Background
Renal cell carcinoma (RCC) consists of distinct subtypes with characteristic histologic features, genetic mutations and clinical behaviors [
1]. The RCC subtype harboring an Xp11.2 chromosomal rearrangement (Xp11 Translocation RCC,
TFE3-fusion RCC, TfRCC) comprises 1–5% of all RCC cases [
2‐
5]. Rearrangements include an inversion or translocation of the
TFE3 gene (Xp11.2), which is a member of the Microphthalmia-associated transcription factor (MiT) family that regulates growth and differentiation [
6]. The resulting gene-fusion product links the TFE3 C-terminus with the N-terminus of a fusion partner [e.g.
PRCC (1q23),
ASPSCR1 (17q25),
SFPQ (1p34),
NONO (Xq13) or
CLTC (17q23)] [
6]. Introduction of a constitutively active promoter upstream of the 3′
TFE3 gene portion is thought to promote carcinogenesis through increased TFE3 C-terminus expression, nuclear localization and transcriptional activity [
6]. Characteristic clinical features include common diagnosis in early or mid-adulthood, frequent metastasis at presentation [
7] and other atypical risk factors for RCC, including female gender and childhood chemotherapy [
3,
7‐
9]. Defining histologic features include clear and eosinophilic cells, papillary and/or nested architecture, and occasional psammoma bodies [
8,
10]. The diagnosis is suggested by young age, tumor histology and nuclear immunoreactivity for the TFE3 C-terminus; however, confirmation of diagnosis requires cytogenetic or molecular evidence of an Xp11 rearrangement or fusion transcript [
8,
10,
11].
Effective drug therapies are yet to be identified for TfRCC, and there is no clinical standard for systemic treatment. Prospective drug trials in metastatic TfRCC patients have not been performed due to the lack of known agents with preclinical efficacy. Retrospective studies suggest rapid progression with cytokine therapy and only occasional, partial responses to rapalogs or anti-angiogenesis therapies [
2,
12‐
17]. Mouse models of xenografted TfRCC patient tumor cell lines are established and provide a promising tool for preclinical drug discovery [
6].
Novel drug discovery for TfRCC will benefit from identification of key molecular pathways driving this disease [
6]. A variety of cellular functions are governed by wild-type TFE3, and the simultaneous dysregulation of these functions might be sufficient to promote carcinogenesis. Key pathways regulated by TfRCC may involve TGFβ, ETS transcription factor, E-cadherin, MET tyrosine kinase, insulin receptor, folliculin, Rb and other cell cycle proteins [
6]. Intriguingly, a common connection among these pathways/proteins is the involvement of Akt, a key regulator of cell growth, metabolism and cytoskeletal reorganization. Akt activation is common in many cancers and the target of ongoing clinical trials [
18,
19]. We and others have previously described common phosphorylation of Akt in clear cell RCC (ccRCC) tumors and cell lines, including constitutively in the absence of exogenous growth factor stimulation, but similar investigation in TfRCC models is lacking [
18‐
21].
An important downstream target of Akt signaling is the mTOR-containing protein complex, mTORC1, a master regulator of protein synthesis, cellular metabolism and autophagy. Activation of mTORC1 is thought to promote ccRCC carcinogenesis, at least in part, through increased cap-dependent translation of the hypoxia-inducible factor alpha (HIFα) transcript [
22]. Selective pharmacologic inhibition of mTORC1 with temsirolimus is approved by the FDA for treatment of high risk metastatic RCC patients and prolongs their survival [
23]. However, clinical resistance to mTORC1 inhibition limits its long-term efficacy and may be mediated by several mechanisms, including a feedback loop involving a second mTOR-containing complex, mTORC2, which phosphorylates Akt in response to mTORC1 inhibition [
24,
25]. Concomitant targeting of mTORC1 and mTORC2 is an intriguing therapeutic strategy that has been evaluated in several malignancies, including ccRCC, with promising preclinical results [
26]. Previous studies have described increased activation of mTORC1 in TfRCC tumors [
27,
28], which supports the Akt/mTOR pathway to be a potential pharmacological target for TfRCC [
28].
