Differential activation of JNK1 isoforms by TRAIL receptors modulate apoptosis of colon cancer cell lines

Tumour necrosis factor-related apoptosis-inducing ligand (TRAIL) induces apoptosis on binding to its receptors, death receptor 4 and 5 (DR4, DR5). TRAIL can also activate c-Jun N-terminal kinase (JNK) through the adaptor molecules, TNF receptor-associated factor 2 (TRAF2) and receptor-interacting protein (RIP). The role of JNK in TRAIL-induced tumour cell apoptosis is unclear. In this study, we demonstrate that JNK is activated by TRAIL in colon cancer cells. Inhibition of JNK with L-JNKI reduced rhTRAIL-induced cell death but enhanced cell death induced by selective activation of DR4 or DR5. This difference was unrelated to receptor internalisation or differential activation of c-Jun, but activation of different JNK isoforms. Our data demonstrate that JNK1, but not JNK2 is activated by rhTRAIL in the examined colon cancer cell lines. Although rhTRAIL activated both the long and short isoforms of JNK1, selective activation of DR4 or DR5 led to predominant activation of the short JNK1 isoforms (JNK1α1 and/or JNK1β1). Knockdown of JNK1α1 by shRNA enhanced apoptosis induced by TRAIL, agonistic DR4 or DR5 antibodies. On the other hand, knockdown of the long JNK1 isoforms (JNK1α2 and JNK1β2) had the opposite effect; it reduced TRAIL-induced cell death. These data indicate that the short JNK1 isoforms transmit an antiapoptotic signal, whereas the long isoforms (JNK1α2 or JNK1β2) act in a proapoptotic manner.

Tumour necrosis factor-related apoptosis-inducing ligand (TRAIL) is a member of the TNF ligand superfamily (Ashkenazi and Dixit, 1998). TRAIL induces apoptosis on binding to its transmembrane, death domain-containing receptors, TRAIL receptors 1 (death receptor 4; DR4) and 2 (death receptor 5; DR5). Three other TRAIL-binding receptors exist, but they are unable to transmit an apoptotic signal and thus considered to be 'decoy receptors'. Decoy receptor 1 (DcR1) lacks the transmembrane and intracellular domains and is anchored to the plasma membrane via a glycosylphosphatidylinositol-tail. Decoy receptor 2 (DcR2) possesses a truncated, non-functional death domain, whereas the third decoy receptor, osteoprotegerin is a secreted, soluble receptor (Ashkenazi and Dixit, 1998).
Binding of homotrimeric TRAIL to DR4 and DR5 induces receptor trimerisation and activation leading to recruitment of adaptor proteins and formation of the death-inducing signalling complex (DISC). Procaspase-8 and/or -10 are recruited to the DISC, leading to their oligomerisation and activation. Active caspase-8/-10 can activate the executioner caspases (procaspase-3, -6 and -7) and/or initiate the mitochondrial apoptotic pathway by cleaving the BH3-only protein Bid.
Generally, caspase activation is the main outcome following activation of DR4/5 by TRAIL. However, TRAIL, via adaptor molecules such as TNF receptor-associated factor 2 (TRAF2), receptor-interacting protein (RIP), and the mitogen-activated protein kinase kinases (MKK)-4 and -7, can also activate the c-Jun N-terminal kinase (JNK) pathway (Hu et al, 1999;Lin et al, 2000). JNK activation is also regulated by scaffold proteins, JNKinteracting protein (JIP) and JNK stress-activated protein kinaseassociated protein 1 (JSAP1) (Whitmarsh et al, 1998;Ito et al, 1999). Depending on the cell type and the stimulus, JNK can activate a number of diverse downstream targets including members of the activator protein-1 (AP-1) family, c-Jun, JunD, activating transcriptional factor 2 (ATF2), Bcl-2 proteins, c-Myc and p53 (Bode and Dong, 2007;Lin et al, 2007). Whether JNK induces or suppresses apoptosis is largely dependent on the molecules it activates. For example, JNK can both phosphorylate antiapoptotic Bcl-2 proteins to promote apoptosis or phosphorylate proapoptotic Bcl-2 proteins (e.g. BAD) to inhibit apoptosis (Maundrell et al, 1997;Yu et al, 2004). JNK proteins are encoded by three genes, jnk1, jnk2 and jnk3. Jnk1 and jnk2 encode ubiquitously expressed JNK proteins whereas the jnk3 protein product is primarily found in the brain, heart and to a lesser extent in the testis (Bode and Dong, 2007). Alternative splicing of the jnk transcripts results in 10 different JNK isoforms each of which may target different transcription factors (Gupta et al, 1996). JNK1 and -2 have four isoforms each (a1, a2, b1 and b2). JNK1a1, JNK1b1, JNK2a1 and JNK2b1 isoforms all have a molecular weight of 46 kDa, whereas JNK1a2, JNK1b2, JNK2a2 and JNK2b2 isoforms are larger due to an extended C-terminus and their molecular weight is 54 kDa. JNK3 has two isoforms, JNK3a1 (46 kDa) and JNK3a2 (54 kDa). The aand b-isoforms correspond to two alternative stretches of sequences in the kinase subdomains IX and X (Gupta et al, 1996).
