Targeting the α4-α5 interface of RAS results in multiple levels of inhibition

ABSTRACT

Generation of RAS-targeted therapeutics has long been considered a “holy grail” in cancer research. However, a lack of binding pockets on the surface of RAS and its picomolar affinity for guanine nucleotides have made isolation of inhibitors particularly challenging. We recently described a monobody, termed NS1, that blocks RAS signaling and oncogenic transformation. NS1 binds to the α4-β6-α5 interface of H-RAS and K-RAS thus preventing RAS dimerization and nanoclustering, which in turn prevents RAS-stimulated dimerization and activation of RAF. Interestingly, NS1 reduces interaction of oncogenic K-RAS, but not H-RAS, with RAF and reduces K-RAS plasma membrane localization. Here, we show that these isoform specific effects of NS1 on RAS:RAF are due to the distinct hypervariable regions of RAS isoforms. NS1 inhibited wild type RAS function by reducing RAS GTP levels. These findings reveal that NS1 disrupts RAS signaling through a mechanism that is more complex than simply inhibiting RAS dimerization and nanoclustering.

RAS proto-oncogenes are among the most frequently mutated proteins in human cancer.¹ These mutations lead to chronic and inappropriate RAS activation resulting in a plethora of pro-oncogenic signaling events. RAS has a multitude of effectors which regulate growth, survival, cytoskeletal dynamics, cell cycle progression, migration, endocytosis and calcium signaling.² There are 3 distinct mammalian RAS genes: H-RAS, N-RAS, and K-RAS. In contrast, to H- and N-RAS, each of which encode a single major protein isoform, K-RAS is alternatively spliced to produce 2 distinct protein isoforms, K-RAS4A and K-RAS4B.

Each RAS isoform consists of 2 major domains: the G-domain (residues 5–166), which shows more than 90% identity between isoforms, and the highly divergent hyper variable region (HVR) (residues 167–188/9).³ The RAS G-domain consists of the catalytic domain responsible for GTP hydrolysis, the guanine nucleotide binding pocket (residues 10–17, 57–61 and 116–119), and the effector binding region consisting of 2 “switch” regions (Switch I, residues 30–40 and Switch II, residues 60–76). The HVR mediates membrane targeting and proper localization of RAS. This region contains a CAAX motif at the extreme C-terminus, which is modified with a membrane anchoring farnesyl lipid in all RAS isoforms. In addition, isoform-specific sequences upstream of the CAAX motif further define the localization of each isoform. Two cysteines in H-RAS (Cys 181 and 184) are palmitoylated whereas N-RAS and K-RAS4A are each palmitoylated at a single Cys residue, Cys 181 and Cys 180, respectively. In contrast, the major K-RAS isoform, K-RAS4B (referred to hereafter as K-RAS), lacks these additional Cys residues and instead possesses a polybasic region comprised of an uninterrupted chain of lysine residues that stabilize K-RAS4B membrane anchoring and geometry.⁴

RAS GTPases elicit their function by cycling between the GTP-loaded “active” and GDP-loaded “inactive” states. GTP is hydrolyzed to GDP by the intrinsic GTPase activity, which is accelerated by RAS GTPase activating proteins (GAPs). Once activated by an upstream stimulus, RAS guanine nucleotide exchange factors (GEFs), such as SOS1, promote GDP release from RAS. Given the ∼10-fold molar excess of GTP to GDP in cells coupled with the high affinity of RAS for nucleotides, RAS reloads with GTP. This results in the active conformation in Switch I and Switch II, allowing binding and activation of specific effectors such as RAF and PI3Ks.³

RAS activating mutations occur in around 30% of human cancers.¹ These mutations impair the ability of RAS to hydrolyze GTP to GDP, locking RAS in the GTP loaded “active” state. RAS mutations occur in both an isoform-specific and tissue-specific manner. H-RAS mutations occur predominantly in cancers of the cervix and urinary tract, whereas N-RAS mutations are associated with cancers of the skin and endometrium.^1,5,6 K-RAS is the most frequently mutated RAS isoform in human cancers, with mutations predominating in cancers of the pancreas, lung, and colon.^1,5,6 For these reasons RAS has become one of the highest priority targets in human cancer as illustrated by establishment of the National Cancer Institute RAS Initiative.⁷ Despite this priority, there are still no drugs in the clinic that directly target and inhibit RAS activity.⁷ This is largely due to a lack of deep binding pockets on the surface of RAS and its picomolar affinity for GTP.⁸ Thus, early attempts to competitively block GTP binding produced only modest results.^9,10 Indirectly targeting RAS has also been attempted by blocking the attachment of the C-terminal farnesyl group with farnesyltransferase inhibitors (FTIs) resulting in mislocalization of RAS from the plasma membrane.¹¹ Although FTIs show efficacy against H-RAS, the ability of K-RAS and N-RAS to undergo alternative lipid modification by geranylgeranyltransferases renders them insensitive to FTIs.¹² Thus, identification of new strategies for inhibiting RAS is greatly needed.

