Rictor/mTORC2 Drives Progression and Therapeutic Resistance of HER2-Amplified Breast Cancers

Akt phosphorylation drives cell survival, motility, and proliferation in several diseases, including cancers, causing increased interest in upstream regulators of Akt. Many cancer studies focus on PI3K-to-PDK1 signaling as the primary Akt activation pathway, via Akt T308 phosphorylation. PI3K inhibition reduces P-Akt T308, decreasing Akt signaling to its downstream effectors, including mTOR complex 1 (mTORC1). In many cancers, PI3K/Akt/mTORC1 inhibition decreases tumor cell growth and survival. However, mTORC1 inhibition releases endogenous restraints upon the PI3K pathway, causing resurgent PI3K/Akt signaling and dampening net treatment response (1). A greater understanding of signaling pathways both upstream and downstream of Akt is required for effective Akt inhibition in cancers.

In addition to PI3K-mediated Akt T308 phosphorylation, Akt S473 phosphorylation activates Akt. Several serine-threonine kinases can phosphorylate Akt S473, including inhibitor of κB-kinase (IKK)-α, DNA protein kinase (DNA-PK), integrin-linked kinase (ILK), and mTOR complex 2 (mTORC2; ref. 2). Interestingly, this places the serine/threonine kinase mTOR in a position to activate Akt (mTORC2) and to be activated by Akt (mTORC1; ref. 3). The two structurally distinct complexes of mTOR are defined by the mTOR-associated cofactors. Raptor is a required cofactor for mTORC1, while Rictor and Sin1 are required for mTORC2. Functional distinctions are also activated downstream of PI3K/Akt and regulate cell growth, protein translation, and metabolism (4), which controls cell polarity and cytoskeletal dynamics (5). mTORC2 lies upstream of Akt, directly phosphorylating Akt S473 (6, 7). Given that mTORC1 and mTORC2 utilize distinct cofactors and substrates, mTORC1 and mTORC2 may have distinct physiologic roles, and their dysregulation may produce distinct pathologic consequences. This notion is supported by recent findings that Rictor/mTORC2, but not Raptor/mTORC1, is required during mammary gland development for ductal branching, mammary epithelial cell (MEC) motility, and MEC survival (8).

Breast cancers often hijack many signaling pathways used by normal MECs to support tumor cell growth, survival, and metastasis (9, 10). It is possible, therefore, that requirements for mTORC2 in MECs are paralleled in breast cancers. However relatively little is known about distinct roles of mTORC2 in breast cancer formation, progression, and treatment, despite its known role in activation of Akt, a key signaling node and critical effector of RTKs, including HER2. We used genetic and pharmacologic models of Rictor/mTORC2 blockade to determine whether Rictor/mTORC2 supports spontaneous tumor formation, tumor cell survival, and therapeutic response to HER2 inhibition. Our study uncovers a previously unreported role for Rictor/mTORC2 in HER2-amplified breast cancers.

Human tissue microarray (deidentified breast carcinoma specimens with normal tissue controls) was purchased from Cybrdi (#CC08-10-001) and stained with antibodies against Rictor or Raptor. Average positive staining expressed in percentage and the intensity were calculated for each core. For mouse studies, tumors were resected from mice and paraffin sections (5 μm) were stained with hematoxylin and eosin or with the ApopTag TUNEL Analysis Kit (Calbiochem). IHC on paraffin-embedded sections or on human tissue microarrays (TMA) was performed as described previously (11) using: Rictor (Santa Cruz Biotechnology), Raptor (Abcam), Ki67 (Santa Cruz Biotechnology), P-S6 (Cell Signaling Technology); P-Akt S473 (Cell Signaling Technology); Rictor (Santa Cruz Biotechnology). Immunodetection was performed using the Vectastain Kit (Vector Laboratories), AF488-conjugated anti-rabbit, or AF621-conjugated anti-mouse (Life Technologies), according to the manufacturer's directions.

