Beyond growth signaling: apoptotic sensor MERTK activates AKT by a novel mechanism

ABSTRACT

Canonically the oncogenic kinase AKT is activated by growth signals. Our work suggests apoptotic materials, abundant in tumors, also contribute to AKT activation by stimulating MERTK that in turn phosphorylates Y26 in the AKT PH domain. pY26 reverses binding of an AKT endogenous, WW-domain containing inhibitor, SAV1, allowing AKT responsiveness to classic growth signals. This novel mechanism may contribute to drug resistance.

RCC (Renal cell carcinoma), accounting for ~90% of kidney cancer with a high-mortality rate due to its resistance to chemotherapies or radiotherapies. Genetic analyses indicate that loss-of-function mutations of key genes are tightly associated with RCC pathogenesis, including VHL (von Hippel-Lindau), PBRM1 (protein polybromo-1), BAP1 (BRCA1-associated protein 1 or ubiquitin carboxyl-terminal hydrolase), SETD2 (set domain containing protein 2), PTEN (phosphatidylinositol 3,4,5-trisphosphate 3-phosphatase and dual-specificity protein phosphatase) and others.¹ For example, VHL loss/mutation contributes to the vast majority of sporadic ccRCC (clear cell renal cell carcinoma) by elevating HIF signaling and vasculature signatures. In addition, p53 (Tp53, best known as p53) and PTEN mutations are the most significantly mutated genes in all RCC subtypes; consistent with that loss-of-function of these two tumor suppressors contributes to cancer progression. PTEN dephosphorylates PI(3,4,5)P₃, thus antagonizing PI3K (phosphatidylinositol 3-kinase) and promoting PI3K/mTOR/AKT signaling, which is indispensable for tumor cell growth, survival and metabolism. AKT is highly conserved and its activation is tightly controlled via a multi-step process. Oncogenic activation of the PI3K/AKT pathway is observed in virtually all solid tumors and is considered a candidate for targeted therapy. However, inhibitors targeting AKT signaling have shown toxicity in clinical trials,² suggesting that directly targeting AKT may not be an ideal option. To this end, identification of cancerous upstream AKT hyperactivation mechanisms may facilitate designs of new therapies to suppress PI3K/AKT signaling caused by PTEN inactivation and other RCC oncogenic signaling.

To search for additional AKT binding proteins we reasoned that the two pathways that control cell size and cell number (AKT and Hippo), may have common elements. We examined Hippo pathway constituents discovering that one Hippo signaling component, SAV1 (protein salvador 1), binds and suppresses AKT activation in RCC.³ Specifically, the WW domain of SAV1 binds a proline motif in the PH (pleckstrin homology) domain of AKT and suppresses AKT activation, an action independent of SAV1’s function in Hippo signaling. Our results demonstrate that SAV1 binding to AKT-PH domain impedes the plasma membrane attachment as well as AKT binding to its upstream activating kinases such as PDK1 and mTORC2 (mechanistic target of rapamycin complex 2). Our sense that this was important was heightened when we discovered that some cancer patients harbor SAV1-WW domain mutations. When we engineered these cancer-relevant mutated SAV1 molecules, they were deficient in binding AKT, and led to AKT hyperactivation facilitating RCC growth.³ Another SAV1 linkage was shown in that SAV1 can bind directly to the protein phosphatase PP2A (protein phosphatase 2A) suppressing its phosphatase activity. SAV1 is also co-purified with the protein phosphatase PP1A and both of these phosphatases are linked with AKT dephosphorylation through direct or indirect mechanisms⁴ Thus, SAV1 binding may bring PP1A/PP2A to dephosphorylate AKT-pT308 and pS473, which warrants further investigations. Moreover, in RCC, the HOTAIR lncRNA directly binds SAV1 promoting Hippo activation,⁵ which might provide an additional intersection between the two pathways.

The “₂₄PxY₂₆” motif in AKT-PH domain that mediates SAV1 binding is evolutionarily conserved, but how is it controlled? We observed that AKT1-Y26 phosphorylation attenuates SAV1 binding. Thus, we tested several canonical receptor tyrosine kinases and they were inefficient or unable to phosphorylate Y26; however, activation of MERTK produced pY26 and led to release of SAV1 binding, thus activating AKT.³

MERTK is a member of the TAM (TYRO3, AXL and MERTK) family of RTKs (receptor tyrosine kinase). The physiological roles for MERTK activation in macrophage and other myeloid cells have been extensively studied, including promoting rapid and efficient clearance of phosphatidyl serine (PtdSer) exposed on apoptotic cells and exosomes. In this process MERTK also signals to the transcriptional machinery to suppress inflammatory M1 cytokines⁶ and promote polarization to an M2 anti-inflammatory phenotype.⁷ It detects PtdSer in additional physiologic processes such as the second phase of platelet aggregation or the pruning of misaligned axons in neurodevelopment. The linkage bridged between the exposed lipid (PtdSer) and TAM RTKs occurs through a Gla domain containing ligand such as GAS6 (growth arrest specific 6) or PROS (protein S). This normal TAM RTK PtdSer sensing system can be subverted by pathogens such as viruses (Ebola, Zika) that expose PtdSer on their surface or in tumors where the abundance of apoptotic material leads to a MERTK-dependent, immunosuppressive myeloid cell infiltrate (refer to⁸ for review).

