Caspase-9b interacts directly with cIAP1 to drive agonist-independent activation of NF-κB and lung tumorigenesis

Abstract

Alternate RNA processing of caspase-9 generates the splice variants caspase 9a (C9a) and caspase 9b (C9b). C9b lacks a domain present in C9a, revealing a tumorigenic function that drives the phenotype of non-small cell lung cancer (NSCLC) cells. In this study, we elucidated the mechanistic underpinnings of the malignant character of this splice isoform. In NSCLC cells, C9b expression correlated with activation of the canonical arm of the nuclear factor κB pathway, a major pathway linked to the NSCLC tumorigenesis. Mechanistic investigations revealed that C9b activates this pathway via direct interaction with cellular inhibitor of apoptosis 1 (cIAP1) and subsequent induction of the E3 ligase activity of this IAP family member. The C9b:cIAP1 interaction occurred via the BIR3 domain of cIAP1 and the IAP-binding motif of C9b, but did not require proteolytic cleavage of C9b. This protein:protein interaction was essential for C9b to promote viability and malignant growth of NSCLC cells in vitro and in vivo, broadly translating to diverse NSCLC oncogenotypes. Overall, our findings identified a novel point for therapeutic invention in NSCLC that may be tractable to small molecule inhibitors, as a new point to broadly address this widespread deadly disease.

Introduction

Nuclear factor-κB (NF-κB) refers to a family of transcription factors, the Rel family, which plays essential roles in various biological processes including inflammation, immune response, cell growth control, apoptosis and tumor development(1,2,3). In mammalian cells, five NF-κB proteins have been identified: RelA (p65), RelB, c-Rel, NF-κB1 (p105/p50) and NF-κB2 (p100/p50). These NF-κB proteins form different homo- or heterodimers that can enter the nucleus, bind to κB-enhancer sites and trans-activate numerous responsive genes. NF-κB dimers are normally sequestered in the cytoplasm by a family of inhibitors, comprising IκBα, IκBβ, IκBγ, IκBε, NF-κB1 p105 (precursor) and NF-κB2 p100 (precursor). The NF-κB inhibitors mask the nuclear localization signals (NLS) within NF-κB proteins and capture them in the cytoplasm.

There are two main arms of the NF-κB signaling, the canonical and noncanonical NF-κB pathway. The canonical NF-κB pathway is activated by a broad range of stimuli (e.g. inflammatory cytokines, lipopolysaccharides). The limiting step in the canonical NF-κB pathway is the inducible degradation of IκB proteins, particularly IκBα, allowing for nuclear transport of various NF-κB complexes, predominantly the RelA-containing dimer. Upon canonical NF-κB activation, IκBα is phosphorylated by the IκB kinase (IKK), which results in its poly-ubiquitination and proteasome-mediated degradation. The non-canonical NF-κB pathway is activated via TNF receptor family members that bind to the TNF receptor associated factors, TRAF2 and/or TRAF3, (e.g. LTβR, CD40, RANK). Activation of the non-canonical pathway relies on NIK, IKKα-mediated phosphorylation and the processing of NF-κB2 p100 precursor to p52, which then undergoes nuclear translocation of p52-containing dimers.

The NF-κB pathway is considered a pro-survival/oncogenic pathway. In particular, NF-κB transcription factors have been reported to regulate the expression of more than 200 genes, many of which encode for anti-apoptotic proteins(4), regulatory factors in cell cycle control(5,6), angiogenesis factors, and regulatory factors involved in cell migration and adhesion(7,8). Importantly, the NF-κB pathway has been shown to be constitutively activated in a variety of cancers (e.g. lung cancer)(9,10), and contributes to chemoresistance(9,10,11,12,13). NF-κB activation also drives lung tumor formation, progression, and invasion in mouse models (6,14,15,16,17). In this study, we explore a previously undescribed molecular mechanism for the activation of the NF-κB pathway in cancer cell signaling.

Materials and Methods

Cell culture

A549, H358, H460, H838, H1299, H1792 and H1869 cells (NSCLC cell lines) were obtained from ATCC, and Primary human bronchial epithelial primary cells (HBECs) were obtained from Cell Applications. All cell lines were used within 6 months and verified by the company with characteristic morphology consistent (epithelial origin) and positive result for epithelial cell marker cytokeratin 18, and were applicable by mutational analysis and genotyping. NSCLC cell lines were cultured in 50% DMEM/50% RPMI with 10%FBS, penicillin(100U/ml), streptomycin sulfate(100μg/ml). HBECs were cultured in Bronchial/Tracheal Epithelial Cell Growth medium (Cell Applications). All cells were maintained at less than 80% confluency under standard incubator conditions.

siRNA transfection

Cells (2–4 × 10⁵) were transfected with siRNAs (Tbl.S4 for sequences) using the Dharmafect 1 transfection reagent as described(18,19,20). 48 hrs after siRNA transfection, cells were collected for further analysis, transfected with plasmid or adenovirus for additional 24hrs, or sq injected for tumor formation in SCID mice (using ON-TARGET/siSTABLE dsRNAs).

