Combination Analysis of Activator Protein-1 Family Members, Sp1 and an Activator Protein-2α-Related Factor Binding to Different Regions of the Urokinase Receptor Gene in Resected Colorectal Cancers

Patients and Methods

Patients and tumors. One hundred and three prospectively followed patients underwent surgery for colorectal cancer between August 1996 and October 2000. Seventy one of these patients were part of the series published by Schewe et al. (24). Patient and tumor characteristics are given in Table 1. Follow-up [physical exam, ultrasound, chest X-ray, and tumor markers (carcinoembryonic antigen, Ca 19-9)] was carried out 6, 12, 18, and 24 months after surgery and in 1-year intervals thereafter. Tumor recurrence was diagnosed by biopsy or explorative surgery, if possible. Causes of death were evaluated clinically. The study was approved by the institutional ethical standards committee and done with the patients' informed consent. Tissue specimens from tumor and corresponding normal mucosa distant from the tumor site were collected after macroscopic verification by a pathologist, and frozen immediately in liquid nitrogen.

Preparation of nuclear extracts. These were essentially done as described in our previous publication (24). Resected tissue specimens were snap-frozen in liquid nitrogen, mechanically pulverized, resuspended in PBS buffer, centrifuged, and processed for nuclear extraction as described in ref. (24). Protein concentrations were determined using the bicinchoninic acid kit (Pierce, Rockford, IL).

Electrophoretic mobility shift assay and supershift analysis. Our previously established gelshift and supershift method (24) was used, and was found to give clear and reproducible transcription factor binding results in resected tissue material. Nuclear extracts (25 μg) were incubated in a buffer [25 mmol/L HEPES (pH 7.9), 0.5 mmol/L EDTA, 0.5 mmol/L DTT, 0.05 mol/L NaCl, 4% (v/v) glycerol] with 20,000 cpm of the [³²P]ATP end-labeled (phage T4 polynucleotide kinase, ICN***) oligonucleotide containing region −190/−171 (region 1) of the u-PAR promoter (5′-GTGATCACAACTCCATGAGTCAGGGCCGAG-3′; ref. 25) and region −152/−135 (region 2; 5′-CCAGCCGGCCGCGCCCCGGGAAGGGA-3′; ref. 24), respectively, for 10 minutes in the absence or presence of a 200-fold excess of unlabeled oligonucleotide to show specificity of the binding (room temperature). A total of 0.5 μg of poly(deoxyinosinic-deoxycytidylic acid) was present in each reaction to block nonspecific binding. The reactions were subjected to gel electrophoresis on a 5% polyacrylamide gel containing 5% glycerol and exposed to Kodak MR film for 24 hours (−80°C, intensifying screen). For AP-1 supershift analysis, 1 μg of antibody (anti-cFos, anti-FosB, anti-Fra1, anti-Fra2, anti-p-c-Jun, anti-JunD, anti-JunB, and rabbit IgG control, Santa Cruz sc-253, sc-48, sc-605, sc-604, sc-822, sc-74, sc-46, sc-2027) was added to the reactions 10 minutes after nuclear extracts and oligonucleotide had been incubated. For region −152/−135, antibodies were used as published previously (24). Supershift reactions were incubated for 60 minutes at 4°C and electrophoresed at 4°C to ensure complex stability. Binding of the AP-2-related factor was competed with an AP-2α consensus oligonucleotide as published previously (24). As negative controls, a lane without nuclear extract and a second lane without poly(deoxyinosinic-deoxycytidylic acid) and nuclear extract were processed in each electrophoretic mobility shift assay (EMSA).

Quantification of EMSA results for statistical analysis with u-PAR-ELISA. Binding activities of transcription factors were quantified exactly as described previously (24) using ScanPack 3.0 (Whatman-Biometra, Goettingen, Germany) which determined the area under the densitometric curve in each tissue sample for AP-1, AP-2α, and Sp bands. Binding was expressed as a ratio between the densitometric activities of the respective bands in the tissues, and activities from equal amounts of a standardized RKO nuclear extract, which was processed in each EMSA in parallel. RKO binding activity was thereby regarded as 100%.