Significant progress has been made toward this goal in recent years. Two groups have isolated small molecules against K-RAS that bind to a hydrophobic pocket encompassing regions of Switch I and Switch II, including residues important for SOS binding.^13-15 These compounds block K-RAS interaction with SOS and moderately reduce nucleotide exchange.^13,14 This same hydrophobic pocket has been used to prevent RAF interaction, which led to increased apoptosis in an N-RAS-mutant (Q61K) non-small cell lung cancer line, H1299.^16,17 SOS interaction has also been targeted using a cell permeable peptide based on the RAS-interacting α-helix 1 of SOS that blocks the interaction of K-RAS with SOS1 and further prevents GTP association, resulting in inhibition of K-RAS function.¹⁸ However, none of these antagonists possess sufficient affinity and efficacy to potently block RAS function in a therapeutically relevant setting.

Recently, 2 studies have shown promising results by targeting a previously uncharacterized pocket beneath Switch II in K-RAS(G12C). These compounds block its function by shifting the nucleotide binding preference of K-RAS(G12C) to GDP.^19,20 These antagonists require a cysteine at position 12 for binding and thus do not bind to wild type (WT) K-RAS, or other oncogenic mutations. Although their efficacious doses are in the low micromolar range, further improvements to the chemical and pharmacodynamic properties of these compounds may lead to more potent in vivo inhibition of mutant K-RAS. Thus, this binding pocket in Switch II represents a promising target for development of K-RAS mutant-specific drugs, at least against the G12C mutant.

We have discovered another approach to blocking RAS function: inhibition of RAS dimerization/nanoclustering.²¹ We recently reported the development of a synthetic binding protein termed NS1 monobody (referred to hereafter as NS1) that potently inhibits oncogenic RAS-driven signaling and transformation by binding the α4 and α5 helices of the G-domain (referred to as the α4-α5 interface hereafter).²¹ Although the isolated G domain of RAS does not detectably form dimers in solution, growing experimental evidence suggests that activation of RAS at the plasma membrane promotes dimers and nanoclusters that contribute to activation of downstream effectors such as RAF (reviewed in ref. 22). Indeed, numerous studies using quantitative imaging technologies such as electron microscopy (EM)-spatial mapping and fluorescence lifetime imaging microscopy-fluorescence resonance energy transfer (FLIM-FRET) reveal that RAS proteins segregate into nanoclusters of ∼9 nm radius containing 6–7 molcules of RAS, which function as highly dynamic platforms for RAS signaling.²³ Furthermore, superresolution microscopy supports the premise that RAS proteins form dimers at the plasma membrane.¹⁷ However, these dimers were independent of the activation state of RAS and mediated by the hypervariable region.¹⁷ Our analysis of H-RAS crystal structures reveals the presence of a unique α4-α5 dimeric structure in “active” H-RAS that is not present in “inactive” state H-RAS structures, suggesting weak propensity of the G-domain to dimerize using the α4-α5 interface.²¹ Similar dimers have been reported in N-RAS and K-RAS.^21,24,25 NS1 represents the only reagent available as yet that binds to the α4-α5 interface and reduces dimerization and nanoclustering of full-length H- and K-RAS on the cell membrane.²¹

Unexpectedly, mutations of residues predicted to weaken dimer formation through the α4-α5 interface (e.g., charge reversal mutations of R135, D154, and R161)²⁴ do not affect oncogenic H-RAS-mediated signaling.²¹ These results suggest that RAS dimerization does not require the formation of a single, precisely defined dimer interface and these mutations still allow RAS protomers to stay within sufficient proximity to drive dimerization of RAS-associated RAF molecules. In contrast, binding of NS1, a 10 KDa globular protein, may provide sufficient steric bulk to interfere with RAS-driven dimerization of RAF. Nevertheless, these findings suggest that the formation of RAS G-domain dimers requires synergistic contributions of membrane tethering, G-domain interaction, and potentially additional interactions mediated by the HVR. Our results suggest a new approach to therapeutically inhibit RAS by targeting the α4-α5 interface thereby blocking RAS dimerization/nanoclustering and inhibiting effector activation.

