Background
Renal cell carcinoma (RCC), one of the ten most common cancers, occurs in both sporadic and inherited forms. MiT family translocation RCC, is an aggressive form of the disease [
1‐
3] which accounts for 1–5% of sporadic RCC and represents 42% of kidney cancer in children and young adults [
4]. MiT family translocation RCC is characterized by gene fusions resulting from somatic chromosomal translocations involving microphthalmia-associated transcription factor (MiT) family members TFE3, TFEB or MITF. These proteins are basic helix-loop-helix leucine zipper transcription factors that undergo homo- or hetero-dimerization and then bind to E-box sequences in target genes to promote transcription. MiT proteins are considered master regulators of lysosomal biogenesis, with additional diverse and tissue specific functions often related to cell growth and differentiation [
5].
TFE3-fusion RCC, the most common form of MiT translocation RCC, is characterized by Xp11.2 chromosomal translocations involving TFE3 that result in a fusion of the C-terminal portion of TFE3 with the N-terminal portion of specific partner genes. Consequently, activity of the TFE3 C-terminus is dysregulated resulting in constitutive nuclear localization and increased transcriptional activity of the protein. The most common Xp11.2 translocation is t(X;1)(p11.2;q21), resulting in a
PRCC–TFE3 fusion protein [
6]; other rearrangements include t(X;1)(p11.2;p34) leading to
SFPQ‐
TFE3 fusion, inv(X)(p11.2;q12) leading to
NONO‐
TFE3 fusion, t(X;17)(p11;q25) leading to
ASPSCR1‐
TFE3 fusion, and t(X;17)(p11;q23) leading to
CLCT‐
TFE3 fusion [
5].
MiT translocation RCC is characterized by heterogeneous architectural and cytologic features that overlap with clear cell and papillary RCC as well as oncocytic tumors [
7], and have historically been frequently underdiagnosed. Recent TCGA data confirmed that MiT translocation RCC is more common than appreciated and is found in up to 12% of type 2 papillary RCC in adult patients [
8].
TFE3-fusion RCC is a highly aggressive form of renal cancer and affected patients often present with metastasis at initial diagnosis. There is currently no clinical standard of effective systemic treatment and no effective targeted drug therapies have been identified to date. Activity of tyrosine kinase inhibitors (TKI) against TFE3-RCC has recently been reported in in vitro organoid-based experiments [
9], and a clinically complete response to the multi-targeted TKI sunitinib was reported in a child with metastatic
TFE3 translocation RCC [
10]. A few retrospective studies of adult patients with metastatic TFE3 fusion RCC have reported incomplete responses to TKIs and mTOR inhibitors [
11,
12]. Despite genomics or transcriptomics analyses of TFE3-fusion RCC, information leading to an effective therapeutic approach targeting key signaling pathways that drive these tumors is lacking [
13,
14].
A known biomarker and potential therapeutic target for TFE3-fusion RCC is glycoprotein nonmetastatic melanoma B (GPNMB), a highly glycosylated transmembrane protein that regulates a variety of physiological processes, notably osteoclast differentiation and melanosome maturation [
15].
GPNMB is overexpressed in a number of cancer types [
16,
17] and was first shown to be transcriptionally regulated by MITF [
18,
19], but subsequently was identified as a direct transcriptional target of TFE3 and is highly expressed in TFE3-fusion RCC [
20,
21].
Unbiased high-throughput drug screens are a potential first line approach to identifying key signaling pathways for the development of therapies. Herein we have conducted quantitative high-throughput screens of multiple TFE3-fusion RCC cell lines derived from TFE3-fusion renal tumors and identified classes of drugs with cytotoxicity against TFE3-fusion RCC. We further validated the cytotoxicity and increased apoptosis of selected agents from each class and their effects on cell cycle and downstream pathway members in 2D and 3D in vitro assays and TFE3-fusion derived xenograft models. Notably, we identified the TFE3 transcriptional target GPNMB as a cell surface biomarker that is upregulated in TFE3-fusion RCC, and evaluated a clinically relevant antibody-drug conjugate (ADC) that targets this protein in in vitro and in vivo assays. Finally, combinations of these drugs with the ADC were assessed for potential synergy as therapeutic agents in the TFE3-fusion cell lines and xenograft models.
Materials and methods
Patients
Patients were seen at the Urologic Oncology Branch (UOB) of the National Cancer Institute (NCI), National Institutes of Health (NIH) for clinical assessment. This study was approved by the Institutional Review Board of the National Cancer Institute and patients provided written informed consent on either Urologic Oncology Branch protocol NCI-89-C-0086 or NCI-97-C-0147.
