Research ArticlePharmacology

The transcription factor SP3 drives TNF-α expression in response to Smac mimetics

See allHide authors and affiliations

Science Signaling  29 Jan 2019:
Vol. 12, Issue 566, eaat9563
DOI: 10.1126/scisignal.aat9563

SP3 is key to Smac mimetic efficacy

Finding drugs that selectively target tumor cells and spare healthy tissue is a challenging goal. Smac mimetic compounds have emerged as promising cancer therapies by blocking anti-apoptotic machinery and inducing the production of the cytokine TNF-α. Beug et al. found not only that the transcription factor SP3 critically mediates the molecular and cellular effects of Smac mimetics but also that it is more abundantly expressed in tumors than in normal tissues from the same patients. These findings suggest that using SP3 as a biomarker may identify patients that will best respond to and tolerate these drugs.

Abstract

The controlled production and downstream signaling of the inflammatory cytokine tumor necrosis factor–α (TNF-α) are important for immunity and its anticancer effects. Although chronic stimulation with TNF-α is detrimental to the health of the host in several autoimmune and inflammatory disorders, TNF-α—contrary to what its name implies—leads to cancer formation by promoting cell proliferation and survival. Smac mimetic compounds (SMCs), small-molecule antagonists of inhibitor of apoptosis proteins (IAPs), switch the TNF-α signal from promoting survival to promoting death in cancer cells. Using a genome-wide siRNA screen to identify factors required for SMC–to–TNF-α–mediated cancer cell death, we identified the transcription factor SP3 as a critical molecule in both basal and SMC-induced production of TNF-α by engaging the nuclear factor κB (NF-κB) transcriptional pathway. Moreover, the promotion of TNF-α expression by SP3 activity confers differential sensitivity of cancer versus normal cells to SMC treatment. The key role of SP3 in TNF-α production and signaling will help us further understand TNF-α biology and provide insight into mechanisms relevant to cancer and inflammatory disease.

INTRODUCTION

The importance of tumor necrosis factor–α (TNF-α) in biology and disease is underscored in that it is the second most studied gene in the human genome (1). Although TNF-α was identified in 1962, it is only within the last two decades that we gained an understanding of the signaling components and pathways that mediate TNF-α–induced activation of the classical nuclear factor κB (NF-κB) pathway or triggering of programmed cell death in cancer cells. These pathways are critically dependent on receptor interacting protein 1 (RIP1), which is ubiquitinated by two inhibitor of apoptosis proteins (IAPs), cellular IAP1 (cIAP1) and cIAP2, to form a signalosome scaffold complex. The ubiquitinated complex is required for the downstream signaling of kinases, ultimately resulting in the transcription of NF-κB gene targets. However, when cIAP1/2 expression is suppressed, de-ubiquitinated RIP1 can form distinct death complexes, leading to TNF-α–induced cell death through caspase-8–mediated apoptosis or RIP3/MLKL-mediated necroptosis.

A class of IAP small-molecule antagonists termed Smac mimetic compounds (SMCs) have emerged for the potential treatment of cancer (2). SMCs were designed to mimic the binding of a conserved IAP binding motif of the pro-apoptotic mitochondrial protein second mitochondrial activator of caspases, which binds to the BIR2 and BIR3 domains of cIAP1, cIAP2, and X-linked IAP (XIAP) with differing affinities. The interaction of SMCs with cIAP1 and cIAP2 (hereafter, cIAP1/2) induces rapid activation of their ubiquitin E3 ligase RING domains, leading to K48-linked auto- and trans-ubiquitylation and subsequent proteasomal degradation of the cIAPs (3). In the absence of cIAP1/2, RIP1 is no longer ubiquitinated in response to stimulation by TNF-α, forming a ripoptosome platform that can include Fas-associated protein with death domain (FADD), RIP1, and caspase-8, and thereby leading to activation of downstream effector caspases and ultimately cancer cell death (4, 5). Moreover, the SMC-mediated degradation of the cIAPs leads to stabilization of NF-κB–inducing kinase (NIK), a protein that is continuously targeted for proteasomal degradation by the cIAPs, which then activates the alternative NF-κB pathway (6, 7).

To gain insight into the mechanism of SMCs, we undertook a functional genome-wide small interfering RNA (siRNA)–based screen and identified targets that were required for SMC efficacy in cancer cells. We report here that the transcription factor specificity protein 3 (SP3) is a critical positive regulator of SMC efficacy by stimulating the production of TNF-α and downstream engagement of the NF-κB pathway. Moreover, the promotion of TNF-α expression by SP3 activity also confers differential sensitivity of cancer versus normal cells to SMC treatment. Collectively, our findings demonstrate that SP3 functions as a critical transcription factor of SMC-mediated death of cancer cells by driving the transcription of the gene encoding TNF-α.

RESULTS

A genome-wide siRNA screen identifies SP3 as a critical factor for SMC-mediated cancer cell death

To explore the mechanism underlying SMC-mediated cancer cell death, we performed a rescue screen in which all genes were individually down-regulated using siRNA before treatment with an SMC. We used an Alamar blue assay, which measures the activity of the mitochondrial respiratory chain (8), as a readout of cell viability. For the screen, we used the SMC-sensitive triple-negative breast cancer cell line, MDA-MB-231 (Fig. 1A). This cell line produces TNF-α in an autocrine fashion and is dependent on RIP1 for SMC-induced cell death (912), which enabled us to assess the robustness of this genome-wide assay by comparing SMC sensitivity of MDA-MB-231 cells treated with control, “nontargeting” (NT) siRNA or with RIP1-targeted siRNA (fig. S1, A and B). The assay provided a wide dynamic range and negligible data variability of cell viability as measured by Alamar blue, resulting in a Z factor of 0.54 (fig. S1C), a measure of statistical effect size indicating that the screen was suitable for identifying likely mediating factors (13). The efficiency of siRNA targeting RIP1 was confirmed by Western blotting (fig. S1D). In total, we identified 429 genes with Z scores (a measurement of the number of standard deviations from the mean) above 2.05, suggesting that these targets are critical for SMC efficacy in MDA-MB-231 cells (Fig. 1B and table S1). Among these candidates were factors known to be involved in SMC-induced death, including TNF-α and RIP1, indicating that the screen was successful in identifying genes required for cancer cell death induced by SMC treatment.

Fig. 1 A genome-wide siRNA screen identifies SP3 as a sensitizer to SMC efficacy in cancer cells.

(A) Alamar blue viability assay of MDA-MB-231 cells treated with vehicle or SM-164 (100 nM) for 48 hours. (B) Quantile-quantile plot showing Z score for siRNA pools (ranks) in the genome screen. The siRNA pool targeting SP3 is indicated by the blue circle. The dashed line indicates the proximity of the lowest Z score of the positive control (siRNA targeting RIP1). (C) Alamar blue viability assay of MDA-MB-231 cells transfected with control (NT) siRNA, pooled SP3 siRNA, or deconvoluted SP3 siRNA for 72 hours and subsequently treated with vehicle or SM-164 (100 nM) for 48 hours. (D) Knockdown efficacy of siRNA targeting SP3 from the experiment depicted in (C). (E) Alamar blue viability assay of MDA-MB-231 cells transfected with plasmids encoding wild-type (WT) or siRNA-resistant (RiR) SP3 cDNA and NT or SP3 #9 siRNA and for 48 hours, and subsequently treated with vehicle or 100 nM SMC for 48 hours. (F) Alamar blue viability assay of cells transfected with NT or SP3 siRNA for 48 hours and subsequently treated with vehicle or LCL161 (500 nM). Bottom, Western blots showing efficacy of SP3 knockdown for the indicated cancer cell lines. Data are means ± SD from n = 3 or 4 experiments (individual data indicated); *P < 0.05, **P < 0.01, and ***P < 0.001 by unpaired Student’s t test (A) or two-way analysis of variance (ANOVA) using Dunnett’s multiple comparison test (C, E, and F).

