Epigenomic Inactivation of RasGAPs Activates RAS Signaling in a Subset of Luminal B Breast Cancers

The RAS family genes HRAS, KRAS, and NRAS encode small GTPase signaling proteins and are among the most frequently mutated oncogenes in human cancer (1). Their activation feeds downstream signaling pathways that control cell proliferation, metabolism, and invasive cell behavior, and the task of “drugging” mutant RAS proteins has become a national effort. Activating RAS mutations are found in on average 30% of human tumors, but only approximately 1% of breast tumors carry a RAS mutation. However, high RAS activity is detected in about 50% of breast cancer cell line and tumor samples (2, 3). As depicted in Fig. 1, this can be due to elevated signaling of receptor tyrosine kinases, including HER2 and EGFR, that activate guanine-nucleotide exchange factors (GEF) such as SOS to increase RAS GTP binding or loss of negative regulator GTPase-activating proteins (GAP) that accelerate the hydrolysis of GTP to GDP (4). A prototypic RasGAP, the neurofibromatosis type 1 gene, NF1, is lost in approximately 3% of breast tumors and is associated with constitutive activation of RAS and sensitivity to MEK inhibition (5).

Now, Olsen and colleagues have uncovered two RasGAPs that are frequently epigentically silenced in a subset of the aggressive luminal B subtype of breast cancer (6). Breast cancer is currently subdivided into histologic and molecular subtypes based on estrogen/progesterone hormone receptor expression, HER2 receptor tyrosine kinase expression, and gene expression classifiers (7). The most common form of breast cancer is estrogen receptor–positive (ER⁺), molecularly classified as luminal, and this subtype generally represents the best prognosis. However, within ER⁺ tumors, the luminal B subtype is more aggressive and less responsive to endocrine therapy, with a risk of relapse similar to HER2⁺ and triple ER/PR/HER2-negative basal classified tumors. Mechanisms underlying the aggressive behavior of luminal B tumors are poorly understood, and clinical histologic markers to inform care are primarily based on an increased proliferative index. Thus, a better understanding of what drives this lethal form of ER⁺ breast cancer and a more refined clinical biomarker could lead to more sophisticated identification and treatment of these at-risk patients.

Olsen and colleagues examined the expression of the 14 RasGAPs in the human genome and discovered that the DAB2IP RasGAP, previously described to be a tumor suppressor in prostate cancer, is suppressed selectively in luminal B breast cancer due to promoter hypermethylation (8). Importantly, they showed luminal B breast cancers could be stratified according to relapse-free survival based on DAB2IP expression. Specifically, patients with tumors showing reduced DAB2IP expression experienced significantly shorter relapse-free survival. Although infrequent, deleterious mutations in DAB2IP were also identified in breast cancer in two functional domains, the catalytic RasGAP domain and the period-like domain, previously shown to negatively regulate NF-κB activity through binding TRAF2 and preventing TRAF2 activation of IkB kinase (IKK). The latter phosphorylates the inhibitor of NF-κB (IkB) to release free active NF-κB to the nucleus (9). Olsen and colleagues also performed gain-of-function and loss-of-function studies in luminal cell lines to show that DAB2IP controls anchorage-independent growth dependent on the RasGAP catalytic domain but not on proliferation, and that this involves regulation of multiple RAS family members.

