Abstract
Background Resistance to endocrine therapy in estrogen receptor-positive (ER+) breast cancer remains a significant clinical problem. Riluzole is FDA-approved for the treatment of amyotrophic lateral sclerosis. A benzothiazole-based glutamate release inhibitor with several context-dependent mechanism(s) of action, Riluzole has shown anti-tumor activity in multiple malignancies, including melanoma, glioblastoma, and breast cancer. In several (but not all) of these studies, Riluzole-mediated growth inhibition is attributed to increased expression of metabotropic glutamate receptors (mGluRs, GRMs). We recently reported that acquisition of Tamoxifen resistance in a cellular model of invasive lobular breast cancer is accompanied by the upregulation of GRM mRNA expression and growth inhibition by Riluzole.
Methods In the current study, we tested the ability of Riluzole to reduce cell growth, alone and in combination with endocrine therapy, in a diverse set of ER+ invasive ductal and lobular breast cancer-derived cell lines, primary breast tumor explant cultures, and the estrogen-independent, ESR1-mutated invasive lobular breast cancer patient-derived xenograft model HCI-013EI.
Results Single-agent Riluzole suppressed the growth of ER+ invasive ductal and lobular breast cancer cell lines in vitro, inducing a histologic subtype-associated cell cycle arrest (G0-G1 for ductal, G2-M for lobular). In an invasive lobular, endocrine resistant model, Riluzole induced apoptosis and reduced phosphorylation of multiple pro-survival signaling molecules, including Akt/mTOR, CREB, and Src/Fak family kinases. Riluzole in combination with either Fulvestrant or 4-hydroxytamoxifen additively or synergistically suppressed ER+ breast cancer cell growth in vitro. The combination of Riluzole plus Fulvestrant significantly reduced proliferation in primary breast tumor explant cultures, and inhibited HCI-013EI xenograft growth in vivo significantly earlier than Fulvestrant alone.
Conclusions These findings suggest Riluzole combined with endocrine therapy may offer therapeutic benefit in diverse ER+ breast cancers, including lobular breast cancer.
Background
Estrogen receptor-positive (ER+) breast cancer is the most commonly diagnosed cancer among women in the United States [1]. Endocrine therapies ranging from selective estrogen receptor modulators and downregulators (SERMs, SERDs) to aromatase inhibitors are the backbone of our current standard of care for clinical management of ER+ breast cancers [2]. Although these treatments have significantly improved disease-free and overall survival for individuals with ER+ breast cancer, endocrine resistance remains a persistent, multifactorial problem [3] that current efforts aim to address through combination therapy with endocrine agents plus a myriad of other molecularly targeted therapies. Further complicating these efforts is the fact that ER+ breast cancer is not a single disease. Invasive lobular breast cancer (ILC) is a distinct histologic subtype of breast cancer that is overwhelmingly ER+ yet has distinct genomic, transcriptomic and proteomic features [4–6]. These distinctions have important implications for endocrine therapy response, since when compared to the more common invasive ductal breast cancer (IDC), ILC carries a greater risk for late recurrence (evident >6 years after initial diagnosis) [7,8], and is less responsive to the SERM Tamoxifen [9,10] and potentially the steroidal aromatase inhibitor exemestane [11]. A recent preclinical study suggests models of ILC are less responsive to the second-generation SERD AZD9496 than Fulvestrant, while these drugs are equipotent in preclinical models of IDC [12].
Our group [13,14] and others [12,15–20] have identified a number of potential mechanisms that contribute to endocrine therapy resistance in ILC. We recently identified upregulation of multiple metabotropic glutamate receptors (mGluRs, GRMs) in Tamoxifen-resistant ILC cells [14]. This, coupled with other studies that directly or indirectly implicate altered amino acid metabolism and signaling in ILC pre-clinical models [21] and clinical disease [22,23], led us to consider whether glutamate signaling is functionally relevant to endocrine resistance in this setting. Initially in melanoma [24,25], and now other malignancies [26], it has been shown that pro-tumorigenic signaling through GRMs can be inhibited by Riluzole, a benzathiole-based glutamate release inhibitor that is FDA-approved for the treatment of amyotrophic lateral sclerosis (ALS). Riluzole’s proposed mechanism of action in the central nervous system in ALS and in melanoma is that, by blocking glutamate release into the extracellular space, GRMs are starved of their glutamate ligand and thus functionally inhibited; this reduces glutamate excitotoxicity and inhibits tumor cell growth, respectively. However, in breast cancer, Riluzole action is not dependent on GRMs [27,28], although GRMs are implicated in breast tumor progression and Riluzole can still exert anti-tumor effects [29–31].
ER+ breast cancer, and especially ILC, has not been a major focus for prior studies of Riluzole repurposing in cancer. Therefore, the goal of this study is to more broadly test the efficacy of Riluzole, alone and in combination with multiple endocrine therapies, in a diverse set of ER+ in vitro and in vivo models enriched for ILC.
