Vanderbilt Institute of Chemical Biology



Discovery at the VICB







Targeting RAS Through SOS1


By: Carol A. Rouzer, VICB Communications
Published:  October 3, 2018


Potent agonists of SOS1-mediated guanine nucleotide exchange provide a novel approach for suppression of RAS signaling in cancer


RAS is a small (~21 kD) guanine nucleotide binding protein that acts as an on/off switch to modulate various growth, differentiation, and survival signaling pathways. RAS is activated through the exchange of bound GDP for GTP through the action of a guanine nucleotide exchange protein such as SOS1 (son-of-sevenless homologue 1). Inactivation of RAS results from the hydrolysis of GTP to GDP through the protein's weak intrinsic GTPase activity. A number of mutations of RAS result in a permanently active state, resulting in excessive cellular proliferation and survival. As these mutations contribute to the malignant behavior of multiple forms of cancer, an ongoing drug discovery effort has attempted to find inhibitors that block aberrant RAS signaling. Unfortunately, this effort has been stymied, primarily by the lack of well-defined pockets to which small molecules can bind and modulate RAS activity. This road block led Vanderibilt Institute of Chemical Biology members Steve Fesik and Alex Waterson, along with their colleagues, to seek an alternative approach for suppressing RAS-dependent signaling. They have previously discovered several distinct series of small molecules that bind to a hydrophobic pocket in SOS1 in a RAS-SOS1-RAS ternary complex. When bound, these molecules promote SOS1-dependent GDP-to-GTP exchange and, at low concentrations, stimulate RAS activity. At high concentrations, however, the more potent molecules trigger a feedback mechanism that leads to reduced RAS signaling. Now, the researchers report a new series of benzimidazoles that also promote SOS1-mediated nucleotide exchange on RAS via binding to this same hydrophobic pocket. This series includes the most potent compounds yet discovered, and further supports the potential of this approach to suppress RAS-dependent signaling in cancer cells (T. R. Hodges, et al., (2018) J. Med. Chem., published online September 11, DOI: 10.1021/acs.jmedchem.8b01108).


The investigators began their work by screening the Vanderbilt University compound collection for molecules that can facilitate SOS1-dependent nucleotide exchange on RAS. The screen identified two dihydrobenzimidazoles (compounds 1 and 2, Figure 1), which, together with information derived from active compounds in other series, inspired the synthesis of benzimidazole compound 3. The compounds were assessed on the basis of the concentration that resulted in 50% of the maximal stimulation of nucleotide exchange (EC50). In addition, the researchers evaluated the overall effectiveness of the activation, which was assessed using a relative percent activation (Act) – defined as the rate of nucleotide exchange at a 100 μM compound concentration expressed as a percent of the exchange rate achieved by the same concentration of a previously described control compound. Based on these criteria, compound 3 was the most active of the three early leads (Figure 1), and further work to optimize potency and efficacy began with this compound.




FIGURE 1. Structures of compounds 1, 2, and 3.



An X-ray co-crystal structure of compound 3 bound to a RAS:SOS1:RAS ternary complex (Figure 2) revealed that the molecule inserts into the same hydrophobic pocket utilized by the previously discovered series of molecules. The structure also revealed potential ways that compound 3's potency might be improved. Specifically, the investigators identified several hydrophobic subpockets that might be exploited by adding substituents to the molecule. In addition, the investigators hypothesized that strengthening a water molecule-mediated polar interaction between compound 3 and Asp-887 of SOS1 and/or establishing interactions with three additional nearby polar residues (Tyr-884, Glu-902, and His-905) should improve activity.



FIGURE 2. X-Ray co-crystal structure of compound 3 bound to SOS1 in a RAS:SOS1:RAS ternary complex. SOS1 is shown in gray and RAS in yellow space-filling mode. Compound 3 is in yellow sticks. The blue oval highlights the water-mediated interaction of 3 with Asp-887 of SOS1. Red arrows indicate a hydrophobic subpocket beneath Phe-890. The pink rectangle indicates a hydrophobic region along the CDC25 domain of SOS1. The green circle highlights a hydrophobic subpocket beneath Glu-902 and adjacent to Phe-890, and the white arrows indicate the locations of residues with the potential for interaction with substituents in the 4-position of the benzimidazole of 3. Image reproduced with permission from T. R. Hodges, et al., (2018) J. Med. Chem., published online September 11, DOI: 10.1021/acs.jmedchem.8b01108. Copyright 2018 American Chemical Society.



To begin testing these hypotheses, the investigators synthesized compound 10 (Figure 3), in which the amino group at the 2-position of the benzimidazole core is replaced with a piperazino group. The researchers hypothesized that the piperazine ring would establish a strong interaction with Asp-887 of SOS1, and an X-ray co-crystal structure of compound 10 in the RAS:SOS1:RAS complex confirmed this to be the case. The 4-fold increase in potency and 3-fold increase in activity of compound 10 over compound 3 confirmed the value of this interaction.



FIGURE 3. Structure of 10.



The researchers next tried to vary the substituents on the benzyl ring of compound 10 in an attempt to gain access to a hydrophobic pocket in the region of Phe-890 (Figure 2). The resulting compounds failed to exhibit improved potency or activity relative to those of compound 10, suggesting that steric constraints limit access to the pocket in that region. Similarly, efforts to replace the methyl groups at the 5- and 6-positions of compound 10 produced limited improvements in potency or efficacy. However, when they combined results from all of their tested analogs at this point, they discovered compound 29 (Figure 4), which exhibited substantially better potency than compound 10, though with somewhat reduced activity. The higher potency of compound 29 necessitated the use of an assay that directly measured the affinity of the molecule for the binding site (Kd). This became the standard for compound comparisons going forward.



FIGURE 4. Structure of 29.



