody shows that compound addition has no general effect on protein levels. The data were first normalized to Vinculin and subsequently with PAR2 levels at 100 μM. (B) Isothermal dose−response fingerprints of PAR2 stabilization by AZ8838 (at 53 °C) and AZ3451 (at

50 °C). All experiments were reproduced with at least two biological replicates, and representative Western Blots are shown.

ACS Chemical Biology Articles DOI: 10.1021/acschembio.9b00399

ACS Chem. Biol. 2019, 14, 1913−1920 1916 GPCR. GPCRs form the largest human membrane protein family and have been of long-standing interest as pharmaco- logical targets, mostly due to their substantial involvement in human pathophysiology combined with the exposure of druggable sites at the cell surface.29 The GPCR in question was PAR2, a seven-pass transmembrane receptor that plays a critical role in inflammation and metabolism.30,31 Inhibition of

PAR2 activation has been explored for treatment of pain in osteoarthritis. PAR2 is cleaved by proteases at the N terminus, generating a new N terminus that can then act as a tethered ligand which autoactivates the receptor.32 AZ8838 and

AZ3451 have been identified as antagonists of PAR2 (Figure

3A). X-ray crystallography showed that these two compounds bind to different sites of the PAR2 receptor (Figure 3B), and

PAR2 binding was further confirmed by SPR and by a radioligand binding assay.32 Additionally, a thermal shift assay based on detergent-solubilized PAR2 showed a thermal unfolding temperature of 43 °C in the absence of inhibitors, while the presence of 75 μM of AZ8838 or AZ3451 increased the thermal unfolding temperature to 49 and 53 °C, respectively (Figure 3C,D).

These data confirm that solubilized PAR2 can be stabilized by ligands toward thermal denaturation and that this can be followed by assessment of levels of monodisperse PAR2 measured by fluorescent size exclusion chromatography.

We next investigated target engagement of PAR2 in a live cell setting for both AZ3451 and AZ8838 using CETSA. In addition to being a multipass transmembrane protein, PAR2 presents an additional challenge when it comes to its detection because the cellular form is abundantly glycosylated. This results in multiple, fuzzy bands on Western blot analysis, thus complicating accurate signal quantification. To minimize these challenges, we first established a deglycosylation step, applied after the transient heating, to resolve the Western blot signal into one major band, a process that greatly facilitated probing of target engagement (Supporting Information Figures 6, 7, and 8). To maximize PAR2 detection, we used 1321N1 cells engineered to transiently overexpress PAR2, and cells were incubated for 90 min with either DMSO, AZ3451 (50 μM), or

AZ8838 (50 μM). Following harvest, the cells were subjected to a 3 min heat shock followed by rapid cooling to RT. The cells were then lysed using optimized conditions at 0.2% v/v

NP-40 followed by three freeze−thaw cycles; similar to TSPO and SERCA2, we observed a steep thermal transition,9 in this case with an apparent Tagg of 50 °C for the major thermal transition (Figure 4A). In contrast, the protein levels of vinculin, which is monitored as a loading control, were unaffected. Interestingly, treatment with AZ3451 and AZ8838 resulted in increased levels of solubilized PAR2 also at temperatures well above 50 °C, and these remained constant up to the highest tested temperature. Residual levels of extractable receptor were observed also in the absence of the ligand.

Dose dependency of the increased levels of PAR2 was confirmed by the isothermal dose−response fingerprints of

AZ3451 and AZ8858 at 53 and 50 °C, respectively, in the sub- micromolar to high micromolar range (Figure 4B). As shown in Figure 4B, all data were normalized based on compound responses at a 100 μM concentration. We conclude that, whereas the PAR2 residual levels are clearly impacted by ligands in a concentration-dependent fashion, there is only a limited impact on the apparent Tagg.

