Evasion of cell death is one crucial capability acquired by tumour cells to ward-off anti-tumour therapies and represents a fundamental challenge to sustaining clinical efficacy for currently available agents. Inhibitor of apoptosis (IAP) proteins use their ubiquitin E3 ligase activity to promote cancer cell survival by mediating proliferative signalling and blocking cell death in response to diverse stimuli. Using immunoaffinity enrichment and MS, ubiquitination sites on thousands of proteins were profiled upon initiation of cell death by IAP antagonists in IAP antagonist-sensitive and -resistant breast cancer cell lines. Our analyses identified hundreds of proteins with elevated levels of ubiquitin-remnant [K-GG (Lys-Gly-Gly)] peptides upon activation of cell death by the IAP antagonist BV6. The majority of these were observed in BV6-sensitive, but not-resistant, cells. Among these were known pro-apoptotic regulators, including CYC (cytochrome c), RIP1 (receptor-interacting protein 1) and a selection of proteins known to reside in the mitochondria or regulate NF-κB (nuclear factor κB) signalling. Analysis of early time-points revealed that IAP antagonist treatment stimulated rapid ubiquitination of NF-κB signalling proteins, including TRAF2 [TNF (tumour necrosis factor) receptor-associated factor 2], HOIL-1 (haem-oxidized iron-regulatory protein 2 ubiquitin ligase-1), NEMO (NF-κB essential modifier), as well as c-IAP1 (cellular IAP1) auto-ubiquitination. Knockdown of several NF-κB pathway members reduced BV6-induced cell death and TNF production in sensitive cell lines. Importantly, RIP1 was found to be constitutively ubiquitinated in sensitive breast-cancer cell lines at higher basal level than in resistant cell lines. Together, these data show the diverse and temporally defined roles of protein ubiquitination following IAP-antagonist treatment and provide critical insights into predictive diagnostics that may enhance clinical efficacy.
Inhibitor of apoptosis (IAP) proteins play critical roles in cellular survival by blocking cell death, modulating signal transduction and affecting cellular proliferation . Through their interactions with inducers and effectors of cell death, IAP proteins suppress apoptosis triggered by diverse stimuli including death-receptor signalling, irradiation, chemotherapeutic agents or growth-factor withdrawal . Evasion of apoptosis, in part due to the action of IAP proteins, enhances resistance of cancer cells to treatment with chemotherapeutic agents and contributes to tumour progression . c-IAP1 (cellular IAP1) and c-IAP2 are components of TNFR1 [TNF (tumour necrosis factor) receptor 1] complexes where they modulate apoptotic signalling and caspase-8 activation [4–6]. XIAP (X-chromosome-linked IAP) is an endogenous inhibitor of caspases that uses the linker region between its BIR1 (baculoviral IAP-repeat domain 1) and BIR2, as well as the BIR2 domain, to inhibit caspases 3 and 7 and the BIR3 domain to inhibit caspase-9 [7,8]. XIAP and c-IAP proteins are also RING (really interesting new gene) domain-containing E3 ligases that promote the assembly of polyubiquitin chains on themselves and on several cell-death mediators including caspases and SMAC (second mitochondrial activator of caspases), the endogenous IAP antagonist polypeptide . The E3 ligase activity of IAP proteins allows them to modulate cell death by regulating the stability of cell-death mediators. This E3 ligase activity is particularly important for promoting RIP1 (receptor-interacting protein 1) ubiquitination and activating intracellular cell-death complexes [6,10,11]. Intriguingly, E3 ligase activity of c-IAP proteins is also critical for the activation of NF-κB (nuclear factor κB) pathways: in the canonical pathway, c-IAP-mediated RIP1 ubiquitination is required for activation, whereas in the non-canonical NF-κB pathway c-IAPs promote degradative ubiquitination of NIK (NF-κB-inducing kinase) to suppress signalling .
