Multiple mechanisms contribute to acquired TRAIL resistance in multiple myeloma

Abstract

Multiple Myeloma (MM) prognosis has recently improved thanks to the incorporation of new therapies to the clinic. Nonetheless, it is still a non-curable malignancy. Targeting cancer cells with agents inducing cell death has been an appealing alternative investigated over the years, as is the case of TRAIL, an agonist of DR4 and DR5 death receptors. This pathway, involved in apoptosis triggering, has demonstrated efficacy on MM cells. In this research, we have investigated the sensitivity of a panel of MM cells to this agent and generated TRAIL-resistant models by continuous culture of sensitive cells with this peptide. Using genomic and biochemical approaches, the mechanisms underlying resistance were investigated. In TRAIL-resistant cells, a strong reduction in cell-surface receptor levels was detected and impaired the apoptotic machinery to respond to the treatment, enabling cells to efficiently form the Death Inducing Signalling Complex. In addition, an upregulation of the inhibitory protein c-FLIP was detected. Even though the manipulation of these proteins was able to modify cellular responses to TRAIL, it was not complete, pointing to other mechanisms involved in TRAIL resistance.

Introduction

Multiple myeloma (MM) is a clonal B-cell hemopathy, characterized by the accumulation of malignant plasma cells in the bone marrow [1, 2]. This neoplasia accounts for approximately 2% of all cancers and 20% of hematologic ones, being the second most frequently diagnosed hematological malignancy [3]. Even though nowadays it is still considered a non-curable disease, its prognosis has significantly improved thanks to the development of new therapies, such as proteasomal inhibitors or immunomodulatory drugs [4,5,6]. Moreover, the field of immunotherapy has irrupted with strength in this field with the incorporation to the clinic of antibodies targeting CD38, bispecific antibodies targeting B-cell maturation antigen (BCMA) or G-protein coupled receptor family C group 5 member (GPRC5D) and CD3 [7, 8], or antibody–drug conjugates directed to BCMA, such as belantamab mafodotin [9]. Moreover, interest has been paid to the development of antimyeloma therapies based on CAR-T cells directed to BCMA [10]. Currently, selection of the treatment for a patient with MM is considered multifactorial [11], taking into account the patient's physical condition, efficacy of strategies and drug resistance.

Apoptosis is a critical event for tissue homeostasis and cellular development, and its deregulation often occurs during tumor progression and/or chemoresistance by mechanisms not yet fully elucidated [12]. The apoptotic machinery is mainly driven by two different routes, namely the intrinsic and extrinsic pathways. The intrinsic pathway initiates at the mitochondria [12, 13], while the extrinsic pathway is triggered at the cell membrane upon binding of death-inducing polypeptides to their receptors expressed at the cell surface. Both signaling pathways converge on a common machinery which includes members of a family of cysteine proteases called caspases [14]. There are two types of these proteins: initiator and executor caspases. Initiator caspases such as caspases 2, 8, 9 and 10 are activated in response to stress signals or cell damage, while executor caspases such as caspases 3, 6 or 7, are responsible for direct proteolysis of different substrates leading to cell death [15].

Inducing apoptotic cell death has been proposed as an antitumoral strategy [16, 17]. In fact, the BCL2 inhibitor venetoclax has received approval for the treatment of acute myeloid leukemia [18]. Death receptor activation by specific ligands has received attention due to the fact that receptors for these ligands may be more expressed in tumoral tissues with respect to normal tissue counterparts [19]. Such is the case for the TNF-Related Apoptosis-Inducing Ligand (TRAIL) [15, 20]. This ligand is assembled as a soluble homotrimer that interacts with its receptors. Five TRAIL receptors have been described: TRAIL-R1 or DR4 (TNFRSF10A, Tumor Necrosis Factor Receptor SuperFamily member 10A), TRAIL-R2 or DR5 (TNFRSF10B, Tumor Necrosis Factor Receptor SuperFamily member 10B), TRAIL-R3 or DcR1, TRAIL-R4 or DcR2 and a soluble receptor called osteoprotegerin (OPG) [21]. Importantly, DcR1, DcR2 and OPG lack a cytoplasmic death domain and are therefore unable to transduce signaling. Moreover, since they may interact with soluble TRAIL but do not transduce a signal, they are expected to antagonize the proapoptotic action of TRAIL. Therefore, only DR4 and DR5 present this complete structure and can efficiently bind TRAIL and transduce the death signal to target cells. Upon binding to the ligand, DR4 or DR5 may form homo- or heterotrimeric receptor structures that are then able to recruit proteins that interact with the intracellular region of the receptor, including Fas-Associated Death Domain (FADD) protein and pro-caspase 8. This membrane-associated complex, known as DISC (Death Initiating Signaling Complex) will trigger the death pathway [22,23,24]. In addition to this complex, cytoplasmic inhibitors such as c-FLIP (FADD-like IL-1β-converting enzyme)-inhibitory protein, can restrict caspase activation and cause resistance to TRAIL-mediated apoptosis [25].

