Role of PLZF in melanoma progression - Nature

Oncogene (2004) 23, 4567–4576

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

Role of PLZF in melanoma progression Federica Felicetti1, Lisabianca Bottero1, Nadia Felli1, Gianfranco Mattia1, Catherine Labbaye1, Ester Alvino2, Cesare Peschle1, Mario P Colombo3 and Alessandra Care`*,1 1 Department of Hematology, Oncology and Molecular Medicine, Istituto Superiore di Sanita`, 00161 Rome, Italy; 2Institute of Neurobiology and Molecular Medicine, CNR, 00100 Rome, Italy; 3Department of Experimental Oncology, Immunotherapy and Gene Therapy Unit, Istituto Nazionale Tumori, 20133 Milan, Italy

The promyelocytic leukemia zinc finger (PLZF) protein has been described as a transcriptional repressor of homeobox (HOX)-containing genes during embryogenesis. As we previously demonstrated a functional link between overexpression of HOXB7 and melanoma progression, we investigated the lack of PLZF as the possible cause of HOXB7 constitutive activation in these neoplastic cells. Accordingly, we found PLZF expression in melanocytes, but not in melanoma cells, a pattern inversely related to that of HOXB7. PLZF retroviral gene transduction was then performed in a panel of melanoma cell lines, and tumorigenicity was compared with that of empty vector-transduced control cell lines. Evaluation of in vitro migration, invasion and adhesion indicated that PLZF gene transduction induced a less malignant phenotype, as confirmed through in vivo studies performed in athymic nude mice. This reduced tumorigenicity was not coupled with HOXB7 repression. In order to find more about the molecular targets of PLZF, the gene expression profiles of PLZF- and empty vector-transduced A375 melanoma cells were analysed by Atlas Cancer macroarray. Among several genes modulated by PLZF enforced expression, of particular interest were integrin avb3, osteonectin/SPARC and matrix metalloprotease-9 that were downmodulated, and the tyrosinase-related protein-1 that was upregulated in all the analysed samples. This profile confirms the reduced tumorigenic phenotype with reversion to a more differentiated, melanocyte like, pattern, thus suggesting a suppressor role for PLZF in solid tumors. Moreover, these results indicate that PLZF and HOXB7 are functionally independent and that their coupled deregulation may account for most of the alterations described in melanomas. Oncogene (2004) 23, 4567–4576. doi:10.1038/sj.onc.1207597 Published online 12 April 2004 Keywords: PLZF; homeobox gene; cDNA expression array; melanoma; tumor progression

*Correspondence: A Care`; E-mail: [email protected] Received 3 September 2003; revised 29 December 2003; accepted 28 January 2004; Published online 12 April 2004

Introduction The promyelocytic leukemia zinc finger (PLZF) was firstly discovered as a fusion protein with the retinoic acid receptor a (RARa) locus in a retinoid acid resistant form of acute promyelocytic leukemia (APL) carrying the t(11;17) chromosomal translocation (Licht et al., 1995). The PLZF gene encodes for a zinc-finger transcription factor, which is localized in nuclear speckles, but becomes delocalized in APL patients (Reid et al., 1995). PLZF is a DNA-binding phosphoprotein characterized by the presence of a BTB/POZ domain involved in its abilities to homodimerize (Dong et al., 1996) and in repressing gene transcription (Li et al., 1997). Besides several data showing the dominant-negative role of the rearranged PLZF gene into PLZF-RARa and RARa-PLZF fusion proteins conferring a survival advantage to the leukemic cells (Chen et al., 1993; He et al., 1998), only recent works analysed the functional role of the normal PLZF product. During normal hematopoiesis, PLZF is expressed in CD34 þ immature cells and declines when cells are induced to erythroid, granulocytic and monocytic differentiation, indicating that PLZF might play a role in the maintenance of the quiescent state of stem/progenitor cells; conversely, PLZF expression is initially unaffected and then upmodulated during megakaryocytic maturation (Krug et al., 2002; Labbaye et al., 2002). In situ analysis in mouse embryos revealed an expression pattern correlated with that of the homeobox (HOX)-containing gene family. Moreover, targeting disruption of Zfp145 gene (encoding for PLZF) originated knockout mice displaying defects in limb patterning and skeletal elements along the proximodistal axis (Barna et al., 2000). Several studies on HOX genes and functional data derived from gain- or loss-offunction mutations in mouse embryos indicated that their expression is stage related and tissue or region specific and that they play a fundamental role in the determination of positional identity (Dubrulle and Pourquie, 2002). PLZF could therefore regulate HOX gene expression during embryogenesis; in this regard, PLZF canonical binding sites have been identified within the Abdb HoxD gene complex, suggesting that PLZF might act as a direct transcriptional repressor of HOX genes (Barna et al., 2000).

