Covalent Fragment Inhibits RhoA Activation by Guanine Exchange Factor

Abstract

Ras homolog gene family member (RhoA) is a GTPase and a member of the RAS superfamily of GTPases. RhoA is a master regulator of the actin cytoskeleton. It inhibits axon growth preventing repair and recovery following spinal cord and traumatic brain injuries. Despite decades of research into the biological function of Rho GTPases, there exists no small-molecule Rho inhibitors. Here, we screen a library of cysteine electrophiles to explore whether covalent bond formation at Cys-107 leads to inhibition of RhoA activation by guanine exchange factor (GEF) Trio. Two fragments, propiolamide 1 (ACR-895) and acrylamide 2 (ACR-917), inhibited RhoA nucleotide exchange by Trio in a time-dependent manner. The fragments formed a covalent bond with wild-type RhoA but not Cys107Ser RhoA mutant. Time- and concentration-dependent studies led to equilibrium constants K_Is and reaction rates that correspond to t_1/2 values in the single-digit hour range. One fragment was selective for RhoA over Rac1 GTPase and had no effect on K-RAS nucleotide exchange by SOS1. The fragments did not inhibit RhoA binding to ROCK effector protein. This work establishes Cys-107 as a suitable site for Rho GTPase inhibition and provides fragment starting points for the future development of Rho GTPase covalent inhibitors that could have profound implications for the treatment of patients with injuries of the central nervous system.

INTRODUCTION

Rho GTPases are master regulators of actin cytoskeleton with prominent roles in cell polarity, microtubule dynamics, trafficking, and transcription factor activity (1,2). Rho proteins activate a large number of signal transduction pathways that have been associated with impairing repair and recovery following spinal cord and traumatic brain injuries (3–5), promoting neurodegenerative diseases (6), contributing to immune system related diseases (7), and enhancing tumor growth and metastasis (8–10). Among the 20 members of the Rho GTPase family, RAS homolog gene family member (RhoA), Ras-related C3 botulinum toxin substrate 1 (Rac1), and cell division cycle 42 (Cdc42) are the most well studied. RhoA generates actin filaments known as stress fibers (11); Rac1 is responsible for the formation of actin filaments known as lamellipodium; Cdc42 controls the formation of filaments known as filopodium. Like all small GTPases, Rho GTPases cycle between inactive GDP-bound and active GTP-bound states (12). In their active state, Rho GTPases bind to effector proteins such as ROCK to activate a variety of signal transduction pathways. Guanine exchange factors (GEFs) such as Trio, LARG, Dbs, and VAV1 activate Rho GTPases by facilitating nucleotide exchange between GDP- to GTP-bound states (13). Guanine activating proteins (GAPs) inhibit Rho GTPases by catalyzing the hydrolysis of the phosphodiester bond of GTP (14).

Among Rho GTPases, RhoA is the most extensively studied, particularly in the context of central nervous system (CNS) injuries. Following spinal cord and traumatic brain injuries, scarring often results due to the release of factors that inhibit growth such as chondroitin sulfate proteoglycans (CSPGs) and myelin (15–18). RhoA is the central node through which growth inhibitory factors converge (19,20). To that end, substantial efforts have been devoted towards inhibiting RhoA to promote functional recovery following these injuries. Since there are no small molecules that directly bind to RhoA and inhibit its activity, most efforts have concentrated on the use of ROCK kinase small-molecule inhibitors. Unfortunately, RhoA has multiple effector proteins in addition to ROCK, so these compounds only partially inhibit RhoA signaling. Another approach has been the use of biologics, such as the Rho GTPase inhibitor VX-210, which was tested in clinical trials in spinal cord injury patients, but the trials ended early due to lack of efficacy (21).

Despite several decades of research on Rho GTPases there exists no small-molecule Rho inhibitors. There is intense interest in the development of small-molecule Rho GTPase inhibitors. Like RAS GTPases, Rho GTPases are considered undruggable due to the absence of a large druggable pocket that is accessible by small molecules. Although the nucleotide binding pocket is considered druggable, but the high affinity of the nucleotide to the GTPase, along with the millimolar concentration of nucleotides in cells, rule out the possibility of using this pocket for drug development (22). A recent breakthrough in the development of a covalent inhibitor of a relatively rare G12C mutation of the K-RAS GTPase oncogene that was approved by the US food and drug administration (FDA) confirms that covalent inhibition can overcome the lack of druggable pocket (23–25). Another example of covalent drugs that bind to a pocket that is not considered druggable are penicillin-based antibiotics. These drugs have low affinity to their target, but their irreversible covalent bond makes them highly efficacious drugs. Covalent inhibitors are driven by a binding and a reaction event, and the measure of potency of these compounds is provided by a second order rate constant that consists of a ratio between compound reaction rate and binding constant. Interestingly, RhoA, unlike wild-type K-RAS, has several cysteine amino acids on its surface.