Preclinical studies in MM have shown that TRAIL can trigger cell death of malignant plasma cells while sparing hematopoietic stem cells [26, 27]. Moreover, it has been reported that TRAIL may be used to overcome resistance to conventional antimyeloma treatments [28, 29]. Mechanistically, activation of TRAIL receptors in myeloma using agonistic antibodies promoted cleavage of the antiapoptotic protein MCL-1L [30]. Interestingly, the antimyeloma action of TRAIL did not appear to correlate with the pro-caspase 8/c-FLIP ratio [31]. These precedents stimulated the testing of TRAIL for the therapy of MM.

Several clinical trials, using circularly permuted TRAIL have shown antimyeloma activity in refractory MM, either used alone or in combination with thalidomide and dexamethasone [32, 33]. The results of the latter phase II trial showed that the combination of TRAIL with thalidomide and dexamethasone resulted in an overall response rate of 38.3%, while in the case of thalidomide and dexamethasone it was of 25%. Moreover, the TRAIL combination doubled the progression free survival time with respect to the two standard of care drugs combined. These data have been updated in a recent phase III trial that has demonstrated the beneficial effects of combining the TRAIL derivative aponermin with dexamethasone and thalidomide in MM refractory patients [34]. These clinical results are promising, but also showed that not all patients respond to TRAIL, and those which respond ultimately relapse. That fact opens the question of how to better optimize the therapy of MM patients with TRAIL and how to identify patients refractory versus those that may respond.

To better understand the molecular basis of TRAIL action and resistance in the context of MM, we generated TRAIL-resistant models by continuously exposing MM cells to this agent. These studies allowed the identification of mechanisms of resistance to the death-inducing action of TRAIL. Moreover, genomic analyses of the resistant cells allowed identification of pathways deregulated in those cells, that could be used to develop strategies to overcome TRAIL-induced resistance.

Materials and methods

Reagents and antibodies

Culture media, fetal bovine serum (FBS), trypsin, penicillin and streptomycin were purchased from GIBCO BRL; Protein-A Sepharose, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and puromycin were from Sigma-Aldrich; Immobilon®-P transfer membranes (PVDF) from Millipore; isoton from Beckman; jetPEI™ from Polyplus-transfections SA; Annexin V-FITC from BD Biosciences, Talon resin from Clontech, BlueSafe staining reagent from NZYTECH. Other general reagents were purchased from Millipore Corp, Sigma-Aldrich, Roche Biochemicals or Merck.

The anti-FLIP (#8837), anti-Caspase 8 (#9746), anti-cleaved Caspase 3 (#9664), anti-Caspase 9 (#9502), anti-Caspase 7 (#9492) and anti-BID (#2002) antibodies were purchased from Cell Signaling technologies; the anti-Caspase 3 (#610,323) and anti-XIAP (#610,716) from BD Biosciences; anti-FADD (#06-711) from Millipore; anti-Calnexin (SPA-860) from Stressgen Bioreagents corporation, and anti-GAPDH (sc-166574), anti-PARP (sc-8007), anti-MCL1 (sc-20679) and anti-BAX (sc-493) and anti-BCL-2 (sc-7382) were purchased from Santa Cruz Biotechnology. For flow cytometry, phycoerythrin (PE)-conjugated anti-DR4 and -DR5 antibodies were obtained from eBiosicence (#12–6644-71 and #12-9908-42, respectively). In addition to the commercial antibodies described, we have used anti-DR4 and anti-DR5 antibodies previously generated and characterized in our laboratory [35].

Cell lines and culture, transfections, generation of retroviruses, lentiviruses and infection

Multiple myeloma cell lines were grown in RPMI (Roswell Park Memorial Institute 1640) medium and HEK-293T (or HEK293T, RRID: CVCL_0063) cells in DMEM (Dulbecco's Modified Eagle Medium), both supplemented with 10% FBS and antibiotics (penicillin 100 U/ml, streptomycin 100 $\mu$g/ml). All cells were cultured at 37°C in a humidified atmosphere and in the presence of 5% CO₂. MM cell lines included MM1S (RRID: CVCL_8792), MM1R (RRID: CVCL_8794), OPM2 (RRID: CVCL_1625), RPMI-8226 (RRID: CVCL_0014), MM1-144 (RRID: CVCL_EI97), RPMI-8226/LR5 (shortened as RPMI-LR5; RRID: CVCL_J433), U266 (or U266B1, RRID: CVCL_0566) and NCI-H929 (RRID: CVCL_1600). The source of these cell lines has already been described [36]. Cell authenticity was analysed at origin or by STR at the Hematology Service of the University Hospital of Salamanca and verified lines were expanded one to five 100 mm dishes and, when confluent, frozen 1 dish:1 vial ratio. Only one cell line at a time underwent this freezing protocol. Authenticated cells maintained in culture were regularly (every 3–6 months) replaced for new ones by thawing a frozen vial.