Next-Generation RNA Sequencing and Expression Analysis

Total RNA was extracted from cells and subjected to Next-Generation Sequencing as described(21). FASTQ files were analyzed using the Galaxy software and Tuxedo protocol(21). Quality control of fragments was determined using the FASTQC tool. Fragments were mapped to a human reference genome using Tophat and merged using Cuffmerge. BAM files were generated using Cufflinks and statistically significant transcripts were identified using Cuffdiff. The resulting files were analyzed using Ingenuity Pathway Analysis (IPA) to identify differentially regulated pathways. Deep RNA sequencing data were deposited to NCBI’s Sequence Read Archive (SRA) database (SRA accession number is PRJNA278375).

Human Tissue Specimens

Frozen human lung tissue was obtained from the Virginia Commonwealth University Tissue and Data Acquisition and Analysis Core under a VCU IRB-approved protocol (#HM12702). Total RNA was extracted from multiple 10-μm frozen sections as described(22). For laser capture microdissection (LCM), tissues were sectioned (8 μm thickness) and placed onto glass slides, immediately frozen and stored at −80°C for <7 days. LCM was performed to isolate tumor and non-tumor epithelium on an Arcturus Microdissection System. Caps with adherent tissue were placed on 0.5 ml tubes containing 30μl RNA extraction buffer and incubated at 42°C,30 minutes. Extraction buffer from multiple (8–10) caps was combined, and RNA was extracted using the PicoPure RNA extraction kit. RNA quality was assessed on the Experion Bioanalyzer (BioRad); all samples were of high quality with RQI>7.4.

Plasmid transfection and selection of stable cell lines

For transient and stable transfections, cells (1.5×10⁵) were transfected with plasmids as described(18). For transient transfections, cells were used in preparation of the protein extracts for Western immunoblotting or cell lysates for immunoprecipitation/co-immunoprecipitation after 24hrs. For stable transfections, cells were passaged into zeocin selection medium (with concentration of 100 μg/μl for H838 cells, 300 μg/μl for A549 cells and 400 μg/μl for H460 and H1299 cells) 24hrs after transfection and selected for 2wks.

Nuclear and cytoplasmic extraction

Cytoplasmic and nuclear extracts were separated and prepared from cells using NE-PER Reagents following manufacture’s protocol.

Quantitative RT-PCR

Quantitative RT-PCR (RT-qPCR) were performed as described (19,20,23,24) utilizing primers for the genes listed in Table S1. The relative mRNA expression was normalized to 18S except for human tissues where relative mRNA expression was normalized to ESD.

DNA binding enzyme-linked immunosorbent assay for activated NF-κB transcription factors

The DNA-binding capacity of NF-κB p52- and p65-containing complexes were evaluated by using TransAM® NFκB p52/p65 ELISA Kits (Active Motif). Nuclear extract (3–5 μg) was added to each well of 96-well plates with oligonucleotides containing NF-κB consensus binding site were immobilized. The specific competitors (SC), wild-type consensus oligonucleotides, and the non-specific competitors (NSC), mutated consensus oligonucleotides, were utilized as controls for binding specificity. The Jurkat or Raji nuclear extract was used as positive controls for p65 and p52 activation.

Immunoprecipitation (IP) and coimmunoprecipitation (coIP)

For co-IP, cells were lysed in IP buffer (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 1mM EDTA pH 8.0,0.5% NP-40,1X phosphatase/protease inhibitor cocktail). For RIP and NIK IP, in order to remove co-binding proteins, cells were lysed in IP buffer plus 1% SDS followed by heating at 92°C for 2 minutes. Lysates were diluted 10-fold with IP buffer and placed on ice. RIP1 or NIK monoclonal antibody (Cell signaling) added (2μl) to each sample (0.5–1mg of lysates) and incubated at room-temperature with gentle agitation (45 minutes). Pre-washed protein A/G magnetic beads (15μl) were added and incubated at room temperature (10 minutes) with gentle agitation. The bead complexes were washed 3X IP buffer, pelleted, resuspended in Laemmli buffer, incubated at room temperature (10 minutes) and the resulting supernatant subjected to SDS-PAGE/immunoblotting for K48 or K63-likage specific ubiquitin, NIK, or RIP1. For NIK IP, cells were treated with proteasome inhibitor MG-132 (10 μM) for 1 hr before collection. For cIAP1 coIP, cIAP1 monoclonal antibody (3μl) was utiilzed. For myc or FLAG coIP, pre-washed anti-FLAG M2 or anti-Myc tag magnetic beads (15ul) were placed in each cell lysate sample for 4 hrs at 4°C with gentle agitation. The bead complexes were washed 3X with IP buffer, pelleted, and incubated with myc or 3X FLAG peptide (Sigma) solution for 45 minutes at 4°C to elute the myc/FLAG fusion proteins. Final concentration of peptides in elution solution was 0.5 μg/ul. The eluates were used directly in the in vitro ubiquitination assay.

Cell survival assay (colony formation in liquid plates)

A549, H460, H1299 (1.5×10²) or H838 (4×10²) cells were assayed for cell survival over 12 days as described(20). Colonies were fixed with methanol and stained with 0.05% crystal violet solution and colonies then counted.

Colony formation assay (soft-agar)

A549 (2×10³) or H838 (5×10³) cells were assayed for anchorage-independent growth over 3 weeks as described(18). Microscopic images from each well of the soft-agar plates were taken for evaluation of colony number and size distribution. Image J was used to count the number of stained colonies with diameter more than 100 μm (or area more than 0.00785 mm²) and calculate the average size of counted colonies.