The mechanism and role of JNK activation in TRAIL-induced tumour cell apoptosis has not been fully elucidated. Some reports suggest that JNK is not activated by TRAIL in colon cancers regardless of their sensitivity to TRAIL , whereas others suggest that JNK activation augments TRAILinduced apoptosis in a number of other tumours (Herr et al, 1999;Li et al, 2006;Mikami et al, 2006). The reason for this discrepancy is currently not known. It was therefore of interest to investigate which JNK isoforms are activated by which TRAIL receptor and how the different JNK isoforms contribute to TRAIL-induced colon cancer cell death.

Cell culture and treatments
Colo205 cells were obtained from American Tissue Culture Collection (ATCC, Manassas, VA, USA). HCT15 and HCA7 cells were a kind gift from Professor L Egan (University College Hospital, Galway). Colo205 and HCT15 cells were maintained in RPMI-1640 medium and HCA7 in DMEM medium, both media supplemented with 10% fetal bovine serum (FBS), 2 mM glutamine, 50 U ml À1 penicillin and 50 mg ml À1 streptomycin at 371C, 5% CO 2 in a humidified incubator. Cells were treated with rhTRAIL (nontagged, fragment of amino acids 114 -281; Triskel Therapeutics, Groningen, The Netherlands), agonistic anti-DR4 or anti-DR5 antibodies (Novartis Pharmaceuticals, Basel, Switzerland). To inhibit JNK activation, L-JNKI (Calbiochem, San Diego, CA, USA), a cell-permeable, 21-amino-acid peptide inhibitor of activated JNK was added 30 min before treatment with TRAIL or agonistic antibodies. UV treatment was done at 25 J m À2 for 3 min as a positive control. All reagents were from Sigma-Aldrich (St Louis, MO, USA) unless otherwise stated.

Cell death assay
Cell death was monitored by labelling phosphatidyl serine on the surface of apoptotic cells with Annexin-V-FITC. Following treatment, cells were collected by gentle trypsinisation and incubated for 10 min at 371C to allow membrane recovery after trypsinisation, pelleted by centrifugation at 350 Â g and incubated with Annexin-V-FITC in calcium buffer (10 mM HEPES/NaOH, pH 7.4, 140 mM NaCl and 2.5 mM CaCl 2 ) for 15 min on ice in the dark. A wash step in calcium buffer was carried out before acquisition on a FacsCalibur flow cytometer (Becton Dickinson, Oxford, England).
In-vitro kinase assay (GST p-c-Jun) JNK activity was measured using a specific kit (Cell Signaling Technology) following the manufacturer's instructions and using GST fusion peptide as the specific substrate for JNK. In brief, cell lysates (100 mg protein) were incubated overnight at 41C with GST-c-Jun fusion protein beads. After washing, the beads were resuspended in kinase buffer containing ATP and kinase reaction was allow to proceed for 30 min at 301C. Reactions were stopped by the addition of polyacrylamide gel electrophoresis (PAGE) sample loading buffer. Proteins were separated by electrophoresis on a 10% PAGE gel, transferred on PVDF membrane and incubated with phospho-c-Jun (Ser63) antibody. Finally, blots were subjected to enhanced chemiluminescence and kinase activity determined by densitometric analysis.