All animals were housed under pathogen-free conditions, and experiments were performed in accordance with The Association for Assessment and Accreditation of Laboratory Animal Care guidelines and with Vanderbilt University Institutional Animal Care and Use Committee approval. Rictor^FL/FL mice (12) were kindly provided by Dr. Mark Magnuson (Vanderbilt University, Nashville, TN) and were inbred to FVB for >10 generations. MMTV-NIC mice (generated in FVB) have been previously described (13). All analyses of Rictor^FL/FL X MMTV-NIC mice were performed on age-matched siblings.

BT474, MDA-MB-361, and SKBR3 cells were purchased in 2012 from ATCC (cell identity verified by AACR using genotyping with a Multiplex STR assay) and cultured at low passage in DMEM with 10% FCS. MCF10A and MCF10A-Rictor^ZFN cells (Sigma-Aldrich) were cultured in DMEM:F12 plus insulin (4 μg/mL), cholera toxin (1 μg/mL), EGF (100 ng/mL), hydrocortisone (2 μg/mL), and 5% horse serum and transduced with lentiviral HER2-IRES-RFP (GenTarget) and selected with 10 μg/mL blasticidin. AZD5363 and lapatinib were purchased from SelleckChem. Lapatinib-resistant cell lines were previously described (14).

Lentiviral shRNA-encoding plasmids (Rictor shRNA #1853, 1854; Raptor shRNA #1857, 1858; scramble shRNA #1864, Addgene) were transfected into 293FT cells plus packaging vectors. Cultured media containing viral load were used to infect cells. Cells were selected and maintained at low passage with puromycin (2 μg/mL).

Cells were homogenized in ice-cold lysis buffer [50 mmol/L Tris pH 7.4, 100 mmol/L NaF, 120 mmol/L NaCl, 0.5% NP-40, 100 μmol/L Na₃VO₄, 1X protease inhibitor cocktail (Roche)] and cleared by centrifugation (4°C, 13,000 × g, 10 minutes). Protein concentration was determined using BCA Assay (Pierce). Proteins separated by SDS-PAGE were transferred to nitrocellulose membranes. Membranes were blocked and probed with antibodies as described previously (10) using primary antibodies α-actin (Sigma-Aldrich), and Rictor (Santa Cruz Biotechnology), and the following from Cell Signaling Technology: phospho-cocktail; AKT, P-Akt S473, P-Akt T308, S6, P-S6, and Raptor.

β-Estradiol pellets (0.17 mg; 14-day release; Innovative Research of America) were implanted subcutaneously into 4-week-old female BALB/c athymic nu/nu mice (Harlan Laboratories). Right and left inguinal mammary fat pads were injected with 10⁶ MDA-MB-361 shScramble or shRictor cells in 100-μL Matrigel. Tumors were measured with calipers twice weekly.

Cells (5,000) seeded in 12-well plates were cultured up to 14 days. Growth media (with 1 μmol/L lapatinib, PP242, or equal volumes of DMSO) was changed every 3 daya. Cells were washed with ice-cold PBS, fixed with 10% formalin, stained with crystal violet, imaged, and quantitated using the Odyssey fluorescence scanner and Odyssey software. For three-dimensional growth assay, 5,000 cells were suspended in 100-μL Matrigel and plated in 96-well plates. After Matrigel solidified, 100-μL growth media (supplemented with 1 μmol/L lapatinib, PP242, or DMSO) was layered over cultures and changed every 3 days. Colonies were photographed after 14 days.

Cells (5,000) were plated in 96-well plates in triplicate and cultured overnight. Annexin V-FITC (5 μL; Invitrogen, 807876) in 95 μL complete or serum-free DMEM was added to each well. Live cell fluorescence was captured at 6 hours and quantitated using ImageJ.

Cells (5,000) were plated in 96-well plates and cultured overnight. Caspase-Glo 3/7 reagent (Promega) was added (1:1 ratio) and incubated for 1 hour and luminescence was measured.

Experimental groups were compared with a control group using Student unpaired, two-tailed t test. Multiple groups were compared across a single condition using one-way ANOVA. To compare the response of two agents combined to either single agent alone, two-way ANOVA was used. P < 0.05 defined significant differences from null hypothesis.

