Oct-4 Regulates Alternative Platelet-derived Growth Factor a Receptor Gene Promoter in Human Embryonal Carcinoma Cells*

Expression of the platelet-derived growth factor alpha-receptor (PDGFalphaR) gene is tightly controlled in mammalian embryogenesis. A well established model system to study human embryogenesis is the embryonal carcinoma cell line Tera2. We have shown previously that retinoic acid-differentiated Tera2 cells express two PDGFalphaR transcripts of 6.4 kilobase pairs (kb) (encoding the full-length receptor) and 3.0 kb, respectively, whereas in contrast, undifferentiated Tera2 cells express PDFGalphaR transcripts of 1.5 kb and 5.0 kb. Here we show that this switch in PDGFalphaR expression pattern during differentiation of Tera2 cells results from alternative promoter use. In undifferentiated cells, a second promoter is used, which is located in intron 12 of the PDGFalphaR gene. Functional analysis shows that this promoter contains a consensus octamer motif, which can be bound by the POU domain transcription factor Oct-4. Oct-4 is expressed in undifferentiated Tera2 cells but not in retinoic acid-induced differentiated cells. Mutation of the octamer motif decreases promoter activity, while ectopic expression of Oct-4 in differentiated Tera2 cells specifically enhances the activity of this PDGFalphaR promoter. Therefore, we suggest that an important aspect in the maintenance of the undifferentiated state of human embryonal carcinoma cells results from Oct-4 expression, which thereupon activates this PDGFalphaR promoter.

E x p ressio n of th e p latelet-d eriv ed grow th facto r «-re c e p to r (PDGFaR) gene is tig h tly controlled in m am m a lia n em bryo genesis. A w ell estab lish ed m odel system to stu d y h u m a n em bryogenesis is th e em bryonal carci nom a cell lin e Tera2. We have show n previously th a t re tin o ic acid -d ifferen tiated T era2 cells express tw o PDGFarR tra n s c rip ts of 6.4 kilobase p a irs (kb) (encoding th e full-length re c e p to r) a n d 3.0 kb, respectively, w h ereas in co n trast, u n d iffe re n tia te d Tera2 cells ex p ress PD FG aR tra n s c rip ts of 1.5 kb a n d 5.0 kb. H ere we show th a t th is sw itch in P D G FaR expression p a tte rn d u rin g d iffe re n tia tio n of T era2 cells resu lts from a lte r n a tiv e p ro m o te r use. In u n d iffe re n tia te d cells, a second p ro m o te r is used, w hich is lo cated in in tro n 12 of th e P D G F aR gene. F u n c tio n a l analysis show s th a t th is p ro m o te r co n tain s a co nsensus o ctam er m otif, w hich can be b o u n d by th e POU dom ain tra n s c rip tio n fac to r Oct-4. Oct-4 is ex p ressed in u n d iffe re n tia te d Tera2 cells b u t n o t in re tin o ic acid -in d u ced d iffe re n tia te d cells. M uta tio n of th e o ctam er m o tif d ecreases p ro m o ter activity, w hile ectopic ex p ressio n of Oct-4 in d ifferen tiated Tera2 cells specifically en h an ces th e activ ity of th is PD G FaR p ro m o ter. T herefore, w e suggest th a t an im p o rtan t as p ec t in th e m a in te n a n c e of th e u n d iffe re n tia te d sta te of h u m a n em bryonal carcin o m a cells resu lts from Oct-4 expression, w h ich th e re u p o n activ ates th is PD G FaR p ro m o ter. P la te le t-d eriv e d g ro w th factor (PD G F)1 an d its receptors p lay a p ro m in e n t role d u rin g early m a m m a lia n developm ent. A lready in th e p re im p la n ta tio n em bryo of th e m ouse, from the two-cell sta g e onw ards to th e b lasto cy st stage, th e PDGF-A ch ain is expressed (1), w hile bo th th is gene and th e cognate P D G F a-recep to r (P D G F aR ) gene a re expressed in early p o stim p la n ta tio n em bryos (2). M u rin e em bryonal carcinom a (EC) cells in c u ltu re secrete PDGF-AA (3) and express the P D G F a R following d ifferen tiatio n by retinoic acid (4), T he im * The costs of publication of this article were defrayed in part by the payment of page charges, This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequenceisj reported in this paper has been submitted to the GenBank™ /EMBL Data Bank with accession number(s) X95095.
