and taxol-mediated cell death via PI3K/AKT - Nature

Oncogene (2006) 25, 3565–3575

& 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc

ORIGINAL ARTICLE

Evi1 is a survival factor which conveys resistance to both TGFb- and taxol-mediated cell death via PI3K/AKT Y Liu1, L Chen, TC Ko2, AP Fields1 and EA Thompson1 1 Department of Cancer Biology, Mayo Clinic Comprehensive Cancer Center, Jacksonville, FL, USA and 2Department of Surgery, University of Texas Medical Branch, Galveston, TX, USA

In hematopoietic cells the transforming potential of the ecotropic viral integration site 1 (Evi1) oncogene is thought to be dependent upon the ability to inhibit TGFb signaling. Although Evi1 has recently been implicated in certain epithelial cancers, the effects of Evi1 on transformation and TGFb signaling in epithelial cells are not completely understood. Herein, we have determined the effects of Evi1 on TGFb signaling in intestinal epithelial cells. Stable expression of Evi1 in non-transformed intestinal epithelial cells inhibited induction of some Smad3-dependent TGFb target genes, such as PAI1. However, TGFb-mediated induction of cellular adhesion signaling components such as integrin1 and paxillin was not inhibited by Evi1; nor did Evi1 inhibit TGFb-mediated epithelial to mesenchymal transition. Likewise, Evi1 did not inhibit TGFb-mediated downregulation of cyclin D1 or block TGFb-mediated growth inhibition. However, Evi1 did inhibit TGFb-mediated apoptosis by a process that involves phosphoinositide-3-kinase (PI3K) and its downstream effector AKT. The ability of Evi1 to suppress apoptosis is not restricted to TGFb-mediated cell death, since Evi1 also protects intestinal epithelial cells from taxol-mediated apoptosis. Evi1 is overexpressed in some human colon cancer cell lines, and overexpression is associated with amplification of the Evi1 gene. Knockdown of Evi1 by siRNA inhibited AKT phosphorylation in HT-29 human colon cancer cells and increased their sensitivity to taxol-mediated apoptosis. These data indicate that Evi1 functions as a survival gene in intestinal epithelial cells and colon cancer cells, activating PI3K/ AKT and conveying resistance to both physiological and therapeutic apoptotic stimuli. Oncogene (2006) 25, 3565–3575. doi:10.1038/sj.onc.1209403; published online 6 February 2006 Keywords: apoptosis; carcinogenesis; colon cancer; cell signaling; chemotherapy

Correspondence: Dr EA Thompson or Dr AP Fields, Department of Cancer Biology, Mayo Clinic, 4500 San Pablo Road, Griffin Cancer Research Building, RM310, Jacksonville, FL 32224, USA. E-mails: [email protected] or fi[email protected] Received 23 September 2005; revised 1 December 2005; accepted 14 December 2005; published online 6 February 2006

Introduction Retroviral activation of the ecotropic viral integration site 1 (Evi1) results in leukemia in mice (Morishita et al., 1988). Evi1 activation has been implicated in hematopoietic malignancies in humans (Bartholomew and Ihle, 1991; Ogawa et al., 1996), and the oncogenic role of Evi1 has been studied extensively in myeloid and lymphoid leukemia. Evi1 is located on human chromosome 3q26, and the 3q25–27 region is amplified in cancer of the cervix (Sugita et al., 2000), ovary (Sonoda et al., 1997), lung (Brass et al., 1996; Racz et al., 1999; Imoto et al., 2001), head and neck (Bergamo et al., 2000), and prostate (Sattler et al., 2000). The relationship between 3q26 amplification and Evi1 expression has not been systematically studied, although it is known that Evi1 is overexpressed in some ovarian cancers (Brooks et al., 1996). These data suggest that Evi1 may play a role in the initiation and/or progression of solid tumors, as well as hematopoietic malignancies. Evi1 is a transcriptional repressor (Tanaka et al., 1994; Hirai, 1999; Mochizuki et al., 2000; Palmer et al., 2001; Wakefield and Roberts, 2002), which inhibits TGFb signaling by binding Smad3 and recruiting corepressors of the CtBP family (Kurokawa et al., 1998; Izutsu et al., 2001; Alliston et al., 2005). CtBP recruits histone deacylases, and thereby converts Smad3 from a TGFb-responsive transcriptional activator to a transcriptional repressor (Izutsu et al., 2001; Alliston et al., 2005). The ability of Evi1 to transform hematopoietic cells may result, at least in part, from the ability to inhibit Smad3-mediated transactivation of gene expression, although Evi1 fusion proteins and alternatively spliced forms Evi1 may have different effects on TGFb signaling (Sood et al., 1999). TGFb is an important regulator of proliferation, migration, differentiation, and death of epithelial cells, and loss of TGFb-mediated growth control is thought to be an obligatory step during colon cancer progression (Roman et al., 2001). Loss of TGFb responsiveness during colon carcinogenesis has been associated with genetic and epigenetic inactivation of the TGFb type II receptor (Markowitz et al., 1995; Parsons et al., 1995), Smad mutations (Riggins et al., 1997), oncogenic activation of K-Ras (Higashidani et al., 2003), and induction of protein kinase C-beta II (Murray et al., 2002). However, the possible role of Evi1 in TGFb

Evi1 is a survival factor in intestinal epithelial cells Y Liu et al

3566

a 0.016 3TP/luc Relative Light Units

signaling in colon cancer is poorly understood. In this report, we assessed the role of Evi1 in TGFb-mediated signaling in non-transformed intestinal epithelial cells. Our data demonstrate that Evi1 inhibits some, but not all TGFb responses. The most significant effect of Evi1 appears to be suppression of TGFb-mediated apoptosis. Our results also reveal that Evi1 functions as a survival gene in colon cancer cells and plays a key role in resistance to chemotherapeutic drugs such as taxol.