Here we examined Akt/mTOR pathway activation and the preclinical efficacy of dual mTORC1/2 inhibition compared to selective mTORC1 inhibition in TfRCC preclinical in vitro and in vivo models. The results support an important role for Akt/mTOR activation in TfRCC carcinogenesis and identify dual mTORC1/2 inhibition as a systemic therapeutic strategy with in vivo preclinical efficacy against this cancer.
Methods
Cell lines and culture
The UOK109, UOK120, UOK124 and UOK146 cell lines had previously been derived from tumors excised from four TfRCC patients who were treated at the National Cancer Institute (NCI, Bethesda, MD), and had been shown to harbor the
NONO-TFE3 or
PRCC-TFE3 gene fusions [
29‐
31]. The UOK111, UOK139 and UOK150 cell lines had been derived from ccRCC tumors excised from RCC patients treated at the NCI and were shown to harbor
VHL gene mutations [
32,
33]. Collection of this material was approved by the Institutional Review Board of the National Cancer Institute and all patients had provided written informed consent. RCC4 was obtained from ECACC General Cell Collection (Salisbury, UK; Cat Nr. 03112702) and the human renal cortical epithelial (HRCE) cell line was obtained from ATCC (Manassas, VA; Cat Nr. PCS-400-011). All cell lines were maintained in vitro in DMEM media supplemented with L-glutamine (4 mM), sodium pyruvate (110 mg/l), glucose (4.5 g/l), and 1X essential amino acids (Gibco, Gaithersburg, MD), with or without 10% fetal bovine serum (Sigma Aldrich, St. Luis, MO). Cell lines were authenticated using short tandem repeat DNA profiling (Genetica DNA Laboratories, Burlington, NC) and confirmed to be mycoplasma-free by LookOut® Mycoplasma qPCR Detection Kit (Sigma Aldrich).
Immunoblotting
Phosphorylated and total levels of Akt/mTOR pathway proteins were measured by immunoblot in TfRCC and ccRCC cell lines. ccRCC cell lines were used for comparison since we have previously shown that this RCC subtype has frequent constitutive activation of the Akt/mTOR pathway [
20]. Akt kinase activation was evaluated by measurement of phosphorylated levels of Akt (Thr308) and Akt (Ser473), the latter also served as a reporter for mTORC2 activation [
25], in addition to levels of phosphorylated GSK3β, which is an Akt kinase target. Activation of mTORC1 was assessed by measuring phosphorylated levels of S6 ribosomal protein (Ser240/244) and 4EBP1 (Thr37/46 and Ser65); levels of HIF1α protein, whose translation is suppressed by hypophosphorylated 4EBP1 through its interaction with eIF4E, provided an indirect measure of mTORC1 activity [
34]. Levels of phosphorylated mTOR provided additional measures of mTORC1 and mTORC2 activity, where mTOR Ser2448 is activated by S6K1 kinase and reflects amino acid and nutrient status [
35] and mTOR Ser2481 autophosphorylation site correlates with intrinsic mTOR catalytic activity [
26,
36]. Protein lysates were harvested from cell lines at 60–70% confluency using RIPA buffer (Thermo-Fischer Scientific, Waltham, MA) supplemented with 1 mM PMSF protease inhibitor (Sigma Aldrich). Two-dimensional electrophoretic separation of proteins was performed using 10 μg protein/well in 4–20% gradient polyacrylamide gels (Biorad, Hercules, CA) and transferred onto PVDF membranes (BioRad). Membranes were blocked for 1 h at room temperature in 5% fat-free milk with 0.1% tween, followed by overnight incubation at 4 °C with primary antibody in either fat-free milk and 0.1% tween or TBS with 5% bovine serum albumin and 0.1% tween. Primary antibodies included rabbit anti-P-mTOR (Ser2448), rabbit anti-P-mTOR (Ser2481), rabbit anti-mTOR (total), rabbit anti-P-Akt (Thr308), rabbit anti-P-Akt (Ser473), mouse anti-Akt (total), rabbit anti-P-GSK3β (Ser9), rabbit anti-GSK3β total, rabbit anti-P-S6 (Ser240/244), rabbit anti-S6 (total), rabbit anti-P-4EBP1 (Thr37/46), rabbit anti-P-4EBP1 (Ser65), rabbit anti-4EBP1 (total), rabbit anti-VHL, and mouse anti-β-actin (all from Cell Signaling Technology, Danvers, MA); mouse anti-HIF1α (BD Biosciences, San Jose, CA); and goat anti-TFE3 (Santa Cruz Biotechnology, Santa Cruz, CA). All primary antibodies were incubated at a 1:1000 dilution, with the exception of the anti-VHL and anti-HIF1α, for which a 1:500 dilutions were used. Primary antibody-stained membranes were incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibody, including goat anti-mouse 1:2000 (Cell Signaling Technology), goat anti-rabbit 1:5000 (Cell Signaling Technology) or donkey anti-goat 1:5000 (Santa Cruz Biotechnology). Secondary-antibody stained membranes were developed using a chemiluminescence kit (Pierce, Rockford, IL) followed by radiographic film exposure.