PLZF in melanoma progression F Felicetti et al

4568

Moreover, PLZF binds to and represses the cyclin A2 promoter and is involved in the regulation of cell cycle progression and growth suppression (Yeyati et al., 1999). The lack of this normal coordinated regulation of PLZF and cyclin A, proper of normal B lymphocytes, has been associated to B-chronic lymphocytic leukemia (B-CLL) progression and a 100-fold reduction of PLZF has been reported in B-CLL compared to normal B cells (Parrado et al., 2000). Aiming to identify genes that might restore the HOX gene control lost in tumors (Abate-Shen, 2002), we analysed PLZF endogenous expression and the effects of PLZF retroviral gene transduction in melanoma cells. Here we show that PLZF is expressed in melanocytes, but not in melanoma cells, a pattern that inversely correlates with that previously reported for HOXB7 (Care` et al., 1996, 1998). Although we found a perfect PLZF DNA consensus sequence in HOXB7 promoter, we could not demonstrate its functional binding and no downregulation of HOXB7 mRNA was obtained through PLZF enforced expression. Nonetheless, in melanoma cell lines retrovirally transduced with PLZF gene, a clear modulation of several crucial genes was obtained, inducing a more differentiated, less malignant phenotype and suggesting a potential role for PLZF as a tumor suppressor gene for tumors other than leukemias.

Results PLZF endogenous expression in normal and neoplastic melanocytes We evaluated PLZF expression in normal human epidermal melanocytes from foreskin (Promocell, Heidelberg, Germany) and in a panel of melanoma cell lines derived from tumors at different stages of progression, including radial growth phase (RGP) and vertical growth phase (VGP) primary melanomas as well as subcutaneous and lymph node metastases. As determined by RT–PCR, RNase protection and Western blot analyses, PLZF was barely or not expressed in all the examined melanoma samples, even after 40 cycles of amplification and Southern blot analysis, whereas it was expressed in the normal melanocytic counterpart at both mRNA and protein levels (Figures 1 and 2). The absence of PLZF mRNA in melanoma cells was confirmed in primary and metastatic samples analysed at low passages (Figure 1). PLZF enforced expression in melanoma cell lines We investigated the effects of PLZF gene transduction in melanoma cells. The retroviral vector LPLZFSN was utilized to transduce five melanoma cell lines WM983A, Me665/1, A375, Me1811 and Me1402/R corresponding to different stages of progression including one primary tumor VGP, one metastatic melanoma, two lymph node metastases and one recurrence of primary tumor (listed in Care` et al., 1996). Supernatants from Am12/ LPLZFSN and control Am12/LXSN producer cell lines Oncogene

Figure 1 RT–PCR analysis of PLZF in a panel of primary melanomas, melanoma cell lines and normal human melanocytes. Top: Real-time analysis. Bottom: Traditional PCR; PLZF was evaluated at 40 cycles, while control GAPDH remained in the exponential phase at 18 cycles. () is the negative control, including all the reaction reagents except cDNA. RNAs from an 8-week-old fetal liver and from the erythroleukemic cell line HEL were used as positive controls

were used to infect the target cells that in turn were selected with 0.8 mg/ml of G418 and tested for the transgenic expression at mRNA as well as protein level. Real-time RT–PCR, RNase protection and Western blot analyses revealed the correct transcription and translation of PLZF mRNA and protein (Figure 2). Since PLZF pattern of expression inversely correlates with that of the HOX-containing gene, HOXB7 (Care` et al., 1996), and being a PLZF putative binding site in HOXB7 promoter, we firstly tested whether PLZF gene transduction could affect HOXB7 transcription in melanoma cells. No relevant modulation of this gene was observed by either the traditional or the real-time RT–PCR and confirmed by RNase protection in PLZFtransduced vs control melanoma cell lines (Figure 2a, b and not shown). These results were confirmed by bind-shift and cotransfection experiments. Nuclear extracts obtained from PLZF- or vector-transduced melanoma cell lines were used to run electrophoretic mobility shift assay (EMSA) with an oligomer encompassing the PLZF putative DNA-binding site (position 945) present in the HOXB7 promoter (Meccia et al., 2003). Antibody supershift did not show any protein complex containing PLZF (not shown). Moreover, transactivation experiments were performed: the PLZF cDNA coding sequence was cotransfected either with the whole HOXB7 promoter, containing the PLZF DNA-binding site or, as a negative control, with a shorter 30 region with no putative PLZF sites (Meccia et al., 2003). No significant changes of CAT activity were observed in both cases (not shown). Role of exogenous PLZF on melanoma cell proliferation, migration and invasion in vitro In order to determine whether PLZF presence could affect important biological properties of different melanomas, we used some in vitro specific assays. LXSN- and PLZF-transduced melanoma cell lines were