Regulation of protein stability by the ubiquitin–proteasome system is vital for maintenance and modification of numerous cellular processes . Ubiquitination involves covalent modification of target proteins with the 76-amino-acid ubiquitin protein and requires an E1 ubiquitin-activating enzyme, an E2 ubiquitin-conjugating enzyme and an E3 ubiquitin ligase [12,13]. Since each ubiquitin molecule contains seven lysine residues and a free N-terminus, a variety of ubiquitin–ubiquitin linkages can be formed to yield a structurally diverse array of polyubiquitin signals . This diverse topology provides a means for ubiquitin to transmit complex biological signals in a temporally-controlled or spatially defined manner . In this context, Lys63-linked chains, N-terminally linked linear chains and in some cellular pathways, Lys11-linked chains, provide scaffolding for the recruitment and assembly of signalling complexes [9,16]. In contrast, Lys48-linked chains predominantly target substrates for proteasomal degradation . Recent work further suggests that forked ubiquitin chains may display enhanced degradation targeting and signalling activities [17,18]. The vast combinatorial complexity of the ubiquitin system stems from the activities of tens of different E2 enzymes, which mostly dictate the selectivity of ubiquitin chain assembly, and hundreds of E3 ligases which provide substrate specificity .
The pro-survival activity of IAP proteins can be antagonized by the SMAC/Diablo (direct IAP-binding protein with low pI) protein or by SMAC-mimetic IAP antagonist compounds. Importantly, binding of IAP antagonists to the BIR3 domain of c-IAP1 triggers conformational changes in c-IAP1 protein that greatly enhance its E3 ligase activity . As a result, several proteins known to complex with c-IAPs are directly ubiquitinated by c-IAP1, with ubiquitination of RIP1 leading to the activation of canonical NF-κB signalling [6,21]. However, this outburst of E3 ligase activity also causes the rapid auto-ubiquitination and proteasomal degradation of c-IAPs themselves [6,22,23]. Elimination of c-IAP proteins clears the way for NIK accumulation and activation of the non-canonical NF-κB signalling [22,23]. These events stimulate TNFα production and lead to TNFR1-mediated death of tumour cells lacking the protective effects of c-IAP proteins [6,22–24]. As such, activation of IAP E3 ligase by SMAC mimetic/IAP antagonist compounds is integral to their ability to induce cell death in tumour cells.
IAP proteins have been implicated in human malignancies because of their elevated expression levels, anti-apoptotic activity and the ability to engage survival signalling . Nevertheless, IAP antagonists efficiently inhibit tumour growth and induce tumour-cell-death in only a subset of cell types. Currently, there is no explanation for this selective sensitivity to IAP antagonist treatment. Herein, we used the ability of IAP antagonists to boost IAP E3 ligase activity as a molecular indicator of sensitivity to this anti-cancer treatment. Our analysis of IAP antagonist-sensitive and -resistant breast-cancer cell lines demonstrated vastly different ubiquitination profiles, with notable increases in protein ubiquitination specifically in sensitive cells. This method revealed a startling depth and breadth of the ubiquitination response during the cell-death process. Prominent ubiquitination of NF-κB pathway proteins was only evident in sensitive cells and remarkably the basal level of RIP1 ubiquitination appears to correlate with sensitivity to IAP antagonist treatment. Knockdown of RIP1 and other mediators of NF-κB signalling blocked IAP antagonist-induced cell death, confirming the functional relevance of these findings. Collectively, our data identify ubiquitination of NF-κB proteins and, particularly, RIP1 as indicators of sensitivity to IAP antagonists and pave the way for future development of biomarker(s) for IAP antagonist anti-cancer treatment.
Cells [(1–1.5) × 104 per well] were seeded into 96-well dishes. After 8–12 h, the media were changed and cells were treated as indicated in the Figure legends. Cell viability was measured by Neutral Red uptake, as described previously .
Western blot analysis and immunoprecipitation
Western blot analyses were performed, as described previously, with the following lysis buffer: 20 mM Tris/HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100 and protease inhibitor cocktail (Roche) . Detection of endogenous Lys63-linked RIP1 ubiquitination was performed as described previously [16,27].