Drug agents
The dual mTORC1/2 inhibitor, AZD8055 (AstraZeneca, London, UK), was prepared for in vitro assays by dissolution in DMSO to 10 mM (4.65 mg/mL), per manufacturer instructions. The selective mTORC1 inhibitor, sirolimus (Selleckchem, Houston, TX), was prepared for in vitro assays by dissolution in 100% ethanol to 10.9 mM (10 mg/mL). For in vivo assays, AZD8055 was dissolved by sonication in 30% Captisol (CyDex Pharmaceuticals, Lenexa, KS) to a working concentration of 2 mg/ml and pH of 5.0 per manufacturer instructions. For in vivo assays, sirolimus was dissolved in 5% Tween-80 (Sigma Aldrich) and 5% PEG-400 (Hampton Research, Aliso Viejo, CA) to a working concentration of 0.4 mg/ml. Doses of ~ 200 μl drugs were administered to each animal.
Cell viability assay
Cell viability in vitro was measured using the tetrazolium salt 3-(4,5-dimethylthiazol-2-yl)-2,5- diphenyltetrazolium bromide (MTT, Sigma Aldrich) in a 96-well plate format after 72 h of treatment as previously described [
20].
Cytotoxicity assay
Cell cytotoxicity in vitro was measured with the lactate dehydrogenase (LDH)-based Cytotoxicity Detection Kit (Roche, Indianapolis, IN) using the modified protocol described by Smith et al. [
37]. Briefly, 1–5 × 10
3 cells were plated onto a 96-well plate to achieve approximately 20% cell confluency 1 day after plating, and drug treatment was initiated in pyruvate-free media. Media without cells served as a control for baseline LDH levels in serum (“media control”). After 48 h of treatment, 4 μl Triton X-100 detergent was added to half of the wells for each drug concentration to lyse all live cells (“high controls”). Reaction mixture was made per manufacturer instructions and added to all wells, and absorbance was measured at 490 nm wavelength (Abs
490). Cytotoxicity for each concentration was calculated as [Abs
490 (condition) – Abs
490 (media control)] / [Abs
490 (condition high control) – Abs
490 (media control)] [
37]. The drug LY294002 was used as a positive control for cytotoxicity induction.
Cell cycle analysis
Cell cycle analysis was performed following 24-h drug treatment as previously described [
38].
