Oncogene (1998) 16, 3057 ± 3068  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Transition from SCLC to NSCLC phenotype is accompanied by an increased TRE-binding activity and recruitment of speci®c AP-1 proteins

Gundula Risse-Hackl1,JuÈrgen Adamkiewicz2, Anja Wimmel1 and Marcus Schuermann1

1Zentrum fuÈr Innere Medizin, Abteilung HaÈmatologie/Onkologie, Philipps-UniversitaÈt Marburg, Baldingerstrasse, D-35033 Marburg; 2Institut fuÈr Molekularbiologie und Tumorforschung (IMT), Philipps-UniversitaÈt Marburg, Emil-Mannkop€-Str. 2, D-35037 Marburg, Germany

Transitions from small cell (SCLC) to non-small cell lung lung cancer are supposed to originate from di€erent cancer (NSCLC) cells have been documented both in vitro cell types of the bronchial epithelium and alveoli. In and in vivo and are thought to be an important step during clinical practise, however, lung cancer frequently tumor progression of human small cell lung cancer exhibits more than one histologic pattern (Kalemker- towards a treatment-resistant tumor state. We have ian and Mabry, 1993; Mabry et al., 1988, 1991; Roggli screened NSCLC and SCLC cell lines for di€erences in et al., 1985). Thus, mixtures of cells with SCLC and the composition of nuclear transcription factors using NSCLC histologies appear in SCLC tumors and, consensus oligonucleotide sequences (SRE, Ets, TRE, moreover, both SCLC and NSCLC phenotypic CRE, B-motif, GAS, E-box). We found NSCLC cells to markers have been detected simultanously in indivi- exhibit signi®cantly higher AP-1 binding activity than dual cells. These observations have led to the SCLC cells consistent with the increased expression of hypothesis that in vivo transitions from an SCLC to CD44, an AP-1 target gene. To gain more insight into the an NSCLC phenotype might occur. Further evidence molecular mechanisms underlying these di€erences, we in favour of this hypothesis came from in vitro analysed SCLC cell lines (NCI-N592 and NCI-H69) experiments in which overexpression of activated H- which were phenotypically transformed into NSCLC-type ras and c-myc oncogenes in SCLC cell lines resulted in cells by transfection with activated H-ras and c-myc the loss of SCLC-speci®c markers and gain of oncogenes. In these cells, ras-induced transition is phenotypic characteristics resembling those of NSCLC accompanied by a strong induction of AP-1-binding cells (Mabry et al., 1988, 1991 BuÈ rger et al., 1994). activity along with increased expression of CD44 mRNA Such cell type transitions may be crucial for the and protein. When analysing the composition of the AP-1 progression of each lung cancer type and its resistance complex in more detail and comparing ras-induced versus to current modes of therapy. phorbol ester-induced changes, we found Fra-1 to be the Although the molecular events underlying these cell major component induced in ras-transfected but not in type transitions have not yet been elucidated, it is likely phorbol-ester treated or non-treated parental SCLC cells. that distinct transcription factors are involved in this This ®nding is paralleled by the observation that among process. This is exempli®ed by the aforementioned in the various members of the Fos and Jun family analysed vitro observation that the combined action of H-ras (c-Fos, FosB, Fra-1, Fra-2, c-Jun, JunD, JunB) fra-1 is and c-Myc is sucient to promote `transdi€erentiation' the only gene to be exclusively expressed in NSCLC cells into a di€erent phenotype. c-Myc itself is a member of but not in cells of SCLC origin. Our data, thus, point to a a larger family of transcription factors binding to the histiotype-related mechanism of recruitment among AP-1 E-box element (for review see Amati and Land, 1994; proteins which may have bearings on the fate of lung Bernards, 1995) while H-ras as a signal transduction cancer development. molecule transmits its activity via the Raf-MAP kinase pathway (reviewed in Karin and Hunter, 1995; Keywords: AP-1; Jun; Fos; Fra-1, Ras; lung cancer; Marshall, 1995) and modulates the activity of distinct transdi€erentiation nuclear transcription factors, such as the activity of AP-1, Ets/TCF and NF-kB proteins. Alteration of AP- 1 activity has been particularly associated with increased Ras activity and therefore has been Introduction extensively characterized in the past (for review see Angel et al., 1987; Hill and Treisman, 1995; Marshall, Lung cancer is the leading cause of tumor-related death 1995). AP-1 consists of proteins of the Jun and Fos in both men and women. Tumors arising in the lung families which associate as homo-(Jun/Jun) or hetero- form a heterogenous group by histology which can be dimers (Fos/Jun; reviewed in Angel and Karin, 1991; classi®ed into four major subgroups: small cell Cohen and Curran, 1990). To date, three Jun proteins carcinoma (SCLC), accounting for 25% of human (c-Jun, JunB and JunD) and four Fos family members lung tumors, squamous cell carcinoma, large cell (c-Fos, FosB, Fra-1 and Fra-2) have been identi®ed as carcinoma and adenocarcinoma, collectively termed components of the AP-1 factor (Bohmann et al., 1987; non-small cell lung carcinoma (NSCLC). Due to their Cohen and Curran, 1988; Hirai et al., 1989; Matsui et di€erent phenotypic properties, the various types of al., 1990; Nakabeppu et al., 1988; Nishina et al., 1990; Ryder et al., 1989; Zerial et al., 1989). Fos/Jun hetero- or Jun/Jun homodimers recognize a cis-acting element, the phorbol 12-myristate 13-acetate (TPA)-responsive Correspondence: M Schuermann Received 22 October 1997; revised 19 January 1998; accepted 20 element (TRE: TGACTCA), which is found in a January 1998 number of cellular promoters such as the genes coding Histiotype-dependent recruitment of Fra-1 in lung cancer GRisse-Hacklet al 3058 for interstitial collagenase, transin, typeIV collagenase, component in various di€erentiation processes invol- procathepsin L or CD44 (Angel et al., 1987; Hofmann ving growth arrest (Dixit et al., 1993; Melamed et al., et al., 1993; Kerr et al., 1988; Sato and Seiki, 1993; 1993) or apoptosis (Estus et al., 1994; Goldstone and Troen et al., 1991). Lavin, 1994; Smeyne et al., 1993). Most of these e€ects While AP-1 activity is generally associated with have been reported in cell systems of diverge tissue proliferation (Angel and Karin, 1991; Kovary and origin and thus, the function of AP-1 seems to be Bravo, 1992) it is also an important signaling dependent not only on the pathway of signal

a NSCLC b SCLC NHBE 32M1 U1752 97TM1 103H U1810 H23 H69 H82 H841 H146 H187 86M1 DMS79 – + ––––––++++++ PMA – + ––––––++++++ PMA