§ 1 The abbreviations used are: PDGF, platelet-derived growth factor; PDGFaR and PDGF/3R, platelet-derived growth factor-a and -j3 recep tor, respectively; RA, retinoic acid; EC, embryonal carcinoma; kb, kilobase pair(s); kFGF, Kaposi's fibroblast growth factor; EMSA, electro phoretic mobility shift assay. portance of th e P D G F aR in m am m a lian developm ent is also exem plified by th e Patch (Ph) m ouse m u ta n t. T he Ph m u ta n t lacks p a rt of the P D G F aR gene (5,6) and displays severe developm ental defects in m esoderm al a n d neu ro ecto d erm al tis sues, often resu ltin g in p re n a ta l le th a lity (7,8).
A n im p o rtan t m odel system for studying h u m a n early em bryogenesis is th a t of te stic u la r germ cell tum ors. T h ese tu m o rs are derived from a d eran g em en t of a prim ordial g erm cell in early life, which first develops into a noninvasive carcinom a-ira situ and subsequently grow out as a sem inom a or a nonsem inom atous tum or (9). T he stem cells of n o n sem in o m ato u s t u m ors, also referred to as E C cells, strongly resem ble cells of th e early p reim p lan tatio n h u m a n embryo. V arious estab lish ed h u m an EC cell lines, am ong others th e T era2 cell line, can be induced to differentiate in vitro into a v a rie ty of m a tu re , nontum origenic cell types by th e m orphogen retinoic acid (10). W e have recently show n th a t differen tiatio n of T era2 E C cells by retinoic acid (RA) is accom panied by a shift in expression of PD G F aR mRNA v a ria n ts ( l l ) . 2 F o u r h u m a n P D G F aR t r a n scripts have been identified as a re s u lt of a com bination of altern ativ e splicing and prom oter use. Two P D G F a R m R N A species of 1.5 and 5.0 kb, respectively, are ex p ressed in early h u m a n embryonic cells, including th e u n d iffe re n tia te d T era2 EC cells. S tudies on surgically rem oved te stic u la r g erm cell tu m o rs have show n th a t th e 1.5-kb P D G F aR tra n s c rip t can be used as a selective m a rk e r for carcinom a-i/i situ, sem inom a, and undifferentiated nonsem inom atous tu m o rs in th e h u m a n te stis.2 In differentiated cells, including R A -differentiated T era2 (Tera2 RA) cells, two o th er P D G F aR tra n s c rip ts of 6.4 kb, which encodes th e functional full-length receptor, an d of 3.0-kb, w hich potentially encodes a d o m in an t n e g a tiv e isoform , have been identified. A b e rra n t expression of th e fu ll-len g th PD G F aR receptor, encoded by th e 6.4~kb tra n sc rip t, h a s also been im plicated in tum origenesis, i.e. it is overexpressed in various tum ors, including gliom as (12).
In a previous stu d y we cloned and ch aracterized th e h u m a n P D G F aR gene prom oter (P I), w hich gives rise to th e 6.4-a n d 3.0-kb tran scrip ts. A ctivity of th is P I p ro m o ter can be s tim u lated strongly by RA a n d cAMP (13). S im ilar stu d ie s w ith respect to th e m ouse an d r a t P D G F aR p ro m o ter h av e b een published recently (14,15). In th e p re se n t stu d y we h av e cloned an d characterized th e second P D G F aR gene p ro m o ter (P2), w hich gives rise to th e 1.5-and 5.0-kb tra n s c rip ts in e a rly em bryonic cells. We show here th a t the P 2 prom oter, located in in tro n 12 of the P D G F aR gene, is active in u n d iffe re n tia te d T era2 cells an d is controlled by th e P O U dom ain tra n s c rip tio n factor Oct-4. Oct-4 expression is detected in T era2 E C cells b u t not in T era2 RA cells.