Qualitative presence of transcription factor binding was defined as soon as a band could be detected by densitometry above the background of the EMSA. The term high-binding was defined for a transcription factor as soon as the binding intensity of its corresponding band was ≥40-fold density of the background of the individual shift (examples in ref. 24; Fig. 1). In contrast, the term low binding was defined by the intensity of the transcription factor band ≤10-fold of the background of the individual shift, a density corresponding to a band just visible as a specific band to the human eye (examples in ref. 24; Fig. 1). The term tumor-specific binding was attributed if high transcription factor binding according to the definition above was seen in the tumor in parallel to an absent or low binding (according to the definition above) in the corresponding normal tissue, as already described in Schewe et al. (24).

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

A, EMSA using an oligonucleotide corresponding to region −190/−171 of the u-PAR promoter, and nuclear extracts of tumor (T) and corresponding normal tissues (N) of colorectal cancer patients (examples). In cases 1 and 3, AP-1 binding can be detected in the tumor tissue only. In case 2, AP-1 binding is found in the tumor and in the normal tissue, whereas in case 4, no binding is observed, neither in the tumor nor in the normal tissue. Further explanations are given in the text. B, identification of AP-1 family members bound to region −190/−171 of the u-PAR promoter by supershift analysis in resected tumor (T) or normal (N) tissue of patients (examples) with colorectal cancer. Antibody binding results in complexes migrating more slowly as compared with the original complex (arrows). C, u-PAR mRNA amounts are associated with u-PAR protein levels. Colorectal cancer (T) and normal tissues (N) of 19 patients investigated for u-PAR mRNA in addition to u-PAR ELISA. U-PAR RT-PCR was done as described in Patients and Methods (top), and normalized for glyceraldehyde-3-phosphate dehydrogenase as a loading control (second panel, ratios given in the graph below the panel). Corresponding u-PAR ELISA values can be seen in the graph (bottom).

ELISA to determine u-PAR protein amounts, u-PAR reverse transcriptase-PCR. u-PAR protein in the tissue cytosols was assayed using Imubind-u-PAR-ELISA kit (American Diagnostica, Greenwich, CT). High and low u-PAR expression was defined as an amount greater or lower than the median as in ref. (24).

In a set of 19 patients, u-PAR mRNA was measured in addition, applying a reverse transcriptase-PCR standard protocol and Superscript II reverse transcriptase following Trizol RNA extraction (Invitrogen, Carlsbad, CA), using the primers u-PAR sense 5′-ACCAACAAG ACC CTG AGC-3′, u-PAR antisense 5′-CAT CCA GGC ACT GTT CTT C-3′, G3PDH sense 5′-GTC TTC ACC ACC ATG GAG AA-3′, G3PDH antisense 5′ ATC CAC AGT CTT CTG GGT GG-3′ (positive control).

Statistical analysis. Analysis was done using StatView 5.0 for Macintosh/Windows. The correlation between u-PAR amounts and transcription factor binding in tissues was determined by linear regression. For correlation analysis, transcription factor–binding was dichotomized (presence/absence, see above), and χ² tests (Bonferroni corrected) were done, considering the variables as follows: Dukes stage as A-D, Laurén's classification as intestinal versus diffuse/mixed, lymphangiosis carcinomatosa as present versus absent, pT as pT_1-2 versus pT_3-4, pN as pN₀ versus pN_1-3, and M, Unio Internationale Contra Cancrum, G, and Borrmann stages according to the established groups. Differences between tumor and normal tissues in u-PAR content were calculated with the Wilcoxon test for independent values. P < 0.05 was defined as significant. Group-oriented life-table curves were calculated by Kaplan-Meier analysis and confirmed by Mantel-Cox log-rank statistics.