The observation of higher extraction yields of PAR2 throughout all elevated temperatures, whether treated with ligand or not, prompted further investigations as to the cause of this behavior. These studies were also motivated by recently reported similar observations for solute carrier proteins in the literature; i.e., residual proteins levels are detected and stabilized also at temperatures well after the thermal transition.12 The persistent stabilization of PAR2 at high temperatures after the thermal transition prompted the idea that the ligands may exert their effect on the PAR2 receptor even after the first thermal transition (Figure 5).

In order to probe this hypothesis, we modified the CETSA experimental protocol. Instead of first treating live cells with compounds and then subjecting them to heat shock, we changed the order to first subject the cells to the heat shock and did not add AZ3451 until the final 30 s of the heat shock.

In practice, this means the compound is present during cooling and subsequent sample workup, but not before heating occurs, thus showing an effect of the ligand on the receptor during cooling and sample workup. In doing so we observed the same elevation of baseline levels of extractable stabilized PAR2 at high temperatures as in the previous experiment (Figure 4), confirming an impact on soluble PAR2 levels also when added late in the transient heat pulse in the presence of unfolded protein. We believe this observation to be unique to each system on a case-by-case basis since when we performed these steps to other systems such as the above-reported SERCA2, no such changes to the baseline levels were observed (data not shown).

■DISCUSSION CETSA has been well established as a method for targeted as well as unbiased identification of compound target engagement for “in-lysate” settings and/or downstream target modulations for “in-cell” settings.4,5,7,9,11,34 Herein, we reported three case studies of applying live cell CETSA to probe target engagement/modulation on multipass transmembrane recep- tors. In each case, CETSA gave a readout that linked the cellular target modulation of submicromolar compounds with a difference in thermal melting behavior, while highlighting the distinct characteristic behaviors for each target.

TSPO gave a large thermal stabilization, with a thermal shift of at least 5 °C and upward to 10° in the case of Alpiderm. In contrast, we observed a modest but consistent thermal shift for

SERCA2 in the presence of thapsigargin. We were not able to detect target engagement of the agonist CDN1163, and this is likely due to the weak interaction of this compound with

SERCA evidenced by the modest functional activity (EC50 ∼

10 μM).

Figure 5. Mechanistic study of the ligand impact on detergent extracted PAR2 from heated live cells. The ligand AZ′3451 was added at a concentration of 50 μM only after an initial heat pulse of 150 s, meaning it was only present during the final 30 s of the total 180 s heat pulse, during cooling and during sample workup. All experiments were reproduced with at least two biological replicates, and representative Western Blots are shown.

ACS Chemical Biology Articles DOI: 10.1021/acschembio.9b00399

ACS Chem. Biol. 2019, 14, 1913−1920 1917 PAR2 presented the most unique challenges in that the protein itself was heavily glycosylated, which resulted in multiple faint bands on the Western blot and the inability to accurately quantify thermal stabilization. A deglycosylation treatment successfully consolidated the signal into one major band allowing the observation of thermal stabilization upon treatment with compounds AZ′8838 and AZ′3451. PAR2 did not demonstrate a clear thermal shift upon treatment with modulators but resulted in a ligand-stabilizing effect that could be observed through a change in protein baseline levels that persisted even at the highest temperatures (above 70 °C). We thus altered the experimental procedures for CETSA by removing the initial compound incubation step in cells and treating the cells with the compound only after 150 s of the heat shock but 30 s before the rapid cooling. We expected the

150 s of heat shock to thermally denature most PAR2, and thus if any effect of compound was seen, it would be exerted on the remaining native protein during the cooling and sample workup or alternatively aiding the refolding of denatured protein. Utilizing this procedure, we observed the same stabilization at the highest temperatures as before, which is consistent with the compound preventing complete aggrega- tion and precipitation from taking place. Given the observation of a positive thermal stabilization on solubilized PAR2 (Figure

4A), the simplest models involve binding to and stabilization of the remaining native PAR2 receptor in a live cell setting during cooling and sample workup. During the heating process, there exists a dynamic equilibrium between the folded form of PAR2 and the denatured form. PAR2 modulators preferentially bind to the folded form and, in turn, shift the equilibrium toward the folded form, thus resulting in the observed stabilization even at high temperatures.35

The in-cell CETSA protocol described here has also been successfully applied in our laboratories to additional ion channels and transporters as the solute carrier proteins family.