fp — fp — 12345678910 11 12 13 14 12345678910 11 12 13 14

H125 H820 H1573 H2009 H2077 H2126 H526 H1092 H510 SW210 ––––––++++++ PMA PMA ––––++++

fp — fp —

15 16 17 18 19 20 21 22 23 24 25 26 15 16 17 18 19 20 21 22

c SCLC NSCLC

H69 H82 H841 H146 32M1 103H H1573 H2077 PMA ––––––––+ + ++++ + +

Figure 1 AP-1 DNA binding activity in NSCLC (a) and SCLC cell lines (b). Whole cell extracts (10 mg) were prepared from NHBE cells and NSCLC and SCLC cell lines as described under Materials and methods either in the presence (+) or absence (7) 78 of 5610 M PMA added for 24 h (indicated above the lanes). Aliquots were incubated with 32P-labeled, double-stranded consensus AP-1 DNA binding sequence. The bound complexes were resolved on a 4% polyacrylamide gel (fp=free probe). (c) Same experimental approach using a double-stranded consensus fp — sequence for SP-1 as control for equal quality of extracts (shown only for eight cell lines) Histiotype-dependent recruitment of Fra-1 in lung cancer G Risse-Hackl et al 3059 transduction itself but also on the context of cellular di€erentiation. a NSCLC In this study, we have screened a number of transcription factors with respect to their binding 97TM1 32M1 H125 H820 103H H23 U1752 U1810 potential in lung cancer cell lines of di€erent NHBE H1573 H2009 H2126 H2077 histological origin. Among these, we found AP-1 to be a protein complex which is abundantly present in CD44 — NSCLC but not in SCLC cells. Using H-ras/c-myc transfected SCLC cells, we can further show that upon transition from SCLC- to NSCLC phenotype these cells also gain strong AP-1 DNA binding capacity along with GAPDH — elevated AP-1 dependent gene expression. The strong TRE-binding activity in these NSCLC-like cells is largely achieved by the recruitment of Fra-1 into the complex the expression of which also is a hallmark of b SCLC NSCLC cells. This ®nding, thus, may point to a histiotype-related mechanism of regulation among 24H SW210 86M1 H841 H146 H187 H526 H510 H69 H82 members of the AP-1 family in lung cancer cells. DMS79 H1092

CD44 — Results

Comparison of sequence-speci®c DNA-binding activities GAPDH — in NSCLC and SCLC cell lines

A collection of 12 NSCLC and 11 SCLC cell lines was Figure 2 Expression of CD44 mRNA in lung cancer cell lines as examined in electrophoretic mobility shift assays analysed by RT ± PCR. Total RNA was isolated from the (EMSA) for the presence of proteins binding to indicated cell lines and reverse transcribed. Aliquots of the resulting cDNA were then used as a template for PCR di€erent synthetic consensus sequences. The latter ampli®cation using gene-speci®c primers. [33P]dCTP was added either corresponded to consensus sites for nuclear to the PCR reaction to allow the sensitive detection of ampli®ed factors involved in the Ras-MAP kinase signaling DNA fragments using only a limited number of PCR cycles. The pathway such as the TRE (TPA-responsive element, reaction products were electrophoresed on 6% polyacrylamide gels and visualized by autoradiography. In case of CD44 a 166 bp also referred to as AP-1 binding site), Ets element fragment (CD44 non-spliced transmembrane region) is ampli®ed (binding site of TCF), the SRE (serum-responsive (25 PCR cycles). GAPDH transcription served as control for element), E-box (binding site of Myc/Max proteins), equal cDNA content (300 bp fragment, 20 PCR cycles) kB site (binding site of NF-kB) or to transcription factors involved in other signal transduction pathways such as the CRE (cAMP-responsive element) and GAS element (binding site of STAT factors). While the are induced to adopt properties of an NSCLC binding activities for most of these target sequences did phenotype by transfection and thus overexpression of not vary signi®cantly between NSCLC and SCLC cells c-myc and activated H-ras oncogenes (BuÈ rger et al., (data not shown), we observed a marked di€erence in 1994). The transfected cells are characterized by the level of AP-1 DNA binding activity being abundant distinct NSCLC-like properties such as adherent in the majority of the NSCLC cell lines but barely growth, gain of expression of NSCLC-speci®c genes detectable in most of the SCLC cell lines analysed and loss of expression of SCLC-speci®c genes (BuÈ rger (Figure 1). This ®nding prompted us to further et al., 1994). Comparing extracts from transfected and investigate whether the level of AP-1 complex found non-transfected cells in analogous EMSA experiments, would also correlate with functional transcriptional we found that the TRE-binding capacity is highly activity in these cells. As an example, we monitored the upregulated upon transition from the SCLC to the H- expression of the CD44 gene by RT ± PCR since this ras/c-myc induced NSCLC phenotype. (Figure 3a). gene is expressed in cells of epithelial origin and known Competition experiments using a mutant TRE- to harbor an AP-1 consensus element within the oligonucleotide indicated that the AP-1 DNA binding proximal promoter region (Ho€mann et al., 1993). As activity present in the transfected cells was speci®c for shown in Figure 2, the majority of NSCLC cells but not this sequence (Figure 3b). To make sure that the of SCLC cells revealed clearly detectable CD44 mRNA di€erences observed for the parental and transfected levels which largely corresponded with the relative cells with respect to their TRE-binding activities were amount of TRE-binding activity in this cell-type. not due to a reduced quality of the extracts, we performed EMSA using a consensus oligonucleotide for the transcription factor SP-1 instead of AP-1 TRE-binding activity and AP-1 dependent gene (Figure 3c). Next, we also monitored the extent of transcription in SCLC cells phenotypically transformed CD44 mRNA induction as previously done for to NSCLC NSCLC and SCLC lung cancer cell lines. As To gain more insight into the molecular mechanisms anticipated, all three H-ras/c-myc transfected sub- underlying these histiotype-borne di€erences, we took clones reveal a markedly increased expression of advantage of an established in vitro system, in which CD44, contrary to a non-adherent clone transfected SCLC cell lines (cell lines NCI-N592 and NCI-H69) with the empty expression vector (N592C, Figure 4a). Histiotype-dependent recruitment of Fra-1 in lung cancer GRisse-Hacklet al 3060 Interestingly, the extent of gene induction correlates parental and transfected N592 and H69 cell lines by with the amount of nuclear factors binding to the immunohistochemistry. As shown in Figure 4c, tumor TRE-consensus element (cf. Figures 3 and 4; tissue derived from transfected NSCLC-like subclones H69.54H69.444H69; N592.544N592). Finally, exhibits a much stronger CD44-related membrane we examined the level of CD44 glycoprotein present staining than tissue of the non-transfected counter- at the cell surface in order to test whether the parts, which contains only marginally detectable CD44 di€erence in AP-1 content would lead to persistent immunoreactivity. Thus, in these cells, AP-1 content, altered protein expression. This time we analysed transcription of AP-1 target genes and subsequent sections of xenograft tumors derived from both protein expression seemed to be closely linked.

a probe TRE mut. TRE TRE mut. TRE H69.4 H69.5 H69.4 H69.5 H69 H69 probe probe probe probe N592.5 N592.5 N592 N592 N592C N592C