Here, we explore whether RhoA Cys-107 is accessible for covalent bond formation, and whether the amino acid could be a site for the development of a covalent inhibitor of RhoA. To explore this question, we screened a library of cysteine electrophile fragments for candidates that form a covalent bond at Cys-107 and inhibit RhoA activation by Trio GEF. Two fragments were identified that formed a covalent bond with RhoA. The fragments were tested for inhibition of RhoA, Rac1, and K-RAS GEF activation, as well as RhoA binding to its effector protein ROCK. These studies were carried out using wild-type and mutant RhoA and Rac1 proteins. Time- and concentration-dependent studies using both whole protein mass spectrometry and nucleotide exchange was used to gain further insight into the reaction and inhibition kinetics of the fragments.

RESULTS

Three-Dimensional Structures of RhoA Reveal Surface Cysteine Residues.

RhoA harbors several cysteine amino acids on its surface, such as Cys-20, Cys-107, and Cys-190. Cys-20 is located within hydrogen-bonding distance of the nucleotide phosphate oxygen atoms. Although Cys-20 was shown to have a slightly depressed pKa in Rac1, and the amino acid is prone to oxidation (26), there are no available binding pockets near this amino acid that can accommodate a small molecule. The amino acid is surrounded by amino acids of the switch I loop, making it even more difficult for electrophiles to access the site. Cys-190 is the prenylation site of RhoA. It is located at the end of a highly flexible polybasic tail that protrudes out of the globular domain of the RhoA protein. Although we expect this amino acid to be reactive and accessible for covalent bond formation, the lack of a binding pocket near the amino acid makes it likely unsuitable for small-molecule covalent bond formation and drug development. Cys-107 on the other hand is located on the globular domain of RhoA in a pocket sandwiched between the C-terminal polybasic tail of RhoA, and the Switch II loop of the GTPase (Fig. 1A). Cys-107 is surrounded by hydrophobic amino acids creating a binding pocket that is likely suitable for binding of small-molecule organic compounds (Fig. 1B). A crystal structure of RhoA in complex with Dbs GEF shows that a loop from the GEF, known as the β3/β4 loop, extends to the Cys-107 pocket. This loop was shown to be critical for its ability to carry out nucleotide exchange (Fig. 1A).

Figure 1. Screening of Fragment Library Identifies Covalent Inhibitors.

(A) Stereo view of the crystal structure of RhoA in complex with Dbs GEF (PDB code: 1LB1). RhoA is shown in grey Connolly surface representation; amino acids of the Switch II pocket are shown in marine blue, Cys-107 is shown in yellow, and amino acids of the Switch I loop in red. Dbs is depicted in ribbon representation with the PH domain shown in green, and the DH domain in magenta. (B) Strereoview of the binding site surrounding Cys-107. RhoA is depicted in grey ribbon representation, except the Switch II loop is shown in blue. Sidechains of the Switch II loop are shown in blue capped-sticks, and other RhoA sidechains are shown in grey capped-sticks, except for Cys-107, which is depicted in color-coded capped- sticks (yellow and gold for carbon and cysteine, respectively). The β3/β4 loop of Dbs is shown in magenta. (C) GDP-bound RhoA^WT is incubated with fragments from an in-house library of 783 cysteine electrophiles at 4 °C for 24 h followed by addition of Trio GEF and BODIPY-FL-GDP. Percent inhibition is determined relative to DMSO. (D) Fragments that inhibited exchange by more than 20% were tested for inhibition of GDP-bound RhoA^WT nucleotide exchange after 1 and 24 h at 4 °C followed by addition of Trio and BODIPY-FL-GDP; mean ± s.d., n = 2 replicates. (E) Fragments (200 μM) showing time-dependent inhibition of GDP-bound RhoA^WT nucleotide exchange were tested for inhibition of nucleotide exchange of GDP-bound RhoA mutants, namely RhoA^C107S,C190S, with 2 μM RhoA protein for 24 h at 4°C prior to adding Trio (100 nM) and BODIPY-FL-GDP (250 nM); mean ± s.d., n = 2 replicates.

Fragment Screening Identifies Covalent Inhibitors.

To investigate if Cys-107 is accessible for covalent bond formation, an in-house electrophile library of 783 acrylamide fragments was screened for candidates that inhibit RhoA^WT activation by the Rho GEF Trio (Fig. 1C). RhoA^WT (2 μM) was incubated with each of the fragments at 4 °C for 24 h. Trio was added to each well along with BODIPY-FL-GDP to monitor nucleotide exchange by total fluorescence. Fragments that inhibited by at least 20% were selected for follow-up time-dependent studies at 1 and 24 h to identify covalent fragment inhibitors (Fig. 1D). This was done by incubation of RhoA^WT (2 μM) with each fragment (200 μM) for 1 or 24 h, at 4 °C. Among the 49 fragments tested, 20 showed increased inhibition from 1 to 24 h, suggesting that they are covalent inhibitors. To identify whether fragments reacted at Cys-107, RhoA^WT (2 μM) and RhoA^C107S,C190S (2 μM) were incubated with these fragments (200 μM) for 24 h and 4 °C, followed by nucleotide exchange by Trio. Two fragments had reduced inhibition of RhoA^C107S,C190S mutant exchange compared to RhoA^WT (Fig. 1E), suggesting that these fragments likely form a covalent bond at Cys-107.