Nearly 20% of all breast cancers exhibit HER2 gene amplification, driving PI3K signaling and Akt phosphorylation. Notably, HER2-amplified breast cancers rely on PI3K/Akt signaling. Because Akt is a critical signaling effector of HER2 in HER2-amplified breast cancers, we examined AKT1/2/3 mRNA and/or upregulation of P-Akt S473 in a panel of HER2-enriched breast tumor specimens curated by The Cancer Genome Atlas (TCGA; Fig. 1A). Overexpression (>1.5 SD) of AKT1, AKT2, or AKT3 and/or P-Akt S473 upregulation occurred more frequently in HER2-enriched tumors (25/58, or 43%) as compared with luminal A/B breast cancers (20%, or 64/321 tumors) and basal-like breast cancers (33%, or 26/81 tumors, Fig. 1B and Supplementary Fig. S1A and S1B). AKT-activating mutations and amplifications were rarely seen in HER2-enriched breast tumors, including one AKT1 E17K mutation, one AKT2 amplification, and four AKT3 amplifications within the 58 tumors reported. Patients whose HER2-enriched tumors exhibited AKT1/2/3 mRNA or P-Akt S473 upregulation displayed substantially reduced overall survival (OS) as compared with the remaining HER2-enriched breast cancer patients (Fig. 1C, P = 4.999e−5, log-rank test). In contrast, AKT1/2/3 mRNA and/or P-Akt S473 upregulation did not correlate with OS in patients with luminal A/B or basal-like breast cancers (Supplementary Fig. S1A and S1B).

Activating phosphorylation of Akt occurs at T308 and S473. While the importance of PI3K-mediated Akt T308 phosphorylation in HER2-amplified breast cancers is clear, less is known about how kinases governing Akt S473 phosphorylation participate in the genesis, maintenance, progression, and treatment of HER2-amplified breast cancers. Consistent with a potential role for mTORC2 in HER2-amplified breast cancers, RICTOR mRNA levels assessed by qRT-PCR revealed elevated RICTOR expression in 3 of 3 HER2-amplified breast cancer cells assessed, as compared with what was seen in untransformed MCF10A mammary epithelial cells (MECs, Fig. 1D). RICTOR expression was higher in 2 of the 3 HER2-amplified breast cancer cells as compared with luminal (N = 4) or basal-like (N = 2) breast cancer cells that were tested. We confirmed these findings using Western blot analysis of whole-cell lysates, revealing Rictor upregulation in 4 of 4 HER2-amplified breast cancer cell lines over untransformed MCF10A (Fig. 1E).

We used IHC to assess the mTORC2 obligate cofactor Rictor in clinical breast cancer TMAs. Rictor was expressed below the level of detection in normal breast and ductal carcinoma in situ specimens (Fig. 1F and Supplementary Fig. S2A; Supplementary Table S1), in contrast to our previous findings that Rictor/mTORC2 (but not Raptor/mTORC1) is required for mammary ductal branching and MEC survival during puberty. This may indicate that Rictor expression is higher during dynamic architectural changes in the mammary epithelium, such as during puberty, but its expression may be lower in mature, quiescent MECs, although this hypothesis has not been tested. Rictor was significantly upregulated in invasive breast carcinomas (Fig. 1G), 37% of which scored positive for Rictor staining (Supplementary Table S1). Rictor was localized in the cytoplasm in a granular pattern, focused primarily at perinuclear location. The coexpression of positivity for both markers was 43% in IDC and 88% in ILC.

Strikingly, Rictor staining intensity was higher in grade II/III breast tumors as compared with what was seen in grade I/II tumors (Supplementary Table S2; Supplementary Fig. S2B). Analysis of TCGA-curated invasive breast cancers revealed that 8% of all breast cancers assessed display RICTOR gene overexpression, mutation, or amplification (Supplementary Fig. S3), correlating with decreased overall survival (Fig. 1G). These data suggest that Rictor, a key factor for mTORC2 signaling, may play a role in tumor progression, an idea motivating further studies into Rictor/mTORC2 signaling in breast cancers.