Methods
Cell Culture and Reagents
SUM44 and LCCTam cells were maintained under serum-free conditions in improved minimal essential media (IMEM, #A1048901, ThermoFisher, Grand Island, NY) supplemented with insulin, hydrocortisone, and other supplements as previously described [13], with the addition of 500 nM 4-hydroxytamoxifen (#H7904, Sigma Aldrich, St. Louis, MO) to LCCTam cells. For selected experiments, SUM44 and LCCTam cells were maintained under serum-free conditions in a base media of Ham’s F12 (#11765062, ThermoFisher) supplemented as above. MDA-MB-134VI cells were maintained in IMEM supplemented with 5% fetal bovine serum (FBS). BCK4 cells were maintained in IMEM supplemented with 10% FBS, insulin, and nonessential amino acids as previously described [32]. MCF7 and LCC9 cells were maintained in IMEM supplemented with 5% FBS, or phenol red-free IMEM supplemented with 5% charcoal-cleared serum (CCS), respectively. For selected experiments, MDA-MB-134VI and MCF7 cells were short-term hormone-deprived by culturing in phenol red-free IMEM (#A1048801, ThermoFisher) supplemented with 5% CCS for 72 hours. The immortalized mammary epithelial cell line MCF10A was maintained as previously described [33] All cell lines were authenticated by short tandem repeat (STR) profiling and regularly tested to ensure that they remained free of Mycoplasma spp. contamination. Unless otherwise noted, general cell culture supplements and reagents were purchased from either ThermoFisher or Sigma Aldrich. Fulvestrant and Riluzole were purchased from Sigma Aldrich, Tocris Bio-Techne (Minneapolis, MN) or Selleckchem (Houston, TX).
Cell Proliferation Assays
On Day 0 cells were seeded in 96-well plates at the following densities: 1,000 cells/well (MCF7); 2,000 cells/well (LCC9, MCF10A); 10,000 cells/well (SUM44, LCCTam, MDA-MB-134VI); 15,000 cells/well (BCK4). Forty eight hours later, on Day 2, cells were treated with the indicated concentration of compound(s) or solvent control (DMSO) for an additional 7 or 8 days, with media replaced and fresh compound(s) on Day 5 or 6. Plates were then stained with crystal violet, dried, rehydrated, and read as previously described in [14]. Data are presented as mean ± standard error of the mean (SEM, Riluzole growth curves) or median with upper/lower quartiles (effect of 10 μM Riluzole on cell line pairs) of % growth (vehicle = 100%) for 5-6 technical replicates, and are representative of 2-4 independent biological assays. For assays of Riluzole in combination with Fulvestrant, data are presented as median with upper/lower quartiles of % growth for 5-6 technical replicates, and are representative of 2-4 independent biological assays. For assays of Riluzole in combination with 4-hydroxytamoxifen, data are presented as the mean of % growth for 5-6 technical replicates, and are representative of 2-4 independent biological assays.
Cell Cycle Assays
On Day 0 cells were seeded in 6-well plates at the following densities: 150,000 cells/well (MCF7, LCC9); 300,000 cells/well (SUM44, LCCTam, MDA-MB-134VI, BCK4, MCF10A). Forty eight hours later, on Day 2, cells were treated with 10 μM Riluzole or DMSO control for the additional indicated times before collection, fixation, staining, and cell cycle analysis by flow cytometry as described in [34]. Data are presented as mean ± standard deviation (SD) for 3-4 independent biological assays.
Annexin V Apoptosis Assays
On Day 0, cells were seeded in a 6-well plate at 300,000 cells/well (SUM44, LCCTam). Forty eight hours later, on Day 2, cells were treated with either 10 μM Riluzole or the solvent, DMSO, as a control for an additional 48 hours. On Day 4, cells were collected and stained with 4μL of propidium iodide (PI) and 4μL of annexin V conjugated with fluorescein isothiocyanate (FITC) in 100μL of 1X binding buffer. Control cells were either left unstained or stained with either PI or annexin V dye conjugated with FITC as described in [35]. Live (PI-, annexin V-), early apoptotic (PI-, annexin V+), late apoptotic(PI+, annexin V+), and necrotic cells (PI+, annexin V-) were quantified by flow cytometry. The PI/Annexin V-FITC apoptosis detection kit was purchased from BioLegend (#640914, San Diego, CA, USA). Data are presented as mean ± SD for 3 (SUM44) or 4 (LCCTam) independent biological assays.
Human Phospho-Kinase Proteome Profiler™ Array
On Day 0, cells were seeded in a 6-well plate at 500,000 cells/well (SUM44, LCCTam). Forty eight hours later, on Day 2, cells were treated with either 10 μM Riluzole or the solvent, DMSO, as a control for an additional 48 hours. On Day 4, cells were collected in 100 μL lysis buffer/well prior to determining total protein concentration by bicinchoninic acid (BCA) assay (#23225, ThermoFisher). Five hundred micrograms of whole cell lysate were then assayed using the Human Phospho-Kinase Proteome Profiler™ Array (#ARY003B, Bio-Techne) according to manufacturer’s instructions. Array membranes were visualized using chemiluminescence detected by HyBlot CL autoradiography film (#E3018, Thomas Scientific, Swedesboro, NJ), then films were scanned and analyzed using FIJI [36]. A ratio of background-corrected intensity values for targets (phospho-kinase spots) to references (control spots) was created for each condition (DMSO and Riluzole) within each cell line. Data are presented as the mean of the Riluzole:DMSO ratio for 2 technical replicates from a single experiment for each cell line. Gene symbols corresponding to the kinases showing decreased phosphorylation in response to Riluzole in LCCTam cells were analyzed using Metascape [37] to identify top functional enrichments.
