THE JOURNAL OF GENE MEDICINE J Gene Med 2000; 2: 334±343.
NF±kB kinetics predetermine TNF-a sensitivity of colorectal cancer cells Ralf M. Zwacka* Lesley Stark Malcolm G. Dunlop Department of Oncology, University of Edinburgh, Edinburgh, UK *Correspondence to: Ralf M. Zwacka, Department of Oncology, University of Edinburgh, Western General Hospital, MRC Human Genetics Unit, Crewe Road, Edinburgh EH4 2XU, UK. E-mail: [email protected]
Abstract Background Tumour necrosis factor (TNF)-a has considerable anti-tumour activity and may have potential as a treatment for metastatic colorectal cancer. However, TNF-a responses in patients and cell lines are variable and TNF-a treatment is associated with dose limiting clinical toxicity. Activation of NF±kB is protective against TNF-a induced cell death, and this may explain tumour resistance. Methods In order to provide further understanding of determinants of TNFa responses, we studied TNF-a induced NF±kB activation and variable tumour responses. We analysed the kinetics of TNF-a induced NF±kB activation in colorectal cancer cells and determined whether it is possible to sensitize colorectal tumour cells to TNF-a by modulation of NF±kB signalling. Results We demonstrated that sustained NF±kB activation exceeding 16 h was observed in HRT18 and SW480 cells and was associated with TNF-a resistance. In contrast, transient NF±kB activation in HCT116 cells was associated with sensitivity to cytotoxic TNF-a effects, suggesting that NF±kB kinetics may have utility as clinical marker of TNF-a tumour resistance. Despite variable TNF-a responses and NF±kB kinetics, all three colorectal cancer cell lines were highly sensitive to treatment with the TNF-related apoptosis-inducing ligand (TRAIL) which induced only transient NF±kB activation. This further supports the notion of a pre-determined NF±kB response in¯uencing receptor-mediated cell death. We also show that stable transfection and adenoviral-mediated expression of IkB(A32/36) can be used to confer TNF-a sensitivity to colorectal tumour cells previously resistant. Conclusions These ®ndings indicate that a combined approach using gene therapy and recombinant TNF-a merits further appraisal. Furthermore, the kinetics of the TNF-a response could be determined using a `test-dose' to indicate whether individual patients might bene®t from this gene therapy approach. Copyright # 2000 John Wiley & Sons, Ltd. Keywords
NF±kB; TNF-a; colorectal cancer; TRAIL; gene therapy
Introduction Received: 23 May 2000 Revised: 19 June 2000 Accepted: 28 June 2000 Published online: 4 July 2000
Copyright # 2000 John Wiley & Sons, Ltd.
Current treatment regimens for hepatic colorectal metastases rarely effect a cure and only marginally prolong survival. Hence, there is a clear need for novel approaches to treat such established disease. Combination chemotherapeutic approaches that incorporate gene therapy are attractive since genetic manipulation has the potential to sensitize cancer cells. Initial excitement surrounding the use of tumour necrosis factor-a (TNF-a) as an
NF-kB Activation Kinetics in TNF-a Resistance
anti-neoplastic agent [1,2] waned as it became clear that some tumours are relatively resistant to TNF-a and its use is limited by toxicity [3±5]. Recently, a molecular mechanism, which can account for at least part of the observed TNF-a resistance has become evident [6±8]. TNF-a induces a pro-apoptotic molecular cascade mediated by death domain proteins that associate with the TNF-a receptor . This results in activation of a cascade of caspase proteases . However, tumour cells also initiate a protective antiapoptotic pathway mediated by activation of NF±kB. Thus, the ability of TNF-a to elicit cell death depends on the balance between apoptotic and protective signals. Several NF±kB target genes such as A20, MnSOD, the bcl2 homologue b¯-1/A1 as well as TRAF-1 and 2, in conjunction with c-IAP-1 and 2, have been implicated in the anti-apoptotic functions. Overexpression of these genes has been shown to protect cells from TNF-a cytotoxicity [11±13]. Since NF±kB activation plays such a pivotal role in cellular protective responses it might also have a role in determining TNF-a resistance in colorectal and other cancer cells. Furthermore, NF±kB responses may have relevance to drug resistance in other chemotherapeutic approaches such as CPT11 . TNF-related apoptosis-inducing ligand (TRAIL) is another factor shown to activate NF±kB and to induce cell death in several tumour cell lines , but seems to induce very limited NF±kB mediated anti-apoptotic responses . It has also been shown to have antitumour activity in xenografts of human colon cancer cell lines . TRAIL (also known as Apo2), induces