Cell surface expression of TRAIL receptors
Cells were washed twice in PBS containing 1% BSA and then incubated with monoclonal antibodies to DR4 or DR5 (Alexis, Lausen, Switzerland) for 40 min. After two wash steps with PBS/ BSA, anti-mouse IgG-FITC (Sigma) secondary antibody was added for 30 min. All incubations were carried out on ice. Negative controls contained isotype control antibody. Cells were analysed by FacsCalibur flow cytometer (Becton Dickinson).

Measurement of receptor internalisation by flow cytometric analysis
To measure cellular uptake of receptor bound TRAIL and agonistic DR4/5 antibody, 2 Â 10 5 Colo205 cells were incubated at 4 or 371C in the presence of 50 ng ml À1 FITC-conjugated TRAIL or agonistic DR4/5 antibody cross-linked with FITC-labelled anti-mouse antibody for 30 min. Samples were rapidly chilled on ice to inhibit endocytosis and cells were collected by a brief centrifugation at 41C. After washing twice in prechilled wash buffer (20 mM Hepes, pH 7.4, 150 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 ), cell surface-bound ligand/antibody was removed by resuspension in prechilled acid wash solution (0.2 M NaCl, 0.2 M acetic acid) for 5 min on ice. Cells were subsequently washed three times in wash buffer and resuspended in cold PBS containing 2% (w/v) FBS before immediate quantification of ligand internalisation using a FacsCalibur flow cytometer (Becton Dickinson).

Immunoprecipitation of JNK
Cells were lysed in phosphate lysis buffer (PLB) containing 20 mM sodium phosphate, 137 mM NaCl, 25 mM sodium-b-glycerol phosphate, 2 mM sodium pyrophosphate, 2 mM EDTA, 1% Triton X-100, 10% glycerol, 1 mM DTT, 1 mM PMSF, 5 mg ml À1 aprotinin, 2.5 mg ml À1 leupeptin, 1 mM Na 2 VO 3 and 1 mM NaF with sonication. To cross-link mouse monoclonal JNK1 (BD Pharmigen) and p-JNK (Cell Signaling Technology) antibodies to protein G-sepharose beads (Sigma), 2 mg antibody was incubated with 30 ml of beads for 1 h at 41C, washed twice with PLB, resuspended in 0.2 M sodium borate pH 9.0 containing 20 mM dimethylpimelidate (DMP) and incubated for 45 min at room temperature. The reaction was stopped by washing the beads once with 0.2 M ethanolamine/HCl pH 8.0 and incubating at room temperature for 2 h. The beads were washed twice in 200 mM glycine and 200 mM NaCl pH 2.5 followed by a wash in 500 mM Tris/HCl pH 8.0. Protein (300 mg) was incubated for 2 h at 41C with cross-linked JNK1 or p-JNK (all isoforms) antibodies bound to protein G-Sepharose. After two rounds of washes with PLB, the beads were resuspended in 30 ml of 1X Laemmli buffer. The supernatant was loaded on a 10% SDS-PAGE acrylamide gel and transferred to nitrocellulose membranes. Membranes containing the JNK1 immunoprecipitates were incubated with mouse monoclonal JNK1 (BD Pharmigen) and rabbit monoclonal p-JNK (Cell Signaling Technology) antibodies to identify which JNK1 isoforms were activated. Membranes of the p-JNK immunoprecipitates were probed with JNK2 (Santa Cruz) and rabbit monoclonal total p-JNK (Cell Signaling Technology) antibodies to identify the JNK2 isoforms activated. Detection and visualisation was carried out as described above.