We assessed the impact of Rictor/mTORC2 in the genesis of HER2-amplified breast tumors using a transgenic mouse model, MMTV-Neu-IRES-Cre (NIC; ref. 15), which expresses a bicistronic transcript comprised of oncogenic Neu (the rat HER2 homolog), followed by internal ribosomal entry site (IRES) and Cre recombinase expression cassettes. These mice develop mammary tumors with an average latency of 6 months (13). Mice harboring floxed Rictor alleles (Rictor^FL/FL) were crossed with NIC mice to generate Rictor^FL/FLNIC, ensuring Cre-mediated Rictor elimination in Neu-transformed cells. Rictor mRNA levels were reduced in Rictor^FL/FLNIC mammary samples versus Rictor^+/+NIC samples (Supplementary Fig. S4). We analyzed hematoxylin and eosin (H&E)-stained mammary glands of 12-week old virgin female mice, a time point prior to palpable tumor formation (Fig. 2A). Prominent epithelial structures seen in Rictor^+/+NIC samples were diminished in Rictor^FL/FLNIC samples (Fig. 2B). Whole mount hematoxylin-stained mammary glands showed diffuse hyperplasia in Rictor^+/+NIC, but not Rictor^FL/FLNIC samples (Fig. 2A). Importantly, focal neoplasias seen in whole mounted Rictor^+/+NIC mammary glands were reduced or absent in Rictor^FL/FLNIC samples (Fig. 2B).

IHC revealed abundant Rictor expression in Rictor^+/+NIC mammary glands, which was eliminated in Rictor^FL/FLNIC samples, although occasional cells retained Rictor expression (Fig. 2C, yellow arrows). P-Akt S473 was decreased in Rictor^FL/FLNIC mammary glands as compared with Rictor^+/+NIC. Histologic evidence of apoptotic bodies in Rictor^FL/FLNIC samples (Fig. 2C, black arrows) was investigated further by TUNEL analysis, revealing >2-fold more apoptotic cells in Rictor^FL/FLNIC over Rictor^+/+NIC (Fig. 2D and E). Ki67 staining, used as a marker of cellular proliferation, revealed similar proportions of proliferating cells in Rictor^FL/FLNIC and Rictor^+/+NIC mammary glands (Fig. 2D and E).

Rictor^+/+NIC mice (N = 11) formed tumors with 100% penetrance and an average latency of 135 days, but few Rictor^FL/FLNIC animals ever formed tumors (5/14, or 35%), doing so with a substantially delayed latency (Fig. 3A) and decreased tumor multiplicity per mouse (Fig. 3B), suggesting that Rictor promotes HER2-mediated mammary tumorigenesis. Rictor^FL/FLNIC tumors were distinct histologically from controls, harboring a lower nuclear-to cytoplasmic ratio, more serous fluid, and more acellular debris (Fig. 3C). Tumor cellularity was quantitated (excluding large areas of acellular debris from analysis) revealing decreased tumor cells per 600×-field in Rictor^FL/FLNIC samples versus Rictor-expressing controls (Fig. 3D). Tumor volume measured at 28 days after initial tumor palpation revealed decreased tumor burden in tumor-bearing Rictor^FL/FLNIC as compared with Rictor^+/+NIC mice (Fig. 3E, tumor-free mice were excluded from analysis).

Rictor^FL/FLNIC tumors harvested 28 days after tumor palpation displayed reduced Rictor and decreased P-Akt S473, as assessed by Western blot analysis of whole tumor lysates (Fig. 3F and G). These results were confirmed by IHC for P-Akt S473 (Fig. 3H). Ki67 staining was similar in Rictor^+/+NIC and Rictor^FL/FLNIC tumors (Fig. 3H and I). These data are consistent with the important role of Akt as an effector of HER2 signaling in breast tumorigenesis and suggest that Rictor/mTORC2 is required for Akt phosphorylation in HER2-amplified tumor cells.