Our studies have also revealed isoform specific differences in the effects of NS1 on H-RAS vs K-RAS. For example, NS1 reduces the localization of K-RAS, but not H-RAS, at the plasma membrane.²¹ In addition, NS1 inhibits the association of K-RAS with RAF but has little effect on H-RAS:RAF association.²¹ However, our crystallography and NMR studies indicate that NS1 does not sterically interfere with the switch regions nor cause significant allosteric changes in these regions. Furthermore, biochemical experiments indicate that NS1 does not alter GTP-loaded H-RAS_1–166 binding to RAF-RBD in vitro.²¹ Thus, these data suggest that NS1 does not alter the affinity of the G-domain for effectors. Here, we demonstrate that these isoform-specific effects of NS1 are due to differences in the HVRs of the RAS isoforms. Furthermore, we found that NS1 reduces the level of GTP-loaded WT H-RAS following growth factor stimulation. Thus, our results further illustrate the utility of targeting the α4-β6-α5 region to block RAS function and reveal differences in the inhibitory effects of NS1 on oncogenic and WT RAS isoform function.

NS1 blocks oncogenic H-RAS and K-RAS signaling and transforming activity.²¹ To determine the potency of this effect, we performed a transfection-titration analysis. NS1 reduced oncogenic H-RAS activation of ERK by ∼80% at a molar ratio of less than 2:1 NS1:RAS (Fig. 1A and B). In addition to blocking oncogenic RAS, NS1 inhibited growth factor activation of ERK-MAPK.²¹ Indeed, CFP-NS1 but not CFP alone precipitated both WT H-RAS and K-RAS from cell lysates (Fig. 1C), consistent with the in vitro binding specificity of NS1. Thus, NS1 binds and inhibits both WT as well as oncogenic RAS proteins.

Figure 1.

NS1 inhibits WT RAS. (A and B). NS1 dose-dependently inhibits RAS-mediated ERK activation. (A) Titration of CFP-NS1 effects on H-RAS(G12V)-mediated MYC-ERK activation in HEK293 cells. H-RAS(G12V) and MYC-tagged ERK were co-expressed with varying amounts of CFP-NS1. Following MYC-ERK immunoprecipitation, ERK activation was detected by Western blot with a phosphospecific ERK (pERK) antibody. Whole cell lysates (WCL) were examined for expression of transfected constructs (bottom panels). (B) Quantification of (A) with densitometry using NIH ImageJ. Results are the mean +/−s.e.m. of 3 independent experiments. p value was determined by a Student's t-test using Graphpad. ***p < 0.001. (C) NS1 binds WT H-RAS and K-RAS in cell lysates. HEK293 cells were transiently transfected with the indicated HA-tagged WT RAS expression constructs along with FLAG-tagged CFP-NS1 or CFP alone. CFP and CFP-NS1 were immunoprecipitated with an anti-FLAG antibody and examined for binding to HA-tagged WT H-RAS or K-RAS by Western blot analysis (top 2 panels). Whole cell lysates were examined for expression of transfected constructs (bottom panel). (D) Cells were transfected with WT H-RAS and CFP or CFP-NS1 as in Fig. 1A. GST-RAF-RBD or GST-alone were used to precipitate HA-tagged H-RAS(WT) from whole cell lysates. Bound proteins were fractioned on gels and probed with the indicated antibodies (top panels). (E) Quantification of (D) with densitometry using NIH ImageJ. Results represent the average of 3 independent experiments +/−s.e.m. p value was determined by a Student's t-test using Graphpad. *,p < 0.05

Our previous work demonstrates that NS1 does not affect the in vitro nucleotide binding or SOS-mediated nucleotide exchange on H-RAS.²¹ In addition, NS1 does not block interaction of oncogenic H-RAS with full length RAF or GST-RAF RBD²¹ suggesting that NS1 does not affect oncogenic H-RAS GTP levels.²¹ These results are consistent with the observation that NS1 does not affect binding of GTP-loaded H-RAS_1–166 with the GST-RAF RBD in vitro. However, when expressed in cells CFP-NS1, but not CFP alone, reduced the amount of GTP-loaded WT H-RAS following EGF stimulation (Fig. 1D and E). Together, these data suggest that NS1 may inhibit EGF-stimulated nucleotide exchange on full length RAS in cells. Alternatively, it is possible that NS1 may decrease H-RAS GTP levels by facilitating RAS interaction with GAPs resulting in enhanced GTPase activity. Nevertheless, these data reveal an added level of complexity to the effect of NS1 on inhibiting WT RAS-mediated signaling.

