Amiloride, An Old Diuretic Drug, Is a Potential Therapeutic Agent for Multiple Myeloma

Despite improvements in the survival of multiple myeloma patients thanks to the introduction of novel therapeutic agents (1, 2), it remains an incurable disease (3). Multiple myeloma initially responds to chemotherapy but relapse and chemoresistance usually occur (4), so subsequent recurrences are part of its natural history. Therefore, the search for new drugs that control the disease continues to be required.

Great efforts to develop new agents against multiple myeloma have been made in recent years, to the extent that a wide array of new agents with different mechanisms of action have recently been approved. These include new mAbs, proteasome inhibitors, immunomodulatory drugs and histone deacetylase inhibitors, among others (5). However, their approval processes required several years of research and major investment. An interesting alternative by which this long process might be shortened is the drug-repositioning approach, which involves using old drugs approved for noncancerous diseases (6). One of the advantages of this strategy is that the pharmacokinetic and pharmacodynamic properties and toxicity profiles tend to be well known. The diuretic drug, amiloride, is one such agent.

Amiloride is a potassium-sparing diuretic that has been employed clinically for more than three decades in the treatment of hypokalemia, hypertension, edema and congestive heart failure (7). Some studies demonstrated its significant antitumor and antimetastasis activities that were initially associated with the inhibition of Na⁺/H⁺ exchangers (8). Recently, amiloride was found to modify alternative splicing (AS) in various human cancer cells (9). Pre-mRNA alternative splicing is a highly regulated process, and numerous studies have demonstrated its aberrations to be associated with cancer, tumor progression, and metastasis. This mechanism has recently gained attention as a potential therapeutic target for cancer due to the differential splicing patterns identified in tumor cells and metastatic tumor populations (10–14).

In this study, we evaluated for the first time the antimyeloma (anti-multiple myeloma) effect of amiloride using in vitro, ex vivo, and in vivo models. We found that amiloride had potent activity against a broad panel of multiple myeloma cell lines regardless of TP53 status. In addition, RNA-Seq experiments showed a strong alteration of spliceosome functionality. These encouraging findings, in conjunction with the manageable toxicity profiles of amiloride, provide a framework for evaluating its utility in clinical trials.

The human myeloma cell lines, NCI-H929, MM1S, MM1R, and U266 were acquired from ATCC, RPMI-8226, KMS12-BM, KMS12-PE, and JJN3 from DMSZ (Deutsche Sammlung von Mikroorganismen and Zellkulturen). RPMI-LR5 cell line was kindly provided by Dr. W.S. Dalton (Moffitt Cancer Center, Tampa, FL). All cell lines were cultured in RPMI1640 medium supplemented with 10% FBS and antibiotics (Gibco). Cells were routinely checked for the presence of mycoplasma with MycoAlert kit (Lonza). Cell line identity was confirmed periodically by STR analysis with PowerPlex 16 HS System kit (Promega) and online STR matching analysis (www.dsmz.de/fp/cgi-bin/str.html). All multiple myeloma samples from patients and cells from healthy donors were cultured in AIMV medium supplemented with 20% FBS (Thermo Fisher Scientific). CD138⁺ plasma cells from bone marrow samples of 8 patients with multiple myeloma were isolated using an autoMACS separation system (Miltenyi-Biotec). Clinical information of the patients included in the study is summarized in Supplementary Materials.

Amiloride and melphalan were purchased from Sigma-Aldrich, bortezomib was from LC Laboratories, dexamethasone was from Merck KGaA, and lenalidomide and pomalidomide from Selleckchem. All multiple myeloma patients as well as healthy donors involved in the study provided written informed consent in accordance with the Helsinki Declaration. The research ethics committee of the University Hospital of Salamanca approved the study.

Cell viability and proliferation were evaluated using CellTiter-Glo (Promega) and 3 (4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) colorimetric assay (Sigma-Aldrich), respectively, as described previously (15, 16). Synergism between amiloride and other drugs was evaluated with CalcuSyn software (Biosoft; ref. 17; Supplementary Material).