Cell lines and cell culture
The TFE3-fusion RCC cell lines UOK109, UOK120, UOK124, UOK145, and UOK146 were developed within the Urologic Oncology Branch (NCI) from surgically resected specimens (
2,
22,
23). The clear cell RCC-derived UOK140 cell line (
22) was used as negative control in CDX-011 experiments. Cells were maintained at 37ºC with 5% CO
2 and were cultured in high glucose (4.5g/l) Dulbecco’s modified Eagle’s medium (DMEM; Gibco, Gaithersburg, MD) supplemented with L-glutamine (4 mM), sodium pyruvate (110mg/l), and 10% fetal bovine serum (Sigma Aldrich, St. Luis, MO).
Sanger sequencing
DNA was extracted from tumor tissue as above using the Maxwell Tissue Kit (Promega, WI, USA). PCR reactions to amplify the sequences across the gene fusions were performed using KAPA2G Fast DNA polymerase (Roche, Indianapolis, IN) according to the manufacturer’s specifications. Bidirectional Sanger DNA sequencing of the PCR products was performed using the Big Dye Terminator v.1.1 Cycle Sequencing Kit (Applied Biosystems, CA, USA) according to the manufacturer’s specifications and run on an ABI 3130xl or 3730 Genetic Analyzer (Applied Biosystems). Sanger Sequencing was conducted at the CCR Genomics Core at the National Cancer Institute, NIH, Bethesda, MD 20,892. Forward and reverse sequences were evaluated using Sequencher 5.0.1 (Genecodes, MI, USA). Fusion partner-specific primers to characterize the gene fusions were as follows: TFE3_Rv: GCAGGAGTTGCTGACAGTGA; NONO_Fw: ATCAAGGAGGCTCGTGAG; PRCC_Fw1: AGGAAAGAGCCCGTGAAGAT (UOK120, UOK146): PRCC_Fw2: ATGCCGCTGGTGCTTATTAT (UOK124); SFPQ_Fw: CTTTTGCGCCAAGATCTGA.
Quantitative high-throughput screening
In collaboration with the National Center for Advancing Translational Sciences, NIH, quantitative high-throughput screening (qHTS) of 1912 pharmacologically defined small molecule compounds was conducted on three TFE3-fusion cell lines bearing three different fusion partners (UOK109, UOK124, UOK145) as outlined in
Supplemental Methods.
Cell viability assays
2D cell viability following 48–72 h drug treatment was evaluated using CellTiter-Glo reagent (Promega) according to manufacturer’s instructions. Assays were performed in triplicate and experiments were repeated three times in all five cell lines. Spheroids were successfully developed in three of the cell lines, UOK109, UOK120, and UOK124, using a previously described methodology [
24] to evaluate 3D cell viability. Spheroids were incubated with drug for five days, before assessing viability by CellTiter-Glo 3D Cell Viability Assay (Promega). Calculation of drug synergy was performed with Compusyn [
25], according to software instructions, including at least 6 concentration points for each drug alone and in combination. Cell cytotoxicity in vitro was measured with the lactate dehydrogenase (LDH)-based Cytotoxicity Detection Kit (Roche) as previously described [
26].
Reagents and resources
Commercially available reagents and resources included Mithramycin A (Tocris) and NVP-BGT226 (Selleck Chemicals). EC-8042 (EntreChem SL), antibody-drug conjugate CDX-011 and non-conjugated antibody CR011 (Celldex Therapeutics) were obtained from the manufacturers through collaborative agreements. All other compounds were generously provided by the NCI Development Therapeutics Program, NIH.
Flow cytometry assays
Cell cycle and cell apoptosis analyses were performed by flow cytometry as previously described [
27] using anti-cleaved PARP (BD Horizon) and anti-cleaved Caspase 3 (Cell Signaling Technologies) antibodies. Cell surface expression of GPNMB was measured by flow cytometry using anti-GPNMB (R&D Systems AF2550) or CR011 (Celldex Therapeutics) antibodies following 30-minute treatment with fixative solution or fixation/permeabilization solution (BD Biosciences). Cells were washed and resuspended in MACS buffer (PBS, 0.5% BSA, 2mM EDTA) for analysis. All samples were run on a BD FACS Canto II flow cytometer (Becton Dickinson, NJ) and analyzed with FlowJo Software (FlowJo, OR).
SP1 transcription factor reporter assay
SP1 reporter assay was performed with Cignal Luciferase Reporter Assay Kit (Qiagen) following manufacturer’s instructions and outlined in Supplementary Methods.
Immunoblots
Protein lysates were prepared from cell lines or frozen tissues using RIPA buffer (Thermo Fisher Scientific, Waltham, MA) supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific). Western Blot analyses were performed with standard techniques using the following antibodies: β-Actin, total Src, phosphoSrc (T416), phospho-ERK1/2 (T202/204), total and phospho-Akt (S473), total and phospho-S6 (S240/244), phospho-mTOR (S2448), phospho-4EBP1(T37/46), LC3B (all from Cell Signaling Technology), goat anti-human Osteoactivin/GPNMB antibody, p62 (both R&D Systems). Protein bands where visualized with IRDye secondary antibodies diluted in Odyssey blocking buffer containing 0.2% Tween and 0.1% SDS and signals were visualized with an Odyssey imager and analyzed with Image Studio software (all LI-COR Bioscience).