Fragments Inhibit RhoA Activation through Covalent Bond Formation at Cys-107.

To further confirm that these two fragments inhibit due to covalent bond formation at Cys-107, a concentration-dependent nucleotide exchange study was carried out. RhoA^C190S (2 μM) or RhoA^C107S,C190S (2 μM) were incubated with each fragment at increasing concentration for 24 h and 4°C followed by addition of Trio and BODIPY-FL-GDP. Only two fragments with a similar core structure, namely propiolamide 1 (ACR-895) and acrylamide 2 (ACR-917) [Fig. 2A], revealed concentration-dependent inhibition of RhoA nucleotide exchange by Trio. Raw exchange curves for the nucleotide exchange process are shown for 1 (ACR-895) and 2 (ACR-917) in Fig. 3A and 3B, respectively. The fragments inhibited RhoA^C190S nucleotide exchange in a concentration-dependent manner with IC₅₀s of 20 and 46 μM for 1 (ACR-895) and 2 (ACR-917), respectively (Fig. 2B and Fig. 3C, D). To confirm that the inhibition of nucleotide exchange is due to covalent bond formation at Cys-107, the nucleotide exchange study was repeated using RhoA^C107S,C190S (Fig. 3C and D). Raw exchange curves for this mutant are provided in Fig. S1. Both fragments did not inhibit RhoA^C107S,C190S nucleotide exchange by Trio suggesting that covalent bond formation at Cys-107 is essential for inhibition. Additional time-dependent studies were carried out to further establish that the fragments are covalent inhibitors for 1 (ACR-895) [Fig. 3E] and 2 (ACR-917) [Fig. 3F]. Both 1 (ACR-895) and 2 (ACR-917) did not inhibit exchange following 1 h incubation with RhoA^C190S, yet robust inhibition was observed at 24 h (Fig. 2B). These time-dependent data further confirm that the fragments are covalent inhibitors.

Figure 2. Chemical Structures and Reaction Kinetics Parameters.

(A) Chemical structure of fragments. (B) IC₅₀ values obtained from concentration-dependent nucleotide exchange studies for fragments 1-2.

Figure 3.

(A) Time-dependent nucleotide exchange for RhoA (2 μM) incubated with increasing concentration of fragment 1 (ACR-895) following 24 h pre-incubation at 4 °C prior to adding Trio (100 nM) and BODIPY-FL-GDP (250 nM). (B) Time-dependent nucleotide exchange for RhoA (2 μM) incubated with increasing concentration of fragment 2 (ACR-917) following 24 h pre-incubation at 4 °C prior to adding Trio (100 nM) and BODIPY-FL-GDP (250 nM). (C) Inhibition of RhoA^C190S and RhoA ^{RhoAC107S,C190S} nucleotide exchange with increasing concentration of 1 (ACR-895) determined from the time-dependent exchange curves; mean ± s.d., n = 2 replicates. (D) Inhibition of RhoA^C190S and RhoA ^{RhoAC107S,C190S} nucleotide exchange with increasing concentration of 2 (ACR-917) determined from the time-dependent exchange curves; mean ± s.d., n = 2 replicates. (E) Time-dependent inhibition of RhoA^C190S (2 μM) nucleotide exchange by 1 (ACR-895) after incubation of the GTPase with the fragment for 1 h and 24 h at 4 °C followed by Trio (100 nM) and BODIPY-FL-GDP (250 nM); mean ± s.d., n = 2 replicates. (F) Time-dependent inhibition of RhoA^C190S (2 μM) nucleotide exchange by 2 (ACR-917) after incubation of the GTPase with the fragment for 1 h and 24 h at 4 °C followed by Trio (100 nM) and BODIPY-FL-GDP (250 nM); mean ± s.d., n = 2 replicates. (G) Intrinsic nucleotide exchange curves for GDP-bound RhoA^C190S incubated with 200 μM 1 (ACR-895) for 24 h at 4°C followed by addition of BODIPY-FL-GDP; mean ± s.d., n = 2 replicates. (H) Intrinsic nucleotide exchange curves for GDP-bound RhoA^C190S incubated with 200 μM 2 (ACR-917) for 24 h at 4°C followed by addition of BODIPY-FL-GDP.

The nucleotide binding pocket is large and can accommodate a small molecule, but it is not accessible in cells due to the picomolar affinity of the nucleotide to RhoA and the millimolar concentration of GDP and GTP. An intrinsic nucleotide exchange study was carried out for both fragments to rule out that inhibition is due to direct competition with nucleotide binding (Fig. 3G and H). GTPases can experience nucleotide exchange, although at a substantially slower rate than in the presence of a GEF. RhoA^C190S (2 μM) was incubated with fragments 1 (ACR-895) or 2 (ACR-917) at 4 °C and 24 h followed by addition of BODIPY-FL-GDP. Interestingly, 1 (ACR-895) appears to slightly enhance the exchange rate compared with DMSO control, but the fragment did not affect nucleotide binding to RhoA. Fragment 2 (ACR-917), on the other hand, had no effect on the initial exchange rate, although the fragment appears to slightly inhibit nucleotide loading likely due to its effect on the Switch II loop.