In vitro ubiquitination assay

FLAG tagged cIAP1 or myc tagged Wt/Mut C9b was purified by IP of FLAG or myc fusion proteins from HEK 293 cells transfected with FLAG tagged cIAP1 plasmids or myc tagged WT/Mut C9b plasmids for 24hrs in normal growth medium and placed in serum-free medium for additional 6hrs. Purified products of FLAG or myc immunoprecipitation from cells transfected with parental empty plasmids were utilized as negative controls. In vitro ubiquitination reactions were performed at 37°C for 1hr. The reaction mixtures (20 μl) included 50 nM E1 (Boston Biochem E-305), 0.5 μM Ubc H5a E2 (E2-616), FLAG tagged cIAP1 as E3 ligase or FLAG negative control, myc-tagged WT/Mut C9b or myc negative control, 100 ng recombinant tagged RIP1 (Abcam), 10 μM ubiquitin (Boston Biochem), 5 mM ATP, 50 mM Tris 7.5, 5mM MgCl₂ and 2mM DTT. After incubation time, the product mixtures were analyzed by SDS-PAGE/Western immunoblotting.

Western immunoblotting

Western immunoblotting was accomplished as previously described (25,26) using the following primary antibodies: anti-caspase-9 (Assay Designs); anti-FLAG and anti-β-actin (Sigma); anti-p65, anti-NF-κB2, anti-IκBα, phospho-IκBα, anti-laminA/C, anti-α-tubulin, anti-NIK, anti-RIP1, anti-cIAP1, anti-cIAP2, anti-Apaf1, anti-myc (Cell Signaling Technology); anti-K48, anti-Ki-67, and K63-linkage specific ubiquitin, and anti-Ki-67 (Abcam). Secondary antibodies were horseradish peroxidase-conjugated anti-rabbit IgG and anti-mouse IgG antibodies (Cell Signaling Technology).

Adenovirus

Adenovirus control, adenovirus expressing IκBα or dominant negative IκBαS (from Vector Biolabs) were used at 20–50 MOI for transfection. Adenovirus control or expressing WT or Mut C9b were generated by using Adeno-X CMV Adenoviral System 3 (Clontech) and following the manufacturer’s protocols (used at 50–200 MOI for transfection).

Animals tumor models

Seven-week-old male C.B-17 SCID (IcrHsd-Prkdc^scid) mice received subcutaneous injections of A549 (1×10⁶), H460 or H1299 (2×10⁶) cells into the right hind flank. Tumors were allowed to develop for 2–4 wks and tumor size measured by caliper. These experiments were conducted under the VCU approved IACUC proposal, AD10000534 (Assurance number: A3281-01), and the McGuire VAMC approved IACUC protocol, #01790.

Results

C9b regulates the NF-κB pathway

C9b, in contrast to 9a, lacks the catalytic domain, while retaining the caspase recruitment/APAF1 association domain (CARD). In opposition to the pro-apoptotic function of C9a, C9b is an apoptosis inhibitor by competing with C9a for binding to APAF1 and consequently suppressing intrinsic apoptosis (27,28). More recently, we reported that C9b is expressed mainly in transformed cells and NSCLC tumors, and importantly, C9b is required for anchorage-independent growth (AIG) and the tumorigenicity of human NSCLC cells(18). These recent findings were the basis of the hypothesis that C9b provides a “gain of function” versus inhibition of C9a activation via competition for APAF1 binding. To interrogate this hypothesis, we utilized deep RNA sequencing (RNAseq) to determine whether C9b regulated specific cell signaling pathways. Both down-regulation of C9b in the NSCLC cell line, A549 cells and ectopic expression of C9b in primary human bronchial epithelial cells (HBEC) demonstrated profound effects on the NF-κB pathway (Fig. 1A,1B;Fig.S1,Tbl.S2). Specifically, downregulation of C9b in A549 cells led to modulation of 77 genes associated with the NF-κB pathway. Ectopic expression of C9b in HBECs, normally devoid of C9b, affected 133 genes associated with the NF-κB pathway in a contrasting manner. Even though A549 cells have a different epithelial lineage than HBECs (i.e. type II pneumocytes), 15 overlapping NF-κB associated genes were identified between the two cell lines (Fig. 1C). Fig. 1D depicts the contrasting regulation of 7 of these genes (Fig. 1D;Tbl.S3), which was further validated by RT-qPCR (Fig. 1E). Overall, a strong link between C9b and the NF-κB pathway was revealed.

Figure 1. Identification of the NF-κB pathway as a plausible target of C9b.