To manipulate the cellular levels of proteins of interest, viral infections were performed as previously described [35, 37, 38]. Retroviral systems were used to increase levels, based on the vectors pLZR-IRES-GFP-puro (DR4, DR5S or DR5L) and p-Babe (FLIP), that was generously gifted by Dr. Abelardo Lopez-Rivas. To decrease protein levels, lentiviral systems based on pLKO with small hairpin RNA sequences (shRNA) were used. In both manipulations 293T cells were transfected using JetPEI™ and viruses collected to transduce target cells as described.

Protein extraction, immunoprecipitation, Western blotting, and TRAIL treatment

For protein extraction, cells were harvested by centrifugation, washed with cold PBS and lysed in ice-cold lysis buffer. Clarified protein extracts were then quantified by Bradford assay and IP and WB were performed with the indicated antibodies as described [38,39,40]. Cell fractionation experiments were carried out as described [41].

Recombinant TRAIL (rTRAIL) was produced and quantified as previously described [42], using the pET-28b plasmid containing the TRAIL gene (amino acids 95-281). This plasmid was provided by Dr. Abelardo López-Rivas. For TRAIL treatment experiments, cells were maintained in the presence of 1 µg/ml TRAIL, unless otherwise explained. To biotin-label TRAIL, the resin-bound rTRAIL was suspended in 1.5 ml of PBS containing 200 μg/ml of Biotin-7-NHS (Biotinamidocaproate NHS ester) and the labeling reaction performed at 4 °C for 1 h with shaking. Biotin-labeled TRAIL was then eluted from the resin for ulterior use.

DR4 and DR5 cell surface detection and DISC pull-down assay

Fifty-thousand cells were washed with PBS containing 2% BSA before adding 10 μL of PE-labeled anti-DR4 or DR5 antibodies, followed by incubation at room temperature for 30 min. Cells were next washed and read on a FACScalibur flow cytometer. The signal for each cell line was normalized to its isotypic control.

For DISC pull-down experiments, fifty million cells were collected and resuspended in complete media containing 1 μg Biotin-labeled TRAIL and incubated for the indicated times at 37 °C before lysis in ice-cold lysis buffer. Supernatants were cleared by centrifugation and incubated with 50 μl of streptavidin-agarose (Sigma-Aldrich) at 4 °C for 3–12 h. Agarose was then recovered, washed with Lysis Buffer and prepared for WB in sample buffer.

Cell proliferation, cell cycle, apoptosis and caspase activity assays

To analyze cell proliferation, conventional MTT assays were performed [36, 43]. Cell cycle analysis and apoptotic cell death were assessed as previously described [37]. Caspase activity was measured in fluorometric assays in which 50 $\mu$g of previously prepared and quantified cell lysates were incubated for 1 h at 37 °C with the correspondent caspase substrate (Ac-IETD-AFC, 10 mM) in caspase activity buffer (25 mM HEPES pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.1% CHAPS, 10% sucrose). Signals were measured at 400/505nm in a multi-well fluorescent reader (BioTek).

Gene expression analyses

Total RNA extraction was carried out using the Qiagen miRNeasy Mini Kit according to the manufacturer's instructions. The amount and integrity of RNA was assessed by high resolution electrophoresis on an Agilent 2100 Bioanalyzer. For transcriptomic analyses, previously isolated RNA was hybridized to HG-U133 plus 2.0 GeneChip oligonucleotide arrays (Affymetrix) and scanned as described [35]. The raw files were normalized using the RMA algorithm implemented in Affymetrix Expression Console 4. The cutoff point for differentially expressed genes (DEGs) was ≥ twofold change and a p-value ≤ 0.05. Gene set enrichment analysis (GSEA) was performed to identify genesets with expression alterations. The network of genesets interactions was constructed using Cytoscape software (version 3.6.0). Only genes with maximum 0.05 p-value differential expression were selected. Data will be accessible without restriction from the date of publication on GEO Datasets.

qPCR, PCR arrays and apoptosis antibody array

To perform real-time PCR (qRT-PCR), RNA was primed with oligo (dT) and previously synthesized cDNA with MMLV reverse transcriptase (Promega). qRT-PCR reactions were performed using SYBR Green (Bio-Rad Laboratories) and recorded with the Bio-Rad iCycler IQ5 software. To quantify the levels of DR4 and DR5 receptors we used real-time RT-PCR using specific oligonucleotides as previously described [35]. Relative quantitation of gene expression was normalized to GAPDH.

To analyze apoptosis pathway proteins the Human Apoptosis the database program Pathway Interaction Database available through the application NDEX (www.ndexbio.org) was used. Relative gene expression was normalized with GAPDH and compared to the parental cells using the 2ˆ(-ΔΔCt) method, representative graphs of the information were then generated.

Statistical analyses

In proliferation experiments each condition was analyzed in triplicate and data presented as mean ± SD of at least 2 independent experiments. Comparisons of continuous variables between two groups were performed using a two-sided Student's t-test. p < 0.05 was considered statistically significant.