Mass Spectrometry Confirms Cys-107 as Reaction Site and Provides Inhibition Kinetics.

To further establish that the fragments are forming a covalent bond at RhoA Cys-107, RhoA^C190S (2 μM) was incubated with each fragment (200 μM) at 4 °C for 24 h, followed by dialysis for another 24 h (Fig. 4A). This was followed by nucleotide exchange with Trio. Despite the absence of compound in solution, RhoA^C190S nucleotide exchange was substantially inhibited, confirming that the compounds are irreversible inhibitors. Further evidence of covalent inhibition came from whole protein mass spectrometry for 1 (ACR-895) and 2 (ACR-917) [Fig. 4B and C]. RhoA^C190S with DMSO showed a peak that matched the protein molecular weight. Incubation of RhoA^C190S with fragment resulted in a mass spectrum with an additional peak with a mass difference of Δ229 for 1 (ACR-895) [Fig. 4B] and Δ331 for 2 (ACR-917) [Fig. 4C], in both cases matching fragment molecular weight. These experiments were also conducted at a pH of 7.4 at both 4 °C and room temperature and adduct formation was also observed in these conditions (Fig. S2). We also tested 1 (ACR −895) for inhibition of RhoA nucleotide exchange with LARG and found it to be a weaker inhibitor of LARG exchange compared to Trio (Fig. S3). All nucleotide exchange studies in this work were carried out with His-tagged RhoA protein. To confirm that the His-tag has no effect on nucleotide exchange inhibition by the fragments, we removed the tag to compare Trio exchange with His-tagged RhoA (Fig. S4). To confirm that the adducts were occurring at Cys-107, RhoA^C107S,C190S mutant was incubated with 1 (ACR −895) or 2 (ACR-917) for 24 h at 4 °C (Fig. 4D and E). In each case, there was no adduct for the covalent complex in the mutant confirming that the fragments are reacting at Cys-107.

Figure 4. Time- and Concentration-Dependent Studies for Reaction Kinetics Parameters.

(A) Inhibition of fragments 1 (ACR-895) and 2 (ACR-917) after 24 h incubation at 4 °C with RhoA^C190S followed by dialysis for another 24 h at 4 °C. (B) RhoA^C190S•GDP (2 μM) were incubated with 1 (ACR-895) [200 μM) or (C) 2 (ACR-917) [200 μM] for 24 h at 4°C. The reactions were quenched with formic acid, followed by collection of MS. (D) RhoA^C107SC190S•GDP (2 μM) were incubated with 1 (ACR-895) [200 μM) or (E) 2 (ACR-917) [200 μM] for 24 h at 4°C. The reactions were quenched with formic acid, followed by collection of MS. (F) Extent of adduct formation of 1 (ACR-895) with RhoA^C190S (2 μM) is determined at increasing concentration of the fragment over time (30 mins to 6 h) at 4 °C; mean ± s.d., n = 2 replicates. (G) Extent of adduct formation of 2 (ACR-917) with RhoA^C190S (2 μM) is determined at increasing concentration of the fragment over time (30 mins to 6 h) at 4 °C; mean ± s.d., n = 2 replicates (H) k_obs was obtained from time-dependent studies and plotted against concentration of compounds to obtain binding constant and reaction rates for each fragment; mean ± s.d., n = 2 replicates. (I) Binding constant K_I and reaction rate k_inact for 1 (ACR-895) and 2 (ACR-917). (J) Time-dependent inhibition of nucleotide exchange of RhoA^C190S (2 μM) at ten concentrations fragment 1 (ACR-895) starting at 200 μM using two-fold dilution at 4 °C. (K) Time-dependent inhibition of nucleotide exchange of RhoA^C190S (2 μM) at ten concentrations of fragment 2 (ACR-917) starting at 200 μM using two-fold dilution at 4 °C. (L) k_obs rate constants obtained from fitting of time-dependent RhoA^C190S nucleotide exchange inhibition curves for 1 (ACR-895). (M) k_obs rate constants obtained from fitting of time-dependent RhoA^C190S nucleotide exchange inhibition curves for 2 (ACR-917).

Time- and concentration-dependent whole protein mass spectrometry was carried out to get insight into the reaction rates and binding constants of the fragments (Fig. 4F–I). The percent occupancy over time at different concentrations are shown in Figs. 4F and G, for 1 (ACR-895) and 2 (ACR-917), respectively. The fragment reaction was plotted against time for each concentration affording a pseudo first-order rate constant k_obs (Fig. 4H). These rate constants were plotted against concentration of compound to extract the binding constant K_I and the inactivation rate constant k_inact (Fig. 4I). In each case, the binding constant K_I was approximately 500 μM. It is important to note that K_I is not the binding constant, but the concentration required to achieve half of the reaction rate k_inact. Regardless, it is an important measure of the affinity required for a compound to react with the target at the concentrations of the fragments. The reaction half-life, t_1/2, obtained for 1 (ACR-895) and 2 (ACR-917) from the k_inact values were 0.8 and 3.5 h, respectively at 4 °C (Fig. 4I).