MATERIALS AND METHODS
Cell Culture -Tcra2 clone 13 (Tera2) cells were grown in a-modification of minimal essential medium lacking nucleosides and deoxynucleosides, supplemented with 10% (v/v) fetal calf serum and 44 mM NaHCOa in a 7.5% C 02 atmosphere at 37 °C. Differentiation of cells was induced by the addition of RA (5 /¿M) 16 h after the cells were seeded at low density {5.0 x 10:* cells/cm2) and maintained at this medium for 7 days, prior to further analysis or transfection.
PDGFaR P2 Promoter Constructs -Nucleotide sequence analysis was performed using the Pharmacia T7 sequencing kit. PDGFaR P2 promoter constructs were generated by standard cloning procedures (16), using either restriction fragments or DNA fragments obtained by the polymerase chain reaction. Mutation of the octamer binding site was performed with the Altered Sites system kit (Promega), according to the manufacturer's protocol. All fragments were inserted in the multiple cloning site of the pSLA4 luciferase reporter plasmid (13).
Transfection} Luciferase, and fi-Galactosidase Assays-Promoterluciferase constructs were transiently transfected into either undiffer entiated (Tera2 EC) or differentiated (Tera2 RA) cells using the calcium phosphate coprecipitation method (16). Luciferase activity was deter mined 48 h (Tera2 EC) or 72 h (Tera2 RA) post-transfection (Luciferase assay kit, Promega). The luciferase activity was corrected for transfec tion efficiency by measuring the j3-galactosidase activity of a lacZ gene driven by an SV40-promoter of a cotransfected pCHHO plasmid (17). Every experiment was done in duplicate and repeated at least twice with two batches of DNA.
RNA Analysis-Total RNA was isolated from undifferentiated (EC) or differentiated (7 days of RA treatment) Tera2 cells, using the isothiocyanate method (18). After poly(A)+ isolation, the mRNA was quan titated spectrophotometrically and subjected to 1% agarose gel electro phoresis in form amide. The amount and integrity of loaded mRNA was controlled by ethidium bromide staining, after which it was transferred to Hybond-N (Amersham Corp.). Hybridization and washing proce dures were carried out as described (11). A mouse Oct-4 cDNA probe (kindly provided by Dr. P. C. van der Vliet, University of Utrecht), was labeled by random priming (19) using a labeling kit (Amersham).
Electrophoretic Mobility Shift Assay (EMSA)-DNA restriction frag ments were filled in by Klenow polymerase treatment in the presence of [a-32P]dCTP. Oligonucleotides for EMSAs were end-labeled using [7-32P]ATP and T4 polynucleotide kinase. Double-stranded oligonucleo tides were separated from single-stranded oligonucleotides by poly acrylamide gel electrophoresis. The oct-lc consensus oligonucleotide was purchased from Promega. Nuclear extracts were prepared as de scribed (20). Binding reactions and gel electrophoresis were performed essentially as described (21). The mouse anti-Oct-4-antibody has been described by Rosfjord and Rizzino (22).

RESULTS
Sequence Determination and Characterization of the P2 Pro moter of the PDGFaR Gene-Two alternative PDGFaR tran scripts, of 1.5 and 5.0 kb, respectively, are specifically ex pressed in the undifferentiated Tera2 embryonal carcinoma cells. Although, the transcripts terminate differently as a result of alternative splicing, both transcripts initiate in intron 12 of the PDGFaR gene ( ll ).2 In order to functionally characterize this putative promoter, which was designated P2 promoter, the region was cloned and sequenced ( Fig. 1). The sequence up stream from the transcription initiation site lacks a TATA box, which is also the case for the PDGFaR PI promoter in human (13), mouse (14), and rat (15), and is also not extremely GCrich. In the promoter region, several consensus binding sites for transcription factors could be detected, including API, AP2, and PE A3 motifs (23). A consensus octamer binding site is located in the transcribed region, at positions +28 to + 35. We first set out to determine the functional relevance of these putative binding sites and of other regions required for activity of the PDGFaR promoter. It is of interest to note that the P2 promoter region is demethylated in both undifferenti ated and RA-induced differentiated Ter a 2 cells.3 A series of progressive deletion mutants of the PDGFaR P2 promoter was 3 H. J. Kraft, unpublished results.  cloned in front of a luciferase reporter gene, transiently trans fected into undifferentiated Tera2 cells, and assayed for pro moter activity (Fig. 2). The luciferase activity of the complete intron 12 promoter sequence of approximately 2.5 kb (clone -2500/+182) was comparable with that of the much smaller clone -668/+182, indicating that no important expression in formation is pertained upstream from nucleotide -668, up to exon 12 of the PDGFaR gene. A further deletion, down to position -102 (clone -102/+182), displayed only fractionally lower activity than the clones -2500/+182 and -668/+182, which further limits the region necessary for control of high level expression. The reverse orientation of the -668/+182 fragment in the pSLA4 vector (clone -668/+182R) almost com pletely abolished activity, demonstrating the orientation de pendence of the P2 promoter. These data show that intron 12 of the PDGFaR gene contains a bona fide promoter.