programmed cell death following binding to surface receptors  by utilizing the same cellular death pathway as TNF-a . The NF±kB transcription factor  is normally sequestered in the cytoplasm by IkB±a, or other members of the IkB family . Following stimulation by extracellular signals such as TNF-a, IL-1 or LPS, IkB is phosphorylated at serine residues 32 and 36 by the IKK(IkB±a kinase) complex [22±24]. Such phosphorylation leads to proteolytic degradation of IkB which in turn unmasks a nuclear target sequence within NF±kB and leads to its translocation to the nucleus. The expression of a super-repressor form of IkB±a mutated at residues 32 and 36, designated IkB(A32/36) or IkB-SR has been shown to block the nuclear translocation of NF±kB in Hela, Jurkat and ®brosarcoma cells leading to increased TNF-a sensitivity [6±8,14]. Inhibition of NF±kB by this super-repressor prevents the activation of anti-apoptotic genes such as cIAP-1 and 2 that block a pro-apoptotic response by inhibition of caspase-8 . Hence, a gene therapy strategy could be envisaged that exploits mutant IkB to selectively increase in vivo tumour sensitivity but with a concomitant TNF-a dose reduction to minimize clinical toxicity. This strategy requires only transient expression of IkB (A32/36), and so replication-defective adenoviral vectors are an attractive gene delivery system. Several suicide gene approaches are currently in experimental and clinical trials with some degree of success Copyright # 2000 John Wiley & Sons, Ltd.
. Adenoviral mediated IkB (A32/36) tumour transduction followed by TNF-a treatment might have particular advantages, because TNF-a increases tumour sensitivity to the chemotherapeutic agent 5-FU . A combination approach is attractive in view of potential synergism between agents. In this article we show that the kinetics of NF±kB activation have a major in¯uence on the TNF-a response. Sustained NF±kB activation was associated with TNF-a resistance and transient NF±kB activation with TNF-a sensitivity. We provide further evidence that receptormediated apoptosis is in¯uenced by NF±kB activation kinetics since treatment with TRAIL induced a transient NF±kB response that resulted in apoptosis in all colorectal cancer cells studied. The NF±kB response could be used as a biological marker to predict individual TNF-a responses with respect to its anti-neoplastic function and could aid in the tailoring of TNF-a based therapies. We show that colorectal cancer cells that are resistant to TNF-a can be sensitized by expression of a dominant negative IkB, which inhibits the NF±kB mediated anti-apoptotic cellular programme.
Materials and methods Cell culture and viral infection HCT116, HRT18 and SW480 human colorectal cancer cells were grown in McCoy, RPMI and L-15 medium (Gibco BRL, Paisley, UK), respectively, containing 10% fetal calf serum (FCS) and 1% penicillin/streptomycin. For functional studies, cells were treated with TNF-a (R&D Systems, Minneapolis, MN, USA) of concentrations of 0, 1, 10 and 100 ng/ml in 2% serum-containing medium and harvested at different time-points. TRAIL (R&D Systems) was used at a concentration of 10 ng/ml. Stable transfectants were generated by transfection of a pcDNA construct containing an IkB(A32/36) cDNA under the control of a CMV early promoter/enhancer element . A standard lipofection method was used (Gibco). The transfected cells were selected under 1.5 mg/ml G418 (Gibco) and 20 individual clones picked, expanded and tested for transgene expression. Despite several attempts we have not been able to identify any IkBexpressing SW480 clones. We believe that such clones might have signi®cant survival disadvantages during G418 selection. Adenovirus infections were performed in medium containing 2% FCS and 1% penicillin/streptomycin for 24 h, after which cells were harvested for transgene analysis, or TNF-a treatment was started. Multiplicity of infection (MOI) of 0, 1000, 5000 and 10 000 particles/ cell corresponding to 0, 40, 200 and 400 pfu/cell (25 particles required to infect one cell of the reference cell line 293), respectively were used. For growth studies and EMSA analysis an MOI of 5000 was used. Growth was measured by seeding 104 cells into each well of a 24-well plate, and treated with TNF-a, as described above, 2 days J Gene Med 2000; 2: 334±343.
later. The cell number/well was counted prior to TNF-a application and then every 24 h for the next 5 consecutive days for parental cells and stable transfectants, and 3 days for adenovirus transduced cells. All experiments were performed in triplicate and non-TNF-a treated controls were always run in parallel. The Ad.IkB was kindly provided by David Brenner, University of North Carolina .