In addition to whole protein mass spectrometry, we used our nucleotide exchange assay to explore inhibition kinetics to compare with reaction kinetics obtained from whole protein mass spectrometry. A time- and concentration-dependent nucleotide exchange study was carried out by incubation of RhoA^C190S (2 μM) at 4 °C for varying concentrations of 1 (ACR-895) and 2 (ACR-917) [Fig. 4J–M]. From these studies an inhibition reaction rate constant k_inact and binding constant K_I were obtained for each fragment (Fig. 4I). For fragment 1 (ACR-895), a reaction half-life t_1/2 of 1.3 h was obtained along with a K_I of 223 μM. For fragment 2 (ACR-917), the reaction half-life was 3.7 h and the binding constant 177 μM. It is interesting to note that the reaction half-lives and K_I of the fragments for the exchange and mass spectrometry studies were similar, further confirming that the inhibition of the compounds is due to direct engagement of Cys-107.

Fragments Effect on K-RAS and Rac1 Exchange.

Further evidence of specific binding and reaction at Cys-107 is provided by testing the effect of the two fragments on K-RAS nucleotide exchange by Son-of-Sevenless (SOS1) guanine exchange factor (Fig. 5A and B). Unlike Rho GTPases, K-RAS does not harbor a cysteine at the equivalent position of RhoA Cys-107. Therefore, we expect that the compounds should have substantially less effect on K-RAS nucleotide exchange by SOS1. This was tested for both fragments. As expected, both 1 (ACR-895) and 2 (ACR-917) had minimal effect when K-RAS (2 μM) was incubated with the fragments following addition of SOS1 and BODIPY-FL-GDP and measurements of fluorescence.

Figure 5.

(A) Percent Inhibition from nucleotide exchange of RhoA^C190S and KRASG12D incubated with increasing concentrations of 1 (ACR-895) for 24 h at 4 °C; mean ± s.d., n = 2 replicates. (B) Percent Inhibition from nucleotide exchange of RhoA^C190S and KRASG12D incubated with increasing concentrations of 2 (ACR-917) for 24 h at 4 °C; mean ± s.d., n = 2 replicates. (C) Time-dependent inhibition from nucleotide exchange of Rac1 incubated with increasing concentrations of 1 (ACR-895) for 24 h at 4 °C; mean ± s.d., n = 2 replicates. (D) Time-dependent inhibition from nucleotide exchange of Rac1 incubated with increasing concentrations of 2 (ACR-917) for 24 h at 4 °C; mean ± s.d., n = 2 replicates. (E) Comparison of the binding region of the fragments surrounding Cys-107 in RhoA and Cys-105 in Rac1. The amino acids believed to form binding interactions with the fragments are labelled 1 to 5. (F) Development of AlphaScreen assay for RhoA binding to ROCK1. (G) Fluorescence was measured following titration of increasing concentration of GST-ROCK1 to GMPPMP-bound biotinylated RhoA (25 nM); mean ± s.d., n = 2 replicates. (H) GDP was used to confirm binding of RhoA to ROCK1 in the AlphaScreen assay. Biotinylated RhoA•GMPPNP (25 nM) is incubated with increasing concentration of GDP nucleotide followed by addition of and GST-ROCK1 (6.25 nM), streptavidin- and anti-GST-antibody-loaded beads, and reading of fluorescence; mean ± s.d., n = 2 replicates. (I) Biotinylated RhoA•GMPPNP (25 nM) is incubated with increasing concentration of 1 (ACR-895) or 2 (ACR-917) followed by addition of GST-ROCK1 (6.25 nM), streptavidin- and anti-GST-antibody-loaded beads, and the reading of fluorescence; mean ± s.d., n = 2 replicates.

Rac1 harbors a cysteine at the same position as RhoA Cys-107 (Cys-105 on Rac1). Despite the high sequence identity between RhoA and Rac1, there are some differences in the amino acids surrounding the nucleophile cysteine. Rac1 (1–177) was incubated at varying concentration of fragment 1 (ACR-895) and 24 h at 4 °C, followed by addition of Tiam1 and BODIPY-FL-GDP (Fig. 5C and D). Fragment 1 (ACR-895) showed time-dependent inhibition of Rac1 although the IC₅₀ of 37 μM at 24 h was slightly higher than that we obtained following RhoA incubation with the fragment. Fragment 2 (ACR-917) did not show much inhibition within the 100 μM range considered in the study in contrast to RhoA exchange, which was inhibited by the same fragment with an IC₅₀ of 46 μM. This is likely due to the fact that 2 (ACR-917) engages more amino acids in the pocket therefore making the fragment more susceptible to changes in amino acids between RhoA and Rac1 (Fig. 5E).

Fragments Do Not Inhibit RhoA Binding to Effector Protein.