4569

plated and cell proliferation measured at different times. Melanoma cells, expressing PLZF, showed a reduced proliferative rate ranging from 40 to 60% when compared with parental or empty vector-transduced cells (Figure 3). Proliferation was further reduced in the A375/PLZF high-expresser clone G3 showing a 70% of decrease in the growth rate. The A375/PLZF lowexpresser clone G9, showing a proliferative rate comparable with the PLZF-transduced bulk population, is also reported (Figure 3). We then analysed whether such enforced expression of PLZF could interfere with the invasive ability of melanomas and, as shown in Figure 4a, a clear reduction was observed. By using an in vitro Matrigel assay, we found a reduction of 40–60% for all the examined melanoma cell lines when compared with LXSN-transduced control cells. In the chemotaxis assay, we observed essentially similar results (Figure 4b), indicating that PLZF expression affected the migration toward the lower chambers. In contrast, we could not find any clear effect due to PLZF expression on cell adhesion to a matrigel substrate (not shown). It was interesting to note that adhesion on vitronectin-coated substrates was strongly reduced in PLZF-expressing cells, thus suggesting an involvement of integrin avb3, the adhesive glycoprotein most frequently involved in melanoma invasion and metastasis (Seftor, 1998). In three independent experiments performed in plates coated with 50 or 200 ng/cm2 of soluble vitronectin, A375/PLZF and Me1811/PLZF melanoma cells showed a residual spreading capacity of 40% when compared to LXSN-transduced cells (not shown). Growth in semisolid medium The effect of PLZF expression on melanoma cell lines capacity of forming foci in semisolid medium was also investigated. As reported in Table 1, a 10-fold reduction of foci was observed for the PLZF-transduced A375, Me665/1 and Me1811 cell lines compared to the untransduced cells. No significant difference was observed in the total number of foci for Me1402/R cells, but Me1402/R/PLZF cells originated approximately 75% of melanotic colonies, whereas in parental cells most colonies were white thus indicating a more differentiated phenotype in PLZF-expressing cells. WM983A cell line, at least in our hands, was unable of growing in agar even in its parental condition. In vivo tumor growth To confirm PLZF function in regulating melanoma progression, we analysed its potential role in an in vivo Figure 2 PLZF and HOXB7 expression studies in LXSN- and PLZF-transduced melanoma cell lines. (a) RNase protection analysis. Riboprobes (R) and protected fragments (P) are shown on the right. Molecular size markers (MWM) are shown on the left. (b) Real-time RT–PCR. (c) Western blot analysis. Actin is the internal control. (d) Immunofluorescence analysis of LXSN- and PLZF-transduced A375 melanoma cells and normal human melanocytes Oncogene


4570 Table 1

Effects of PLZF expression on the number of colonies grown in semisolid medium

Melanoma cell lines Me665/1 A375 Me1811 Me1402/R

LXSN

PLZF

233.3747.1 108.079.2 285.1715.0 111.1713.2

27.372.0 13.274.3 26.074.3 78.077.2

over 6 weeks. As shown (Figure 5), both tumor take and volume of PLZF-expressing melanomas were clearly reduced when compared to control melanoma cell lines. Expression of PLZF-regulated genes

Figure 3 Growth kinetics of LXSN- and PLZF-transduced melanoma cell lines. Cell growth was monitored over the indicated time intervals by the XTT-based colorimetric assay, as indicated in Materials and methods. Values are reported as mean7s.e. of three separate experiments

Figure 4 Invasion (a) and chemotaxis (b) of LXSN-and PLZFtransduced melanoma cell lines. Results, expressed as % of migrated cells/well and standardized to empty vector-transduced cells, are reported as mean7s.e. Each experiment was performed in triplicate

model. LXSN- and PLZF-infected cells were inoculated s.c. into athymic nude mice for Me665/1 and Me1811 cell lines; tumor growths were compared and followed Oncogene

To find further information on PLZF effect on melanoma cells phenotype, we investigated genes differentially expressed in PLZF-negative vs PLZFpositive melanoma cells by probing the cDNA macroarray (Atlas Human Cancer 1.2, Clontech, containing 1176 human genes involved in cancer) with the A375/ LXSN and A375/PLZF cDNAs. Results are reported in Tables 2 and 3; the expression level of some selected genes was confirmed by RT–PCR. As expected, considering the PLZF repressor role, many of the modified genes were decreased. Particularly important for an experimental validation of the array was ITGb3 gene downregulation that we had observed at both mRNA and protein levels, as well as functionally through the reduced adhesion to vitronectin. Integrin avb3 is one of the most important cell surface receptors involved in melanoma progression, affecting cell survival, proliferation, invasion and migration (Seftor, 1998). RT–PCR studies on our panel of control and PLZF-transduced melanomas showed a clear and constant decreased expression of the b3 subunit at mRNA level in the PLZF-positive cells (Figure 6a). Integrin expression pattern was confirmed by Western blot analysis (Figure 6b). We observed decreased levels of both b3 and av integrin subunits in the total and cytoplasmic extracts obtained from PLZFpositive cells as compared to the empty vector-transduced melanoma cells (Figure 6b and not shown). In order to clarify these data, we analysed transduced and not transduced melanomas by immunofluorescence using anti-b3 and anti-avb3 Abs (Figure 6c and not shown). Figure 6c shows the expression of these cell surface receptors in the control melanoma cells and their strong reduction in PLZF-transduced cells. The same regulated pattern was observed by RT– PCR for osteonectin (SPARC) gene, known to be critical for melanoma progression and metastases (Sturm et al., 2002), which was downmodulated in PLZF-transduced melanoma cells (not shown). Since a coordinated interplay between adhesion molecules and metalloproteases has been reported in both vascular remodelling and tumor invasion (Hofmann et al., 2000), matrix metalloprotease (MMP)-2 and MMP-9 expression was analysed in the transduced vs control cell lines. Interestingly, MMP-9 was downregulated in all PLZFexpressing melanomas as compared with the empty


4571 Table 2 Primers, PCR products and annealing conditions Gene SKY RAD21 Erf-1 Integrin beta3 KIAA0022 TRP-1 SPARC MMP-2 MMP-9 gapdh

Upper primer (50 to 30 )

Lower primer (50 to 30 )

Product size (bp)

AT (1C)

gcaccagccagagagtcc cagcagcatcagcagcgtga aatccttctccaccgcacag atcgagttcccagtgagt accttccactcaactcccat tgaatggaacagggggacaa gctctgccttaaacacacat tcgcccatcatcaagttc aaccaatctcaccgacag acatcaagaaggtggtgaagcagg

gtgtggtaatagctgggc tgggggaagctctacaggtg ctaagtgtgtgggtctgttg tactggtgagctttcgca atggcgaaaccccgtctcta gtacactaaactcccgactt ttctctcagttacagcctca gtgatctggttcttgtcc caaaggcgtcgtcaatca ctcttcctcttgtgctcttgctgg