Although the mechanism of action of NS1 on both RAS isoforms is centered on blocking dimerization/nanoclustering, NS1 has isoform-specific effects on RAS with regard to RAF binding and plasma membrane localization.²¹ NS1 did not block binding of the isolated G-domains of either H-RAS or K-RAS to the RAF RBD in vitro (Fig. 2). In contrast, NS1 blocks the interaction of oncogenic K-RAS with RAF in cells without affecting mutant H-RAS:RAF interaction.²¹ Given that the HVRs of RAS isoforms represent the regions of greatest sequence divergence, we examined whether the HVRs contribute to this differential effect of NS1 on H-RAS vs K-RAS. We constructed chimeric proteins where the HVRs of oncogenic H-RAS(G12V) and K-RAS(G12V) were switched (Fig. 3A). The binding pattern of NS1 to these chimeric proteins corresponded to the respective G-domain. For example, H-RAS(G12V) and H-RAS(G12V)-K-tail bound similar amounts of NS1 whereas K-RAS(G12V) and K-RAS(G12V)-H-tail bound lower amounts of NS1, consistent with the lower affinity of NS1 for K-RAS than for H-RAS G-domain in vitro (Fig. 3B).²¹ Additionally, H-RAS(G12V)-K-tail and K-RAS(G12V)-H-tail activated ERK comparatively to H-RAS(G12V) and K-RAS(G12V), respectively, demonstrating the functionality of these chimeras. Finally, NS1 inhibited ERK activation by each RAS protein to similar extents (Fig. 3B).

Figure 2.

NS1 does not affect G-domain binding to RAF RBD in vitro. Following loading with either GDP or GTPγS (indicated as GTP), H-RAS_1–174 and K-RAS_1–174 were incubated with either GST or GST-RAF RBD bound to Glutathione Sepharose beads in the presence or absence of NS1. Proteins bound to the Glutathione beads (top 2 panels) were analyzed by Western blot using anti-HIS or anti-GST antibodies. Total input of RAS and NS1 is shown in the bottom 2 panels. The results shown are representative of at least 3 independent binding experiments with similar results.

Figure 3.

The RAS HVR determines the effect of NS1 on RAS:RAF interaction. (A) Schematic representations of oncogenic H-RAS and K-RAS proteins. Amino acid 167–188/9 were swapped between H-RAS and K-RAS G-domains. C, designates sites of palmitoylation in H-RAS (Cys 181 and 184); KKKKKK, polybasic domain of K-RAS; CAAX, site of farnesylation. (B) The RAS G-domain determines the affinity for CFP-NS1. HA-tagged RAS constructs were co-expressed with CFP or CFP-NS1. Association of NS1 with the various RAS proteins was determined by co-immunoprecipitation and Western blot analysis (top 2 panels). ERK activation was assessed by Western blot analysis of whole cell lysates (WCLs) with antibodies to phosphorylated ERK (pERK) and total ERK. Levels of each protein are shown in the bottom 4 panels. (C) Effect of NS1 on RAS:BRAF interaction is dependent on the HVR. HEK-293T cells were transiently transfected with the indicated HA-tagged RAS expression constructs along with FLAG-tagged CFP-NS1 or CFP alone as in (B). Following serum starvation, RAS proteins were immunoprecipitated and examined for association with endogenous BRAF. The top 2 panels represent different exposures of the same blot. Whole cell lysates (WCL) were examined for expression of transfected constructs (bottom 2 panels). (D) Quantification of (C) with densitometry using NIH ImageJ. Results represent the average of 2 independent experiments +/−s.e.m.