TfRCC mouse xenograft experiments
Animal studies were approved by the NIH Institutional Animal Care and Use Committee (IACUC; Protocol Number: PB-029) and conducted in accordance with US and International regulations for protection of laboratory animals. TfRCC tumor xenografts were generated using the UOK120 and UOK146 cell lines in female immunocompromised athymic Nude mice (Foxn1nu; Jackson Laboratory, Bar Harbor, ME) at 4–6 weeks of age. Mice were housed under specific pathogen free conditions. Briefly, 5 × 106 cells in PBS suspension with 30% (UOK120) or 50% (UOK146) Matrigel (BD Biosciences, Franklin Lakes, NY) were injected subcutaneously into the mouse right flank. When UOK120 (N = 34) or UOK146 (N = 40) tumors were palpable (volume 0.05–0.20 cm3), treatment was initiated with doses of 4 mg/kg sirolimus intraperitoneal (IP) weekly, IP vehicle control weekly (5% Tween-80 and 5% PEG-400), AZD8055 20 mg/kg oral (PO) daily, or PO vehicle control daily (30% Captisol, pH 5.0). 24 UOK120 mice were randomly assigned to receive either AZD8055 (N = 12) or PO control (N = 12), while 10 UOK120 mice were randomly assigned to receive sirolimus (N = 5) or IP control (N = 5). 40 UOK146 mice were randomly assigned to receive AZD8055 (N = 10), PO control (N = 10), sirolimus (N = 10), or IP control (N = 10). Mouse weights were monitored weekly. Tumor dimensions were measured every 2 days and volume was calculated using the formula: 0.4 × (width)2× (length). Mice were sacrificed by CO2 asphyxiation and cervical dislocation when the longest tumor diameter reached 2 cm per institutional regulations. An additional 8 mice xenografted with UOK120 or UOK146 tumors underwent the same treatments (N = 2 mice per treatment) and were sacrificed at 6 h after their first drug dose for analysis of tumor protein. Protein lysates were prepared by mincing tissue and solubilization in RIPA Buffer (Thermo Fisher Scientific). Immunoblotting was performed as described above, with the exception that detection was performed with a Licor Odyssey Imager (LI-COR Biosciences, Lincoln, NE).
Tumor growth of mouse xenografts was compared by calculating linear regressions of growth curves over the treatment period and calculation of p-values through a Mann-Whitney test. Survival times were analyzed through a log-rank test and graphed with GraphPad Prism 7.01 (La Jolla, CA).
Discussion
TfRCC is an aggressive RCC subtype with no known effective therapy in the clinical or preclinical setting [
2,
12‐
17]. TfRCC incidence has been historically underestimated because of frequent misdiagnosis as either ccRCC or papillary RCC due to overlapping histologic features, particularly when clinical suspicion for TfRCC (i.e., young age) is otherwise lacking [
8]. Retrospective identification of
TFE3-fusion gene mutations by the TCGA project in several patients diagnosed originally with ccRCC or papillary RCC is consistent with the 1–5% incidence of retrospective identification reported among nephrectomy patients by others [
2‐
5] and may be even higher among metastatic RCC patients. Development of novel therapeutic strategies for TfRCC patients warrants investigation, and identification of key molecular pathways driving TfRCC carcinogenesis is a critical first step.
The current study reveals Akt/mTOR pathway activation in TfRCC cell lines. Akt and mTORC1 pathway activation is common in many human cancers, including ccRCC [
18‐
22] and is mediated by phosphoinositide kinase 1 (PDK-1), the VHL/EGLN suppressive pathway [
41], and the mTORC2 complex. mTORC1 activation, as measured by downstream S6 phosphorylation, is reported to be higher in suspected or genetically confirmed TfRCC tumors compared to ccRCC or papillary RCC tumors [
27,
28]. We similarly observed high levels of phosphorylated S6 in TfRCC cell lines, comparable to levels in ccRCC cell lines. Levels of Akt activity in TfRCC cell lines generally surpassed those in ccRCC cell lines evaluated and were partly independent of exogenous growth factor stimulation, as previously described for ccRCC [
20]. Persistent phosphorylation of mTOR targets in the absence of exogenous growth factor stimulation is consistent with some level of constitutive activation of the mTORC1 and mTORC2 complexes in TfRCC cells. These results suggest that dysregulated Akt and mTOR activation may play an important role in TfRCC carcinogenesis.
To further explore this possibility, we evaluated the efficacy of a dual mTORC1/2 inhibitor, AZD8055, and compared it with a selective mTORC1 inhibitor, sirolimus, in TfRCC cell lines, observing consistently greater growth inhibition with dual mTORC1/2 inhibition. The inhibitory mechanism for both AZD8055 and sirolimus included cell cycle arrest without significant cytotoxicity induction, consistent with the effect of rapalogs reported in other cancer types [
42]. Both drugs caused less growth inhibition in benign renal epithelial cells compared to TfRCC cells, indicating a largely cancer-specific effect. Greater growth suppression with AZD8055 than sirolimus in vitro was validated in vivo using two separate mouse xenograft models of TfRCC. These results are consistent with another preclinical study that recently reported PI3K/mTOR pathway dysregulation in TfRCC and suggested that more complete inhibition of this pathway with a dual TORC1/2 and PI3K inhibitor (BEZ-235) results in a greater antiproliferative effect than a selective TORC1 inhibitor [
28].