509 225 302 283 358 276 273 301 317 282

56 56 51 54 53 50 52 53 55 56

AT, annealing temperature

As listed in Table 3, the differential level of expression observed by the Atlas assay has been confirmed for some other genes by semiquantitative RT–PCR; in particular, a decreased expression has been confirmed for Sky tyrosine kinase receptor (Ohashi et al., 1994) and Erf-1 transcription factor (McPherson et al., 1997). On the contrary, RAD21, a factor involved in DNA repair and apoptotic cell death (Pati et al., 2002), and the anonymous cDNA KIAA0022 (Nomura et al., 1994) resulted in upregulation (not shown). Further analysis on these and other genes modulated by PLZF will be performed.

Discussion

Figure 5 In vivo tumor growth into athymic nude mice. Tumor take and volume at day 40 after s.c. injection of 5 106 cells for Me665/1 and 106 for Me1811 of LXSN- or PLZF-transduced melanoma cell lines. Data are representative of two independent experiments carried out with 10 mice for each cell line

vector-transduced cells, whereas MMP-2 was downregulated in three out of five cell lines, Me665/1/PLZF, Me1402/R/PLZF and Me1811/PLZF (Figure 7 and not shown). Another differentially expressed gene was the tyrosinase-related protein 1 (TRP-1), known to have a role in melanin synthesis (Cohen et al., 1990). Most melanoma cell lines become amelanotic during tumor progression and this loss of pigmentation correlates with the loss of expression of TRP-1. Interestingly, in all the analysed PLZF-transduced cell lines, we observed a clear increase of TRP-1 expression that could be associated with the observed regression of the neoplastic phenotype (Figure 7). Finally, cyclin A2, already reported as a target gene for PLZF repression in B lymphocytes, was not modulated in PLZF-transduced melanoma cells (Figure 7a).

Several molecular events occur during the transition from normal melanocytes to metastatic melanoma, through dysplastic nevi, RGP and VGP melanomas (Satyamoorthy and Herlyn, 2002). However, the mechanisms underlying this multistep progression have not been fully understood so far. PLZF silencing could represent one of these changes occurring during melanoma progression when the neoplastic cells acquire the ability to invade the dermis, intravasate and survive at distant sites. PLZF protein is a transcriptional regulator (Cook et al., 1995); when its physiological control is lost, PLZF seems to be involved in tumorigenesis (Parrado et al., 2000). Using knockout mice, PLZF has been demonstrated to regulate the spatial expression of the AbdB HoxD gene complex through cis regulatory elements and histone deacetylases recruitment (Barna et al., 2000). A balance between activating signal and transcriptional repressors, represented by retinoic acid and PLZF, regulates HoxD gene expression boundaries in the limb bud (Barna et al., 2002). We had previously reported the functional role of HOXB7 as a key factor for melanoma progression and activation of tumor-associated neo-angiogenesis (Care` et al., 2001). In particular, we found a clear increase, up to 10-fold, of VEGF, GROa, IL-8 and Ang-2 in the HOXB7-transduced cells, besides the induction of bFGF, whereas Ang-1 was totally abrogated (Care` Oncogene


4572 Table 3 Expression pattern obtained by array hybridization GENE MIG5 SKY CDKN1A S6KII-alpha3 S100 RAD 21 Erf-1 Integrin beta3 CDC34 MDA-7 HEM45 RGS4 KIAA0022 GENE gravin TRP-1 SPARC HLA-DR associated invariant chain HLA class IIa chain precursor

GenBank ID

Function

Ratio PLZF/LXSN

Z30183 D17517 U09579 U08316 M80563 D38551 U85658 J02703 L22005 U16261 U88964 U27768 D14664 M96322 X51420 J03040 X00497 K01171

Oncogene Tyrosine kinase receptor CDK inhibitor Intracellular kinase Calcium-binding protein DNA repair protein Transcription regulator Transmembrane receptor protein Protein turnover Apoptotic gene Translation Regulator of G-protein signalling Unknown A-kinase anchoring protein Signalling intermediate Calcium-binding glycoprotein Major histocompatibility complex class II Major histocompatibility complex class II

0.5 0.5 Down 0.6 Down 1.6 0.6 0.6 Up Down Up Up Up 0.6 Up 0.5 0.3 Down

RT–PCR

Differential gene expression was confirmed by RT–PCR when indicated by . The reported ratio indicates the observed fold increase or decrease of the A375/PLZF gene expression over A375/LXSN