A co-crystal structure of compound 29 in complex with the RAS:SOS1:RAS complex revealed the presence of a large hydrophobic space near 29's benzimidazole ring (Figure 5). The investigators hypothesized that addition of a hydrophobic moiety into this space would further enhance binding to SOS1. Consequently, they synthesized a series of compound 29 analogs of which compound 38 (Figure 6), with a chlorobenzyl group at position-4, was particularly potent. Unfortunately, 38 proved to be a highly lipophilic compound (as indicated by its AlogP of 7.3), suggesting that it might not be a good choice for use in vivo.




FIGURE 5. X-Ray co-crystal structure of compound 29 bound to SOS1 in a RAS:SOS1:RAS ternary complex. SOS1 is shown in gray and RAS in yellow space-filling mode. Compound 29 is in teal sticks. The red spheres and surrounding dots indicate the positions and space occupied by water molecules Image reproduced with permission from T. R. Hodges, et al., (2018) J. Med. Chem., published online September 11, DOI: 10.1021/acs.jmedchem.8b01108. Copyright 2018 American Chemical Society.




FIGURE 6. Structure of 38.



A co-crystal structure of compound 38 bound to the RAS:SOS1:RAS complex indicated that Glu-902 had rotated relative to its position in earlier structures to provide room for 38's chlorobenzyl group. This led the researchers to synthesize compound 43 (Figure 7) containing a piperazino group in place of 38's chlorobenzyl group. They hypothesized that the piperazino group provided a protonated nitrogen atom that could interact with Glu-902. This modification did not lead to an increase in binding affinity, but it did reduce the lipophilicity of the compound. The finding that compounds 40 and 41 (Figure 7), containing a piperidinyl and morpholinyl substituent, respectively, retained affinities similar to that of 43 regardless of their inability to interact with Glu-902 suggested that establishing a polar interaction with Glu-902 would not necessarily lead to a major improvement in potency.



FIGURE 7. Structures of 40, 41, 43, 47, and 58.



As creating contacts with Glu-902 provided no significant advantage, the researchers next turned to Tyr-884. Attempts to create contacts with this amino acid led to creation of compound 47 (Figure 7), containing an aminomethyl-cyclopentane substituent at position-4 of the benzimidazole. This compound was designed to direct the amino group of the substituent toward Tyr-884. Further work to create similar structures led to compound 58 (Figure 7), which exhibited a Kd value below the limits of detection of the assay being used at the time. Ironically, structural data indicated that, despite its remarkably high affinity, this compound did not make polar contacts with any of the residues in the vicinity of Tyr-884.

With compound 58, the researchers had achieved an affinity of <100 nM, requiring a modification of their assay to enable them to measure the affinities of more potent molecules. At this point, they also decided to revisit the structure-activity relationships at other points in the molecule, based on the hypothesis that substitutions added late in the process might have altered interactions that had been explored previously. This effort led to compounds containing substituents at the 4-position of the benzimidazole ring, as found in 47 and 58, but with a diazaspiro[3.3]heptane moiety replacing the piperazinyl group at position-2. Many of these compounds, such as 64 and 65 (Figure 8), exhibited exceptionally high affinity. X-Ray co-crystal structures of these compounds bound to the RAS:SOS1:RAS complex (Figure 9) verified an interaction between the diazaspiro[3.3]heptane group and Asp-887. Interestingly, the methylpiperidinyl group of 64 established a polar contact with Glu-902, whereas this residue rotates away from the dimethylpyrazole group of 65, enabling it to fill the hydrophobic pocket.



FIGURE 8. Structures of 42, 64, and 65.




FIGURE 9. X-Ray co-crystal structures of compound 64 (top) and compound 65 (bottom) bound to SOS1 in a RAS:SOS1:RAS ternary complex. SOS1 is shown in gray and RAS in yellow space-filling mode. Compound 64 is in light gray sticks and compound 65 is in yellow sticks. Yellow arrows indicate the rotation of Glu-902 to either interact with 64 or provide room for 65 without direct interaction.  Image reproduced with permission from T. R. Hodges, et al., (2018) J. Med. Chem., published online September 11, DOI: 10.1021/acs.jmedchem.8b01108. Copyright 2018 American Chemical Society.


With high affinity compounds in hand, the researchers tested them for their ability to affect RAS-dependent signaling in two cancer cell lines. As one consequence of RAS activation is phosphorylation of the extracellular signal-regulated kinase (ERK), they used this as their endpoint. Specifically, they assessed compounds for their ability to stimulate ERK phosphorylation at lower concentrations and then suppress it at higher concentrations. Of the compounds tested, they found 12 that met these criteria. In general, the more potent compounds elicited the biphasic response, but this was not always the case. The researchers proposed that other factors, such as lipophilicity, solubility, protein binding, membrane permeability, and export from the cell also likely impact a compound's potency in this assay. Compounds 64 and 42 (Figure 8) were the most potent molecules from any series yet observed in this cellular assay, enabling both a reduction in concentration used and a decrease in treatment time.


These efforts provided a number of highly potent compounds that modulate SOS1-mediated nucleotide exchange in RAS. They provide critical tool compounds that can be used to assess the potential to exploit this mechanism to suppress RAS signaling in cancer and may ultimately serve as the starting point for clinical translation to provide a novel chemotherapeutic drug.


View J Med Chem article: Discovery and Structure-Based Optimization of Benzimidazole-Derived Activators of SOS1-Mediated Nucleotide Exchange on RAS







The Vanderbilt Institute of Chemical Biology, 896 Preston Building, Nashville, TN 37232-6304, phone 866.303 VICB (8422), fax 615 936 3884
Vanderbilt University is committed to principles of equal opportunity and affirmative action. Copyright © 2014 by Vanderbilt University Medical Center