Apoptosis using Annexin V-FITC/propidium iodide (PI) double staining, mitochondrial membrane depolarization using DilC1(5) (Immunostep) and cell-cycle analysis, were performed by flow cytometry using Infinicyt software (Cytognos S.L.), as described previously (15). Caspases-3/7, 8, and 9 activities were evaluated by Caspase Glo 3/7, Caspase-Glo 8, and Caspase-Glo 9 assays (Promega), respectively, according to the manufacturer's protocol.

The experiments with patient's cells were performed in total bone marrow samples from patients with multiple myeloma. Immediately after extraction, total bone marrow samples were lysed with ammonium chloride to remove red blood cells (erythrocytes); the remaining white blood cells were maintained for 48 hours in AIMV medium supplemented with 20% FBS (Thermo Fisher Scientific) in the absence or presence of different concentrations of amiloride. Then, the activity of amiloride was investigated on plasma cells (PC) and on the main bone marrow cell populations separately. To evaluate the cytotoxicity of amiloride on PCs, samples were analyzed using Annexin V (Immunostep) in combination with three markers that allowed for the identification of pathologic PCs present in the sample. With that aim, we used a fix combination of two mAbs (CD38 and CD45) plus a third one, chosen depending on the specific phenotype of each patient's clonal plasma cells (usually CD56 or CD19). The cells were incubated for 15 minutes at room temperature in the dark. A total of 5 × 10⁵ cells were acquired on a FACSCanto II flow cytometer (BD Biosciences). Finally, apoptosis was analyzed in pathologic PCs (gated on CD38⁺⁺ and CD56^+/−, or CD19^+/− and FSC/SSC) using the Infinicyt software. Annexin V–positive events among the target populations were considered apoptotic cells.

Amiloride cytotoxicity on the other bone marrow cell populations, that is B and T lymphocytes, NK cells and granulocytes, was assessed with the same aforementioned protocol described, but including a panel of 5 antibodies in combination with Annexin V to identify T lymphocytes (CD3⁺), B lymphocytes (CD19⁺), NK cells (CD56⁺/CD3⁻), and granulocytes (SSC^high/CD45+^dim). Among each of these populations separately, we identified as apoptotic the percentage being Annexin V positive using the Infinicyt software.

All animal experiments were performed according to the institutional guidelines and the protocol previously approved by the ethical committee of the University of Salamanca (Salamanca, Spain). For the human subcutaneous plasmacytoma model, 65 CB17-SCID mice (The Jackson Laboratory) were subcutaneously inoculated into the right flank with 3 × 10⁶ MM1S cells in 100 μL of RPMI1640 medium and 100 μL of Matrigel (BD Biosciences). Treatment was initiated immediately after tumor cell inoculation and mice were randomized to the following treatment cohorts, each of five animals: vehicle-alone PBS (C); amiloride, 10 mg/kg (A10); amiloride, 15 mg/kg (A15); dexamethasone, 0.5 mg/kg (D); melphalan, 2.5 mg/kg (M), dexamethasone + melphalan (DM); dexamethasone + amiloride, 10 mg/kg (DA10); melphalan + amiloride, 10 mg/kg (MA10); dexamethasone + melphalan + amiloride, 10 mg/kg (DMA10); dexamethasone + amiloride, 15 mg/kg (DA15); melphalan + amiloride, 15 mg/kg (MA15); and dexamethasone + melphalan + amiloride, 15 mg/kg (DMA15). Amiloride was administered orally daily, and dexamethasone and melphalan intraperitoneally (i.p.) two days a week. Tumor burden estimation and toxicity monitoring were performed as described previously (18).