AP1 —

fp —

12345678 9 10 11 12 13 14 15 16

bc Competitor probe SP1 AP-1 Random × × × × × × H69.5 H69 H69.4 probe N592.5 N592C N592 No Competitor 1000 1000 100 100 10000 10000

fp —

Figure 3 AP-1 DNA binding activity in parental and H-ras/c-myc transfected SCLC cell lines. (a) Whole cell extracts prepared from parental and H-ras/c-myc transfected NCI-N592 and NCI-H69 cells were analysed for AP-1 DNA binding activity as described in the legend to Figure 1 (N592C=mock transfected clone, N592.5, H69.4 and H69.5=H-ras/c-myc transfected subclones of N592 and H69, respectively). Probes indicated above the lanes correspond to either the consensus TPA-responsive element (TRE) or a mutant TRE form (mut. TRE) lacking speci®c AP-1 protein binding activity (Risse et al., 1989). Speci®c retarded DNA-protein complexes are indicated by `AP-1'. (b) Competition experiments using a 100, 1000 and 10 000-fold excess of the unlabeled corresponding TRE olgonucleotide (AP-1) or of an unmatching unlabeled sequence (Random). (c) Binding to the transcription factor SP-1 consensus sequence as a control for the quality of the di€erent whole cell extracts used Histiotype-dependent recruitment of Fra-1 in lung cancer G Risse-Hackl et al 3061 JunD, c-Fos, FosB, Fra-1 and Fra-2. These antisera had Characterization of the AP-1 complex induced in c-myc been raised against fusion proteins linked to b- and Ha-ras transfected SCLC cells galactosidase or against peptide epitopes of di€erent Having established a correlation between NSCLC-type Fos and Jun proteins and had been previously shown to and augmented AP-1 activity, we next examined the speci®cally interfere with the formation of Fos/Jun composition of the AP-1 complex more precisely by complexes in electrophoretic mobility shift assays performing AP-1 DNA binding assays in the presence (Adamkiewicz et al., 1990, 1993). We ®rst incubated of inhibiting or supershifting antibodies directed against extracts of H-ras/c-myc transfected SCLC cells with Jun and Fos family proteins, i.e. against c-Jun, JunB, these di€erent antisera. Figure 5a shows that several antisera a€ected the binding potential signi®cantly. As an example, antibody a Pan AP-1, directed against the highly conserved DNA-binding domain of AP-1 a proteins, almost completely inhibited the formation complexes in N592.5 as well as in H69.5 cells (data only shown for clone N592.5). Two antisera directed against c-Fos, a c-Fos (455) and a c-Fos (P3), did not interfere with the formation of speci®c protein-DNA complexes H69.4 H69.5 N592C N592 primer control H69 N592.5 while with a Fos-B antibody a weak supershifting activity could be observed. Three antisera directed — CD44 against c-Jun, on the other hand, largely interferred with DNA binding. Antisera K30 (rabbit polyclonal against entire c-Jun), jp 5/6 (rabbit polyclonal against c- Jun DNA binding region) strongly inhibited complex — GAPDH formation while another antiserum to c-Jun (Santa Cruz Biotechnology Inc.) supershifted part of the TRE- protein complex. Similar e€ects were seen with antibodies directed against JunD and Fra-1. Antisera directed against JunB and Fra-2 had no detectable b e€ect. From these data we can conclude that the AP-1 DNA binding complex in extracts from transfected NSCLC-like cells is mainly composed of c-Jun, JunD and Fra-1 (data summarized in Table 1). In order to be able to distinguish ras-induced alterations of AP-1 components in these cells from the `normal' AP-1 content, we next analysed the AP-1 complex in the parental H69 and N596 cell lines using the same approach. As in these cells the basal level of TRE-binding was very low, we used extracts from cells which had been stimulated with PMA and therefore exhibited an induced TRE-binding activity. A compar- able level of total AP-1 protein could be achieved only in N592 cells following 6 h of PMA treatment (Figure c N592 N592.5 5b). Under these conditions, we found c-Fos, FosB and c-Jun to be the major components of the AP-1 complex (Figure 5a, lower panel). JunD and Fra-1 could not be detected in the AP-1 DNA binding complex. Finally, in order to be able to directly compare the composition of the AP-1 complex in parental and H-ras/c-myc transfected subclones, we also analysed extracts of PMA-stimulated N592.5, H69.4, and H69.5 cells (results for clone N592.5 shown in Figure 5a, middle panel). As in case of the non-stimulated transfected Figure 4 Upregulation of CD44 transcripts and CD44 protein in cells, the major components of the AP-1 complex were H-ras/c-myc transfected cells compared to the non-transfected c-Jun, JunD and Fra-1 while an induction of c-Fos and SCLC counterparts. (a) mRNA expression of CD44 was analysed FosB activity could not be observed. This ®nding as described in the legend to Figure 2, for nomenclature of cell lines see legend to Figure 3(a). (b) Densitometry of the therefore indicates that transfection by H-ras and c- autoradiography in (a). CD44 mRNA levels are 7 ± 10-fold more myc resulted in a persistent change in the composition abundant in H-ras/c-myc transfected SCLC cells than in the of AP-1 proteins along with a decreased sensitivity to parental cells. (c) CD44 expression at the surface of H-ras/c-myc PMA. A summary of these ®ndings is given in Table 1. transfected SCLC and parental cells analysed by immunostaining Although one has to take into account the di€erent (shown only for cell line N592 and subclone N592.5). H69, N592 and H-ras/c-myc transfected subclones were grown as xenografts anities of the antisera used it becomes evident from in nu/nu mice for 21 days. Indirect immuno¯uorescence was these data that stimulation by PMA and H-ras/c-myc performed on 4 mm sections of the resulting tumors using a mouse overexpression have markedly di€erent e€ects with monoclonal antibody against the CD44 standard form (clone regard to the recruitment of speci®c AP-1 proteins. SFF-2) and alkaline phosphatase conjugated rabbit a-mouse second antibody. Counterstaining was performed with Hoechst To further test whether the change in AP-1 activity 33258 dye (stained nuclei) and composition in transfected SCLC cells was Histiotype-dependent recruitment of Fra-1 in lung cancer GRisse-Hacklet al 3062 occurring at the level of protein or gene activity we survey are shown in Figure 6. While transcripts coding analysed mRNA levels coding for the di€erent AP-1 for c-Jun, JunB and Fra-2 are comparable among constituents by RT ± PCR analysis. The results of this transfected and non-transfected cell lines, minor

a Pan AP-1 (P10) c-Fos (455) c-Fos (P3) FosB (K30) c-Jun (K31) c-Jun (jp5/6) c-Jun* JunD* JunB* Fra-1* Fra-2* α α α α α α α α α α α No Antibody Rabbit lgG