Intriguingly, Olsen and colleagues found that a second RasGAP, RASAL2, is frequently silenced in the same luminal B breast cancers as DAB2IP. This is remarkable because molecules with the same function usually are mutually exclusively altered in individual tumors. Instead, the authors found in The Cancer Genome Atlas datasets that nearly half of the 46% of luminal B tumors that had reduced DAB2IP expression also had reduced RASAL2. These statistics were confirmed by immunohistochemistry on 63 primary luminal B human tumors. They also found that luminal B tumors with low levels of both RASAL2 and DAB2IP more frequently presented at stage III/IV, whereas tumors with high levels presented at stage I. This striking cooperativity suggests complementary activities of these two RasGAPs in the aggressive behavior and poor outcome of luminal B tumors. Olsen and colleagues explored mechanisms by which DAB2IP and RASAL2 might cooperate to drive tumor progression. They found through studies of human tumor xenografts that loss of each gene individually promoted tumor growth and that this was not augmented by combined inactivation. Because coinactivation of these genes in human tumors was associated with late-stage disease, the authors tested the effects of their loss on aggressive behavior, including epithelial–mesenchymal transition (EMT) and invasion assays. In this case, concomitant loss dramatically increased the invasion of human mammary epithelial cells and luminal B cancer cells. Further, only the combined loss of RASAL2 and DAB2IP resulted in robust EMT marker expression. Consistent with the role of EMT in metastasis, both loss-of-function and gain-of-function studies showed a striking role for the combined activity of these proteins in regulating metastasis, where loss of both RASAL2 and DAB2IP dramatically increased metastasis and reconstitution of both completely prevented metastasis in xenograft experiments. In addition, knockout of Rasal2 in a luminal B mammary tumor mouse model enhanced metastasis, and 60% of primary tumors from this model spontaneously lost DAB2IP in association with increased metastasis. When DAB2IP was not lost in the primary tumor, where examined, it was selectively lost in the metastasis.

Olsen and colleagues then explored why loss of two RasGAPs with presumed overlapping functions would cooperate to this degree in the metastatic propensity of luminal B tumors. The authors first looked at RAS signaling and observed that suppression of each RasGAP activated KRAS, HRAS, ERK, and AKT, but loss of both substantially enhanced activation. They found that increased MEK/ERK and PI3K/AKT signaling were critical for invasion triggered by DAB2IP/RASAL2 loss, which could be rescued by wild-type DAB2IP or RASAL2, but not by RasGAP dead mutants. Intriguingly, the induction of EMT by DAB2IP/RASAL2 loss did not involve these pathways, or the catalytic GTPase-activating activity. This observation led the researchers to the novel observation that DAB2IP uniquely contains a period-like domain that binds TRAF2 and suppresses NF-κB activity (8). Their analysis of the role of this second DAB2IP function in luminal B breast cancer demonstrated that DAB2IP loss activated NF-κB whereas RASAL2 loss did not, but combined loss further activated NF-κB. This is consistent with RAS activity feeding into NF-κB signaling (10) via a DAB2IP-dependent mechanism in which the period-like domain-mediated inhibition of NF-κB is dominant. Moreover, the activation of NF-κB by combined loss of DAB2IP and RASAL2 was reversed by wild-type or GTPase catalytically dead DAB2IP, but not by DAB2IP with a deleterious mutation in the period-like domain. Finally, the authors show that it is the NF-κB activation resulting from loss of DAB2IP and RASAL2 that drives the EMT phenotype. Consistent with this finding, expression of an IkBα super suppressor that potently inactivates NF-κB reduced metastasis in the DAB2IP- and RASAL2-deficient luminal B mouse model.

Overall, these studies suggest that invasion and metastasis of a subset of aggressive luminal B breast cancers is driven by the concomitant inactivation of the RasGAPs DAB2IP and RASAL2. Inactivation of both proteins increases RAS activity and drives invasion, and inactivation of DAB2IP specifically promotes NF-κB–mediated EMT. This study suggests biomarkers that enable identification of a RAS-driven subpopulation of luminal B cancers that may be effectively controlled by RAS pathway–targeted therapies. This is a timely discovery given the effort now under way at Frederick National Laboratory for Cancer Research to accelerate the development of RAS pathway–targeted therapies (11). Both DAB2IP and RASAL2 function as tumor suppressor genes, subject to epigenomic downregulation. This suggests the possibility that epigenomically targeted drugs and/or drugs that target the NF-κB pathway might be deployed to improve overall survival in aggressive luminal B tumors. The study also suggests the importance of assessing expression of RasGAP proteins in a broader range of aggressive human tumors.

No potential conflicts of interest were disclosed.

This work was supported by NIH grants U54 CA112970 (J.W. Gray) and R01 CA196228 (R. Sears).