Although mTORC1-selective inhibitors and mTORC1/2 dual kinase inhibitors currently exist, there are no mTORC2-selective inhibitors, and the impact of mTORC2-specific inhibition in the context of HER2-driven breast cancers is unclear. We addressed this by transducing HER2-amplified breast cancer cells with lentivirus encoding Rictor shRNA sequences (shRictor), or a scrambled control shRNA sequence (shScr). We confirmed Rictor depletion in SKBR3 (HER2^amp), MDA-MB-361 (PIK3CA^E545KHER2^amp), and BT474 (PIK3CA^K111NHER2^amp) cells. Cells expressing shRictor displayed robust downregulation of Rictor protein and decreased P-Akt S473 (Fig. 4A). Rictor knockdown decreased growth of SKBR3, MDA-MB-361, and BT474 cells grown in monolayer (Fig. 4B and Supplementary Fig. S5B), or embedded in three-dimensional (3D) Matrigel (Fig. 4C and Supplementary Fig. S5C). To deplete Rictor using an independent method, zinc-finger nucleases (ZFN) with specific homology to Rictor genomic sequences (ZFN-Rictor) were used to engineer homozygous RICTOR genomic deletions in MCF10A-HER2 cells. This approach confirmed that Rictor ablation decreased P-Akt S473 (Supplementary Fig. S5A). MCF10A-HER2 cells expressing ZFN-Rictor exhibited decreased growth in monolayer and in 3D Matrigel as compared with parental MCF10-HER2 cells (Supplementary Fig. S5D–S5F).

Because mTORC2 activates Akt, which in turn activates mTORC1, it is possible that the effects of Rictor/mTORC2 targeting seen here are due to inhibition of Akt-mediated mTORC1 signaling. Therefore, it was important to distinguish the relative impacts of mTORC1 inhibition versus mTORC2 inhibition in this scenario. Thus, we assessed HER2-amplified breast cancer cells with knockdown of Raptor, the mTORC1 obligate cofactor. We confirmed Raptor depletion in SKBR3, MDA-MB-361, and BT474 cells (Fig. 4A, right), finding decreased phosphorylation of the mTORC1 effectors ribosomal protein S6 and 4-EBP1. Raptor knockdown cells were assessed in parallel to Rictor knockdown cells in growth assays. Interestingly, S6 phosphorylation was not affected by inhibition of mTORC2. While Raptor ablation decreased growth of SKBR3 cells in monolayer, Raptor knockdown did not impact growth of MDA-MB-361 or BT474 cells in monolayer (Fig. 4B and Supplementary Fig. S5B). In 3D Matrigel cultures, shRaptor decreased colony formation in SKBR3 and BT474 cells, but not MDA-MB-361 cells (Fig. 4C and Supplementary Fig. S5C). In all cases, genetic Rictor/mTORC2 targeting decreased HER2-amplified breast cancer cell growth to an equal or greater extent than Raptor/mTORC1 targeting.

Xenografted MDA-MB-361 cells expressing shRictor formed palpable tumors 7 to 14 days after injection, similar to tumor formation in controls, suggesting that Rictor depletion did not affect tumor take rate in this model. Tumor measurements revealed similar tumor volumes in shScr and shRictor samples at 21 days after injection, but decreased volume of shRictor tumors at 52 days after injection (Fig. 4D). Tumors assessed at day 52 confirmed decreased Rictor and P-Akt S473 in shRictor tumors (Fig. 4E and F). Cell proliferation was not affected by Rictor ablation in this xenograft model (Fig. 4F and G), although shRictor tumors displayed fewer nuclei per 400× field (Fig. 4F–H).