N592.5 basel —

N592.5 6 h PMA —

N592 6 h PMA —

12345678910 11 12 13

b N592 N592.5 H69 H69.4 24 h 24 h 24 h 24 h PMA 0 h 6 h 0 h 6 h 0 h 6 h 0 h 6 h

fp —

Figure 5 Composition of AP-1 complexes in N592 SCLC and H-ras/c-myc transfected N592.5 derivatives as compared to phorbol ester treated analogous cells. (a) Characterization of the AP-1 complex in the transfected subclones and the parental cells by supershift analysis. Whole cell extracts derived from non-stimulated N592.5 subclone, PMA-treated N592.5 or PMA-treated N592 cells were preincubated with the respective anity puri®ed antibodies prior to EMSA analysis. Rabbit IgG: protein A puri®ed preimmune rabbit IgG. a Pan AP-1(P10), rabbit antibody directed against oligopeptide P10, covering the DNA binding region conserved within all Fos family members; a c-Fos(455), rabbit antibody directed against a b-gal/Fos fusion protein; a c-Fos(P3), rabbit antibody directed against Fos oligopeptide P3; a FosB(K30): rabbit antibody directed against a b-gal/FosB fusion protein; a c-Jun(K31): rabbit antibody directed against a b-gal/c-Jun fusion protein; a c-Jun(jp5/6): rabbit antibody directed against Jun oligopeptide jp5/6. All antibodies indicated by an asterisk are peptide-speci®c, anity-puri®ed antibodies especially recommended for gel shift analysis (purchased from Santa Cruz Biotechnology). Similar results were obtained for subclones H69.4 and H69.5. (b) Induction of AP-1 binding activity in N592, H69 and H-ras/c-myc transfected clones in response to PMA treatment. The cell lines 78 indicated were stimulated with PMA (5610 M) for 6 h and 24 h, respectively. Seen is the induction of AP-1 binding activity in cell line N592 (maximum after 6 h) and transfected derivatives of both cell lines Histiotype-dependent recruitment of Fra-1 in lung cancer G Risse-Hackl et al 3063 Table 1 Level and composition of AP-1 binding complexes in H-ras/c-myc transfected (N592.5) versus phorbol ester treated N592 SCLC cellsa cFos FosB Fra1 Fra2 cJun JunD JunB N592.5 7 7+ ++++ 7 ++++ ++ 7 N592.5, 6 h PMA + + ++++ 7 ++++ ++ 7 N592, 6 h PMA ++++ ++++ 7 7 ++++ 7 7 aValues are based on autoradiographic data shown in Figure 5. Symbols denote the degree of antibody-mediated binding inhibition as determined by densitometry: (7) no change, (+) 1 ± 5%, (++) 5 ± 15%, (++++) 550% inhibition H69.4 H69.4 H69.4 H69.4 H69 H69 N592.5 N592.5 N592 N592 N592C N592C

— c-jun c-fos —

— junD fosB —

— junB fra-1 —

— GAPDH fra-2 —

Figure 6 Expression of fos and jun family members in N592, H69 cells and derived H-ras/c-myc transfected subclones. RT ± PCR analysis was performed as described in the legend to Figure 2 over 30 cycles for ampli®cation of fos and jun cDNA and 20 cycles in case of GAPDH. For nomenclature of cell lines see legend to Figure 3(a). Due to the location of the primers used two fra-1 fragments representing naturally occurring transcripts have been detected (Matsui, et al., 1990). Parental N592 and H69 SCLC cells do not exhibit detectable fra-1 transcription

di€erences in the level of expression are observed for c- cell lines, we monitored the level of expression of all fos and junD and signi®cant di€erences with respect to AP-1 family members in the NSCLC and SCLC cell fosB and fra-1 mRNA. Variations in junD, c-fos and lines which had been analysed for their TRE-binding fosB mRNA levels, however, do not coincide completely capacity before. As method of choice we performed with the H-ras/c-myc-induced phenotype since e.g. c-fos RT ± PCR analysis in order to look at di€erences in is upregulated also in cells transfected with a mock the activity of the respective genes. The result of this plasmid (N592C), fosB is downregulated in the same survey is given in Figure 7. While no marked cells, and junD expression not induced in H69.4 cells. di€erences were observed for the expression of c- The most obvious, speci®c change is seen with respect to fos, fra-2, c-jun and junB, we found a generally fra-1 mRNA expression. mRNA coding for Fra-1 is elevated number of transcripts coding for junDin hardly detectable in the parental cell lines N592 and most of the NSCLC cells. We also noted varying H69 and the control cells N592C but signi®cantly expression with respect to fosB mRNA ranging from induced in H-ras/c-myc-transfected subclones N592.5, no detectable expression to relative abundance in H69.4 and H69.5. This result is consistent with the both SCLC and NSCLC cell lines. Again, the most previous ®nding that the AP-1 complex in these three obvious di€erence between NSCLC and SCLC cells subclones but not in the parental cells contains was seen with respect to fra-1 expression. mRNA functionally active Fra-1. Thus, fra-1 expression coding for Fra-1 was practicably not detectable in all seemed to be the most consistent parameter di€ering SCLC lines analysed. On the contrary, all NSCLC between the two states of cells examined. cell lines with one exception (NCI-H2077) do express signi®cant amounts of this gene. These ®ndings indicate that elevated fra-1 expression is not only Fra-1 is di€erentially expressed in NSCLC and SCLC found in phenotypically converted SCLC cells as a cell lines consequrence of H-ras and c-myc overexpression, but Finally, to see whether components of AP-1 might is more generally a distinct marker of NSCLC also be di€erentially regulated in other lung cancer character. Histiotype-dependent recruitment of Fra-1 in lung cancer GRisse-Hacklet al 3064

b SCLC

a NSCLC 24H DMS79 H1092 H69 H82 SW210 86M1 H841 H146 H187 H526 H510

— c-fos H1573 H2009 H2126 H2077 H23 97TM1 U1752 U1810 H125 H820 NHBE 32M1 103H

c-fos —

— fosB

fosB —

— fra-1

fra-1 —

— fra-2 fra-2 —

— c-jun c-jun —

— junD junD —

junB — — junB

GAPDH — — GAPDH

Figure 7 Expression of fos and jun family members in NSCLC compared to SCLC cell lines. Conditions of RT ± PCR analysis were the same as described in the legend to Figure 6. Seen is the virtual absence of fra-1 transcripts in all SCLC-derived cell lines