Results

Generation of TRAIL-resistant myeloma cells

The sensitivity of different MM cells to TRAIL was evaluated by performing dose–response experiments. Those studies were carried out in eight different MM cell lines which included cells resistant to dexamethasone (MM1R) or melphalan (RPMI-LR5). These dose–response curves showed that OPM2 cells represented the most sensitive cell line studied, followed by MM1R and MM1S cells (Figs. 1A and B). The most resistant cell lines were RPMI-LR5, MM1-144, NCI-H929 and U266. In the latter cases, the dose–response curve failed to show substantial inhibition of the proliferation of those MM cell lines, as their IC₅₀ values fell above the maximal dose of TRAIL used (1000 ng/ml)(Fig. 1B). These cell lines were therefore considered de novo (primary) resistant to the action of TRAIL.

Fig. 1

Effect or rTRAIL on several multiple myeloma cellular models. A Different multiple myeloma cell lines were incubated for 24 h in the presence of increasing amounts of rTRAIL. MTT assays were carried out to measure rTRAIL cytotoxicity. Data were normalized to vehicle-treated condition. B IC₅₀ values for the cell lines analyzed, calculated from data shown in A. C Schematic representation of the generation of TRAIL resistant MM cells. To generate TRAIL resistance models from a sensitive line, an continuous exposure protocol was followed. It included chronic treatment of MM cells for three months with 1 μg/ml of recombinant TRAIL until viable subpopulations (MM-TR) were obtained. Once potentially resistant subpopulations were generated, their resistance to TRAIL was verified using MTT viability assays. D Effect of rTRAIL on different MM models of acquired resistance to TRAIL. Several MM cell lines were grown in the presence of rTRAIL for 3 months until viable subpopulations (TRAIL Resistant or -TR) were obtained and the acute response to this peptide was assessed and plotted as described in A. All graphs in this figure show the mean ± SD of a representative experiment that was repeated at least 2 times

To generate models of resistance to TRAIL, we followed a protocol consisting in the chronic treatment of MM cells for three months with 1 µg/ml of recombinant TRAIL (Fig. 1C). Once the potentially resistant subpopulations were generated, their resistance to TRAIL was verified by performing dose–response studies (Fig. 1D). These studies showed that the chronic treatment approach resulted in the generation of TRAIL-resistant subpopulations in the case of OPM2, MM1S, and MM1R cells. As expected from their intrinsic resistance to TRAIL, no major differences in the MTT metabolization values were observed in the case of naïve RPMI-8226, RPMI-LR5, NCI-H929, MM1-144 and U266 cells when compared to the populations obtained after the three months chronic treatment period with TRAIL.

Transcriptomic analysis of TRAIL-resistant myeloma cells

To gain insights into the mechanism of resistance to TRAIL, transcriptomic studies were performed. For these studies we selected the OPM2 cell line, since it showed the highest sensitivity to TRAIL and a resistant subpopulation to the drug (OPM2-TR) was obtained. Principal component analysis (PCA) obtained from microarray assays showed differences between both cell lines (Fig. 2A). Comparison of the transcriptomes of OPM2 and OPM2-TR cells, considering fold changes ≥  ± 2 and p-values ≤ 0.05, detected 147 deregulated genes of which 81 were up-regulated and 66 down-regulated (Fig. 2B).

Fig. 2

Genomic characterization of TRAIL resistant MM cells. A Principal component analysis (PCA) of OPM2 versus OPM2-TR cells. RNA extracted from OPM2 and OPM2-TR cells was hybridized with Affymetrix HuGene 1.0 arrays and PCA analysis performed, representing OPM2 as blue dots and OPM2-TR in red. B Differentially expressed genes (DEGs) in OPM2 versus OPM2-TR. Cut-off values for DEGs (grey background areas) were ≥ twofold change and p value ≤ 0.05 (represented as -log p-value). Genes meeting both criteria are colored blue if downregulated or red if upregulated. C Analyses of apoptotic pathway proteins deregulated in TRAIL resistant cells. Using the list of deregulated genes obtained in the microarray analyses and the cytoscape software, a schematic representation of TRAIL pathway proteins was generated. Upregulated genes are shown in red and downregulated in blue. Red lines indicate ligand binding to the receptor, green lines activation control of the protein and purple lines phosphorylation. D Using the Pathway Interaction Database software, deregulated genes were similarly selected and represented, demonstrating the downregulation of DR4 and DR5 receptors (p < 0.001) and the overexpression of the inhibitor FLIP (p < 0.01)