(A–D) A549 cells were transfected with the indicated siRNAs,48hr. HBECs were transfected with adenovirus (Ad) control (Con) or expressing C9b,24hr. RNA was isolated and analyzed by RNAseq. (A) Network computationally generated on the basis of IPA knowledge memory indicated a high relevance to the NF-κB pathway. Red, increased expression; green, reduced expression; higher color intensity indicates higher difference in expression level. Cell localization of NF-κB-related factors is indicated (Fig. S1, larger legend). (B) Total protein was subjected to SDS-PAGE/immunoblotting. (C) The network for A549 cells comprises 77 NF-κB-related genes significantly different between control siRNAs and C9b siRNA samples. The network for HBECs comprises 133 NF-κB-related genes significantly different between AdCon and AdC9b samples. There are 15 overlapping genes; (Table S2). (D) Heatmap showing hierarchical clustering of 7 overlapping NF-κB genes that expression is contrastingly affected by C9b siRNA (A549) versus up-regulation of C9b (HBEC); red=increased expression; green=reduced expression (Table S3). Results were repeated on 2 separate occasions. (E) Total RNA from C9b-downregulated A549 cells was subjected to RT-qPCR for 7 overlapping NF-κB genes in (D). Data in E are shown as mean ± SD; n=6: one-way ANOVA followed by a Tukey’s HSD test (##p<0.01, **p<0.001); C9b-siRNA sample was compared to each control siRNA sample. (F) RT-qPCR for C9b and 5 NF-κB target genes was performed on lung tissues of 22 NSCLC patients. Values=means of triplicates; each dot represents a patient; Pearson correlation coefficient (r value) and corresponding significance level (p value) are indicated (Fig.S2). (G) LCM images of pre-LCM slide, post-LCM slide after isolating tumors and post-LCM cap view of purified tumor epithelium; tumor and normal epithelial cells are indicated in pre-LCM slides. (H) RT-qPCR for C9b and BIRC5 was performed from RNA prepared from LCM purified epithelium in (G); data are shown as means ± SD (n=3). (I) RT-qPCR for C9a, C9b or BIRC5 was performed with RNA purified from the NSCLC cell lines in Table S3.; Values=means of triplicates; each dot represents a NSCLC cell line.

Previously, our laboratory demonstrated that caspase-9 RNA splicing was dysregulated in favor of C9b expression in 78% of NSCLC tumors (N=79)(19), and 22 of the 79 tumors possessed a strong comparative analysis criteria of >70% tumor, <10% necrosis and <10% of infiltrating immune cells. These tumors were re-analyzed for C9b by RT-qPCR along with 2 well-documented NF-κB pathway genes, BIRC5 and IER3, and the 7 additional genes identified in Fig. 1C,D. Of these 9 genes, 5 correlated with C9b expression when normalized to ESD expression (Fig. 1F,Fig.S2), a validated normalizing gene for NSCLC (Fig.S2). Importantly, none of these genes correlated with C9a expression (Fig.S2). To further confirm the association of C9b expression with the NF-κB pathway, Laser Capture Microdissection (LCM) was performed to isolate tumor and non-tumor epithelial cells from lung tissues of 2 representative and randomly selected patient tumors (one tumor with high C9b expression and one tumor with low C9b expression). Importantly, the NSCLC tumor with high expression of C9b demonstrated increased expression of BIRC5 as compared to normal lung epithelial cells (Fig. 1G,H). Lastly, BIRC5 expression also strongly correlated with expression of C9b, but not with C9a, in a panel of multiple NSCLC cell lines (Fig. 1I,Tbl.S4). Hence, the link between the expression of C9b and NF-κB target genes translates to human NSCLC tumors and cell lines.

C9b, but not C9a, activates the canonical and inhibits the non-canonical NF-κB pathways

Due to our deep RNAseq findings, we examined whether modulation of C9b affected the intracellular signaling of the NF-κB pathway. As shown in Fig. 2A, downregulation of C9b by specific siRNA (Tbl.S5) resulted in a significant reduction of p65 in the nucleus; in contrast, increased levels of p52 were observed in the nucleus. The contrasting effect on p65 and p52 nuclear content was observed in response to ectopic expression of C9b (Fig. 2B), which translated to the expression of genes associated with the canonical (BIRC5,IER3, BMI1) and non-canonical NF-κB pathways (CXCR4,IRF3,CCL21) (Fig. 2C,D,Tbl.S5). The effect of C9b modulation on DNA trans-factor content in the nucleus and NF-κB associated gene expression also translated to the binding of p65 and p52 to specific DNA cis-elements (Fig. 2E,F).

Figure 2. C9b activates the canonical and inhibits the noncanonical arm of the NF-κB pathway.

(A,B) Cytosolic/nuclear protein extracts from A549 cells transfected with Con or C9b siRNA,48hr (A) and from A549 Con cells or stably expressing C9b (B) were subjected to SDS-PAGE/immunoblotting. Graphs depict the quantitative changes in p65 and p52 localization as means ± SD (n=6) and expressed as %control (either against random control siRNA (siRan) or Con cells. One-way ANOVA followed by a Tukey’s HSD test (##p<0.01, *p<0.005) for (A) and Student’s t-test (#p<0.05) for (B). (C,D) RNA was isolated from A549 cells, Con or C9b siRNA. mRNA expression of the canonical (C) or noncanonical (D) NF-κB targets was determined by RT-qPCR (Table S1). (E) DNA binding ELISA for NF-κB p65 or p52 was performed using nuclear extracts from A549 cells transfected with Con or C9b siRNA,48hr. Specific or nonspecific competitors utilized. Nuclear extracts from Jurkat or Raji cells were used as positive control for p65 or p52 activation. (F) Nuclear extract input for (E) was subjected to SDS-PAGE/immunoblotting. (G,I) A549 cells transfected with Con, C9a or C9b siRNA (Table S5),48hrs. (H,J) A549 cells were transfected with Con or C9b-expressing plasmids,24hr. Total proteins were subjected to SDS-PAGE/immunoblotting. (K) A549 cells were transfected with Con or C9b siRNA,48hr. RIP1 or NIK was IPed from cell lysates and resolved by SDS-PAGE/immunoblotting. Data in (C–E) are means±SD (n=3–6 in (C) and (D), n=5 in (E) from 2 independent occasions; one-way ANOVA followed by a Tukey’s HSD test (##p<0.01, *p<0.005, **p<0.001). C9b siRNA sample was compared to each Con.

