Breast Cancer Targeting through Inhibition of the Endoplasmic Reticulum-Based Apoptosis Regulator Nrh/BCL2L10

Apoptosis is a protective program promoting the clearance of aberrant cells. Failure to undergo apoptosis also enables tumor cells to escape conventional chemotherapy and targeted therapies (1, 2). Thorough understanding of the apoptotic machinery is thus critical to overcome patient relapse. Bcl-2 family proteins act as master gatekeepers of the mitochondrial pathway of apoptosis (3) by regulating outer mitochondrial membrane (OMM) permeability, the increase of which is a point of non-return in apoptosis (4). Apoptosis inhibitors are known to trap the BH3 domains of apoptosis drivers, such as Bax and Bak, which prevents their insertion into the OMM and subsequent cytochrome c release, activating caspase-dependent cell death. Regarding anticancer therapy, BH3 mimetics are currently being developed to inhibit antiapoptotic Bcl-2 proteins, although such molecules present drawbacks and side effects limiting clinical use (5, 6).

In fact, several Bcl-2 homologs colocalize at the endoplasmic reticulum (ER), where they control Ca²⁺ fluxes by interacting with Ca²⁺ channels, including the inositol 1,4,5-trisphosphate receptors (IP₃R; refs. 7, 8). Indeed, Bcl-2 was found to interact with the modulatory and transducing domain (MTD) and the coupling domain (CD) of IP₃R (9, 10), whereas Bcl-xL preferentially interacts with IP₃R CD (11, 12). The Bcl-2 homolog Nrz was reported to interact with IP₃R1 ligand binding domain (LBD), and control in this way actin cytoskeletal dynamics during zebrafish development (13–15). Actually, nrz is the zebrafish ortholog of chicken nr-13, mouse diva/boo and human nrh (also referred to as bcl-2l10/bcl-B). Although nrh is considered as a genuine antiapoptotic bcl-2 homolog (16–20), its actual function remains enigmatic, largely owing to the fact that no phenotype could be detected in diva/boo knockout mice (21). Furthermore, Nrh protein displays limited capacity to inhibit Bax/Bak-dependent apoptosis (22) together with a restricted tissue nrh gene expression pattern (16, 17).

Actually, nrh was reported to be overexpressed in breast, prostate, and lung cancers, as well as myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML; refs. 23, 24). In such diseases, its expression was correlated with poor prognosis and chemoresistance (25). However, little is known about the involvement of Nrh protein in breast cancer. Systematic studies of the pathologic parameters associated with Nrh protein expression and its role at the molecular level are also lacking.

Here, we assess the contribution of Nrh to breast tumor development and chemotherapy resistance. Based on the analysis of a large cohort of breast cancer patients, we report that Nrh protein is present in invasive breast carcinomas and is an independent poor prognostic marker correlated with metastatic relapse. We demonstrate that Nrh protein preferentially localizes at the ER, inhibits Ca²⁺ release, and prevents apoptosis by interacting with the LBDs of IP₃R channels. Finally, we show that disruption of Nrh/IP₃Rs complexes using BH4-mimicking peptides primes Nrh-addicted cells to apoptosis and inhibits tumor growth both in vitro and in vivo. Collectively, our results reveal that Nrh is a poor prognostic marker linked with chemotherapy resistance and is a promising target for breast cancer treatment.

A total of 393 consecutive patients with operable breast cancer who were given adjuvant/neoadjuvant treatment at the Centre Léon Bérard (CLB) between 2001 and 2003 were tested for Nrh protein expression. Axillary lymph node (LN) invasion was assessed by sentinel node and/or level I and II axillary dissection. Tumor size was defined as the maximum tumor diameter measured on the tumor specimens at the time of surgery. Tumors were considered ERα- or PR-positive if they displayed a nuclear staining in 10% or more of tumor cells, as detected by immunohistochemistry. Tumors were considered positive for HER2 expression if they had 3+ staining by immunohistochemistry or 2+ staining by HER2 amplification detected using FISH. The data exported from the patients' files for analysis are consigned in Supplementary Table S1.

All studies including mice and deidentified human samples were conducted according to the CIOMS ethical guidelines and approved by the ethic review board of the Centre Léon Bérard Hospital. Written informed consent was obtained from each patient and the study protocol was approved by the institutional ethics committee. This study is reported according to the REMARK criteria (26).

To analyze Nrh protein expression, a human normal tissue array (TMA-1205, Protein Biotechnologies, Clinisciences) and paraffin-embedded breast tumor tissues containing invasive carcinomas (n = 393) were used. After deparaffinization and rehydration, endogenous peroxidases were blocked by incubating the slides in 5% hydrogen peroxide in sterile water. For heat-induced antigen retrieval, tissue sections were boiled in 10 mmol/L citrate buffer pH 8.0 (Dako). The slides were then incubated at room temperature for 60 minutes with an antibody against Nrh (1/200, Anti-Nrh rabbit HPA042222, Sigma). After washing with PBS, the bound antibody was incubated with OmniMap anti-rabbit HRP kit (Roche), then with chromoMap DAB kit (Roche). Sections were counterstained with hematoxylin.