Apoptosis assay Programmed cell death was measured following TNF-a and TRAIL treatment using a Cell Death ELISA system (Roche Diagnostics, Mannheim, Germany). Cells were seeded at 104 cells/well in a 24-well plate, grown for 2 days and then treated for 24 h with TNF-a or TRAIL (10 ng/ml). Treated cells were trypsinized, washed and lysed in lysis buffer provided by the manufacturer. Aliquots (5 ml) of cytoplasmic extract were analysed according to the manufacturer's instructions for quantitative determination of apoptosis.
Cytoplasmic and nuclear extracts Nuclear extracts were prepared from cells grown on 100 mm plates. Following treatment with 10 ng/ml TNFa or TRAIL for 1, 5, 16 and 48 h, the cells were washed twice with ice-cold PBS and then harvested in 1 ml PBS. The cells were pelleted in a microfuge at 3000 rpm for 5 min at 4uC. After resuspending the cells in 100 ml Dignam A buffer (10 mM Hepes, pH 7.9, 1.5 mM MgCl2, 10 mM KCl, 1 mM DTT, 1 mM PMSF) they were incubated on ice for 15 min and lysed by vigorous pipetting. Nuclei were then pelleted by centrifugation at 13 000 rpm for 1 min at 4uC. The supernatant was harvested and used as cytoplasmic extract for IkB-a Western blots, after protein concentrations had been measured by the method of Bradford (BioRad, Hercules, CA, USA). The nuclei were resuspended in 50 ml of Dignam C (20 mM HEPES, pH 7.9, 25% glycerol, 0.42 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 1 mM DTT, 1 mM PMSF) and incubated for 30 min at 4uC with gentle agitation. Cell membranes were pelleted and the supernatant transferred to a fresh tube prior to measuring protein content of nuclear extracts by a Bradford assay (BioRad).
Western blot analysis Cytoplasmic protein (100 mg) was separated on a denaturing 10% SDS-PAGE and transferred to a PVDF membrane (BioRad) by semidry electroblotting for 30 min at 10 V. The membrane was stained with 0.5% PonceauS to check for ef®cient transfer and subsequently blocked in 4% non-fat dry milk solution in PBS, supplemented with 0.3% Tween20. This solution was used for all other antibody incubations and washing steps. The primary monoclonal antibodies against IkB-a and the PK-tag (kindly provided by Ron Hays, University of Copyright # 2000 John Wiley & Sons, Ltd.
R. M. Zwacka et al.
StAndrews) were diluted 1 : 1000 and incubated for 1 h at room temperature after which the membrane was washed four times. A 1 : 2000 diluted goat anti-mouse antibody coupled to horseradish peroxidase (Amersham, Little Chalfont, UK) was used as secondary antibody and incubated for 1 h at room temperature. The anti-human CuZnSOD-horseradish-peroxidase-conjugated antibody (The Binding Site, Birmingham, UK) used as a loading control was diluted 1 : 2000. Following antibody incubations, the membrane was washed four times in blocking buffer and ®nally twice in PBS. Proteins were visualized, using enhanced chemiluminescence mix (ECL, Amersham, Little Chalfont, UK) on X-ray ®lm (Kodak, Rochester, NY, USA). The bands were quanti®ed using the Herolab E.A.S.Y.Win32 image analysis software.
Electrophoretic mobility shift assays Nuclear extracts (6 mg) were incubated on ice for 1 h with 1 ml anti-NF±kB antibodies (p50, p65 and c-Rel at 1 mg/ ml) for supershift assays (Santa Cruz, Santa Cruz, CA, USA). Following pre-incubation, 3.4 ml of 5rEMSA buffer (250 mM KCl, 100 mM HEPES, pH7.9, 25% glycerol, 5 mM EDTA, 5 mM DTT), 1 ml BSA (1 mg/ml), 1 ml poly(dI-dC) oligo (1 mg/ml), 5.6 ml H2O, and lastly 2 ml of double stranded 32P end-labeled oligo (100 000 cpm/ ml) were added. For normal EMSAs the antibodies were omitted and the volumes adjusted with water. The mix was then incubated for 30 min at room temperature and separated on a 4% polyacrylamide gel run in Tris/glycine buffer (125 mM Tris, pH 8.5, 950 mM glycine, 5 mM EDTA). The oligo sequence used for double stranded probes was as follows: NF±kB wt: 5k-AGT TGA GGG GAC TTT CCC AGG C-3k (Santa Cruz).