RNA isolation, cDNA synthesis, RT -PCR
Total RNA was isolated from cells using the GenElute Mammalian Total RNA Extraction kit (Sigma). Reverse transcription was carried out with 2 mg total RNA and oligo(dT) (Invitrogen, Paisley, Scotland) using 20 U/25 ml reaction avian myeloblastosis virus reverse transcriptase. cDNAs for genes of interest were amplified during 32 cycles of 30 s denaturation at 941C, 30 s annealing at 561C, and 60 s extension at 721C, with the following primers:

Statistical analysis
Differences in Annexin V staining between the treatment groups were analysed used a non-paired Student's t-test, with a significance of Po0.05. Error bars are shown as standard error of mean (s.e.m.). All statistical analysis was carried on Graphpad Prism 4 (GraphPad Software Inc., La Jolla, CA, USA).

Colo205, HCT15 and HCA7 colon cancer cells are sensitive to TRAIL
To examine the sensitivity of colon cancer cells to TRAIL, Colo205, HCT15 and HCA7 cells were treated with increasing concentra-tions of rhTRAIL for 24 h and cell viability assessed by MTT assay (Figure 1). All three cell lines express both DR4 and DR5 on their surface (Supplementary Figure 1; Figure 2C). The viability of all three cell lines decreased in a dose-dependent manner. Colo205 cells were the most sensitive to rhTRAIL, with 10 ng ml À1 of rhTRAIL sufficient to decrease cell viability by 61.8 ± 2.3% ( Figure 1A). HCT15 and HCA7 cells were less sensitive to rhTRAIL. In these cell lines, a maximal decrease in cell viability to 65.5±3.6% and 80.3±3.0%, respectively, was achieved following treatment with 50 ng ml À1 of rhTRAIL. No further decrease in TRAIL activates the JNK pathway in colon cancer cell lines via both DR4 and DR5 To examine whether the JNK pathway was activated during TRAIL-induced colon cancer cell death, phosphorylation of JNK and its target, c-Jun were assessed by western blot analysis following treatment with rhTRAIL. rhTRAIL (20 ng ml À1 for Colo205 and 50 ng ml À1 for HCT15 and HCA7 cells) resulted in phosphorylation of JNK in all three cell lines ( Figure 2A). Phosphorylation of c-Jun followed a similar pattern ( Figure 2B). To elucidate whether JNK phosphorylation was mediated by DR4 or DR5, Colo205 and HCT15 cells were treated with agonistic antibodies against DR4 or DR5. Both cell lines were found to express DR4 and DR5 on the cell surface and ligation of both  Figure 2C and D). Colo205 cells were more sensitive to the DR5-agonistic antibody (44.3%±3.5 and 27%±1.5 cell death induction by anti-DR5-and anti-DR4 antibodies, respectively). The DR4-and DR5 antibodies (10 nM) induced almost equal levels of apoptosis in HCT15 cells (DR4, 49.8 ± 3.1% and DR5, 45.3 ± 5.7%) measured at 5 h after treatment. Ligation of both DR4 and DR5 could induce JNK phosphorylation ( Figure 2E) in both cell lines. In Colo205 cells, DR5 ligation induced a stronger JNK phosphorylation. This may correlate with the stronger apoptosis-inducing ability of DR5 in these cells ( Figure 2E). Agonistic antibody treatment of HCT15 cells demonstrated that the DR4 and DR5 receptors can equally trigger JNK phosphorylation, which was detected maximally after 2 h treatment ( Figure 2E). These data suggest that both DR4 and DR5 can induce JNK activation.
Inhibition of JNK potentiates apoptosis-induction by selective activation of DR4 or DR5, but reduces apoptosis induced by rhTRAIL In colon cancer cells, the role of JNK in death receptor-induced apoptosis has not been fully elucidated. One of the most selective JNK inhibitors currently available is the JNK-inhibitory peptide analogue, L/D-JNKI (Barr et al, 2002) and thus was chosen to inhibit JNK to examine the role of JNK in TRAIL-induced colon carcinoma cell apoptosis. Inhibition of JNK by L-JNKI was confirmed by measuring JNK activity in TRAIL-and UV-treated HCT15 cells with an in vitro kinase assay ( Figure 3A).