As a follow-up, we repeated the genome-wide siRNA screen by assessing for the ability of the top 300 candidate genes to rescue SMC-induced cancer cell death using a deconvoluted pool of siRNAs. Among these candidates, all four deconvoluted siRNAs targeting SP3 blunted SMC death in MDA-MB-231 cells (fig. S2). We choose to study SP3 further because of the limited knowledge around that transcription factor and TNF-α or SMC biology. To confirm that SP3 is a critical factor for SMC efficacy, we transfected MDA-MB-231 cells with a separate dissimilar deconvoluted pool of siRNAs targeting SP3. All four individual SP3 siRNAs prevented MDA-MB-231 cells from dying in response to SMC (Fig. 1, C and D). The major isoforms derive from alternative translational start sites and posttranslational modifications, denoted as long, medium, and short SP3 (L-SP3, M-SP3, and S-SP3). To provide additional evidence that SP3 mediated cell death upon SMC treatment, we conducted a rescue experiment: Transfection of MDA-MB-231 cells with a single deconvoluted siRNA (#9 from Fig. 1D) (14) and a plasmid expressing wild-type SP3 or siRNA-resistant SP3 revealed that cell death occurred in cells expressing cDNA encoding siRNA-resistant but not with wild-type SP3 (Fig. 1E). We next assessed whether the dependence of SMC efficacy on SP3 is evident in other SMC-sensitive cancer cell lines. Similar to that which was found for MDA-MB-231 cells, these cell lines are sensitive to SMC treatment because of the autocrine production of TNF-α, which represents ~5 to 15% of cultured cancer cells (1519). In seven of eight cancer cell lines of differing histological origins tested, the knockdown of SP3 substantially rescued viability in the presence of SMC (Fig. 1F), indicating that SP3 is critical for SMC efficacy in various types of cancer cells.

Of the SP transcription family members, SP1 and SP3 share the same consensus binding sequence, with about 12,000 SP1/SP3 binding sites in the human genome (20, 21). Although SP1 was not identified as a candidate from the genome-wide siRNA screen, we nevertheless determined whether the related transcription factor SP1 might have a similar functional role in mediating cancer cell death upon SMC treatment. Knockdown of SP1 did not prevent MDA-MB-231 cell death in response to SMC treatment, although it induced considerable cell death on its own (Fig. 2, A and B). Consistent with these results, the application of an SP1 chemical inhibitor, mithramycin A (2224), did not prevent the death of cancer cells from SMC treatment (fig. S3A). SP3-mediated knockdown markedly rescued MDA-MB-231 cells from SMC-induced cell death; however, in other cell lines (such as M059J), the knockdown of SP3 only partially prevented SMC-induced cell death (Fig. 1F and fig. S3, C and D). To further investigate the requirement of SP1 in SMC-mediated cell death, we down-regulated SP1 and SP3 in M059J cells. In these cells, SP1 has a minor role in the efficacy of SMC treatment, because there was a proportionally smaller rescue in SMC-treated M059J cells that were transfected with SP1-siRNA alone or SP1- and SP3-siRNA combined (Fig. 2, A and B). However, treating M059J cells with mithramycin A did not affect SMC efficacy (fig. S3A). Together, these data indicate that the transcription factor SP3, but not SP1, is a critical factor that determines SMC efficacy.

Fig. 2 cIAP1, cIAP2, and XIAP cooperatively protect cancer cells from SP3-mediated cancer cell death.

(A) Alamar blue viability assay of MDA-MB-231 and M059J cells transfected with control (NT), SP1, or SP3 siRNA for 48 hours and subsequently treated with vehicle or LCL161 (500 nM) for 48 hours. (B) Knockdown efficacy of NT, SP1, or SP3 siRNA from the experiment described in (A). Blots are representative of three independent experiments. (C) Alamar blue viability of cells transfected with NT, SP3, cIAP1, cIAP2, or XIAP for 48 hours. (D) Representative siRNA efficacy Western blots for the experiment depicted in (C). Blots are representative of two independent experiments. (E) Viability of cells transfected with NT or SP3 siRNA and treated with vehicle or the indicated SMC (1 μM) for 48 hours. Viability was assessed by Alamar blue. Data (A, C, and E) are means ± SD from n = 3 experiments; *P < 0.05, **P < 0.01, and ***P < 0.001 by two-way ANOVA using Dunnett’s multiple comparison test.

Cellular depletion of SP3 prevents SMC-induced apoptosis

The ability of SMCs to antagonize the IAPs—namely, cIAP1, cIAP2, and XIAP—is critical for sensitizing cancer cells to cytokine-mediated cell death. To assess the specificity of SP3 for IAP-mediated cell death, we individually and collectively knocked down SP3, cIAP1, cIAP2, and XIAP abundance by transfecting cells with siRNAs. In MDA-MB-231 and M059J cells, the presence of SP3 was required for death of cancer cells upon depletion of all three IAP proteins (Fig. 2, C and D). In addition, SP3 promoted the death response of cancer cells treated with different monovalent and bivalent SMCs (Fig. 2E). To further assess the specificity of SP3 for the induction of cancer cell death, we analyzed the ability of SP3 to contribute toward the death of cancer cells caused by different apoptotic triggers. The down-regulation of SP3 did not block the induction of cancer cell death by VP16 (etoposide), staurosporine, or cycloheximide (fig. S4). Overall, these results suggest that SP3 is a specific mediator of SMC-induced cancer cell death through triple antagonism of cIAP1, cIAP2, and XIAP.

A major mode of action for cell death in SMC-sensitive cancer lines is by the induction of TNF-α and concurrent antagonism of the cIAPs to induce caspase-8–mediated apoptosis (12, 16, 17, 25). Accordingly, to determine the underlying mechanistic role of SP3 in SMC treatment, we assessed for the presence of obligate proteins for cytokine-mediated apoptosis. As expected, SMC treatment of MDA-MB-231 and M059J cells that was transfected with NT siRNA resulted in processing and down-regulation of full-length FLIP (an endogenous caspase-8 inhibitor) and activation of the downstream caspase-8 and caspase-3 (Fig. 3A). We also observed the presence of a smaller fragment of SP3, which may indicate that SP3 is a caspase substrate (26). In contrast, knocking down SP3 blocked the ability of cells to process cFLIP and to activate caspase-8 and caspase-3 (Fig. 3A). The ability of SMCs to induce caspase-dependent processing through engagement by TNF-α is dependent on the formation of a ripoptosome-containing death complex that includes FADD, RIP1, and caspase-8. Accordingly, we conducted endogenous coimmunoprecipitation experiments to determine whether SP3 promotes the formation of this death complex. In MDA-MB-231, M059J, and SNB75 cells, SMC treatment in cells transfected with control siRNA led to the formation of the death complex, but knocking down SP3 abundance abrogated its formation (Fig. 3B).

Fig. 3 Down-regulation of SP3 rescues cancer cells from SMC-mediated apoptosis.

