Exploiting c-RAF dependency in RAS mutant cancer: beyond catalytic activity

KEYWORDS:

• RAS
• c-RAF
• kinase-dependent
• kinase-independent
• oncology

1. Introduction

A major downstream effector of RAS, c-RAF (RAF1) is considered the preferential RAF isoform responsible for supporting KRAS tumor initiation [1–3]. Strengthening this narrative are two convincing pre-clinical studies, where-by systemic ablation of c-RAF induced robust tumor suppression in mouse (PDX and GEM) models of KRAS mutant (KRAS^MT) pancreatic and lung adenocarcinomas; with limited observable toxicity [4,5]. These studies, and others since [6,7], highlighted a non-essential role for c-RAF kinase activity in KRAS^MT tumor progression. Instead, c-RAF kinase-independent mechanisms were believed to be primarily responsible for the observed pro-oncogenic activity. Consequently, KRAS^MT cancer continues to be viewed as c-RAF ‘addicted’ and novel strategies to selectively targeting the inhibition of c-RAF (beyond its catalytic activity) are becoming increasingly attractive.

2. Targeting c-RAF catalytic activity

To date, no specific inhibitor of c-RAF has been developed. Existing small molecule kinase inhibitors capable of targeting c-RAF activity represent sub-optimal therapeutic approaches to treating RAS^MT cancer, and have been reviewed extensively (Table 1) [8–12].

Table 1. Mechanism of pharmacological inhibitors capable of exploiting c-RAF activity. Compounds in blue are FDA approved for the treatment of cancer. SM, small molecule; PD, peptide disruptor; TAT, cell-penetrating cationic peptide; st, stearic acid (short chain fatty acid).

Inhibitor	Class	Mechanism	Example Compound(s)
Kinase	Type I (SM)	Binds closed RAF at ATP pocket: αC-IN, DFG-IN	GW5074, Agerafenib,
	Type I½ (SM)	Binds closed RAF at ATP pocket: αC-OUT, DFG-IN	Vemurafenib, Dabrafenib, PLX4720
	Type II (SM)	Binds closed RAF at ATP pocket: αC-IN, DFG-OUT	Sorafenib, Regorafenib, Belvarafenib, Tovorafenib, Naporafenib, Lifirafenib, LY3009120, AZ-628, TAK-632, RAF-709, RAF-265, BI-882370, CCT196969
	Paradox Breakers (SM)	Binds closed RAF at ATP pocket: αC-OUT, DFG-IN. Additional B-RAF: Leucine505 binding, adjacent to αC-helix.	PLX7904, PLX8394
	c-RAF – MEK (SM)	Binds RAF – MEK dimer interface, trapping the heterodimer in an inactive conformation or disrupting heterodimer formation entirely	Trametiglue, ROS5126766 Trametinib
Kinase Independent	RAS-RAF Disruptor (SM)	Disrupts the interaction between GTP-active RAS and c-RAF	MCP110
	RAF Dimer Disruptor (PD)	Cell-penetrating peptide disruptor of the RAF intact dimer interface. Dimer disruption subsequently promotes B-RAF, c-RAF and MEK protein degradation. BRAF Peptide: T508-M517	TAT-Braftide
	c-RAF – PDE8A Disruptor (PD)	Cell-penetrating peptide disruptor of the c-RAF – PDE8A PPI, derived from the c-RAF binding epitope on PDE8A. Disruption promotes PKA-mediated inhibitory c-RAF phosphorylation (Ser259). PDE8A Peptide: R454-T465	RRLSGNEYVLST – St PPL-008C/N/CSS/NSS
	c-RAF − 14-3-3 Stabiliser (SM)	Disulphide tethering of c-RAF to 14-3-3σ, aimed at stabilizing c-RAF in a closed ‘inactive’ conformation. Compound 5 derived from c-RAF phospho peptide: Q^255RSTpSTPNVH^264	Compound 5