Discussion Transitions between small cell and non-small cell carcinoma have been observed both in vivo and in vitro In this study we have analysed molecular events (Barr et al., 1996; BuÈ rger et al., 1994; Falco et al., underlying the transition from SCLC to NSCLC 1990; Kalemkerian and Mabry, 1993; Mabry et al., phenotype of two lung cancer cell lines expressing H- 1988, 1991). A model system to study the molecular ras and c-myc transfected oncogenes. As a result, we events underlying these transitions has been recently found that these cells di€er from their corresponding established by transfecting an activated Ha-ras gene parental cells by an increased AP-1 DNA binding into myc-expressing SCLC cells which causes transition activity (see under Results) contrary to the binding to a NSCLC phenotype resembling large-cell, undiffer- activities of other transcription factors analysed (data entiated cancer (Barr et al., 1991, 1996; BuÈ rger et al., not shown). Moreover, we can show that in SCLC cells 1994; Falco et al., 1990; Mabry et al., 1988). Since expressing these oncogenes the composition of the AP- overexpression of H-ras in SCLC cells alone does not 1 complex di€ers markedly from complexes induced by lead to full phenotypic conversion, it has been activators of protein kinase C such as phorbol esters. necessary in the past to either select SCLC cells which While the former complex contains signi®cant amounts overexpress an endogenously ampli®ed c-myc or N-myc of Fra-1 protein the phorbol ester induced complex gene, or to co-express a transfected human c-myc gene, does not. We can further provide evidence that this thus permitting progression of SCLC into NSCLC ®nding is not restricted to H-ras/c-myc transfected cell (Barr et al., 1996; BuÈ rger et al., 1994; Falco et al., clones but seems to be a general characteristic for 1990). This raises the question as to which of the two nearly all NSCLC cell lines analysed. Contrary to oncogenes expressed in the transfected SCLC subclones SCLC cells and in line with the aforementioned contributes more to the altered AP-1 binding activity. transfected SCLC clones, most cell lines of NSCLC Although this aspect has not been clari®ed in the origin exhibit (i) a detectable amount of constitutive present study, several lines of evidence suggest that this AP-1 binding activity which is readily inducible by e€ect is more a consequence of Ras function rather phorbol esters, (ii) constitutive expression of fra-1 and than that of Myc. First of all, the role of Ras in signal (iii) expression of CD44, a target gene of AP-1 activity. transduction pathways that lead to the activation of Histiotype-dependent recruitment of Fra-1 in lung cancer G Risse-Hackl et al 3065 AP-1 has been fundamentally established in numerous fra-1 was originally isolated as an immediate-early reports (reviewed in Karin and Hunter, 1995; Marshall, gene which encodes a Fos-related antigen cross- 1995). Members of the myc family, on the other hand, reacting with c-Fos antibodies (Cohen and Curran, are frequently overexpressed in SCLC including several 1988). Serum growth factors stimulate transcription SCLC lines analysed in this study (c-myc, cell line from the promoter of this gene in a delayed-early type SCLC-24H, NCI-H82; N-myc, cell lines NCI-N592, of response contrary to that of c-fos and fosB SCLC-86M1; L-myc, cell lines SCLC-24H, SCLC- induction which both are of the immediate-early type 86M1, NCI-H69, and NCI-H82; see (BuÈ rger et al., (Cohen and Curran, 1988; Kovary and Bravo, 1992). 1994). In this collection of SCLC cell lines analysed, Recently, it has been shown in NIH3T3 cells that fra-1 however, we could not observe any correlation between is absent from cycling cells but accumulates in cells myc overexpression and AP-1 content or CD44 gene expressing oncogenic Ras protein (Lallemand et al., expression. Moreover, our analysis of E-box binding 1997; Mechta et al., 1997). In these cells, c-Jun and activity in SCLC cell lines including N592 and H69 Fra-1 are the major components of AP-1 activity and, and oncogene transfected derivatives revealed a as shown by co-transfection experiments, probably the relatively uniform pattern (data not shown). Thus, mediators of Ras-transformation. This ®nding is in line activated Ras may be the likely component to achieve with our own data based on the analysis of SCLC cells. altered TRE-binding and CD44 target gene activation Similarly, we ®nd Ras to have the same profound in the cell lines examined although a permissive e€ect e€ects on the composition of the AP-1 complex as in of c-myc cannot be excluded. NIH3T3 cells leading to the accumulation of Fra-1 and The upregulation of TRE-binding activity in H-ras/ c-Jun. Our results di€er only with respect to JunD c-myc transfected SCLC cells implies several possibi- which is moderately upregulated in the three H-ras/c- lities of AP-1 complexes to form. Due to the relative myc transfected SCLC clones analysed but is not increase in Fra-1, cJun and JunD we expect either Fra- a€ected in Ras-transformed NIH3T3 cells (Mechta et 1/c-Jun or Fra-1/JunD heterodimers to form or c-Jun al., 1997; Pfarr et al., 1994), a function which may not or JunD homodimers. Since both homodimers possess account for SCLC cells. The second important much weaker DNA-binding activities than the hetero- implication according to Lallemand et al. is that fra-1 dimers (Ryseck and Bravo, 1991), it is likely that is absent from cycling cells. Although this observation heterodimeric proteins make up the bulk of AP-1 does not imply that Fra-1 has no function in cycling binding activity. It is noteworthy that these two NIH3T3 cells (see Kovary and Bravo, 1992) it would heterodimers have di€erent transcriptional regulatory explain our own observation that SCLC cell lines function, the Fra-1/c-Jun heterodimer being an which show no fra-1 expression have cell doubling extremely poor, the Fra-1/JunD a very ecient times which are comparable to the majority of NSCLC transactivator of AP-1 regulated gene expression lines which do express this gene. These data, therefore, (Suzuki et al., 1991). The Fra-1/JunD heterodimer make it more likely that Fra-1 in lung cancer cells may has been shown to play an important regulatory role in not be linked to cell cycle activity but may rather the AP-1 dependent expression of the human constitute a marker in the cell type transition pathway urokinase-type plasminogen activator receptor gene in coinciding with the expression of NSCLC markers. colon cancer cells (Lengyel et al., 1996) and of the One consequence which most likely results from the human involucrin gene in keratinocytes (Welter et al., induction of TRE-binding activity in phenotypically 1995). Interestingly, neither Fra-1 nor JunD were converted SCLC cells is the elevated expression of found in AP-1 complexes of PMA-induced parental CD44. This gene has been shown to be AP-1 SCLC cells. Thus we can conclude that the high TRE- responsive due to a TRE in its promoter (Hofmann binding activity we observe in ras/myc transfected et al., 1993). In addition CD44 expression is SCLC cells phenotypically transformed to NSCLC cells upregulated in cells transformed by either v-src or is achieved by speci®c recruitment of distinct AP-1 activated H-ras (Hofmann et al., 1993; Jamal et al., proteins to a functional AP-1 complex. 