Control TGFβ

0.014 0.012 0.010 0.008 0.006 0.004 0.002 0.000 RK5

Results

b

Oncogene

RIE/pBabe

RK5-dnSmad3

RIE/Evi1 Evi1 TBRII Actin phosphoSmad2 Smad2 Smad3

-

+

-

+

TGFβ

c 100 PAI1 mRNA Abundance

Evi1 inhibits transactivation of TGFb target genes in RIE cells Evi1 is known to inhibit Smad3-dependent transactivation in hematopoietic cells (Kurokawa et al., 1998). Our first objective was to confirm this response in nontransformed intestinal epithelial cells. RIE-1 cells were transiently cotransfected with an Evi1 expression vector (pRK5/Evi1) plus the 3TP/luc TGFb reporter construct (Wrana et al., 1994). As shown in Figure 1a, RIE-1 cells exhibited a robust transcriptional response to TGFb when the 3TP/luc reporter was cotransfected with an ‘empty’ pRK5 expression vector. Cotransfection of the pRK5/Evi1 expression vector profoundly inhibited TGFb-dependent transactivation of 3TP/luc in the transient transfection assay. The magnitude of inhibition (75%) was similar to that observed when TGFbmediated transcriptional activity was blocked by expression of dominant-negative Smad3 (89%), indicating that Evi1 attenuates TGFb-mediated, Smad-dependent promoter transactivation in intestinal epithelial cells. Interpretation of transient transfection data is often confounded by the fact that supra-physiological levels of expression of putative transcription factors can be achieved, leading to squelching artifacts that may not prevail under more physiological conditions. To circumvent these potential problems, we constructed a cell line that stably expresses Evi1. As a control, we constructed an RIE-1 derivative that was stably transformed to puromycin resistance with the pBabe/ puro-3 retrovirus (RIE/pBabe). As shown in Figure 1b, Evi1 protein was not detectable in RIE/pBabe but was stably expressed in RIE/Evi1 cells. TGFb is known to downregulate the TGFb type II receptor (TBRII), as illustrated by a decrease in TBRII expression in TGFbtreated RIE/pBabe cells, relative to control RIE/pBabe cells (Figure 1b). Evi1 expression had no effect on ligand-mediated downregulation of TBRII, indicating that the early steps in activation of TGFb signaling are intact in RIE/Evi1 cells. This conclusion was confirmed by the observation that TGFb-dependent phosphorylation of Smad2 was not inhibited by Evi1 (Figure 1b). Furthermore, Evi1 had no effect on Smad2 or Smad3 expression in RIE cells. Although RIE/Evi1 cells retain the ability to activate the TGFb receptor complex and phosphorylate Smads in response to TGFb, these cells are defective in activation of classical Smad3-responsive genes, such as

RK5-Evi1

Control TGFβ

80 60 40 20 0 RIE/pBabe

RIE/Evi1

RIE/dnSmad3

Figure 1 Evi1 inhibits TGFb-mediated transactivation in RIE cells. (a) RIE1 cells were transiently transfected with an ‘empty’ pRK5 expression vector or pRK5 that expresses Evi1 or Smad3DSSVS (dnSmad3) along with the 3TP/luc reporter gene and an SV40 Renilla luciferase internal standard. Cells were treated with vehicle (Control, black bars) or 3 ng/ml TGF-b1 (TGFb, gray bars) for 24 h. Firefly luciferase was measured and normalized to Renilla luciferase activity. The bars represent the mean and s.d. of three replicate dishes. (b) Extracts were prepared from RIE/pBabe or RIE/Evi1 cells treated with vehicle or 3 ng/ml TGF-b1 for 24 h. Western blotting was carried out using specific antibodies. (c) QPCR was used to measure expression of PAI1 mRNA in RIE/ pBabe, RIE/Evi1 and RIE/dnSmad3 cells which had been treated with vehicle or 3 ng/ml TGF-b1 for 24 h. RNA abundance was normalized to GAPDH and calibrated to PAI1 mRNA expression in untreated RIE/pBabe cells, which was set to 1.0.

PAI1 (Figure 1c), demonstrating that this oncogene blocks TGFb-mediated induction of endogenous genes, as well as transiently transfected reporters. The magnitude of inhibition by Evi1 (93%) was comparable to that observed in RIE-1/dnSmad3 cells (97% in Figure 1c). Having demonstrated that Evi1 inhibits TGFb-mediated


3567

Evi1 does not block TGFb-mediated growth inhibition in RIE cells Inhibition of epithelial cell proliferation is thought to be one of the principal physiological roles of TGFb, and rat intestinal epithelial cells such as RIE-1 and IEC6 are exquisitely sensitive to TGFb-mediated growth inhibition (Ko et al., 1994). In such cells, TGFb-mediated

Integrinα1 (re.GAPDH)

6 Control TGFβ

5 4 3 2 1 0 RIE/pBabe

b RIE/pBabe

RIE/Evi1

RIE/Evi1

RIE/dnSmad3

RIE/dnS3 Paxillin pPaxillin Actin

-

+

c

-

-

+

TGFβ

TGFβ

RIE/pBabe

Control

+

RIE/Evi1

Evi1 has no effect on TGFb-mediated EMT TGFb regulates cell motility during organogenesis (Hay, 1995 ), and it is believed that regulation of cell motility is involved in tumor promotion by TGFb (Akhurst and Derynck, 2001; Wakefield and Roberts, 2002). Changes in motility of epithelial cells are associated with a characteristic morphological transition, known as EMT, which is accompanied by changes in integrin/focal adhesion kinase signaling, as well as changes in the actin cytoskeleton. In RIE-1 cells, one of the earliest manifestations of EMT is induction of integrina1 mRNA. As shown in Figure 2a, integrina1 mRNA abundance increased four- to five-fold within 8 h after addition of TGFb to RIE/pBabe cells. The induction of integrina1 mRNA in TGFb-treated RIE/Evi1 cells was comparable to that observed in RIE/pBabe cells. These results indicate that the early effects of TGFb on integrin signaling are intact in RIE/Evi1 cells. During EMT, TGFb induces changes in focal adhesion activity. As shown in Figure 2b, TGFb induces the focal adhesion protein paxillin and stimulates tyrosine phosphorylation of paxillin by focal adhesion kinase in RIE/pBabe cells. Paxillin was induced by TGFb in RIE/Evi1 cells, and TGFb promoted tyrosine phosphorylation of paxillin in RIE/Evi1 cells. Dominant-negative Smad3 (dnSmad3) blocks almost all known TGFb signaling events, and therefore was used as a negative control. As shown in Figure 2a and b, dnSmad3 inhibited induction of integrina1 and paxillin, as well as stimulation of paxillin phosphorylation in TGFb-treated cells. The morphological hallmark of EMT is formation of actin stress fibers. As shown in Figure 2c, addition of TGFb to RIE/pBabe cells induced a characteristic EMT response with formation of prominent actin stress fibers. Stress fiber formation in TGFb-treated cells was not inhibited by Evi1. However, induction of actin stress fibers was inhibited in TGFb-treated cells that express dnSmad3 (RIE/dnSmad3, consistent with the observation that dnSmad3 blocks activation of integrin signaling and focal adhesion kinase. These data indicate that even though Evi1 blocks TGFb-dependent activation of some classical target genes such as PAI1, induction of other genes, such as integrina1 and paxillin is not affected by Evi1, and Evi1 does not block TGFbmediated EMT in intestinal epithelial cells.

a

RIE/dnSmad3

gene induction in intestinal epithelial cells, we assessed the effects of Evi1 on cellular responses to TGFb. The three most prominent cellular responses to TGFb are epithelial to mesenchymal (EMT) transformation, inhibition of proliferation, and induction of apoptosis.