et al., 1996, 2001). Because HOXB7 might represent a sort of master gene responsible of tumor progression, we were interested in searching an upstream inhibitory gene able to repress its function. PLZF has been described as a transcriptional repressor of HOX genes during embryogenesis and, considering the presence of a PLZF canonical binding site in HOXB7 promoter (Meccia et al., 2003), we hypothesized that PLZF could directly repress the transcription of HOXB7 and that the constitutive abnormal expression of HOXB7 in melanomas might result from the inactivation of PLZF. The presence of PLZF mRNA in normal melanocytes, and its absence in both melanoma primary lesions and cell lines, further suggested that a similar mechanism might operate in these tumors. However, our experimental results demonstrated that PLZF-expressing melanoma cell lines were not able to bind directly the PLZF putative sequence in the HOXB7 promoter and HOXB7 expression did not show the expected decrease (Figure 2a, b and not shown). It is possible that the coordinated regulation of PLZF and HOXB7 was lost in melanomas, as reported for PLZF and cyclin A in B-cell tumors (Parrado et al., 2000). In these leukemic cells, PLZF is 100-fold downregulated and it is no more able to control the cell cycle through repression of cyclin A as occurs in normal B lymphocytes. Recent data have suggested the existence of functional heterodimeric complexes containing different zinc-finger proteins, including PLZF (Hoatlin et al., 1999; Dhordain et al., 2000). According to our results, the incompetence of the ectopic PLZF in repressing HOXB7 in melanoma cells could also be due to the lack of its interacting partners. The in vitro and in vivo evidences of the less tumorigenic phenotype of PLZF-transduced melanoma cells suggested that we should search for other biological pathways independent of HOXB7. We used the Atlas Cancer macroarray assay (Clontech, Palo Alto, CA, Oncogene

USA) to profile a gene expression that might explain the reduced malignancy of transduced melanoma cells. As reported in Table 3, 18 genes resulted differentially expressed in the A375/LXSN and A375/PLZF compared cell lines when analyzed by the computer program Atlas Image 2.0. Particularly important for the validation of the array was ITGb3 downregulation in PLZFtransduced cells. The expression of integrin avb3, and in particular of the b3 subunit, correlates with tumor thickness and with the ability to invade and metastasize. ITGb3 adenovector-mediated gene transduction in melanomas induces a more aggressive phenotype associated with an increase in the osteonectin/SPARC level (Sturm et al., 2002). Osteonectin/SPARC was also found to be involved in melanoma progression, and its inhibition by antisense oligos considerably reduced tumor outgrowth, both in vitro and in vivo (Ledda et al., 1997), likely through the regulation of avb3. Melanoma cell adhesion is based on a balance between the expression of vitronectin receptor avb3 and that of vitronectin endogenously produced. Accordingly, in PLZF-transduced melanoma cells, we observed a downregulated expression of both INTb3 and osteonectin/ SPARC. PLZF induction of the melanocyte-specific gene TRP1 was also observed. In most melanoma specimens and cell lines, TRP-1 is silenced, possibly by a melanomaspecific repressor(s) that selectively inhibits the binding of the microphtalmia-associated transcription factor (MITF) to TRP-1 promoter (Fang et al., 2002). PLZF protein might compete with this unknown repressor for the same binding site or it might be able to downregulate the repressor itself. TRP-1 promoter studies will clarify this point. In conclusion our data show that several genes involved in melanoma progression are apparently coordinated, either directly or indirectly, by PLZF. We found PLZF and TRP-1 transcription in normal


4573

Figure 7 Gene regulation in LXSN- and PLZF-transduced melanoma cell lines. Traditional and Real-time RT–PCR analyses: MMP-9 and TRP-1 modulation was evaluated. CycA2 constant expression level is also shown. GAPDH housekeeping gene expression is the internal control

melanocytes, but not in melanoma cells, whereas the reciprocal pattern was observed for integrin avb3 (Van Belle et al., 1999), osteonectin/SPARC (Sturm et al., 2002) and MMP-9 (van den Oord et al., 1997) mRNAs that are present in melanomas and absent in melanocytic cells. PLZF-enforced expression in melanoma cells induces a less tumorigenic phenotype, through the upregulation of TRP-1 and the downregulation of avb3, osteonectin/SPARC and MMP-9. Of interest, none of these genes seems to be modulated by HOXB7, thus suggesting that PLZF- and HOXB7-coupled deregulation may account for most of the altered gene expression profile described in melanomas. Whether or Figure 6 INTavb3 expression analysis. (a) Real-time and traditional RT–PCR analysis of INTb3 expression in LXSN- and PLZF-transduced cell lines. () is the negative control. GAPDH housekeeping gene expression is shown as endogenous control. (b) Western blot analysis of INTav and INTb3 subunits in A375/ LXSN and A375/PLZF cell lines. Cytoplasmic and total extracts from A375/LXSN (lanes 1 and 3) and A375/PLZF (lanes 2 and 4) have been used. Actin is the internal control. (c) Immunofluorescence studies showed INTavb3 distribution on LXSN- and PLZFtransduced melanoma cell lines Oncogene


4574

not HOX genes other than HOXB7 can be regulated by PLZF outside embryogenesis remains to be investigated. Further analysis on PLZF-regulated genes is in progress in order to elucidate the specific role of PLZF in the molecular mechanisms underlying the normal melanocytic cell growth.