Type I and I½ ATP-competitive RAF inhibitors (RAFi) are differentiated by their ability to bind RAF in a closed conformation presenting as either DFG-IN:αC-helix-IN or DFG-IN:αC-helix-OUT, respectively. These therapies are largely limited in their therapeutic utility to B-RAF^V600E/K mutant cancer (i.e. constitutively active B-RAF monomers), where acquired resistance occurs within 12 months (even when combined with a downstream MEK inhibitor). In the presence of mutant RAS, type I and I½ RAFi have been shown to drive paradoxical MAPK activation through promoting dimerization of drug-bound (inhibited) B-RAF with unbound RAF (particularly c-RAF); fostering transactivation of the RAF dimer partner and hyperactivation of the MAPK pathway.

Type II RAFi were developed as a potential strategy to mitigate against paradoxical MAPK activation. Type II RAFi can be categorized as pan-RAF, binding to RAF in a DFG-OUT:αC-helix-IN conformation and targeting the inhibition of all kinase active RAF monomers and dimers with similar potency. However, clinical responses have been variable and dose-limiting toxicities are a concern (largely due to lack of selectivity compared with type I½ RAFi and indiscriminate binding of wild-type RAF).

Paradox breaker RAFi are derived from Type I½ RAFi, binding RAF in a DFG-IN:αC-helix-OUT confirmation. What separates paradox breakers from their earlier type I½ analogues is their ability to also strongly bind B-RAF at Leucine 505, adjacent to the RTKR motif of the αC-helix. Paradox breakers disrupt the B-RAF dimer interface, preventing B-RAF homodimerisation and B-RAF – c-RAF heterodimerisation. Although these inhibitors can prevent paradoxical MAPK activation, they cannot target c-RAF homodimers or c-RAF – A-RAF heterodimers. Therefore, paradox breaker RAFi are not considered a viable approach to treating RAS^MT driven cancers, where c-RAF homodimerisation and heterodimerisation with A-RAF is significantly upregulated vs. B-RAF homo/heterodimers [7,13].

Finally, RAF-MEK complex inhibitors – capable of disrupting RAF-MEK heterodimers or locking RAF-MEK in an inactive state – prove highly effective in abolishing RAF-driven MAPK activity. However, adverse side effects induced by these inhibitors are serious – resulting in significant dose-limiting efficacy in patients.

Given that most RAFi approaches are directed at inhibiting RAF kinase activity and preventing paradoxical MAPK activation, coupled with the consistent observation that c-RAF’s kinase-independent mechanisms are capable of driving RAS^MT tumourigenesis, novel strategies to exploiting c-RAF activity are urgently needed [14].

3. Targeting c-RAF independent of its catalytic activity

Though our understanding of c-RAF’s kinase-independent mechanisms remain within their infancy, several studies have demonstrated c-RAF’s ability to regulate cellular apoptosis, cell cycle progression and cell migration, independent of its catalytic activity (Figure 1). Said regulation is managed through specific c-RAF protein-protein interactions (PPIs) and has been shown to play a direct role in promoting RAS-driven oncogenesis [8,15,16].

Figure 1. Diagram highlighting studied c-RAF kinase-dependent and kinase-independent signaling pathways. Created with BioRender.com.

For instance, c-RAF possesses the ability to evade programmed cell death through its direct interaction with pro-apoptotic kinases ASK1 and MST2, and its indirect role as a scaffolding protein that promotes PKCθ-mediated inhibitory phosphorylation of BAD (Bcl-2 agonist of cell death).

Beyond its negative regulation of apoptosis, c-RAF can bind to and inhibit the activity of the cytoskeleton-based Rho effector kinase ROKα. c-RAF – ROKα PPI antagonizes ROKα, promoting cancer cell migration and survival. More, c-RAF – ROKα binding potentiates STAT3 and Myc expression, subsequently suppressing differentation (i.e. increased de-differentiation) and promoting RAS-induced epidermal tumourigenesis. Though no evidence of a direct interaction between c-RAF and STAT3 has been observed, increasing evidence of signaling cross-talk between c-RAF and STAT3 highlights a potential codependency in RAS – c-RAF dependent cancer [17].

c-RAF can also regulate the cell cycle in a kinase-independent manner through its interaction with kinases PLK1 and Aurora A at the mitotic spindle. Phosphorylation of c-RAF at Ser338 potentiated the formation of this protein complex, consequently upregulating PLK1 kinase activity and promoting tumourigenesis through driving mitotic progression. Additionally, c-RAF Ser338 phosphorylation (but not its kinase activity) promotes the interaction with CHK2, a checkpoint kinase that regulates the cell cycle via DNA repair mechanisms and is directly linked with radiotherapeutic resistance.