To estimate survival, mice were sacrificed when the diameter of their tumor reached 2 cm or when they became moribund. Time to endpoint (TTE) was estimated from the day of treatment initiation. For in vivo mechanistic studies, six and four mice, respectively, were subcutaneously inoculated in the right flank with 3 × 10⁶ MM1S cells and 3 × 10⁶ RPMI cells. When the tumor attained a large volume, mice were randomized to receive the vehicle-alone PBS (control group) or amiloride (20 mg/kg) orally for two consecutive days. On the third day, mice were sacrificed and the tumors retrieved for analysis.

Poly A⁺ RNA from KMS12-BM and JJN3 cells untreated or treated with amiloride (0.1 mmol/L and 0.4 mmol/L, respectively) for 24 hours was isolated and prepared for RNA sequencing (RNA-Seq). Libraries were constructed following a TruSeq Stranded mRNA Sample Preparation Guide (Illumina). The final cDNA library was sequenced using Illumina HiSeq 2500 in combination of 100 Paired-End at Lifesequencing S.L. (Supplementary Material).

Paired-end FASTQ files for 12 samples were used in the RNA-Seq analyses. We analyzed the data in three stages: gene expression, isoform level, and splicing events. First, in the analysis of differential expression at the gene level, the genes were considered to be differentially expressed for an absolute n-fold change (FC) of ≥2 and a false discovery rate (FDR) of <0.05. Second, in the analysis of isoform level, we focused on the isoforms with an absolute value of FC ≥ 2 and that corresponded to genes without altered total expression. The criteria used to assign genes as “no-change” were FDR > 0.05 and an absolute value of FC < 2, when the DESeq2 package was applied. Third, differential alternative splicing events were detected using rMATS version 3.0.9 (19), classifying these events into five major types of pattern: skipped exon (SE), alternative 5′ splice site (A5SS), alternative 3′ splice site (A3SS), mutually exclusive exons (MXE), and retained introns (RI). rMATS also calculates the difference in the ratio of these events between two conditions, and produces an estimate of the FDR. Finally, all the enrichment analyses were conducted using the Webgestalt web tool (20), employing the Gene Ontology and KEGG databases as data sources. The dataset is available at the Gene Expression Omnibus (GEO) repository (http://www.ncbi.nlm.nih.gov/geo) under the accession number GSE95077.

Further details are provided in the Supplementary Material.

RNA was extracted using the RNeasy Plus Mini kit (Qiagen). The RNA integrity was assessed with an Agilent 2100 Bioanalyzer (Agilent Technologies). Total RNA (1 μg) was reverse-transcribed to cDNA using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Gene expression was quantified by TaqMan qRT-PCR mRNA assays (Applied Biosystems) and normalized relative to 18S5 using the 2^–ΔC_t method.

Western blot methods and the preparation of protein lysates have been described elsewhere (15). The sources of the mAbs are described in the Supplementary Material.

For immunofluorescence, cells were fixed in 4% paraformaldehyde, permeabilized with 0.25% Triton X-100/PBS, stained with primary mouse anti-SC35 (Abcam) and goat anti-mouse IgG (H+L) secondary antibody, and Alexa Fluor 488 conjugate (Thermo Fisher Scientific). Fluorescence was measured under a Leica confocal microscope.

All statistical analyses were carried out with IBM SPSS Statistics 22.0 (IBM Corp.) and the Simfit package (W.G. Bardsley, University of Manchester, Manchester, UK; v7.0.9 Academic 32-bit, http://www.simfit.org.uk/). P values were corrected for multiple testing using the FDR, with values of <0.05 being considered to be statistically significant. Differences in the in vitro experiments are expressed as the mean ± SD of at least three determinations and were assessed by the two-sided Student t test or the Mann–Whitney U test. Differences in tumor volumes between groups were evaluated fitting an exponential regression model and the regression parameters were compared using a t test for unequal variances. Survival curves were plotted using the Kaplan–Meier method, and compared using the log-rank test.