(A) Cells were transfected with control (NT) siRNA or siRNA targeting SP3 for 48 hours and then treated with vehicle or LCL161 (500 nM). Cells were harvested for Western blotting at the indicated posttreatment times. Blots are representative of two independent experiments. (B) Cells were treated with vehicle or LCL161 (500 nM) for 24 hours, and endogenous caspase-8–associated complexes were isolated by immunoprecipitation (IP), resolved by SDS–polyacrylamide gel electrophoresis, and probed for the presence of proteins by the indicated antibodies. Blots are representative of two independent experiments. (C and D) Cells were transfected with NT or SP3 siRNA for 48 hours, treated with vehicle or LCL161 (500 nM) for 18 hours, and then processed for flow cytometry with fluorescein isothiocyanate–conjugated annexin V (ANXA5-FTIC) and 7-aminoactinomycin D (7-AAD). Representative flow cytometry plots (C); data (D) are means ± SD from n = 3 experiments. (E) MDA-MB-231 cells were transfected with NT or SP3 siRNA for 48 hours, reseeded in equal numbers, treated with vehicle or SM-164 (100 nM) for 7 days, and stained with crystal violet. Scale bars, 5 mm. Images are representative of two independent experiments.

These results indicate that SP3 positively regulates SMC-induced cell death. We next measured apoptosis using flow cytometry by identifying the number of cells stained for annexin V (ANXA5) and 7-AAD. Consistent with the lack of formation of a death-inducing complex and the downstream activation of caspases in SP3–down-regulated MDA-MB-231 and M059J cells, we detected fewer apoptotic, necrotic, and dead cells by flow cytometry (Fig. 3, C and D). Furthermore, knocking down SP3 abundance enabled increased clonogenic survival of MDA-MB-231 cells that were treated with SMC, as compared to controls (Fig. 3E). Together, these results are consistent with the requirement for the presence of SP3 to induce caspase-mediated cell death upon SMC treatment.

SP3 promotes the production of TNF-α

Both cIAP1 and cIAP2 have critical regulatory roles in the control of the NF-κB pathways (6, 9, 27, 28). The cIAPs positively activate the classical NF-κB pathway upon stimulation by TNF-α by coordinating the formation of a regulatory signalosome. On the other hand, the cIAPs negatively regulate the activation of the alternative NF-κB pathway (which is mediated by other TNF superfamily ligands and their cognate receptors) by down-regulating NIK. Accordingly, we assessed whether SP3 has a role in the activation of either of the NF-κB pathways. Because SP3 is a ubiquitous transcription factor that has about 6000 SP1/SP3 consensus sites within the human genome (29), we first probed for the presence of factors required for activation of NF-κB by Western blotting. The down-regulation of SP3 in MDA-MB-231 cells blunted SMC-induced up-regulation of the TNF-α receptor 1 (TNF-R1), and SP3 deficiency did not lead to the absence of RIP1, or the NF-κB dimers, p105/p50, RelA (a.k.a. p65), p100/p52, and RelB (Fig. 4A). However, SP3 down-regulation in SMC-treated MDA-MB-231 cells impaired the processing of p105 to p50 (a component of the classical NF-κB pathway) but did not affect the processing of p100 to p52 [a component of the alternative NF-κB pathway (Fig. 4A)].

Fig. 4 SP3 promotes the production of TNF-α.

(A) Western blotting of cells transfected with control (NT) or SP3 siRNA for 48 hours and then treated with vehicle or LCL161 (SMC; 500 nM) for 8 hours. Blots are representative of two independent experiments. (B) Detection of TNF-α at the mRNA level, assessed by reverse transcriptase quantitative polymerase chain reaction (RT-qPCR), in cells transfected with NT or SP3 siRNA for 48 hours and then treated with vehicle or LCL161 (500 nM). (C) Cells were transfected with NT or SP3 siRNA and vehicle or LCL161 (500 nM) for 24 hours. TNF-α in supernatants was measured by enzyme-linked immunosorbent assay (ELISA). (D) Cells were transfected with NT or SP3 siRNA for 48 hours and then treated with vehicle or LCL161 (500 nM) for 9 hours, and cellular abundance of TNF-α was measured in viable cells (Zombie Green negative) by flow cytometry. Data are representative of two independent experiments. (E) Alamar blue viability assays of cells transfected with NT or SP3 siRNA for 48 hours and treated with combinations of vehicle, 100 nM SM-164, 0.01% bovine serum albumin (BSA), and interleukin-1β (IL-1β; 10 ng/ml). (F) Alamar blue viability assay of cells transfected with NT or SP3 siRNA (48 hours for SNB75 and 24 hours for EMT6) and subsequently treated with vehicle or 1 μM LCL161 and 0.01% BSA, lipopolysaccharide (LPS; 1 μg/ml), polyinosinic-polycytidylic acid [poly(I:C)] (1 μg/ml), or interferon-β (IFN-β; 250 U/ml). (G) Cells were transfected with NT or SP3 siRNA for 48 hours and subsequently treated with vehicle or 500 nM LCL161 and 0.01% BSA, IL-1β (10 ng/ml), or IFN-β (250 U/ml) for 24 hours. Supernatants were processed for the presence of TNF-α by ELISA. (H) Efficacy of siRNA transfections from the experiments depicted in (E) and (F). Data are means ± SD from n = 3 experiments; *P < 0.05 and ***P < 0.001 by one-way ANOVA using Dunnett’s multiple comparison test.

The ubiquitination of RIP1 upon activation of the classical NF-κB pathway is dependent on the presence of TNF-α. However, whereas SMC treatment and cIAP depletion blunt the activation of the classical NF-κB pathway (shifting the prosurvival signalosome into a ripoptosome death complex), SMC-mediated antagonism of cIAP1/2 activates the alternative NF-κB pathway, leading to increased TNF-α expression (16). We observed that SMC treatment leads to the activation of the classical NF-κB pathway as seen by the processing of p105 to p50 and the ubiquitination of RIP1 (Fig. 4A), as previously reported (16, 19, 30). However, the siRNA-mediated down-regulation of SP3 blunted SMC-induced RIP1 ubiquitination and classical NF-κB activation. We also observed a decrease in the SMC-induced proteolysis of RelB in the absence of SP3, which can be attributed to caspase-3–mediated cleavage of RelB and/or proteolysis of RelB upon activation of the NF-κB pathway (31, 32). We then investigated the expression of TNF-α in SMC-sensitive cell lines, namely, MDA-MB-231 and M059J. We observed an increase in TNF-α transcript expression upon SMC treatment, which was abrogated by SP3 knockdown (Fig. 4B). We concurrently observed reduced secretion of TNF-α into the cell culture supernatant and less intracellular TNF-α in SMC-treated cells in the absence of SP3 (Fig. 4, C and D). Thus, SP3 is required for the induction of TNF-α by SMC treatment.

Several immunostimulatory ligands have been documented to induce SMC-mediated cancer cell death by the secondary production of pro-inflammatory ligands such as TNF-α or prodeath ligands such as TNF-related apoptosis-inducing ligand (TRAIL) (33, 34). For instance, IL-1β treatment induces NF-κB–dependent production of TNF-α, which subsequently triggers RIP1- and caspase-8–dependent death of SMC-treated cancer cells (33). We investigated whether SP3 has a prodeath role with SMCs and IL-1β. In SNB75, U118, and U2OS cancer cells, the combination of SMC and IL-1β leads to cell death, but this death was at least partially prevented by the down-regulation of SP3 (Fig. 4, E and H). This rescue of cell death coincided with the reduced secretion of TNF-α in IL-1β and SMC-treated cancer cells (Fig. 4G). Furthermore, some cancer cells can be induced to die with cotreatment of SMC and LPS, poly(I:C), or IFN-β (33, 34). The down-regulation of SP3 prevented cells from death induced by the cotreatment of these agents with SMC in human SNB75 and mouse EMT6 cancer cells and was associated with a decrease in the amount of TNF-α (Fig. 4, F to H).