Results TNF-a and TRAIL effects in colorectal cancer cells The cellular responses to TNF-a and TRAIL of HCT116, HRT18 and SW480 cells, using a Cell Death Detection ELISA were compared by apoptotic indexes expressed as the ratio of apoptosis measurements of treated cells over non-treated cells (Figure 1). HCT116 cells exhibited 4.5fold increased cell death when treated with 10 ng/ml TNF-a (Figure 1) while SW480 and HRT18 were almost completely resistant (1.4- and 1.1-fold, respectively) against TNF-a induced cell death (Figure 1). In contrast to TNF-a, TRAIL (10 ng/ml) induced apoptosis in all three cell lines (Figure 1). HCT116, HRT18 and SW480 cells showed an increase in apoptosis of 3.3-, 3.7- and 4.9-fold, respectively, following treatment with TRAIL. These ®ndings demonstrate a substantial difference between the action of TNF-a and TRAIL in colorectal cancer cells. Since both TNF-a and TRAIL can activate the same caspase-mediated pro-apoptotic pathway, we were interJ Gene Med 2000; 2: 334±343.
NF-kB Activation Kinetics in TNF-a Resistance
Figure 1. Apoptosis assay for HCT116, HRT18 and SW480 colorectal cancer cells following 24 h treatment with 10 ng/ ml TNF-a and 10 ng/ml TRAIL, respectively. Apoptosis was measured using an ELISA Cell Death Assay (Roche Diagnostics) which detects cytosolic DNA fragments, a hallmark of apoptotic processes. The apoptotic index is expressed as the ratio of readings for treated over untreated cells, i.e. an index of 1 means the absence of induced programmed cell death. It shows that HCT116 cells undergo substantial apoptosis following TNF-a treatment (4.5-fold increase) whereas HRT18 and SW480 are relative resistant with 1.4- and 1.1fold increases, respectively. In contrast TRAIL signi®cantly induces apoptosis in all three cell lines. The increases are 3.3-fold in HCT116, 3.7-fold in HRT118 and 4.9-fold in SW480 cells
ested to elucidate the basis of the difference between cell lines in the NF±kB-mediated anti-apoptotic response.
NF±kB kinetics following TNF-a and TRAIL treatment To investigate the observed differential TNF-a sensitivity and the contrast between TNF-a and TRAIL effects, we investigated individual NF±kB responses by electrophoretic mobility shift assay (EMSA). All three cell lines examined showed immediate-early NF±kB activation at 1 h post-TNF-a treatment (Figure 2A). The EMSA analysis also revealed that NF±kB activation in the TNF-a resistant cell lines HRT18 and SW480 was sustained for at least 16 h (Figure 2A). In contrast, in HCT116 cells, NF±kB was activated only transiently, peaking at 1±2 h post-TNFa treatment and disappearing by 5 h. It reappeared, characteristic of a biphasic activation, at 16 h, but completely disappeared again by 48 h (Figure 2A). These ®ndings led us to hypothesize that the physiological response to TNF-a depends on the duration and magnitude of NF±kB activation. Persistent up-regulation of NF±kB was associated with a dominant protective TNF-a-induced response, while short-term activated NF±kB was not capable of eliciting the anti-apoptotic cellular pathways. Next, we examined the kinetics of NF±kB activation following TRAIL treatment. TRAIL induced exclusively transient NF±kB activation with NF±kB activated for 1±2 h in HRT18 and SW480 cells and only at the 5-h time point in HCT116 cells. NF±kB Copyright # 2000 John Wiley & Sons, Ltd.