We next examined the effects of NS1 on RAF interaction with each of these RAS proteins. In keeping with previous results,^21,26 K-RAS(G12V) bound RAF to a greater extent than H-RAS(G12V), and NS1 reduced K-RAS(G12V) interaction with RAF but did not affect RAF interaction with H-RAS(G12V) (Fig. 3C). This effect of NS1 was essentially reversed with the chimeric proteins. H-RAS(G12V)-K-tail bound RAF to a similar extent as K-RAS(G12V) and NS1 reduced the interaction of RAF with both proteins (Fig. 3C and D). In contrast, K-RAS(G12V)-H-tail bound RAF at levels similar to H-RAS(G12V) and this binding was unaffected by NS1 (Fig. 3C and D). These results demonstrate that the HVR and not the G-domain, dictates the differential effects of NS1 on RAS:RAF interaction.

How might NS1 differentially affect the HVR of RAS isoforms? Comparison of the NS1:H-RAS_1–166 structure²¹ to full length RAS structures^13,27 reveals that NS1 extends beyond the G-domain of RAS and projects toward the HVR (Fig. 4A and B). This spatial arrangement suggests that NS1 may affect the interactions of the G-domain with the HVR and/or the plasma membrane. NS1 binding to α5, which extends beyond amino acid 166 to residues 172–174,^13,27-29 may stabilize or facilitate α5 interactions with the G- domain, HVR and plasma membrane. In addition, NS1 may disrupt the interaction of the polybasic domain of K-RAS with the plasma membrane thereby decreasing affinity of K-RAS for the membrane. Consistent with this possibility, NS1 decreases binding of K-RAS, but not H-RAS, to the plasma membrane.²¹ This result likely stems from the weaker association of K-RAS with the plasma membrane due to a single lipid attachment vs H-RAS which has 3 lipid anchoring sites.³⁰ NS1 disruption of K-RAS dimerization/nanoclustering may reduce the avidity of K-RAS for the plasma membrane vs H-RAS as we previously proposed.²¹ Despite not substantially affecting H-RAS membrane association,²¹ NS1 may alter the interaction or orientation of RAS with the lipid bilayer through steric hindrance thereby affecting RAS function.²⁹ Indeed, the G-domains of each RAS isoform interacts with the plasma membrane in distinct orientations.³¹ Additional studies will determine contributions of these and other factors to the specific effects of NS1 on the HVR and interaction of the G-domain with the plasma membrane.

Figure 4.

Extension of the α5 helix influences NS1 binding. (A and B) Comparison of H-RAS:NS1 structure to full length K-RAS structures. NS1 (gold) binds the α4-α5 dimer interface of H-RAS and K-RAS and projects toward the extended α5 helix and the HVR. The structures of K-RAS (gray) were aligned to H-RAS in the H-RAS:NS1 structure (PDB ID: 5E95; blue) using Pymol.³⁹ The SW1 and SW2 regions are highlighted yellow and polybasic region in the HVR is highlighted red. (A) K-RAS (PDB ID: 4DSO)¹³ was aligned with H-RAS:NS1. The C-terminal 8 amino acids did not exhibit sufficient electron density to determine their structure. Note that the VIM sequence is removed upon processing of RAS in vivo. (B) K-RAS from the K-RAS:PDE6 structure (PDB ID: 5TB5)²⁷ was aligned with H-RAS:NS1. Note that 3–5 amino acids in HVR of K-RAS do not exhibit interpretable electron density suggesting flexibility in this region consistent with NMR and crystallographic data.^13,28,29 Although the α5 helix appears to be extended throughout the entire HVR in an alternative K-RAS:PDEδ crystal structure (PDB ID: 5TAR), it is likely that this extended helical structure may be due to crystal packing effects. The farnesyl group is highlighted green. The PDE6 structure has been removed for clarity. (C and D). Truncation of the α5 helix alters NS1 binding to RAS. Binding titration of NS1 displayed on the surface of yeast cells with K-RAS_1–166 (C) or H-RAS_1–166 (D) isoforms bound to nucleotide using flow cytometry detection. Affinity values (K_D) are as follows: K-RAS GDP, 490 ± 75 nM; K-RAS GTPγS, 410 ± 79 nM.; H-RAS GDP, 6.1 ± 0.23 nM; H-RAS GTPγS, 6.0 ± 0.6 nM. Error bars on graphs represent s.d. from n = 3. The dissociation constants shown are the average from 3 independent experiments +/−s.d.