We have robustly observed both stabilization and destabiliza- tion events when performing in-cell CETSA on other membrane proteins, demonstrating the broad applicability of the technology. Our analysis around these observations will be disclosed in forthcoming publications. In this article, we present three representative cases of unique thermal stabilization behaviors observed for multipass transmembrane proteins after cellular treatment with small molecule modulators. These cases show that each protein will behave differently, and in contrast to soluble proteins where one protocol generally fits all, each membrane protein will require efforts for optimizing the best detection including detergents’ nature and quantities and the temperature range. We also predict that a CETSA experiment coupled to mass spectrometric read out, even with a detergent extraction step after the heating step, will limit the robust detection of several membrane proteins due to the necessary case-by-case optimization. Our observations expanded the otherwise limited cases of applying CETSA to membrane proteins and will lead to more informed application of this technique for observing target engagement or target modulation, particularly when there is a lack of alternative appropriate biochemical or biophysical methods.

■MATERIALS AND METHODS Chemicals. Alpiderm, Ro5-4864, Thapsigargin, CDN1163,

AZ3451, AZ8838 were obtained from Sigma and AstraZeneca collection. These chemicals were dissolved in DMSO.

Generation of 1321N1-PAR2 Transfected Cell Line. The human wild type PAR2 (Uniport P55085) sequence was optimized for mammalian and insect cell expression. The sequence included the

N-terminal Kozac sequence (GCCACC) and a C-terminal decahistidine tag. This construct was inserted into a pcDNA3.1(+) vector, prepared to 2−5 mg mL−1 DNA concentration in dH2O and incubated at 65 °C for 20 min. Adherent 1321N1 cells (ECACC

86030402) were expanded in DMEM (ThermoFisher 31966) with

10% fetal bovine serum (FBS, ThermoFisher 10270) to be in log phase on the day of transfection. Just prior to electroporation, the cells were washed and resuspended in MaxCyte buffer to a density of 100 million cells mL−1 with 50 μg mL−1 DNA added. The solution was run through the MaxCyte STX electroporator using the built-in

1321N1 protocol and then transferred to a culture vessel at 4 million cells/cm2 and incubated at 37 °C for 15 min. The cells were then resuspended in growth medium, counted, and centrifuged. The cell pellet was then resuspended to 5−8 million cells mL−1 in freezing medium (DMEM with 20% FBS (Sigma F2442) and 8% DMSO (Sigma) before being cryopreserved using a Kryo 560 (Planar PLC,

UK) controlled rate freezer.

General Protocol for Live Cell Cellular Thermal Shift Assay.

Live Adherent Cells CETSA. Cells were seeded 1 day before the experiment with fresh medium in 15 cm cell culture plates (1 × 10 6 cells per well). On the day of the experiment, cells were exposed to compounds at the indicated concentrations for the indicated time (30−90 min). All incubations were performed at 37 °C and 5% CO2.

Controlled cells were incubated with an equal volume of a vehicle.

Following incubation, the cells were washed with PBS to remove excess drug/control, trypsinized, and taken up with growth medium.

This suspension was centrifuged at 340g and 25 °C for 5 min, washed with PBS (2 × 10 mL), and taken up to 100 million cells mL−1 with

PBS. This suspension was aliquoted into a series of PCR tubes (compound vs vehicle) after which they were subjected to a 3 min heat shock to the appropriate heat cycle (37 to 85 °C) for generating melt curves followed by rapid cooling to 25 °C. NP40 was then added to the suspensions to give a 0.2% v/v final concentration, and the suspensions were mixed, and the cells were lysed by three freeze− thaw cycles for 3 min in liquid nitrogen. The precipitated proteins and cell debris were then pelleted by centrifugation at 11 800g for 20 min at 4 °C. The supernatants were transferred to gel loading buffers (LDS), and protein amounts were analyzed by SDS-PAGE followed by Western Blot analysis.