Statistical significance between two groups was analyzed using the 1-tailed Student t test. The correlation between Nrh protein expression and the clinicopathologic factors was calculated using Pearson χ² test. Distant metastasis-free survival (DMFS) was defined as the time from histologic diagnosis of breast cancer to the date of distant relapse or death. Survival curves were derived from Kaplan–Meier estimates and were compared using the log-rank test. Hazard ratios and 95% CIs were calculated using Cox regression model for univariate and multivariate analyses. All statistical tests regarding clinical data analysis were two-sided. A P value equal to or under 0.05 was considered statistically significant.

MDA-MB-231, CAL-51, MCF10A, and HeLa cells were obtained from the ATCC and grown in DMEM, supplemented with 10% FBS, 1% penicillin/streptomycin or for MCF10A only in DMEM/F12 supplemented with 5% horse serum, 20 ng/mL EGF, 0.5 μg/mL hydrocortisone, 100 ng/mL cholera toxin, 10 μg/mL insulin, and 1% penicillin/streptomycin. Cell lines were routinely tested for mycoplasma contamination using Mycoalert kit (Lonza) and authenticated by single nucleotide polymorphism profiling (Multiplexion GmbH). Experiments were carried out within 6 weeks from cell thawing. Viral production for generating CRISPR/Cas9 modified cell lines were carried out as previously described (27).

Whenever stated, MDA-MB-231 cells treated with FITC TAT-Nrh 1-23 or FITC TAT-Nrh 1-23 Y16F were incubated, 24 hours after plating on Nunc Labtek chambers, with the corresponding peptide diluted to 10 μmol/L in regular medium for 2 hours. Each chamber was then irradiated, or “photoactivated” whenever stated with a fluorescent bulb (λ_{488 nm}) for 2 minutes to trigger peptide endosomal leakage into the cytosol. Cells were grown for another 16 hours, before being used for proximity ligation assay or apoptosis assay experiments.

For treatment with tetra-methyl-rhodamine (TMR) dfTAT-Nrh 1-23 or TMR dfTAT-Nrh 1-23 Y16F, cells were seeded, onto regular 6-well, 24-well, or 96-well plates (Corning) and grown for 24 hours. Corresponding peptides were diluted to 10 μmol/L in Ca²⁺-free balanced salt solution (BSS), from a 100 μmol/L peptide stock solution in water prepared 24 hours prior to the experiment. Cells were incubated for 1 hour in the 10 μmol/L peptide/BSS solution, then washed and placed in regular medium before subsequent experiments.

For the MDA-MB-231 cell line, 4.5 × 10³ cells were seeded onto a 96-well plate for 24 hours before incubation with dfTAT peptides. MDA-MB-231 cells were treated with 4 μmol/L etoposide or 0.04 μmol/L doxorubicin. For CAL-51 cell line, 8 × 10³ cells were seeded in a 96-well plate 24 hours before incubation with dfTAT peptides for. CAL-51 cells were treated with 25 μmol/L azacytidine or 0.1 μmol/L doxorubicin. Images were acquired using an IncuCyte ZOOM over a 96 hours timeframe, and cell proliferation was measured as the percentage of cell density observed over this period.

Cells were treated either with staurosporine (1 μmol/L for 8 hours), thapsigargin (ranging from 0.1 to 10 μmol/L, for 24 to 36 hours), etoposide (50 μmol/L for 36 hours), or doxorubicin (5 μmol/L for 24 hours).

For kinetic experiments, 1 × 10⁴ cells were seeded onto a 96-well plate 24 hours before transfection. Cells were transfected with 50 nmol/L of the different siRNAs using Lipofectamine RNAiMAX reagent (Invitrogen), then incubated for another 48 hours. Apoptosis was assessed by following the caspase activity using the CellPlayer Kinetic Caspase-3/7 Apoptosis reagent (Essen Bioscience) diluted 1:2,000. Images were acquired using an IncuCyte ZOOM. Caspase activity was corrected to cell density to obtain % of apoptotic cells.

For endpoint experiments, 1 × 10⁵ cells were seeded in a BD Falcon culture chamber 24 hours before transfection. Cells were transfected with 0.5 μg of pCS2⁺-EGFP vector mixed with 1 μg of the indicated vectors using XtremeGene HP transfection reagent (Roche), then grown for another 24 hours. Apoptosis was assessed at the end of the corresponding drug treatment using the SR-FLICA Poly Caspase Assay staining (ImmunoChemistry Technologies). Images were acquired using a Nikon NiE microscope and green transfected cells positive for FLICA staining were counted with a custom-made macro on the ImageJ software to calculate the percentage of green cells displaying caspase activity.