Primary Breast Tumor Explant Cultures
Patient-derived explants (PDEs) from five (5) ER+ primary breast tumors were processed and cultured as described in [38]. PDEs were treated with 100 nM Fulvestrant, 10 μM Riluzole, the combination, or solvent control (DMSO) for 48 hours prior to formalin fixation, paraffin embedding, sectioning, and staining for PCNA (1:1000, #sc-56, SCBT, Santa Cruz, CA), cleaved caspase 3 (1:300, #9661, Cell Signaling Technology, Danvers, MA), and Ki67 (1:500, #ab16667, Abcam, Cambridge, MA). Stained sections were then visualized and scored as described in [38]. Data are presented as change relative to vehicle (set to 0) for each of the PDEs.
HCI-013 and HCI-013EI Xenograft Experiments
All animal studies were ethically conducted in accordance with our approved Institutional Animal Care and Use Committee (IACUC) protocols #2018-0005 and #2018-0006. For the comparison of time to tumor formation between HCI-013 and HCI-013EI in the presence vs. absence of supplemental estrogen pellets, twenty four (24) 5-6 week-old intact (non-ovariectomized) female non-obese diabetic, severe combined immunodeficient mice (NOD.CB17-Prkdcscid/NCrCrl, purchased from Charles River, Wilmington, MA) were orthotopically implanted in the right 4th mammary gland with a single 1-3 mm3 PDX fragment per mouse as follows: HCI-013EI (n=6); HCI-013EI+E2 (n=6); HCI-013 (n=6); HCI-013+E2 (n=6). “+E2” denotes co-implantation of a 1 mg estrogen pellet under the skin on the back between the shoulder blades. Mice were followed until measurable tumor development (by calipers), and data are presented as a survival plot with n=6 mice per group. For treatment studies of HCI-013EI tumors, forty eight (48) 5-6 week-old intact (non-ovariectomized) female mice were orthotopically implanted in the right 4th mammary gland with a single 1-3 mm3 HCI-013EI PDX fragment per mouse without estrogen supplementation, then followed until tumors reached ~100 mm3 before enrollment to one of the four (4) treatment arms: control (n=8), 10 mg Fulvestrant in castor oil SQ (2 5-mg injections per week, n=9), 10 mg/kg Riluzole PO in corn oil (5 days per week, n=9), or the combination (n=9). Mice were monitored for tumor growth (measured by calipers) and body weight twice per week. Tumor volumes were calculated by the modified ellipsoid formula V=1/2(XY2), where X is the longest axis and Y is the longest perpendicular axis. Tumor volume data are presented as mean ± SEM for n=8-9 mice per treatment group. The baseline measurement represents the measurement at the point of enrollment to a treatment group based on the a priori tumor volume as calculated above. The subsequent measurements are those taken while on treatment at the twice per week frequency. The slope of tumor growth for each group is presented as mean ± SD for the same tumors.
Multiplex Immunohistochemistry (mIHC) Staining
HCI-013+E2 (n=5) and HCI-013EI (n=4) tumors from mice in our PDX maintenance colony – independent from experimental animals described above – were resected, formalin fixed, paraffin embedded, then sectioned for staining on the Vectra3 multi-spectral imaging platform (Akoya Biosciences, Marlborough, MA) using OPAL chemistry. The slides were baked at 60°C, deparaffinized in xylene, rehydrated, washed in tap water and incubated with 10% neutral buffered formalin for an additional 20 minutes to increase tissue-slide retention. Epitope retrieval/microwave treatment (MWT) for all antibodies was performed by boiling slides in Antigen Retrieval buffer 6 (AR6 pH6; Akoya, AR6001KT). Protein blocking was performed using antibody diluent/blocking buffer (Akoya, ARD1001EA) for 10 minutes at room temperature. Primary antibody/OPAL dye pairings and incubation conditions for ER, PR, HER2, Ki67, and pan cytokeratin staining are detailed in Table 1. MWT was performed to remove the primary and secondary antibodies between rounds of multiplex IHC. Multiplex IHC was finished with MWT and counterstained with spectral DAPI (Akoya FP1490) for 5 min and mounted with ProLong Diamond Antifade (ThermoFisher, P36961). The order of antibody staining and the antibody/OPAL pairing was predetermined using general guidelines and the particular biology of the panel. General guidelines include spectrally separating co-localizing markers and separating spectrally adjacent dyes. Multiplex IHC was optimized by first performing singleplex IHC with the chosen antibody/OPAL dye pair to optimize signal intensity values and proper cellular expression, followed by optimizing the full multiplex assay.
mIHC Imaging and Analysis
Slides were scanned at 10X magnification using the Vectra 3.0 Automated Quantitative Pathology Imaging System (PerkinElmer/Akoya,). Whole slide scans were viewed with Phenochart (Perkin Elmer/Akoya) which also allows for the selection of high-powered images at 20X (resolution of 0.5m per pixel) for multispectral image capture. Multispectral images of each xenograft tissue specimen were captured in their entirety. Multispectral images were unmixed using spectral libraries built from images of single stained tissues for each reagent using the inForm Advanced Image Analysis software (inForm 2.4.6; PerkinElmer/Akoya). A selection of 10-15 representative multispectral images spanning all nine tissue sections was used to train the inForm software (tissue segmentation, cell segmentation and phenotyping tools). All the settings applied to the training images were saved within an algorithm to allow batch analysis of all the multispectral images particular to each panel. Data are presented as median with upper/lower quartiles of % marker positivity for 5-13 fields per tumor, and as overall mean ± SD of % marker positivity for all tumors.