Considering that our goal was to understand the mechanism(s) involved in TRAIL resistance, we used the Pathway Interaction Database (PID) in the NDEX online application (www.ndexbio.org) to focus on genes described to play a role in TRAIL signaling. Under the “TRAIL signaling pathway” denomination, a group of 25 genes involved in TRAIL signaling was selected (Fig. 2C). This group included the components of Death Induced Signaling Complex (DISC, Fig. 2C yellow mark). Of those 25 genes, 6 of them were found to be down regulated in OPM2-TR cells (Fig. 2C, blue background), while 5 genes were up-regulated (Fig. 2C, red background) in the resistant cells. A radial graphic representation of the transcriptomic map corresponding to the 25 selected genes showed that TNFRSF10A (DR4) and TNFRSF10B (DR5) receptors were among the under-expressed genes in OPM2-TR cells (p < 0.005) (Fig. 2D). Besides, the CFLAR gene, which codes for the TRAIL signaling inhibitory protein FLIP, was overexpressed in the resistant line (p < 0.01). These studies uncovered gene expression changes in several proteins known to mediate the proapoptotic action of TRAIL, and therefore evidenced alterations in components of the extrinsic pathway of apoptosis in the resistant model.

Down regulation of DR4 and DR5 promotes TRAIL resistance

The transcriptomic analyses revealed that the two signaling receptors for TRAIL, namely DR4 and DR5, as well as FLIP, were deregulated in the resistant cells. To explore whether such disbalances mediated resistance to TRAIL, several biochemical and biological studies were carried out.

Initially, the role of DR4 and DR5 receptors in modulating the response to TRAIL was analyzed. qPCR studies confirmed the higher levels of DR4 and DR5 receptors in the parental cells, when compared with the resistant ones (Fig. 3A). Furthermore, analyses of cell surface DR4 and DR5, performed by flow cytometry (Fig. 3B), or total levels of these receptors by Western Blot (Fig. 3C) demonstrated that OPM2-TR cells expressed lower levels of the receptors when compared to naïve OPM2. Western blotting studies performed in MM1R cells and MM1R-TR cells also showed that both DR4 and DR5 receptors levels were down-regulated in MM1R-TR cells when compared to MM1R (Supplementary Fig. 1). In MM144 and RPMI-LR5 cells, which were intrinsically resistant to TRAIL, chronic treatment with TRAIL also decreased DR4 receptor levels. In contrast, such chronic treatment did not affect the amount of DR5 (Supplementary Fig. 2). Therefore, in the case of those two cell lines, their intrinsic resistance to TRAIL does not appear to depend on DR4 and DR5 receptor levels.

Fig. 3

TRAIL receptors are downregulated in TRAIL resistant cells. A Analyses of DR4 and DR5 levels in OPM2 and OPM2-TR cells. Total RNA from both cell lines was isolated and gene expression level determined by qRT-PCR. mRNA of DR4 (left) or DR5 (right) were normalized using GAPDH as a control. Graphs show mean ± SD of triplicates of a representative experiment repeated twice. Asterisk indicates p < 0.05. B Cell surface expression of DR4 and DR5 receptors in TRAIL sensitive and resistant cells. Cell surface receptors were stained with phycoerythrin-labeled anti-DR4 (left) or -DR5 (right) antibodies. The resulting overlapped histograms were plotted. Control: isotype control; blue line, OPM2 cells; Red line, OPM2-TR. C DR4 and DR5 total protein levels. OPM2 or OPM2-TR cells were grown, and protein extracts prepared and quantified. 1mg total protein was immunoprecipitated with specific antibodies against DR4 or DR5. Immune complexes were resolved, and receptors detected with the same antibodies. Red asterisks indicate the IgG band. D DISC formation in TRAIL sensitive and resistant cells. 50 × 10⁶ cells of each of the indicated cell lines were resuspended in 1ml of complete media containing 1$\mu$g/ml of bTRAIL an incubated at 37ºC for 0 or 30 min before being lysed. DISCs were pulled down by streptavidin-agarose and immune complexes subjected to WB and probed with the indicated antibodies

Next, the capability of TRAIL treatment to promote the formation of the DISC functional complex was analyzed. To that end, cells were treated with biotinylated TRAIL (bTRAIL) for 30 min and after cell lysis, streptavidin-agarose was added to the samples to pull down TRAIL-bound proteins. These experiments showed that the DISC complex was correctly assembled after bTRAIL treatment in OPM2 cells, since caspase 8, DR4 and DR5 coprecipitated with bTRAIL (Fig. 3D). In contrast, treatment of OPM2-TR with bTRAIL did not promote the efficient formation of the DISC complex.

Analyses of the induction of cell death, as measured by Annexin-V-propidium iodide (AV-PI) staining indicated that TRAIL increased the amount of AV-PI staining in naïve OPM2 (Fig. 4A). That increase was already detected at 3 h of treatment and reached a peak at 9 h (Fig. 4A and B). In contrast, OPM2-TR cells were largely resistant to the induction of AV-PI staining upon treatment with TRAIL. Biochemically, the induction of proteolytic processing of several proteins involved in the proapoptotic action of TRAIL was also explored. In OPM2 cells, TRAIL provoked cleavage of initiator and executor caspases such as caspase 8, caspase 9, caspase 7 and caspase 3 (Fig. 4C). In addition, TRAIL also provoked PARP processing, which is indicative of apoptotic cell death, and decreased the amount of BID or XIAP. No major changes were observed in the amounts of BCL2, BAX or MCL-1. On the other hand, when OPM2-TR cells were similarly analyzed, no changes in the apoptotic machinery or PARP cleavage were detected.