Colo205 and HCT15 cells were treated with rhTRAIL or agonistic DR4/5 antibodies with or without L-JNKI of 50 mM, a concentration generally used due to the short half-life time of the inhibitor pretreatment (Bonny et al, 2001). Inhibition of JNK activity in Colo205 cells reduced TRAIL-induced apoptosis ( Figure 3B; P ¼ 0.0041), however enhanced agonistic DR4-(P ¼ 0.0002) and DR5 antibody-induced cell death (P ¼ 0.0006; Figure 3B). A similar pattern, albeit to a lesser extent, was observed with HCT15 cells. Inhibition of JNK activity again resulted in a significant increase in DR5 but not DR4 antibody-induced apoptosis (P ¼ 0.310; P ¼ 0.041, respectively). On the other hand (as was observed in Colo205 cells), inhibition of JNK reduced rhTRAIL-induced apoptosis (P ¼ 0.0254; Figure 3C). This surprising result can be explained by either different JNK isoform activation or different JNK compartmentalisation induced by the different treatments.
Receptor internalisation occurs following treatment with rhTRAIL and agonistic DR4/5 antibodies On ligation by TRAIL, the TRAIL receptor-ligand complexes can be internalised and surface bound vs internalised TNF receptors have been shown to induce different signal transduction pathways resulting different cellular responses (Schutze et al, 1999;Schneider-Brachert et al, 2004;Varfolomeev et al, 2005;Kohlhaas et al, 2007). We hypothesised that unlike rhTRAIL, agonistic DR4/ 5 antibodies do not trigger receptor internalisation resulting in JNK activation in a different cellular compartment. Colo205 cells were treated with FITC-labelled rhTRAIL or agonistic DR4/5 antibodies cross-linked by a FITC-labelled secondary antibody for 30 min at either 371C or þ 41C and their internalisation analysed by flow cytometry. An acid wash step was carried out at þ 41C after the incubation to remove all non-internalised, surface bound ligand/antibody ensuring that any fluorescent signal was due to internalised ligand/antibody-receptor complexes. The flow cytometric analysis showed that both rhTRAIL and agonistic antibodies bound to the TRAIL receptors at both 371C and þ 41C and were all internalised to a similar extent, when incubated at 371C but not at þ 41C ( Figure 4A and B). The same samples were also tested for the ability of rhTRAIL and agonistic antibodies to bind to and activate their receptors in these conditions. At the end of the 30 min incubation, unbound rhTRAIL or agonistic antibodies were removed by a wash step. The samples were incubated for an additional 3 h and induction of apoptosis was detected as a measure of receptor activation. The extent of apoptosis was the same regardless of incubation temperature ( Figure 4C), confirming that all treatment conditions enabled ligand/antibody-receptor interaction. These results argue against the compartmentalisation hypothesis.

rhTRAIL and agonistic DR4/5 antibodies phosphorylate distinct JNK1 isoforms
To address the different effects of JNK inhibition on apoptosis, we next investigated which JNK isoforms were phosphorylated by rhTRAIL and the DR4/5 antibodies. JNK1 was immunoprecipitated from rhTRAIL-treated and agonistic antibody-treated Colo205 and HCT15 cells. Immunoprecipitates were electrophoresed on SDS -PAGE and probed for phosphorylated JNK, to identify which JNK1 isoforms were activated by the different treatments. For quantification, blots were also probed for total JNK1 ( Figure 5A) and the densitometric ratio of p-JNK1-long or -short to total JNK1-long or -short was calculated ( Figure 5B). In both cell lines, all treatments could initiate phosphorylation of the short JNK1 isoform (46 kDa, JNK1a1 and/or JNK1b1 isoforms), whereas the long isoform (54 kDa, JNK1a2 and/or JNK1b2) was only phosphorylated after rhTRAIL treatment ( Figure 5A and B). Lysate input from all treated samples confirmed JNK phosphorylation in both cell lines ( Figure 5A). Due to the high homology between the a1/2 and b1/2 isoforms, we could not differentiate between the a1/b1 and the a2/b2 isoforms. To assess which JNK2 isoforms were phosphorylated following selective receptor activation or rhTRAIL treatment, immunoprecipitation with JNK2-specific antibody was attempted, but was unsuccessful due to technical difficulties (no suitable, isoformspecific antibody was available, data not shown). As an alternative strategy, phospho-JNK (including all JNK1 and JNK2 isoforms) was immunoprecipitated, electrophoresed by SDS -PAGE and probed with total JNK2 antibody (this identified whether the short or long JNK2 isoforms were phosphorylated). Total phospho-JNK levels of the same blots were also determined to quantify the levels of phospho-JNK2 isoforms, in a similar manner as for the JNK1 isoforms ( Figure 5C and D). The levels of phosphorylated JNK2 did not increase from the baseline after any of the treatments, suggesting that JNK2 is not activated by TRAIL receptors ( Figure  5C and D). These data demonstrate that JNK1 is the main JNK subtype activated by TRAIL receptors and selective activation of DR4 or DR5 activated predominantly the short isoforms of JNK1 (JNK1a1 and/or JNK1b1) whereas rhTRAIL led to phosphorylation of both the short and long JNK1 isoforms (ie JNK1a2 and/or JNK1b2).