Live Suspension Cells CETSA. Cells were seeded 1 day before the experiment with fresh medium. On the day of the experiment, cells were harvested, washed with PBS, and taken up to 120 million cells mL−1 in growth media. This cell suspension was then transferred to separate tubes and incubated with the desired concentration of compound and vehicle. After incubation, cells were aliquoted into a series of PCR tubes’ compounds and vehicles (50 μL in each PCR tube), which were then subjected to a 3 min heat shock to the appropriate heat cycle (37 to 85 °C) for generating melt curves followed by rapid cooling to 25 °C. NP40 was then added to the suspensions to give a 0.2% v/v final concentration, and the suspensions were mixed, and the cells were lysed by three freeze− thaw cycles for 3 min in liquid nitrogen. The precipitated proteins and cell debris were then pelleted by centrifugation at 11 800g for 20 min at 4 °C. The supernatants were transferred to gel loading buffers (LDS), and protein amounts were analyzed by SDS-PAGE followed by Western Blot analysis.

Immunoblotting. CETSA samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. After the com- pletion of SDS-PAGE, gels were transferred to nitrocellulose membrane using the iBlot apparatus (20 V for 3 min, 23 V for 2 min, 25 V for 2 min). Membranes were then blocked for 2 h at RT with 5% milk, followed by overnight incubation with primary antibody at 4 °C. The primary antibody was removed, and the membrane was washed three times with Tris buffered saline with 0.2% v/v Tween (TBST). The membrane was then incubated with secondary antibody for 1 h at RT and washed three times with TBST. The membrane was exposed with Western Blot Substrate for 3 min, and the luminescence

ACS Chemical Biology Articles DOI: 10.1021/acschembio.9b00399

ACS Chem. Biol. 2019, 14, 1913−1920 1918 signal was read. The primary antibodies and secondary antirabbit

HRP-conjugate were used according to the manufacturer’s recom- mendations.

Data Analysis. The Western blot intensities were obtained by quantifying the chemiluminescence count per square millimeter (I = counts per mm2) using ImageJ-win 64 (Fiji) and normalized to loading control. Signals were also normalized to the lowest temperature to accurately assess the behavior. The normalized data was plotted using Excel or GraphPad Prism. Concentration response curves were fitted using a four-parameter nonlinear regression curve fitting in GraphPad Prism.

■ASSOCIATED CONTENT * S Supporting Information The Supporting Information is available free of charge on the

ACS Publications website at DOI: 10.1021/acschem- bio.9b00399.

Detailed Materials and Methods; Figures S1−S8, and

Tables S1 and S2 (PDF) ■AUTHOR INFORMATION Corresponding Authors

*E-mail: aarti.kawatkar@astrazeneca.com.

*E-mail: andrew.zhang@astrazeneca.com.

*E-mail: paola.castaldi@astrazeneca.com.

ORCID Dean G. Brown: 0000-0002-7130-3928 Thomas Lundbäck: 0000-0002-8145-7808

Andrew X. Zhang: 0000-0003-0406-2404 M. Paola Castaldi: 0000-0003-1959-0007

Author Contributions A.K., M.S., and A.X.Z. designed and conducted CETSA experiments and analyzed the data. M.P.C., N.D., and T.L. interpreted the data and provided strategic directions. A.K.,

A.Z., T.L., and M.P.C coordinated manuscript preparation. N.- O.H. provided cells. A.S. and N.D. carried out tFSEC experiments and analyzed the data. D.G.B. provided inputs on PAR2/TSPO compound selection. All authors discussed results and provided inputs on the manuscript.