The percentage of TUNEL⁺ cells was increased nearly 4-fold in spontaneous Rictor^FL/FLNIC tumors over controls (Fig. 5A). Similarly, the cellular fraction staining with Annexin V, a marker of dying cells, was increased upon Rictor knockdown in MDA-MB-361, SKBR3, and BT474 cells (Fig. 5B and Supplementary Fig. S6A) and in MCF10A-HER2 cells lacking Rictor expression (Rictor-ZFN; Supplementary Fig. S6B and S6C). In contrast, shRaptor did not affect the percentage of Annexin V–positive SKBR3, MDA-MB-361, or BT474 cells (Fig. 5B). Similar results were observed using a luminescent caspase-3/7 assay to measure caspase activity, demonstrating that shRictor, but not shRaptor, increased caspase activity (Fig. 5C and Supplementary Fig. S6A). MDA-MB-361-shRictor xenografts exhibited nearly 10-fold increased percentage of TUNEL⁺ tumor cells over shScr controls (Fig. 5D), suggesting that tumor cell survival is affected downstream of Rictor/mTORC2 ablation, but not Raptor/mTORC1 ablation.

To test the hypothesis that Rictor-to-Akt signaling drives survival of HER2⁺ breast cancer cells, we expressed active Akt phospho-mimetic (Ad.Akt^DD) in SKBR3-shRictor and MDA-MB-361-shRictor cells, rescuing P-Akt S473 despite sustained Rictor depletion (Fig. 5E). Restoration of P-Akt in Rictor-depleted cells increased cell survival (Fig. 5F and Supplementary Fig. S7A). Conversely, the allosteric Akt kinase inhibitor AZD5363, which decreased phosphorylation of the Akt substrate PRAS40 (Fig. 5G), decreased survival of parental SKBR3 and MDA-MB-361 cells (Fig. 5H and Supplementary Fig. S7B), suggesting that HER2-amplified breast cancer cells require Rictor-mediated Akt phosphorylation for cell survival. These results confirm that cell death occurs in response to loss of Rictor/mTORC2 signaling to Akt, but independently of Akt-mediated mTORC1 activation.

We tested the impact of Rictor/mTORC2 loss on therapeutically induced tumor cell death, using lapatinib to block HER2 kinase activity in SKBR3, MDA-MB-361, and BT474 cells. Within 0.5 hour, lapatinib (1 μmol/L) decreased P-HER2 and P-Akt S473, as expected (Fig. 6A). However, P-Akt S473 reemerged after 4–8 hours of lapatinib treatment, correlating temporally with Rictor (but not Raptor) protein upregulation, despite sustained suppression of P-HER2 and P-S6 (Fig. 6A). RICTOR mRNA levels were increased after 4–8 hours of lapatinib treatment in each cell line tested (Fig. 6B), as was RPTOR (Supplementary Fig. S8).

We next examined the effect of pharmacologic mTORC1/2 inhibition on tumor cell response to lapatinib. As expected, lapatinib treatment blocked P-HER2 in SKBR3, MDA-MB-361, and BT474 cells, and decreased P-S6 and P-Akt S473 (Fig. 6C). Dual mTORC1/2 inhibition using PP242 inhibited P-S6 and reduced P-Akt S473. However, combined treatment with lapatinib and PP242 completely abolished P-Akt S473 and P-S6 in all three cell lines. Although PP242 and lapatinib each decreased growth of SKBR3, MDA-MB-361, and BT474 cells as single agents, the combination of lapatinib with PP242 blocked cell growth to a greater degree than either agent alone (Fig. 6D and Supplementary Fig. S9). Furthermore, the combination of lapatinib with PP242 increased caspase activity to a greater degree than either agent alone. (Fig. 6E).

To define the impact of mTORC2-specific ablation on lapatinib-mediated cell death, we used SKBR3 and MDA-MB-361 cells expressing shRictor. HER2 inhibition combined with Rictor depletion decreased P-Akt S473 to a greater extent than lapatinib alone (Fig. 6F). In contrast, shRaptor had no impact on P-Akt S473, but effectively blocked P-S6 (Fig. 6G). Although lapatinib as a single agent decreased growth of SKBR3 and MDA-MB-361 acini in 3D Matrigel, the combination of lapatinib plus shRictor produced the greatest degree of growth inhibition (Fig. 6H and Supplementary Fig. S10A). The combination of lapatinib plus shRaptor blocked growth more than lapatinib alone, but not to the same extent as shRictor. Lapatinib-mediated cell death as measured by Annexin-V staining was highest in cells expressing shRictor as compared with shScr or shRaptor (Fig. 6I and Supplementary Fig. S10B). These data suggest that Rictor/mTORC2 targeting in HER2-amplified breast cancer cells improves tumor cell response to lapatinib.