Finally, recent characterization of the HSP90-CDC37-c-RAF complex highlights yet another approach to exploiting c-RAF activity [18]. c-RAF binding to the HSP90-CDC37 chaperone complex is critical to stabilizing c-RAF in a folded, inactive conformation with 14-3-3 dimers. Disruption of this complex through site-directed mutagenesis promotes the degradation of c-RAF, inhibits cell proliferation, and highlights yet another vulnerability in c-RAF signaling; independent of its catalytic activity.

Selectively modulating PPIs within the oncoproteome represents a highly attractive and emerging therapeutic strategy, capable of fine-tuning the disease microenvironment and significantly reducing the potential for off-target toxicity [19,20]. However, given the significant size of the RAF PPI interactome, efforts aimed at characterizing c-RAF PPIs and defining their viability as drug targets are deficient [21,22]. This is particularly true for the above c-RAF kinase-independent targets, where PPI disruptor therapeutics are yet to be developed and evaluated in RAS^MT cancer. With that being said, a select number of pre-clinical compounds capable of modulating other c-RAF-associated PPIs have been developed, offering a differentiated approach to exploiting c-RAF activity vs. traditional kinase inhibitors (Table 1).

One such compound is MCP110, a small molecule derived from yeast two-hybrid screens that disrupts the interaction between RAS and c-RAF [23,24]. RAS – c-RAF disruption suppressed MEK-ERK activation inhibited the growth of KRAS^MT lung (A549 KRAS^G12S) and pancreatic (PANC1 KRAS^G12D) human cancer cell lines (not seen in A2058 BRAF^V600E cells), and attenuated HT1080 (NRAS^Q61K) cell invasion.

TAT-Braftide, a cell-penetrating 10-mer peptide designed via an in silico screening strategy to block the intact dimer interface (DIF) of RAF, was developed as an allosteric approach to inhibiting B-RAF and prevent paradoxical MAPK activation [25]. Unlike existing small molecule paradox breakers (e.g. PLX8394), TAT-Braftide-induced dimer disruption resulted in B-RAF and c-RAF protein degradation. RAF protein degradation was linked to the de-stabilization of the RAF-MAPK complex and was dependent upon the ubiquitin proteosome system. TAT-Braftide inhibited cell proliferation of two KRAS^G13D colorectal cancer human cell lines (HCT116, HCT-15).

Notably, no specific c-RAF kinase-independent mechanisms were evaluated with MCP110 or TAT-Braftide. However, given the effectiveness of these compounds in cellular models of RAS^MT cancer, it could be suggested that these compounds are negatively regulating c-RAF beyond its kinase-dependent activity. Thus, further investigation is justified.

Disruption of the c-RAF – PDE8A PPI represents another targeted approach to negatively regulating c-RAF activity [26,27]. PDE8A protects c-RAF from the surrounding cAMP microenvironment, blocking inhibitory PKA phosphorylation and enabling pro-oncogenic c-RAF activity. Targeted disruption of the c-RAF – PDE8A PPI with a cell-penetrating disruptor peptide was shown to displace PDE8A, subsequently upregulating PKA-mediated inhibitory phosphorylation of c-RAF at Ser259 (other inhibitory PKA phospho sites on c-RAF were not assessed: Ser43, Ser233, Ser621). Said disruptor was derived from the c-RAF binding epitope on PDE8A, and was later demonstrated to have a therapeutic role in overcoming BRAF inhibitor-resistant NRAS^Q61L malignant melanoma. Resultingly, efforts to evaluate the therapeutic relevance of c-RAF – PDE8A disruption in other RAS-RAF^MT cancer models is on-going. Whether disrupting the cross-talk between c-RAF and the cAMP-PKA signaling axes influences c-RAF kinase-independent mechanisms is yet to be elucidated. However, it is worth noting that disrupting the c-RAF – PDE8A complex was shown to inhibit T cell adhesion and motility; a finding that may translate in RAS^MT cancer cell models [28].