1994), which suggestes that CD44 expression is The most interesting result was obtained with normally regulated by signaling pathways in which respect to the di€erential fra-1 expression in NSCLC these oncogenes participate. The product of the CD44 and SCLC cell lines pointing to a di€erential gene itself is a cell surface transmembrane proteoglycan histiotype-related mechanism of regulation. Fra-1 has and appears to mediate a diverse range of functions in been identi®ed in heart and skeletal muscle (Hannan di€erent cell types (Haynes et al., 1991; Sherman et al., et al., 1993; Park et al., 1992), brain (Persico et al., 1994). CD44 has been shown to function in both cell- 1993), leukocytes (Boise et al., 1993), ®broblasts cell and cell-ECM interactions through binding to (Braselmann et al., 1993), adipocytes (Stephens et hyaluronan, ®bronectin, collagen and osteopontin al., 1993), the vascular system (Miano et al., 1993), in (Aru€o et al., 1990; Jalkanen and Jalkanen, 1992; St. skeletal tissues (Wang et al., 1993), in keratinocytes John et al., 1990; Weber et al., 1996). Our ®nding that (Welter et al., 1995) and in colon cancer (Lengyel et CD44 is preferentially expressed in NSCLC cell lines al., 1996). Based on our data, we now describe a also is in line with previous reports which show that possible tissue-speci®c function of Fra-1 in human CD44 is preferentially expressed in NSCLC-type lung cancer cells. Evidence for such a role comes not tumors but not in SCLC (Givehchian et al., 1996; only from the di€erential gene expression in lung Penno et al., 1994; Wimmel et al., 1997) and as such cancer cell lines, but also from the altered transcrip- can be viewed as an NSCLC-speci®c marker. In the tional status detected in the phenotypically trans- light of these data also the NSCLC character of H-ras/ formed lung cancer cells. Thus, Fra-1 is highly c-myc transfected SCLC clones is supported. Whether expressed in H-ras/c-myc transfected SCLC but the gain of additional CD44-related adhesion functions hardly detectable in the parental SCLC cells. might be responsible for the adherent growth of Histiotype-dependent recruitment of Fra-1 in lung cancer GRisse-Hacklet al 3066 NSCLC in contrast to SCLC cells, however, remains to (random sequence): 5'-GCGACTAACATCGATCG-3' be shown. (sense) and corresponding antisense strand, SP-1 (kindly In conclusion, our data suggest, that SCLC cells provided by Dr G Suske, IMT, Marburg): 5'-AGCTTCC- upon H-ras/c-myc-induced morphological conversion GTTGGGGCGGGGCTTCACG-3' (sense) and 5'-TCGA- to NSCLC cell phenotype reveal enhanced AP-1 CGTGAAGCCCCGCCCCAACGGA-3' (antisense strand). All oligonucleotides were labeled and applied in electro- activity, upregulation of fra-1 expression, recruitment phoretic mobility assays as described earlier (Risse et al., of Fra-1 protein to the TRE-binding complex, and 1989). activation of AP-1 dependent gene expression, e€ects, which are also a hallmark of nearly all NSCLC lines analysed. These results, therefore, may point to a Electrophoretic mobility shift assay (EMSA) general involvement of AP-1 proteins, namely that of Whole cell extracts were prepared according to a protocol Fra-1, in the transition from SCLC to NSCLC describedinSchoÈler et al. (1989). Binding reactions were phenotype and in the maintainance of di€erentiation performed by incubating 5 or 10 mg of whole cell extract speci®c functions in NSCLC cells. with 0.5 or 1 mg of polydI-dC (Pharmacia Inc.) in a bu€er containing 10 mM HEPES, pH 7.9, 60 mM KCl, 4% Ficoll, 1 mM EDTA and 1 mM DTT for 30 min at room temperature (Barberis et al., 1987). One to two Materials and methods femtomoles (5 ± 106103 c.p.m.) of 32P-labeled double- stranded oligonucleotide was then added and incubation Cell lines was continued for 30 min at room temperature. Protein- DNA complexes were resolved on 4% native polyacryla- The human lung carcinoma cell lines used in this study mide gels in 0.256TBE bu€er and visualised by include the NSCLC lines EPLC-32M1, U1752 (squamous autoradiography. cell carcinomas), LCLC-103H, U1810, LCLC-97TM1 Competition assays were performed by adding the (large cell carcinomas), NCI-H23, -H125, -H820, -H1573, indicated amounts of unlabeled speci®c or random oligonu- -H2009, -H2077, -H2126 (adenocarcinomas) and the SCLC cleotides to the binding reaction 30 min before the addition lines NCI-H69, -H82, -H841, -H146, -H187, -H526, - of 32P-labeled probe. For speci®c antibody binding, whole cell H1092, -H510, -24H, 86M1, DMS79 and SW210. The extracts were preincubated with the respective antisera (10 mg origin and characteristics of these cell lines have been in case of anity-puri®ed antibody, 50 mg in case of described previously (Wimmel et al., 1997). All cells were antiserum) for 45 min at 48C followed by the addition of grown in RPMI 1640 medium (GIBCO ± BRL) supplemen- the labeled oligonucleotide. ted with 10% heat-inactivated fetal calf serum (GIBCO ± BRL), nonessential amino acids and 4 mmol/L glutamine. Normal bronchial epithelial (NHBE) cells were obtained Reverse transcriptase-polymerase chain reaction (RT ± PCR) from Clonetics Corp. and grown in BEGM medium according to the supplier's instructions. Total RNA was prepared from all cell lines using RNAzol Whenever necessary, PMA (phorbol 12-myristate 13 B (CINNA/BIOTECX Laboratories INC.) as ready-to-use acetate, Sigma, dissolved in DMSO) was added to a ®nal solution or RNeasy RNA extraction kit (Qiagen) according to the suppliers' protocols. 1 mg RNA was then reverse concentration of 561078 M for the indicated period of time. transcribed using Oligo (d)T as primer and aliquots of the resulting cDNA were used for RT ± PCR. The cDNA was Antibodies ampli®ed by subsequent PCR for 20 ± 30 cycles (948Cfor 1min,608Cfor1minand728C for 1 min) in the presence Antibodies indicated by an asterisk (see Figure 5) were of a-[33P]dCTP. For ampli®cation of c-jun fragments, 5% peptide-speci®c, anity-puri®ed mouse antibodies and DMSO was added to the reaction mix. The number of were obtained from Santa Cruz Biotechnology, Inc. (a c- cycles was kept limited in order to ensure a linear Jun*/sc-45, a JunD*/sc-74, a JunB*/sc-46, a Fra-1*/sc-183, ampli®cation of PCR products. The following primers a Fra-2*/sc-171). All other antibodies used were generated were used for PCR ampli®cation: and anity-puri®ed as described before (Adamkiewicz et al., 1990; 1993). These comprised the following antisera: a GAPDH (Arcari et al., 1984): Pan AP-1(P10), rabbit antibody directed against oligopep- 5'primer, 5'-CGTCTTCACCACCATGGAGA tide P10 (c-Fos aa 135 ± 162), covering the DNA binding 3'primer; 5'-CGGCCATCACGCCACAGTTT region conserved within all Fos family members; a c- Fos(455), rabbit antibody directed against a b-gal/Fos CD44 (transmembrane region; Screaton et al., 1992) fusion protein containing aa 151 ± 292 of c-Fos; a c- 5'primer, 5'-GAGACCAAGACACATTCCACC Fos(P3), rabbit antibody directed against Fos oligopeptide 3'primer, 5'-GCCAAGAGGGATGCCAAGATGA P3 (c-Fos aa 214 ± 226); a FosB(K30): rabbit antibody directed against a b-gal/FosB fusion protein containing aa c-fos (Van Straaten et al., 1983): 1 ± 169 of FosB; a c-Jun(K31): rabbit antibody directed 5'primer, 5'-ACGCAGA-CTACGAGGCGTCA against a b-gal/c-Jun fusion protein containing aa 80 ± 334 3'primer, 5'-TTCACAACGCCAGC-CCTGGA of c-Jun; a c-Jun(jp5/6): rabbit antibody directed against Jun oligopeptide jp5/6 (c-Jun aa 265 ± 280). Rabbit IgG fosB (Siderovski et al., 1990): was obtained from a preimmune serum by protein A 5'primer, 5'-AAAAGCAGAGCTGGAGTCGG puri®cation. 3'primer, 5'-GTACGAAGGGTTAACAACGG