Figure 2. (A) EMSA analysis of nuclear extracts from TNF-a treated HCT116, HRT18 and SW480 cells. The TNF-a induced NF±kB band is marked with an arrow on the right side. HCT116 cells show transient NF±kB activation. NF±kB disappears after 5 h and weakly re-emerges after 16 h. SW480 and HRT18 cells have nuclear NF±kB for at least 16 and 48 h post-a treatment, respectively. The faster migrating bands are non-speci®c bands, as shown by competition experiments and supershift analysis with NF±kB speci®c antibodies (Figure 3). (B) Human recombinant TRAIL, in contrast to TNF-a, leads to transient NF±kB activation in HCT116, HRT18 and SW480 cells. The EMSA analysis shows that TRAIL (10 ng/ml) activates NF±kB at 1±5 h post-treatment in HRT18 and SW480 cells, while in HCT116 cells it is only activated at 5 h. At 16 h the nuclear NF±kB levels have returned to basal levels. The induced NF±kB band is marked by an arrow, while faster migrating bands are non-speci®c
returned to basal levels at 16 h post-TRAIL treatment (Figure 2B). Taken together, these ®ndings suggest that the kinetics of the NF±kB response play a crucial role in determining whether a pro-apoptotic or an anti-apoptotic response predominates. Furthermore, the response of particular tumour cells appears to be predetermined. In contrast to the varying TNF-a induced responses, TRAIL signals transient NF±kB activation and consequently death in cells irrespective of their TNF-a sensitivity.
Characterization of the TNF-a induced NF±kB complex Supershift analyses using antibodies against different members of the Rel-family (p50, p65 and c-Rel) revealed that the induced form of NF±kB contained mainly p65 and c-Rel subunits with p50 only a minor constituent (Figure 3). Previous reports indicate that the most frequently found form of activated NF±kB consists of p50/p65 heterodimers, and so our ®nding of the involvement of c-Rel in the complex in colorectal cancer cells is intriguing. Usually the expression of cRel is best described in association with lymphoid cell J Gene Med 2000; 2: 334±343.
R. M. Zwacka et al.
Figure 3. Supershift analysis of 1 h post-TNF-a nuclear extracts from HCT116, HRT18 and SW480 cells. It identi®es p65 (RelA) and c-Rel as the major components of the activated NF±kB complex in the three colorectal cancer cell lines as indicated by the arrow. A p50 homodimer predominantly found in HRT18 cells is also pointed out. The supershifted bands are indicated by dashed arrows
differentiation and growth stimulation in a variety of cell types, and associated with lymphoid malignancies . Interestingly, only HRT18 cells contained any signi®cant levels of p50/p50 homodimers, however the functional signi®cance of these ®ndings remains to be determined.
Generation of IkB(A32/36) expressing colorectal cancer cell clones In order to determine whether it is possible to render resistant colorectal cancer cells sensitive to TNF-a, we generated stable transfectants of HCT116 and HRT18 cells expressing the super-repressor (SR) IkB(A32/36) driven by a CMV early promoter/enhancer element . These cell lines could further aid in the elucidation of the role of NF±kB kinetics as a determinant for sensitivity to TNF-a and potentially other cytotoxic agents. Several G418 resistant clones were isolated and analysed for IkB(A32/36) expression by Western blot (Figure 4A). We selected two IkB expressing clones and one non-expresser from each cell line. The expressing HCT clones were HCT-SR8 and HCT-SR7, while the nonexpressing control clone was HCT-C11. Two HRT clones expressing IkB(A32/36) were designated HRT-SR1 and HRT-SR28, and the non-expressing clone HRT-C22. HCTSR7 expresses signi®cantly lower levels of IkB A32/36 than HCT-SR8, and HRT-SR28 expresses slightly lower levels of IkB(A32/36) than IkB-SR1 (Figure 4A). Clones and their respective parental lines were used in growth and survival experiments. First, we tested the capability of IkB(A32/36) to suppress TNF-a, induced NF±kB activation in the expressing clones. As shown in Figure 4, NF±kB activation is markedly decreased at 1 h post-TNF-a treatment in the high-expressers (HCT-SR8, HRT-SR1 and HRT-SR28) while the non-expressers (HCT-C11 and HRT-C22) are indistinguishable from parental HCT116 and HRT18 cells, respectively. Furthermore, in HRT-SR1 and to a lesser extent in HRT-SR28, Copyright # 2000 John Wiley & Sons, Ltd.