Although NS1 was isolated using H-RAS_1–166, extension of the α5 helix by an additional 8 amino acids altered the affinity of NS1 for both H-RAS and K-RAS (Fig. 4C and D).²¹ K-RAS_1–166 bound NS1 with approximately 7-fold lower affinity than reported for K-RAS_1–174 (Fig. 4C).²¹ In contrast, the affinity of NS1 for H-RAS_1–166 was only slightly better than reported for H-RAS_1–174 (∼6 nM vs ∼15 nM²¹) (Fig. 4D), revealing different effects of extending the C terminus between H- and K-RAS. While these differences most likely stem from enhanced folding or stability of the α5 helix in the longer K-RAS constructs, extension of the α5 helix nevertheless alters NS1 binding, especially with regard to K-RAS. These results also suggest the possibility that NS1 binding may promote the formation of a defined structure of the region C-terminal to residue 166. These results support the premise that NS1 interacts with the extended C-terminus of RAS. However, future studies using targeted mutations in the α5 helix and HVR region will be needed to define the importance of these regions for NS1 binding and inhibition of RAS function.

In recent years, work outlining the requirement for RAS multimerization has been at the forefront of research in the field.^23,24,32,33 Indeed, blocking K-RAS self association is one of the goals of the National Cancer Institute RAS Initiative.⁷ Our work with the NS1 monobody demonstrates that this goal can be achieved by targeting the α4-α5 interface of H-RAS and K-RAS, leading to potent reduction in RAS-driven signaling and oncogenic transformation.²¹ Although NS1 does not bind or inhibit N-RAS due to sequence-specific differences in the α4-α5 interface of N-RAS compared with H-/K-RAS,²¹ we anticipate that N-RAS function would also be inhibited through targeting this region given its conservation among RAS isoforms.²⁴

The α4-α5 interface lies within the RAS allosteric lobe. In addition to encompassing the RAS dimer interface,^21,24 the allosteric lobe also contains the so called “Switch III” region, which encompasses residues 40–56 of the β2/3 loop and residues 152–166 of the α5 helix.³⁴ This region, along with Switch I and Switch II, essentially integrate the effector lobe, allosteric lobe and HVR during RAS activation.²³ Upon GTP-loading, conformational changes in Switches I and II result in a change in the orientation of RAS with the plasma membrane, priming RAS for effector interaction.^35,36 Although Switch III has been described in all major RAS isoforms, sequence differences in the allosteric lobes and HVRs of RAS isoforms lead to distinct orientations of RAS isoforms at the plasma membrane. These orientations also differ between the GDP- and GTP-loaded states.^35,36 These differences lead to isoform-specific effector interactions,^35,36 and, along with the HVR, contribute to the segregation of different RAS isoforms to distinct membrane microdomains.^37,38 Indeed, mutations in the α4 helix that stabilize RAF and PI3K binding to H-RAS have the opposite effect and reduce K-RAS:effector interactions.³⁵ Thus, NS1 binding to the α4-α5 interface may alter the orientation of RAS isoforms with the plasma membrane through interfering with Switch III interactions.

Given the orientation of NS1 with respect to the α4-α5 interface, it is not surprising that NS1 interferes with aspects of RAS signaling in addition to reducing RAS dimerization. Indeed, our results demonstrate that targeting this interface with NS1 disrupts the interaction of oncogenic K-RAS, but not H-RAS, with RAF²¹ and that these effects can be reversed by switching the HVR regions (Fig. 3C and D). NS1 also reduces the localization of oncogenic K-RAS, but not H-RAS at the plasma membrane.²¹ Our present findings demonstrate an additional level of RAS inhibition in which NS1 reduced WT RAS GTP levels (Fig. 1D and E). This inhibition may occur through interfering with GTP loading or enhancing GTP hydrolysis. The first possibility is in contrast to our previous findings showing that NS1 does not affect nucleotide cycling of the WT H-RAS G-domain in solution.²¹ However, NS1 binding to WT RAS in cells may result in conformational changes at the plasma membrane that interfere with nucleotide exchange.²⁹ Alternatively, NS1 may potentiate GAP-mediated GTP hydrolysis by WT RAS, although the potential mechanism underlying such an effect is unclear. In contrast, oncogenic H-RAS is constitutively GTP-loaded and insensitive to GAP activity. Thus, NS1 does not affect oncogenic H-RAS GTP levels.²¹ This effect of NS1 on WT RAS-GTP levels highlights the utility of targeting this interface in cancers where RAS is hyperactivated but not mutated. However, targeting the α4-α5 interface may result in RAS inhibition in normal cells leading to possible toxicity issues.