SP3 has the potential to control the transcription of a large number of genes (35). Accordingly, we assessed whether SP3 globally regulates the expression of cytokines and chemokines. The induction of cytokines and chemokines by SMC treatment was partially unaffected by SP3 down-regulation, as we observed down-regulation in 51% (34 of 66) of SMC-induced genes in SP3-deficient cells (Fig. 5A and fig. S5). Together, these results suggest that a major role for SP3 in SMC-treated cancer cells is to stimulate the production of TNF-α and possibly other pro-inflammatory ligands and chemokines. The potential roles of these genes in the induction of various factors secreted by SMC treatment are currently being investigated.

Fig. 5 SP3 positively controls the activity of the TNF-α promoter.

(A) MDA-MB-231 cells were transfected with control NT siRNA or siRNA targeting SP3 for 48 hours and then treated with vehicle or LCL161 (500 nM) for 8 hours. Cells were processed for quantitation of 176 genes encoding cytokines and chemokines by RT-qPCR. Shown are normalized heat maps of two major groups identified by hierarchical clustering. The complete hierarchical cluster profiles are provided in fig. S5. Data represent means from four independent experiments. (B) MDA-MB-231 cells stably expressing NT short hairpin RNA (shRNA) or shRNA targeting SP3 were transfected with plasmids encoding luciferase under the control of the TNF-α promoter for 24 hours. Cells were processed for luciferase assays at the indicated posttreatment times with LCL161 (500 nM). Data are means ± SD from n = 3 experiments. (C) Schematic of predicted or reported binding sites of the human TNF-α promoter for SP1, NF-κB, and IRF1. (D) Cells were treated with vehicle or 500 nM LCL161 for 6 hours and processed for chromatin immunoprecipitation (ChIP) using primers spanning the indicated regions of the TNF-α promoter. (E) Alamar blue viability assays of cells transfected with NT or SP3 siRNA for 48 hours and subsequently treated with vehicle or 500 nM LCL161 and 0.01% BSA or TNF-α (1 ng/ml) for 24 hours. (F) Cells were transfected twice with siRNAs targeting NT or the indicated genes for 24 and 48 hours and then treated with vehicle or LCL161 (500 nM) for 24 hours. Viability was measured by Alamar blue. (G) Luciferase assays of MDA-MB-231 cells stably expressing NT or SP3 shRNA that were transfected with plasmids encoding luciferase under the control of the TNF-α promoter for 24 hours, treated with vehicle or LCL161 (500 nM) for 2 hours and 0.01% BSA or TNF-α (1 ng/ml) for 24 hours. (H) SNB75 cells stably expressing NT or SP3 shRNA were transfected with plasmids encoding luciferase under the control of the TNF-α promoter for 24 hours and were treated with vehicle or LCL161 (1 μM) and BSA (0.01%) or TNF-α (1 ng/ml). At the indicated times, cells were processed for luciferase assays. Data are means ± SD from n = 3 experiments; *P < 0.05 and ***P < 0.001 by two-way ANOVA using Sidak’s (D) or Dunnett’s (E, F, and G) multiple comparison test.

SP3 regulates NF-κB activity at the promoter of the gene encoding TNF-α

Depending on the molecular and cellular context, SP3 can function as a transcriptional activator or repressor (21). Because the SP3 transcription factor is required for the basal and stimulated induction of expression of the gene encoding TNF-α (hereafter, TNF-α), we next assessed whether SP3 activity is affected during SMC treatment. We did not observe a difference in the ability of SP1/SP3 to bind to nuclear extracts derived from SMC-treated cancer cells (fig. S6). We then assessed whether SP3 enhances the promoter activity of TNF-α. We observed that TNF-α promoter activity is elevated in SMC-treated MDA-MB-231 cells; however, TNF-α promoter activity is attenuated by shRNA-mediated down-regulation of SP3 (Fig. 5B and fig. S7A). Because these results suggest that SP3 is a critical transcription factor that influences TNF-α expression, we profiled for known and putative SP1/SP3 consensus DNA binding sites on the promoter of TNF-α. There are seven predicted SP1/SP3 consensus binding sites on the promoter of TNF-α (Fig. 5C), and these regions are closely associated with NF-κB consensus response elements. We evaluated the ability of SP3 to promote TNF-α expression by measuring SP3 binding to the TNF-α promoter in SMC-treated cancer cells using ChIP and qPCR techniques. In MDA-MB-231 and M059J cells, we observed that SMC treatment enhances SP3 binding at the sites −173 and −565/−534 within the SP1/SP3 consensus sequences (Fig. 5D).

We found that the production of TNF-α by immunostimulatory factors is dependent on SP3 (Fig. 4, E to H). Cancer cells still express the core components required for caspase-mediated apoptosis (Fig. 3B). Therefore, we explored whether SMC treatment can induce cancer cell cytotoxicity in response to stimulation with exogenous TNF-α. As expected, down-regulation of SP3 prevented MDA-MB-231 cells from SMC treatment–induced death; however, the application of exogenous TNF-α in SP3-deficient cells did not lead to cell death in SMC-treated MDA-MB-231 and SNB75 cells and is specific for SP3 (Fig. 5E and figs. S3B and S8). In addition, we observed that down-regulation of TNF-α blunted SMC-induced cancer cell death at a level similar to down-regulation of TNF-R1 and SP3 (Fig. 5F). These observations correlate with the reduction of TNF-α promoter activity in MDA-MB-231 and SNB75 cells in which SP3 was silenced by expression by shRNA (Fig. 5, G and H, and fig. S7B).

Collectively, our results indicate that SP3 is a critical transcription factor responsible for TNF-α gene expression. Next, we examined whether SP3 affects activation of the NF-κB pathways, which have critical roles for the induction of TNF-α expression. We first investigated whether down-regulation of nuclear components linked to NF-κB activity or TNF-α expression would affect SMC efficacy. Consistent with a previous report, the down-regulation of IRF1 partially prevented cancer cell death by SMC treatment (Fig. 6, A and B) (36), which has two predicted binding sites on the TNF-α promoter (Fig. 5C). However, there was no rescue of cell from death with exogenous TNF-α treatment. A similar trend was observed with down-regulation of one of the NF-κB constituents, RelA, but not with p105, p100, or RelB (Fig. 6, A and B). These results indicate that SP3 is a critical component for TNF-α–induced death of SMC-treated cancer cells and that SP3 stimulates TNF-α gene expression at the transcription initiation step.

Fig. 6 SP3 stimulates NF-κB activity.