fra-1 (Matsui et al., 1990): Double stranded oligonucleotides 5'primer, 5'-CCAAGCATCAACACCATGAG TRE (TPA responsive element=AP-1 consensus sequence 3'primer, 5'-AGGGAGATACAAGGTACAGG Angel, et al. 1987): 5'-AAGCATGAGTCAGACAC-3' (sense) and corresponding antisense strand, mut. TRE fra-2 (Matsui et al., 1990): (mutant TRE; Risse et al., 1989) 5'-AAGCACGAGGCA- 5'primer, 5'AAGATTAG-CCCCGAGGAGCG GACAC-3' (sense) and coresponding antisense strand, Rd 3'primer, 5'-CAGCAATGCTGATG-GGCTTG Histiotype-dependent recruitment of Fra-1 in lung cancer G Risse-Hackl et al 3067 c-jun (Hattori et al., 1988): immuno¯uorescence was performed on 4 mmsectionsof 5'primer, 5'-TTCACCTT-CTCTCTAACTGC the resulting tumors with murine monoclonal antibody 3'primer, 5' TCACTCACTGAGCGCT-CTTC against the CD44 standard form (clone SFF-2, Bender MedSystems) and alkaline phosphatase conjugated rabbit junD (Nomura et al., 1990): a-mouse second antibody. Enzymatic conversion of 5'primer, 5'-TTGTCGC-CCATCGACATGGACA ¯uorescent substrate followed a protocol provided by the 3'primer, 5'-AGCTCCGTGTTCTGACTCTTGA manufacturer (Molecular Probes, Leiden, The Netherlands) and antibody binding was then visualised by ¯uorescence junB (Nomura et al., 1990): microscopy. Counterstaining was performed with Hoechst 5'primer, 5'-CCAGTCCTTCCACCTCGACGTTTACAA 33258. 3'primer, 5'-GACTAAGTGC-GTGTTTCTTTTCCACAG.