For immunoprecipitation experiments, 6 × 10⁶ HeLa cells were transfected with the indicated vectors and grown for another 24 hours. Cells were lysed in TNE buffer [10 mmol/L Tris-HCl, 200 mmol/L NaCl, 1 mmol/L EDTA (pH 7.4), supplemented with 1 mmol/L β-glycerophosphate, 1 mmol/L sodium orthovanadate, 0.1 mmol/L sodium pyrophosphate, 0.2 % NP-40 and a protease inhibitor, Roche]. Extracts were precleared with protein G-Sepharose beads (Sigma) for 2 hours at 4°C, then incubated overnight with 5 μg IP₃R1 primary antibody. Afterward, extracts were incubated with protein G-Sepharose beads for 3 hours. Immunoprecipitated fractions were washed three times with TNE buffer, then analyzed by immunoblotting. For HA and Flag immunoprecipitation, 1 × 10⁶ HeLa cells were seeded 24 hours before transfection with the indicated vectors and grown for another 24 hours. Two micrograms of primary antibody HA rabbit or Flag rabbit was used.

For pulldown experiments, 6 × 10⁶ HeLa cells were transfected with the indicated vectors and grown for another 24 hours. Lysates were precleared with protein Streptavidin beads (Dynabeads MyOne Streptavidin T1, Thermo) for 2 hours at 4°C split in two and 5 μg of purified biotin-coupled Nrh 1-23 WT or Nrh 1-23 R6A/Y16F peptides was then added when indicated.

Full listing of the antibodies used can be found in the Supplementary Materials and Methods section.

All steps were carried out at 4°C if not otherwise stated. 20 × 10⁶ MDA-MB-231 cells were resuspended in 1 mL cold MB buffer (210 mmol/L mannitol, 70 mmol/L sucrose, 1 mmol/L EDTA, 10 mmol/L HEPES; pH 7.5) and disrupted using a dounce tissue grinder. Extracts were centrifuged at least 5 times at 1,500 × g for 5 minutes to eliminate nuclei, and then centrifuged at 10,600 × g for 10 minutes. The pellet containing mitochondria was washed twice in MB buffer and then resuspended in TNE buffer. The supernatant was then centrifuged at 100,000 × g for 1 hour. The supernatant containing the cytosolic fraction was conserved; the pellet containing ER membranes was resuspended in TNE buffer for immunoblotting.

HeLa cells cultured in Nunc Labtek chambers were incubated with 5 μmol/L of FluoForte Ca²⁺ probe in a Ca²⁺-free balanced salt solution (BSS; 121 mmol/L NaCl, 5.4 mmol/L KCl, 0.8 mmol/L MgCl₂, 6 mmol/L NaHCO₃, 5.5 mmol/L D-glucose, 25 mmol/L HEPES, pH 7.3) for 45 minutes at 37°C. After 10 seconds of measurement, 100 μmol/L histamine in Ca²⁺-free medium was added. Time-lapse fluorescence values were collected using a Zeiss LSM 780 confocal microscope.

For immunofluorescence staining, transmission electron microscopy (TEM) was carried out as described previously (14, 15). Subcellular localization of endogenous Nrh in MDA-MB-231 cells was performed by Nrh immunogold labeling and TEM imaging. Uranyle acetate staining was used to allow for subcellular structure visualization in fixed cell sections. Images were collected using a JEOL 1400JEM transmission electron microscope and Zeiss LSM780 confocal microscope. Colocalization scores were calculated using Pearson coefficient, corrected by subtracting the background intensity in both channels (pCS2+ empty vector). The proximity ligation assay kit was used according to the manufacturer's protocol (Olink). Images were acquired as Z-stacks and then analyzed using the ImageJ software.

Briefly, TAT-NDP and TAT-CTRL were diluted to a concentration of 80 μmol/L in water with 20% 2,2,2-Trifluoroethanol. Circular dichroism spectra were recorded using a Chirascan CD Spectrometer (Applied Photophysics).

Experiments using mouse xenograft models were carried out by Antineo, Lyon, France. MDA-MB-231 cells (6 × 10⁶) were injected into the mammary gland of SCID mice (The Charles River Laboratory). Mice were randomly distributed into treatment groups. When the tumors reached a mean volume of 90 mm³, peritumoral injection of the corresponding peptides or PBS (control) was performed every 2 to 3 days for 4 weeks. When indicated, intraperitoneal injection of doxorubicin (1.2 mg/kg) was performed every week. Tumor size was measured twice a week using calipers, and tumor volume (TV) was calculated using the formula TV = (4/3π) × r³. All animal studies were conducted in accordance with European Union guidelines and approved by the regional ethics committee.