Figure 4. (A) Western blot of isolated stable transfected HCT116 and HRT18 clones. HCT clones IkB-SR7 and IkB-SR8 and HRT clones IkB-SR1 and IkB-SR28 express IkB(A32/36). HCT clone IkB-C11 and HRT clone IkB-C22 are G418 resistant, but had no detectable IkB(A32/36 levels) and were used as controls. The transgenic (tg) PK-tagged IkB(A32/36) has slightly reduced mobility, and therefore runs above the endogenous (eg) IkB in the SDS-PAGE. This was shown in Western blots probed with an IkB-a speci®c antibody labelled IkB. Furthermore, transgene expression of IkB(A32/36) was detected with an antibody directed against the PK-tag, which is labelled Tag. The Western blot was probed with an antibody against human CuZnSOD (Cu) to control for loading and to measure relative expression levels of transgenic IkB(A32/36). (B) EMSA analysis of parental HCT116 and HRT18 and their respective IkB-SR clones. High expressing clones (HCT IkB-SR8; HRT IkB-SR1 and IkB-SR28) inhibit NF±kB activation following 1 h of TNF-a treatment. The HCTSR7 clone does not show any signi®cant inhibition, indicating that the levels of transgene expression in HCT-SR7 are not suf®cient to block NF±kB activation. The NF±kB band is indicated by an arrow
NF-kB DNA binding was still inhibited at the longer time points of 16 and 48 h post-TNF-a (data not shown). The lower expression of IkB(A32/36) transgene achieved in HCT±SR7 is not suf®cient to block NF±kB activation.
IkB(A32/36) sensitizes colorectal cancer cells to TNF-a HRT-SR1 and HRT-SR28 showed markedly increased apoptosis (Figure 5B, upper panel) and growth retardation following TNF-a treatment at concentrations of 1, 10 and 100 ng/ml (Figure 5B, lower panels) compared to parental HRT18 and non-expressing control cells HRTJ Gene Med 2000; 2: 334±343.
NF-kB Activation Kinetics in TNF-a Resistance
Figure 5. (A) An apoptosis assay in HCT116 cells shows endogenously high levels of TNF-a induced programmed cell death (10 ng/ml). Clone HCT-SR8 exhibits a slightly higher apoptotic index (6.1-fold) as compared to parental HCT116 (4.5-fold) and non-expressing control HCT-C11 (3.5-fold), while clone HCT-SR7 (3.8-fold) shows no sensitization. The results depict the mean (tSEM). These ®ndings are in accordance with results from the EMSA analysis, which showed that low levels of IkB(A32/36) expressions as seen in HCT-SR7 cannot inhibit TNF-a induced NF±kB activation. The growth curves of HCT116 cells and clones at different concentrations of TNF-a (2, 0 ng/ml; &, 1 ng/ml; +, 10 ng/ml; r, 100 ng/ml) corroborate the ®ndings from apoptosis tests. Parental HCT116, control HCT-C11 and also low-IkB(A32/36) expressing clone HCT-SR7 do not differ in their growth behaviour following TNF-a treatment. Increasing concentrations of TNF-a cause a marked retardation of cellular growth over 5 days in a dose dependent manner. Clone HCT-SR8 was slightly sensitized as illustrated by a greater effect in terms of lower cell counts at the lowest TNF-a concentration of 1 ng/ml in comparison to parental cells and the other two clones. (B) In contrast, HRT18 cells and the non-expressing clone IkB-C22 demonstrate almost complete resistance against 10 ng/ml TNF-a (1.4- and 1.3-fold, respectively), whereas HRT-SR1 and HRT-SR28 exhibit increased apoptosis (2.4and 2.8-fold). The growth curves of HRT clones IkB-SR1 and IkB-SR28 are markedly sensitized to increasing TNF-a concentrations (2, 0 ng/ml; &, 1 ng/ml; +, 10 ng/ml; r, 100 ng/ml) as compared to parental HRT18 and HRT-C22
C22 (Figure 5B, lower panels). It should be noted that we measured apoptosis in adherent cells following TNF-a treatment, and we therefore may have underestimated Copyright # 2000 John Wiley & Sons, Ltd.
the number of apoptotic cells. HCT-SR8 was only marginally sensitized beyond its natural responsiveness as measured by apoptosis (Figure 5A, upper panel) and J Gene Med 2000; 2: 334±343.