Results

Patients. One hundred and three patients (Table 1) with primary colorectal cancer were analyzed for transcription factor binding to region −190/−171 of the u-PAR promoter in their primary tumors and corresponding normal mucosa. A subgroup of 71 patients (subseries 2) were also analyzed for transcription factor–binding to region −152/−135 of the u-PAR promotor (24), so that a combination analysis of transcription factor–binding to this region and the AP-1-binding region −190/−171 could be done.

The median time of follow-up for all patients is presently 30 months (range, 1-77). During this time, 55 patients died (12 deaths were not tumor-related). In subseries 2 (n = 71), there were 23 tumor-related, and 6 non–tumor-related deaths.

In the curatively resected (R0) patients (n = 86, n = 56 in subseries 2), 22 recurrences occurred (16 in subseries 2), 2 (both in subseries 2) of them being locoregional, 6 (4 in subseries 2) liver metastasis, 14 (10 in subseries 2) distant metastasis (lung, bone, brain, generalized).

Qualitative results of EMSA and supershift analysis. Our previously established gelshift and supershift method (24) was used, and was found to give clear and reproducible transcription factor binding results in resected tissue material. All EMSA analyses were done using a ³²P end-labeled oligonucleotide corresponding to region −190/−171 of the u-PAR promoter, and normalized equal protein amounts of normal and tumor tissue extracts as well as equal amounts of a standardized nuclear extract of RKO colon cancer cells as a reference. Lengyel et al. (25) showed the binding of transcription factors of the AP-1 family to region −190/−171 of the u-PAR promotor in RKO cell lines using EMSA technology in vitro. This binding was represented by a large complex of slower mobility in a similar appearance as seen in our gelshift results in tissues (25), and supershift experiments could identify Fra-1, JunD, c-Jun, and c-Fos within this complex (25).

Using our modified gelshift protocol for in vivo resected tissues (ref. 24, see Patients and Methods), the binding of AP-1 family proteins was evident in patients' tumor and normal tissues at varying intensities (Fig. 1A, lanes 7, 9, 11, and 15), as slower mobility complexes analogous to the complexes seen by Lengyel et al. (25). The identities of the AP-1 members bound to region −190/−171 were investigated applying specific antibodies against c-Jun, JunD, JunB, c-Fos, Fra-1, and Fra-2. All antibodies were directed against a non–DNA-binding domain of the individual transcription factor and resulted in a supershift of the AP-1 complexes, as shown previously (25). Supershifts in vivo are shown in Fig. 1B: c-Fos in lane 7, Fra-1 in lane 9, c-Jun in lane 11, and JunD in lane 13. Transcription factor binding was specific, as a 200-fold excess of the nonlabeled oligonucleotide corresponding to region −190/−171 resulted in an elimination of the bands (Fig. 1A, lanes 4, 6, 8, 10, 12, 14, 16, 18, and 20). As shown in our previous report for Sp1, we did not observe an association of AP-1 binding in gelshift analysis with AP-1 protein amounts detected in Western blotting (data not shown).

As we have shown previously (22), it is possible to reproducibly show the binding of Sp1, Sp3, and an AP-2-related factor to region −152/−135 of the u-PAR promoter in vivo (24). However, because binding of Sp3 can be seen less consistently in resected tissues than in Sp1- and AP-2-related proteins (24), we focus on the latter two transcription factors for combination analysis of the present study.

Distribution of transcription factor binding in resected tumors and normal tissue. An overview on transcription factor–binding to region −190/−171 of the u-PAR promoter in the resected tissues analyzed is given in Table 2A. In all colorectal cancer patients (n = 103), high binding of AP-1 to region −190/−171 of the u-PAR promoter was detected in the tumor tissue in 69.9% of cases, the subgroup additionally being analyzed for binding to u-PAR promoter region −152/−135 (n = 71) demonstrating high AP-1-binding in 76.1% of cases.