Full details and listing of vector constructions, siRNA, reagents, and antibodies can be found in Supplementary Materials and Methods section. Nrh constructs targeting the mitochondria, using the ActA domain (Nrh-ActA), the ER, using the cb5 ER-anchoring domain (Nrh-cb5), or a truncated Nrh lacking the BH4 domain (Nrh-ΔBH4) were cloned into the pCS2+ vector. Two Nrh ER-associated mutants feature with the loss of a functional BH3 binding pocket, with either a point mutation in the Bcl-2 family conserved motif NWGR of the BH1 domain (Nrh G85A) or deletion of half of the BH1 and BH2 domains (Nrh 1-97) were cloned into the pCS2+ vector.

In order to evaluate the possible correlation between Nrh protein expression and clinical parameters in breast cancer, we performed a systematic analysis using immunohistochemistry staining in a large cohort of patients (n = 393) with invasive breast carcinomas (Fig. 1A; Supplementary Table S1). Accordingly, 180 patients (45.8%) displayed Nrh-positive staining while 213 patients (54.2%) were Nrh-negative (Fig. 1B; Supplementary Table S1). Furthermore, using a panel of 33 human healthy tissues, we observed that Nrh is restricted to the kidneys (proximal tubules), stomach (glandular layer), liver and testis (Leydig cells), but is not expressed in healthy breast tissues (n = 3; Supplementary Fig. S1A). Our results also show that Nrh expression is enhanced in liver tumors compared with normal liver tissue (Supplementary Fig. S1B). Though the presence of Nrh was significantly correlated with the premenopausal status and with larger tumors, there was no association with LN invasion, SBR grade, ERα, progesterone receptor (PR), HER2 status or breast cancer subtype (Supplementary Table S2). Nrh-positivity was associated with shorter DMFS, as evidenced by the higher rate of metastasis recurrence at 10 years, with a DMFS of 69.5% versus 80.6% in the Nrh-negative subgroup (P = 0.02; Fig. 1B). Moreover, after adjusting the significant prognostic factors in the univariate model, larger tumors, a high SBR grade, and Nrh protein expression were the only predictors of shorter DMFS, as shown in the final multivariate model (Supplementary Table S3). Altogether, these data support the hypothesis that Nrh is a novel independent marker of poor prognosis in breast cancer, a predictor of metastatic relapse, and might be an indicator of resistance to chemotherapy.

Having shown the association of Nrh and breast cancer, we analyzed endogenous Nrh levels in a panel of breast cancer cell lines and found that the protein was present in all cell lines tested, but remained undetectable in the non-tumorigenic epithelial MCF10A and MCF12A cell lines, although this was not correlated with mRNA levels (Fig. 1C; Supplementary Fig. S1C).

We selected the MDA-MB-231 cell line to test the effect of nrh silencing, using two specific siRNAs (Fig. 1D), on the cell response to death-inducing compounds. In contrast to staurosporine, thapsigargin showed a limited death-inducing effect in MDA-MB-231 cells over the time course of the experiment (36 hours), because only 17.1% ± 1.1% of the cells displayed caspase activity when treated with siSCR (control siRNA). However, in nrh-silenced cells, thapsigargin treatment triggered a significant increase in cell death, as evidenced by 43.6% ± 1.4% (si3-Nrh) and 44.8% ± 1.2% (si6-Nrh) of MDA-MB-231 cells displaying caspase activity (Fig. 1E and F; Supplementary Fig. S2A–S2D).

Based on these observations, we decided to delete the nrh gene in MDA-MB-231 cells, using the CRISPR/Cas9 technology, to evaluate its contribution to tumor growth (Supplementary Fig. S3A and S3B). In effect, nrh^−/− homozygotes could not be obtained, suggesting that MDA-MB-231 cells could be addicted to nrh. However, the nrh^+/− cells proved to be viable and showed a markedly reduced proliferation rate after 4 days of treatment with etoposide, doxorubicin, or azacytidine (Fig. 1G), compared with control cells. Furthermore, xenograft assays revealed that partial deletion of nrh reduced tumor growth over a long time frame and increased potency of doxorubicin (Fig. 1H and I). Of note, reduced tumor size appeared to be correlated with increased apoptosis events as shown by TUNEL assays (Supplementary Fig. S3C–S3E). Thus, Nrh seems to be involved in a mechanism preventing thapsigargin-mediated cell death and could contribute to the tumorigenicity of cancer cells in vivo.