RhoA•GTP binds to ROCK effector protein to initiate a series of downstream protein-protein interaction and phosphorylation events that culminate in transcriptional activity. Cys-107 is not located at the RhoA protein-protein interaction interface with ROCK, but we wondered if the fragments inhibited RhoA•GTP binding to ROCK through a potential allosteric mechanism. To that end, we developed an AlphaScreen assay to investigate this matter (Fig. 5F). The assay consisted of biotinylated RhoA•GMPPNP and GST-tagged ROCK1. Following incubation of biotinylated RhoA•GMPPNP with fragments for 24 h at 4°C, GST-tagged ROCK1 was added along with streptavidin- and anti-GST-antibody-loaded beads. Robust binding of RhoA•GMPPNP to ROCK was detected (Fig. 5G). The binding was successfully inhibited when RhoA•GMPPNP was incubated with increasing concentration of GDP nucleotide (Fig. 5H). Following incubation of RhoA•GMPPNP with 1 (ACR-895) or 2 (ACR-917) at varying concentration of fragment for 24 h and 4 °C, we detected no inhibition of RhoA•GMPPNP binding to ROCK1 (Fig. 5I).

DISCUSSION

RhoA plays a crucial role in actin dynamics. Following central nervous system (CNS) injuries, factors that inhibit axon growth converge through RhoA to block repair and regeneration. Small-molecule RhoA inhibitors are expected to promote recovery following injury from traumatic injuries of the brain and spinal cord. Like K-RAS, RhoA is considered undruggable and despite more than three decades since its discovery, there are no small molecules that bind and inhibit its function. Unlike K-RAS, however, Rho GTPases possess several cysteine amino acids on their surface, providing a possible path for the development of targeted covalent inhibitors. The development of K-RAS G12C mutant therapeutic agents that were FDA-approved suggests that the lack of a druggable pocket can be overcome with covalent inhibition. However, until now, it was not clear if covalent bond formation at Rho cysteines will affect Rho GTPase nucleotide exchange, or Rho GTPase binding to effector proteins.

Our analysis of RhoA crystal structures either in the apo form, or in complex with GEFs revealed that Cys-107 is in a binding site that is at least partially located at the interface between the GTPase and the β3/β4 loop of GEFs. We hypothesized that covalent bond formation at Cys-107 may result inhibitors of RhoA exchange by GEFs. To that end, we carry out a covalent fragment screen to explore if Cys-107 is accessible for covalent bond formation. We identify two fragments, 1 (ACR-895) and 2 (ACR-917), which form a covalent bond at Cys-107. Testing of nucleotide exchange at 1 and 24 h showed time-dependent inhibition, confirming that the fragments were covalent inhibitors of RhoA exchange. Mutation of Cys-107 to serine resulted in complete loss of inhibition by the fragment confirming that adduct formation at Cys-107 was essential for inhibition. Intact mass spectrometry showed robust adduct formation for RhoA^C190S but not for RhoA^C190S,C107S. This establishes that the fragments form a covalent bond at Cys-107. The fragments were tested against Rac1, which possesses a cysteine at position 105, equivalent to RhoA Cys-107. Fragment 1 (ACR-895) inhibited Rac1 exchange but 2 (ACR-917) was a much weaker Rac1 inhibitor. This is likely due to the fact that 2 (ACR-917) occupies a larger portion of the pocket and engages more residues, making it more susceptible to the differences in amino acid composition between RhoA and Rac1 in the pocket. Both fragments had no effect on K-RAS exchange, as K-RAS does not have a cysteine at the same position as RhoA Cys-107, despite the structural similarities.

Cys-107 is in a binding pocket that is partially created by the Switch II loop of RhoA (Fig. 1A and B) on one side, and the C-terminus polybasic tail of RhoA. The pocket is at least partially located at the interface between Trio GEF and RhoA. GEFs like Dbs and Trio extend into the Cys-107 pocket through their β3/β4 loops, which is known to be critical for the binding of GEFs to Rho GTPases (27). It is likely that covalent bond formation at Cys-107 results in orthosteric inhibition of RhoA binding to Trio. It is also possible that covalent bond formation at Cys-107 leads to inhibition of RhoA exchange by altering the structure of the switch II loop, which may affect the interaction of RhoA with GEFs. The intrinsic exchange studies that we conducted show that the fragments have no appreciable effect on the extent of nucleotide binding, ruling out direct competition with nucleotide as a mechanism of inhibition. Comparison of the K_I and k_inact values obtained from whole protein mass spectrometry and nucleotide exchange also provide some insight into the mechanism by which the compounds inhibit exchange. Whole protein mass spectrometry measures direct engagement of the target by the fragments, while the exchange study reflects inhibition of Trio binding to RhoA. The similarity in the K_I and k_inact values obtained from whole protein mass spectrometry and nucleotide exchange support an orthosteric inhibition mechanism.