Key to stabilizing c-RAF in a closed-inactive conformation is its binding with the scaffolding protein 14-3-3. 14-3-3 dimers bind to phospho-serines on c-RAF, Ser233, Ser259 and Ser621 (e.g. following cAMP activated PKA phosphorylation), structurally constraining c-RAF in a closed-inactive conformation; occluding the catalytic site. Thus, Kenanova and colleagues set out to identify compounds capable of increasing 14-3-3-mediated PPI stabilization, including 14-3-3 – c-RAF stabilizers [29]. Utilising a disulfide tethering – FBDD (fragment-based drug discovery) approach, they developed small molecule mimetics of a known 14-3-3 binding c-RAF phospho-peptide (Q²⁵⁵RSTpSTPNVH²⁶⁴). These mimetics act as molecular glues, selectively stabilizing the c-RAF − 14-3-3 complex. Unfortunately, the activity of these mimetics in cellular models of RAS^MT cancer are yet to be evaluated and warrant investigation.

4. C-RAF dependency and RAS mutant heterogeneity

Activating RAS mutations are detected in over 20% of human cancers, predominantly occuring in three hotspot locations: G12, G13 or Q61 [30]. The pro-oncogenic regulatory behavior(s) of mutant RAS is remarkably complex, depending on which RAS family member is mutated (K/N/HRAS), what specific RAS amino acid is substituted, and what secondary mutations exist [31]. Such heterogeneity in the RAS^MT landscape will undoubtedly influence the degree to which a given cancer is dependent upon c-RAF. Therefore, systematic characterization of the relationship between RAS mutational status and c-RAF dependency is critical to the effective development of next-generation c-RAF inhibitors against appropriately stratified patient cohorts.

Utilising publicly available bioinformatic datasets available through the DepMap Portal database (https://depmap.org/portal/), further insights into RAS-RAF mutational status and relative c-RAF dependency can be made (Figure 2). Firstly, human cancer cell lines harboring a KRAS^MT, NRAS^MT or HRAS^MT are significantly more dependent upon c-RAF than RAS-RAF^wild-type (Figure 2a). This was also observed in RAS^G12 and RAS^Q61 mutant cancer cell lines, but not RAS^G13 or RAS^other (i.e. non-G12, G13 or Q61, Figure 1a). Secondly, B-RAF^V600 mutant cancer cell lines were expectedly less dependent upon c-RAF (Figure 2a). However, no significant differences between c-RAF dependency and A-RAF^MT, B-RAF^non-V600 or c-RAF^MT were observed (Figure 2a). Moreover, no correlation (Table 2) was shown between RAS-RAF mutational status and c-RAF gene expression (Figure 2b), c-RAF protein expression (Figure 2c) or c-RAF phospho-Ser338 protein expression (Figure 2d, representative of relative c-RAF kinase activity). These data not only suggest that targeted degradation of c-RAF is unlikely to prove efficacious in these RAF^MT cancer cell lines, but that relative c-RAF gene expression, protein expression and kinase activity represent unsuitable biomarkers for predicting c-RAF inhibitor sensitivity (Figure 2b-d). On the other hand, RAS^MT cancer cell lines are significantly more c-RAF dependent than wild-type (Figure 2a). However, the extensive variability within all human cancer cell cohorts (including RAS-RAF^wild-type) cannot be ignored (Figure 2a). Therefore, given the far from straight forward relationship between RAS mutational status and c-RAF dependency, it is unlikely that RAS^MT can be wielded as a standalone marker for future c-RAF inhibitor therapies.

