Identification of AF17 As a Downstream Gene of the β-catenin/T-Cell Factor Pathway and Its Involvement in Colorectal Carcinogenesis¹

APC³ was isolated as a gene responsible for familial adenomatous polyposis of the colon and was subsequently shown to be involved also in sporadic colon tumors(1). In association with β-catenin, APC acts as a negative regulator of the Wnt signaling pathway by modulating cytoplasmic and nuclear levels of β-catenin. APC facilitates the phosphorylation of β-catenin by glycogen synthase kinase 3β. With a scaffold protein, axin/conductin, APC leads to degradation ofβ-catenin via the ubiquitin-proteosome pathway (2, 3, 4, 5). Abnormal accumulation of β-catenin as a consequence of mutation(s) in APC, β-catenin, or AXIN1 genes has been observed in a variety of human cancers including colorectal and hepatocellular carcinomas (6, 7). Accumulated β-catenin interacts with the Tcf/LEF transcription complex, translocates to the nucleus, and transactivates target genes such as c-myc, cyclin D1, matrilysin, WISP, c-jun, fra-1, NBL4, and MDR1 (8, 9). However, the precise mechanisms remain unknown.

Various cytogenetic abnormalities play critical roles in leukemogenesis. For example, translocations involving chromosomal band 11q23, e.g., t(4;11), t(6;11), t(9;11), and t(11;19), are observed in 10% of patients with acute lymphoblastic leukemia and more than 5% of myeloid leukemias (10). The MLL(ALL-1, HRX, TRX) gene, located at the breakpoint on band 11q23, is cleaved by these translocations, and its fusion to specific genes on partner chromosomes results in production of chimeric proteins. The AF17 gene at chromosome 17q21 is a fusion partner of a less frequent translocation of the MLLgene, t(11;17)(q23;q21). The predicted amino acid sequence of AF17 contains three zinc-finger domains at the NH₂ terminus, and a leucine-zipper dimerization motif located 3′ of the fusion point. Although AF17 is thought to function as a transcriptional regulator (10),its role in leukemogenesis remains to be explained.

In this study, we report identification of AF17 as a possible downstream target of the β-catenin-Tcf/LEF transcriptional complex. We also document its involvement in cell-cycle regulation and discuss its possible role in the mechanisms of colorectal carcinogenesis.

Human fibroblasts (NHDF), mouse fibroblasts (NIH3T3), monkey fibroblasts (COS7), and human colon-cancer cell lines SW480 and DLD1 were obtained from the American Type Culture Collection (Rockville,MD). All of the cell lines were grown in appropriate media and maintained at 37°C in a humidified atmosphere with 5%CO₂ (for NHDF, NIH3T3, COS7, and DLD1) or without CO₂ (for SW480). Cancerous tissues and corresponding noncancerous mucosae were excised during surgery from 12 patients with colon cancer after informed consent had been obtained.

Fabrication of the cDNA microarray slides and construction of recombinant adenovirus have been described elsewhere (8, 11). Duplicate sets of cDNA microarray slides containing 9216 cDNA spots in all were used for each analysis of expression profiles to reduce experimental fluctuation. Briefly, SW480 cells were infected with adenovirus expressing APC, Ad-APC, or control virus,Ad-LacZ. Total RNAs were extracted 72 h after infection,mRNA was purified from the samples, and T7-based RNA amplification was carried out. Aliquots (5-μg) of amplified RNA from SW480/Ad-APC and SW480/Ad-LacZ were labeled with Cy5-dCTP and Cy3-dCTP, respectively (Amersham Pharmacia Biotech, Piscataway, NJ). Hybridization, washing, and detection were carried out as described previously (11). Genes were excluded from further investigation when the intensities of both Cy3 and Cy5 were below 250,000 fluorescence units. Those with Cy3/Cy5 signal ratios greater than 2.0 were selected for further evaluation.