To unravel the underlying mechanisms, we evaluated the cellular localization of endogenous Nrh using subcellular fractionation and TEM. Remarkably, Nrh was detected at the ER but not at the mitochondria in MDA-MB-231 cells (Fig. 2A and B). We confirmed these findings by confocal microscopy, using ectopic expression of Flag-tagged Nrh in HeLa cells (Fig. 2C and D). Taken together, these results suggested that Nrh may have a predominant ER-related function. In order to assess the importance of Nrh subcellular localization to regulate apoptosis, we generated Nrh mutants, in which the C-terminus TM domain was replaced with an ER (Nrh-cb5) or mitochondria (Nrh-ActA) targeting sequence (Fig. 2E). As shown in Fig. 2F, expression of Nrh-cb5 in HeLa cells prevented caspase activation upon thapsigargin treatment with the same efficacy as native Nrh, contrary to cells transfected with the Nrh–ActA construct. In addition, the ΔBH4 Nrh deletion mutant, lacking the N-terminal α-helical domain, was unable to protect HeLa cells against thapsigargin-induced apoptosis, suggesting that, at the ER, Nrh prevents lethal Ca²⁺ insults in a BH4-dependent manner.

Bcl-2–like apoptosis inhibitors are known to trap the BH3 domains of the apoptosis activators Bax and Bak (3). To examine whether the antiapoptotic activity of Nrh was also linked to such BH3-dependent interactions, we analyzed the effects of two Nrh mutants unable to bind Bax or Bak. These mutants, referred to as Nrh G85A and Nrh 1-97 (Fig. 2E), respectively, bear a point mutation, hindering BH3 binding (28, 29), or a truncation at the C-terminus, resulting in the same loss of function (15). However, both mutants were able to prevent caspase activation in HeLa cells following thapsigargin treatment, similarly to native Nrh (Fig. 2G). Hence, at the level of the ER, the antideath activity of Nrh does not appear to be dependent on binding with Bax-like proteins. To further characterize the molecular events that could be affected by Nrh, we analyzed its effect on the unfolded protein response (UPR), a major ER-dependent response to stress-inducing agents, including thapsigargin. Accordingly, CHOP accumulation and increased eIF2α phosphorylation, typical UPR events (30, 31), were observed within 4 hours in control HeLa cells, upon exposure to thapsigargin. Conversely, CHOP accumulation and eIF2α phosphorylation were weaker in Nrh-stably expressing cells (Fig. 2H), indicated that Nrh might behave as an inhibitor of the PERK-mediated UPR pathway. Furthermore, the previously observed kinetic of caspase activation (within 36 hours) following thapsigargin exposure of nrh-silenced cells appears to substantiate the reported time course of UPR-induced cell death (Supplementary Fig. S2A–S2C and S2H). Of note, the overall status of two other UPR pathways, as measured by ATF6 cleavage and xbp1 splicing, was not significantly changed in Nrh-stably expressing cells (Supplementary Fig. S4A and S4B). Together, these data suggest that Nrh might modulate UPR initiation and subsequent cell death via an effect on ER Ca²⁺ fluxes.

Recently, we reported that Nrz, the zebrafish ortholog of Nrh, directly interacts with the IP₃R1 Ca²⁺ channel and controls the release of Ca²⁺ from the ER into the cytosol (15). We therefore investigated the ability of Nrh to interact with IP₃R1 domains using coimmunoprecipitation (co-IP) assays. This revealed that Nrh interacted with the LBD, as well as the MTDII domain of IP₃R1 (Fig. 3A and B). GST pulldown assays identified residues 227 to 448 of the LBD of IP₃R1 as critical for this interaction (Supplementary Fig. S5A). Moreover, the ΔBH4 Nrh deletion mutant was unable to bind endogenous IP₃R1 in HeLa cells, indicating that the N-terminal BH4 domain mediates Nrh/IP3R1 complex formation (Fig. 3C). Interestingly, among Bcl-2 homologs, Nrh appears to exhibit the highest affinity for the IP₃R1 LBD (Supplementary Fig. S5B). It should also be noted that Nrh interacted with the LBD of IP₃R3, but not of IP₃R2 (Fig. 3D).

On this basis, we analyzed the effect of Nrh on ER Ca²⁺ release by monitoring cytosolic Ca²⁺ levels in HeLa cells treated with Histamine. As shown in Fig. 3E and F, Nrh prevented Ca²⁺ release from the ER, as evidenced by a reduction in the maximum fluorescence intensity from 4.2 ± 0.2 A.U. to 3.1 ± 0.2 A.U., whereas ΔBH4 Nrh (maximum fluorescence intensity of 4.2 ± 0.2 A.U.) had similar effects as the empty control vector. Using the same experimental setup, we observed that Nrh G85A and Nrh 1-97 prevented Ca²⁺ release from the ER with the same efficiency as native Nrh (Fig. 3G and H). Together, these findings confirm that the inhibition of ER Ca²⁺ release by Nrh depends on IP₃R1, but not on Bax-like proteins.