Statistical Analysis
Statistical analyses were performed using GraphPad Prism 8.0 (San Diego, CA) at α≤0.05, except for Riluzole/4-hydroxytamoxifen combination experiments, which were analyzed by synergyfinder [39]. Single-agent Riluzole experiments were analyzed by nonlinear regression ([inhibitor] vs. normalized response), and response to 10 μM Riluzole in endocrine therapy sensitive/resistant cell line pairs (SUM44 vs. LCCTam and MCF7 vs. LCC9) was compared by Mann-Whitney test. Riluzole/Fulvestrant combination experiments were analyzed by Kruskal Wallis test followed by Dunn’s multiple comparison test. Riluzole/4-hydroxytamoxifen combination experiments were analyzed using the zero interaction potency (ZIP) method [40] in synergyfinder. Cell cycle experiments were analyzed by two-way Analysis of Variance (ANOVA) followed by either Sidak’s (single time point) or Dunnett’s (multiple time points) multiple comparisons tests. Annexin V apoptosis assays were analyzed by two-way ANOVA followed by Sidak’s multiple comparisons test. Staining for each marker in primary breast tumor explant cultures was analyzed by one-sample t test vs. 0 (vehicle). In the xenograft experiment comparing time to tumor formation between HCI-013 and HCI-013EI in the presence vs. absence of supplemental estrogen pellets, data were analyzed by log-rank (Mantel-Cox) test. In the xenograft experiment testing Fulvestrant, Riluzole, the combination, or control in HCI-013EI, tumor volume and mouse body weight data were analyzed by mixed effects analysis followed by Dunnett’s multiple comparisons tests at each timepoint vs. control, while differences in the slope of tumor growth between treatment groups was compared by Browne-Forsyth and Welch ANOVA followed by Dunnett’s T3 multiple comparisons tests. mIHC data were analyzed by Mann-Whitney test.
Results
Riluzole has shown anti-tumor activity in preclinical models of multiple cancers, including melanoma, glioblastoma, and breast cancer [25–31]. In several (but not all) of these reports, Riluzole-mediated growth inhibition is attributed to increased expression of metabotropic glutamate receptors (mGluRs, GRMs). We recently reported that acquisition of Tamoxifen resistance in a cellular model of invasive lobular breast cancer (ILC, [13]) is accompanied by the upregulation of GRM mRNA expression and growth inhibition by Riluzole [14]. Here, our goal was to more broadly test the efficacy of Riluzole, alone and in combination with multiple endocrine therapies, in a diverse set of estrogen receptor-positive (ER+) in vitro and in vivo models enriched for ILC. The in vitro models used include ER+ ILC cell lines (SUM44, LCCTam, MDA-MB-134VI, and BCK4), ER+ IDC cell lines (MCF7 and LCC9), and the ER-negative (ER-) non-transformed cell line MCF10A as a control. The in vivo models include a series of ER+ patient-derived explants, and patient-derived xenografts (HCI-013 and HCI013EI).
Growth suppression of ER+ breast cancer cell lines by Riluzole
We performed dose-response assays of Riluzole (33 nM to 100 μM) in three ILC- and two invasive ductal breast cancer (IDC)-derived cell lines, and the ER-non transformed breast epithelial cell line MCF10A, using crystal violet staining as a proxy for total cell number [14] (Figure 1A). Nonlinear regression analysis calculated the Riluzole IC50 for all six cell lines to be ~10-100 μM, consistent with published studies in other malignancies. Direct comparison of growth inhibition by 10 μM Riluzole in two endocrine-responsive and -resistant cell line pairs (Figure 1B) confirmed [14] that the Tamoxifen-resistant ILC cell line LCCTam was significantly more responsive to Riluzole than its Tamoxifen-responsive counterpart SUM44 (Mann-Whitney test, **p=0.002). This was not the case for the MCF7/LCC9 IDC cell line pair [41], in which MCF7 cells showed greater Riluzole-mediated growth inhibition (*p=0.024) than Fulvestrant-resistant/Tamoxifen-cross-resistant LCC9 cells. The MDA-MB-134VI (MM134) ILC cell line showed a similar response as SUM44, while MCF10A cells were not growth inhibited by 10 μM Riluzole vs. DMSO control.
Response of ER+ cell lines to growth inhibition by small molecules can be influenced by the presence vs. absence of steroid hormones and estrogenic compounds. Experiments presented in Figure 1 were, except for LCC9 cells, performed under hormone-replete conditions, so assays in selected cell lines were repeated under hormone-deprived conditions (Figure S1). The SUM44/LCCTam cell line pair is cultured in serum-free media, but a phenol red-containing base (IMEM, 10 mg/L), while MCF7 and MM134 cells are cultured in phenol red-containing IMEM supplemented with 5% FBS. While individual differences within cell lines were observed, hormone deprivation – reduced phenol red media for SUM44/LCCTam (Ham’s F12, 1.2 mg/L) or phenol red-free IMEM supplemented with 5% CCS for MCF7 and MM134 – did not consistently enhance or impair Riluzole-mediated growth inhibition.
Riluzole induces a histologic subtype-associated cell cycle arrest
To corroborate results of the cell proliferation assays, we tested the effect of Riluzole on cell cycle progression (Figure 2), and expanded our studies to include BCK4, a third model of ER+ ILC [32]. All ILC cell lines showed a significant accumulation of cells in G2-M phase (two-way ANOVA followed by Sidak’s (single time point) or Dunnett’s (multiple time points) multiple comparisons tests, see figure legend). However, both IDC-derived cell lines showed a significant accumulation of cells in G0-G1 phase, while non transformed MCF10A cells showed no significant cell cycle arrest in response to Riluzole. Together with the results presented in Figures 1 and S1, these data suggest that while all ER+ breast cell lines tested can be growth inhibited by Riluzole, ILC cells preferentially undergo G2-M arrest while IDC cells arrest in G0-G1.