(A) Alamar blue viability of cells transfected with control NT or the indicated siRNA for 48 hours and subsequently treated with vehicle or LCL161 (500 nM) for 24 hours. (B) Verification of siRNA-mediated efficacy by Western blotting for the experiment depicted in (A). (C) Cells were transfected with NT or SP3 siRNA for 48 hours and subsequently treated with vehicle or LCL161 (1 μM) for 6 hours. Cells were harvested for electrophoretic mobility shift assay (EMSA) using a consensus NF-κB probe. Blots are representative of three independent experiments. (D) Cells were transfected with NT or SP3 siRNA for 48 hours and treated with vehicle or LCL161 (500 nM). Cells were processed at the indicated times for an ELISA to test the ability of the indicated proteins to bind to a consensus NF-κB probe. (E) Cells were transfected with NT or SP3 siRNA for 48 hours and treated with vehicle or LCL161 (500 nM) for 6 hours and processed for ChIP using RNA Pol II and primers spanning the transcriptional start site (TSS) of the TNF-α promoter. (F) Cells were treated as in (E) and processed for ChIP using the indicated NF-κB antibodies. The three NF-κB regions on the TNF-α promoter are denoted by κB1 (−873), κB2a/b (−627 and −598), and κB3 (−98). Data are means ± SD from n = 3 experiments; *P < 0.05, **P < 0.01, and ***P < 0.001 by two-way ANOVA using Dunnett’s (A) or Tukey’s (D, E, and F) multiple comparison test.

To further explore the function of SP3 in activation of the TNF-α promoter, we measured the effects of IAP antagonism on SP1/SP3, NF-κB, and IRF1 binding activity by EMSA. Consistent with the lack of enhanced SP1/SP3 binding activity in SMC-treated MDA-MB-231 cells (fig. S6), we did not observe a difference in SP1 or SP3 binding ability to a consensus SP1/SP3 probe upon SMC treatment (fig. S9A). However, congruent with our finding that SP3 predominates over SP1 in SMC-mediated expression of TNF-α, the down-regulation SP3 did not blunt the ability of SP1 to bind to an SP1/SP3 consensus probe in MDA-MB-231 cells (fig. S9B). Similarly, SMC treatment or down-regulation of SP3 did not perturb the ability of nuclear lysates derived from vehicle or SMC-treated cancer cells to bind to the consensus sequence of IRF1 (fig. S9, C and D). These results indicate that IRF1 activity is not modulated by SMC treatment or SP3 activity. On the other hand, SMC treatment enhances the ability of nuclear proteins to bind to an NF-κB consensus sequence, and this binding is attenuated with the removal of SP3 (Fig. 6C). Because these findings indicate that the presence of SP3 is required for full activation of the NF-κB pathway(s), we then determined which NF-κB components are affected. Of the four NF-κB factors tested, we observed substantial impairment of RelA and RelB to bind to a consensus NF-κB probe upon SP3 down-regulation (Fig. 6D).

Our results thus far ascribe SP3 as a primary driver for the regulation of cytokine expression in SMC-treated cells. To elucidate additional mechanisms as to how SP3 initiates the transcription of TNF-α, we conducted ChIP-qPCR analysis in SMC-treated and SP3-deficient cancer cells with an RNA Pol II–specific antibody. We observed that SMC treatment enhanced binding of RNA Pol II to the TSS of TNF-α, and this enhancement was inhibited by the down-regulation of SP3 (Fig. 6E). These results suggest that SP3 stimulates transcriptional initiation of the pro-inflammatory cytokine TNF-α through the recruitment of Pol II to the TSS. We next asked whether SP3 facilitates the recruitment of NF-κB constituents to the three NF-κB response elements (NREs) within the TNF-α promoter. Among the three NF-κB regions in the TNF-α promoter (κB1, κB2a/b, and κB3), we observed that SMC treatment most strongly increased the association of all the NF-κB proteins at the middle κB2a/b region (Fig. 6F). The down-regulation of SP3 abolished this SMC-mediated recruitment to middle and proximal κB2a/b and κB3 regions. On the other hand, SMC treatment did not markedly increase the association of NF-κB proteins to the distal κB1 and proximal κB3 regions. Collectively, the results indicate that basal TNF-α expression is positively mediated by SP3 at the κB3 site and that SMC treatment up-regulates TNF-α expression at the κB2a/b NRE region, which is also promoted by SP3.

SP3 expression correlates with cancer versus normal cell sensitivity to SMCs

The identification of mechanisms that confer the ability of SMCs to selectively induce the death of cancer cells while sparing the majority of non-cancer cells is ideal for clinical application. The combination of SMC and TNF-α did not lead to death of normal established and primary human and mouse cells (Fig. 7A). SMC treatment of these normal cells does not lead to up-regulation of TNF-α expression (Fig. 7B). However, these cells express TNF-α in response to the exogenous application of TNF-α, indicating that the cells are capable of inducing the expression of TNF-α. We noted that cancer cells sustained SMC-induced TNF-α expression over time (Fig. 4B), whereas the expression of SMC-induced TNF-α returns to basal levels within 9 hours in normal cells. These results indicate that a potential reason for the inability of SMCs to induce death of normal cells is through the induction of feedback mechanisms that down-regulate TNF-α expression. To examine whether SP3 may have a role in determining the differential response between cancer and normal cells, we looked at the expression profiles of SP3 and its associated SUMO E3 ligase, PIAS1. We observed consistent reduction of SP3 expression in non-cancer cells (Fig. 7C). This down-regulation of SP3 was also consistent with the decreased presence of PIAS1. We also observed a consistent differential profile of SP1 expression between cancer and non-cancer cells (Fig. 7C). With these findings, we hypothesized that the overexpression of SP3 would result in the death of normal cells in the presence of an SMC. However, we were unable to generate stable lines (fig. S10), indicating that forced overexpression of SP3 is cytotoxic to normal cells.

Fig. 7 SP3 expression correlates with susceptibility of cancer cells to death by SMC treatment.

(A) Cancer cell lines (MDA-MB-231 and SNB75), cultured normal cells (GM38, HEL299, 1059SK, and HMEC), or primary normal cells [human foreskin fibroblast (HFF) or mouse embryonic fibroblasts (MEFs)] were treated with vehicle or LCL161 (5 μM) and BSA (0.01%) or TNF-α (1 ng/ml). Cell viability was assessed 48 hours after treatment by Alamar blue assay. Data are means ± SD from n = 3 experiments; ***P < 0.001 by two-way ANOVA using Dunnett’s multiple comparison test. (B) Abundance of mRNA-encoding TNF-α, assessed by RT-qPCR, in cells treated as described in (A) for the indicated time (h, hours). Data are means ± SD from n = 3 experiments; ***P < 0.001 by two-way ANOVA using Dunnett’s multiple comparison test. (C) Cancer cell lines (MDA-MB-231 and EMT6) and normal cells (GM38, HEL299, HFF, and MEF) were treated with vehicle or LCL161 (1 μM) for 8 hours and processed for Western blotting using the indicated antibodies. Blots are representative of two independent experiments. (D) SMC-resistant cancer cell lines were treated and processed for Western blotting as in (C). Blots are representative of two independent experiments. (E) Abundance of mRNA encoding SP3 in cancer and patient-matched normal tissues. ACC, adrenocortical carcinoma; BLCA, bladder urothelial carcinoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; CHOL, cholangio carcinoma; COAD, colon adenocarcinoma; DLBC, lymphoid neoplasm diffuse large B cell lymphoma; ESCA, esophageal carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; KICH, kidney chromophobe; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LAML, acute myeloid leukemia; LGG, brain lower grade glioma; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; OV, ovarian serous cystadenocarcinoma; PAAD, pancreatic adenocarcinoma; PCPG, pheochromocytoma and paraganglioma; PRAD, prostate adenocarcinoma; READ, rectum adenocarcinoma; SARC, sarcoma; SKCM, skin cutaneous melanoma; STAD, stomach adenocarcinoma; TGCT, testicular germ cell tumors; THCA, thyroid carcinoma; THYM, thymoma; UCEC, uterine corpus endometrial carcinoma; UCS, uterine carcinosarcoma. *Log2FC = 0.5, q < 0.01; **Log2FC = 1.0, q < 0.01; and ***Log2FC = 1.5, q < 0.01. (F) The median RNA sequencing (RNA-seq)–based expression of the indicated genes in cancers was analyzed by hierarchical clustering. Shown are the normalized heat maps. (G) Gene expression variances between brain tissue, LGG, and GBM for the genes in (F) displayed as a t-distributed stochastic neighbor embedding (t-SNE) of RNA-seq transcript counts.