Figure 2 Evi1 does not inhibit TGFb-mediated epithelial to mesenchymal transition in RIE cells. (a) RIE/pBabe, RIE/Evi1, or RIE/snSmad3 cells were treated with 3 ng/ml TGF-b1 for 8 h. RNA was extracted and integrina1 mRNA was measured by QPCR. Data were normalized to GAPDH and calibrated to expression of integrina1 in control RIE/pBabe cells, which was set to 1.0. (b) Cells were treated with 3 ng/ml TGF-b1 for 24 h. Protein was extracted and immunoblotted using specific antibodies for paxillin or phospho-paxillin. (c) Cells were treated with 3 ng/ml TGF-b1 for 24 h. After fixation, cells were stained with Texas Red phalloidin to visualize the actin cytoskeleton.

growth inhibition is due to inhibition of cyclin D1 expression (Ko et al., 1995, 1998). As expected, cyclin D1 mRNA (Figure 3a) and protein (Figure 3b) expression was inhibited by TGFb in RIE/pBabe cells, but not in RIE/dnSmad3 cells. Interestingly, cyclin D1 mRNA and protein expression was also inhibited by TGFb in Oncogene


a

Cyclin D1 mRNA Abundance

3568 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

Control TGFβ

RIE/pBabe

b

RIE/pBabe

RIE/Evi1 RIE/dnSmad3

RIE/Evi1

RIE/dnS3 Cyclin D1 Actin

-

c

+

-

-

+

+

TGFβ

2.5e+6

Cells/Plate

2.0e+6

Control TGFβ

1.5e+6 1.0e+6

RIE/Evi1 cells, suggesting that Evi1 may not affect TGFb-mediated G1 arrest in RIE cells. Therefore, we measured culture growth rates of RIE/pBabe, RIE/Evi1, and RIE/dnSmad3 cells (Figure 3c). Addition of TGFb inhibited proliferation of RIE/pBabe cells, as previously reported (Ko et al., 1994), whereas TGFb did not inhibit proliferation of RIE/dnSmad3 cells. Evi1 did not block TGFb-mediated growth inhibition of RIE/Evi1 cells, consistent with the effects of TGFb on cyclin D1 expression in RIE/Evi1 cells. The results shown in Figure 3c were confirmed using flow cytometry to measure DNA content in TGFbtreated RIE/pBabe and RIE/Evi1 cells, as shown in Table 2. TGFb treatment of RIE/pBabe cells caused a significant increase G1-phase cells (Po0.01) with a concomitant decrease in S-phase cells (Po0.001). Cell cycle distribution of RIE/Evi1 cells was not significantly different from that of RIE/pBabe cells, consistent with the observation that the population doubling times of these cultures are similar. Furthermore, TGFb caused a significant increase in REI/Evi1 cells in G1 phase (Po0.03) with a corresponding decrease in S-phase cells (Po0.006). Thus, in all respects, the antiproliferative effects of TGFb are identical in RIE/pBabe and RIE/Evi1 cells.

5.0e+5 RIE/pBabe

2.0e+6 Control TGFβ

Cells/Plate

1.5e+6 1.0e+6 5.0e+5 RIE/Evi1 3.0e+6

Cells/Plate

2.5e+6

Control TGFβ

2.0e+6 1.5e+6 1.0e+6 5.0e+5

RIE/dnSmad3 0

1

2

3

Days Figure 3 Evi1 does not inhibit TGFb-mediated growth inhibition in RIE cells. Cells were treated with 3 ng/ml TGF-b1 for 24 h and RNA and protein were extracted. (a) QPCR was used to measure cyclin D1 mRNA abundance. Cyclin D1 expression was calibrated to control RIE/pBabe cells, in which expression was set to 1.0. The bars represent the mean and s.d. of three replicates. (b) western blotting was used to measure cyclin D1 protein in the extracts described above. (c) Culture growth was measured by counting cell number in control and TGFb-treated cultures. Control culture growth is shown in filled circles, whereas growth in 3 ng/ml TGF-b1 is shown in open circles. Each point represents the mean and s.d. for three replicate samples.

Oncogene

Evi1 inhibits TGFb-mediated apoptosis in intestinal epithelial cells TGFb-mediated inhibition of RIE cell growth in culture is due, in part, to induction of apoptosis (Conery et al., 2004 ). As shown in Figure 4a, TGFb treatment of RIE/ pBabe cells resulted in a decrease in total caspase 3 and a corresponding increase in cleaved caspase 3 (Casp3*). Likewise, TGFb treatment of RIE/pBabe cells resulted in a decrease in total poly(ADP-ribose) polymerase (PARP) with a corresponding increase in cleaved PARP (PARP*), as shown in Figure 4b. Both Evi1 and dnSmad3 blocked TGFb-mediated cleavage of caspase 3 (Figure 4a) and PARP (Figure 4b), indicating that Evi1 inhibits TGFb-mediated apoptosis. This result was confirmed by measuring release of nuclear histones (Figure 4c). Nuclear histone release was stimulated by TGFb in RIE/pBabe cells. However, both basal and TGFb-induced release of nuclear histones were inhibited in RIE/Evi1 cells relative to RIE/pBabe. We have previously reported that induction of apoptosis in RIE cells involves Smad3 (Conery et al., 2004), and, consistent with these data, RIE/dnSmad3 cells are resistant to induction of nuclear histone release by TGFb (Figure 4c). However, unlike Evi1, dnSmad3 had no effect on basal apoptosis in RIE cells. The observation that Evi1 caused a significant reduction in basal apoptosis in RIE cells suggested that this oncogene might have a general affect on cellular survival. To test this hypothesis, we measured apoptosis in cells treated with taxol (paclitaxel). Addition of 10 nM taxol caused release of nuclear histones in RIE/pBabe cells (Figure 4c). Although RIE/dnSmad3 cells were insensitive to TGFb-mediated apoptosis, these cells were as sensitive to taxol as RIE/pBabe cells. However,


a RIE/pBabe

RIE-1 (tax)

3569 RIE/dnS3

RIE/Evi1

Caspase3 Casp3* +

-

b RIE/pBabe

+

-

RIE/Evi1

+ RIE-1 (tax)

-

-

TGFβ

RIE/dnS3 PARP PARP*

-

c

+

-

+

-

-

+

TGFβ

1.6 Control TGFβ Taxol

Absorbance 405nm

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

RIE/pBabe RIE/Evi1 RIE/dnSmad3

d Cell Death (re. 0hr)

10 RIE/pBabe (1nM) RIE/Evi1(1nM) RIE/pBabe (10nM) RIE/Evi1 (10nM) RIE/pBabe (100nM) RIE/Evi1 (100nM)

8 6 4 2 0 0

2

4 6 8 Hours after Taxol

concentration (Figure 4d). RIE/pBabe cells are exquisitely sensitive to taxol, exhibiting almost complete cell death within 16 h after addition of 10 nM taxol (Figure 4d). RIE/Evi1 cells were significantly more resistant to taxol-mediated cell death. These data indicate that Evi1 protects RIE cells from cell death mediated by physiological mediators such as TGFb or by cytotoxic drugs such as taxol.