Aliquots of the PCR reaction were electrophoresed in 6% Acknowledgements polyacrylamide for 2 h at room temperature, the gels were We are grateful to M GoÈ tzfried for skilful technical dried thereafter and exposed to X-ray ®lm (Kodak X-OMAT assistance and to Dr M Krause and S KlingelhoÈ fer for AR) for 12 ± 72h. the synthesis of oligonucleotides. We are also indebted to Dr A Stark for critically reading the manuscript. This work has been supported by the Dr Mildred Scheel-Stiftung fuÈ r Immunostaining Krebsforschung (W 65/93/Schu 1 granted to MS) and in N592 and H-ras/c-myc transfected N592.5 cells were grown part by the Deutsche Forschungsgemeinschaft (donated to as xenografts in nu/nu mice for 21 days. Indirect JA, SFB 215).

References

Adamkiewicz J, BruÈ ller H-J, Gausepohl H, Frank R and Goldstone SD and Lavin MF. (1994). Oncogene, 8, 2305 ± MuÈ ller R. (1990). Oncogene, 5, 525 ± 533. 2311. Adamkiewicz J, Behn M, BruÈ sselbach S, Havemann K and Hannan RD, Stennard FA and West AK. (1993). J. Mol. MuÈ ller R. (1993). Int. J. Oncol., 2, 503 ± 512. Cell. Cardiol., 25, 1137 ± 1148. Amati B and Land H. (1994). Curr. Opin. Genet. Dev., 4, Hattori K, Angel P, LeBeau MM and Karin M. (1988). Proc. 102 ± 108. Natl. Acad. Sci. USA, 85, 9148 ± 9152. Angel PE, Imagawa M, Chiu R, Stein B, Imbra RJ, Haynes BF, Liao X-H and Patton KL. (1991). Cancer Cells, RahmsdorfHJ,JonatC,HerrlichPandKarinM. 3, 347 ± 350. (1987). Cell, 49, 729 ± 739. Hill CS and Treisman R. (1995). Cell, 80, 199 ± 211. Angel PE and Karin M. (1991). Biochim. Biophys. Acta, Hirai S-I, Ryseck R-P, Mechta F, Bravo R and Yaniv M. 1072, 129 ± 157. (1989). EMBO J., 8, 1433 ± 1439. Arcari P, Martinelli R and Salvatore F. (1984). Nucl. Acids Hofmann M, Rudy W GuÈ nthert U, Zimmer, SG, Zawadzki Res., 12, 9179 ± 9189. V, ZoÈ ller M, Lichtner RB, Herrlich P and Ponta H. (1993). Aru€o A, Stamenkovic I, Melnick M, Underhill CB and Seed Cancer Res., 53, 1516 ± 1521. B. (1990). Cell, 61, 1303 ± 1313. Jalkanen S and Jalkanen M. (1992). J. Cell. Biol., 116, 817 ± Barberis A, Superti-Furga G and Busslinger M. (1987). Cell, 835. 50, 347 ± 359. Jamal HH, Cano-Gauci DF, Buick RN and Filmus J. (1994). Barr LF, Campbell SE, Penno MB, Ball DW and Baylin SB. Oncogene, 9, 417 ± 423. (1996). Cell Growth Di€., 7, 1149 ± 1156. Kalemkerian GP and Mabry M. (1993). In: Roth JA, Cox JD Barr LF, Mabry M, Nelkin BD, Tyler G, May WS and and KiHong W (eds). Lung Cancer. Blackwell Scienti®c Baylin SB. (1991). Cancer Res., 51, 5514 ± 5519. Publications. pp. 57 ± 84. Bernards R. (1995). Curr. Biol., 5, 859 ± 861. Karin M and Hunter T. (1995). Curr. Biol., 5, 747 ± 757. Bohmann D, Bos TJ, Admon A, Nishimura T, Vogt PK and Kerr LD, Holt JT and Matrisian LM. (1988). Science, 242, Tijan R. (1987). Science, 238, 1386 ± 1392. 1424 ± 1427. Boise LH, Petryniak B, Mao X, June CH, Wang CY, Kovary K and Bravo R. (1992). Mol. Cell. Biol., 12, 5015 ± Lindsten T, Bravo R, Kovary K, Leiden JM and 5023. Thompson CB. (1993). Mol. Cell. Biol., 13, 1911 ± 1919. Lallemand D. Spyrou G, Yaniv M and Pfarr CM. (1997). Braselmann S, Graninger P and Busslinger M. (1993). Proc. Oncogene, 14, 819 ± 830. Natl. Acad. Sci. USA, 90, 1657 ± 1661. Lengyel E, Wang H, Stepp E, Juarez J, Wang Y, Doe W, BuÈ rger C, Sche‚er S, ElsaÈ sser HP and Adamkiewicz J. Pfarr CM and Boyd D. (1996). J. Biol. Chem., 271, (1994). Mol. Cell. Di€., 2, 373 ± 398. 23176 ± 23184. Cohen DR and Curran T. (1988). Mol. Cell. Biol., 8, 2063 ± Mabry M, Nakagawa T, Nelkin BD, McDowell E, Gesell M, 2069. Eggleston JC, Casero RA and Baylin SB. (1988). Proc. Cohen DR and Curran T. (1990). Crit. Rev. Oncogenesis, 1, Natl. Acad. Sci. USA, 85, 6523 ± 6527. 6184 ± 6191. Mabry M, Nelkin BD, Falco JP, Barr LF and Baylin SB. Dixit VM, Marks RM, Sarma V and Prochovnik EV. (1989). (1991). Cancer Cells, 3, 53 ± 58. J. Biol. Chem., 264, 16905 ± 16909. Marshall CJ. (1995). Cell, 80, 179 ± 185. Estus S. Zaks WJ, Freeman RS, Gruda M, Bravo R, Johnson Matsui M, Tokuhara M, Konuma Y, Nomura N and EM Jr. (1994). J. Cell. Biol., 127, 1717 ± 1727. Ishizaki R. (1990). Oncogene, 5, 249 ± 255. Falco JP, Baylin SB, Lupu R, Borges M, Nelkin BD, Jasti Mechta F, Lallemand D, Pfarr CM and Yaniv M. (1997). RK, Davidson NE and Mabry M. (1990). J. Clin. Invest., Oncogene, 14, 837 ± 847. 85, 1740 ± 1745. Melamed D, Resnitzky D, Haimov J, Levy N, Pfarr CM, Givehchian M, Woerner SM, Lacroix J, ZoÈ llerM,DringsP, Yaniv M and Kimchi A. (1993). Cell Growth Di€., 4, 689 ± Becker H, Kayser K, Ridder R and von-Knebel-Doeberitz 698. M. (1996). Oncogene, 12, 1137 ± 1144. Histiotype-dependent recruitment of Fra-1 in lung cancer GRisse-Hacklet al 3068 Miano JM, Vlasic N, Tota RR and Stemerman MB. (1993). Sherman L, Sleeman J, Herrlich P and Ponta H. (1994). Curr. Am.J.Pathol.,142, 715 ± 724. Opin. Cell Biol., 6, 726 ± 733. Nakabeppu Y, Ryder K and Nathans D. (1988). Cell, 55, Siderovski DP, Blum S, Forsdyke RE and Forsdyke DR. 907 ± 915. (1990). DNA Cell Biol., 9, 579 ± 587. Nishina H, Sato H, Suzuki T, Sato M and Iba H. (1990). Smeyne RJ, Vendrell M, Hayward M, Baker SJ, Miao GG, Proc. Natl. Acad. Sci. USA, 87, 3619 ± 3623. Schilling K, Robertson LM, Curran T and Morgan JI. NomuraN,IdeM,SasamotoS,MatsuiM,DateTand (1993). Nature, 363, 166 ± 169. Ishizaki R. (1990). Nucleic Acids Res., 18, 3047 ± 3048. St John T, Meyer J, Idzerda R and Gallatin WM. (1990). Park K, Chung M and Kim SJ. (1992). J. Biol. Chem., 267, Cell, 60, 45 ± 52. 10810 ± 10815. Stephens JM, Butts M, Stone R, Pekala PH and Bernlohr Penno MB, August JT, Baylin SB, Mabry M, Linnoila RI, DA. (1993). Mol. Cell. Biochem., 123, 63 ± 71. Lee VS, Croteau D, Yang XL and Rosada C. (1994). Suzuki T, Okuno H, Yoshida T, Endo T, Nishina H and Iba Cancer Res., 54, 1381 ± 1387. H. (1991). Nucleic Acids Res., 19, 5537 ± 5542. Persico AM, Schindler CW, O'Hara BF, Brannock MT and Troen BR, Chauhan SS, Ray, D and Gottesman, MM. Uhl GR. (1993). Mol. Brain. Res., 20, 91 ± 100. (1991). Cell Growth Di€er., 2, 23 ± 31. Pfarr CM, Mechta F, Spyrou G, Lallemand D, Carillo S, Van Straaten F, MuÈ ller R, Curran T, Van Beveren C and Yaniv M. (1994). Cell, 76, 747 ± 760. Verma I. (1983). Proc. Natl. Acad. Sci. USA, 80, 3183 ± Risse G, Jooss K, Neuberg M, BruÈ ller HJ and MuÈ ller R. 3187. (1989). EMBO J., 8, 3825 ± 3832. Wang ZQ, Grigoriadis AE and Wagner EF. (1993). J. Bone Roggli VL, Volmer RT, Greenberg SD, McGavran MH, Miner. Res., 8, 839 ± 847. Spjut HJ and Yesner R. (1985). Hum. Pathol., 16, 569 ± Weber G, Ashkar S, Glimcher M and Cantor H. (1996). 579. Science, 271, 509 ± 512. Ryder K, Lanahan A, Perez-Albuerne E and Nathans D. Welter JF, Crish JF, Agarwal C and Eckert RL. (1995). J. (1989). Proc. Natl. Acad. Sci. USA, 86, 1500 ± 1503. Biol. Chem., 270, 12614 ± 12622. Ryseck RP and Bravo R. (1991). Oncogene, 6, 533 ± 542. Wimmel A, Schilli M, Kaiser U, Havemann K, Ramaswamy Sato H and Seiki M. (1993). Oncogene, 8, 395 ± 405. A, Branscheid D, Kogan E and Schuermann M. (1997). SchoÈ ler HR, Hatzopoulos AK, Balling R, Suzuki N and Lung Cancer, 16, 239 ± 257. Gruss P. (1989). EMBO J., 8, 2543 ± 2550. Zerial M, Toschi L, Ryseck R-P, Schuermann M, MuÈ ller R Screaton GR, Bell MV, Jackson DG, Cornelis FB, Gerth U and Bravo R. (1989). EMBO J., 8, 805 ± 813. and Bell JI. (1992). Proc. Natl. Acad. Sci. USA, 89, 12160 ± 12164.