Taken together, our prior results²¹ combined with data in this current study reveal that targeting the α4-α5 interface with NS1 has multiple inhibitory effects on RAS function (Table 1). As we previously published, the primary mechanism of action of NS1 is to block RAS dimerization, which holds true for both H-RAS and K-RAS.²¹ However, NS1 also reduces WT RAS-GTP levels and, in the case of K-RAS, reduces RAS:RAF association and K-RAS plasma membrane localization (Table 1). These additional inhibitory effects of NS1 on K-RAS may explain why NS1 can potently block K-RAS activity in spite of binding K-RAS with lower affinity than H-RAS (Fig. 4C, D and Table 1).²¹ Nevertheless, our work demonstrates that targeting the allosteric lobe of RAS may be key to developing effective and potent inhibitors of RAS function and provides new hope of drugging this “undruggable” GTPase.

Table 1.

Effects of NS1 on RAS isoforms.

RAS Isoform	NS1 Binding	RAS Dimer	RAS:BRAF Binding	PM Localization	RAS·GTP Levels
H	YES	↓	—	—	↓
K	YES	↓	↓	↓	Not tested
N	NO	N/A	N/A	N/A	N/A

^*with respect to WT RAS

↓, decreases; -, no effect; N/A, not applicable

Methods

Cell culture, subcloning, transfections and reagents

HEK293 and HEK293T cells were cultured in DMEM (Corning, Tewksbury, MA, USA) with 10% FBS (Gemini, West Sacramento, CA, USA). These cells were transiently transfected using empirically determined concentrations of polyethyleneimine (PEI) along with Opti-Mem reduced serum medium (Life Technologies, Carlsbad, CA, USA) and serum-free DMEM. H-RAS-K-tail and K-RAS-H-tail were subcloned as follows: H-RAS(12V)_1–166 and K-RAS(12V)_1–166 were PCR amplified from the pCMV-VN173 vector using 5′ primers, 5′-GCCTGGGAGGACCTTCTAGCGGATCCACCATGACAGAATACAAGCTTGT –3′ (H-RAS) and 5′-GCCTGGGAGGACCTTCTAGCGGATCCACCATGACTGAATATAAACTT–3′ (K-RAS) and 3′ primers 5′-CCAAACTCACCCTGAAGTTCTCAGGA TCCTTACATAATTACACACTTTGTCTTTGACTTCTTTTTCTTCTTTTTACCATCTTTGCTCATCTTTTCTTTGTGCTGCCGGATCTCACGCAC-3′ (H-RAS-K-tail) and 5′-CCAAACTCACCCTGAAGTTCTCAGGATCCTCAGGAGAGCACACACTTGCAGCTCATGCAGCCGGGGCCACTCTCATCAGGAGGGTTCAGCTTCCGCAGCTTATGTTTTCGAATTTCTCGAAC-3′ (K-RAS-H-tail), digested with BamHI and cloned into the BamHI site of the HA-tagged, pCGN vector for mammalian cell expression.

Antibodies

Antibodies to ERK1/2, 1:2,000 dilution (9102S), phospho-ERK1/2, 1:2,000 dilution (9101L), MEK1/2, 1:1,000 dilution (9126S), phospho-MEK1/2 (Ser217/221), 1:1,000 dilution (9121S), were purchased from Cell Signaling Technology. Monoclonal Flag-M2 antibody, 1:4,000 dilution (F3165–1MG) and anti-Flag polyclonal antibody, 1:2,000 dilution (F7425-.2MG) were purchased from Sigma-Aldrich. Anti-HA.11 monoclonal antibody, 1:4,000 dilution (16B12) and anti-HA.11 polyclonal antibody (Poly9023) were purchased from BioLegend. Anti-MYC, 1:2,000 dilution was purchased from EMD Millipore (05–724); anti-GST (sc-459), 1:5,000 dilution, anti-BRAF (sc-5284), 1:500 dilution were purchased from Santa Cruz Biotechnology.

Cell signaling assays

Activation of transfected ERK was measured as described previously.²¹ Activation was quantified by densitometry using ImageJ. Quantification of ERK activation was determined by dividing the level of phospho-ERK by total-ERK. Relative CFP-NS1 to RAS expression was quantified by dividing the expression level of CFP-NS1 by the expression level of VN H-RAS(G12V). Activation of endogenous MEK and ERK in whole cell lysates was measured using the indicated phosphospecific antibodies. All cells were serum starved in serum-free DMEM overnight and stimulated for 10 minutes with 100 ng/ml EGF where indicated. These assays were performing in duplicate or triplicate.