Materials and methods Cell line cultures and transduction The human melanoma cell lines used in the current study were described by Care` et al. (1996). The early passage melanoma cells were kindly provided by Dr S D’Atri (IDI IRCCS, Rome). Two cell lines were obtained from primary tumors (GR-mel, ST-mel) and three from lymph node metastases (CRmel, CN-mel, MR/C-mel) (Lacal PM et al., 2000; Alvino et al., 2002). Normal human epidermal melanocytes from foreskin were obtained from Promocell (Heidelberg, Germany). The PLZF cDNA encompassing its complete coding sequence was cloned into the retroviral vector LXSN. The LXSN empty vector was used as an internal control. The ecotropic GP þ E86 and the amphotropic GP þ env AM12 packaging cell lines were produced. Amphotropic cells were selected in the presence of 0.8 mg/ml of G418 plus 0.2 mg/ml of hygromycin and used to generate helper-free virus-containing supernatant. For target infection, melanoma cells were treated twice for 4 h with undiluted packaging cell supernatants in the presence of 8 mg/ml of polybrene. Cells were grown for 48 h and then selected with G418 (0.8 mg/ml). A375/PLZF clones were obtained by limiting dilutions; single clones were selected for PLZF expression level as demonstrated by RT–PCR, Western blot and immunofluorescence. Two A375/PLZF cellular clones (clone G3 and clone G9, PLZF high and low expressers, respectively), out of the 20 clones produced, were grown and utilized for functional studies. RNA analysis RT–PCR analysis and RNase protection were performed according to standard procedures. For traditional RT–PCR, the sequences of primers, the PCR products and the annealing conditions are reported in Table 2 and in Care` et al. (1996) and Labbaye et al. (2002). All the amplified fragments were hybridized to a 30-base end-labelled oligonucleotide according to Southern blot standard procedures. Real-time RT–PCR was performed by TaqMan technology, using the ABI PRISM 7700 DNA Sequence Detection System (Applied Biosystems, Foster City, CA, USA) according to standard procedures (Livak and Schmittgen, 2001). Thermal cycling was performed using 40 cycles of 951C for 15 s and 601C for 1 min. Original input RNA amounts were calculated with relative standard curves for both the RNAs of interest and the GAPDH control. Duplicate assays were performed with RNA samples obtained from at least two independent experiments. SEs smaller than 5% are not shown. Gene expression values are reported as the normalized quotient derived by dividing each single gene copy number by the GAPDH copy number. Commercial ready-to-use primers/ probe mixes were used (Assays on Demand Products, Applied Biosystems, Foster City, CA, USA). For the RNase protection assays, DNA fragments to be transcribed consisted of a specific PLZF sequence of 270 bp and a specific 139 bp sequence 30 to the box for HOXB7. The b-actin insert was a 93-bp RsaI fragment, always labelled in the Oncogene

presence of cold GTP to obtain a low specific activity. The expression levels were analysed by the Image Quant software (Molecular Dynamics, Sunnyvale, CA, USA). Western blot Ectopic PLZF protein was evaluated in nuclear cell lysates. The anti-PLZF polyclonal antibody (a kind gift of Prof. PG Pelicci, European Institute of Oncology, Milan, Italy) was diluted 1 : 3000 in TBST, 1% milk. The expression of integrin av and b3 subunits was evaluated in total and cytoplasmic lysates according to standard procedures. Anti-av and anti-b3 Abs (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) were diluted at 1 : 1000 and 1 : 250, respectively. As internal controls of total, nuclear and cytoplasmic proteins, actin (Oncogene Research Products, Boston, MA, USA), nucleolin (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) and tubulin (Sigma, Saint Louis, MI, USA) were used. Immunofluorescence Immunofluorescence was performed essentially as described. Briefly, cells were fixed with 2% paraformaldehyde in PBS and stained with anti-avb3 fluorescein-conjugated antibody. Mouse anti-human integrin avb3 (clone LM609) (Chemicon International, Temecula, CA, USA) was used at 1 : 300 in PBS–0.2% BSA. For the PLZF immunofluorescence experiments, cells were permeabilized with methanol, fixed with acetone and then incubated with anti-PLZF monoclonal Ab (Oncogene Research Products, Boston, MA, USA), diluted 1 : 20 in PBS–0.2% BSA. Finally, cells were stained with the goat anti-mouse IgG rhodamine conjugated antibody (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) and observed under a Olympus IX81 (Olympus, Tokyo, Japan) confocal microscope. EMSA EMSAs were performed as described (Meccia et al., 2003). An end labelled oligomer (3 104 cpm per sample) containing the PLZF putative binding sequence (GGGCAGGCATAAAGGTGGAGAGTGCAGGC) was bound to 10 mg of nuclear extracts obtained from PLZF- or vector-transduced melanoma cells. Binding reactions were incubated per 90 min at 41C; PLZF antibody, when present, was added to the mixtures 30 min before the other reagents. Transfections and CAT assays For the transactivation experiments, two different promoter fragments were utilized, corresponding to the whole HOXB7 promoter [1802/ þ 90] and, as negative control, to the 30 region [566/ þ 90], which does not contain putative PLZF DNA-binding sites (Meccia et al., 2003). A measure of 10 mg of the different constructs was transfected into the 293 cells by Lipofectamin 2000 Reagent (Life Technologies, Inchinnan, Scotland) and CAT activity was quantitated by an ELISA detection kit (Boehringer, Mannheim, Germany). bgal levels were used to normalize. In vitro growth assay The proliferative rate of parental and PLZF-transduced melanoma cells was evaluated by an XTT-based colorimetric assay (Roche Molecular Biochemicals, Mannheim, Germany). Briefly, cells were grown for different times in a 96-well tissue culture plate; wells were thus incubated with the yellow XTT solution for 3 h. After this incubation period, orange formazan