Fig. 4

Cell death induction of rTRAIL on sensitive and resistant cells. (A) Cytometric determination of TRAIL effect. Cells were treated with 1$\mu$g/ml TRAIL for the indicated times, and cell death determined by double staining with annexin V and PI and flow cytometry analysis. The percentage of cells entering apoptosis for each cell line and condition was determined and plotted in B. C DISC downstream proteins after TRAIL treatment. OPM2 or OPM2-TR cells were treated and lysed. The level and status of the indicated proteins was assessed by WB on 50 $\mu$g total protein extract. GAPDH was used as a loading control. (D) Caspase 3 activity induced by TRAIL on sensitive or resistant cells. Cells were treated with 1$\mu$g/ml TRAIL as indicated, and caspase 3 activity determined in a fluorometric assay. Data are shown as mean ± SD of triplicates of a representative experiment repeated twice; u.a. arbitrary units

Evaluation of caspase activity in both cell lines by fluorometric assays specifically detecting caspase 3 activation showed that such activity increased in naïve OPM2 cells after the third hour of treatment with TRAIL, with a maximal activity reached after 6 h (Fig. 4D, white bars). In contrast, in the resistant OPM2-TR line, only a very slight increase in caspase 3 activity was observed (Fig. 4D, black bars). Taken together, the above results indicated that TRAIL-induced death signaling was profoundly compromised in the OPM2-TR cells.

Genetic manipulation of DR4 and DR5 levels regulates TRAIL sensitivity

The above results showed that DR4 and DR5 were down-regulated in OMP2-TR and such characteristic could be responsible for the decreased sensitivity to TRAIL in those cells, as compared to naïve OPM2. Thus, recovering TRAIL receptor levels in OPM2-TR cells should resensitize cells to TRAIL action. To analyze that possibility, OPM2-TR cells were transduced with cDNAs coding for DR4 and the two versions of DR5, namely DR5-L and DR5-S, for DR5 long and short isoforms, respectively [35]. OPM2-TR cells were infected with retroviral vectors coding for either of these receptors and the expression of the three receptors analyzed by Western. In OPM2-TR cells transduced with DR4, bands of the expected size (approximately 50 KDa) and not visualized in the empty vector transduction (pLZR-IRES-GFP, transfection control) were observed (Fig. 5A). Similarly, when cells transduced with DR5 isoforms (DR5-L and DR5-S) were analyzed, bands corresponding to DR5-L and DR5-S isoforms were detected (Fig. 5B). Of note, the expressed receptor proteins appeared as double or triple bands, suggesting different degrees of posttranslational modifications. Once the expression of these proteins in OPM2-TR cells was verified, viability was determined by MTT uptake assays to investigate their sensitivity to TRAIL (Fig. 5C). In OPM2-TR cells infected with the control empty vector, TRAIL was unable to provoke an effect on cell survival (Fig. 5C). Cells transduced with DR4 or DR5-S receptors showed a higher sensitivity to TRAIL compared to those transduced with the empty vector. In the case of DR5-L we did not observe these differences and these cells behaved, in terms of resistance, as parental cells transduced with the empty control vector.

Fig. 5

TRAIL receptor levels are a key component in MM response to this agent. A and B Overexpression of DR4 and DR5 receptors in OPM2-TR cells. cDNAs coding for DR4 (A), or DR5-L or DR5-S (B) and subcloned into retroviral expression vectors were transfected into HEK-293T cells and conditioned media collected and used to transduce OPM2-TR cells. GFP positive cells were sorted by flow cytometry and once grown, lysed. The increase in DR4 or DR5-L or -S was tested in WB assays. C Reexpression of DR4 and DR5 receptors resensitize OPM2 resistant cells to TRAIL action. Transduced cells expressing the different receptors were stimulated with 1$\mu$g/ml TRAIL for 48 h, and cell proliferation was analyzed 48 h later in MTT assays. Data were normalized to their corresponding vehicle treated controls and shown as mean ± SD of triplicates of a representative experiment repeated twice. D and E Importance of DR receptors was validated in loss of function experiments. DR4 (D) and DR5 (E) levels were diminished in OPM2 cells by lentiviral transduction with commercial plasmids carrying shRNAs for the indicated receptors. The performance of the silencing was tested in IP and WB. F Besides, the sensitivity to TRAIL of the generated populations was measured in MTT assays as described. Data were analyzed and plotted as in C. Asterisk indicate statistically significant differences (p < 0.05)

On the other hand, we analyzed whether downregulation of DR4 and DR5 receptors in the sensitive line OPM2 was sufficient to induce TRAIL resistance. Lentiviral plasmid vectors containing short hairpin RNAs (shRNAs) to specifically block the expression of target genes were used. Preliminary experiments performed to identify the most efficient shRNAs from a set of five for each receptor showed that sh08 and sh11 were the most efficient in down-regulating DR4, while for DR5, the most efficient knockdown was achieved with sh05 (Fig. 5D and E and data not shown). Cell viability assays were then performed on the different populations of OPM2 cells transduced with the different vectors. Cells infected with viruses containing the DR4-targeting vectors were significantly more resistant to TRAIL than those transduced with the empty vector pLKO (Fig. 5F). Similarly, sh05, which decreased DR5 levels, significantly induced resistance to TRAIL.