It is interesting that the fragments show different binding and inhibition between RhoA and Rac1 suggesting that it may be possible to develop RhoA or Rac1 selective inhibitors through this strategy. A comparison between the Cys-107 pocket on RhoA and the equivalent Cys-105 pocket on Rac1 shows a similar pocket shape but with some noteworthy changes in amino acid composition around the cysteine residue (Fig. 5E). The most significant change near the cysteine is the presence of a neutral glutamine in Rac1 instead of Asp-76 of RhoA. Two hydrophobic residues at Ile-80 and Val-110 on RhoA are replaced with more polar amino acids on Rac1, Ser-78, and Thr-108, respectively. Finally, aromatic Phe-106 of RhoA is occupied by His-104 on Rac1. It is possible that these differences at the Cys-107 may be responsible for the higher level of inhibition of RhoA activation by the compounds compared with Rac1. Unlike RhoA or Rac1, K-RAS does not have a cysteine at the same position as Cys-17 of RhoA, which was reflected by a lack of inhibition of the fragments.

The discovery of Cys-107 as a site for covalent bond formation on RhoA and for inhibition of activation of the GTPase is a significant development towards developing potential therapeutic agents targeting Rho GTPases. The binding site near Cys-107 may not be considered druggable, but the ability to form a covalent bond presents an opportunity to develop a compound that could engage RhoA with sufficient potency to exhibit suitable therapeutic index.

MATERIALS and METHODS

Expression and Purification.

Plasmids encoding His-RhoA, His-RhoA^C190S, and His-RhoA^C107S,C190S were expressed in E. Coli BL21 (DE3). A single colony was inoculated in 40 mL LB media using 100 μg/mL ampicillin and grown overnight at 37 °C. The overnight culture was diluted into 4 L LB media and grown at 37 °C to an OD₆₀₀ values of 0.6–0.8. IPTG was added at a final concentration of 0.5 mM and the culture was grown at 37 °C for 4 h with continuous shaking. The cells were harvested at 4500 rpm for 15 mins. Cells were resuspended in buffer A consisting of 20 mM Tris, pH 8.0 and 300 mM NaCl. Cells were lysed using a microfluidizer and clarified by centrifugation at 35000 rpm for 1 h. The supernatant was loaded onto a Ni-NTA column (His trap FF column) and washed using Buffer A. The protein was eluted using a 0–300 mM imidazole gradient in buffer A. The fractions from the Ni-NTA column were pooled and concentrated to run into a superdex-75 pg size exclusion chromatography column for further purification using 20 mM Tris, pH 8.0 and 100 mM NaCl. The purified protein was concentrated using an Amicon 10 kDa cut-off concentrator and the concentration was measured using a Themo Fisher Scientific Nanodrop One.

Nucleotide exchange assay.

The nucleotide exchange assay was performed in a black 384 microplate, round bottom, polystyrene, with a maximum volume of 20 μL in each well. 10 μL His-RhoA (2 μM) or the mutants (2 μM) were added in each row (24 wells) in exchange buffer (20 mM Tris, pH 8.0, 100 mM NaCl, 0.01% IGEPAL, and 5 mM MgCl₂). 2 μL compound at varying concentrations or 2% DMSO in exchange buffer were added into each well and incubated for 24 h at 4 °C. 5 μL MBP-Trio (100 nM) or only buffer was added into each well. The exchange of GDP to BODIPY-FL-GDP was initiated by adding 3 μL BODIPY-FL-GDP at a final concentration of 1.7 μM. The exchange was read immediately in a Biotek Neo2 Multimode plate reader using excitation and emission wavelengths of 485 nm and 535 nm. The fluorescence increase was fitted to an exponential function:

$$\text{Fluorescence}\text{intensity}=\text{Initial}\text{fluorescence}+\text{extent}\text{of}\text{binding}(1-{\text{e}}^{-\text{Rate}\times \text{Constant}\times \text{Time})}$$

The rate constant was calculated by fitting experimental values for total fluorescence intensity and corresponding time. Percent inhibition was calculated by comparing the rate constant of the compound inhibited sample versus the DMSO control and the minimal control without GEF (Trio). Based on the plot of the percent inhibition versus compound concentration, a four-parameter logistic curve was fit to determine the ${\text{IC}}_{50}$ values at 24 h incubation time.

$$\begin{gathered} \text{Percent}\text{Inhibition}=\text{Minimum}\text{Inhibition} \\ +(\text{Maximum}\text{Inhibition}-\text{Minimum}\text{Inhibition})/1+{(\text{Compound}\text{Concentration}/{\text{IC}}_{50})}^{-\text{HillSlope}} \end{gathered}$$

Maximum inhibition was set at 100% as no clear plateau was achieved. Minimum inhibition was found to be near 0%.

Protein mass spectrometry.

20 μL His-RhoA (2 μM) or the RhoA mutants (2 μM) were incubated with the fragments (200 μM) for 24 h at 4 °C. The samples were centrifuged at 20000 g for 10 mins to remove any precipitants. The sample was loaded onto a Zorbax 300-SB-C3 column using an autosampler on an Agilent 1290 liquid chromatography system using a gradient of buffer A (H₂O with 0.1% formic acid) and buffer B (H₂O with 0.1% acetonitrile), and masses were detected and analyzed using an Agilent 6545 Q-TOF mass spectrometry.

Time and concentration dependent protein mass spectrometry.