Riluzole induces apoptosis and inhibits phosphorylation of pro-survival signaling molecules
We then performed Annexin V assays to measure the effect of Riluzole on apoptosis in the SUM44/LCCTam cell line pair. The percentage of live, early apoptotic, late apoptotic, and necrotic cells were measured in response to Riluzole for each cell line (Figure 3A), and the percentage of early apoptotic cells were regraphed separately (Figure 3B). LCCTam cells showed a significant increase in the percent of cells in early apoptosis when treated with Riluzole (two-way ANOVA followed by Sidak’s multiple comparisons test, ****p<0.0001), but SUM44 cells did not. These data are consistent with those presented in Figure 1B, which show that LCCTam cells were significantly more growth-inhibited by Riluzole than SUM44 cells.
To identify molecular signaling events that accompany Riluzole-mediated apoptosis in the SUM44/LCCTam cell line pair, we used the Human Phospho-Kinase Proteome Profiler™ Array to detect changes in 43 phosphorylation sites across 40 different kinases or substrates (Figure 3C). In SUM44 cells, Riluzole reduced phosphorylation of mutant p53 [42] (S92 and S392), and Akt T308. In LCCTam cells, Riluzole reduced phosphorylation of markedly more sites, in kinases and substrates with significantly enriched ontology clusters including EGFR (ERBB1) and MET signaling, and cellular response to hormone stimulus (Figure S2A). Notable inhibition of Akt/mTOR (Akt S437, TOR S2448, PRAS40 T246), CREB (Msk S376/360, CREB S133), and Src/Fak (Lyn Y397, Yes Y426, Fak Y397, Figure S2B) signaling pathways was observed specifically in LCCTam cells. Prior studies in melanoma and glioblastoma have shown that Riluzole inhibits Akt phosphorylation, and that Riluzole combined with mTOR inhibition can synergistically decrease xenograft growth [26,43]. However, inhibition of Src/Fak family kinases has not to our knowledge been previously reported.
Additive/synergistic suppression of ER+ breast cancer cell line growth by Riluzole in combination with endocrine therapies
Endocrine therapies ranging from selective estrogen receptor modulators and downregulators (SERMs, SERDs) to aromatase inhibitors represent the standard of care for clinical management of ER+ breast cancers [2]. We therefore tested the activity of Riluzole in combination with the SERD Fulvestrant or SERM Tamoxifen (4-hydroxytamoxifen) in ILC- and IDC-derived ER+ breast cancer cell lines, and the ER-non transformed breast epithelial cell line MCF10A as a negative control. These experiments were conducted under hormone-replete conditions. In the ILC-derived SUM44, LCCTam, MM134, and BCK4 cell lines, the combination of Fulvestrant plus Riluzole caused significantly additive or better growth inhibition (Figure 4, Kruskal Wallis test followed by Dunn’s multiple comparison test, see figure legend). In the IDC-derived MCF7 cell line, Fulvestrant alone was highly effective, so the addition of Riluzole caused no further decrease in cell growth. LCC9 cells were resistant to Fulvestrant, and the combination of Fulvestrant plus Riluzole was not better than Riluzole alone. MCF10A cells were not significantly growth-inhibited by either drug or the combination. Combinations of 4-hydroxytamoxifen and Riluzole (Figure S3) were moderately to strongly synergistic [39,40] in SUM44, LCCTam, and MCF7 cells, but not LCC9 cells. MCF10A cells were again not significantly growth-inhibited by either drug or the combination. Altogether, these data suggest that the combination of endocrine therapy and Riluzole can additively or synergistically suppress the growth of a variety of ER+ breast cancer cell line models.
Riluzole plus Fulvestrant inhibits proliferation in primary breast tumor explant cultures
There are limitations to pre-clinical testing of combination therapies in two-dimensional (2D) cell culture models on tissue culture plastic using established cell lines. Patient-derived explants (PDEs) allow medium-throughput testing of short-term response to therapeutic agents in fragments of fresh tumor tissue, thereby overcoming some of these limitations. Specifically, PDEs capture and preserve the initial tissue architecture, tumor microenvironment, and inter-person heterogeneity [38]. We tested the efficacy of Fulvestrant, Riluzole, or the combination vs. solvent control (DMSO) in five (5) PDEs from ER+/PR+/HER2-negative primary tumors (Figure 5A). Using proliferating cell nuclear antigen (PCNA) staining as a proxy for cell proliferation, the combination of Riluzole plus Fulvestrant significantly reduced PCNA (Figures 5B and 5C, one-sample t test vs. 0 (Vehicle), *p=0.013 Vehicle vs. Combination), with 4 of 5 PDEs showing better growth inhibition by the combination than either drug alone. Staining for Ki67 was inconsistent in response to either drug alone or the combination, while staining for cleaved caspase 3 suggested a modest induction of apoptosis by either drug alone or the combination in some of the PDEs (Figure S4).