Because Nrh prevents cell death by lowering IP₃R1 and IP₃R3 Ca²⁺ permeability, we speculated that the disruption of Nrh/IP₃R complexes could sensitize cells to death-inducing stimuli acting on Ca²⁺ homeostasis. To test this hypothesis, we analyzed the ability of BH4-derived peptides (Fig. 4A), to prevent Nrh Co-IP with IP₃R1. Indeed Co-IP assays revealed that the ectopic expression in HeLa cells of the Nrh 1-23 peptide compromised the formation of the Nrh/IP₃R1 complex, whereas peptides harboring R6A or Y16F point mutations were less efficient in this respect (Fig. 4B). Of note, the residual Nrh/IP3R1 interaction normalized against peptide expression levels indicated that the impact of the loss-of-function of the Y16F mutation was greater than that of R6A (Fig. 4B). As the Tyr to Phe mutation is less likely to hinder the helical structure Nrh BH4 domain, this mutation was selected as a control in downstream experiments, even if some residual activity was observed.

With respect to cell death, the Nrh 1-23 peptide, but not the Y16F mutant, increased HeLa cells response to thapsigargin, as measured by caspase activity (control: 31% ± 2.1%, Nrh 1-23: 60% ± 1.3%, Y16F 30.8% ± 0.1%, see Fig. 4C). Moreover, Nrh 1-23, but not the Y16 mutant, suppressed the inhibiting effect of Nrh on histamine-triggered Ca²⁺ release (Fig. 4D and E). Interestingly, potentiation of thapsigargin effect by Nrh 1-23 can be reversed by the PERK inhibitor GSK2606414, suggesting that Nrh regulates UPR-mediated cell death via the PERK pathway (Supplementary Fig. S6A).

At the molecular level, we investigated whether the Nrh 1-23 peptide could compete with Nrh for IP₃R1 binding. In contrast to full-length Nrh (Fig. 4F and G), Nrh 1-23 interacted neither with the IP₃R1 LBD nor with Beclin-1, a known interactor of Nrh (29) used here as a control. In fact, Nrh 1-23 interacted directly with the Nrh protein, suggesting that disruption of the Nrh/IP₃R1 complex might be due to the association of this peptide with Nrh but not with IP₃R1 (Fig. 4F and G). Pulldown assays revealed that this peptide could interact with a truncated form of Nrh lacking residues 98 to 194, hence ruling out the requirement for the BH3 binding pocket (Supplementary Fig. S6B). In addition, Nrh 1-23 peptide failed to inhibit the interaction between the autophagy regulator Beclin-1 and Nrh (Supplementary Fig. S6C), which is in accordance with previous findings attributing this interaction to the canonical BH3 binding pocket of Nrh (29). Of note, residue Y16 also appeared to be essential for Nrh 1-23/Nrh interaction (Fig. 4H and I). Thus, Nrh 1-23 does not seem to act as a decoy peptide at the Nrh binding site on IP₃R1, but instead binds directly to Nrh to prevent the formation of the Nrh/IP₃R1 complex.

Having highlighted the effect of Nrh 1-23 in transfected HeLa cells, we wondered whether a synthetic membrane-permeable peptide could have a similar effect in breast cancer cells. We used the PLA technology on MDA-MB-231 cells to directly detect endogenous Nrh/IP₃R1 complexes in cellulo. For calibration purposes naïve and nrh-silenced cells were used as positive and negative controls, respectively (Supplementary Fig. S7A).

A FITC tag linked to a TAT sequence was added to the N-terminus of the 1-23 Nrh-derived peptide (NDP), hence called TAT-NDP, the Y16F mutant being used as a control (TAT CTRL peptide; Fig. 5A). Of note, both peptides adopted an α-helical conformation in solution (Fig. 5B). To prevent the trapping of the internalized TAT-NDP peptide in the endosomes, we used a light source at 488 nm to excite the FITC fluorophore and induce endosomal leakage (32), thus increasing cytosolic availability of the peptide (Supplementary Fig. S7B). Under such conditions, Nrh/IP₃R1 interactions were clearly detected in control MDA-MB-231 cells (vehicle) with 80.3 ± 5.6 dots per cell, as well as in cells incubated with TAT-CTRL (65.2 ± 4.4 dots per cell), whereas only a weak PLA signal (11.8 ± 1.3 dots per cell) was detected in cells incubated with TAT-NDP (Fig. 5C and D). Furthermore, we showed that thapsigargin triggered caspase activation in 62.4% ± 10% of the MDA-MB-231 cells incubated with TAT-NDP, compared with only 11.5% ± 4.4% of the cells incubated with TAT-CTRL (Fig. 5E). Altogether, these data support the hypothesis that disruption of the Nrh/IP₃R1 complex by TAT-NDP may enhance the cell response to death-inducing agents, in particular agents that disturb Ca²⁺ homeostasis.