To assess the requirement of mTORC1/2 in lapatinib-resistant cells, we used lapatinib-resistant SKBR3, MDA-MB-361, and BT474 cells. These cells were rendered lapatinib-resistant by culturing in progressively increasing concentrations of lapatinib for >6 months, and have been previously described (14). After withdrawing lapatinib for 48 hours, we treated SKBR3-LR, MDA-MB-361-LR, and BT474-LR cells with lapatinib in the presence or absence of PP242. While P-HER2 and P-AktS473 were unaffected by lapatinib treatment, PP242 abolished P-Akt S473 (Fig. 7A). PP242, but not lapatinib, decreased growth of SKBR3-LR, MDA-MB-361-LR, and BT474-LR cells (Fig. 7B). Caspase-3/7 activity was increased in cells treated with PP242, but not in cells treated with lapatinib (Fig. 7C).

We examined mTORC2 complex assembly in BT474 and BT474-LR cells using coimmunoprecipitation (co-IP) for mTOR with Rictor. Rictor coprecipitated with mTOR in both parental and BT474-LR cells (Fig. 7D), confirming mTORC2 complex assembly. Interestingly, Raptor coprecipitated with mTOR to a lesser extent in BT474-LR cells as compared with what was seen in parental BT474 cells.

As PP242 blocks both mTORC1 and mTORC2, we assessed the impact of Rictor/mTORC2-specific targeting in LR cells by Rictor knockdown, which was confirmed by Western blot analysis (Fig. 7E). Lapatinib did not affect the growth of SKBR3-LR, MDA-MB-361-LR, and BT474-LR cells expressing shScr, while cells expressing shRictor grew at a decreased rate (Fig. 7F and G). Furthermore, LR cells expressing shRictor exhibited increased cell death over shScr. (Fig. 7H), suggesting that lapatinib-resistant cells are exquisitely sensitive to Rictor/mTORC2 targeting.

Taken together, these data highlight the critical role of Rictor/mTORC2 in activating Akt to support genesis, growth, and survival of HER2-amplified breast cancers.

Approximately 20% of breast cancers exhibit overexpression of HER2, a marker of aggressive disease (16). The studies presented here uncover the key roles played by mTORC2/Rictor in models of HER2-amplified breast cancer initiation, maintenance, and progression. Our results show that RICTOR ablation in a transgenic mouse model of HER2-driven breast cancer decreased tumor formation and tumor multiplicity, suggesting that Rictor promotes the genesis of HER2-overexpressing tumors. Consistent with the idea that mTORC2 is a key factor in Akt phosphorylation, Rictor depletion decreased Akt S473 phosphorylation and tumor cell survival in this model of spontaneous breast cancer. Furthermore, RICTOR loss decreased Akt activation (P-S473) and cell survival in multiple HER2-amplified human breast cancer cell lines, including lines with activating PIK3CA mutations. Other intracellular serine-threonine kinases are capable of Akt S473 phosphorylation, including IKK-α, DNA-PK, and ILK (2). However, our studies suggest that HER2⁺ human breast cancer cells and mouse mammary tumors specifically require Rictor/mTORC2 for Akt S473 phosphorylation, as Rictor ablation eliminated Akt phosphorylation and decreased cell survival. Interestingly, this is distinct from what is seen in untransformed MECs, which rely on an Akt-independent, Rictor–PKCa–Rac1 signaling axis for cell survival, invasion, and morphogenesis (8).