Manipulation of the protein levels of the apoptotic inhibitor FLIP

The TRAIL pathway analysis performed with the microarray data showed that in addition to a decrease in TNFRSF10A (DR4) and TNFRSF10B (DR5), an increase in another DISC component, CFLAR (FLIP) was present in OPM2-TR (Fig. 2C and D). Western blot analyses confirmed up-regulation of FLIP in OPM2-TR cells (Fig. 6A). In addition, a small decrease in two other DISC components, namely caspase 8 and FADD, was observed.

Fig. 6

Importance of cFLIP levels in TRAIL response. A Analysis of DISC components in OPM2 and OPM2-TR cells. Total protein extracts were prepared, separated in SDS-PAGE and membranes probed with the indicated antibodies. B The importance of cFLIP on TRAIL response was analyzed in gain of function experiments. cFLIP levels were increased in OPM2 cells by retroviral transduction and the levels of this protein assessed by IP and WB with anti-Flip antibodies. C The response to TRAIL of the resultant populations were determined by MTT assays as described. D Similarly, loss of function experiments were carried out to decrease FLIP levels in OPM2-TR cells. Lentiviral transfection was accomplished to diminish FLIP protein in OPM2-TR cells and new expression was tested by IP and WB as described. E The sensitivity to TRAIL of the transduced population was analyzed in MTT assays. In all the MTT assays shown, data were normalized to vehicle-treated controls and mean ± SD of triplicates of a representative experiment repeated twice are plotted. Asterisk denote statistically significant differences (p < 0.05)

Because of the role of FLIP as a negative regulator of signaling through the DISC complex, we decided to analyze the regulation of signaling by TRAIL in an analogous way as above, i.e. by loss of function studies in OPM2-TR and gain of function studies in OPM2. For the FLIP gain-of-function experiments, naïve OPM2 cells were retrovirally infected with a vector coding for FLIP. After selection, OPM2 cells expressed exogenously-transfected FLIP (Fig. 6B). MTT metabolization experiments revealed that OPM2 cells infected with the retroviral vector coding for FLIP were resistant to the action of TRAIL (Fig. 6C). Of note, OPM2 cells infected with retroviruses which included the empty vector were more resistant to the action of TRAIL than naïve OPM2. This result suggests that resistance to TRAIL was altered in the empty vector-transduced population even though their resistance to TRAIL was lower than that observed in cells infected with the vector coding for FLIP.

For the loss of function experiments, OPM2-TR cells were infected with retroviruses containing coding sequences to target FLIP. These infections resulted in a decrease in the amount of FLIP as measured by Western (Fig. 6D). MTT metabolization assays showed that cells with less FLIP were more sensitive to TRAIL than cells transduced with the empty vector pLKO (Fig. 6E), even though that decrease did not reach statistical significance.

Discussion

The development of strategies that activate the extrinsic pathway of apoptosis through cell death receptors represents a very attractive antitumoral option [21, 44]. The fact that TRAIL is a death ligand that induces apoptosis in tumor cells with minimal action on normal cells [45,46,47,48] makes it an excellent therapeutic option. In fact, several therapies that use this pathway to kill tumor cells are currently under development [24, 49]. In this regard, phase I, II and III clinical trials are being developed using TRAIL or modified forms of this molecule in monotherapy or in combination with other therapies (dexamethasone and/or thalidomide) in patients with relapsed MM, with quite encouraging results [33, 34, 50, 51].

Many cancer therapies are limited because of the problem of resistance generated to the antitumoral drugs 52, [53]. Given the clinical interest in the use of TRAIL as an antitumoral drug and intending to anticipate to the development of resistance against such treatment, we decided to study the mechanisms involved in the generation of resistance to TRAIL. Initial studies shown in Fig. 1A indicated that treatment of OPM2, MM1S and MM1R cells with TRAIL resulted in a profound decrease in cell viability. In contrast, TRAIL failed to affect the viability of MM1-144, U266, or RPMI-LR5 myeloma cells. These results indicated that the latter cell lines present primary resistance to TRAIL, and advert about the relevance of performing a precise patient selection for potential therapies using TRAIL in MM.