Riluzole plus Fulvestrant inhibits HCI-013EI tumor growth earlier than Fulvestrant alone
Patient-derived xenografts (PDXs) provide a second clinically relevant alternative to 2D culture models for pre-clinical testing of combination therapies. The HCI-013 PDX model was established from a 53 year-old woman with metastatic, multi-therapy-resistant ER+/PR+/HER2-ILC by serial passage through intact (non-ovariectomized) female NOD scid gamma (NSG) mice supplemented with a 1 mg estrogen (E2, 17β-estradiol) pellet [15]. The HCI-013EI (estrogen-independent) variant was established by two weeks of in vitro culture of cells from HCI-013 tumors under hormone-deprived conditions, then reimplanted into intact female NSG mice without estrogen supplementation. Both models harbor the clinically relevant ESR1 activating mutation Y537S, with the HCI-013EI variant reported to have a more abundant variant allele fraction of Y537S.
To directly compare the responsiveness to, and requirement for, supplemental estrogen of HCI-013 vs. HCI-013EI in our hands, six (6) 5-6 week-old intact (non-ovariectomized) severe combined immunodeficient (SCID) female mice per group were orthotopically implanted with a single 1-3 mm3 PDX fragment, then followed until tumor development measurable by calipers (Figure 6A). In the presence of supplemental estrogen pellets, HCI-013 and HCI-013EI both exhibited a 100% tumor take rate, with a median time to tumor formation of 23.5 and 26.5 days, respectively. However, in the absence of supplemental estrogen pellets, HCI-013 PDX fragments were unable to form tumors out to 113 days post implantation, while HCI-013EI PDX fragments exhibited a 50% tumor take rate, with a median time to tumor formation of 39 days (log-rank Mantel-Cox test, ***p=0.0007). These data suggest that supplemental estrogen is necessary for HCI-013 tumor formation, and beneficial but not necessary for HCI-013EI tumor formation, in SCID mice.
We characterized an independent set of HCI-013+E2 (estrogen supplemented, n=5) and HCI-013EI (not estrogen supplemented, n=4) tumors with respect to ER, PR, HER2, and Ki67 expression using Opal chemistries on the Vectra3 multi-spectral imaging platform (Figures 6B and 6C). This approach captured heterogeneity in marker expression between and within tumors. Overall, percent ER positivity (% ER+) was significantly lower in HCI-013EI vs. HCI-013+E2 tumors (Mann-Whitney test, *p=0.032), consistent with a prior report that Y537S mutant ER protein expression can be lower than wild type ER [44]. However, overall percent PR and Ki67 positivity was not significantly different between these PDX variants. No HER2 staining was detected.
We moved forward with the HCI-013EI (not estrogen supplemented) PDX variant to test the anti-tumor activity of Fulvestrant, Riluzole, or the combination relative to vehicle control. Forty eight (48) 5-6 week-old intact SCID female mice were orthotopically implanted with a single 1-3 mm3 HCI-013EI PDX fragment without E2 supplementation, then followed until tumors reached ~100 mm3 before enrollment to one of four (4) treatment arms: control (n=8), Fulvestrant (n=9), Riluzole (n=9), or the combination (n=9) (Figures 6D and S5A). Relative to control, single-agent Riluzole had no effect on tumor volume. Single-agent Fulvestrant and the combination of Fulvestrant plus Riluzole were similarly effective, as both significantly slowed the increase in tumor volume. However, combination treatment yielded a statistically significant growth delay earlier (Measurement 2, ~Day 7) than Fulvestrant alone (Measurement 5, ~Day 17, mixed effects analysis followed by Dunnett’s multiple comparisons tests at each timepoint vs. control, see figure legend). Comparison of the slope of tumor growth for the treatment groups further demonstrated a statistically significant difference between Fulvestrant-vs. combination-treated tumors (Figure 6E, Browne-Forsyth and Welch ANOVA followed by Dunnett’s T3 multiple comparisons tests, *p=0.023), with combination-treated tumors having a more negative slope suggestive of greater tumor shrinkage. Analysis of mouse body weights between the treatment groups showed that there were no significant differences between treatment groups. As seen in Figure S5B, the slope of the graphs for each treatment group is close to zero. Together with the results presented in Figures 4, 5, and accompanying supplementary figures, these data suggest that the combination of Fulvestrant and Riluzole may offer improved therapeutic benefit in ER+ breast cancer.
Discussion
We tested the efficacy of Riluzole, alone and in combination with multiple endocrine therapies, in a diverse set of estrogen receptor-positive (ER+) in vitro and in vivo models enriched for ILC. Single-agent Riluzole suppressed the growth of ER+ ILC and IDC cell lines in vitro, inducing a histologic subtype-associated cell cycle arrest (G0-G1 for IDC, G2-M for ILC). In the Tamoxifen-resistant ILC-derived LCCTam model, Riluzole induced apoptosis and reduced phosphorylation of multiple pro-survival signaling molecules, including Akt/mTOR, CREB, and Src/Fak family kinases. Riluzole in combination with either Fulvestrant or 4-hydroxytamoxifen additively or synergistically suppressed ER+ breast cancer cell growth in vitro. The combination of Riluzole plus Fulvestrant significantly reduced proliferation in primary breast tumor explant cultures, and inhibited HCI-013EI ILC PDX growth in vivo significantly earlier than Fulvestrant alone.