Notes The authors declare no competing financial interest.

■ACKNOWLEDGMENTS We are grateful to J.A. Hendricks from AstraZeneca and Pelago

Bioscience and particularly D. Martinez Molina for helpful discussions on classic CETSA method development.

■ABBREVIATION CETSA, cellular thermal shift assay; ITDRF, isothermal drug- response fingerprint; Tm, apparent melting temperature; TSA, thermal shift assay; Tagg, aggregation temperature; GPCR, G- protein coupled receptor; DDM, n-dodecyl-B-D-maltoside;

NP-40, Nonidet P-40; TSPO, translocator protein; PAR2, protease-activated receptor 2; SERCA2, sarco(endo)plasmic reticulum calcium ATPase

■REFERENCES (1) Hughes, J. P., Rees, S., Kalindjian, S. B., and Philpott, K. L. (2011) Principles of early drug discovery. Br. J. Pharmacol. 162,

1239−1249. (2) Lomenick, B., Hao, R., Jonai, N., Chin, R. M., Aghajan, M.,

Warburton, S., Wang, J., Wu, R. P., Gomez, F., Loo, J. A.,

Wohlschlegel, J. A., Vondriska, T. M., Pelletier, J., Herschman, H.

R., Clardy, J., Clarke, C. F., and Huang, J. (2009) Target identification using drug affinity responsive target stability (DARTS). Proc. Natl.

Acad. Sci. U. S. A. 106, 21984−21989. (3) Senisterra, G., Chau, I., and Vedadi, M. (2012) Thermal denaturation assays in chemical biology. Assay Drug Dev. Technol. 10,

128−136. (4) Molina, D. M., Jafari, R., Ignatushchenko, M., Seki, T., Larsson,

E. A., Dan, C., Sreekumar, L., Cao, Y., and Nordlund, P. (2013)

Monitoring drug target engagement in cells and tissues using the cellular thermal shift assay. Science 341, 84−87. (5) Jafari, R., Almqvist, H., Axelsson, H., Ignatushchenko, M.,

Lundback, T., Nordlund, P., and Molina, D. M. (2014) The cellular thermal shift assay for evaluating drug target interactions in cells. Nat.

Protoc. 9, 2100−2122. (6) Martinez Molina, D., and Nordlund, P. (2016) The Cellular

Thermal Shift Assay: A Novel Biophysical Assay for In Situ Drug

Target Engagement and Mechanistic Biomarker Studies. Annu. Rev.

Pharmacol. Toxicol. 56, 141−161. (7) Savitski, M. M., Reinhard, F. B. M., Franken, H., Werner, T.,

Savitski, M. F., Eberhard, D., Molina, D. M., Jafari, R., Dovega, R. B.,

Klaeger, S., Kuster, B., Nordlund, P., Bantscheff, M., and Drewes, G. (2014) Tracking cancer drugs in living cells by thermal profiling of the proteome. Science 346, 1255784. (8) Asial, I., Cheng, Y. X., Engman, H., Dollhopf, M., Wu, B.,

Nordlund, P., and Cornvik, T. (2013) Engineering protein thermo- stability using a generic activity-independent biophysical screen inside the cell. Nat. Commun. 4, 2901. (9) Reinhard, F. B., Eberhard, D., Werner, T., Franken, H., Childs,

D., Doce, C., Savitski, M. F., Huber, W., Bantscheff, M., Savitski, M.

M., and Drewes, G. (2015) Thermal proteome profiling monitors ligand interactions with cellular membrane proteins. Nat. Methods 12,

1129−1131. (10) Ashok, Y., Nanekar, R., and Jaakola, V. P. (2015) Defining thermostability of membrane proteins by western blotting. Protein

Eng., Des. Sel. 28, 539−542. (11) Huber, K. V., Olek, K. M., Muller, A. C., Tan, C. S., Bennett, K.