Cell lines and reagents
MDA-MB231, KPL4, MDA-MB453, MDA-MB175, HCC-1395, CAL51 and MDA-MB361 human breast cancer cell lines were obtained from American Type Culture Collection. EVSA-T and EFM192A human breast carcinoma cells were obtained from DSMZ (German Collection of Micro-organisms and Cell Cultures). All cells were grown and maintained in 50:50 RPMI 1640/DMEM (Dulbecco's modified Eagle's medium) medium supplemented with 10% FBS, penicillin and streptomycin. Human recombinant soluble TNFα, TNFR2–Fc fusion protein and BV6 were produced at Genentech, Inc. MG132 was purchased from American Peptide Company, z-VAD-Fmk (benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone) from Calbiochem, GEM (gemcitabine), 5-FU (5-flourouracil), CARB (carboplatin), MITO (mitomycin), THIO (thiotepa), VINO (vinorelbine), taxol, DAU (daunorubicin) and DOX (doxorubicin) from Sigma. Antibodies against actin, PGAM5 (phosphoglycerate mutase family member 5) (Sigma), IKKα [IκB (inhibitor of NF-κB) kinase α], IKKβ, ubiquitin, ERK1/2 (extracellular-signal-regulated kinase 1/2), TAB1 (TAK1-binding protein 1) (Cell Signaling Technology), RIP1, TRAF2 (TNF receptor-associated factor 2), XIAP, RIP2, NEMO (NF-κB essential modifier), BRUCE (baculoviral IAP repeat-containing ubiquitin-conjugating enzyme), SMAC, CYC (cytochrome c) (BD Biosciences), c-IAP1, TAK1 [transforming growth factor β (TGFβ)-activated kinase 1] (R&D Systems), p100 (Millipore), HOIP (HOIL-1-interacting protein) (Novus), HOIL-1 (haem-oxidized iron-regulatory protein 2 ubiquitin ligase-1) (Santa Cruz Biotechnology) were purchased from the suppliers indicated, and Lys63-linkage-specific anti-ubiquitin antibody was as described previously .
K-GG immunoaffinity enrichment and MS
Immunoaffinity enrichment of K-GG (Lys-Gly-Gly) peptides and LiME (linear mixed effects) analysis were carried out as previously described [29,30]. EFM192A, EVSA-T and KPL4 cells were treated for 24 h with DMSO vehicle GEM (20 mM), IAP antagonist (BV6; 10 nM for EFM192A and EVSA-T cells and 100 nM for KPL4 cells) or the combination of GEM and BV6, each in the presence of pan-caspase inhibitor z-VAD-Fmk (10 μM). For the time course study, EVSA-T cells were treated for 0, 5 or 20 min with BV6 (100 nM) or for 4 h with GEM (20 mM). Immunoaffinity enrichment of K-GG peptides and MS analysis was carried out using PTMscan reagents and protocols (Cell Signaling Technology) essentially as described in [29,30]. Cells were lysed under denaturing conditions (9 M urea and 20 mM Hepes, pH 8.0) and proteins reduced and alkylated using standard methods. From each sample, 60 mg of clarified protein lysate was diluted to a final concentration of 2 M urea and lysates digested overnight at 37°C with trypsin. Digested peptides were de-salted using SepPak C18 cartridges (Waters) then lyophilized for >48 h. Dry peptide samples were resuspended in IAP buffer and incubated with anti-K-GG-coupled resin for 2 h at 4°C. IAP samples were washed twice with IAP buffer and four times with water prior to elution in 0.15% TFA (trifluoroacetic acid) (twice for 10 min each). Enriched peptides were prepared for MS analysis using standard protocol for C18 STAGE tips  with elution performed in 60% acetonitrile/0.1% TFA. Phosphopeptide containing samples were dried completely and subsequently resuspended in 5% acetonitrile/0.1% TFA for MS analysis.
A detailed overview of the MS and data analysis methods used in this article are presented in the Supplementary material.