The limiting step in the activation of the canonical NF-κB pathway is the phosphorylation-mediated turn-over of IκB proteins, particularly IκBα, and the activation of the non-canonical pathway relies on stabilization of NIK and processing of NF-κB2 p100 precursor to p52. C9b expression promoted the phosphorylation of IκBα and degradation of both NIK and IκBα while inhibiting the processing of NF-κB2 100 (Fig. 2G–J). To demonstrate the specificity for a “gain of function” role for C9b versus a mechanism of inhibiting C9a activation, C9a was also specifically downregulated using siRNA. The downregulation of C9a did not affect any of these mechanistic functions on the NF-κB pathways attributed to C9b (Fig. 2G). Hence, the function of C9b on modulating NF-κB pathway signaling is independent of C9a corroborating the lack of correlation of C9a expression with NF-κB gene expression shown in Fig. 1.

Based on the above findings, we next examined both RIP1 and NIK ubiquitination. Indeed, RIP1 is a crucial factor in NF-κB signaling, of which K63-linked ubiquitination positively regulates the canonical NF-κB pathway(29,30,31), and K48-linked ubiquitination of NIK is a major regulatory step to degrade the factor and inhibit the non-canonical pathway. Downregulation of C9b induced the loss of ubiquitination for both RIP1 and NIK, specifically K63-linkage for RIP1 and K48-linkage for NIK (Fig. 2K). Importantly, RIP1 was shown to mediate the role of C9b in the canonical NF-κB pathway (Fig.S3). These findings validate the hypothesis of the dual actions and specificity of C9b in NF-κB signaling. Specifically, C9b stimulates RIP1 ubiquitination and enhances phosphorylation-mediated degradation of the IκBα to free p65-containing complexes. Concomitantly, C9b facilitates ubiquitination-mediated degradation of NIK and prevents the processing/nuclear transport of the NF-κB2 (i.e. the non-canonical NF-κB pathway is kept in an “off-state”).

The translatability of this mechanism was examined using a separate NSCLC cell line with a contrasting oncogenotype, H838 cells and also in HBECs. Modulation of C9b expression demonstrated more robust effects on the canonical NF-κB pathway as compared to the A549 cells (Fig.S4,S5). Also of note, C9b but not C9a expression inversely correlated with IκBα level in a panel of NSCLC cell lines(Fig.S6). Interestingly, the modulation of the non-canonical NF-κB pathway was not observed in either the H838 cells or HBECs when C9b was modulated (data not shown). This null effect on the non-canonical pathway in H838 may be due to the undetectable/low basal levels of non-canonical activity and NIK in these cell lines (Fig.S4C). These data demonstrate that C9b activation of the canonical NF-κB pathway translates to diverse epithelial cell lines in contrast to the non-canonical pathway, which correlates with NIK expression.

C9b enhances the survival and AIG capability of NSCLC cells by activation of the canonical NF-κB pathway

Next, we tested the hypothesis that C9b “drives” the AIG and enhanced cell survival of NSCLC cells via the NF-κB pathway by ectopically expressing the NF-κB suppressor, IκBα, or the dominant-negative/non-degradable super-repressor, IκBαS, in NSCLC cells followed by C9b expression. C9b expression significantly affected the cell survival of NSCLC cells (Fig. 3A,B), but repression of canonical NF-κB by ectopic expression of either IκBα or the IκBαS significantly decreased cell survival and AIG(Fig. 3C–E). The effect of IκBα expression, but not IκBαS expression, on the survival/AIG of A549 cells was “rescued” by ectopic expression of C9b (Fig. 3C–E). Importantly, cells stably expressing C9b had a lower level of IκBα but the same level of IκBαS compared to control cells (Fig. 3C,D). These findings show that C9b augments the cell survival and AIG by facilitating degradation of IκBα and subsequent activation of the canonical NF-κB pathway.

Figure 3. Activation of the canonical NF-κB pathway contributes to C9b-mediated enhancement of the survival and AIG of NSCLC cells.

(A–D) Colony formation assays were performed with A549 cells (Con or C9b siRNA,48 hr) (A), with A549 Con cells or stably expressing C9b (B), with A549 cells transfected with ConAd, AdIκBα or AdIκBαS,24hr followed by 24hr transfection with ConAd or AdC9b (C), with A549 Con or C9b-stably-expressing cells transfected with ConAd or AdIκBα or AdIκBαS,24hr (D). Western immunoblotting of total proteins was also performed. (E) Colony formation assays (soft-agar) using A549 cells as in (D). Representative fields are shown in the images below the graphs, (*) scale bar, 2mm. Data are means ± SD (n=6 on 2 independent occasions); one-way ANOVA followed by Tukey’s HSD test (#p<0.05, ##p<0.01, *p<0.005, **p<0.001).