growth (Figure 5A, lower panels), even though the IkB(A32/36) expression levels were comparable to those seen in HRT18 clones. HCT116 parental, control HCT-C11 and low-expressing HCT-SR7 all exhibited similar responses (Figure 5A). The changes in apoptotic responses and growth correlate with the IkB(A32/36) mediated NF±kB inhibition, i.e. cell growth in HRT-SR1 was 1.3-fold more inhibited than in HRT-SR28 cells, broadly in line with the difference in transgene expression levels which were 1.4-fold higher in HRT-SR1 cells. The difference in IkB(A32/36) transgene levels in HCTSR8 over HCT-SR7 was 5.7, but since HCT-SR7 did not inhibit NF±kB activation and consequently showed no sensitization such a correlation can not be made for HCT116 clones. These results indicate that parental HCT116 cells do not elicit a protective response following TNF-a treatment, due to the transient nature of NF±kB activation, and so further sensitization cannot be achieved by blocking the NF±kB pathway. The ®ndings in HCT116 cells are in line with recent ®ndings , and it is likely that the TNF-a, induced pro-apoptotic pathways, such as the caspase cascade, are rate limiting. Interestingly, when we treated IkB(A32/36) expressing clones, in which NF±kB activation is blocked, with TRAIL we did not observe any signi®cant degree of sensitization (data not shown). This indicates that, due to its transient activation, NF±kB plays no role in the inhibition or regulation of TRAIL-induced pro-apoptotic pathways.
Ad.IkB expresses high levels of IkB(A32/36) in colonic cancer cells and sensitizes to TNF-a In order to extend the ®ndings in the stable transfectants studies, we transduced HCT116, HRT18 and SW480 with a replication defective adenovirus containing a CMV promoter/enhancer IkB(A32/36) expression cassette . Western analysis (Figure 6A) shows the level of expression of mutant IkB in the three colorectal cancer cells at MOI of 0, 5000 to 10 000 particles/cell. Substantially, higher levels of mutant transgene expression over endogenous can be seen, and expression levels were somewhat higher in HCT116 cells (Figure 6A). Control infections with Ad.LacZ indicated that an MOI of 5000 particles/cell corresponds to a 70% infection rate in all of the colorectal cancer cell lines. EMSA analysis (Figure 6B) shows that adenoviral infection itself (Ad.LacZ) activates NF±kB. In contrast, infection with Ad.IkB blocks the adenoviral-induced NF±kB activation. TNF-a treatment 24 h post-infection further increased the levels of nuclear NF±kB in Ad.LacZ infected cells at 1 h post-TNF-a treatment, while no such increase could be detected in Ad.IkB± transduced cells (Figure 6B). In Ad.IkB-transduced HRT18 and SW480 cells, NF-kB activation was still inhibited at 16 h (Figure 6B) and 48 h (data not shown) post-TNF-a treatment, while in HCT116 cells NF-kB DNA binding was decreasing from its peak at 1 h at Copyright # 2000 John Wiley & Sons, Ltd.
R. M. Zwacka et al.
these time points anyway. However, Ad.IkB transduction reduced these levels even further (Figure 6B). These data show that adenoviral mediated expression of IkB(A32/36) can inhibit sustained TNF-a-induced NF±kB activation. Apoptosis studies on Ad.LacZ and Ad.IkB demonstrated that expression of IkB(A32/36) sensitized HRT18 and SW480, but not HCT116 cells to TNF-a (10 ng/ml) treatment (Figure 6C). The apoptotic indexes of Ad.IkB infected HRT18 and SW480 cells increased more than two-fold as compared to Ad.LacZ transduced cells. Growth curves of Ad.LacZ and Ad.IkB infected HRT18 and SW480 cells showed that IkB expression rendered these cells TNF-a sensitive (Figure 6D) at concentrations as low as 1 ng/ml. These ®ndings re¯ect the high levels of IkB(A32/36) achieved by adenoviral mediated expression. Despite very high IkB(A32/36) expression levels the TNF-a sensitive HCT116 cells could only be slightly further sensitized to TNF-a by infection with Ad.IkB, supporting our ®ndings in the stable transfection experiments.