Time and concentration dependent ESI-MS was done at varying concentrations of fragments for 10 mins to 24 h at 4 °C. Percent adduct formation of 1 (ACR-895) and 2 (ACR-917) was plotted against time to derive the k_obs. k_obs for each of the fragment was then plotted against concentration to determine K_I and k_inact.

Conversion of GDP to GNP and Biotinylation.

His-RhoA GDP was converted to His-RhoA.GMPPNP (GNP) using alkaline phosphatase enzyme in buffer consisting of 20 mM Tris, pH 8.0, 100 mM NaCl, 2.5 mM MgCl₂, 300 mM (NH₄)₂SO₄, and 5% glycerol. His-RhoA GDP was incubated with 5–10 times molar excess of GNP and rotated for 3 h at 4 °C. The reaction was quenched using MgCl₂ and ran onto a size exclusion chromatography column (superdex 75 pg) using 20 mM Tris, pH 8.0, 100 mM NaCl, and 2.5 mM MgCl₂ for further purification of the His-RhoA.GNP. Protein was biotinylated using Thermo Scientific’s EZ-Link Micro NHS-PEG₄ biotinylation kit for 1 h at 25 °C before desalting using Zeba spin columns (7K MWCO, 2.0 mL) into 20 mM HEPES (pH 7.5), 5 mM MgCl₂, and100 mM NaCl. Protein concentration was determined using a Thermo Fisher Scientific Nanodrop One.

AlphaScreen assay.

AlphaScreen was conducted in a white 384 microplate (Perkin Elmer), round bottom, polystyrene, with a maximum volume of 20 μL. 10 μL Biotinylated RhoA.GNP at a concentration of 80 nM was added in each well in AlphaScreen buffer (20 mM Hepes, pH 7.5, 100 mM NaCl, 5 mM MgCl₂, and 0.01% Triton X-100) and incubated with increasing concentrations of 1 (ACR-895) or 2 (ACR-917) for 24 h at 4 °C. After 24 h GST-ROCK1 (6.5 nM) was added in each well and incubated at RT for 30 mins. The AlphaScreen buffer was supplemented with 0.25% BSA in the GST-ROCK1 solution. After 30 minutes anti-GST-antibody acceptor beads (20 μg) and streptavidin donor beads (20 μg) [Promega] were added in each well and incubated at RT in the dark for 45 minutes before reading the plate in an Envision plate reader at an excitation and emission spectrum of 680 nm and 570 nm. Percent inhibition was calculated by comparing the fluorescence intensity of the compound inhibited sample versus the DMSO control and the control without the target protein.

Supplementary Material

1

Figure S1. (A) Time-dependent nucleotide exchange for RhoA^C107S,C190S (2 μM) incubated with increasing concentration of fragment 1 (ACR-895) following 24 h pre-incubation at 4 °C prior to adding Trio (100 nM) and BODIPY-FL-GDP (250 nM). (B) Time-dependent nucleotide exchange for RhoA^C107S,C190S (2 μM) incubated with increasing concentration of fragment 2 (ACR-917) following 24 h pre-incubation at 4 °C prior to adding Trio (100 nM) and BODIPY-FL-GDP (250 nM).

Figure S2.(A) RhoA^C190S•GDP (2 μM) were incubated with 1 (ACR-895) [200 μM) for 24 h at 4°C using Tris buffer at pH 7.4 (20 mM Tris, pH 7.4, 100 mM NaCl, 10 mM MgCl2). (B) RhoA^C190S•GDP (2 μM) were incubated with 1 (ACR-895) [200 μM) for 24 h at RT using Tris buffer at pH 7.4 (20 mM Tris, pH 7.4, 100 mM NaCl, 10 mM MgCl2). The reactions were quenched with formic acid, followed by collection of MS.

Figure S3. Inhibition of RhoA^C190S, Trio and RhoA^C190S, LARG nucleotide exchange with increasing concentration of 1 (ACR-895) incubated for 24 h at 4°C determined from the time-dependent exchange curves; mean ± s.d., n = 2 replicates.

Figure S4. (A) Inhibition of RhoA^C190S and RhoA (untagged) nucleotide exchange with increasing concentration of 1 (ACR-895) determined from the time-dependent exchange curves; mean ± s.d., n = 2 replicates. (B) Inhibition of RhoA^C190S and RhoA (untagged) nucleotide exchange with increasing concentration of 2 (ACR-917) determined from the time-dependent exchange curves; mean ± s.d., n = 2 replicates.

Figure S5. (A) Biotinylated GST-RhoA•GMPPNP (80 nM) is incubated with increasing concentration of 1 (ACR-895) or (B) 2 (ACR-917) for 24 h at 4°C followed by addition of HisArhgap11a (100 nM), streptavidin- and anti-His-antibody-loaded beads, and the reading of fluorescence; mean ± s.d., n = 2 replicates.

Table 1.

Inhibition of RhoA and Rac1 Exchange by Fragments

	RhoAC190S GDP		Rac1 GDP
Compound	IC50 (μM)	Max (%)	IC50 (μM)	Max (%)
1 (ACR-895)	19.6 ± 3.1	100	37.0 ± 11.0	100
2 (ACR-917)	45.8 ± 4.2	100	<1000	100