Deletion mutants generated at the 3' end (clones; -668/+14; -102/ + 14), which still included an intact transcription initia tion site, reduced activity 4-5 times, compared with the paren tal clones -668/+182 and -102/+182. Thus, a cis-element determining high promoter activity is located in the tran scribed part, within the region +14 to +182. The consensus octamer motif ATGCTAAT at position +28 to +35, which is present in all the constructs that show high promoter activity, was thereupon mutated to the sequence ACGCCAAT (clones -668/+182M and -102/+182M, respectively). This mutation is expected to abrogate all fortuitous binding of POU domainspecific proteins (24,25). Upon changing these two nucleotides in the octamer motif, promoter activity drops by a factor of 3-4, comparable with deletion of the complete +14/+182 region (Fig. 2). This indicates that the octamer motif is indeed in volved in directing P2 promoter activity in Tera2 EC cells, In order to demonstrate that Tera2 EC nuclear proteins can actually bind to the octamer motif, an EMSA was performed. It is shown in Fig. 3A that the intact -102/+182 promoter frag ment forms a complex with nuclear extracts of Tera2 EC cells. The formation of this complex can be specifically competed by excess (100 times) unlabeled probe itself, but not by the -102/ + 182M fragment containing the mutated octamer motif (lane 4). Moreover, the -102/+182M fragment does not form a com plex in this EMSA (Fig. 3A, lanes 5-8). In addition, the -102/ +182 fragment competed efficiently the four specific complexes of a consensus oct-lc oligonucleotide (Promega) with nuclear extracts of EC cells, while the " 102/+182M fragment was refractory to competition (Fig. SB).
In conclusion, an octamer motif is involved in the regulation of the P2 promoter of the PDGFaR gene in undifferentiated human embryonal carcinoma cells.
Oct-4 Binds to the PDGFaR P2 Promoter Octamer Mo tif-As described previously, the major change in constitu ents of octamer binding proteins during retinoic acid-induced differentiation of murine embryonal carcinoma cells involves the down-regulation of Oct-4 expression (26). We hence hy pothesized that also in the undifferentiated human embryo-* nal carcinoma Tera2 cells the POU domain transcription factor Oct-4 is present and occupies the promoter P2 octamer motif, oct-P.
In order to test this hypothesis a series of EMSAs were performed with a double-stranded oligonucleotide (oct-P), which results in a higher resolution compared with the long promoter fragment. The oct-P contains the octamer sequence and flanking 7 nucleotides at the 3' side and 8 nucleotides at the 5' side of the P2 promoter (+20/+42). The resulting com plexes were compared with the complexes formed by the con sensus oct-lc oligonucleotide (see above). The oct-P as well as the oct-lc oligonucleotide gave rise to the formation of several complexes with nuclear proteins of Tera2 EC cells, Tera2 RA cells, or mouse F9 EC cells, which could be specifically com peted by excess of the respective cold probe itself ( Fig. 4; see also Fig. 5). In the EMSAs, oct-lc and oct-P displayed identical bandshift patterns (not shown). Based upon the complexes formed with the nuclear extracts of mouse F9 EC cells (27,28) a positive identification of the Oct-4 complex with Tera2 EC or RA extracts was made possible, and is indicated in Fig. 4A. This shows that also during RA-induced differentiation of hu man Tera2 embryonal carcinoma cells the POU domain tran scription factor Oct-4 is down-regulated, To confirm the results obtained with the above described EMSAs, a Northern blot analysis was performed on mRNA of Tera2 EC and Tera2 RA cells. The blot was probed with a labeled Oct-4 cDNA, which showed that Oct-4 mRNA is present in Tera2 EC cells, contrasting with the absence of any detect able Oct-4 mRNA in Tera2 RA cells (Fig. 4B). Conclusive evi dence that the indicated Oct-4 complex in the EMSAs is formed with this transcription factor comes from a supershift analysis. This analysis was performed with an anti-Oct4-antibody (22), which only supershifted the Oct-4-containing complex and not the Oct-1 complex (Fig. 4C).