Immunohistochemistry, FISH analysis, and spectral karyotyping
Hematoxylin and eosin staining was performed by standard methods. Histology was reviewed by a pathologist experienced in evaluating kidney cancer. Immunohistochemistry for TFE3 and GPNMB was performed as previously described [
28]. Primary antibodies were as follows: Human Osteoactivin/GPNMB anti goat (R&D Systems; 1:800); TFE3 anti rabbit (Sigma Aldrich; 1:800). Spectral karyotyping and TFE3 was performed as previously described [
29] and outlined in Supplementary Methods.
Gene expression analysis by real-time-PCR
RNA was extracted from frozen tumor tissue and from cell lines with TRIzol™ Reagent (Thermo Fisher Scientific) according to manufacturer’s instructions and converted to cDNA with SuperScript IV VILO cDNA synthesis kit (Invitrogen). Gene expression was measured by Real-Time PCR using TaqMan® Gene Expression Assays on a ViiA7 Real-Time PCR system (Thermo Fisher Scientific) per manufacturer’s instructions. All assays were run in triplicate and gene expression was calculated as comparative CT (ΔΔCT) values. Gene expression was evaluated for GPNMB (Hs01095669_m1) and BIRC5 (Hs04194392_s1), using ACTB (Hs01060665_g1) as reference gene.
In Vivo studies
Tumor xenografts were generated by injecting UOK124 or UOK146 cells subcutaneously into flanks of athymic nude mice (Charles River). Mice were randomized into treatment groups (n = 10 mice/group) based on tumor volume and treated with the following agents, singly or in combination: (1) NVP-BGT226, (2) Mithramycin A, (3) Dasatinib, (4) Carfilzomib, (5) CDX-011, (6) respective vehicle. Details are outlined in Supplementary Methods. Body weights and tumor measurements were taken to determine response. Mouse survival was calculated as Log-rank test in Graphpad Prism and xenograft growth rates were compared based on rate-based T/C metric according to Hather et al. [
30] and as outlined in Supplementary Methods.
Statistical analysis
Values are expressed as mean ± standard deviation. Where appropriate, data were analyzed using a two-tailed t (parametric) test or Mann-Whitney (non-parametric) test, with a p value < 0.05 considered significant. All experiments were performed three times, with exception of the animal studies which were performed once using 2 cell lines.
Discussion
MiT-RCC constitutes a subset of primarily sporadic RCC characterized by genomic translocations leading to tumor-driving fusion proteins involving members of the MiT family of transcription factors [
4]. Although infrequently diagnosed in the adult population, MiT-RCCs represent a significant proportion of RCC in children and young adults. While these aggressive tumors have a propensity toward early metastasis to regional lymph nodes [
4], currently no effective standard of care therapy is available. Therapeutic regimens developed to treat metastatic ccRCC, such as multi-kinase inhibitors and immune checkpoint inhibitors, have shown limited response in patients with metastatic MiT-RCC [
11,
38,
39]. In an era of personalized medicine, the identification of a targeted therapeutic approach for the treatment of advanced MiT-RCC is needed. In the present study we have used a combination of an unbiased drug screening and biologically targeted approach.
The development of treatments for advanced rare tumor types always presents a challenge due to the scarcity of patients and, in this case, the severity of the disease and its rapid progression. For MiT-RCC patients, surgery remains the most effective treatment for localized disease since an effective systematic treatment for advanced, metastatic disease is still lacking. We have developed a series of TFE3 translocation RCC derived cell lines that represent different TFE3 fusions [
2,
22,
23], which were used here in a high-throughput broad spectrum drug screen and to evaluate a GPNMB-targeted antibody-drug conjugate therapy.
The initial high-throughput drug screen and subsequent 2D and 3D spheroid-based validations highlighted the PI3K/mTOR inhibitor NVP-BGT226, the RNA synthesis inhibitor Mithramycin A, and the SRC inhibitor Dasatinib as promising therapies for MiT-RCC. On-target effects were confirmed for all three agents in the in vitro cell line models. Evaluation of these drugs in mouse xenografts derived from two different TFE3-RCC cell lines demonstrated growth suppression in both models with NVP-BGT226 and Mithramycin A, but growth suppression in only one model with Dasatinib. The efficacy of the PI3K/mTOR inhibitor NVP-BGT226 correlates well with recent studies that proposed Akt/mTOR inhibition as a promising therapeutic target for TFE3-fusion RCC [
26,
40]. In line with these previous studies, NVP-BGT226 decreased the S-phase of the cell cycle while not significantly inducing apoptosis in TFE3-RCC cell lines [
26,
40]. These studies support the potential use of PI3K/mTOR inhibitors to treat MiT-RCC, most likely as a component of combination therapy.