More importantly, a high transcription factor binding in the resected tumors in contrast to low or absent binding in corresponding normal mucosa (tumor-specific binding) was observed for AP-1 in 39.8% of all patients (n = 103), and in 45.1% of the 71 patients additionally investigated for binding to region −152/−135. These data indicate that the potential transactivation of the invasion-related gene u-PAR by AP-1 is tumor-specific in ∼40% to 45% of colorectal cancer patients, whereas in our previous work, a tumor specificity of almost 60% was seen for Sp1/AP-2–related factors binding to region −152/−135 (24).

In a subgroup of 26 patients with colorectal cancer showing AP-1 complexes in nuclear extracts of their tumor tissues, we did additional supershift analyses to identify binding frequencies of specific AP-1 family members in tumors and normal tissues (Table 2B). We identified c-Fos (84.6% of cases analyzed), c-Jun (92.3%), JunD (61.5%), and Fra-1 (3.8%) as the AP-1 family members bound in resected tumors. We observed tumor-specific binding for c-Fos in 57.7%, for c-Jun in 50.0%, for JunD in 38.5%, and for Fra-1 in 3.8% of these 26 cases. There was no Fra-1-binding in normal tissues and no binding of Fra-2, Jun B, and Fos B. The results indicate that, in vivo, c-Fos, c-Jun and JunD are the main AP-1 members bound to region −190/−171 of the u-PAR promoter in resected tumor tissues, c-Jun and c-Fos showing the highest percentages of tumor-specific binding.

Distribution of transcription factor binding to both regions −152/−135 and −190/−171 in a subgroup analysis. Next, we investigated binding to both u-PAR promoter motifs −190/−171 and −152/−135 in a subgroup of 71 colorectal cancers (subseries 2). These tissues had been additionally analyzed for binding of Sp1 and an AP-2α-related factor to region −152/−135 of the u-PAR promoter by our group (24). A detailed overview of combinatorial analysis of transcription factor binding to both promoter regions is given in Table 3. Tumor-specific binding of AP-2α (not considering a simultaneous binding of other transcription factors) could be observed for 53.5% of the patients with colorectal cancer, and a tumor-specific binding of Sp1 was seen in 50.7%. This was in the range of tumor-specific binding percentages published previously, corroborating these results (24). Tumor-specific binding of AP-1 complexes to promoter region −190/−171 was seen less often in 32 (45.1%) of 71 patients with colorectal cancer, as already discussed above (Table 3, lane 1).

More importantly, we now analyzed how often the binding of exclusively one transcription factor was found in these 71 patients, as opposed to the binding of a combination of factors. High binding in tumor tissues of one individual transcription factor complex (AP-1, Sp-1 or the AP-2α-related factor) was exclusively found in 0% to 4.2% of the colorectal cancer patients only, whereas in all other cases, the binding of at least two of these transcription factor complexes could be seen. High and exclusive binding of a combination of two factors could be detected in 0% to 15.5% of the patients, depending on the combination of transcription factors analyzed (Table 3, third row). The highest percentage of high transcription factor binding in tumor tissues, however, was found for the combination of all three transcription factors (AP-1, Sp1, and the AP-2α-related factor bound simultaneously), which was seen in 44 of 71 (62.0%) colorectal cancer patients (Table 3, third row). Moreover, the presence of binding of AP-1 and Sp1, AP-1 and the AP-2α-related factor, as well as Sp1 and the AP-2α-related factor in the tumors correlated significantly with each other (P < 0.001, linear regression analysis).

Finally, subgroups with bound transcription factor combinations were analyzed for the criterion of tumor-specificity (Table 3, second row). The combination of binding of the AP-2-like protein and Sp1 reached the highest level of tumor-specificity (45.1%; Table 3, second row), as opposed to combinations involving AP-1. A tumor-specific simultaneous binding of AP-1, Sp1, and the AP-2α-related factor together was observed in 25.3% of colorectal cancer patients.