Amiloride exhibited potent in vitro antimultiple myeloma activity in a dose- and time-dependent manner, as demonstrated in a panel of seven myeloma cell lines with a wide range of cytogenetic abnormalities and p53 status (Supplementary Table S1; Supplementary Fig. S1A). The viability was significantly reduced in both the TP53 wild-type (WT; H929, MM1S) and the mutated TP53 cell lines (KMS12-BM, KMS12-PE, U-266, and RPMI-8226) after exposure to amiloride (P < 0.01; Fig. 1A), although viability reduction in the p53-null cell line JJN3 required a higher dose and a longer time course (P < 0.01). The antimultiple myeloma effect of amiloride was also observed in melphalan- and dexamethasone-resistant cell lines, RPMI-LR5 and MM1R, respectively (Supplementary Fig. S2A).

In the ex vivo study using bone marrow cells from 10 patients with multiple myeloma (six newly diagnosed and four relapsed/refractory), we observed significant apoptosis induction in plasma cells, even in three patients bearing deletion of 17p, with minor cytotoxicity toward B and T lymphocytes, NK cells, and neutrophils (Fig. 1B). Myeloma cell cytotoxicity of amiloride was also confirmed on isolated CD138⁺ plasma cells from eight multiple myeloma patients (Supplementary Fig. S1B). Interestingly, amiloride did not induced cytotoxicity on normal plasma cells from healthy donors (Supplementary Fig. S1C).

Next, we evaluated the cytotoxicity of double combinations of amiloride with melphalan and dexamethasone, employing a constant ratio between them. Subsequent isobologram analysis revealed a combination index (CI) in the synergistic range for the double combinations of amiloride with dexamethasone or melphalan, ranging from 0.2 to 0.8, depending on the doses and cell lines used (Fig. 1C; Supplementary Fig. S2B). Furthermore, amiloride overcame the melphalan and dexamethasone resistance of RPMI-LR5 and MM1R, respectively (Supplementary Fig. S2C and S2D). Amiloride was also combined with new agents such as lenalidomide, pomalidomide and bortezomib. A significant synergism was observed between amiloride and lenalidomide or pomalidomide plus dexamethasone (Fig. 1D and E). In contrast, the combination of amiloride with bortezomib was antagonistic in all cell lines tested (Supplementary Fig. S2E).

To test whether amiloride was able to inhibit the protective effect of the bone marrow microenvironment, MM1S-luc cells were cocultured with mesenchymal stem cells (MSC) from six multiple myeloma patients and treated with increasing concentrations of amiloride for 48 hours. Despite the proliferative advantage to multiple myeloma cells conferred by MSCs, amiloride abrogated the protective effect conferring by MSCs. In contrast, MSCs were resistant to the cytotoxic effect of amiloride (Supplementary Fig. S3).

To elucidate the mechanisms leading to the decrease of cell growth induced by amiloride, we analyzed cell cycle and apoptosis in multiple myeloma cell lines treated with increasing concentrations of the drug (0.1–1.0 mmol/L). Amiloride induced significant apoptosis after 24 and 48 hours in H929, KMS12-BM and JJN3 cell lines (Fig. 2A; Supplementary Fig. S4A), as well as in MM1S, U-266 and RPMI cell lines after 48 hours (Supplementary Fig. S4B). The apoptosis induction was dose-dependent in all cell lines, with the highest levels in the KMS12-BM cell line, even at 0.1 mmol/L after 24 hours of treatment (Fig. 2A). It is of particular note that the apoptosis induced by amiloride was also observed in cells with del(17p) or TP53 mutations (JJN3 and KMS12-BM, respectively). No significant effect of amiloride on cell cycle was observed (Supplementary Fig. S5).

To evaluate the involvement of mitochondria in cell death, the membrane potential (ΔΨ_m) was measured. Amiloride caused a decrease in ΔΨ_m, particularly significantly in KMS12-BM, H929, and RPMI-8226 cells (Fig. 2B; Supplementary Fig. S6). Using a luminescent-proteolytic assay, we observed that the caspases 3/7, 8, and 9 were significantly activated in all the cell lines tested (Fig. 2C; Supplementary Fig. S7A). The involvement of a caspase-independent mechanism was also observed, as Z-VAD-FMK, a pan-caspase inhibitor, was able to inhibit caspase-3/7 activity, but unable to inhibit apoptosis induced by amiloride (Supplementary Fig. S7B).