This study confirms growth suppression by Mithramycin A in MiT-RCC cells. A selective growth-inhibitory effect of Mithramycin A has been observed in
FLCN-deficient tumor cells as well [
41],which are characterized by activated wild-type TFE3 [
20], suggesting that TFE3-driven tumor cells may be specifically sensitive to this drug. Mithramycin A is an anti-tumoral antibiotic that binds to GC-rich regions of DNA and inhibits RNA synthesis by blocking the DNA binding capacity of transcription factors, such as Specificity Protein 1 (SP1) [
42] or E2F [
43], thereby leading to reduced cell proliferation and survival. A known transcriptional target of SP1,
BIRC5 (survivin) [
44], was upregulated in TFE3-RCC. This study confirms the expected decrease in SP1 transcriptional activity including
BIRC5 expression in response to Mithramycin A treatment. The treated TFE3-RCC cells demonstrated a G2/M phase block in the cell cycle and increased apoptosis as was previously observed in ovarian carcinoma cells in response to a Mithramycin A analog [
43]. Mithramycin A has shown relatively high toxicity in patients at therapeutic doses and Mithramycin A analogs with better toxicity profiles, such as EC-8042 (EntreChem SL), have been developed [
34]. Although TFE3-RCC cell lines were less sensitive to EC-8042 in comparison to Mithramycin A, the improved toxicity profile suggests its potential use as a component of combination therapy without excessive toxicity. Additionally, these results support further investigation of additional mithralogs as potential therapeutic agents for MiT-RCC.
The unbiased high-throughput drug screen identified pharmacological targets for TFE3-RCC from an array of known anti-tumor agents but was not directed against specific dysregulated pathways in the tumor. Improved understanding of MiT-RCC tumor biology may elucidate specific and untested therapeutic targets or vulnerabilities. In this study the observed tumor specific expression of the cell surface marker GPNMB led to the preclinical evaluation of Glembatumumab vedotin (CDX-011), a fully human antibody-drug conjugate (ADC) that targets GPNMB and delivers a cytotoxic dolostatin-like tubulin inhibitor, Monomethyl auristatin E (MMAE) [
36,
37]. CDX-011 has been previously shown to be safe for clinical use and demonstrates some pharmacologic effects against breast cancer, melanoma, and osteosarcoma [
45‐
50]. CDX-011 induced growth inhibition of TFE3-RCC both in vitro and in vivo, while having no effect in vitro on a cell line derived from a clear cell RCC. In support of GPNMB as a therapeutic target in cancer, a previous study of ASPSCR1-TFE3 fusion-driven alveolar soft part sarcoma also demonstrated increased GPNMB expression, and
Gpnmb silencing in a mouse model of this disease inhibited cell migration, suggesting a role in metastasis [
51].
While NVP-BGT226, Mithramycin A and CDX-011 induced growth inhibition as single therapies in this study, dual agent combinations demonstrated improved responses compared to single agents in both TFE3-RCC xenograft models studied. Recent advances in treating RCC have placed a substantial emphasis on combination therapies, even as a first line therapy [
52]. Combination therapies allow for different tumorigenic pathways to be targeted simultaneously with an increased likelihood of therapeutic response and decreased opportunity for cancer cells to develop resistance. Advanced MiT-RCC is very aggressive and may benefit greatly from combination therapy if the side effects can be limited, for example, by using precision therapies such as CDX-011.
A limitation of this study is the use of cell line models in evaluating potential therapies. These cell lines models are artificially cultured, may have acquired additional genetic alterations, and may not accurately represent the wide variety of tumors classified as MiT-RCC. For MiT-RCC, this is particularly notable as the tumors are driven by gene fusions involving several members of the MiTF gene family fused to a diverse number of other partner genes. To partially counteract some of these disadvantages, this study utilized multiple cell lines that represented three different TFE3 fusion partners, NONO, PRCC, and SFPQ. Several other fusion combinations have been identified and it is possible that these other gene fusions may not respond in the same way to the potential therapies described in this study. Even the subset of TFE3-RCC tumors show high clinical and histological variability, which may in part correlate with the TFE3 fusion partner. Successful drug design and treatment strategies will therefore require a personalized medicine approach involving the molecular identification of the MiT-RCC subtype and potentially additional biomarkers (e.g., GPNMB expression levels) for each patient. Further investigations using patient derived xenografts (PDXs), potentially in humanized mice, may provide a more comprehensive representation of MiT-RCC in preclinical studies.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.