The results of these combination analyses, showing highly significant correlations between the binding of all three transcription factors to both motifs, together with the fact that the majority of cases showed binding of all three or at least two of the transcription factors, support our in vitro findings of a synergism between both elements in u-PAR regulation (22). However, a synergism between AP-1 and the AP-2/Sp1 motif might not necessarily be as tumor-specific in colorectal cancers, because tumor-specificity for transcription factor combinations involving AP-1 was less than for the combination of Sp1 and the AP-2-like protein.

Correlation of transcription factor binding to regions −190/−171 and −152/−135 of the u-PAR promoter with established tumor characteristics. To assess the association of AP-1 binding to region −190/−171, and of the combination of transcription factor binding to both promoter regions with established clinical tumor variables, χ² analysis (Bonferroni-corrected) was done. In all 103 colorectal patients, there was a significant correlation between the binding of AP-1 in tumor tissues with positive pN stage (lymph node metastasis, P = 0.023) and Dukes/Unio Internationale Contra Cancrum stage (P = 0.038). No correlations were found for AP-1 binding in normal tissues. Analysis of both promoter regions revealed a significant correlation between binding of both Sp1 and the AP-2α-related factor in tumor tissues with lymphangiosis carcinomatosa in 71 patients with colorectal carcinoma (subseries 2, P = 0.003). Considering combinations of transcription factors, a significant correlation between the binding of all three factors together (AP-1, Sp1 and the AP-2α-related factor) in the tumor tissues with advanced pT stages and lymphangiosis carcinomatosa (P = 0.018) was found.

Correlation of transcription factor binding with u-PAR protein amounts in tumor and normal tissue. To corroborate the biological relevance of transactivation of u-PAR gene expression by region −190/−171 in tumor and corresponding normal tissue, and to potentially further support the hypothesis of a potential synergism between regions −190/−171 and −152/−135, correlations between binding of individual transcription factors and u-PAR protein amounts as measured by ELISA were investigated. EMSA results were quantified by densitometric scanning of the transcription factor complexes in relation to the densitometric intensity of the bands resulting from equal amounts of a standardized RKO-nuclear extract. The method was applied previously by others (29) and our group (24) for a relative quantitation of EMSA data.

Median u-PAR protein amounts were 1.0 ng/mg protein (range 0.0-4.6) in tumor tissues, 0.1 ng/mg (range 0.0-1.2) in normal tissues (all 103 colorectal patients), 1.1 ng/mg (range 0.0-4.6) in tumor tissues, and 0.2 ng/mg (range 0.0-1.2) in normal tissues in the 71 patients with colorectal carcinoma analyzed for binding to both promoter motifs, respectively. The differences between tumor and normal tissue in u-PAR content were significant with P < 0.001 (Wilcoxon). In 19 resected tumor and normal colon cancer tissues, we did additional reverse transcriptase-PCR analyses for u-PAR mRNA (Fig. 1C). There was a significant correlation between u-PAR mRNA, and u-PAR protein amounts as measured by ELISA in these tissues (P = 0.010, linear regression), implicating that u-PAR protein measurements reflect the amount of gene transcripts at the mRNA level in resected tissues.

In contrast to the AP-2-like factor and Sp1 bound to region −152/−135, which correlated significantly with u-PAR protein amounts in tumor tissues only (24), AP-1 binding correlated with high u-PAR protein amounts in both tumor and normal tissues (linear regression analysis, Table 4). Our previously published results were confirmed by the present analysis, in which a significant correlation between the binding of Sp1 and the AP-2α-related factor with u-PAR protein amounts was found for tumor tissues only (P < 0.001 in both cases), but not for the corresponding normal mucosa (P > 0.05, Table 4).

These data confirm our previous hypothesis (24) that transcription factors bound to region −152/−135 of the promoter are of special in vivo relevance for u-PAR gene expression in colorectal carcinoma tissue in contrast to normal tissue. They further indicate that the binding of AP-1 family members to region −190/−171 of the u-PAR promoter, correlating significantly with u-PAR protein amounts in tumors and normal mucosa, could be less tumor-specific in terms of regulating u-PAR gene expression.