Combination of IAP antagonists and chemotherapeutic agents activates cell death in breast cancer cells
IAP antagonists can promote cell death in a number of cancer cell types, either as single agents or in combination with other pro-apoptotic stimuli [3,32]. To better understand the factors affecting IAP antagonist sensitivity, we treated a panel of breast cancer cell lines with the IAP antagonist BV6  individually and in combination with standard-of-care chemotherapeutic agents. Our experiments confirmed that single agent BV6 treatment elicits significant cell death in a subset of cell lines and no cell death in others (Supplementary Figure S1A). Using this information, cell lines were classified as either sensitive or resistant to IAP antagonist treatment. As chemotherapeutic agents are used at a wide range of concentrations, dose ranges capable of stimulating low-level apoptosis across a panel of cell lines were established for each agent individually prior to combination studies with BV6 (Supplementary Figure S1B). Pairwise combinations of BV6 with GEM, 5-FU, CARB, DOX, THIO, EPI (epirubicin), VINO, DAU and MITO each caused cell death that exceeded the activity of IAP antagonist or chemotherapeutic agents administered alone (Figure 1; Supplementary Figure S1C). As reported in previous papers, BV6 specifically induced c-IAP1 degradation, whereas treatment with GEM or other chemotherapeutic agents (i.e. DOX and 5-FU) did not (Supplementary Figure S1D).
IAP antagonist BV6 activates cell death in breast cancer cells in combination with GEM
Given that IAP antagonist-induced cell death relies on TNF signalling, we next investigated whether TNF was necessary for pro-apoptotic combination of IAP antagonists and chemotherapeutic agents. In cells treated with the TNFR1–Fc fusion protein, secreted TNF is unable to engage its cognate receptor and elicit its pro-apoptotic effects . In most instances, treatment with TNFR1–Fc showed limited effects on the ability of IAP antagonist to sensitize cells to chemotherapeutic agents (Figure 1 and Supplementary Figure S1C). A notable exception was the combination of BV6 and 5-FU, where treatment with TNFR1–Fc fusion abrogated cell death in multiple cell lines. In the case of BV6/GEM combinations, the dependence of TNF signalling was cell-line-dependent (Figures 1A–1C). For the BV6-sensitive breast cancer cell line EFM192A, dramatic enhancement of cell death was observed when 1 nM BV6 was combined with GEM (20 μM) and was unaltered by TNFR1 co-administration (Figure 1A). The MDA-MB231 cell line showed intermediate sensitivity to the BV6/GEM combination and was fully rescued by TNFR1–Fc (Figure 1B). In the most sensitive cell line, EVSA-T, addition of GEM did not enhance BV6-induced cell death (Figure 1C) and TNFR1–Fc failed to provide any protection against this combination. In contrast with these lines, KPL4 cells appear refractory to BV6-induced cell death at concentrations exceeding 1000 nM (Figure 1D). Thus, the combinations of IAP antagonist and chemotherapeutic agents elicit cell death in selected breast cancer cell lines through TNF-dependent and TNF-independent mechanisms.
IAP antagonist treatment stimulates ubiquitination in breast-cancer cell lines
A critical feature of IAP proteins is their E3 ligase activity, which regulates the formation of cell-death-inducing intracellular-signalling complexes and prevents caspase activation . The ubiquitin ligase activity of IAP proteins is also crucial for the induction of cell death by IAP antagonists . Thus, we postulated that an understanding of the proximal and downstream ubiquitination events triggered by IAP antagonists would elucidate cellular processes that mediate activation of cell death. For this reason, two IAP antagonist-sensitive breast-cancer cell lines (EVSA-T and EFM192A) were treated for 24 h with vehicle, GEM, IAP antagonist (BV6) or the combination of GEM and BV6 (Figure 2A). To prevent overt apoptotic cell-death, cells were co-treated with the pan-caspase inhibitor z-VAD-Fmk. Lysates were prepared, subjected to proteolytic digestion and ubiquitin-remnant peptides enriched using anti-K-GG immunoaffinity purification [33,34]. MS analysis was performed to profile site-specific ubiquitination events under each of these conditions and determine the relative abundance of each K-GG peptide across the treatment groups (Figure 2A). For comparison, the IAP antagonist-resistant cell line (KPL4) was analysed in the same fashion. In total, >13000 unique K-GG peptides were identified via >102000 K-GG PSMs (peptide spectral matches) with false discovery rates of 0.2% and 2.3% at the peptide and protein levels respectively (Figure 2 and Supplementary Dataset S1). To assemble site-specific ubiquitination data at the protein level and identify proteins whose ubiquitination status was most significantly affected by the treatment, LiME modelling was utilized .