We evaluated the in vivo efficacy of amiloride (A) in monotherapy and in combination with melphalan (M) and dexamethasone (D). As to the best of our knowledge, there are no data concerning the antitumor efficacy of amiloride in the animal model used here, we evaluated two doses of amiloride (10 mg/kg and 15 mg/kg). Treatment of MM1S-inoculated CB17-SCID mice with a double or triple combination of A (regardless of the A dose applied), together with D and/or M, enhanced tumor growth inhibition, although the differences were only statistically significant for the combinations DA10, DA15, and DMA15 (Fig. 3A; Supplementary Fig. S8A). With respect to survival, we observed a significant improvement in TTE in the group of mice treated with the double combinations, DA10 and MA10, and the triple combination, DMA10, compared with the D, M, and DM groups, respectively (P < 0.05; Fig. 3B). The mice treated with the triple combination DMA10 had a median overall survival (OS) of 115 days compared with 99 days for the combination DM (P < 0.05; Fig. 3B). The double combinations, DA15 and MA15, also showed a statistically significant benefit (P < 0.01) in terms of OS compared with the single drugs (Supplementary Fig. S8B). The triple combination DMA15 also had a longer OS (median, 110 days) than the double combination DM (median, 99 days), although the difference was not statistically significant (P = 0.053; Supplementary Fig. S8B). No significant toxicity, measured as body weight loss, was observed in the mice receiving combinations with amiloride (Supplementary Fig. S8C).

To determine the molecular basis of the anti-myeloma activity of amiloride, we performed RNA-Seq analysis in KMS12-BM and JJN3 cell lines, the most and the least sensitive, respectively, at the beginning of apoptosis (15%–25% cell death, assessed by CellTiter-Glo luminescent assays) after amiloride treatment. The study design is shown in Supplementary Fig. S9. RNA-Seq data were analyzed at three levels: gene, isoform and splicing events. Although there were clearly more deregulated genes in KMS12-BM (almost 5,000) than in the JJN3 cell line (almost 1,000; Fig. 4A), significant enrichment of functional categories, such as metabolic, MAPK, and Jak–STAT signaling pathways, and endocytosis (Supplementary Fig. S10A), were found among the genes deregulated in both cell lines after amiloride treatment. The analysis of differential expression at the isoform level identified a similar number of deregulated transcript isoforms (over 15,000) in both cell lines (Supplementary Fig. S10B).

Next, we focused our analysis on those genes with a total expression that was not differentially modified after amiloride treatment, but whose transcript isoforms were differentially expressed. We found a considerable number of genes that were significantly deregulated at the isoform but not the gene level in both cell lines (Fig. 4B). Among the most significantly enriched pathways with deregulated transcript isoforms, in both cell lines, were those of the spliceosome, apoptosis, metabolic pathways, and those associated with protein-processing in ER, oxidative phosphorylation, cell cycle, RNA transport, and endocytosis (Fig. 4C). Transcript isoforms belonging to different components of the spliceosome and that are involved in the assembly and regulation of the spliceosomal machinery were significantly deregulated after amiloride treatment (Supplementary Table S3). For example, the transcript ENST00000269601, which encodes the canonical protein TXNL4A, is upregulated in myeloma cells treated with amiloride, whereas the transcript ENST00000588162, which encodes a smaller protein, was only expressed in untreated cells. Notably, the p53 pathway was only highly enriched in p53-expressing cell line, KMS12-BM, but not in the p53-null cell line, JJN3 (Fig. 4C; Supplementary Table S4).

Finally, using Multivariate Analysis of Transcript Splicing software we identified thousands of alternative splicing (AS) events in both cell lines after amiloride exposure (Fig. 4D). Most of the significant AS events (FDR < 0.05) involved genes whose total expression was not modified by amiloride, indicating that the deregulation of alternative splicing could be a specific mechanism of action of amiloride. The most common AS event in both cell lines was the SE, which is the most common splice event in mammalian pre-mRNAs.