cIAP1 plays critical roles in C9b-mediated NF-κB activation/inhibition

The role of C9b in the canonical NF-κB pathway infers that C9b interacts with regulatory factors that cross-talk between the two arms of this signaling pathway. Cellular inhibitors of apoptosis (cIAPs) are strong possibilities due to their functions in both the canonical and non-canonical NF-κB pathways as well as being shown to directly bind and inhibit processed C9a (3,32,33,34,35,36). Thus, we downregulated either cIAP1 or 2 via specific siRNAs. In contrast to downregulation of cIAP2 (Fig.S7), downregulation of cIAP1 had similar effects as C9b on NF-κB signaling as well as the AIG capacity of NSCLC cells: increased the processing of the NF-κB2 p100 to p52, elevated the level of NIK and IκBα (Fig. 4A), decreased nuclear p65 level whereas elevated nuclear p52 level(Fig. 4B), reduced both RIP1 and NIK ubiquitination (K63-likage for RIP1 and K48-linkage for NIK;Fig. 4C) and decreased the AIG of NSCLC cells(Fig. 4D). Importantly, downregulation of cIAP1 abolished the activating/inhibitory effects of C9b on the NF-κB pathways, particularly on the levels of NIK and IκBα(Fig. 4E,F) and the ubiquitination of RIP1 and NIK (K63-linkage for RIP1 and K48-linkage for NIK;Fig. 4G). cIAP1 depletion also abrogated the enhanced cell survival conferred by C9b(Fig. 4H). cIAP2 could not compensate for the loss of cIAP1 as the elevated cIAP2 level induced by cIAP1 downregulation(Fig.S7) did not ‘rescue’ the effect of cIAP1 downregulation on C9b-mediated reduction of IκBα and NIK level as well as the ubiquitination of RIP1 and NIK. These data strongly suggest that cIAP1 is required for the action of C9b on the NF-κB signaling to enhance the survival of NSCLC cells.

Figure 4. cIAP1 mediates the effects of C9b on the NF-κB pathway.

(A–D) A549 cells were transfected with Con or cIAP1 siRNA,48hr. (A) Total proteins, (B) Cytosolic/nuclear protein extracts, (C) Endogenous IPed RIP1 or NIK from cell lysates were resolved by SDS-PAGE/immunoblotting. (D) Soft-agar assays. (E–H) A549 cells were transfected with Con or cIAP1 siRNA,48hr followed by 24hr transfection with ConAd or AdC9b. (E) Total proteins were subjected to SDS-PAGE/immunoblotting. (F) Graphs of the results from (E). (G) Endogenous IPed RIP1 or NIK from cell lysates were resolved by SDS-PAGE/immunoblotting. (H) Colony formation assays were undertaken. Data in (D,F,H) are shown as means±SD; n=3 in (F) and n=6 in (D,H) from 2 independent occasions; one-way ANOVA followed by a Tukey’s HSD test (#p<0.05,##p<0.01,*p<0.005,**p<0.001).

C9b directly binds to cIAP1

cIAP1 associates with processed C9a via an IAP-binding motif (IBM) (37,38). The truncated C9b retains the IBM, and thus, we hypothesized that C9b binds directly to cIAP1 to modulate the NF-κB signaling. Indeed, this direct interaction was validated by co-immunoprecipitation (coIP) of the endogenous cIAP1 in NSCLC cells(Fig. 5A) or cells overexpressing Myc-tagged C9b(Fig. 5B). Furthermore, mutation on C9b at the IBM(Fig. 5C) ablated the association between cIAP1 and C9b(Fig. 5D), but did not interfere the C9b and APAF1 interaction(Fig. 5E) showing structural integrity.

Figure 5. C9b directly binds to cIAP1.

(A,B) Endogenous cIAP1 or myc-tagged C9b was coIPed from A549 cells (A) or from A549 cells transfected with Con or plasmid expressing myc-tagged C9b (B) and resolved by SDS-PAGE/immunoblotting. IgG=control. (C) Schematic representation of C9a and C9b (conserved IAP-binding motif (IBM), caspase recruitment domain (CARD), linker region (LR) (closed triangle in LR indicates the auto-cleavage site and opened triangle indicates the site for caspase-3-mediated cleavage)). Sequences of WT and Mut IBM are indicated. (D) A549, H460 or H1299 cells were co-transfected with FLAG-tagged cIAP1 and Con plasmid or plasmid expressing myc-tagged WT or Mut C9b,24hr. FLAG-tagged cIAP1 was coIPed from cell lysates and resolved by SDS-PAGE/immunoblotting. (E) A549 cells were co-transfected with FLAG-tagged APAF1 and Con plasmid or plasmid expressing myc-tagged WT or Mut C9b,24hr. FLAG-tagged APAF1 was coIP from cell lysates and resolved by SDS-PAGE/immunoblotting. (F) Schematic of cIAP1 structure. Reported critical residues for caspase-9 binding are depicted (white rectangles)(37). cIAP1 mutations utilized are indicated. (G) A549 cells were co-transfected with myc-tagged C9b and Con plasmid or plasmid expressing FLAG-tagged WT or Mut cIAP1,24 hrs. FLAG-tagged cIAP1 was coIP from cell lysates and resolved by SDS-PAGE/immunoblotting, n≥3.

C9a (47kDa) has been reported to bind to XIAP or cIAP1/2 only following proteolytic cleavage(37,38). Interestingly, proteolytic processing of C9b (37kDa) was unnecessary for cIAP1 binding as mutations of the proteolytic cleavage sites had no effect on C9b/cIAP1 interaction(Fig.S8). Therefore, C9b can interact directly with cIAP1 without the need for proteolytic cleavage in contrast to C9a.