4575 solution is formed, which is spectrophotometrically quantified using an ELISA plate reader (Wallac VICTOR2, Turku, Finland). In vitro migration and invasion assay Migration was assayed, as previously described (Albini et al., 1987), using uncoated cell culture inserts (Corning Costar Corporation, Cambridge, MA, USA) with 8 mm pores. Cells (5 104 ) were placed in the upper compartment in 100 ml of serum-free DMEM , while 600 ml of DMEM supplemented with 10% FBS was placed into the lower compartment of the chamber. For invasion studies, membranes were coated with 100 mg/cm2 of Matrigel growth factor reduced (Becton Dickinson, Bedford, MA, USA) as a barrier, and 105 cells were placed in the upper compartment. Assays were incubated at 371C in 5% CO2. After 24 and 48 h, the cells attached to the upper side of the membrane were removed with a cotton swab; each membrane was fixed and stained with crystal violet solution (Niu et al., 2002). Chemotaxis and invasiveness were evaluated, as relative number of cells on the undersurface of the membrane, by a colorimetric assay at 595 nm in a microplate reader (Wallac VICTOR2, Turku, Finland). The data were expressed as the mean absorbance7s.e. for triplicate wells. Adhesion assay Cell adhesion assays were performed essentially as described (Boukerche et al., 1995). Briefly 96-well flat-bottom microtiter plates were coated with 150 ng/cm2 of matrigel and 50 or 200 ng/cm2 of vitronectin for 2 h at 371C; melanoma cell lines were seeded at a density of 104 cells/well in 200 ml of serum-free DMEM containing 0.35% BSA. Nonadherent cells were removed by gentle washing with PBS and the attached cells were fixed and stained with crystal violet solution as in the in vitro migration and invasion assays.

cDNA expression array Commercially available cDNA expression arrays (Atlas human cancer cDNA expression array; Clontech, PaloAlto, CA, USA) were used to compare the gene expression of a melanoma cell line, retrovirally transduced with LXSN or with LPLZFSN constructs. Arrays were screened according to the manufacturer’s protocol. Total RNA (15 mg) from each cell line was radiolabelled using MMLV reverse transcriptase and a specific primer set. Probes were purified, the specific activity measured and 5 106 cpm/ml used to hybridize the filters overnight at 681C. Blots were then washed three times with 2 SSC/0.5% SDS at 681C and twice with 0.1 SSC/0.5% SDS. The damp membrane was wrapped immediately and exposed both to an X-ray film and to a phosphorimager screen. The Atlas Image 2.0 computer program was used to interpret the results. In vivo assay LXSN- or PLZF-transduced melanoma cells in exponential growth phase were injected s.c. at doses of 106 and 5 106 into adult athymic nude mice purchased from Charles River (Calco, Italy) and maintained at the Istituto Nazionale Tumori (Milan, Italy) under standard conditions according to institutional guidelines. Tumor growths were monitored twice a week for at least 6 weeks. Points represent the mean tumor size (as measured by three perpendicular diameters)7s.e.

Abbreviations HOX, homeobox; PLZF, promyelocytic leukemia zinc finger; Abd-B, Abdominal-B; MMP, matrix metalloprotease; TRP-1, tyrosinase related protein-1; APL, acute promyelocytic leukemia; RAR, retinoic acid receptor; B-CLL, B-chronic lymphocytic leukemia; RGP, radial growth phase; VGP, vertical growth phase.

Growth in semisolid medium Base layers of complete DME medium containing 0.5% agar were set in 60 mm plastic dishes. The bottom agar was overlaid with 1.5 ml of 0.33% agar containing suspensions ranging from 103 to 104 cells. Cultures were incubated for 3–4 weeks at 371C and the colonies counted using an inverted microscope. Two experiments were performed for each cell line and results were calculated as the average7s.e. of three dishes for each condition.

Acknowledgements PG Pelicci is acknowledged for providing us with PLZF polyclonal antibody and S D’Atri for some melanoma samples. We thank G Loreto for the preparation of the figures. This work was supported by The Italy-USA Special Project on Therapy of Tumors and by The Italian Ministry of Health. FF was supported by a fellowship from the Italian Foundation for Cancer Research (FIRC).

References Abate-Shen C. (2002). Nat. Rev. Cancer, 2, 777–785. Albini A, Iwamoto Y, Kleinman HK, Martin GR, Aaronson SA, Kozlowski JM and McEwan RN. (1987). Cancer Res., 47, 3239–3245. Alvino E, Marra G, Pagani E, Falcinelli S, Pepponi R, Perrera C, Haider R, Castiglia D, Ferranti G, Bonmassar E, Jiricny J, Zambruno G and D’Atri S. (2002). J. Invest. Dermatol., 118, 79–86. Barna M, Hawe N, Niswander L and Pandolfi PP. (2000). Nat. Genet., 25, 166–172. Barna M, Merghoub T, Costoya JA, Ruggero D, Branford M, Bergia A, Samori B and Pandolfi PP. (2002). Dev. Cell, 3, 499–510. Boukerche H, Berthier-Vergnes O, Tabone E, Bailly M, Dore JF and McGregor JL. (1995). Br. J. Cancer, 72, 108–116.