Analyses of ubiquitination in breast-cancer cell lines following prolonged treatment with IAP antagonist BV6
Profound differences in profiles of K-GG peptides were observed between IAP antagonist-sensitive and -resistant cells (Figure 2B, Supplementary Figure S2A and Supplementary Datasets S1–S4). In the sensitive EVSA-T and EFM192A lines, a sizeable fraction of proteins displayed increased ubiquitination following treatment with BV6 alone or in combination with GEM (Figure 2B and Supplementary Figure S2A). Changes in K-GG peptide abundance upon BV6 treatment were highly correlated between the two BV6-sensitive cell lines (Figure 2C), suggesting that this effect is driven by the IAP antagonist. Results for the BV6/GEM combination from cell viability assays (Figures 1A and 1C) and K-GG proteomics (Supplementary Datasets S2 and S3) both indicate that EFM192A cells are affected by the addition of GEM on to the BV6 treatment, whereas in EVSA-T cells, IAP antagonist is the primary driver of the effects seen with the combination. In resistant KPL4 cells, little or no correlation was observed in K-GG peptide abundance changes for these same proteins following comparable treatments (Supplementary Figure S2B). For example, the abundance of K-GG peptides from NQO1 [NAD(P)H dehydrogenase (quinone) 1] and PUR2 (purine biosynthesis 2) increased >5-fold (log2 ratio=2.43) in response to BV6 alone or up to 13.5-fold (log2 ratio=3.76) in the BV6/GEM combination for EVSA-T, but not KPL4 cells (Supplementary Figures S3A–S3D). A subset of proteins, such as aldolase A, displayed constitutive ubiquitination in the basal state across sensitive and resistant cell-lines that remained essentially unaffected by the treatments (Supplementary Figures S3E and S3F). Demonstrating that these effects were not limited to IAP antagonist treatment, GEM-specific changes were seen in the abundance of K-GG peptides for the replication factor RFA1 [replication factor A protein 1 (three of three cell lines)], as well as the known GEM target, RIR1 [ribonucleotide reductase 1 (two of three cell lines)] (Supplementary Datasets S2–S4).
Notable among the proteins showing elevated levels of ubiquitinated peptides were known modulators of cell death such as RIP1 [RIPK1 (receptor-interacting protein kinase 1)], PGAM5 and CYC (Figures 2B and 2C). Enhanced CYC ubiquitination was observed across a series of lysine residues (Figure 2D) including Lys87 (Figures 2E and 2F), consistent with the release of this protein from the mitochondrial inner-membrane-space . A series of additional mitochondrial proteins displayed robust increases in ubiquitination 24 h after BV6/GEM combination treatment, including those involved in fatty acid synthesis [ACSL1 (acyl-CoA synthetase 1)], maintenance of cristae architecture [CHCH3 (coiled-coil–helix–coiled-coil–helix domain-containing 3)], mitochondrial DNA replication [SSBP (ssDNA-binding protein)] and oxidative phosphorylation [NDUA8 (NADH-ubiquinone oxidoreductase 19 kDa subunit)] (Supplementary Datasets S2–S4: LiME plots for EVSA-T, EFM192A and KPL4). The modification of these mitochondrial proteins is striking given that the enzymes of the ubiquitin system are not believed to reside within the confines of this organelle , suggesting that compromized integrity of the mitochondrial outer/inner membranes or disruption of the mitochondrial-protein import machinery exposes certain mitochondrial proteins to the ubiquitination machinery within the cytosolic compartment.