L., Colinge, J., and Superti-Furga, G. (2015) Proteome-wide drug and metabolite interaction mapping by thermal-stability profiling. Nat.

Methods 12, 1055−1057. (12) Hashimoto, M., Girardi, E., Eichner, R., and Superti-Furga, G. (2018) Detection of chemical engagement of solute carrier proteins by cellular thermal shift assay. ACS Chem. Biol. 13, 1480−1486. (13) Yin, H., and Flynn, A. D. (2016) Drugging Membrane Protein

Interactions. Annu. Rev. Biomed. Eng. 18, 51−76. (14) Overington, J. P., Al-Lazikani, B., and Hopkins, A. L. (2006)

How many drug targets are there? Nat. Rev. Drug Discovery 5, 993−

996. (15) Alexandrov, A. I., Mileni, M., Chien, E. Y., Hanson, M. A., and

Stevens, R. C. (2008) Microscale fluorescent thermal stability assay for membrane proteins. Structure 16, 351−359. (16) Liu, W., Hanson, M. A., Stevens, R. C., and Cherezov, V. (2010) LCP-Tm: an assay to measure and understand stability of membrane proteins in a membrane environment. Biophys. J. 98,

1539−1548. (17) Ashok, Y., and Jaakola, V. P. (2016) Nanodisc-Tm: Rapid functional assessment of nanodisc reconstituted membrane proteins by CPM assay. MethodsX 3, 212−218. (18) Rupprecht, R., Papadopoulos, V., Rammes, G., Baghai, T. C.,

Fan, J., Akula, N., Groyer, G., Adams, D., and Schumacher, M. (2010)

Translocator protein (18 kDa) (TSPO) as a therapeutic target for neurological and psychiatric disorders. Nat. Rev. Drug Discovery 9,

971−988. (19) Midzak, A., Zirkin, B., and Papadopoulos, V. (2015)

Translocator protein: pharmacology and steroidogenesis. Biochem.

Soc. Trans. 43, 572−578. (20) Midzak, A., Denora, N., Laquintana, V., Cutrignelli, A.,

Lopedota, A., Franco, M., Altomare, C. D., and Papadopoulos, V. (2015) 2-Phenylimidazo[1,2-a]pyridine-containing ligands of the 18- ACS Chemical Biology

Articles DOI: 10.1021/acschembio.9b00399 ACS Chem. Biol. 2019, 14, 1913−1920

1919 kDa translocator protein (TSPO) behave as agonists and antagonists of steroidogenesis in a mouse leydig tumor cell line. Eur. J. Pharm. Sci.

76, 231−237. (21) Papadopoulos, V., Aghazadeh, Y., Fan, J., Campioli, E., Zirkin,

B., and Midzak, A. (2015) Translocator protein-mediated pharmacol- ogy of cholesterol transport and steroidogenesis. Mol. Cell. Endocrinol.

408, 90−98. (22) Trapani, G., Franco, M., Ricciardi, L., Latrofa, A., Genchi, G.,

Sanna, E., Tuveri, F., Cagetti, E., Biggio, G., and Liso, G. (1997)

Synthesis and binding affinity of 2-phenylimidazo[1,2-alpha]pyridine derivatives for both central and peripheral benzodiazepine receptors.

A new series of high-affinity and selective ligands for the peripheral type. J. Med. Chem. 40, 3109−3118. (23) Marangos, P. L., Pate, J., Boulenger, J. P., and Clark-Rosenberg,

R. (1982) Characterization of peripheral-type benzodiazepine binding sites in brain using [3H]Ro 5−4864. Mol. Pharmacol. 22, 26−32. (24) Benavides, J., Quarteronet, D., Imbault, F., Malgouris, C., Uzan,

A., Renault, C., Dubroeucq, M. C., Gueremy, C., and Le Fur, G. (1983) Labelling of ″peripheral-type″ benzodiazepine binding sites in the rat brain by using [3H]PK 11195, an isoquinoline carboxamide derivative: kinetic studies and autoradiographic localization. J.