JNK1a1 has an antiapoptotic function
The above results indicated that the short isoforms of JNK1 (JNK1a1 and/or JNK1b1) may play an antiapoptotic role on  (B) Quantification of receptor internalisation. The graph shows the difference of isotype control-normalised MFI measured at 4 and 371C (normalised MFI at 371C minus normalised MFI at þ 41C). The first set of bars shows the level of receptor internalisation occurred at 371C after treatment with rhTRAIL or agonistic antibodies (normalised MFI at 371C/acid wash -normalised MFI at þ 41C/acid wash). The second set of bars indicates that rhTRAIL and agonistic DR4-and DR5 antibodies bound to the TRAIL receptors to a similar level regardless of the incubation temperature. (C) Induction of apoptosis by treatments detailed in point a. After the 30 min incubation with rhTRAIL or antibodies, the unbound molecules were removed by a wash step and the cells were incubated in normal growth medium for an additional 3 h after which induction cell death was measured with Annexin V. The data shown are representative of three independent experiments.
Role of JNK1 isoforms in TRAIL-induced apoptosis D Mahalingam et al selective activation of DR4 or DR5. In an effort to dissect its role, Colo205 cells were transiently transfected with an shRNA expression plasmid to JNK1a1. RT -PCR demonstrated knockdown of JNK1a1 following transient transfection of JNK1a1 shRNA, without affecting total JNK1 and JNK1b1 mRNA levels confirming isotype-specific knockdown ( Figure 6A; isoform specific primer design is depicted in Supplementary Figure 2). Accordingly, a decrease in the protein levels of the short JNK1 isoform was induced by the JNK1a1 shRNA observed by western blotting ( Figure 6B). Induction of apoptosis by TRAIL or selective induction of DR4 or DR5 by agonistic antibodies was assessed in JNK1a1 shRNA transfectants by Annexin V binding. Knockdown of JNK1a1 significantly increased apoptosis induced by TRAIL, as well as agonistic DR4-and DR5-antibodies at 3 h of exposure (TRAIL, P ¼ 0.040; DR4, P ¼ 0.042; DR5, P ¼ 0.046) compared to cells transfected with a scrambled shRNA expressing plasmid ( Figure 6C). These data suggest that JNK1a1 has an antiapoptotic effect.