In addition to apoptotic regulators and mitochondrial targets, elevated levels of ubiquitin-remnant peptides were observed for several NF-κB pathway proteins in sensitive, but not resistant breast cancer cells (Figure 2B and Supplementary Datasets S1–S4). Included among this group are NEMO, IKKβ, OTUB1 [OTU (ovarian tumour) domain-containing deubiquitinase 1] and TRAF2. Interestingly, some of these proteins were ubiquitinated in untreated cells, suggesting a potentially pre-primed cellular state (Supplementary Datasets S2–S4). Most notable among this group is RIPK1, known to modulate both cell death and NF-κB signalling, for which ubiquitination was observed in the untreated EFM192A and EVSA-T cells and increased upon IAP antagonist treatment (Supplementary Datasets S1–S3). By contrast, ubiquitination of RIP1 was not observed in KPL4 cells (Figure 2, Supplementary Figure S2 and Supplementary Dataset S4). The lack of ubiquitination in KPL4 cells may reflect a low level or lack of expression of signalling proteins whose ubiquitination was detected in sensitive cells. To investigate this possibility, we examined the relative protein expression levels for a number of key signalling proteins (i.e. CYC, PGAM5, TRAF2, RIP1, IKKβ and NEMO) within the same protein samples used for K-GG immunoaffinity enrichment. Western blotting results demonstrated comparable protein expression levels in the examined cell lines and ruled out protein level alterations as an explanation for increases in K-GG peptides observed (Supplementary Figure S4).
Whereas profound changes in ubiquitination were seen 24 h after IAP antagonist treatment, we posited that many of these events reflected downstream consequences of the cell death programme rather than the direct targets of the IAP E3 ligase activity. Although IAP antagonists induce rapid activation of IAP E3 ligase, proteasomal degradation of c-IAP1 and c-IAP2 occurs within minutes of treatment [22,23]. To capture ubiquitination events resulting from the early outburst of IAP E3 ligase activity, sensitive EVSA-T cells were treated with BV6 for 0, 5 or 20 min (Figures 3A–3C). A fourth group was treated with GEM for 4 h as a control (Figure 3D). Following K-GG enrichment, a total of >13000 unique K-GG peptides were identified via >71000 K-GG- PSMs, with false discovery rates of 0.1% and 1.0% at the peptide and protein levels respectively. Similar to long-term treatments, 5 and 20 min treatments with BV6 stimulated increased ubiquitination of several proteins regulating NF-κB activation (Figure 3), with the most prominent hits being the E3 ligase c-IAP1 [BIRC2 (baculoviral IAP repeat-containing 2), Figures 3B, 3C and 3G] and its constitutive binding partner, TRAF2 (Figures 3B, 3C, 3E and 3F, and Supplementary Dataset S5). HOIL-1, a component of LUBAC (linear ubiquitin assembly complex) and its substrate NEMO exhibited elevated ubiquitination levels following brief BV6 administrations as well (Figures 3B and 3C). XIAP, the TRAF6-binding protein TAXB1 and several proteins associated with clathrin-mediated internalization of cell-surface receptors including ITSN2 (intersectin 2), EPS15 (epidermal growth factor receptor substrate 15) and its homologue EP15R also displayed higher ubiquitination levels following short BV6 treatments (Figure 3).
Analyses of ubiquitination in EVSA-T cells after short BV6 treatments
Of note, short BV6 treatments did not affect the ubiquitination profile of mitochondrial regulators of cell death, such as CYC or PGAM5, for which ubiquitinated peptides were observed in the basal state. This suggests that the inducible ubiquitination of these proteins is probably not a direct consequence of IAP E3 ligase activity, but instead is carried out by downstream E3 ligases. Expression levels for these signalling proteins were confirmed by Western blotting samples treated with BV6 and/or GEM (Supplementary Figure S4). A427 lung carcinoma cells treated with MEKi [MEK (MAPK/ERK kinase) inhibitor] G-963 (Argenta Discovery) and Bcl-2 (B-cell lymphoma 2) antagonist ABT-263 were similarly analysed by K-GG immunoaffinity enrichment, revealing inducible ubiquitination of mitochondrial cell-death regulators CYC and PGAM5, but not NF-κB pathway modulators (Supplementary Figure S5 and Supplementary Dataset S6). These data further support our findings from EFM192A and EVSA-T cells that ubiquitination of mitochondrial cell-death regulators is probably a consequence of general stress or anti-proliferative cell-death signalling and argue that enhanced ubiquitination of NF-κB regulators correlates with the treatment and sensitivity to IAP antagonists.