Neurochem. 41, 1744−1750. (25) Vangheluwe, P., Raeymaekers, L., Dode, L., and Wuytack, F. (2005) Modulating sarco(endo)plasmic reticulum Ca2+ ATPase 2 (SERCA2) activity: cell biological implications. Cell Calcium 38, 291−

302. (26) Lytton, J., Westlin, M., and Hanley, M. R. (1991) Thapsigargin inhibits the sarcoplasmic or endoplasmic reticulum Ca-ATPase family of calcium pumps. J. Biol. Chem. 266, 17067−17071. (27) Dahl, R. (2017) A new target for Parkinson’s disease: Small molecule SERCA activator CDN1163 ameliorates dyskinesia in 6- OHDA-lesioned rats. Bioorg. Med. Chem. 25, 53−57. (28) Kang, S., Dahl, R., Hsieh, W., Shin, A., Zsebo, K. M., Buettner,

C., Hajjar, R. J., and Lebeche, D. (2016) Small Molecular Allosteric

Activator of the Sarco/Endoplasmic Reticulum Ca2+-ATPase (SERCA) Attenuates Diabetes and Metabolic Disorders. J. Biol.

Chem. 291, 5185−5198. (29) Hauser, A. S., Attwood, M. M., Rask-Andersen, M., Schioth, H.

B., and Gloriam, D. E. (2017) Trends in GPCR drug discovery: new agents, targets and indications. Nat. Rev. Drug Discovery 16, 829−842. (30) Adams, M. N., Ramachandran, R., Yau, M. K., Suen, J. Y.,

Fairlie, D. P., Hollenberg, M. D., and Hooper, J. D. (2011) Structure, function and pathophysiology of protease activated receptors.

Pharmacol. Ther. 130, 248−282. (31) Ramachandran, R., Noorbakhsh, F., Defea, K., and Hollenberg,

M. D. (2012) Targeting proteinase-activated receptors: therapeutic potential and challenges. Nat. Rev. Drug Discovery 11, 69−86. (32) Cheng, R. K. Y., Fiez-Vandal, C., Schlenker, O., Edman, K.,

Aggeler, B., Brown, D. G., Brown, G. A., Cooke, R. M., Dumelin, C.

E., Dore, A. S., Geschwindner, S., Grebner, C., Hermansson, N. O.,

Jazayeri, A., Johansson, P., Leong, L., Prihandoko, R., Rappas, M.,

Soutter, H., Snijder, A., Sundstrom, L., Tehan, B., Thornton, P.,

Troast, D., Wiggin, G., Zhukov, A., Marshall, F. H., and Dekker, N. (2017) Structural insight into allosteric modulation of protease- activated receptor 2. Nature 545, 112−115. (33) Hattori, M., Hibbs, R. E., and Gouaux, E. (2012) A fluorescence-detection size-exclusion chromatography-based thermo- stability assay for membrane protein precrystallization screening.

Structure 20, 1293−1299. (34) Franken, H., Mathieson, T., Childs, D., Sweetman, G. M.,

Werner, T., Togel, I., Doce, C., Gade, S., Bantscheff, M., Drewes, G.,

Reinhard, F. B., Huber, W., and Savitski, M. M. (2015) Thermal proteome profiling for unbiased identification of direct and indirect drug targets using multiplexed quantitative mass spectrometry. Nat.

Protoc. 10, 1567−1593. (35) Matulis, D., Kranz, J. K., Salemme, F. R., and Todd, M. T. (2005) Thermodynamic Stability of Carbonic Anhydrase: Measure- ments of Binding Affinity and Stoichiometry Using ThermoFluor.

Biochemistry 44, 5258−5266.

ACS Chemical Biology Articles DOI: 10.1021/acschembio.9b00399

ACS Chem. Biol. 2019, 14, 1913−1920 1920