10

12

Figure 4 Evi1 inhibits both TGFb-induced and taxol-induced apoptosis in RIE cells. (a) cells were treated with 3 ng/ml TGF-b1 for 24h. Protein was extracted, resolved by electrophoresis, and probed with antibodies against caspase 3. (b) The extracts used in (a) were probed with PARP antibodies. The cleaved forms of Caspase 3 and PARP are identified as Casp* and PARP*, respectively. Taxol-treated RIE cells (RIE-1 tax) were used as positive controls for cleavage of caspase 3 and PARP. (c) cells were treated with TGFb for 24 h, as described above, and nuclear histone release was measured by ELISA. In addition, cells were treated with 10 nM taxol (paclitaxel) for 12 h and nuclear histone release was measured. Bars represent the mean and s.d. of three replicates. (d) RIE/pBabe or RIE/Evi1 cells cells were treated with various concentrations of taxol and cell death was assayed as a function of time by measuring nuclear histone release. Closed symbols correspond to RIE/pBabe and open symbols to RIE/Evi1. Each point represents the mean and s.d. from three replicate samples.

RIE/Evi1 cells were resistant to taxol-mediated apoptosis (Figure 4c). These data were confirmed by measuring nuclear histone release as a function of time and taxol

Evi1 suppresses apoptosis by a phosphoinositide-3-kinase (PI3K)/AKT-dependent mechanism dnSmad3 blocks TGFb-mediated apoptosis, suggesting that the antiapoptotic effects of Evi1 may be mediated by inhibition of Smad signaling. However, there is no known role of Smads in taxol-induced apoptosis. Since Evi1 inhibits taxol-dependent cell death, Evi1 inhibition of taxol-mediated cell death may be unrelated to inhibition of Smad3 signaling. Therefore, we measured the activity of other signal transduction pathways that have been implicated in TGFb signaling and/or apoptosis. Both p38 MAPK and JNK are known to be regulated by TGFb under some circumstances and both have been implicated in apoptosis. Evi1 had no effect on the activity of p38 MAPK, as measured by the phosphorylation state of this enzyme (Figure 5a). Evi1 is known to inhibit JNK activity (Kurokawa et al., 2000). However, phosphorylation of JNK is below the limits of detection in RIE cells, and it was not possible to determine if JNK phosphorylation is affected by Evi1. Conversely, we observed a significant increase in phosphorylation of ERK1/2 and AKT in RIE/Evi1 cells, compared to RIE/pBabe. TGFb had little or no effect on phosphorylation of ERK1/2 or AKT (Figure 5b). (Although there appears to be less phospho-ERK1/2 in TGFb-treated RIE/Evi1 cells, this apparent decrease is due to loading, as evidenced by reduced total ERK1/2 in this lane.) The observation that TGFb does not regulate ERK1/2 or AKT phosphorylation in these cells indicates that differences in basal ERK1/2 and AKT phosphorylation in RIE/pBabe and RIE/Evi1 cells are not due to changes in TGFb signaling. Next, we used a panel of inhibitors to determine if any of these signaling pathways is involved in TGFbmediated apoptosis in RIE/Evi1 cells. As shown in Figure 5c, inhibition of p38 MAPK (by SB203580 or SB202190) had no effect on TGFb-mediated apoptosis in RIE/pBabe or RIE/Evi1 cells. Likewise, inhibition of JNK (SP600125) or MEK1/2 (U0126 or PD98059) had no significant effect on TGFb-mediated apoptosis. However, two inhibitors of PI3K (LY294002 and wortmanin) restored TGFb-mediated cell death in RIE/Evi1 cells. Thus, when RIE/Evi1 cells were treated with TGFb in the presence LY294002, TGFb promoted caspase 3 cleavage (Figure 5d), in contrast to the results obtained when such cells were treated with TGFb in the absence of LY294002 (Figure 4a). Although RIE/Evi1 cells are resistant to taxol (Figures 4c and 5), taxol caused massive nuclear histone release and caspase 3 cleavage in RIE/Evi1 cells in the presence of LY294002 Oncogene


2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

RIE/Evi1

RIE/pBabe

b RIE/pBabe RIE/Evi1 pERK1/2

pp38

pERK1/2

ERK1/2

p38

ERK1/2

pAKT

pJNK

pAKT

AKT

JNK

AKT

RIE Control RIE+TGFβ RIE/Evi1 Control RIE/Evi1+TGFβ

-

+

+

-

TGFβ

d RIE/pBabe RIE/Evi1 Caspase3 Casp3* -

+

-

+

TGFβ

C o SB ntr 20 ol SB 35 20 80 SP 219 60 0 01 2 U 5 0 PD 12 9 6 LY 805 29 9 W 40 or 0 tm 2 an in

Absorbance 405nm

c

RIE/Evi1

a

RIE/pBabe

3570

Absorbance 405nm

e

f 2.5 2.0

Control Taxol LY294002 LY+Taxol

1.5

RIE/pBabe

RIE/Evi1 Caspase3 Casp3* Taxol LY294002

1.0 0.5 0.0 RIE/pBabe RIE/Evi1 RIE/dnSmad3

Figure 5 Evi1 inhibits apoptosis by activation of PI3K/AKT signaling. (a) Extracts were prepared from mid-log phase cultures of RIE/pBabe or RIE/Evi1, and these extracts were blotted using antibodies against phosphorylated and total p38 MAPK, JNK, ERK1/ 2, or AKT. (b) RIE/pBabe or RIE/Evi1 cells were treated with 3 ng/ml TGF-b1 for 24 h, and proteins were extracted and assayed for total and phosphorylated ERK1/2 and AKT. (c) Cell death was assessed by ELISA measurement of nuclear histone release from RIE/ pBabe or RIE/Evi1 cultures treated with 3 ng/ml TGF-b1 for 24 h in the presence or absence of SB203580 (5 mM), SB202190 (5 mM), SP600125 (1 mM), U0126 (5 mM), PD98059 (5 mM), LY294002 (25 mM) or wortmanin (50 nM). (d) Cells were treated with vehicle or with 3 ng/ml TGF-b1 for 24 h in the presence of 25 mM LY294002. Proteins were extracted and Western blotting was carried out to measure caspase 3 cleavage. (e) RIE/pBabe, RIE/Evi1, or RIE/dnSmad3 cultures were treated with 10 nM taxol in the presence or absence of 25 mM LY294002 for 6 h. Cell death was determined by measurement of nuclear histone release. Bars represent the mean7s.d. of three replicate experiments. (f) Cells were treated with 10 nM taxol alone or with 25 mM LY294002 for 6 h. Proteins were extracted and caspase 3 cleavage was assessed by Western blotting.