Several preclinical studies demonstrate that Akt/mTOR inhibition is critical for full therapeutic response to HER2 inhibition in HER2-amplified breast cancers (17, 18). However, most clinical and preclinical studies have focused primarily on mTOR associated with mTORC1, perhaps because mTORC1 is activated downstream of HER2/PI3K signaling. While mTORC1 blockade produces therapeutic benefit, mTORC1 inhibition also induces resurgent PI3K signaling in tumor cells, reducing net antitumor activity, and driving therapeutic resistance. mTOR kinase inhibitors, which target both mTORC1 and mTORC2, have generated intense interest, although little is known about mTORC2-specific signaling in breast cancers. Our analyses, using novel transgenic mouse models and genetic inhibition of mTORC2 reveal a key role for Rictor/mTORC2 in HER2-amplified breast cancers, including those with acquired resistance to lapatinib.

Our studies confirm Rictor overexpression in invasive human breast tumors as reported by Zhang and colleagues (19). RICTOR gene amplification and Rictor overexpression was noted in a subset of lung cancer patients with susceptibility to mTORC1/2 dual kinase inhibitors (20). Rictor overexpression was also reported in melanomas (21), glioblastomas (22, 23), gastric cancers (24), and hepatocellular carcinomas (25), supporting a potential role for mTORC2 components in aggressive tumors, and highlighting Rictor/mTORC2 as a potential therapeutic target.

As AKT1/2/3 and RICTOR gene mutations are rare in breast cancer (Fig. 1A and Supplementary Fig. S3), Akt and Rictor possibly are regulated at the levels of gene amplification, transcription, and/or catalytic activity. One recent study identified RICTOR transcripts as a miR-218 target, suggesting cancers may benefit from miR-218 loss in part through upregulation of Rictor/mTORC2. Another study demonstrated glucose-induced Rictor acetylation, which sustained mTORC2 activation in the absence of upstream signaling input, driving therapeutic resistance to inhibitors of the PI3K pathway (PMC4522814). Another study showed Rictor T1135 phosphorylation by the kinase S6K1, which downregulated mTORC2 activity, suggesting that alterations in signaling pathways upstream of S6K1 could affect mTORC2. Because the mTORC2 substrate Akt was more frequently upregulated in HER2-positive breast cancers than in TNBC (Fig. 1), we focused our studies on HER2-amplified breast cancers. However, Rictor expression was more frequently elevated in triple-negative breast cancers (TNBC) than in HER2-positive breast cancers, suggesting that Rictor/mTORC2 signaling is not exclusive to HER2 tumors, but its role in TNBCs may be independent of HER2 and Akt.

In summary, we have defined a previously unreported role for Rictor/mTORC2 in the genesis, maintenance, and therapeutic response of HER2-driven breast cancers. We have shown that Rictor/mTORC2 blockade is effective at each stage in the natural history of HER2-amplified tumors, supporting continued translational investigations of dual mTORC1/2 inhibitors to improve the outcome for patients with HER2-amplified breast cancers, and warranting future efforts to develop mTORC2-specific inhibitors.

No potential conflicts of interest were disclosed.

Conception and design: M. Morrison Joly, M. Williams, B.N. Rexer, R.S. Cook

Development of methodology: M. Morrison Joly, D.J. Hicks, M. Williams, D.D. Sarbassov, D. Brantley-Sieders, R.S. Cook

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Morrison Joly, D.J. Hicks, B. Jones, C. Young, B.N. Rexer, W.J. Muller, D. Brantley-Sieders

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Morrison Joly, B. Jones, M.V. Estrada, C. Young, M. Williams, D. Brantley-Sieders

Writing, review, and/or revision of the manuscript: M. Morrison Joly, M. Williams, D. Brantley-Sieders, R.S. Cook

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Morrison Joly, D.J. Hicks, C. Young, R.S. Cook

Study supervision: M. Morrison Joly, M. Williams, R.S. Cook

Other (performed the immunohistochemistry procedures): V. Sanchez

R.S. Cook received NIH award R01CA143126 and Susan G. Komen for the Cure grant KG100677. M. Morrison-Joly received NRSA F31 predoctoral award CA186329-01 and CTSA award no. UL1TR000445 from the National Center for Advancing Translational Sciences. M. Williams received NRSA F31 predoctoral award CA195989-01. C.D. Young received CDMRP-BCRP award W81XWH-12-1-0026.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.