Six amino acid residues in BIR2 and BIR3 domain of cIAP1 have been reported as critical for the caspase-9 interaction(37)(Fig. 5F), but only one mutation, D320A on BIR3 of cIAP1, abolished C9b binding. This mutation did not affect cIAP1 interaction with RIP1(Fig. 5G), which “ruled out” the possibility of impaired the cIAP1 structure. Thus, C9b directly interacts with cIAP1 via the IBM with BIR3.

C9b regulates the NF-κB pathways and promotes the survival, AIG and tumorigenicity of NSCLC cells via direct interaction with cIAP1

To examine whether the C9b/cIAP1 interaction is essential for C9b-mediated NF-κB activation/inhibition, the IBM-mutant (Mut) C9b (ATPF→GGPF) (Fig. 6C) was expressed and the levels of IκBα were evaluated. Expression of MutC9b did not significantly affect IκBα level in stark contrast to wild-type (WT) C9b. Mutation of C9b at the IBM also resulted in the loss of C9b effects on the survival(Fig. 6B), AIG(Fig. 6C), tumor growth of NSCLC cells(Fig. 6D,E,Fig.S9,S10). Importantly, only the re-expression of WT/siRNA-resistant(SR) C9b, but not Mut/SR C9b, “rescued” the reduced survival and AIG of NSCLC cells as a result of C9b depletion regardless of oncogenotype (Fig. 6F,G,Fig.S10). These in vitro observations also translated in vivo as only re-expression WT/SR C9b, and not Mut/SR C9b, could “rescue” the reduction in tumor volume (equivalent to control cells) as a result of C9b downregulation(Fig. 6H). These data demonstrate that C9b controls the survival, AIG and tumorigenicity of the NSCLC cells via direct interaction with cIAP1.

Figure 6. Direct binding to cIAP1 is essential for C9b-mediated NF-κB activation/inhibition.

(A) A549 cells were transfected with control plasmid or plasmid expressing WT or Mut C9b, 24hr. Total proteins were subjected to SDS-PAGE/immunoblotting. (B) Colony formation assays were performed from control A549 cells or cells stably expressing WT or Mut C9b. (C) Soft agar assay were undertaken using cells as in (B). The graphs depicts: colony size distribution and average# of colonies/well. Representative fields and colonies are shown in the images below the graphs, (*) 2 mm scale bar, (**) 400 μm scale bar. (D) A549 cells used in (B,C), or (E) H460 or H1299 cells stably expressing WT or Mut C9b were injected sq into right flank of the SCID mice. After 2–4wks, tumor volume was assessed. (F,G) A549 cells were transfected with Conl or C9b siRNA,48hr followed by 24hr transfection with Con or C9b-siRNA-resistant plasmid expressing WT or Mut C9b. (F) Colony formation assays were performed, and total protein extracts were subjected to SDS-PAGE/immunoblotting. (G) Soft-agar assay as in (C). (H) H460 cells stably overexpressing, siRNA-resistant WT or Mut C9b were treated with either random control siStable siRNA (siRan) or siStable siC9b (siC9b), sq injected into SCID mice, and tumor volume assessed after 2wks. Data are means ± SD; 7–8 mice/group (A549), 5 mice/group (H460 and H1299) in (E); n=5 in (B and H) and n=6 in (C,F,G); one-way ANOVA followed by Tukey’s HSD test (#p<0.05, ##p<0.01, *p<0.005, **p<0.001; for the top graph in (D), WT C9b versus control or MutC9b.

C9b augments the E3 ligase activity of cIAP1 via direct interaction

The E3 ligase activity of cIAP1 has been reported to mediate the ubiquitination of RIP1 and NIK(3,32,36,38). Therefore, we explored whether the interaction of C9b:cIAP1 interaction induces the E3 ligase activity of cIAP1 and the ubiquitination of NIK and RIP1. Co-expression of WT C9b, but not Mut C9b, in conjunction with cIAP1 induced the K63-linked ubiquitination of RIP1 and K48-linked ubiquitination of NIK(Fig. 7A,B). This effect was recapitulated in vitro as the addition of C9b to cIAP1 dramatically increased E3 ligase activity(Fig. 7C). Furthermore, Mut C9b did not possess this ability (Fig. 7A–C). These findings coupled with the entirety of the presented study support the mechanism that C9b modulates NF-κB pathways through direct binding to cIAP1 and activation of the E3 ligase function, which subsequently induces the ubiquitination of RIP1 for the canonical NF-κB activation and the ubiquitination of NIK for the non-canonical inhibition(Fig. 7D).

Figure 7. C9b enhances E3 ligase activity of cIAP1 via direct interaction.

(A,B) A549 cells were co-transfected with the control or plasmid expressing cIAP1 and control or plasmid expressing WT or IBM-mutated (Mut) C9b,24hr. RIP1 or NIK was IP from cell lysates and resolved by SDS-PAGE/immunoblotting. Input was also immunoblotted. (C) In vitro ubiquitination assay was performed using E1 activating enzyme and E2 conjugating enzyme±FLAG tagged cIAP1 (E3 ligase) and myc tagged WT or MutC9b. FLAG tagged cIAP1, myc tagged WT or MutC9b was purified by IP of FLAG/myc proteins. FLAG- or myc-IP products from cells expressing control plasmid were used as controls. In vitro ubiquitination assay were subjected to SDS-PAGE/immunoblotting. Experiments in (A,B,C) were n≥3. (D) Mechanistic model of how C9b regulates the NF-κB signaling and promotes the survival and AIG capacity of NSCLC cells.