Because the presence of full-length isoforms of SP3 results in the increased transcription of target genes (37), our results suggest that the expression of SP3 drives TNF-α expression, which then leads to death of SMC-sensitive cancer cells. It has been previously reported that ~50% of cell lines can be killed by the combination of SMC and TNF-α or TRAIL, and most of the resistant cell lines can be sensitized to cell death by the down-regulation of the caspase-8 inhibitor, cFLIP (15). We observed that the expression level of SP1, SP3, and PIAS1 was similar in the SMC-resistant cell lines when compared to MDA-MB-231 cells (Fig. 7D). Consistent with previous reports, we observed that the death of these SMC-resistant cancer cells is only observed by the down-regulation of cFLIP and not by the overexpression of SP3 (figs. S11 and S12).

The results from the in vitro studies thus far indicate that the relative levels of SP3 can predict the propensity of cells to induce TNF-α expression. Using the RNA-seq datasets from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) databases, we profiled for the expression of SP3 in cancer and non-cancer tissues. As expected, SP3 is ubiquitously expressed in all cancer types and in normal cells; however, the expression level of SP3 is significantly higher in 32% (10/31) of profiled cancer types as compared to their non-cancer tissues (Fig. 7E). We then next assessed whether the SP3 expression levels correlate with TNF-α mRNA levels. In this analysis, five of nine high–TNF-α–expressing cancers were clustered with elevated levels of SP3 expression, in addition to PIAS1, SP1, and NF-κB (Fig. 7F). We noted that there was a strong association of gene expression of the profiled genes in gliomas (LGG and GBM). All of our profiled glioma cell lines in this study can be killed by SMCs (M059J, M059K, KNS60, and SNB75) and IAP antagonism eradicates primary and established glioma cell lines in vitro and in vivo (3842). We then evaluated whether there was a relationship between the gene expression profiles in gliomas by t-SNE. We observed that the gliomas clustered separately from the normal brain tissues (Fig. 7G). Thus, increased SP3 and TNF-α expression in cancer cells may be a biomarker for the stratification of patients that might respond to SMC therapy.

DISCUSSION

We used an unbiased strategy to identify genes involved in SMC-induced cancer cell death, and we identified a unique requirement for the transcription factor SP3 in mediating SMC efficacy. We found that SP3 activity is globally required for SMC-mediated death of cancer cells and that this induction is not dependent on its closely related family member, SP1. Mechanistically, SP3 positively regulates the SMC-mediated induction of pro-inflammatory cytokine expression through the stimulation of NF-κB–mediated transcriptional activation of TNF-α. Moreover, along with SP1 and NF-κB, the expression of SP3 and its E3 sumo ligase, PIAS1, is positively correlated with the sustained transcriptional expression of TNF-α, which may be a defining characteristic of the sensitivity of cancer cells, and not normal cells, to SMC treatment.

SP1 and SP3 are ubiquitously expressed in mammalian cells and control gene expression involved in diverse cellular processes, including cell growth, death, angiogenesis, and invasion (21, 43, 44). The functions of SP1 and SP3 are dependent on the promoter context and posttranslational modification; SP1 predominantly activates transcription, whereas SP3 activates or represses transcription depending on its posttranslational modification (21). Our report reveals that SP3 is an indispensable transcriptional activator that is required for the full execution of SMC-mediated cell death in cancer cells. Previous studies have implicated the NF-κB and IRF1 pathways in the expression of pro-inflammatory cytokines and the execution of cell death by SMC treatment (36, 45). Consistent with these earlier reports, we observed that NF-κB activity is enhanced with SMC treatment, which is primarily due to the activation of the alternative NF-κB pathway and subsequent transcription of its target genes, including TNF-α. However, despite several reports implicating SP1 as a component involved in TNF-α expression or SMC efficacy (4651) and showing that SP1 and SP3 commonly colocalize to the same promoter (35), we did not observe a functional role for SP1 for the induction of TNF-α expression. Our results demonstrate a unique role for SP3 in positively regulating SMC-induced and NF-κB–mediated transcription of TNF-α. In this context, SMC induces TNF-α expression through the recruitment of SP3 to the TNF-α promoter. The close proximity of the SP3 and NF-κB sites on the TNF-α promoter strongly suggests that SP3 and NF-κB elements cooperatively bind to neighboring sites in the genome. SP3 could act as a molecular chaperone within the nuclear microenvironment, recruiting NF-κB dimers to the TNF-α locus. The interaction of SP3 with the TNF-α promoter is increased when exposed to inflammatory signals that lead to activation of the NF-κB pathways, resulting in accelerated nuclear transport of NF-κB constituents, such as RelA or RelB. Therefore, SP3 may orchestrate a multiprotein complex containing NF-κB dimers, leading to SP3-mediated transcription of TNF-α. NF-κB most commonly acts as a transcriptional activator; the mechanisms by which NF-κB suppresses gene transcription are less well characterized. The activation of the NF-κB pathways leads to the production of TNF-α whereby activation of the classical pathway typically involves the NF-κB dimers p50 and RelA, whereas activation of the nonclassical or alternative pathway is through p52 and RelB. In this context, NF-κB engagement maximally drives TNF-α production only in the presence of SP3, which can be considered as a ligand-dependent factor that recruits additional factors or converts repressors into activators. For instance, activated RelA/RelB can increase SP3 localization at the TNF-α promoter. Supporting this finding, the down-regulation of SP3 impaired the DNA binding ability of RelA and RelB. Overall, these results imply that there is cooperation between SP3, NF-κB, and IRF1 to specifically induce gene expression of TNF-α.

Similar to a previous report that showed an increase in levels of SP3 in tumor tissues compared to non-tumor tissues (52), we found a higher abundance of SP3 in several cancer cells when compared to non-cancer cells. Currently, it is unknown whether SMC treatment leads to posttranslational modification, such as sumoylation, and thus enhanced function of SP3. Sumoylated SP3 has been reported to be transcriptionally active. The activating effects of SP3 on the TNF-α promoter could involve posttranslational modification. Sumoylation has been reported to switch SP3 from a transcriptional activator to a repressor (53). However, we found that the protein abundance of PIAS1 (an E3 ligase responsible for sumoylating SP3) and the mRNA abundance of SP1, PIAS1, and NF-κB genes and TNF-α correlated with the presence of SP3 in cancer cells. These results imply that sumoylation may be required for the transactivity of SP3. Given these observations, we propose that the expression of SP3 might serve as a functional biomarker for SMC responsiveness.

Our findings that SMC-induced activation of the NF-κB pathway recruits SP3 to the TNF-α promoter, where it functions as a transcriptional activator, may provide unique insight into the role of SP3 in inflammation, immunity, and the pathogenesis of various diseases, including cancer. Our results reveal a mechanism linking inflammation-mediated NF-κB activation through IAP antagonism and transcriptional induction of TNF-α production by the transcription factor SP3. In this context, SP3 is required for the SMC-mediated production of pro-inflammatory cytokines, including TNF-α, from other means, such as from type I IFN, IL-1β, and Toll-like receptor agonists. Given the essential role of the NF-κB pathway in immune function, our results indicate that the targeting of SP3 may be a therapeutic avenue for the development of immunotherapies that rely on TNF-α signaling. In addition, the use of biomarkers, such as SP3, for TNF-α abundance and activity could be used to identify patients with cancer who might best respond to SMCs and TNF-α–based immunotherapies.