Next, we analyzed potential correlations of transcription factor combinations with high u-PAR protein amounts (χ² analysis) in the colorectal cancer patients screened for both promoter motifs. Analyzing for tumor-specific binding of combinations of two transcription factors (not considering presence or absence of a third factor), tumor-specific binding of Sp1 and the AP-2α-related factor together correlated significantly with high u-PAR protein amounts (P = 0.004). When analyzing for an exclusive binding of combinations of two transcription factors in tumor tissue (excluding binding of a third factor and not considering tumor-specificity), only the combination of AP-1 and the AP-2α-related factor showed a significant correlation with high u-PAR protein amounts (P = 0.008). For all other transcription factor combinations, no significant association with high u-PAR protein amounts were detected. Again, this finding corroborates the hypothesis of synergism between both promoter regions, specifically between AP-1 and the AP-2-like protein, from data in resected tissues.

Prognostic effect of transcription factor binding. Finally, we attempted a first preliminary prognostic analysis for transcription factor binding to both u-PAR promoter elements. As expected and previously shown for colorectal cancer tissues (17, 30), a high u-PAR protein amount in tumor tissue showed a trend towards a worse prognosis in the complete colorectal cancer group (n = 103), and in subseries 2 (n = 71; Fig. 2A); however, this was not significant (P > 0.05) due to the short median follow-up time in our preliminary analysis. Tumor-specific binding of all three transcription factors Sp1, AP-1, and the AP-2α-related protein together, as compared with combinations of two factors or one transcription factor only, indicated that the strongest trend towards a worse recurrence-free (Fig. 2B) or overall survival (P > 0.05). These findings indicate that, at this early point of follow-up, transcription factor binding does not yet have a statistically significant prognostic effect. However, trends towards a worse overall prognosis and higher recurrences can be seen for the tumor-specific binding of all three transcription factors, in addition to high u-PAR gene expression.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

A, preliminary analysis of prognostic effect of u-PAR gene expression in colorectal cancer patients within subseries 2 (n = 71, Kaplan-Meier analysis), overall survival. Upper curve, u-PAR protein amounts in tumor tissue low (below median); 37 cases, 9 events; mean survival time, 41.6 months; SD, 19.8 months. Lower curve, u-PAR protein amounts high (above median), 34 cases, 13 events; mean survival time, 30.3 months; SD, 15.4 months. B, prognostic trend of transcription factor binding in colorectal cancer patients within subseries 2 (n = 56 curatively resected patients, Kaplan-Meier analysis), recurrence-free survival. Upper curve, no simultaneous binding of AP-1, Sp1, and the AP-2-like protein, 25 cases, 4 events; mean recurrence time, 36.1 months; SD, 34.9 months. Lower curve, binding of AP-1, Sp1, and the AP-2-like protein; 31 cases, 12 events; mean recurrence time, 30.9 months; SD, 28.9 months.

Discussion

This is the first study in which in vivo transcription factor binding to an AP-1 motif (region −190/−171) of the u-PAR promoter, in addition to AP-1 family differentiation in supershift assays, has been analyzed in a large series of patients with resected tumor and normal tissues, and in which, at the same time, a combinatorial analysis of transcription factor binding to this and a further central u-PAR promoter region (−152/−135) was done. In comparison to our previously published article on region −152/−135 (24), to which tumor-specific binding of the AP-2-like protein and Sp1 was seen in 60% of cases, the AP-1-motif −190/−171 seems to be a less tumor-specific regulator in ∼40% of patients. However, when analyzing for combinations of transcription factor bindings, the majority of patients exhibit binding of at least two transcription factors to both promoter motifs, this being paralleled by significant correlations between transcription factor–binding to both motifs, and with high u-PAR protein amounts. Moreover, first preliminary prognostic analyses suggested a trend towards a worse clinical outcome for patients with simultaneous binding of AP-1, the AP-2-like protein, and Sp1. These data provide further support for the hypothesis of a potential synergism between these motifs from resected tumor tissue, which we had implicated in in vitro cell lines a few years ago (22).