Hence, undifferentiated human Tera2 embryonal carcinoma cells express Oct-4, which can complex with the P2-octamer motif. The cells cease to express Oct-4 upon RA-induced differ entiation, which necessarily excludes complex formation be tween Oct-4 and the oct-P motif.
Comparison of the oct-P Binding Site with the Consensus oct-lc Motif-The members of the POU transcription factor family are defined by their ability to bind to the octamer motif. This causes experimental pitfalls to distinguish between the binding of the different members to a particular motif in a specific cell type (29). In addition to the octamer consensus motif, however, the nucleotides juxtaposed to the motif are also important to the affinity and specificity of binding of a given Oct protein (24).4 Therefore, we set out to compare the binding properties of the consensus oct-lc and the PDGFaR genederived oct-P motifs with nuclear extracts from Tera2 EC cells in EMSAs. The oct-P-derived complexes could not be competed by a 500-fold excess of the consensus oct-lc oligonucleotide, while competition with oct-P itself was easily established. A 100-fold excess of oct-P competitor was sufficient for strong competition. No signal was detectable with a 500-fold excess of competitor, even upon prolonged exposure (Fig. 5). This indi cates that the oct-P oligonucleotide is bound stronger by POU proteins from Tera2 cells than the consensus oct-lc oligonu cleotide. The results with the oct-P oligonucleotide are con firmed by the reciprocal experiment, using the oct-lc oligonu cleotide as probe. Even with 100-fold excess, the oct-P competitor abolished the specific binding to the oct-lc oligonu cleotide, while the oct-lc oligonucleotide was needed in larger   •* s* ni.

Oct-4 Regulates PDGFaR Promoter in Human EC Cells
of the octamer flanking sequences of kFGF, REX-1, and PDGFaR shows, however, that the motifs cannot be aligned adequately to explain the preferential binding of Oct-4 to these sites. Accessory proteins might therefore explain Oct-4-dependent regulation (see above). The role of proteins encoded by these alternative PDGFaR transcripts, if any, remains obscure, since presently none of these proteins have been detected in vivo. The sequence of the 5.0-kb messenger suggests a putative oncogene-like action, which may be important for autonomous growth of Tera2 EC cells. Undifferentiated Tera2 cells have been shown to prolif erate in the absence of serum growth factors (34). A possible role in development or differentiation may also be inferred from the expression patterns of alternative PDGFaR transcripts. P2~initiated transcripts have been identified in human oocytes and preimplantation stages5 and in human testis tumors.2 The down-regulation of the 1.5-kb transcript expression in spermat ogenesis suggests that the P2-promoter is active in cells of the female primordial germ cell lineage, probably in the commit ment of cells during embryonal development. Alternative tran scripts, generated by differential promoter use and/or splicing, of other tyrosine kinase receptors have been described, includ ing the genes for PDGF/3R (4), FGF receptors (35), epidermal growth factor receptor (36), c-kit (37) and PDGFaR of the mouse (38). Interestingly, an alternative transcript of approx imately 4.8 kb of the PDGFaR gene has been detected in the mouse embryonal carcinoma cell line F9, but only after RAinduced differentiation (38). This transcript has also been de scribed by Lee et al. (39) and is regulated in a differentiationspecific manner. Any conclusive evidence concerning functions of the corresponding alternative proteins remains lacking, however.
In conclusion, the POU transcription factor Oct-4 controls the developmentally regulated expression by the PDGFaR P2 promoter. Whether the down-regulation of Oct-4 during RAinduced differentiation is a prerequisite for the activation of the PDGFaR promoter PI remains to be determined. We are currently working toward this goal.