Regulators of NF-κB signalling are critical mediators of IAP antagonists-induced cell death and cytokine production
Given that IAP antagonists such as BV6 can activate NF-κB and MAPK (mitogen-activated protein kinase) signalling pathways, we examined the importance of proteins whose ubiquitination was induced by BV6 treatment in downstream signalling. Knockdown of RIP1 or NEMO in EVSA-T cells abrogated BV6 stimulated canonical NF-κB and JNK (c-Jun N-terminal kinase) activation as shown by the lack of IκBα degradation and JNK phosphorylation (Figure 4A). As a control, we examined whether RIP1 or NEMO knockdowns could affect non-canonical NF-κB signalling, a pathway they do not engage. As expected, down-regulation of RIP1 or NEMO did not affect BV6-induced processing of p100 to p52 or subsequent activation of non-canonical NF-κB signalling (Figure 4A). We also investigated BV6-stimulated cytokine and chemokine production and observed that RIP1, NEMO or HOIL-1 knockdowns inhibited BV6 induced TNFα and ccl20 [chemokine (C-C motif) ligand 20] production (Figure 4B and Supplementary Figure S6A).
Regulators of NF-κB signalling are critical for IAP antagonist-induced cell death and cytokine production
Next we tested the role of identified ubiquitinated regulators of NF-κB signalling in BV6-induced cell death. Knockdown of NF-κB proteins p100, p105 or RelA reduced BV6-induced cell death in EFM192A and EVSA-T cells (Supplementary Figure S6B). Similarly, down-regulation of RIP1 almost completely blocked BV6-induced cell death in accordance with the critical role of RIP1 in TNF-stimulated gene expression and cell-death signalling (Figure 4C). These data confirm the functional relevance of NF-κB signalling proteins that were found to be ubiquitinated in IAP antagonist sensitive cell lines for BV6-induced activation of cell death and signalling pathways.
RIP1 ubiquitination status correlates with sensitivity to IAP antagonists
Throughout these studies, RIP1 consistently emerged as a prominent ubiquitinated protein in IAP antagonist sensitive lines, even in the absence of treatment (Figures 2A–2C, 3A–3D and 5A). RIP1 is also functionally essential for BV6-stimulated NF-κB signalling and cell death (Figure 4). Thus, we investigated ubiquitination status of RIP1 in IAP antagonist-sensitive and -resistant breast-cancer cell lines by immunoprecipitating RIP1 with Lys63-linkage-specific anti-ubiquitin antibodies or control isotype antibodies from untreated cells (Figure 5B). RIP1 was found significantly ubiquitinated in IAP antagonist-sensitive MDA-MB231 and EVSA-T cells, but not in the resistant breast cancer lines (Figures 5B–5D). To extend these observations and evaluate whether RIP1 ubiquitination status correlated more broadly with sensitivity, these lysates were compared with lysates from five additional cell lines (Figures 5C and 5D). RIP1 ubiquitination signal was plotted against the viability, revealing a strong correlation between the level of RIP1 ubiquitination and the sensitivity of breast-cancer cell lines to IAP antagonist-induced cell death (Figure 5E). Therefore we propose that RIP1 ubiquitination may serve as a potential biomarker for this promising anti-tumour treatment.
RIP1 ubiquitination status correlates with sensitivity to IAP antagonist treatment
The ability to block cell death induced by a variety of stimuli, combined with their critical role in the regulation of pro-inflammatory MAPK and NF-κB signalling pathways, places IAP proteins in a central position as mediators of cell death and survival . Tumour cells and tissues often rely on the inhibition of cell death to evade anti-tumour therapies, especially those based on the standard-of-care chemotherapeutic agents . Given that IAP proteins are expressed at elevated levels in many human malignancies, it is not surprising that IAP antagonism offers attractive opportunity for the development of anti-tumour therapies . Over a decade of focused efforts on therapeutic targeting of IAP proteins has yielded several small-molecule compounds that are currently investigated in phase 1 and 2 clinical trials . However, in spite of a sound scientific rationale for targeting IAPs and the wealth of data from many groups, it is still not clear why some tumour cells and tissues are susceptible to IAP antagonism, whereas others are not. Unfortunately, the expression of IAP proteins or their binding partners does not seem to differentiate between IAP antagonist-sensitive and -resistant cells or tumours .