(Figure 5e and f). Expression of dnSmad3 had no effect on taxol-mediated apoptosis (Figure 5e), indicating that resistance to taxol is not due to loss of Smad3 function in RIE/Evi1 cells. The observation that Evi1 inhibits induction of apoptosis by physiological stimuli such as TGFb as well as pharmacological agents such as taxol, raised the question of the potential role of Evi1 in transformed intestinal epithelial cells. Evi1 suppresses apoptosis in human colon cancer cell lines Evi1 suppresses apoptosis in RIE cells, suggesting that the oncogenic potential of Evi1 may depend upon its ability to function as a survival gene in tumor cells. To explore this hypothesis, we screened a number of human colon cancer cell lines for expression of Evi1. As shown in Figure 6a, Evi1 protein was detected in HT-29 cells Oncogene

and, at lower levels, in Caco2 cells. The abundance of Evi1 protein in HT-29 was significantly less than that observed in endometrial carcinoma HEC-1B cells, which are known to express very high levels of Evi1 (Morishita et al., 1990). Evi1 mRNA was also abundant in HT-29 cells Figure 6b. Evi1 is located on human chromosome 3q26, and this region is known to be amplified in a number of carcinomas (Brass et al., 1996; Brooks et al., 1996; Racz et al., 1999). Therefore, we used quantitative real-time PCR to measure Evi1 gene copy number in the human colon cancer cell lines. In parallel, we measured gene copy number of protein kinase C-iota (PKCi), which is also located on 3q26 near the Evi1 locus. As shown in Figure 6c, both the Evi1 and PKCi genes were significantly amplified in HT-29 cells, indicating that the 3q26 locus is amplified in these cells and suggesting


P=0.039

2.0 1.5

Caspase3 Casp3* Taxol LY294002

1.0 0.5 0.0

d pAKT

Evi1 PKCι

AKT

Actin

60

pAKT

Taxol LY294002

40

AKT

e

H C T15 H C T1 16

C ac o2

H T29

H EC -1 B

f 3.0 2.5

Evi1 PKCi 6

4

2

H C T1 16

H C T15

C ac o2

H T29

0

Figure 6 Evi1 expression and gene amplification in human colon cancer cell lines. (a) Protein was extracted from human endometrial cancer HEC-1B cells and from four human colon cancer cell lines. Western blotting was carried out to measure Evi1 expression. (b) Total RNA was extracted and Evi1 mRNA abundance measured by QPCR. Data are normalized to GAPDH and calibrated to HCT116, in which Evi1 expression was set to 1.0. Bars represent the mean and s.d. of three replicates. (c) Genomic DNA was extracted from four colon cancer cell lines and gene copy number was measured by QPCR using primers and probes that correspond to intronic sequences within the genes that encode Evi1, PKCi, or RNAseP. Copy number is normalized to RNAseP and calibrated to Evi1 copy number in HCT-15 cells. The bars represent the means and s.d. of three replicate measurements.

that overexpression of Evi1 in HT-29 cells is the result of gene amplification. We next asked if Evi1 is a survival gene in HT-29 cells. HT-29 cells are much more resistant to taxol than are RIE cells, making it necessary use 100 nM taxol for 48 h to measure significant nuclear histone release. However, HT-29 cells treated with LY294002 exhibited significantly more nuclear histone release when treated with

shEvi1 Caspase3

2.0

Casp3*

1.5

Taxol

1.0 0.5 0.0

8

shGFP

Control TGFβ Taxol

P=0.031

20

0

Gene Copy Number (re RNaseP)

2.5

c

Absorbance at 405nm

Evi1 mRNA Abundance

b 80

Control Taxol LY294002 LY+Taxol TGFβ LY+TGFβ

3.0

shEvi1

Actin

Absorbance at 405nm

Evi1

c

b

a 3.5

shGFP

HCT116

HCT-15

Caco2

HT-29

HEC-1B

3571

a

GFP

shRNA

Figure 7 Knockdown of Evi1 conveys increased sensitivity to the cytotoxic effects of taxol. (a) HT-29 cells were treated with 100 nM taxol for 48 h in the presence or absence of 50 mM LY294002. In parallel, cells were treated with 3 ng/ml TGF-b1 for 24 h in the presence or absence of 50 mM LY294002. Cell death was assessed by nuclear histone release. (b) Proteins were extracted from HT-29 cells treated with taxol7LY294002 as described above. Western blotting was carried out to measure caspase 3 cleavage. (c) The extracts used in (b) were probed with antibodies against total AKT and phospho-AKT (pAKT). (d) HT-29 cells were transfected with shRNA expression constructs that correspond to GFP or Evi1. An aliquot of the transfected cells was harvested at 48 h. Extracts were prepared and assayed for expression of Evi1, PKCi, actin, phospho-AKT and total AKT. (e) The remaining shRNAtransfected HT-29 cells were treated with 3 ng/ml TGF-b1 (gray bars) or 100 nM taxol for 48 h (black bars). Nuclear histone release was measured as an indicator of cell death. (f) western blotting was used to confirm caspase 3 cleavage in taxol-treated HT-29 cells transfected with shGFP or shEvi1 expression vectors, as described above.

taxol (Figure 7a). HT-29 cells express very low levels of the TGFb type II receptor (Winesett et al., 1996), and, as a result, are resistant to TGFb-mediated apoptosis. As expected, LY294002 did not restore TGFb sensitivity to HT-29 cells (Figure 7a). Consistent with the nuclear histone release data, LY294002 increased caspase 3 cleavage in taxol-treated HT29 cells (Figure 7b) indicating that PI3K activity confers resistance to taxolmediated apoptosis. Taxol alone had no significant effect on AKT expression or activity in HT-29 cells (Figure 7c), whereas LY294002 effectively blocked AKT phosphorylation in the presence and absence of taxol. RNAi technology was used to determine if Evi1 inhibits apoptosis by a PI3K/AKT-dependent pathway in HT29 cells. As shown in Figure 7d, shRNA against Evi1 reduced Evi1 protein expression by >80% with no effect on expression of PKCi. However, knockdown of Oncogene


3572

Evi1 significantly inhibited AKT phosphorylation, demonstrating that Evi1 activates PI3K/AKT signaling in HT-29 cells, just as it does in RIE cells. To determine if knockdown of Evi1 expression increases the sensitivity of HT-29 cells to apoptotic stimuli, HT-29 cells transfected with shRNA expression vectors against GFP or Evi1 were treated with TGFb and apoptosis was measured (Figure 7e). Evi1 knockdown had no effect on the resistance of these cells to TGFb. However, knockdown of Evi1 resulted in a significant increase in taxol sensitivity as measured by nuclear histone release (Figure 7e) or caspase 3 cleavage (Figure 7f). We conclude that in addition to the ability to inhibit Smadmediated signaling, Evi1 also functions as a survival gene by activating an antiapoptotic PI3K/AKT signaling pathway in both non-transformed intestinal epithelial cells and in colon cancer cells.