Former studies have shown that MM cell lines acquire tolerance to TRAIL during prolonged treatment [53]. By using such approach, we were able to generate several MM cell lines secondarily resistant to TRAIL (Fig. 1B). One of these models, OPM2-TR was selected for further characterization. Gene expression profiling analyses showed clear differences when compared to naïve OPM2 (Fig. 2). When analyzing the overexpressed genes we found, among others, the TNFAIP3 gene, which codes for a zinc finger and ubiquitin modifying enzyme rapidly induced by TNFα. The TNFAIP3 gene product inhibits the NF-kappa B pathway as well as TNFα-mediated apoptosis, and this may contribute to tumor progression in myeloma patients [54]. Another overexpressed gene was the one coding for caspase 1, part of the NLRP3 inflammasome-caspase 1 complex, which is related to increased plasma cells in MM patients [55]. Among the under-expressed genes, NLRP11, which also belongs to the inflammasome complex, was also identified. That protein promotes the maturation of proinflammatory cytokines contributing to susceptibility to inflammatory processes in MM [56]. In addition to NLRP11, the genes coding for the solute carrier SLC7A2, and the transposon regulatory protein TEX19 were also profoundly down-regulated in the resistant cells.

When we searched for genes involved in the TRAIL-induced cell death pathway we found that OPM2-TR cells had lower expression of the mRNAs coding for DR4 and DR5 receptors than the parental cell line. In addition, the resistant cells overexpressed the apoptosis inhibitor FLIP. The decrease in the receptors and the increase in FLIP are likely to be responsible for the inhibition of TRAIL signaling observed in the OPM2-TR cells. In those cells not only the DISC components failed to assembly (Fig. 3), but also downstream signaling was profoundly inhibited (Fig. 4). In our resistance model we did not observe activation of the extrinsic pathway (caspase 8/caspase 3), results in line with previous work [57]. Furthermore, the induction of PARP cleavage, a readout of apoptotic signaling, was profoundly compromised in OPM2-TR cells treated with TRAIL as compared to parental OPM2. These results are in agreement with data reported in studies carried out in other cell lines [58]. Similarly, we observed processing of multiple proteins involved in these pathways after treatment of OPM2 cells, but not of resistant ones. This occurred with caspases 3, 7 or 9 in addition to MCL-1, BID or XIAP. All these data would confirm the importance of these proteins in apoptosis in the OPM2 model [28, 30, 37, 59].

The fact that several proteins involved in TRAIL-induced apoptosis were altered in the resistant cells suggested that efficient resistance may depend on concerted changes in various molecules involved in TRAIL signaling. This conclusion is suggested by the genetic experiments used to manipulate DR4, DR5, or FLIP levels. In fact, expression of DR4 in OPM2-TR cells, which under basal conditions had minimal amounts of this receptor, partially rescued TRAIL sensitivity. These data would agree with what has been described in the literature that speaks about the importance of DR4 in the response to TRAIL [60]. Other groups reported that the most important receptor mediating this response is DR5 [61]. In our model, overexpression of DR5S also achieved rescue of sensitivity to TRAIL. However, we did not observe such phenotypic rescue when we increased DR5L receptor levels. That DR4 and DR5 levels regulated sensitivity to TRAIL in OPM2 cells was confirmed by knockdown experiments, which showed that a decrease in the levels of DR4 or DR5 provoked greater resistance to TRAIL. Therefore, in OPM2 cells, signaling through both DR4 and DR5 appear to control apoptotic signaling induced by TRAIL (Fig. 5).

Regarding FLIP, its overexpression in the OPM2 sensitive line was associated with increased resistance to TRAIL (Fig. 6). Our results, together with others described in the literature [62] demonstrate the importance of FLIP as a regulatory factor of sensitivity to TRAIL. On the other hand, knockdown experiments carried out in OPM2-TR indicated that decreasing the amount of FLIP partially rescued sensitivity to TRAIL. These results open the possibility of acting on this protein to augment MM cell sensitivity to TRAIL. A model of how TRAIL resistance emerges in MM cells is now represented in Fig. 7.

Fig. 7

Modelation of acquired TRAIL resistance in MM. When continuously grown in the presence of TRAIL, naïve MM cells sensitive to TRAIL develop acquired resistance to that agent through the loss of DR4 and DR5 receptors. At the same time an increase in the system inhibitor FLIP is observed. Thus, due to these alterations, cells are able to proliferate in the presence of the death ligand.

It is important to highlight the translational impact of our results, given the ongoing efforts to use TRAIL and its agonists in the clinic [24]. Thus, identifying patients who could benefit from these therapies is critical. Moreover, as we have already discussed, one of the characteristics of MM and other neoplastic diseases is the emergence of resistance to therapies. Therefore, it seems that it is essential to anticipate them and identify the mechanisms that could cause them, as well as to look for possible therapeutic alternatives once they appear. In this way we can identify biomarkers that can predict patient response and maximize therapeutic efficacy through drug combinations that synergize with TRAIL, in addition to understanding and overcoming resistance as it emerges [53].