Given that RNA-Seq results showed deregulation of alternative splicing and spliceosome components in myeloma cells, we next evaluated whether the antimyeloma effect of amiloride was associated with modulation of the splicing machinery. The immunofluorescent staining for the SR (serine/arginine-rich) protein SC35 demonstrated the amiloride-induced modulation of the splicing machinery in myeloma cells with distinct TP53 status. Thus, the modulation of the splicing machinery was accompanied by a reduction in cell viability in the H929 and JJN3 cell lines (Fig. 5). In both settings, the number of speckles was reduced but the remaining speckles increased in size and intensity. This finding was confirmed in vivo (Supplementary Fig. S11A) using xenografts inoculated with other multiple myeloma cell lines. Structural changes of the nuclear speckles induced by amiloride were also observed in CD138⁺ cells from one newly diagnosed multiple myeloma patient (Supplementary Fig. S11B). In addition, mRNA levels of the spliceosome components (SNRNP27, SRSF4, SF3B1, LSM3, LSM14A, PRPF3, and PRPF4) were significantly overexpressed in myeloma cells from patients after amiloride treatment (Supplementary Fig. S11C). Altogether, these results suggest the potential association between the antimyeloma activity of amiloride and the modulation of spliceosomal machinery.

Our results showed that multiple myeloma cells either with WT or mutated TP53 were highly sensitive to amiloride, although higher doses and longer exposure to amiloride were required for p53-null cells. Moreover, pathway enrichment analysis from RNA-Seq data in mutated TP53 cells revealed a subset of deregulated transcript isoforms involved in the p53 pathway. These findings suggest an activation of p53 signaling pathway in multiple myeloma cells after treatment with amiloride. To test this hypothesis, we used qRT-PCR to measure the expression of p53 targets, such as BAK1, BBC3, TNFRSF10B, FAS, CDKN1B, and CDKN1A in multiple myeloma cell lines with different TP53 status. We observed a normal functional p53 response in the WT/WT cell line (MM1S), whereas p53 signaling was abrogated in JJN3 cells with no basal p53 expression (Fig. 6A). Interestingly, the mutated TP53 cell lines (KMS12-BM and U-266) also showed overexpression of p53 targets. Similar results were found by RNA-Seq data analysis (Supplementary Fig. S12). The activation of p53 signaling pathway was confirmed in CD138⁺ cells from eight multiple myeloma patients after treatment with amiloride (Supplementary Fig. S13)

The cell lines with mutated TP53 showed deregulation of p53 targets after treatment with amiloride, so we decided to confirm the involvement of p53 in amiloride-induced cytotoxicity of multiple myeloma cells. The functionality of p53 signaling pathway in mutated TP53 cell lines was confirmed using p53 activity inhibitors: pifithrin-α (PFTα), a reversible inhibitor of p53-mediated apoptosis and p53-dependent gene transcription (21), and pifithrin-μ (PFTμ), an inhibitor of the p53–Bcl-xL interaction that directly inhibits p53 binding to mitochondria (22). Amiloride cytotoxicity was reduced in WT and mutated TP53 multiple myeloma cell lines when used with p53 inhibitors. As expected, the inhibitors had no effect on the death of JJN3 cells, which lack p53 expression (Fig. 6B). These results imply that p53 signaling has an important role in amiloride-induced apoptosis of multiple myeloma cells that express either WT or mutated TP53. On the other hand, the involvement of a mechanism other than p53 signaling activation would explain the antimyeloma effect of amiloride on p53-null cells.