MATERIALS AND METHODS

Reagents

Novartis provided LCL161 (18). SM-164 was provided by S. Wang (University of Michigan, USA) (54). Birinapant was obtained from Tetralogic Pharmaceuticals (55). OICR720 was synthesized by the Ontario Institute for Cancer Research (OICR) (56). AT-406, GDC-0917, and AZD-5582 were purchased from Active Biochem. TNF-α was purchased from Enzo. IFN-β was obtained from PBL Assay Science. All siRNAs were obtained from Dharmacon (ON-TARGETplus SMARTpool) or Ambion (Silencer Select for IFR1). High–molecular weight poly(I:C) was obtained from InvivoGen. VP16, staurosporine, and cycloheximide were obtained from Sigma.

Cell culture

Cells were maintained at 37°C and 5% CO2 in Dulbecco’s modified Eagle’s medium supplemented with 10% heat-inactivated fetal calf serum and 1% nonessential amino acids (Invitrogen). All of the cell lines were obtained from the American Type Culture Collection, with the following exceptions: SNB75 and OVCAR4 (National Cancer Institute, National Institutes of Health) and KNS60 and KYM-1 (Japanese Collection of Research Bioresources, Japan). Cell lines were regularly tested for mycoplasma contamination. Primary MEFs were derived from C57BL/6 mice. For siRNA transfections, cells were reverse-transfected for 48 hours with 10 nmol of siRNAs with Lipofectamine RNAiMAX (Invitrogen) or DharmaFECT I (Dharmacon) as per the manufacturer’s protocol.

Live cell imaging

Measurement of caspase cleavage or membrane permeability in treated cells was performed by incubating cells with 5 μM caspase-3/7 Apoptosis Assay Reagent (Essen BioScience) or 1 μM YOYO-1 (Life Technologies) and treated with vehicle or 100 nM SM-164. Cells were imaged with the Incucyte Zoom microscope (Essen BioScience) for 48 hours, and enumeration of fluorescence signals was processed using the integrated object counting algorithm within the IncuCyte Zoom software.

Genome-wide siRNA screen

To determine the suitability of using RIP1 as a positive control, NT siRNA and siRNA targeting RIP1 (Dharmacon ON-TARGETplus SMARTpool) were printed to 384 wells at 10 nM in a checkered pattern. MDA-MB-231 cells were reverse-transfected using DharmaFECT I (Dharmacon) for 48 hours, followed by treatment with 100 nM SM-164 for 48 hours. Viability was determined by Alamar blue [Resazurin sodium salt (Sigma)] (8). The Z factor was calculated as previously described (13, 57).

For the genome-wide siRNA screen, pools of siRNA SMARTpool duplexes targeting 18,255 human genes (Dharmacon) were spotted into 384-well plates at 10 nM. MDA-MB-231 cells were reverse-transfected using DharmaFECT I (Invitrogen) for 48 hours and subsequently treated with 100 nM SM-164 for 48 hours. Viability was assessed by Alamar blue. Z scores were calculated as [x − median (sample)]/[median absolute deviation × 1.4826].

Stable cell lines

Lentiviruses that express NT or SP3 siRNA were generated by transfecting 293T cells with polyethylenimine and pCMVΔR 8.74 (Addgene), pMD2-G (Addgene), and pGIPZ (SP3 shRNA clones V3LHS_360595, V3LHS_360596, and V3LHS_360598 from Dharmacon). Cells were infected with lentiviruses at a multiplicity of infection (MOI) of 0.1, and stable cells were selected with puromycin (5 μg/ml). Lentiviruses expressing SP3 were generated by transfecting 293T cells with polyethylenimine and pLenti–CMV–Puro–green fluorescent protein (GFP) (Addgene) or pLenti-CMV-Puro-SP3-FLAG, and PLP-VSVG, PLP-1, and PLP-2. Normal cells were infected with lentiviruses at an MOI of 1 and selected with puromycin (2 μg/ml).

In vitro viability assay

Cells were seeded in 96-well plates and incubated overnight. Cells were treated with vehicle (0.01% dimethyl sulfoxide) or the indicated concentration of SMC and, in some cases, were combined with 0.05% BSA, TNF-α (0.1 ng/ml), IFN-β (250 U/ml), IL-1β (1 ng/ml), or poly(I:C; 10 μg/ml) for 24 to 48 hours. Cells were visualized for signs of cell death by microscopy before the determination of cell viability by Alamar blue (data were normalized to vehicle treatment).

Western blotting

Cells were scraped, collected by centrifugation, and lysed in 1% SDS, 50 mM tris-HCl (pH 8.0), 150 mM NaCl, and a protease inhibitor cocktail (Roche). Equal amounts of soluble protein were separated on polyacrylamide gels followed by transfer to nitrocellulose membranes. Individual proteins were detected by Western blotting using the following antibodies: CASP3 (9662), p105 (3035), RelA (8242), p100 (4882), and RelB (4922) from Cell Signaling; SP1 (59), SP3 (644), and FADD (5559) from Santa Cruz Biotechnology; RIP1 (MAB3585), CASP8 (AF705), and IRF1 (MAB4830) from R&D Systems; cFLIP (ALX-804-961 and ADI-AAP-440) and TNFR1 (ADI-CSA-815F) from Enzo; β-actin (AC-15) from Sigma; β-tubulin (E7) from Developmental Studies Hybridoma Bank; and GFP (5A8E5) from GenScript. Rabbit anti-rat IAP1 and IAP3 polyclonal antibodies were used to detect human and mouse cIAP1/2 and XIAP, respectively (27) (Cyclex Co.). All blots were incubated at a dilution of 1:1000 with the primary antibodies overnight at room temperature. Alexa Fluor 680 (Invitrogen) or IRDye800 (Li-Cor) was used to detect the primary antibodies (room temperature for 1 to 2 hours), and infrared fluorescent signals were detected using the Odyssey Infrared Imaging System (Li-Cor). Full-length blots are listed in fig. S13.

For coimmunoprecipitation, cells from a confluent 15-cm dish were lysed in 50 mM tris-HCl (pH 8.0), 10% glycerol, 1% Triton X-100, 150 mM NaCl, and protease inhibitors (Roche). Endogenous CASP8 complex was immunoprecipitated overnight at 4°C with 4 μg of anti–caspase-8 (Santa Cruz Biotechnology, 6136), and complexes were recovered with protein G Dynabeads (Invitrogen) for 1 hour at 4°C. Protein complexes were resolved on polyacrylamide gels.

Reverse transcriptase quantitative polymerase chain reaction

Total RNA was isolated from cells using the RNeasy Plus Mini Kit (Qiagen). Two step RT-qPCR was performed using Superscript III (Invitrogen, 2 μg total RNA) and SsoAdvanced SYBR Green Supermix (BioRad, 1:50 dilution of the cDNA) on a Mastercycler ep realplex (Eppendorf). Primer sequences are detailed in table S2. Data were normalized to the geometric mean of four different reference genes. The cytokine and chemokine PCR array was obtained from realtimeprimers.com. Data obtained from the cytokine and chemokine PCR array were normalized to the geometric mean of eight reference genes.