AP-1 is a family of leucine-zipper transcription factors regulating proliferation, stress response, tumor cell progression, invasion, and metastasis; besides u-PAR regulating other metastasis-related genes such as plasminogen activator inhibitor-1, urokinase type plasminogen activator, matrix metalloproteinase-3 and -9, and vascular endothelial growth factor (31–33). In particular, c-Jun and c-Fos have been described as the cellular counterparts to viral oncogenic proteins involved in tumor cell progression (34). Therefore, it is not surprising that these two factors were found in the highest percentages to bind, and also to bind tumor-specifically, to u-PAR promoter region −190/−171 in our tumor tissues. In general, the AP-1 members found in our series of resected tumor tissues resembled the same isoforms (c-Jun, JunD, c-Fos, and Fra-1) as found in our previous in vitro studies in cultured colon cancer, also leading to the same supershift phenomena from the upper slow-moving complex described in ref. (25). However, there is one important difference deserving attention. Whereas all four AP-1 family members mentioned were identified to bind to region −190/−171 in cell lines in previous studies, and here in resected gastrointestinal tumor tissues, Fra-1 was not found to bind to this region in the corresponding normal tissues. This is in line with previous publications. For example, Chiappetta et al. (35) found Fra-1 in 100% of 50 thyroid carcinomas, whereas none of 12 normal tissues showed any expression. As a consequence, Fra-1 was suggested to differentiate thyroid malignancy from benign tumors (36). The functional reasons as to why Fra-1 is important exclusively in tumor tissue have mainly been seen in target genes regulated by Fra-1 which are nonactive in normal tissues. Thus, Fra-1 has been found to induce matrix metalloproteinase-1 and -9 in breast cancer (37), maintain the K-Ras-transformed phenotype (38), and induce c-met expression in rat mesothelioma (39). Taken together, our finding corroborates a rather tumor-specific role of this AP-1 isoform.

There have been previous studies investigating AP-1 expression in resected tumor tissues. Assimakopoulou et al. (40) show a correlation of c-Jun/c-Fos expression with WHO grade in astrocytoma tissues. Papachristou et al. (41) detected high protein levels of AP-1 in osteosarcoma tissues. The largest study conducted thus far (42) examined 154 normal, premalignant, and malignant laryngeal tissues. However, all of these studies focused exclusively on protein expression of AP-1 (immunohistochemistry/Western blotting). However, as it is generally accepted (43, 44), the biological relevance of a transcription factor is brought about by its activity at a specific target site, rather than by mere protein expression. Thus, in our previous work, Sp1 binding intensities to u-PAR region −152/−135 in resected tissues was not associated with comparable protein levels as measured by Western blotting (24). It is well known that phosphorylation and other modifications are crucial for binding/transactivating activities of AP-1 (45–48). Thus, studies focusing on measurements of mere protein levels are likely to miss functional AP-1 activities in the tissues investigated. Our present study, to our knowledge, is the first investigating AP-1-binding activity towards a specific promoter motif in resected tissues.