The increased combinatorial efficacy of Riluzole and Fulvestrant we observe across diverse cell line models of ER+ breast cancer in vitro is recapitulated in vivo, using primary tumor explant cultures (PDEs) and (to a lesser extent) the ILC-derived HCI-013EI PDX. In the PDX experiment, tumor volume after ~4 weeks of treatment was not significantly different in combination-vs. Fulvestrant-tumors, which we attribute to the good responsiveness of this PDX to Fulvestrant demonstrated in this experiment and by others [45,46]. It should be noted, however, that these published studies report Fulvestrant responsiveness of HCI-013, and not specifically HCI-013EI. However, the combination of Riluzole and Fulvestrant led to a significant growth inhibition vs. control ~10 days earlier than Fulvestrant alone, and inspection of the individual volume trajectories for each tumor shows more consistent suppression in the combination group as compared to Fulvestrant alone (Figure S5A). Furthermore, the slope of tumor growth for the combination was more strongly negative (−4.174, suggestive of tumor shrinkage) and significantly reduced vs. Fulvestrant monotherapy (−1.69, Figure 6E). In the PDE experiment, the combination of Riluzole and Fulvestrant was more effective than in the PDXs, with 80% of primary tumor explants (4/5) showing significant growth inhibition by the combination as measured by a reduction in PCNA (Figure 5), but not Ki67 (Figure S4). Improved combinatorial efficacy may be due to the lower concentration of Fulvestrant used in these studies (100 nM), so future in vivo studies with the HCI-013EI PDX will adopt this approach (i.e. a lower dose of Fulvestrant, as in [45], alone and in combination with Riluzole). A second contributor to improved combinatorial efficacy in the PDE vs. PDX experiment may be Riluzole bioavailability and metabolism (see below). The discrepancy in response to treatment for the PCNA and Ki67 measures is likely due to the fact that baseline Ki67 positivity in vehicle-treated PDEs is low (3.8% −11.6%), which limits the effective range of an assay designed to measure a (further) decrease in Ki67. By contrast, baseline PCNA expression is robust (48.2%-92.8%), so decreases in expression may be more readily detected. An alternative or complementary explanation is that Riluzole induces G2-M arrest in most malignancies tested, and in all of our ER+ ILC cell lines, but a G0-G1 arrest in the ER+ MCF7-derived IDC cell lines (Figure 2). Riluzole’s tendency to induce G2-M arrest may partly explain why a reduction in Ki67 is not observed, since Ki67 expression accumulates in the S through M phases of the cell cycle [47], while by contrast PCNA expression is maximal in late G1 and S phases.
In the Tamoxifen-resistant ILC-derived LCCTam model, Riluzole induces apoptosis concomitant with reduced phosphorylation of multiple pro-survival signaling molecules. That these include components of the Akt/mTOR signaling pathway (Akt S437, TOR S2448, PRAS40 T246) is not surprising, since Riluzole has been shown to inhibit Akt phosphorylation and synergize with mTOR inhibition in melanoma and glioblastoma models [26,43]. However, Src/Fak kinase family members Lyn, Yes, and Fak have not, to our knowledge, been previously implicated in Riluzole action. In vehicle-treated LCCTam cells, baseline phosphorylation of these kinases is increased (Lyn Y397 1.75-fold, Fak Y397 1.6-fold, Yes Y426 1.85-fold) as compared to vehicle-treated SUM44 cells, and each of these sites is markedly dephosphorylated in response to Riluzole treatment only in LCCTam cells, with Yes Y426 showing the largest Riluzole-induced decrease (~3-fold, Figure S2B). All three of these Src/Fak kinase family members are closely connected to cell survival, invasion and migration, while Lyn and Fak have also been previously implicated in endocrine therapy resistance [48,49]. Being functionally E-cadherin-negative, ILC is particularly resistant to anoikis (a form of cell death induced by extracellular matrix detachment) [50], and becomes highly dependent upon a rewired actin cytoskeleton and constitutive actomyosin contractility (reviewed in [51]). A notable downstream target of Yes whose function is also tightly regulated by actin dynamics (Yes-associated protein 1, YAP1) is more frequently activated and localized to the nucleus in ILC [52], and our reanalysis of The Cancer Genome Atlas (TCGA) reverse phase protein array data [53,54] confirms that total YAP1 expression is significantly increased in ER+ ILC vs. IDC (log2 fold-change 0.21, q-value 4.4e-6). The role of Yes or YAP1 in Riluzole-mediated growth inhibition and apoptosis remains to be explored and will be a component of future studies, as will the role of Yes/YAP1 and other Src/Fak family kinases in the additive or synergistic growth suppression achieved by the combination of Riluzole and either Fulvestrant or 4-hydroxytamoxifen. In the PDE experiment, staining for cleaved caspase 3 demonstrates a modest induction of apoptosis by either drug alone or the Riluzole-Fulvestrant combination in some of the PDEs (Figure S4), with the ILC PDE (#1055) exhibiting apoptosis only when treated with the combination.
