SOS signaling in RAS-mutated cancers

June 29, 2020, by Erin Sheffels and Rob Kortum

RAS-mutated tumors were originally thought to proliferate independently of upstream signaling inputs, but we now know that receptor tyrosine kinase-dependent activation of both mutant RAS and non-mutated wild-type (WT) RAS plays an important role in modulating downstream effector signaling and driving therapeutic resistance in RAS-mutated cancers.  The contributions of wild-type RAS to proliferation and transformation in RAS-mutated cancer cells places renewed interest in upstream signaling molecules, including the RasGEFs SOS1 and 2, as potential therapeutic targets in RAS-mutated cancers.

RAS isoforms have a hierarchy of abilities to activate RAS effectors

Mutant RAS-dependent transformation requires both Raf/MEK/ERK and PI3K/AKT effector pathway activation.  However, while HRAS, NRAS, and KRAS can all interact with PI3K and RAF, a series of seminal papers showed that they activate these effectors to different extents, such that there is an inverse relationship in their ability to activate Raf and PI3K: mutant HRAS is a potent activator of PI3K but a poor activator of RAF, and conversely KRAS is a potent activator of Raf but a poor activator of PI3K (1-3).  We are beginning to understand the mechanism for the differential activation of RAF proteins.  The Morrison laboratory recently showed that BRAF preferentially interacts with KRAS via an interaction between the KRAS(4B) polybasic region and an acidic N-terminal region in BRAF (4).  The ability to directly associate with both BRAF and CRAF makes KRAS a more potent activator of the RAF/MEK/ERK cascade.  While the precise mechanism for differential PI3K activation between HRAS and KRAS remains unclear, a major contributor seems to be the polybasic stretch in the hypervariable region of KRAS; mutating basic residues in the KRAS(4B) HVR inhibits Raf/MEK/ERK signaling but enhances PI3K/AKT phosphorylation (5).  These differences in activation abilities impact the dependence of RAS-mutated cancers on upstream signals. For example, PI3K/AKT pathway activation is dependent on RTK signaling in KRAS-mutated colorectal (6) and lung (7) adenocarcinoma cells. A potential role for the WT RAS isoforms is to activate the effector pathways that mutant RAS does not strongly activate, making the cellular outcome a product of signaling by both WT and mutant RAS.  

Mutant RAS can activate WT RAS via SOS

Mutant RAS can activate WT RAS independently of RTK input by at least two interdependent mechanisms.  First, SOS1 can be allosterically activated by RAS, allowing increased activation of WT RAS. When assessing the crystal structure of SOS1, the Kuryian and Bar-Sagi labs found an allosteric RAS^GTP binding pocket distinct from the SOS1 catalytic domain that, when occupied, relieves SOS1 autoinhibition (8).  This RAS^GTP binding increases SOS1 catalytic activity by up to 500-fold, setting up a RAS^GTP−SOS1−WT RAS positive feedback loop that allows for processive localized WT RAS activation at the plasma membrane. Further downstream, PI3K/AKT signaling can phosphorylate eNOS, which can nitrosylate and activate WT HRAS (9). RTK signaling can also activate WT RAS independently from mutant RAS. The McCormick laboratory built on previous work to show that canonical RTK-dependent WT RAS activation supplements basal signaling from mutated RAS to promote proliferation in RAS-mutated tumor cell lines, and combined inhibition of WT and mutated RAS is required to induce cell killing (3, 10).  These mechanisms are not mutually exclusive and may cooperate in some contexts (11-13).

These models all indicate that SOS plays a role in activating WT RAS in mutant RAS cancers. Data from our lab and others suggests that SOS1 and SOS2 may play non-overlapping roles to promote WT RAS activation in RAS mutated tumor cells.  For SOS1, allosteric signaling and RTK-dependent activation are both important for KRAS-mutated cancer cells depending on the cellular context:  SOS1 is required for WT HRAS and NRAS activation in an animal model of KRAS-induced leukemia (14), mutant KRAS−SOS1−WT RAS allosteric signaling promotes growth of KRAS mutant pancreatic cancer cell xenografts (15), and both allosteric signaling and EGFR-SOS1 signaling contribute to growth of KRAS-mutated colorectal cancer cells (16).  In contrast, we found that RTK−SOS2−WT RAS signaling, but not allosteric SOS2 activation, is a critical mediator of PI3K signaling in the context of mutant RAS (17) and protects KRAS-mutated cancer cells from anoikis (18).  

SOS proteins as therapeutic targets in RAS-mutant cancers

In KRAS-mutated cancer cells, single agent MEK inhibitor treatment is ineffective because it relieves ERK-dependent negative feedback signaling, enhancing RTK-SOS-WT RAS signaling to the Raf/MEK/ERK and PI3K/AKT pathways and leading to therapeutic resistance (19-22).  Similar relief of negative feedback signaling drives rapid resistance to KRAS^G12C inhibitors (23, 24).  In both cases, this resistance is driven by multiple RTKs. CRISPR screens revealed that both KRAS^G12C inhibitors (25) and MEK inhibitors (26) require either broad inhibition of proximal RTK signaling or targeting of PI3K/mTOR survival signaling to enhance their efficacy and delay therapeutic resistance.  Recent pre-clinical studies showed that co-treatment with allosteric SHP2 inhibitors can overcome both KRAS^G12C (23, 24) and MEK (27, 28) inhibitor resistance, leading to more durable responses.  Furthermore, we found that SOS2 deletion inhibited RTK-WT RAS-PI3K signaling and synergized with MEK inhibitors in KRAS mutated cell lines (17).

While there are currently no SOS2-specific inhibitors, Bayer Pharmaceuticals published a SOS1 inhibitor suitable for in vitro studies (29).  Furthermore, Boehringer Ingelheim has developed orally available SOS1 inhibitors (30) and started recruiting patients with advanced KRAS-mutated solid tumors in 2019 for a Phase 1 clinical trial (NCT04111458).  SOS1 inhibition is mechanistically most similar to SHP2 inhibition (31), suggesting that SOS1 inhibition could similarly enhance the efficacy of KRAS^G12C- and MEK-inhibitors.  Indeed this appears to be true for combined SOS1/MEK inhibition, as preliminary data from Boehringer Ingelheim showed marked cooperativity between SOS1- and MEK-inhibition in multiple G12 and G13 KRAS-mutated PDX models (30).  Furthermore, since KRAS^G12C allosteric inhibitors can only bind KRAS^GDP, inhibiting SOS1 has the potential advantage of directly enhancing the efficacy of KRAS^G12C inhibitors by increasing the amount of mutant KRAS^G12C accessible to drug (29), in addition to inhibiting feedback activation of WT RAS.  While further studies are required, the possibility of inhibiting SOS1 has enormous clinical potential as a combination therapy.

WT RAS signaling is an important modifier of KRAS-mutated oncogenesis, and inhibition of WT RAS signaling may be required for effective treatment of KRAS-mutated cancers.  Understanding the mechanisms by which the ubiquitously expressed RasGEFs SOS1 and SOS2 promote WT RAS activation is an important step in determining the best ways to limit WT RAS signaling.  The ability to pharmacologically manipulate SOS1/2 signaling may lead to optimized therapeutic combinations that can be used to treat KRAS-mutated cancers.