However, the percentage of patients with tumor-specific binding was less for this AP-1 motif (40%), than it had been for binding of the AP-2-like protein/Sp1 to region −152/−135, as reported previously (60%; ref. 24). In addition, a significant correlation of AP-1 binding with high u-PAR protein amounts was seen in both tumor and normal tissues, this being in contrast to the AP-2/Sp1 motif, transcription factor binding to which correlated with u-PAR protein in tumor tissues only (24). Also, the combination of AP-2/Sp1 binding to region −152/−135 exhibited the highest percentage of tumor-specificity in terms of patient numbers (Table 3, lane 2). These data raise the hypothesis that the AP-1 motif might be a less tumor-specific regulator of u-PAR gene expression than the AP-2/Sp1 motif. One reason for this notion could be that the AP-2-like factor bound to region −152/−135 is closely related to, however not identical with, AP-2α, as shown previously (22, 24). This protein, in its experimental appearance contrasting classic AP-2, has so far been described in only a few studies (22, 24, 49). Thus, in contrast to the more ubiquitous appearance of classic transcription factors, it might be a more tissue-specific, or more tumor-specific regulator. In addition, Sp1 bound to the same region −152/−135 mediated specific induction of u-PAR gene expression via Src (23), rather than constitutive expression, suggesting a rather specific function of this motif. On the other hand, the AP-1 motif −190/−171 had been shown to mediate many general signal transduction pathways leading to constitutive u-PAR gene expression, for example, Ras/mitogen-activated protein kinase, and c-Jun-NH₂-kinase pathways (20, 25). This may account for the notion that, in the present study, we observe less tumor tissue–specificity for the AP-1 motif.

Combination analysis most notably showed highly significant correlations between the binding of all three transcription factors to both u-PAR promoter motifs. Furthermore, the majority of cases showed binding of all three or at least two of the transcription factors (88.8%) in tumor tissues. Also, binding of the AP-2-like protein and AP-1 to both u-PAR promoter elements exhibited the most significant correlation with u-PAR protein amounts. This in vivo correlational data, together with a trend of the combination of transcription factors bound to both elements towards a worse prognosis (see below), support our in vitro findings of a synergism between both elements (22) in resected tumor material. Again, however, a synergism between AP-1 and the AP-2/Sp1 motif might not necessarily be tumor-specific in colorectal cancers because tumor specificity for transcription factor combinations involving AP-1 was less than for the combination of Sp1 and the AP-2-like protein.

Certainly, our approach cannot be a functional proof for an in vivo synergism, as it could be done with, for example, promoter reporter assays. However, to our knowledge, it is not yet technically possible to perform similar reporter analyses with resected tissues. Also, we are certainly aware of the fact that, for investigating transcription factor binding to natural DNA-sequences, chromatin immunoprecipitation assays would be a superior approach. However, in contrast to cell lines, chromatin immunoprecipitation assay analysis has not yet been sufficiently done in solid resected tissues to enable routine applications. There has been one recent attempt to perform chromatin immunoprecipitation assay analysis in resected liver, spleen, and kidney tissue (50)—however, it is technically more feasible to achieve high-quality chromatin immunoprecipitation assay results from liver or spleen, than with solid carcinomas, representing very tight and fibrous tissues. Chromatin immunoprecipitation assay methodology suitable for resected solid cancer tissues still needs to be established for a routine application, and this was the main reason for our present study still being conducted with in vivo gelshifting.

Finally, a potential clinical relevance for our present study should be discussed. For AP-1-binding in tumor tissues, a significant correlation was found with advanced pN and Dukes stages. A significant correlation between the binding of all three transcription factors to both promoter elements with lymphangiosis carcinomatosa and advanced pT stage was observed, suggesting that the transactivation of u-PAR gene expression via both promoter elements might support cancer progression, especially via lymphatic spread. Data are supported by first preliminary prognostic analyses, which indicate the combination of binding of AP-1, the AP-2-like protein, and Sp1, as a potential high-risk group being associated with a trend towards a worse clinical prognosis. Certainly, as also seen for patients with high u-PAR amounts, this association is not yet significant, which can be attributed to a median survival time that is too short to be sufficiently indicative for colorectal cancer. Nevertheless, preliminary data hypothesize a clinical relevance for u-PAR regulation by the two promoter motifs. It is possible to screen larger patient series for a predominantly tumor-specific binding of transcription factors, allowing conclusions as to which u-PAR-regulatory pathways may be preferred in the individual patient. This strategy may allow the selection of individualized specific targeting approaches (51), for the individual inhibition of u-PAR-mediated invasion. Similar individualized targeting based on the status of molecular regulators will most certainly affect clinical tumor staging and tumor therapy in the following years.