The only region that differs in the long JNK1 isoforms from the short JNK1 isoforms is a 5-nucleotide sequence and thus this was the only region targetable by siRNA. We designed two siRNAs against this region with selectivity for the a2/b2 isoforms (the targeted region is highlighted in Supplementary Figure 2). The efficiency of the knockdown was analysed by western blotting. Cell lysates of Colo205 cells transfected with JNK1a2/b2 siRNA or siRNA against GFP as a negative control for 24 h was analysed for JNK1 expression, using a JNK1-specific antibody ( Figure 7A). JNK1a2/b2 siRNAs reduced the expression of the long JNK1 isoforms, without having a non-specific effect on the short JNK1 isoforms. JNK1a2/b2 siRNA transfected Colo205 cells were then treated with rhTRAIL (40 and 60 ng ml À1 ) for 3 h and induction of apoptosis was measured ( Figure 7B). Knockdown of the long JNK1 isoforms reduced TRAIL-induced apoptosis, indicating that these JNK1 isoforms are indeed proapoptotic (40 ng ml À1 rhTRAIL, P ¼ 0.049 and 0.04 for siRNA 1 and 2, respectively; 60 ng ml À1 rhTRAIL, P ¼ 0.047 and 0.04 for siRNA 1 and 2, respectively).    1994;Cahill et al, 1996;Yang et al, 1997;Herr et al, 1999). The role of this JNK activation in apoptosis is unclear and opposing, proand antiapoptotic functions have been proposed (Bode and Dong, 2007;Yoo et al, 2008). Similarly, controversy exists as to the role that activated JNK might play in TRAIL-induced colon carcinoma apoptosis . This study demonstrates that in colon carcinoma cells that express both DR4 and DR5, both receptors are able to trigger JNK activation and c-Jun phosphorylation. To elucidate the role of JNK activation in DR4-and DR5mediated apoptosis in colon carcinoma cells, JNK activity was blocked by L-JNKI. L-JNKI was chosen over the widely used SP600125 (Bennett et al, 2001) as recent studies found that SP600125 is a rather non-specific JNK inhibitor (Bain et al, 2003). Our studies found that inhibition of JNK by L-JNKI reduced rhTRAIL-induced cell death, suggesting a proapoptotic role for JNK. Interestingly, inhibition of JNK potentiated cell death induced by selective activation of DR4 or DR5, suggesting that depending on the type or the total number of receptors activated, a pro-or antiapoptotic JNK signal transduction pathway can be activated.

DISCUSSION
It has been shown that after binding to its receptors, TRAIL is rapidly internalised by both clathrin-dependent and -independent endocytosis (Kohlhaas et al, 2007). Unlike TNF-R1, internalisation is not required for TRAIL-induced apoptosis, as the assembly of the TRAIL DISC already occurs at the cell membrane (Kohlhaas et al, 2007). However, internalisation of the ligated receptor has been suggested to play a role in other TRAIL-mediated signalling events such as activation of JNK or NF-kB (Varfolomeev et al, 2005). In our hands, internalisation of TRAIL receptors ligated by agonistic DR4 and DR5 antibodies or by rhTRAIL revealed no significant differences; in all cases, the ligated receptor complex was rapidly internalised. This indicates that the opposing apoptosis-modulatory effect of JNK activation induced by TRAIL or selective DR4/DR5 activation was not due to receptor internalisation but possibly different isoforms of JNK activated by the individual receptors.
Ten isoforms of JNK are known to exist as a result of alternative splicing of the three genes, jnk1, jnk2 and jnk3 (Gupta et al, 1996). Little is known about the role of these isoforms in apoptosis. Overexpression of JNK1b1 increases resistance to vesicular stomatitis virus-induced cell death in 3T3 fibroblasts, whereas overexpression of JNK1a1 and JNK1b1 potentiates cisplatin-and doxorubicin-induced cell death in sarcoma cell lines (Han et al, 2002;Koyama et al, 2006). Previous studies have demonstrated a role for either JNK1 or JNK2 in TNFa-induced apoptosis (Dietrich et al, 2004;Liu et al, 2004). Our results show that the chief JNK isotype activated by DR4 and DR5 is JNK1. Furthermore, whereas TRAIL-mediated receptor activation led to activation of both the long and short isoforms of JNK1, selective ligation of DR4 or DR5 with cross-linked agonistic antibodies predominantly activated the short JNK1 isoforms (JNK1a1 and/or b1) and the difference in cell death seen following JNK inhibition may be related to the different JNK1 isoforms, with the short isoforms of JNK1 (JNK1a1 and JNK1b1) transmitting an antiapoptotic signal and the long isoforms of JNK1 (JNK1a2 or JNK1b2) mediating a proapoptotic signal. As a reason why activation of individual TRAIL receptors by agonistic antibodies activates different JNK isoforms from rhTRAIL, it is likely that the agonistic DR4-and DR5-specific antibodies trigger different receptor clustering, or intracellular conformational changes from those induced by rhTRAIL. It is also Colo205 cells were treated with agonistic DR4/5 antibodies (5 nM) or rhTRAIL (20 ng ml À1 ) for 3 h, after which percentage of apoptotic cells was determined with Annexin V. The graph shows the average cell death induced ± s.e.m. of five independent experiments (sum of Ann V þ /PI À and Ann V þ /PI þ percentages). The * indicates significant differences (Po0.05).