Care` A, Felicetti F, Meccia E, Bottero L, Parenza M, Stoppacciaro A, Peschle C and Colombo MP. (2001). Cancer Res., 61, 6532–6539. Care` A, Silvani A, Meccia E, Mattia G, Peschle C and Colombo MP. (1998). Oncogene, 16, 3285–3289. Care` A, Silvani A, Meccia E, Mattia G, Stoppacciaro A, Parmiani G, Peschle C and Colombo MP. (1996). Mol. Cell. Biol., 16, 4842–4851. Chen Z, Brand NJ, Chen A, Chen SJ, Tong JH, Wang ZY, Waxman S and Zelent A. (1993). EMBO J., 12, 1161–1167. Cohen T, Muller RM, Tomita Y and Shibahara S. (1990). Nucleic Acids Res., 18, 2807–2808. Cook M, Gould A, Brand N, Davies J, Strutt P, Shaknovich R, Licht J, Waxman S, Chen Z, Gluecksohn-Waelsch S, Oncogene


4576 Krumlauf R and Zelent A. (1995). Proc. Natl. Acad. Sci. USA, 92, 2249–2253. Dhordain P, Albagli O, Honore N, Guidez F, Lantoine D, Schmid M, De The H, Zelent A and Koken MHM. (2000). Oncogene, 19, 6240–6250. Dong S, Zhu J, Reid A, Strutt P, Guidez F, Zhong HJ, Wang ZY, Licht J, Waxman S, Chomienne C, Chen Z, Zelent A and Chen S-J. (1996). Proc. Natl. Acad. Sci. USA, 93, 3624–3629. Dubrulle J and Pourquie O. (2002). Curr. Opin. Genet. Dev., 12, 519–523. Fang D, Tsuji Y and Setaluri V. (2002). Nucleic Acids Res., 30, 3096–3106. He LZ, Guidez F, Tribioli C, Peruzzi D, Ruthardt M, Zelent A and Pandolfi PP. (1998). Nat. Genet., 18, 126–135. Hoatlin ME, Zhi Y, Ball H, Silvey K, Melnick A, Stone S, Arai S, Hawe N, Owen G, Zelent A and Licht JD. (1999). Blood, 94, 3737–3747. Hofmann UB, Westphal JR, Waas ET, Becker JC, Ruiter DJ and van Muijen GN. (2000). J. Invest. Dermatol., 115, 625– 632. Krug U, Ganser A and Koeffler HP. (2002). Oncogene, 21, 3475–3495. Labbaye C, Quaranta MT, Pagliuca A, Militi S, Licht JD, Testa U and Peschle C. (2002). Oncogene, 21, 6669–6679. Lacal PM, Failla CM, Pagani E, Odorisio T, Schietroma C, Falcinelli S, Zambruno G and D’Atri S. (2000). J. Invest. Dermatol., 115, 1000–1007. Ledda MF, Adris S, Bravo AI, Kairiyama C, Bover L, Chernajovsky Y, Mordoh J and Podhajcer OL. (1997). Nat. Med., 3, 171–176. Li JY, English MA, Ball HJ, Yeyati PL, Waxman S and Licht JD. (1997). J. Biol. Chem., 272, 22447–22455. Licht JD, Chomienne C, Goy A, Chen A, Scott AA, Head DR, Michaux JL, Wu Y, DeBlasio A, Miller Jr WH, Zelenetz AD, Willman CL, Chen Z, Chen S-J, Zelent A, Macintyre E,

Oncogene

Veil A, Corets J, Kantarjian H and Waxman S. (1995). Blood, 85, 1083–1094. Livak KL and Schmittgen TD. (2001). Methods, 25, 402–408. McPherson LA, Baichwal VR and Weigel RJ. (1997). Proc. Natl. Acad. Sci. USA, 94, 4342–4347. Meccia E, Bottero L, Felicetti F, Peschle C, Colombo MP and Care` A. (2003). Biochem. Biophys. Acta, 1626, 1–9. Niu J, Dorahy DJ, Gu X, Scott RJ, Draganic B, Ahmed N and Agrez MV. (2002). Int. J. Cancer, 99, 529–537. Nomura N, Miyajima N, Sazuka T, Tanaka A, Kawarabayasi Y, Sato S, Nagase T, Seki N, Ishikawa K and Tabata S. (1994). DNA Res., 1, 27–35. Ohashi K, Mizuno K, Kuma K, Miyata T and Nakamura T. (1994). Oncogene, 9, 699–705. Parrado A, Noguera ME, Delmer A, McKenna S, Davies J, Le Gall I, Bentley P, Whittaker JA, Sigaux F, Chomienne C and Padua RA. (2000). Hematol. J., 1, 15–27. Pati D, Zhang N and Plon SE. (2002). Mol. Cell. Biol., 22, 8267–8277. Reid A, Gould A, Brand N, Cook M, Strutt P, Li J, Licht J, Waxman S, Krumlauf R and Zelent A. (1995). Blood, 86, 4544–4552. Satyamoorthy K and Herlyn M. (2002). Cancer Biol. Ther., 1, 14–17. Seftor RE. (1998). Am. J. Pathol., 153, 1347–1351. Sturm RA, Satyamoorthy K, Meier F, Gardiner BB, Smit DJ, Vaidya B and Herlyn M. (2002). Cancer Res., 62, 226–232. Van Belle PA, Elenitsas R, Satyamoorthy K, Wolfe JT, Guerry 4th D, Schuchter L, Van Belle TJ, Albelda S, Tahin P, Herlyn M and Elder DE. (1999). Hum. Pathol., 30, 562–567. van den Oord JJ, Paemen L, Opdenakker G and de WolfPeeters C. (1997). Am. J. Pathol., 151, 665–670. Yeyati PL, Shaknovich R, Boterashvili S, Li J, Ball HJ, Waxman S, Nason-Burchenal K, Dmitrovsky E, Zelent A and Licht JD. (1999). Oncogene, 18, 925–934.