This study has limitations that are important to consider. First, hormone responsive ER+ models, particularly of ILC, are limited. In this study we used three of the four well-established ER+, hormone responsive ILC lines [50], the ILC-derived HCI-013EI PDX model [15], and one of five PDEs originated from ILC. Importantly, in each of the ILC models tested, Riluzole plus Fulvestrant provided greater growth inhibition or earlier tumor growth suppression than single-agent Fulvestrant. Expansion of the PDE approach will be an important strategy for rapidly diversifying the repertoire of preclinical ILC models to test this and other novel endocrine therapy combinations, essential for a breast cancer subtype that has a significantly greater risk of late recurrence, and worse response to Tamoxifen and the second-generation SERD AZD9496 [12]. A second limitation is that Riluzole bioavailability is variable, leading to mixed efficacy in preclinical and clinical studies. Here (Figure 6D) and in other preclinical studies of triple negative breast cancer [30] and glioblastoma [26], single-agent Riluzole does not have significant anti-tumor activity in vivo, while in preclinical studies of melanoma [24,25,55], Riluzole is active as a single agent. Serum levels of the drug vary widely in ALS patients receiving the drug, and in a phase II trial for advanced melanoma, circulating Riluzole concentrations also exhibited marked inter-patient variability [56]. Next-generation Riluzole analogs or prodrugs, e.g. [57], may ultimately offer better bioavailability by reducing first-pass metabolism by CYP1A2, and these will be important to explore in combination with endocrine therapy. Finally, Riluzole’s multiple proposed or confirmed mechanisms of action – ranging from inhibition of signaling through GRMs [24] and glutamate release via the cystine-glutamate antiporter xC- [58] to blockade of voltage-gated sodium channels [59] and inhibition of internal ribosome entry site (IRES)-mediated protein synthesis [26] – present a third limitation to readily identifying patients who would benefit most from the drug. Some of this variability is cancer type-specific, with Riluzole action tightly coupled to GRM expression in melanoma [24] and to some extent glioblastoma [26], but not in triple negative breast cancer [27,28]. The many interconnected pathways of glutamate release, uptake and signaling, while reasonably well-defined in the central nervous system, remain understudied in epithelial cells and their pathologies. We posit that future preclinical studies of Riluzole in this context (epithelial tumors generally, and ER+ ILC more specifically) should address these limitations.
Conclusions
Our data suggest Riluzole together with endocrine therapy may offer therapeutic benefit in ER+ breast cancers, including ILC, and support optimization and further investigation of such combinations in this setting.
Declarations
Ethics approval and consent to participate
All animal studies were ethically conducted in accordance with our approved Institutional Animal Care and Use Committee (IACUC) protocols #2018-0005 and #2018-0006. For explant experiments, tissues were collected from discarded surgical samples from UT Southwestern Medical Center (UTSW, Dallas, TX) patients for research purposes after obtaining the written informed consent and in accordance with institutional review board-approved protocol (STU-032011-187).
Consent for publication
Not applicable.
Availability of data and materials
All data generated or analyzed during this study are included in this published article and its supplementary information files.
Competing interests
The authors declare that they have no competing interests.
Funding
These studies were supported by the Department of Defense (DoD) Breast Cancer Research Program award W81XWH-17-1-0615 to RBR. Fellowship support for HS was provided by the Tumor Biology Training Grant T32 CA009686 (principal investigator (PI): Dr. Anna T. Riegel). SP received a Georgetown Undergraduate Research Opportunities Program (GUROP) Summer Fellowship. Technical services were provided by the GUMC Animal Models, Flow Cytometry and Cell Sorting, Histopathology and Tissue, and Tissue Culture Shared Resources, which are supported, in part, by NIH/NCI Cancer Center Support Grant P30 CA051008 (PI: Dr. Louis M. Weiner). The content of this article is the sole responsibility of the authors and does not represent the official views of the DoD or NIH.
Authors’ contributions
HS and AOO contributed to study design, performed and analyzed experiments, and played a key role in writing the manuscript. SM contributed to study design, and performed and analyzed experiments. SP performed and analyzed experiments, and played a key role in writing the manuscript. YG, MIC, and CB performed and analyzed experiments. AR, HC, and DLB contributed to study design, and performed and analyzed experiments. BMJ and GVR contributed to study design. RBR contributed to study design, performed and analyzed experiments, and played a key role in writing the manuscript. All authors read and approved the final manuscript.
Supplementary Figure Legends
Acknowledgements
The authors would like to thank members of the Riggins laboratory, Drs. Karen Creswell, Michael Johnson, Marc Lippman, and Dan Xun (Lombardi Comprehensive Cancer Center, Georgetown University), and Drs. David Lum and Alana Welm (Huntsman Cancer Institute, University of Utah) for sharing reagents, scientific insights, technical assistance, and/or editorial comments on the manuscript.
List of Abbreviations
- ER+
- estrogen receptor-positive
- SERM
- selective estrogen receptor modulator
- SERD
- selective estrogen receptor downregulator
- ILC
- invasive lobular breast cancer
- IDC
- invasive ductal breast cancer mGluR,
- mGluR, GRM
- metabotropic glutamate receptor
- ALS
- amyotrophic lateral sclerosis
- IMEM
- improved minimal essential media
- FBS
- fetal bovine serum
- CCS
- charcoal-cleared serum
- STR
- short tandem repeat
- DMSO
- dimethylsulfoxide
- SEM
- standard error of the mean
- SD
- standard deviation
- PI
- propidium iodide
- FITC
- fluorescein isothiocyanate
- BCA
- bicinchoninic acid
- PDE
- patient-derived explant
- PCNA
- proliferating cell nuclear antigen
- IACUC
- institutional animal care and use committee
- PDX
- patient-derived xenograft
- E2
- estradiol
- SQ
- subcutaneous
- PO
- per os, by mouth
- mIHC
- multiplex immunohistochemistry
- MWT
- microwave treatment
- AR6
- antigen retrieval buffer 6
- DAPI
- 4’,6-diamidino-2-phenylindole
- ZIP
- zero interaction potency
- ANOVA
- analysis of variance
- MM134
- MDA-MB-134VI cell line
- 2D
- two-dimensional
- NSG
- NOD scid gamma
- SCID
- severe combined immunodeficiency
- YAP1
- Yes-associated protein 1
- TCGA
- The Cancer Genome Atlas
- IRES
- internal ribosome entry site