Role of JNK1 isoforms in TRAIL-induced apoptosis D Mahalingam et al possible that on rhTRAIL treatment, higher-order heteromeric receptor complexes (receptosomes) including both DR4 and DR5 are formed where the interaction between the death domains of the various receptor trimers allows for the recruitment of more and/or different adaptor proteins. It has been shown that different agonists of DR4 and DR5 (agonistic antibodies and recombinant ligand) trigger different conformational changes in the receptors resulting in differences in adaptor protein recruitment, such as FADD. This can lead to activation of different downstream signal transduction and kinase pathways (Thomas et al, 2004). Our current studies investigate this possibility. Of note, in both Colo205 and HCT15 cells, rhTRAIL was a more efficient inducer of apoptosis than either DR4 or DR5 agonistic antibodies. This may be in line with the hypothesis that full signal transduction requires both DR4 and DR5 activation simultaneously.
To elucidate the roles of the different JNK1 isoforms, JNK1a1 expression as well as JNK1a2/b2 expression was knocked down with shRNA. Knockdown of the short JNK1 isoform potentiated TRAIL-induced apoptosis, whereas elimination of the long JNK1 isoforms blocked rhTRAIL-induced cell death. These results demonstrate that the short JNK1 isoform, JNK1a1 acts against apoptosis, whereas the long JNK1 isoforms promote it. In has to be noted, that knocking down JNK efficiently (as JNK is an abundant protein) is very difficult. Even 30 -50% of the remaining protein may be sufficient to transmit a signal. Selective targeting one JNK isoform is even more complicated and thus our method probably underestimated the real potency of JNK isoforms in modulating TRAIL-mediated apoptosis.
Genes regulated by JNK1a1 and JNK1b1 have been previously identified by Han et al (2002). Overexpression of a constitutively active form of JNK1a1 led to the induction of a number of antiapoptotic/prosurvival genes including proliferin (mitogenregulated protein), Ack protein tyrosine kinase and serum and glucocorticoid regulated kinase (sgk) and repression of proapoptotic proteins, such as insulin-like growth factor binding protein-4 (IGFBP-4) and suppressor of cytokine signalling-3 (SOCS-3) as a strong indication that JNK1a1 indeed triggers antiapoptotic signalling (Han et al, 2002). Identifying which of these genes, or other so far unidentified genes, play important role warrants further studies.
In conclusion, we show for the first time that combined activation of all TRAIL receptors vs selective activation of DR4 or DR5 result in the activation of different JNK isoforms. Moreover, the short isoform of JNK1, JNK1a1, generates an antiapoptotic signal whereas the long isoforms of JNK1 trigger proapoptotic signalling. Our findings shed light on the apparently contradictory results surrounding the role of JNK signalling in TRAIL-induced apoptosis and also suggest a way forward to target cancer cells for sensitisation to killing by inhibition of short isoforms of JNK1. (A) Knockdown of JNK1a2/b2 in Colo205 cells by siRNA. Cells were nucleofected with two different JNK1a2/b2 siRNAs, or an siRNA against GFP as a negative control. Cell lysates were prepared 24 h after transfection and JNK1-short and -long expression was detected with primers western blotting. (B) Knockdown of JNK1a2/b2 reduces TRAIL-mediated apoptosis. Colo205 cells were treated with rhTRAIL (40 and 60 ng ml À1 ) for 3 h, after which percentage of apoptotic cells was determined with Annexin V. The graph shows the average cell death induced ± s.e.m. of three independent experiments (sum of Ann V þ /PI À and Ann V þ /PI þ percentages). The asterisk (*) indicates significant differences (Po0.05).
Role of JNK1 isoforms in TRAIL-induced apoptosis D Mahalingam et al