Immunoprecipitation

Interaction of the indicated HA-tagged RAS GTPases with NS1 in cells was measured by coimmunoprecipitation as described.²¹ RAS interaction with BRAF was also measured by coimmunoprecipitation as described.²¹ Extent of BRAF co-precipitation was quantified by densitometry using NIH ImageJ. Amount (arbitrary units) of co-precipitated BRAF was divided by amount of precipitated RAS. These assays were performing in duplicate or triplicate.

Monobody affinity measurements

Measurements of monobody affinities were performed essentially as described previously.²¹

RAS activation assays

RAS activation was determined essentially as described.⁴⁰ Briefly, HEK293T cells were transiently transfected with the indicated constructs, serum starved overnight in DMEM and stimulated with EGF (100 ng/ml for 10 min), where indicated. HA-tagged RAS(WT) was precipitated from whole cell lysates (WCL) using GST fusion proteins which were expressed and purified from bacteria in 25 mM HEPES (pH 7.5), 150 mM NaCl, 10% glycerol, 1% Triton X-100, 25 mM MgCl2, 1 mM EDTA, 25 mM NaF, 0.25% DOC, 1 mM DTT, along with protease and phosphatase inhibitors. GST-pulldowns and WCLs were fractioned on gels and probed with the indicated antibodies. Extent of H-RAS(WT) co-precipitation was quantified by densitometry using NIH ImageJ. Amount (arbitrary units) of co-precipitated RAS was divided by amount of precipitated GST-RAF RBD. These assays were performed triplicate.

In vitro RAS-RAF binding assays

NS1, H-RAS_1–174 and K-RAS_1–174²¹ were expressed in bacteria as HIS tagged fusions using the pHBT vector and purified using nickel agarose beads. Purified RAS was incubated in Loading Buffer (20 mM Tris-HCl pH 7.6, 50 mM NaCl, 5 mM MgCl₂, 1 mM DTT, 4 mM EDTA) with either 2 mM GDP or 2 mM GTPγS for 90 min on ice followed by the addition of 10 mM MgCl₂ to stop loading reaction. GST and GST-RAF RBD were expressed and purified from bacteria as described. For binding reactions, 125 nM RAS was incubated with 125 nM GST or GST-RAF RBD (immobilized on Glutathione Sepharose beads) in the presence or absence of 375 nM NS1 in reaction buffer (25 mM HEPES,pH 7.5, 10% glycerol, 150 nM NaCl, 1%Trition100, 0.25% sodium deoxycholate, 25 mM MgCl₂, 1 mM EDTA) for 30 minutes at 4°C. Beads were then washed 3 times with reaction buffer and bound proteins analyzed by Western blot.

Abbreviations

CFP

cyan fluorescent protein

EGF

epidermal growth factor

FTI

farnesyltransferase inhibitor

GAP

GTPase accelerating protein

GEF

guanine nucleotide exchange factor

GST

glutathione S transferase

HVR

hyper variable region

IP

immunoprecipitation

PDEδ

phosphodiesterase6δ

RBD

RAS binding domain

WCL

whole cell lysates

WT

wild type

Disclosure of potential conflicts of interest

S.K. and A.K. are inventors on a patent that covers designs of monobody libraries (US Patent 9512199 B2).

Acknowledgments

We wish to thank members of the O'Bryan and Karginov laboratories for comments on this work. R.S.S is supported by an NIH F31 Predoctoral Award (CA192822).

Funding

This work was supported in part by a Merit Review Award (1I01BX002095) from the United States (US) Department of Veterans Affairs Biomedical Laboratory Research and Development Service to J.P.O. and NIH awards to J.P.O (CA116708 and CA201717) and S.K. (GM090324); a Catalyst award from the Chicago Biomedical Consortium with support from the Searles Funds at The Chicago Community Trust (S.K. and J.P.O.). R.S.S is supported by an NIH F31 Predoctoral Award (CA192822). The contents do not represent the views of the US. Department of Veterans Affairs or the United States Government.

Author contributions

R.S.S, S.K., and J.P.O. designed the study; A.K. performed the monobody affinity experiments; R.S.S., L.L., and S.P. performed biochemical and cell biology experiments; R.S.S and J.P.O wrote the manuscript and all authors commented and approved the manuscript.