In this study, we demonstrate for the first time the antimyeloma activity of amiloride, an antihypertensive drug, through two novel mechanisms of action, spliceosome deregulation and p53 signaling pathway activation. Initially, we observed potent in vitro antimyeloma activity of amiloride, both in TP53 wild-type and mutated TP53 cells. Even in p53-null cells, viability was reduced by using higher doses and longer exposures. The ex vivo study of myeloma cells from patients indicated that amiloride induced cytotoxicity in plasma cells, including three cases bearing deletion of 17p, whereas viability of other bone marrow cell populations was not affected. Furthermore, amiloride in combination with dexamethasone and melphalan was clearly synergic in vitro. A synergic effect was also observed when amiloride was combined with lenalidomide or pomalidomide plus dexamethasone. In this context, amiloride has been described as potentiating synergistically the antiproliferative effect of other drugs like imatinib, the first-line therapy for patients with chronic myeloid leukemia (CML; refs. 23, 24). However, we did not find synergism between amiloride and bortezomib, which could limit the use of amiloride in bortezomib-based induction regimens.

The induction of apoptosis by amiloride in CML was accompanied by the increase in levels of caspases 9 and 3. Our results showed that the apoptosis induced by amiloride was mediated by both caspase-dependent and caspase-independent mechanisms, which is consistent with other studies in glioblastoma and breast tumor cells (25–27). Moreover, we observed increased survival of mice bearing human subcutaneous plasmacytomas treated with double or triple combinations including amiloride compared with treatment with melphalan and/or dexamethasone.

The anticancer effect of amiloride has previously been described in several tumors, using in vitro and in vivo models (9, 23–33). There is evidence for multiple mechanisms of action of amiloride, including TRAIL-induced cytotoxicity associated with the PI3K-Akt pathway (24, 28, 30, 33) and alternative splicing deregulation of apoptotic genes (9, 23, 28). The RNA-Seq allowed us to study its mechanism of action in myeloma cells more extensively. In fact, RNA-Seq analysis in two cell lines with distinct patterns of response to the drug, the most and the least sensitive, revealed that amiloride significantly altered the level of transcript isoforms and the alternative splicing events. It should be pointed out that the significant impact on the differential expression of isoforms from genes whose total expression was not changed. In other words, the traditional gene expression profiling would have overlooked the substantial modifications of more than 10,000 transcript isoforms by amiloride treatment. These results are consistent with the reported advantage of the analysis at the genome-wide isoform level compared with gene expression in cancer research (34, 35).

One of the most significantly enriched pathways in the analysis of differentially expressed isoforms after amiloride treatment was spliceosome. We found that amiloride induced a general deregulation of spliceosomal machinery at the gene and transcript isoform levels that affected the early and late stages of spliceosome assembly and several spliceosome-associated proteins, including the catalytic steps of the splicing. These findings, together with the large quantity of total transcript isoforms modified by amiloride, prompted us to investigate further the influence of amiloride in the pre-mRNA splicing machinery. The small nuclear ribonucleoproteins (snRNP) and splicing factors, like the SR protein family, are organized in nuclear speckles (36). The changes in protein SC35-staining speckles are used as a marker for the disruption of the splicing machinery (37–40). Upon the inhibition of the splicing machinery, the number of nuclear speckles decreases but those remaining increase in size and intensity (36, 41). Using this marker, after amiloride treatment, we identified a similar pattern of nuclear speckle modifications that was associated with cell viability inhibition. This finding indicates that amiloride provokes the disruption of the splicing machinery and that this could, in turn, induce cytotoxicity.

The RNA-Seq analysis at the transcript isoform level also identified the p53 pathway as one of the most significantly enriched functional categories. Remarkably, the p53 pathway was highly overexpressed only in the cell line expressing mutated TP53, and not in the p53-null cell line. In addition, upregulation of p53 targets was observed in WT and mutated TP53 myeloma cells treated with amiloride. These results, together with the fact that the inhibition of p53 protein activity prevents amiloride-induced cell death, even in two mutated TP53 cell lines, demonstrate that amiloride-induced apoptosis in myeloma cells is dependent on p53 activation and is independent of the mutational status of TP53. Apart from that, the reduction in cell viability in the amiloride-treated p53-null cell line supports the notion that other mechanisms independent of p53, such as the spliceosomal machinery disruption observed in JJN3 cell line, are involved in amiloride activity.