Figure 2. Utilisation of a large heterogenous panel of human cancer cell lines (derived from > 25 cancer lineages), generated from the publicly available bioinformatic DepMap portal, to compare RAS-RAF mutational status and respective: (a) c-RAF dependency (RNAi, gene effect, DEMETER2, achilles + DRIVE + Marcotte), (b) c-RAF gene expression (public, 23Q2, Log2(TPM+1)), (c) c-RAF total protein expression (protein array, RPPA signal, Log2), and (d) phosphorylated serine 338 (pS338) c-RAF protein expression (protein array, RPPA signal, Log2). Data represented as MEAN ± STDEV. *p < 0.05; **p < 0.01; ****p < 0.001; ns, not significant vs. RAS-RAF WT (one-way ANOVA). # refers to number of cells per sub-group.

Table 2. Correlation between c-RAF RNAi (achilles + DRIVE + Marcotte, DEMETER2), c-RAF CRISPR knockout (public, 23Q2 + score, chronos), c-RAF gene expression (23Q2, Log2(TPM+1)), c-RAF total protein array (BD610151, RPPA, Log2), and c-RAF pS338 protein array (RPPA, Log2). Table represents data generated and published in DepMap Portals publicly available bioinformatic data sets. P^coeff, Pearson’s Coefficient; S^coeff, Spearman’s Coefficient; #, No. Of data points (i.e. no. Of cell lines assessed).

	c-RAF RNAi (Achilles + DRIVE + Marcotte, DEMETER2)			c-RAF CRISPR (Public, 23Q2 + Score, Chronos)
	P^coeff	S^coeff	P-value	P^coeff	S^coeff	P-value
c-RAF RNAi (Achilles+DRIVE+Marcotte, DEMETER2)				0.517	0.447	4.56E–37 #535
c-RAF CRISPR (Public, 23Q2+Score, Chronos)	0.517	0.447	4.56E–37 #535
c-RAF Gene Expression (23Q2, Log2(TPM+1))	−0.107	−0.126	6.68E–3 #642	−0.084	−0.104	7.47E–3 #1019
c-RAF Protein Array (BD610151, RPPA, Log2)	−0.073	−0.094	8.51E–2 #555	−0.069	−0.086	9.06E–2 #606
c-RAF pS338 Protein Array (RPPA, Log2)	−0.192	−0.182	5.44E–6 #555	−0.137	−0.151	7.43E–4 #606

5. Concluding remarks

The need for novel approaches to effectively exploiting pro-oncogenic c-RAF activity in RAS^MT cancer, beyond simply its catalytic activity, is clearer than ever. Selectively targeting pro-oncogenic c-RAF PPIs, and/or inducing the degradation of c-RAF completely, represent increasingly attractive therapeutic strategies in the context of RAS^MT – c-RAF-dependent cancer. However, succesful development of these next-generation c-RAF inhibitors are wholey dependent upon accurate characterization of the complex RAS^MT – c-RAF relationship. Unlocking key details in this relationship will not only allow for identification of actionable c-RAF-associated vulnerabilities in RAS^MT cancer, but guide future synergistic combination strategies capable of overcoming challenges associated with therapeutic resistance; including that observed with current RAS^MT-specific small molecule inhibitors. When considering c-RAF as a target in RAS^MT cancer, this time – the goal posts appear to of widened.

Declaration of interest

C Blair is the co-founder of Disruptyx Therapeutics Ltd. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contribution

C Blair wrote the article, analyzed the bioinformatic data and made the figures/tables.

S Cooke supported C Blair in writing the article and making the figures/tables.

Acknowledgments

We aknowledge DepMap Portal, a publicly available bioinformatic database, containing in-depth polyomic characterisation of a large heterogenous library of human cancer cell lines. Data from this database was yielded to support the narrative of the manuscript submitted.

Data availability statement

All data is available within article figures, tables and supplementary files.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/14737140.2024.2319035

Additional information

Funding

This paper was funded by the UK Research and Innovation > Medical Research Council [MR/X502807/1].