Greater TfRCC suppression with AZD8055 relative to sirolimus is likely due to more complete suppression of the Akt/mTOR pathway. AZD8055- versus sirolimus-treated TfRCC cell lines and mouse xenografts demonstrated clear differences in Akt/mTOR pathway activation. Selective mTORC1 inhibition induced feedback activation of Akt kinase and, consequently, less effective inhibition of downstream S6 phosphorylation, whereas dual mTORC1/2 inhibition suppressed both upstream Akt activation and downstream S6 phosphorylation. Feedback activation of Akt in response to mTORC1 inhibitors is well described in many cancers and may directly mediate clinical resistance in RCC patients [
24‐
26,
39,
40,
43]. Dual mTORC1/2 inhibition blocks this feedback activation and hence provides a promising strategy for overcoming clinical resistance to selective mTORC1 inhibition.
To date, no drug treatment strategy has demonstrated consistent clinical efficacy for metastatic TfRCC patients. Clinical studies are limited by small cohort sizes, retrospective designs, lack of genetic confirmation of
TFE3-fusion, and heterogeneity in treatment parameters [
2,
12‐
17]. Cytokine therapy is largely ineffective [
2,
14‐
16], and the efficacy of angiogenesis inhibitors has been limited, with progression-free survival typically under 1 year [
16,
17]. Similarly, case reports of mTORC1 inhibitors in TfRCC patients suggest rapid progression during treatment [
12,
13]. There is hence a clear need for novel therapeutic strategies that broaden the therapeutic target beyond mTORC1. Combinations of mTORC1 and angiogenesis inhibitors have not yet demonstrated clinical benefit over VEGF pathway antagonists alone, and do not address the resistance mechanism of upstream Akt reactivation [
44]. The combination of Akt and mTORC1 inhibitors has demonstrated synergistic preclinical efficacy in various cancer types [
39,
45]. Dual mTORC1/2 inhibitors such as AZD8055 or Ku0063794 suppress growth of ccRCC cell lines, including those resistant to angiogenesis inhibitors [
26,
40]. Although dual mTORC1/2 inhibition with AZD2014 proved inferior to everolimus in metastatic ccRCC patients [
46], preclinical studies from our group and others suggest that AZD8055 is superior to rapalogs in ccRCC [
40,
47]. The present study extends this prior work to TfRCC, and provides encouraging preclinical rationale for clinical investigation of dual mTORC1/2 inhibition in TfRCC patients [
48].
The mechanism underlying constitutive activation of mTOR and Akt in TfRCC warrants future investigation. Activating mutations in the
MTOR gene have not yet been detected in patient tumors harboring a
TFE3 gene fusion, nor have mutations in
PIK3CA or
PTEN [
4]. Likewise, genetic characterization of commonly mutated cancer genes in the TfRCC cell lines used in this study did not reveal any pathogenic mutations (unpublished results). Both PI3K and PTEN are implicated as upstream activators of mTORC2 [
43]. Given the potential ability of PI3K to activate both mTORC2 and PDK-1, dysregulated PI3K could theoretically explain the high phosphorylation at both Akt (Ser473) and Akt (Thr308) observed in TfRCC. Simultaneous pharmacologic inhibition of PI3K and mTORC1 has demonstrated preclinical efficacy in ccRCC, however dose-limiting toxicity has hindered clinical use [
49,
50]. Dual mTORC1/2 inhibition might have lower toxicity owing to its narrower target spectrum, as suggested by a phase I trial of AZD8055 [
51]. The MET tyrosine kinase, an upstream activator of Akt, has been proposed to mediate TfRCC carcinogenesis [
52], however the putative MET inhibitor, Tivantinib, had no objective responses and poor progression free survival (median 1.9 months) in a small number of RCC patients with a MiT family gene fusion [
53]. Such findings warrant reexamination of the importance of MET in TfRCC and are consistent with our prior work showing no significant baseline MET activation in TfRCC cell lines or growth inhibition of these cell lines in response to biologically relevant concentrations of multiple MET-selective inhibitors [
6,
54].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.