Chromatin immunoprecipitation

Cells from confluent 15-cm plates were cross-linked with 1% formaldehyde for 10 min at room temperature. Cross-linking was stopped by the addition of 125 mM glycine for 5 min, and cells were washed with and scraped in phosphate-buffered saline (PBS). Cells were Dounce-homogenized in swelling buffer [0.1 M tris-HCl (pH 7.6), 10 mM KOAc, 15 mM MgOAC, 1% NP-40, and protease inhibitors (Roche)], and nuclei were lysed with 50 mM tris-HCl (pH 8), 10 mM EDTA, and 1% SDS. Chromatin was fragmented by sonication (15-s pulses and 1-min pauses). Sheared chromatin was precleared with StaphA cells and incubated overnight at 4°C with 1 μg of primary antibodies [immunoglobulin G (IgG) (2027), SP3 (644), and IRF1 (137061)] from Santa Cruz Biotechnology; RNA Pol II (MMS-126R) from Covance; and IgG (DA1E), p50 (D4P4D), p52 (D7A9K), RelA (D14E12), and RelB (D7D7W) from Cell Signaling Technology, in buffer containing 0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM tris-HCl (pH 8), and 167 mM NaCl. Complexes were recovered with StaphA cells; washed with 100 mM tris-HCl (pH 9), 500 mM LiCl, 1% NP-40, and 1% deoxycholic acid; and eluted in 1% SDS and 50 mM NaHCO3. The ChIP and 10% input sample were purified using the QIAquick PCR purification kit (Qiagen), and RT-qPCR was performed with SsoAdvanced SYBR Green Supermix (BioRad) on a Mastercycler ep realplex (Eppendorf) using 4% of the samples. Primer sequences are listed in table S2.

Enzyme-linked immunosorbent assay

Cells were treated as indicated, cell culture media were collected, debris was removed by centrifugation, and cytokines within the supernatant were measured with the TNF-α Quantikine high-sensitivity or TNF-α DuoSet assay kit (R&D Systems). In some cases, the cell culture supernatants were concentrated using Amicon Ultra filtration units.

Flow cytometry

To determine cell death, cells were treated as indicated and stained with 7-AAD and ANXA5 according to the manufacturer’s instructions (eBiosciences). To detect TNF-α, cells were released by scraping in PBS, stained with Zombie Green (1:500, BioLegend), fixed and permeabilized with the Cytofix/Cytoperm kit (BD Biosciences), and stained with TNF-α (MAB11, 1:100) or isotype control (MOPC-21, 1:100) (both are from BioLegend) for 20 min at room temperature. Cells were analyzed on a BD Fortessa (BD Biosciences), and data were analyzed with FlowJo (TreeStar).

Luciferase assay

The promoter region spanning −1050 to +100 of the TNF-α promoter was cloned into pEZX-PG04. Cells were reverse-transfected with Lipofectamine 2000, and the supernatant was harvested to measure Gaussia luciferase and secreted alkaline phosphatase using the Secrete-Pair Dual Luminescence Assay Kit (GeneCopoeia). Luciferase activity was normalized to alkaline phosphatase activity.

Electrophoretic mobility shift assay

Cells were washed in PBS, scraped, and suspended in cytoplasmic extract buffer [10 mM Hepes (pH 7.6), 60 mM KCl, 1 mM EDTA, 0.25% NP-40, and protease inhibitor cocktail]. Nuclei were subsequently extracted with 20 mM tris-HCl (pH 8), 420 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, and 25% glycerol. Nuclear extracts were incubated with poly(dI-dC) for 20 min at room temperature with 32P-labeled probe [NF-κB: 5′-CAGGGCTGGGGATTCCCCATCT-3′ (58), SP1/SP3: 5′-ATTCGATCGGGGCGGGGCGAGC-3′ (59), IRF1: 5′-GGAAGCGAAAATGAAATTGACT-3′ (60)]. In some cases, nuclear extracts were preincubated with 100-fold excess of cold wild-type or mutant competitor probes for 5 min at room temperature (mutant NF-κB: 5′-CAGGGCTGCGGCTTCCCGATCT-3′). Samples were resolved on a 5% polyacrylamide gel and exposed to film. EMSA supershift was performed by incubating nuclear extracts with 1 μg of the following antibodies at room temperature for 15 min: IgG (2027), SP1 (14027), SP3 (644), or IRF1 (497) from Santa Cruz Biotechnology.

Bioinformatic analysis

The GEPIA website was used to analyze the RNA-seq expression data of SP3 from 9736 tumors and 8587 normal samples (61). RNA-seq data from the TCGA and GTEx projects were derived from the UCSC Xena Platform (https://xenabrowser.net). Hierarchical analysis was determined using Morpheus with Pearson correlation (https://software.broadinstitute.org/morpheus). t-SNE analysis was performed using BioVinci (Theta = 02, Perplexity = 50, Learning rate = 200, and Iterations = 5000).

Statistical analysis

Comparison between multiple treatment groups was analyzed using two-way ANOVA followed by post hoc analysis using Dunnett’s, Sidak’s, or Tukey’s multiple comparison test with adjustments for multiple comparison (GraphPad). Estimate of variation was analyzed with GraphPad.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/12/566/eaat9563/DC1

Fig. S1. Evaluation and validation of a rescue genome-wide siRNA screen.

Fig. S2. Top hits from a secondary deconvoluted siRNA screen.

Fig. S3. SP3, and not SP1, is required for SMC efficacy in cancer cells.

Fig. S4. Knocking down SP3 does not prevent cell death induced by cycloheximide, staurosporine, or VP16.

Fig. S5. Expression of cytokines and chemokines that are modulated by SP3 in SMC-treated cancer cells.

Fig. S6. SMC treatment does not affect the DNA binding activity of SP3.

Fig. S7. SMC-induced cancer cell death is prevented by shRNA-mediated knockdown of SP3.

Fig. S8. SP3 deficiency does not induce cancer cell death upon cotreatment with SMCs and TNF-α.

Fig. S9. SP3 is the major transcription factor involved in promoting SMC-mediated cancer cell death.

Fig. S10. SP3 overexpression in normal cells leads to cytotoxicity.

Fig. S11. Overexpression of SP3 does not lead to cell death in SMC-resistant cell lines.

Fig. S12. cFLIP and SP3 are critical mediators for SMC efficacy in resistant cancer cells.

Fig. S13. Full-length Western blots.

Table S1. Data from siRNA screen.

Table S2. Primer sequences.

REFERENCES AND NOTES

Acknowledgments: We thank J. S. Cameron, B. Firestone, and D. Porter of Novartis for providing LCL161; S. Condon of Tetralogic Pharmaceuticals for providing Birinapant; and S. Wang (University of Michigan, USA) for providing SM-164. Funding: This work was supported by grants awarded to R.G.K., S.T.B., and E.C.L. by the Canadian Institutes of Health Research (CIHR), the Ontario Institute for Cancer Research (OICR), the Terry Fox Research Institute (TFRI) Selective Therapies Program, the Canadian Cancer Society Research Institute (CCSRI), and the Brain Canada Foundation (grant no. 704119). R.G.K. is a Fellow of the Royal Society of Canada and a Distinguished Professor of the University of Ottawa. Author contributions: S.T.B., H.H.C., E.C.L., and R.G.K. designed the research; S.T.B., H.H.C., T.S., C.E.B., M.S.-J., H.M., and S.D.B. performed the research; S.T.B., E.C.L., and R.G.K. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Stay Connected to Science Signaling

Navigate This Article