Discussion A number of groups have shown that Evi1 inhibits classical Smad-mediated transactivation of reporter genes. However, the effects of Evi1 on cellular responses to TGFb have received less attention, particularly in non-hematopoietic cells. Alliston et al. (2005) showed that Evi1 inhibits induction of Smad7 in myoblasts, the first demonstration that Evi1 can inhibit induction of an endogenous gene by TGFb. Here we show that Evi1 also inhibits induction of PAI1 in intestinal epithelial cells. Since Evi1 potently inhibits TGFb-dependent transactivation of some classical target genes, such as PAI1, we expected Evi1 to block TGFb-mediated growth inhibition of RIE cells, just as Evi1 blocks TGFb-mediated growth inhibition in Mv1Lu cells (Kurokawa et al., 1998). However, this prediction was not borne out. Since Evi1 inhibits induction of PAI1 by >90%, it seems unlikely that residual Smad activity in RIE/Evi1 cells is responsible for growth arrest. It is more likely that failure to observe growth inhibition is due to the mechanism that accounts for this response in RIE cells. In some cell types, TGFb inhibits growth by Smad3/ Sp1-dependent activation of the p15INK4B or p21Cip1 promoters (Feng et al., 2000; Pardali et al., 2000). For example, in Mv1Lu cells TGFb-mediated growth arrest results from induction of p15INK4B (Hannon and Beach, 1994). Evi1 blocks transactivation of p15INK4B promoter/reporter constructs (Feng et al., 2000), and also blocks inhibition of [3H]-thymidine incorporation in TGFb-treated Mv1Lu cells (Kurokawa et al., 1998). Thus Evi1 blocks Smad-dependent induction of p15INK4B, thereby attenuating TGFb-mediated growth inhibition. However, p15INK4B is undetectable in RIE cells, even in the presence of TGFb. Instead, we have shown that TGFb-mediated growth inhibition of RIE cells is caused by inhibition of cyclin D1 expression (Ko et al., 1998). Evi1 has no effect on TGFb-mediated inhibition of cyclin D1 expression, and, as a consequence, does affect TGFb-mediated growth arrest in intestinal epithelial cells. Oncogene

Although Evi1 blocks induction of some TGFb target genes, such as PAI1, other TGFb-induced genes are not affected by Evi1. For example, Evi1 does not block induction of paxillin or integrina1. Consequently, Evi1 does not block TGFb-mediated activation of focal adhesion kinase, nor does Evi1 block induction of EMT by TGFb. Thus, intestinal epithelial cells that express Evi1 exhibit two major physiological responses to TGFb: growth inhibition and EMT. Interestingly, Evi1 is a potent inhibitor of TGFb-mediated apoptosis. This property of Evi1 appears likely to be integral to its oncogenic potential, since loss of response to apoptotic signals is one of the hallmarks of transformation. We have shown that the apoptotic response to TGFb can be influenced by crosstalk between Smad3 and AKT (Conery et al., 2004), and several reports indicate that activation of PI3K/AKT signaling suppresses TGFbmediated apoptosis (Chen et al., 1998, 1999; Shih et al., 2000). Our data demonstrate that Evi1 activates PI3K/ AKT signaling, and suppresses TGFb-mediated apoptosis. However, the antiapoptotic effects of Evi1 are not limited to TGFb, since Evi1 also inhibits taxol-mediated cell death by a PI3K-dependent mechanism. Our results indicate that Evi1 functions as a survival factor in intestinal epithelial cells, and is likely to convey broad resistance to a number of physiological and pharmacological activators of apoptosis. Thus, the oncogenic effects of Evi1 may be inherent in the ability to activate PI3K and suppress apoptosis in transformed cells. Evi1 resides at the 3q26 locus, which is known to be amplified in a number of solid tumors. This locus contains several potential oncogenes, in addition to Evi1, which may contribute to the transformed phenotype. For example, PKCi is located adjacent to Evi1 at this locus. We have shown that PKCi is an oncogene that is activated by amplification in squamous cell carcinoma of the lung (Regala et al., 2005a). We have also shown that PKC plays a critical role in transformed growth (Zhang et al., 2004; Regala et al., 2005b), but does not appear to function as a survival gene. In contrast, our present results show that Evi1 is a survival gene in human colon cancers that have undergone amplification of the 3q26 locus. Based on these observations, we hypothesize that Evi1 and PKCi may collaborate during colon carcinogenesis, with Evi1 suppressing apoptosis and PKCi stimulating transformed growth. Our data therefore identify Evi1 as a potential target for chemotherapy. Our results indicate that Evi1 may be overexpressed in a subset of human colon cancers, and that Evi1 might affect disease progression and/or sensitivity to chemotherapy. Current studies are focused upon evaluating the extent of 3q26 amplification in human colon cancers and assessing whether 3q26 amplification predicts sensitivity to chemotherapy and or survival of colon cancer patients. Materials and methods Cell culture and antibodies Rat intestinal epithelium cells RIE-1 were maintained in DMEM supplemented with 5%FBS. Human colon cancer cell


3573 lines and HEC-1B cell line were grown under conditions recommended by American Type Culture Collection (ATCC), from which these cells were obtained. The Evi1 antibody was kindly provided by Dr AS Perkins (Yale University School of Medicine). All other antibodies were purchased from Santa Cruz. Cloning and construction of retroviral expression vectors Dominant-negative rat Smad3DSSVS in the pRK5 expression vector was obtained from Rick Derynck (University of California at San Francisco) and cloned into a pBabe/puro-3 derivative that encodes an N-terminal flag tag. This retrovirus was used to transform RIE-1 cells to puromycin resistance. Expression of dnSmad3 was confirmed by immunoblotting with flag antibodies, and function of dnSmad3 was confirmed by suppression of transient activation of the 3TP/luc TGFb reporter. Full-length human Evi1 cDNA was cloned from mRNA extracted from a human colon tumor. Doublestranded cDNA was synthesized using Clonetech SMART cDNA technology, according to the manufacturer’s suggested protocols. Evi1 cDNA was amplified from doublestranded cDNA using primers that amplify from the translation start site in exon 3 (50 GCTTCTGCTGTTCATGAAGAGC) through the stop codon in exon 16 (50 GTCAACCTTGA TAACGTCATAC). The PCR products were cloned into pRK5 and the identity of the human Evi1 cDNA was confirmed by sequencing. In parallel, the Evi1 coding sequence was inserted into a pBabe/puro-3 derivative that encodes an Nterminal flag tag. Protocols for virus production and infection were obtained from the laboratory of Gary Nolan at Stanford University http://www.stanford.edu/group/nolan/retroviral_systems/phx.html. RIE-1 cells were infected with pBabe/ flagEvi1 or empty pBabe/puro-3, and populations of stable transformants were selected for the ability to grow in 5 mg/ml puromycin. Evi1 expression in RIE/Evi1 populations was confirmed by immunoblot analysis. Quantitative real-time RT-PCR Total RNA or genomic DNA was isolated using Ambion RNAqueous or Qiagen DNA Mini kits. QPCR was carried out using target-specific probes and primers obtained from PE Applied Biosystems (ABI). Primers and reporters for rat cyclin D1, integrina1, PAI1, and GAPDH mRNA and for human Evi1 and PKCi genomic DNA were obtained using the ABI Assay by Design procedure. The sequences of these reagents are shown in Table 1. The reagents used to measure human