Next, we assessed whether disrupting Nrh/IP₃R complexes could enhance the potency of apoptosis inducers used in the clinic, because these drugs are also known to alter intracellular Ca²⁺ trafficking (33, 34). To this end, we used a dimeric TAT peptide (Fig. 6A and Supplementary Fig. S8), because such peptides, here referred to as dTAT-NDP, display enhanced endosomal escape capabilities (35). First, using Co-IPs in HeLa cells, we validated the activity of the newly formulated dTAT-NDP on the Nrh/IP₃R1 BD interaction (Fig. 6B). The peptides were then used in cell lines having a high level of nrh expression (MDA-MB-231 and CAL-51) or underexpressing nrh (MCF10A and MDA-MB-231 Nrh^+/−; Fig. 1C and G). We found that dTAT-NDP was able to circumvent thapsigargin resistance in MDA-MB-231 cells, but not in MCF10A or MDA-MB-231 Nrh^+/− cells (Fig. 6C), indicating that the observed sensitization was not an off-target effect of the peptide. Furthermore, the dTAT-NDP, but not the dTAT-CTRL, challenged MDA-MB-231 cells upon exposure to high doses of etoposide (50 μmol/L) or doxorubicin (5 μmol/L; Fig. 6D and E). We confirmed these results using a proliferation assay in CAL-51 cells and observed that in these cells, dTAT-NDP was a strong enhancer of azacytidine (25 μmol/L) and doxorubicin (0.1 μmol/L) activity (Fig. 6F and G). Furthermore, a limited effect of dTAT-NDP alone was observed in CAL-51 cells, although the combination with either azacytidine or doxorubicin led to higher death-inducing effects (Fig. 6F and G; Supplementary Movie 1). Thus, these results suggest that dTAT-NDP might be effective to inhibit cancer cell growth even when used as a single-agent therapy. In addition, dTAT-NDP but not dTAT-CTRL was shown to greatly enhance etoposide activity and prevent colony growth of MDA-MB-231 cells in soft agar (Fig. 7A and B).

Finally, having demonstrated the potency of dTAT-NDP-TMR in vitro, we conducted xenograft experiments to assess its antitumor effects in vivo. We injected MDA-MB-231 cells into the mammary gland of female SCID mice and showed that continuous peritumoral injection of dTAT-NDP at 2.5 or 10 mg/kg, as soon as the tumors reached a volume of 90 mm³, strongly decreased tumor growth and final tumor weight (Fig. 7C and D). In contrast, an exponential increase in the tumor mass was observed in the control group treated with vehicle alone or dTAT-CTRL. It should be noted that in this series of experiments, the dTAT-CTRL peptide appeared to partially affect tumor growth. This is presumably due to residual activity as revealed by Co-IPs (see above Fig. 4B). TUNEL assays showed that dTAT-NDP increased apoptosis events in tumors, but not dTAT-CTRL. However, no significant effect could be detected with specific markers for proliferation (KI67), UPR (CHOP) or vascularization (CD31) (Supplementary Fig. S9A and S9B). Together, these results indicate that the dTAT-NDP has a potent antitumor effect, and that the 2.5 mg/kg dosage is sufficient to reach the desired effect on tumor growth. Presumably, lower doses of peptides might be used to find the minimal effective dose, although no significant difference in body weight could be observed between the different groups, suggesting that the peptide has no apparent toxic side effects and is well tolerated at the concentration used (Supplementary Fig. S9C).

Although Nrh protein expression was correlated to chemoresistance and poor patient prognosis in hematopoietic malignancies (24), the present work provides the first systematic study with a large cohort of patients regarding tumorigenesis and clinical parameters in breast carcinomas. Here, we provide evidence that Nrh is an independent marker of poor prognosis associated with shorter DMFS, hence highlighting the link between Nrh expression and therapy resistance. These observations are in line with the fact that Nrh is more abundant in invasive breast carcinoma than in in situ carcinoma (23). Together, our results highlight the clinical value of Nrh as a prognostic marker in breast cancer that could prove to be a worthy target for next generation therapies. Despite being a genuine Bcl-2-like protein, Nrh actual roles remained poorly characterized. Here, we show that Nrh localizes mainly at the ER in breast cancer cells. We also report that Nrh interacts with the IP₃R1 Ca²⁺ channel via its BH4 domain and decreases ER Ca²⁺ release, following histamine treatment. Of note, Nrh activity at the ER is Bax-independent, in accordance with low affinity for Bax (18). Furthermore, we observed that Nrh jeopardizes UPR initiation, and subsequent slowly arising cell death. In fact, this study provides the first evidence that a Bcl-2 homolog can control the UPR pathway upstream of the mitochondria, by preventing the release of Ca²⁺ from the ER lumen via IP3Rs (Fig. 7E).