Discussion

This study demonstrates that C9b serves additional biological functions outside the previously known mechanistic function of C9b in APAF1 binding and subsequent inhibition of C9a activation. Specifically, this study delineates an agonist-independent mechanism driven by C9b for the activation of the NF-κB pathway. Indeed, we present several key pieces of evidence that C9b, via a “gain of function”, is a regulating factor for the NF-κB pathway. First, the expression of C9b, but not C9a, correlated with NF-κB gene signatures in human NSCLC tumors and with IκBα level in NSCLC cell lines. Second, downregulation of C9a by specific siRNAs did not affect the canonical and non-canonical NF-κB signaling in stark contrast to siRNA specifically directed against C9b. Lastly, we show that proteolytic cleavage of C9b was not required for activation of the NF-κB pathway, which is required for C9a to expose the interactions sites for IAP family members(37,38). Thus, C9b likely possesses an altered configuration to explain the differential binding requirements (i.e. C9b possesses an exposed IBM allowing for immediate interaction with cIAP1 at BIR3 domain).

Also of note, several groups have detailed a role for cIAP1 in an autocrine-feedback mechanism involving TNFα, which is induced by the treatment of cells with Smac mimetics known to bind the BIR3 domain of cIAP1(35,39). Interestingly, 92% of NSCLC cells are resistant to Smac mimetics(40), and the direct binding of C9b to the BIR3 of cIAP1 may explain this observation. Specifically, the tight association of the BIR3 domain to C9b, which is expressed in 78% of NSCLC tumors and cells, may block the ability of these mimetics to antagonize IAP family members. We also “ruled out” a role for the autocrine-feedback mechanism in our studies as modulation of C9b had no effect on TNFα and IL1β transcription (Fig. 1A,Tbl.S2). Furthermore, C9b siRNA did not inhibit TNFα induced reductions in IκBα for canonical NF-κB activation (data not shown), which would have been blocked if the C9b-mediated NF-κB activation utilized the autocrine-feedback mechanism. Thus, the autocrine TNFα feedback loop is possibly suppressed in NSCLC, and our mechanism is related to the findings of Bertrand and co-workers who reported that cIAPs drive oncogenic phenotypes via their E3 ligase function and modulation of RIP(25). Regardless, small molecules specifically targeting the IBM of C9b could be surmised to reactivate the Smac mimetic response (i.e., autocrine TNFα loop) and induce apoptosis in NSCLC broadening the possible therapeutic use of these small molecules.

The targeting NF-κB signaling through interfering C9b/cIAP1 interaction would itself also be a promising direct approach to further compromise NSCLC development and maintenance. For example, the role of the C9b/cIAP1 interaction in the NF-κB signaling broadly translated to diverse oncogenotypes suggesting efficacy would be observed in a large patient population. Furthermore, therapies targeting the C9b/cIAP1 interaction would also be very advantageous to limit toxic side-effects as C9b is not expressed to a large extent in non-transformed cells(19). Additionally, these therapeutics would likely circumvent the serious concerns about using NF-κB inhibitors in cancer therapy (i.e. impairment of innate and adaptive immunity). Although a key role for this interaction in immune responses cannot be completely “ruled out”, the expression of C9b has not been observed to any great extent outside of epithelial malignancies (data not shown).

Another interesting finding was the opposing role of C9b on the non-canonical NF-κB pathway observed in A549 cells. The majority of studies regarding NF-κB signaling in lung tumorigenesis focus on the canonical NF-κB pathway. The contrasting roles of C9b in the canonical versus noncanonical NF-κB pathway coupled to previous reports on the non-canonical pathways inhibiting the canonical pathway(41,42) suggest that the non-canonical pathway may negatively affect the tumor-promoting roles or activation of the canonical pathway. In this regard, we showed that suppression of the canonical NF-κB pathway increased the levels of NIK, indicating elevated non-canonical NF-κB activation (Fig.S11). This effect was abrogated by ectopic expression of C9b, and thus, C9b may play a plausible role in a functional interplay between the two pathways in “fine-tuning” the gene transcription activity in a particular cell for a specific response. However, inhibition of the non-canonical NF-κB pathway by down-regulation of NIK did not have significant impact on both the survival and AIG of NSCLC cells arguing against this mechanism (Fig.S12).

Mechanistically, we have previously reported that C9b expression is induced by activation of the AKT pathway. Hence, expression of C9b in response to activation of the AKT survival pathway is now shown to serve two major survival roles, activation of the NF-κB pathway and inhibition of intrinsic apoptosis. With these functions in mind, ectopic expression of C9b showed strong enhancing effects on the survival and proliferation of non-transformed HBECs (Fig.S10). Based on this study coupled to our previous reports on the biological functions of C9b, one can hypothesize that C9b serves as a key oncogenic factor in NSCLC.

In conclusion, this study demonstrated several intriguing findings in regard to a previously undescribed mechanism for modulation of the NF-κB pathway in human NSCLC. Specifically, we showed that C9b drives the canonical NF-κB activation, which is required for the cancer phenotypes associated with C9b expression. We further demonstrated that modulation of the NF-κB pathway by C9b requires the direct association with cIAP1 but not an external agonist. Hence, these studies now provide a basis as well as the necessary molecular “tools” to launch both structural and therapeutic design studies to possibly provide innovative anti-cancer agents for the treatment of lung cancer.