Table 1

Evi1 and GAPDH mRNA were obtained from the ABI Assay on Demand catalog, and the context sequence against which these primers and reporters are directed is also shown in Table 1. Reagents used to measure rat and human 18S RNA and human RNaseP were obtained from the Assay on Demand catalog, and the context sequences of these reagents may be obtained from the ABI web site. All QPCR reagents were validated by demonstrating a linear relationship between sample concentration and amplification kinetics over a three log range of nucleic acid concentrations, using cDNA made from total RNA or genomic DNA from reference samples and normalizing to 18S RNA (DCT ¼ CTtarget CT18S ) for mRNA and RNaseP (DCT ¼ CTtarget CTRNaseP ) for genomic DNA. Taqman Universal Master Mix (ABI) was used for PCR reactions and amplification data were collected using an ABI Prism 7900 Sequence Detector and analysed using the Sequence Detection System software (SDS V2.0) from ABI. All QPCR reagents were validated by demonstrating a slope of 070.1 when DCT was plotted against log10 nucleic acid concentration Unless stated otherwise, the abundance of mRNA was calculated by normalization to GAPDH (DCT ¼ CTtarget CTGAPDH ) and calibrated to mRNA abundance in untreated RIE/pBabe cells (DDCT ¼ DCTcell line DCTRIE=pBabe ). Data are represented as 2DDCT, such that the abundance of the individual mRNAs in RIE/pBabe is expressed as 1.0 (DCTRIE=pBabe DCTRIE=pBabe ¼ 0, and 20 ¼ 1.0). The abundance of Evi1 mRNA in colon cancer cells was normalized to GAPDH and calibrated to HCT116 cells. Evi1 and PKCi gene copy number in human colon cancer cell lines was normalized against RNaseP and calibrated to Evi1 gene copy number in HCT-15. Statistical analysis for QPCR and other results was carried out using the Mann-Whitney Rank Sum analytical function of Sigma Stat 32 (Table 2). Immunocytochemistry analysis of actin stress fiber formation during EMT RIE cells cultured on cover glasses (Fisher Scientific) were fixed in 3% paraformaldehyde for 30 min, blocked with 3% nonfat milk in PBS for 10 min, and incubated with Texas RedX phalloidin (Molecular Probes) for 30 min in the dark. The cover glasses were then washed three times with PBS and mounted on glass slides (Fisher Scientific) using Aqua Poly/ Mount (Polysciences). Fluorescence images were captured using a Leica DM5000 microscope and compiled into Adobe Photoshop.

QPCR primer sequences

Gene

Forward primer

Rat Cyclin D1

GCACTTTCTTTCCAGAGTCAT GACTCCAGAAGGGCTTCAATCT ACCCGGACTGCCTCC CAAG CCATCCCAGTAAAATAT GCAGCGACTGAAATGT CTCGCAGAACTGTAAAAC GAAGTTGGACT GATGTTC GCCCCGCCTCCTCATC GCCCTCTGAGGTCCACTTCA CTGCCTAAGTTCTCTCTG CTGGCGTCTTCACCACCAT TTTGGCCCCACCCTTCAG CCCCAGCCTTCTCC Context Sequence ¼ GGGCGCCTGGTCACCAGGGCTGCTT Context Sequence ¼ TCCAAATCGCAGGCATATGCTATGA CATCGGAAATGTTTTATTGTT TTTCTTTATCTTCCCATTTGT CAACCATGAAAAAATAAT TATGC GAAGTC CGTTCTTCCGAAATGTTGATTG TCCCCAGAAATATTTGGTT TTGCTCCATCATATCC TAAAGG

Rat Integrina1 Rat PAI1 Rat GAPDH Human GAPDH Human Evi1 Human Evi1 DNA Human PKCi DNA

Reverse primer

Reporter

All reagents were obtained from Applied Biosystems using the Assay by Design protocol or the Assay on Demand catalog, and were validated as described in Materials and methods. Oncogene


3574 Table 2 Cell cycle profile of RIE/pBabe and RIE/Evi1 cells treated with 3 ng/ml TGF-b1 for 24 h Cell line

RIE/pBabe

RIE/Evi1

Phase

G1

S

G2/M

G1

S

G2/M

Control TGFb

79.4974.52 88.9873.69

17.3272.33 6.4373.04

3.2071.25 4.5971.69

75.5975.31 89.5274.35

21.8973.19 7.872.67

2.5271.52 2.6872.89

Mid-log phase cultures were treated with vehicle or 3 ng/ml TGF-b1 for 24 h. Cell cycle distribution was determined by flow cytometric analysis of 7-AAD-stained cells. Data represent the mean and s.d. of three replicate experiments.

Transient transfection assays The 3TP/luc reporter (Wrana et al., 1992) was used to measure TGFb responsiveness. Transfection was carried out in six-well plates and 50 ng of 3TP/luc was added to each well along with 5 ng phRL-SV40 (Promega), which expresses renilla luciferase. In cotransfection experiments, 150 ng of Evi1 or dnSmad3 expression vector was added. Total DNA in transfection assays was adjusted by adding vector plasmid to equalize the DNA concentration. Cells were seeded at a density of 3 105 per well in six-well plates for 18 h. The cells were then transiently transfected using FuGENE 6 Transfection Reagent (3 ml FuGENE/mg DNA transfected), according to manufacturer’s instruction (Roche). Cell extracts were prepared and assayed for firefly and renilla luciferase activity using a Dual Luciferase Assay kit (Promega). Light emission was detected using a microplate luminometer (Turner Design Inc.). Apoptosis and culture growth The Roche Cell Death Detection ELISA kit was used to measure release of nuclear histones during apoptosis. Nuclear histone release was measured by ELISA, according to the manufacturer’s recommendation. A final concentration 3 ng/ ml of TGF-b1 (120 pM) dissolved in 0.1% HCl was added to culture medium containing 0.1% fetal bovine serum. Vehicletreated controls received an equivalent volume of 0.1% HCl in the same medium. Cell proliferation was measured by counting cells using a hemocytometer. Cultures were treated with vehicle or 3 ng/ml TGF-b1 in medium containing 5% fetal bovine serum. For cell cycle analysis, mid-log phase cells were permeablized and stained with 7-amino-actinomycin D (7AAD) using the BD PharMingen Flowkit and protocols. In all, 1 104 cells from each culture were analysed using a Becton Dickinson FACSVantage SE flow cytometer.

Evi1 shRNA knockdown in HT-29 cell We designed and tested several Evi1 shRNAs using the Ambion Hairpin siRNA Template Design program. These were screened for inhibition of Evi1 expression by measuring Evi1 mRNA abundance following transient transfection into HT-29 cells using the Amaxa Nucleofector electroporation device and optimized buffers for HT-29 (CAT VCA-1001) provided by Amaxa Biosystems. The most effective oligonucleotide sequences were as follows: GATCCCTCTAAGGCT GAACTAGCAGTTCAAGAGACTGCTAGTTCAGCCTTA GATTTTTTGGAAA and AGCTTTTCCAAAAAATCTAA GGCTGAACTAGCAGTCTCTTGAACTGCTAGTTCAGCC TTAGAGG. Bold nucleotides correspond to sense and antisense Evi1 sequences. These two oligonucleotides were annealed and ligated into vector pSilencer-H1 (Ambion) at the BamH1 and HindIII sites to yield the pSilencerH1-Evi1 shRNA expression vector. RNA interference knockdown experiments were controlled by the use of pSilencerH1-GFP (Ambion) The shRNA expression vectors were introduced into HT-29 cells by electroporation using the Amaxa Nucleofector, as described above.