Based both on (i) the restricted localization of Nrh at the ER, where it specifically interacts with IP₃R1&3 and (ii) the fact that nrh silencing, or Nrh protein inhibition, resulted in the sensitization of these cells to thapsigargin and death-inducing agents used in the clinic, we speculated that disrupting Nrh/IP₃Rs complexes could challenge Nrh-expressing tumor cells. Indeed, disruption of the Nrh/IP₃R1 complex in vitro, using the Nrh BH4-derived peptide, appears to sensitize in vitro cultured cells to conventional chemotherapeutic agents. Moreover, using an in vivo SCID mouse xenograft tumor model, we report that the Nrh-BH4 mimicking peptide alone compromises tumor growth, without apparent side effects. Together, our data support that Nrh/IP₃Rs complexes represent a novel target for breast cancer treatment. As a matter of fact, Nrh-BH4 might have similar effects on other types of cancers overexpressing Nrh (23–25). Surprisingly, the Nrh-BH4 peptide appears to disrupt the Nrh/IP₃R1 complex by binding to Nrh instead of the IP₃R1 BD. Additional structural analyses must be conducted to identify the Nrh domains involved in this interaction.

Interestingly, the Bcl-2/IP₃R2 complex was also reported to be a potential target in the context of B-cell cancers. Therefore, as proposed by Bultynck and colleagues (36), the systematic evaluation of IP₃Rs expression levels in tumor cells—referred to as “IP₃R profiling,” as opposed to “BH3 profiling” (37)—may help identify the most appropriate BH4 domain to target for patient therapy. Indeed, high IP₃R2 levels would suggest focusing on Bcl-2, whereas high IP₃R1 or IP₃R3 levels may point toward Nrh. In addition, in the present study, Nrh protein could be detected in only 4 of 33 tested healthy tissues, suggesting that pharmacologic inhibition of Nrh may have few off-target effects, contrary to the inhibition of ubiquitously expressed proteins such as Bcl-2 or Bcl-xL.

Finally, most strategies targeting prosurvival Bcl-2 proteins have focused so far on BH3 domains. In this respect, the Bcl-2 inhibitor ABT-199 (venetoclax) was recently approved by the FDA for CLL (38). However, the high functional redundancy between Bcl-2 antiapoptotic proteins may represent the limit of current therapeutic strategies using BH3 mimetics, as tumor cells may adapt and resist to a single specific inhibitor and as the use of nonselective molecules proved to be impractical in trials (5). In fact, amid the Bcl-2 family, the high level of diversity observed among BH4 domains is an opportunity to identify specific inhibitors targeting nonredundant activities among Bcl-2 homologs. In this regard, the uncovered cytoprotective function of Bcl-2-BH4 was recently exploited to successfully develop new drug candidates (39, 40). Interestingly, Bcl-2-BH4 appears to inhibit apoptosis by preventing Bax activation (41), whereas we show here that Nrh-BH4 accelerates the cell death by disrupting Nrh/IP₃R1 complexes. Actually, BH4-derived molecules may offer better specificity and presumably lower residual toxicity, providing a unique opportunity to anticipate BH3 mimetic shortcomings (6). Overall, this study provides a first comprehensive model for Nrh mechanism of action in breast cancer, together with a rationale for developing Nrh inhibitors for clinical applications.

No potential conflicts of interest were disclosed.

Conception and design: A. Nougarede, N. Popgeorgiev, R. Rimokh, G. Gillet

Development of methodology: A. Nougarede, S. Omarjee, S. Borel, I. Mikaelian, J. Lopez, R. Rimokh

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Nougarede, N. Popgeorgiev, L. Kassem, S. Omarjee, R. Gadet, O. Marcillat, I. Treilleux, B.O. Villoutreix, R. Rimokh

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Nougarede, N. Popgeorgiev, L. Kassem, R. Gadet, I. Treilleux, B.O. Villoutreix, R. Rimokh, G. Gillet

Writing, review, and/or revision of the manuscript: A. Nougarede, L. Kassem, O. Marcillat, B.O. Villoutreix, R. Rimokh, G. Gillet

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Kassem, R. Rimokh

Study supervision: R. Rimokh, G. Gillet

Other (designed, performed and analyzed PLA experiments): S. Omarjee

We thank Elisabeth Errazuriz-Cerda and Denis Resnikoff (CIQLE-SFR Lyon-Est), Amélie Colombe and Laetitia Odeyer (CLB) for technical assistance, and Brigitte Manship (CRCL) for manuscript editing. This work is supported by AFM telethon, Ligue contre le cancer, Fondation ARC (to G. Gillet and J. Lopez), and Cancéropole Auvergne Rhônes-Alpes (to R. Rimokh).

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.