Other roles for MERTK (and other TAM RTKs) in cancer are increasingly being recognized and targeted. Overexpression of MERTK has been observed in multiple types of solid tumors (eg. melanoma and head-and-neck cancer), as well as hematological malignancies including ALL and AML (see⁸). Chemical MERTK inhibitors are being developed at UNC both to probe MERTK as a cancer and immune system target, to perform proof-of-principal preclinical cancer therapeutics,⁹ as well as for potential human clinical trials. Mechanistically, MERTK overexpression/activation leads to activation of a handful of oncogenic signaling pathways including PI3K/AKT, MAPK/ERK, JAK/STAT and NFκB, through the usual pathways activated by other RTKs, autophosphorylation sites attracting the signaling SH2 (src homology 2) domain containing proteins. We found that MERTK-mediated AKT1-Y26 phosphorylation impedes SAV1 binding and predisposes AKT for plasma membrane recruitment, providing a mechanism by which MERTK can govern AKT activation- a release of suppression rather than an induced-activation mechanism. Surprisingly, this action of MERTK is not re-capitulated robustly by other RTKs tested nor inhibited by dasatinib, a very broad-spectrum tyrosine kinase inhibitor. Axl has a minimal level of pY26 activity and TYRO3 has even less.

Deficiency in AKT1-Y26 phosphorylation led to retarded tumor growth in vitro and in vivo due to enhanced SAV1 binding and subsequent AKT inactivation. More importantly, in this model, MERTK inhibitors rely on SAV1 for growth inhibitory function – depletion of endogenous SAV1 leads to reduced sensitivity of RCC cells to MERTK inhibitors in vitro. On the other hand, low SAV1 levels sensitize RCC cells to allosteric AKT inhibitors (targeting AKT-PH domain for inhibition) due to increased availability of AKT-PH domain. Thus, SAV1 expression levels may serve as one of the prediction markers of RCC cell sensitivity to either MERTK inhibitors or AKT allosteric inhibitors.

As noted above the physiological stimuli for MERTK are cellular apoptotic materials including phosphatidylserine (PtdSer), and the linking ligands GAS6 or PROS.⁸ Canonically, activation of macrophage MERTK serves as an apoptotic cell clearance mechanism. Now from our studies it seems that apoptotic cellular materials, which are known to activate MERTK in cancer cells, have an additional mechanism, namely enhancing activation of AKT signaling and leading to survival and drug resistance. Considering that in a solid tumor like RCC, drug gradients are commonly observed, activation of the MERTK/AKT signaling in cancer cells close to apoptotic cancer cells, produced by the hypoxic environment or triggered by drug treatment, will provide an additional survival advantage for nearby viable cancer cells through activating the MERTK/AKT signaling. This is yet another potential mechanism for drug resistance.

The use of PI3K inhibitors to treat cancer is so logical given to the overwhelming number of cancers that have mutations activating PI3K or eliminating AKT off-switches. However, the clinical applications to date have been disappointing due to a less than optimal toxicity to efficacy ratio. Recent use of a PI3K inhibitor has shown promise with a caveat; efficacy may be reduced in patients with higher Body Mass Index (BMI) presumably due to hyperinsulinism, which can stimulate and overcome the PI3K inhibition. In addition, SGLT2 (sodium/glucose cotransporter 2) inhibition or the ketogenic diet also released single PI3Kα inhibitor induced resistance.¹⁰ Our work shows an unexpected mechanism which might also affect therapy aimed at this pathway; an intratumorally apoptotic cell, PtdSer stimulated, MERTK-dependent pathway that increases the ability of other oncogenic pathways to keep AKT activated. If this is correct, a combination therapy with MERTK and low-dose AKT active-site but not allosteric inhibitors may overcome drug resistance (Figure 1).

Figure 1.

Mechanism for therapy-resistance driven by MERTK-mediated AKT activation in solid tumors. Similar to oxygen levels, the drug concentration displays a dose gradient away from blood stream in solid tumors. Higher concentrations of drugs would be more efficient in eradicating cancer cells. Resulted apoptotic cancer cells will release apoptotic materials including PtdSer (phosphatidylserine), GAS6 (growth arrest specific 6) and PROS (proteinS) that would trigger MERTK activation in nearby cancer cells to facilitate AKT activation, leading to enhanced cellular survival potential that may contribute to drug resistance in these cancer cells. Thus, MERTK-mediated AKT activation mechanism may create a local drug resistant environment.

We have identified a new signaling cascade governed by a release of SAV1 suppression on AKT-PH domain triggered by MERTK-mediated AKT1-Y26 phosphorylation. This suggests yet another path in RCC to facilitate AKT activation. Perhaps most surprisingly, the result demonstrates that although extensively studied for 25 years, additional layers of AKT controls are still “out there”. And since this pathway is activated by mutations at every level and in almost all cancers, further investigations are needed if we are to overcome therapeutic resistance by targeting the PI3K/AKT pathway.

Funding Statement

This work was supported by the NIH grants (P.L. R00CA181342, H.S.E. CA016086), the V Scholar Research Grant (P.L. V2018-009), the vwr Charitable Foundation Grant (P.L.) and UNC University Cancer Research Fund (P.L.).

Acknowledgments

We sincerely apologize to colleagues whose important work cannot be cited in our manuscript due to space limits. We thank Liu and Earp lab members for critical reading of the manuscript and helpful discussions.

Disclosure of Potential Conflicts of Interest

H.S.E. is a founder of Meryx (a UNC start up) that is developing small molecule inhibitors for MERTK. H.S.E. owns stock in Meryx. The remaining authors declare no competing interests.