Adenovirus infection induces sustained NF±kB activation and protects cells from TNF-a effects Interestingly, we discovered that HCT116 cells, normally TNF-a sensitive, became resistant when infected with Ad.LacZ. EMSA studies revealed that nuclear NF±kB levels were elevated at 24 h post-adenovirus infection, indicating that a sustained activation is induced, suf®cient to elicit an anti-apoptotic programme. Though adenovirus-induced NF±kB activation has been shown before in various cell lines and tissues  this ®nding provides further evidence that sustained NF±kB activation is required for protection from TNF-a cytotoxicity. As already discussed, we did not see this sustained NF±kB activation (activation >16 h) in HCT116 following TNF-a treatment (Figure 2A). Hence, an anti-apoptotic programme, comparable to HRT18 and SW480 cell lines, is activated by Ad.LacZ infection, and HCT116 cells are protected from a subsequent challenge with TNF-a. This observation provides further support for the notion that prolonged NF±kB activation protects against TNF-a mediated cytotoxic effects. One might speculate that these data point to a potentially bene®cial side effect of an IkB(A32/36) gene therapy approach that employs adenoviral mediated gene transfer for colorectal liver metastases. Using IkB(A32/36) driven by a colorectal cancer speci®c promoter in a recombinant adenoviral vector might allow two levels of speci®city. Promoter speci®city would target expression to tumour cells whilst neighbouring hepatocytes may be selectively protected from TNF-a effects by adenoviral mediated induction of a prolonged NF±kB response. Clearly, this potential synergistic bene®t in the clinical setting remains speculative, but the effect does merit further study. J Gene Med 2000; 2: 334±343.
NF-kB Activation Kinetics in TNF-a Resistance
Discussion Hepatic metastases from colorectal cancer remain dif®cult to treat with conventional chemotherapy. In these studies, we set out to investigate possible determinants of TNF-a response in colorectal cancer cells as such knowledge could inform novel gene therapy cytokine strategies. Some encouraging results have been reported in the treatment of hepatic metastases using TNF-a therapy  although the response rates were usually limited in these studies. A strategy that could identify the subset of TNF-a responsive tumours would be clinically useful, while an approach involving sensitization of resistant tumours could greatly enhance the utility of TNF-a based therapies. The data we present here suggest that the kinetics of NF±kB activation may serve as a marker of tumour TNF-a responsiveness. Sustained NF±kB activation over 16 h appears to block cytotoxic effects, while transient activation results in a predominantly pro-apoptotic response. Sensitivity to TRAIL, which corresponded to transient NF±kB activation, further highlighted a potential link between the duration of NF±kB activation and TNF-a resistance. This observation has potential use as a biological marker of pre-treatment tumour biopsies, in order to determine the TNF-a and/or TRAIL sensitivity of individual tumours. Thus, TNF-a/TRAIL therapy could be restricted to patients whose tumours are most likely to respond. It could also aid in identifying those patients who might bene®t from a gene therapy approach involving tumour sensitization approaches to TNF-a or, indeed, to other drug treatments such as CPT-11. It will be of considerable interest to elucidate the pathways that determine whether TNF-a induces a transient or prolonged NF±kB response in a given
Figure 6. (A) Western blot analysis of HCT116, HRT18 and SW480 cells that were transduced with an adenoviral vector expressing the IkB-SR protein (Ad.IkB) at different MOIs: 1000, 5000 and 10 000 particles/cell. Increasing levels of transgenic IkB(A32/36) can be detected and are indicated by Copyright # 2000 John Wiley & Sons, Ltd.
an arrow labelled tg, while the endogenous (eg) form of IkB has a higher gel mobility. (B) An EMSA analysis shows that Ad.IkB (5000 particles/cell) can inhibit the TNF-a induced NF±kB activation at 1 and 16 h post-treatment while Ad.LacZ infected cells show no such effect. This effect is similar in all of the three cell lines. (C) Programmed cell death was measured in HCT116, HRT18 and SW480 cells that were transduced with Ad.IkB and control Ad.LacZ, respectively, at an MOI of 5000 particles/cell and treated with 10 ng/ml TNF-a. The apoptotic index is signi®cantly increased in HRT18 and SW480 cells that were infected with Ad.IkB as compared to those cells treated with control vector Ad.LacZ. In HCT116 cells the expression of adenoviral mediated IkB(A32/36) did not give rise to a signi®cant increase in apoptosis. The apoptotic indexes depict the mean of three independent experiments (tSEM), and the differences between Ad.LacZ and Ad.IkB infected HRT18 and SW480 cells are signi®cant with p