Acknowledgements This work was supported in part by NCI grants R01CA64701 to EAT and R01DK060105 and P01DK35608 to TCK. The experiments described in this project result from a collaboration between the Thompson and Fields laboratories, and both PIs contributed equally to the supervision of Dr Liu, who performed most of the work described in this manuscript.

References Akhurst RJ, Derynck R. (2001). Trends Cell Biol 11: S44–S51. Alliston T, Ko TC, Cao Y, Liang YY, Feng XH, Chang C et al. (2005). J Biol Chem 280: 24227–24237. Bartholomew C, Ihle JN. (1991). Mol Cell Biol 11: 1820–1828. Bergamo NA, Rogatto SR, Poli-Frederico RC, Reis PP, Kowalski LP, Zielenska M et al. (2000). Cancer Genet Cytogenet 119: 48–55. Brass N, Ukena I, Remberger K, Mack U, Sybrecht GW, Meese EU. (1996). Eur J Cancer 32A: 1205–1208. Brooks DJ, Woodward S, Thompson FH, Dos Santos B, Russell M, Yang JM et al. (1996). Br J Cancer 74: 1518–1525. Chen RH, Chang MC, Su YH, Tsai YT, Kuo ML. (1999). J Biol Chem 274: 23013–23019. Chen RH, Su YH, Chuang RL, Chang TY. (1998). Oncogene 17: 1959–1968. Conery AR, Cao Y, Thompson EA, Townsend Jr CM, Ko TC, Luo K. (2004). Nat Cell Biol 6: 366–372. Oncogene

Feng XH, Lin X, Derynck R. (2000). EMBO J 19: 5178–5193. Hannon GJ, Beach D. (1994). Nature 371: 257–261. Hay ED. (1995). Acta Anat (Basel) 154: 8–20. Higashidani Y, Tamura S, Morita T, Tadokoro T, Yokoyama Y, Miyazaki J et al. (2003). Dis Colon Rectum 46: 327–332. Hirai H. (1999). Int J Biochem Cell Biol 31: 1367–1371. Imoto I, Pimkhaokham A, Fukuda Y, Yang ZQ, Shimada Y, Nomura N et al. (2001). Biochem Biophys Res Commun 286: 559–565. Izutsu K, Kurokawa M, Imai Y, Maki K, Mitani K, Hirai H. (2001). Blood 97: 2815–2822. Ko TC, Beauchamp RD, Townsend Jr CM, Thompson EA, Thompson JC. 1994. Am J Surg 167: 14–19 discussion 19–20. Ko TC, Sheng HM, Reisman D, Thompson EA, Beauchamp RD. (1995). Oncogene 10: 177–184. Ko TC, Yu W, Sakai T, Sheng H, Shao J, Beauchamp RD et al. (1998). Oncogene 16: 3445–3454.


3575 Kurokawa M, Mitani K, Irie K, Matsuyama T, Takahashi T, Chiba S et al. (1998). Nature 394: 92–96. Kurokawa M, Mitani K, Yamagata T, Takahashi T, Izutsu K, Ogawa S et al. (2000). EMBO J 19: 2958–2968. Markowitz S, Hines JD, Lutterbaugh J, Myeroff L, Mackay W, Gordon N et al. (1995). Clin Cancer Res 1: 441–445. Mochizuki N, Shimizu S, Nagasawa T, Tanaka H, Taniwaki M, Yokota J et al. (2000). Blood 96: 3209–3214. Morishita K, Parganas E, Douglass EC, Ihle JN. (1990). Oncogene 5: 963–971. Morishita K, Parker DS, Mucenski ML, Jenkins NA, Copeland NG, Ihle JN. (1988). Retroviral activation of a novel gene encoding a zinc finger protein in IL-3-dependent myeloid leukemia cell lines. Cell 54: 831–840. Murray NR, Weems C, Chen L, Leon J, Yu W, Davidson LA et al. 2002. Cell Biol 157: 915–920. Ogawa S, Kurokawa M, Tanaka T, Mitani K, Inazawa J, Hangaishi A et al. (1996). Oncogene 13: 183–191. Palmer S, Brouillet JP, Kilbey A, Fulton R, Walker M, Crossley M et al. (2001). Evi-1 transforming and repressor activities are mediated by CtBP co-repressor proteins. J Biol Chem 276: 25834–25840. Pardali K, Kurisaki A, Moren A, ten Dijke P, Kardassis D, Moustakas A. (2000). J Biol Chem 275: 29244–29256. Parsons R, Myeroff LL, Liu B, Willson JK, Markowitz SD, Kinzler KW et al. (1995). Cancer Res 55: 5548–5550. Racz A, Brass N, Heckel D, Pahl S, Remberger K, Meese E. (1999). Eur J Cancer 35: 641–646. Regala RP, Weems C, Jamieson L, Copland JA, Thompson EA, Fields AP. (2005a). J Biol Chem 280: 31109–31115.

Regala RP, Weems C, Jamieson L, Khoor A, Edell ES, Lohse CM, Fields AP. (2005b). Cancer Res 65: 8905–8911. Riggins GJ, Kinzler KW, Vogelstein B, Thiagalingam S. (1997). Cancer Res 57: 2578–2580. Roman C, Saha D, Beauchamp R. (2001). Microsc Res Tech 52: 450–457. Sattler HP, Lensch R, Rohde V, Zimmer E, Meese E, Bonkhoff H et al. (2000). Prostate 45: 207–215. Shih WL, Kuo ML, Chuang SE, Cheng AL, Doong SL. (2000). J Biol Chem 275: 25858–25864. Sonoda G, Palazzo J, du Manoir S, Godwin AK, Feder M, Yakushiji M et al. (1997). Genes Chromosomes Cancer 20: 320–328. Sood R, Talwar-Trikha A, Chakrabarti SR, Nucifora G. (1999). Leukemia 13: 348–357. Sugita M, Tanaka N, Davidson S, Sekiya S, Varella-Garcia M, West J et al. (2000). Cancer Genet Cytogenet 117: 9–18. Tanaka T, Nishida J, Mitani K, Ogawa S, Yazaki Y, Hirai HJ. (1994). Biol Chem 269: 24020–24026. Wakefield LM, Roberts AB. (2002). Curr Opin Genet Dev 12: 22–29. Winesett MP, Ramsey GW, Barnard JA. (1996). Carcinogenesis 17: 989–995. Wrana JL, Attisano L, Carcamo J, Zentella A, Doody J, Laiho M et al. (1992). Cell 71: 1003–1014. Wrana JL, Attisano L, Wieser R, Ventura F, Massague J. (1994). Nature 370: 341–347. Zhang J, Anastasiadis PZ, Liu Y, Thompson EA, Fields AP. (2004). J Biol Chem 279: 22118–22123.

Oncogene