We carried out semiquantitative RT-PCR using cDNA reversely transcribed from 0.2 μg of total RNA from each cell line. The PCR exponential phase was determined on 20–32 cycles to allow comparison among cDNAs developed from identical reactions. GAPDH served as an internal control. Primers for human AF17 were H-AF17F(5′-GGAGACCTCTGAGAGCAGC-3′), H-AF17R (5′-GGAGTACTTGTCCTCCTCTG-3′); for human GAPDH, H-GAPDH3F (5′-ACAACAGCCTCAAGATCATCAG-3′),H-GAPDH3R (5′-GGTCCACCACTGACACGTTG-3′); and for mouse homologue of AF17, M-Af17F (5′-GAGGTGCCCACTAGGACAG-3′), M-Af17R(5′-GCACAATTCCAGGCTTGGAG-3′). All of the reactions were carried out in 25-μl volumes and amplified for 3 min at 94°C for initial denaturation, followed by 20–32 cycles at 94°C for 30 s, 58°C for 30 s, and 72°C for 1 min on a GeneAmp PCR system 9700 (PE Applied Biosystems, Foster City, CA).

Real-time quantitative RT-PCR (TaqMan PCR; Perkin-Elmer) was carried out using 7700 Sequence Detector (Perkin-Elmer) according to the manufacturer’s recommendations. Quantification and equalization of the amount of cDNA was achieved by amplifying GAPDH as an internal control (TaqMan GAPDH Control Reagents). Primers and the probe for AF17 were AF17F (5′-TCGCTTGGCAACAACACAAG-3′), AF17R(5′-TGGTCTGGGCAGTGAGGACT-3′), and AF17-Probe(5′-Fam-CAGCAGTAGCAGCAGCAGGCGGA-Tamra-3′).

The entire coding region of AF17 was amplified by RT-PCR with primers AF17-forward (5′-AGGAATTCCATGGGAGTATGAAGGAGATGGTAG-3′) and AF17-reverse (5′-TACTCGAGGATATAGCCTTTTTCCTGGTTGGCTG-3′). The product was digested with EcoRI and XhoI and cloned into the appropriate sites of pcDNA3.1(+) (Invitrogen, Carlsbad,CA) and pFlag-CMV-5a (Sigma Chemical Co., St. Louis, MO).

NHDF cells were transfected with pFlag-CMV-5a/AF17 using FuGENE 6 (Boehringer Mannheim, Mannheim, Germany) according to the manufacturer’s instructions and fixed with PBS containing 4%paraformaldehyde. Fixed cells were incubated with a mouse anti-Flag antibody (Sigma Chemical Co.) and stained by a rhodamine-conjugated antimouse secondary antibody (ICN Biomedicals, Inc., Costa Mesa, CA). After nuclei were stained with 4′,6′-diamidine-2′-phenylindole dihydrochloride (Boehringer Mannheim), the preparations were observed under a Nikon Eclipse E800 fluorescence microscope.

We transiently transfected either pcDNA3.1(+)/AF17, expressing AF17, or empty vector into NIH3T3 cells and treated the cultures with G418 (0.8 mg/ml) for 2 weeks. Cells that survived were fixed with 100% methanol and stained by Giemsa’s solution (Merck, Darmstadt, Germany).

NIH3T3 and DLD1 cells stably expressing AF17 were established by transfecting NIH3T3 and DLD1 cells with pcDNA3.1(+)/AF17 plasmid using FuGENE 6. Control and AF17-expressing NIH3T3 and DLD1 cells were seeded on 6-cm plates (1 × 10⁵cells/plate) and counted with a hemacytometer every day.

To examine whether AF17 has a role in cell-cycle progression, SW480 cells expressing abundant amounts of AF17 were transfected with sense(5′-ATGAAGGAGATGGTAG-3′) or antisense (5′-CTACCATCTCCTTCAT-3′) S-oligonucleotide, the latter being designed to suppress expression of AF17, by Lipofectin (Life Technologies, Inc., Rockville,MD) according to the manufacturer’s recommendations and maintained for an additional 24 h. RT-PCR and FACS were performed on a FACScan flow cytometer using CycleTEST reagents and the manufacturer’s protocol (Becton Dickinson, San Jose, CA). To examine the role of AF17 further, DLD1 cells expressing low levels of AF17 were transfected with pcDNA3.1(+)/AF17, and we selected stable clones (DLD1-AF17-1 and DLD-AF17-2) expressing high levels of AF17. These cells, together with DLD1-vector (mock) cells as controls, were growth-arrested in G₁ phase by incubation with 5 μg/ml aphidicolin(Sigma Chemical Co.) for 36 h and released from G₁ by removal of aphidicolin. FACS was performed 0, 4, 8, and 12 h later. We also treated those cells withγ-irradiation (10 Gy) and analyzed the population at each phase of the cell cycle by FACS 12 h and 24 h after irradiation.

Assessment of statistical differences for Figs. 3 and 4 were determined by Fisher’s Protected Least Significant Difference test and Student’s t test, respectively. P < 0.05 was considered statistically significant. Statistical analyses were performed using StatView software.

We reported earlier that adenovirusmediated gene transfer of wild-type APC reduced the accumulation of nuclearβ-catenin and decreased the transactivational activity of Tcf/LEF signaling in (constitutionally APC-deficient) SW480 cells(8). Among the 9216 cDNAs spotted on our microarray slides, we identified 84 candidate genes (Table 1), the expression of which correlated with accumulation of β-catenin. From among those 84,we focused on AF17 because it was known to be involved in the pathogenesis of acute leukemia. By semiquantitative RT-PCR, we confirmed significant down-regulation of AF17 by infection with Ad-APC as compared with that with Ad-LacZ in SW480 and LoVo cells(Fig. 1 A).

Because accumulation of β-catenin is frequently observed in colorectal cancers as a result of mutations in either APC or β-catenin, we examined expression of AF17 in 12 colorectal cancer tissues and their corresponding normal mucosae by real-time quantitative RT-PCR. These experiments detected increased expression of AF17 in eight (66.7%) of the cancer tissues examined (Fig. 1 B). This result was in line with the view that AF17 is up-regulated in response to activation of the Tcf/LEF transcriptional complex.

Flag-tagged AF17 protein was detected in the nuclei of transfected cells by immunocytochemical staining (Fig. 2). Similar results were obtained when we transfected the plasmid into COS7 and LoVo cells (data not shown).

We performed colony-formation assays after transfecting pcDNA3.1(+)/AF17, expressing wild-type AF17, into NIH3T3 cells. The number of colonies that developed from transfected cells was significantly greater than the numbers achieved on plates of control cells transfected with the mock vector, pcDNA3.1(+) (Fig. 3, A and B). We isolated five clones expressing high levels of AF17 compared with the expression of endogenous homologue of AF17 in mouse (Fig. 3,C) and found that exogenously expressed AF17 significantly accelerated growth of NIH3T3 cells (Fig. 3 D). This growth advantage was closely associated with the expression level of exogenous AF17. These results were confirmed in two independent experiments in triplicate plates.

We performed FACS using SW480 cells transfected with an antisense S-oligonucleotide designed to suppress expression of AF17(Fig. 4,A). A larger proportion of SW480 cells were arrested at the G₂-M phase after this treatment in comparison with cells treated with control S-oligonucleotide (16.2%versus 6.7%; P < 0.05; Fig. 4 B).

To confirm a role of AF17 in G₂-M progression, we cloned DLD1 cells that stably expressed high levels of AF17 (DLD1-AF17 cells) and synchronized them at the G₁ phase by aphidicolin treatment. Overexpression of AF17 significantly shortened the G₂-M transition time as shown in Fig. 4,C (12.9% versus 26.6%; P < 0.05). In addition, 24 h after cells were exposed toγ-irradiation (10 Gy), about 20% more of the DLD1-AF17 cells arrested at the G₂-M phase had progressed to G₀/G₁ than had DLD1-mock cells (43.9% remaining in G₂-M versus64.2%; P < 0.05; Fig. 4 D). These results indicated that AF17 accelerated cell-cycle progression by promoting G₂-M transition. All of the results were confirmed in at least two independent experiments in triplicate plates.

To identify downstream genes of the β-catenin/Tcf complex, we compared expression profiles between SW480/Ad-APC cells and SW480/Ad-LacZ cells using cDNA microarray slides prepared in our laboratory. An approximately 3.17-fold higher level of expression of cyclin D1, a known target of this complex (12),was observed in SW480/Ad-LacZ cells compared with SW480/Ad-APC cells. Expression of AF17 was also down-regulated in SW480 cells in response to a decrease of Tcf/LEF-dependent transcription. Expression of AF17 is frequently elevated in colorectal cancer tissues where β-catenin is augmented in the cytoplasm or nucleus(2). Therefore, AF17 appears to be a target of the β-catenin/Tcf complex and likely to be involved in colorectal carcinogenesis.

Because molecules of the Tcf/LEF family interact with the consensus sequence 5′-CTTTGWW-3′, we searched for this binding motif in the 5′flanking region of AF17 and found one located about 523–529 bp upstream of the transcription-initiation site. Although we carried out reporter assays using plasmids containing a DNA fragment encompassing the motif and various deletion mutants of it, we were unable to detect any significant difference in the reporter activities among them. Thus, the putative binding element did not affect the elevated transcription of AF17, a result that suggested two possibilities: either AF17 may be regulated only indirectly by theβ-catenin-Tcf complex or β-catenin may regulate AF17 through other binding motifs coupled with a different transcriptional factor. The latter notion is consistent with a recent report (13) that the WISP1 gene, a known target gene of β-catenin, is transactivated by cyclic AMP-responsive element binding protein through its binding site but not through putative Tcf/LEF-binding elements.

The AF17 gene was initially isolated as a fusion partner of the MLL gene in t(11;17)(q23;q21) translocations present in some acute myeloid leukemias (10). Although chromosomal translocations are usually associated with overexpression or activation of oncogenes, Prasad et al. (10)proposed a model in which MLL rearrangements would result in loss of function of that gene because most of the partner genes encode unrelated proteins except for similarities between AF9 and ENL and between AF10 and AF17(14, 15). However, reciprocal translocations can produce two chimeric proteins. Thus, in addition to inactivation of or interference with MLL, t(11;17) translocations might confer oncogenic activity through abnormal activation of AF17 and produce a malignant phenotype in leukemic cells. AF10 bears significant homology to AF17 within their respective cysteine-rich domains at the NH₂ termini and leucine zipper domains toward the COOH termini, although they diverge outside those regions(15). The first part of the cysteine-rich region in each case contains conserved zinc-finger domains known as LAP/PHD-finger. The remainder contains a cluster of 12 conserved cysteines and histidines. These cysteine-rich domains show similarity with part of BR140, a bromodomain- and PHD finger-containing protein that is homologous to the TAF250 subunit of transcription factor TFIID. Therefore, AF17 and AF10 may both function as transcriptional repressors or activators (16).

Our experiments have supported a view that AF17, like c-myc and cyclin D1, is involved in cell-cycle progression and is regulated byβ-catenin (12, 17). Although the mechanisms by which AF17 regulates the cell cycle are not clear at present, accelerated progression of the G₂-M boundary could result from abrogation of a checkpoint. If that is the case, cells that overexpress AF17 may accumulate genetic alterations in addition to conferring a growth advantage. Additional investigations of its functions and isolation of its target molecules will help to clarify the role of AF17 in colorectal carcinogenesis. Such studies may also provide clues for identifying predictive and prognostic markers for diagnosis and for developing more effective therapeutic strategies for specific cancers.