Amiloride has been used for many years as adjuvant treatment with thiazide diuretics in congestive heart failure and hypertension (42–45). Here, we demonstrate for the first time the therapeutic potential of amiloride in multiple myeloma. The concentration of amiloride used in the in vivo experiments is higher than that commonly used as a potassium-sparing diuretic, suggesting that a higher dose would be needed to produce the anti-multiple myeloma effect. The toxicity profile of this drug is very well known and the main side effect is hyperkalemia, which could be the main factor that limits the use of amiloride as an antimyeloma drug. To minimize this risk, a careful electrolyte monitoring along with the coadministration of a kaliuretic agent or the use of new oral agents for the hyperkalemia treatment, such as patiromer calcium (46–48) and ZS-9 (zirconium cyclosilicate; refs. 48–50), could be required. Moreover, attempts to develop amiloride analogues that show reduced diuretic and antikaliuretic effects retaining or enhancing anticancer activity are currently underway in different laboratories. On the other hand, as reported in other studies (23, 32), we did not find any significant systemic toxicity in the mice treated with amiloride, and the viability of the lymphocyte population either from multiple myeloma patients or healthy donors was not affected, even at the highest dose.

Our results also revealed that the antimyeloma activity of amiloride was mediated through spliceosome modulation and involved the p53 pathway. In fact, p53 signaling was activated after amiloride exposure, independently of the mutational status of TP53. On the other hand, amiloride was also able to induce apoptosis in myeloma cells that did not express p53.

In conclusion, these findings together with the possibility of combining amiloride with melphalan, dexamethasone, or lenalidomide or pomalidomide, support the initiation of clinical trials including amiloride for patients with relapsed and refractory multiple myeloma, particularly for those with 17p deletion or TP53 mutations who display a poor prognosis.

No potential conflicts of interest were disclosed.

Conception and design: E.A. Rojas, I. Misiewicz-Krzeminska, N.C. Gutiérrez

Development of methodology: E.A. Rojas, J.F. Martínez-Blanch, I. Misiewicz-Krzeminska

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.A. Rojas, L. San-Segundo, J.F. Martínez-Blanch, F.M. Codoñer, T. Paíno, N. Puig, M. Victoria Mateos, N.C. Gutiérrez

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.A. Rojas, L.A. Corchete, J.F. Martínez-Blanch, F.M. Codoñer, T. Paíno, N. Puig, E.M. Ocio, N.C. Gutiérrez

Writing, review, and/or revision of the manuscript: E.A. Rojas, L.A. Corchete, F.M. Codoñer, N. Puig, R. García-Sanz, M. Victoria Mateos, E.M. Ocio, I. Misiewicz-Krzeminska, N.C. Gutiérrez

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): N. Puig, R. García-Sanz, N.C. Gutiérrez

Study supervision: R. García-Sanz, I. Misiewicz-Krzeminska, N.C. Gutiérrez

The authors thank Isabel Isidro, Teresa Prieto, Vanesa Gutiérrez, Irene Aires, and José Pérez for their technical assistance; Lorena González for the help with the assays of the combinations of antimyeloma drugs; Mercedes Garayoa for the help with MSC studies; and Phil Mason for his help in reviewing the English language of the manuscript.

This study was partially supported by the Instituto de Salud Carlos III-Cofinanciación with funding from FEDER (PI13/00111 and PI16/01074), Asociación Española Contra el Cáncer (AECC, GCB120981SAN), Gerencia Regional de Salud, Junta de Castilla y León (BIO/SA57/13 and BIO/SA22/15), and the INNOCAMPUS Program (CEI10-1-0010). I. Misiewicz-Krzeminska was supported by a Black Swan Research Initiative® grant from the International Myeloma Foundation. L.A. Corchete was supported by a grant from the Fundación Española de Hematología y Hemoterapia.

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.