IAP proteins that are the primary targets of IAP antagonists in human cancers [c-IAP1, c-IAP2, XIAP, ML-IAP (melanoma IAP)] are also ubiquitin ligases whose E3 activity is often instrumental in their anti-apoptotic activity . As IAP antagonists stimulate the E3 ligase activity of IAP proteins and trigger a variety of signalling events, we investigated the ubiquitination patterns in IAP antagonist-sensitive and -resistant cancer cell lines [20,22,23,40]. IAP antagonist treatment elicited strikingly different ubiquitination responses in sensitive and resistant tumour cells suggesting a strong correlation between the activation of cell death and global ubiquitination profiles. Using immunoaffinity-enrichment methods coupled with MS, we profiled ubiquitination sites on thousands of substrate proteins and identified two major groups of substrates whose ubiquitination status was changed by the IAP antagonist treatment. The first group constitutes mitochondrial proteins, especially mitochondrial cell-death regulators CYC and PGAM5. However, the ubiquitination of mitochondrial proteins is probably not specific to IAP antagonist treatment, as it was not observed in short BV6 treatments. In addition, unrelated anti-proliferative and pro-apoptotic stimuli elicited similar enhancement of mitochondrial-protein ubiquitination. The second group involves regulators of NF-κB signalling whose ubiquitination was preferentially observed in IAP antagonist sensitive cells following long and short treatments with BV6. The importance of NF-κB regulators for IAP antagonist-induced cell death and signalling was functionally confirmed, in turn validating our experimental approach for the identification of IAP antagonist-sensitivity determinants.
Changes in the numbers and abundance of K-GG peptides can be attributed to one of three main mechanistic events: altered E3 ligase activity, de-ubiquitination or changes in protein abundance. Whereas the most rapid changes (i.e. 5 min of BV6 treatment) are probably mediated directly by the E3 ligase activity of IAP proteins, many ubiquitination events occurred at later time-points when c-IAP proteins are degraded. For these events, the downstream proliferative and stress signalling pathways probably engage all three mechanisms, with an aggregate read-out of elevated ubiquitination. Importantly, our study identifies RIP1 ubiquitination status in the basal state as correlating with sensitivity to IAP antagonist treatment. RIP1 is mechanistically implicated in IAP antagonist-induced cell death, signalling and TNF production. These results further demonstrate that RIP1 ubiquitination status can help differentiate between IAP antagonist-sensitive and -resistant cells. We envision that an antibody that efficiently recognizes ubiquitinated RIP1 could serve as a prognostic biomarker for IAP antagonist treatment and enhance our chances of delivering this new therapeutic option to cancer patients who need it the most.
Eugene Varfolomeev designed, performed and analysed cellular viability and signalling studies. Anita Izrael-Tomasevic performed and analysed MS experiments. Kebing Yu, Alexandre Masselot and Corey Bakalarski analysed MS data. Daisy Bustos performed MS experiments. Tatiana Goncharov performed cellular experiments. Lisa Belmont initiated and designed MS experiments. Donald Kirkpatrick and Domagoj Vucic initiated the project, designed the experiments, analysed the data and wrote the paper with the input from the other co-authors.
We thank Kerry Zobel, Kurt Deshayes, Wayne Fairbrother, Cristina de Almagro, members of the Early Discovery Biochemistry and Protein Chemistry departments and the cell depository at Genentech who provided help with insightful discussions, suggestions and reagents. We acknowledge Nguyen Tan for assistance with sample preparation. PTMscan® is performed at Genentech under limited licence from Cell Signaling Technology.
baculoviral IAP repeat-containing 2
haem-oxidized iron-regulatory protein 2 ubiquitin ligase-1
inhibitor of apoptosis
inhibitor of NF-κB
c-Jun N-terminal kinase
linear mixed effects
mitogen-activated protein kinase
NF-κB essential modifier
nuclear factor κB
phosphoglycerate mutase family member 5
peptide spectral match
receptor-interacting protein 1
receptor-interacting protein kinase 1
second mitochondrial activator of caspases
tumour necrosis factor
tumour necrosis factor receptor
TNF receptor-associated factor 2
X-chromosome-linked IAP, z-VAD-Fmk, benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone