THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 264, No. 21, Issue of July 25, pp. 12611-12619, 1989 Printed in U.S.A.

Expression of the Germ Cell Alkaline Phosphatase Gene in Human Choriocarcinoma Cells*

(Received for publication, March 1, 1989)

Shuichiro Watanabe, Takeshi Watanabe, Wu Bo Li, Bing-Wen Soong, and Janice Yang ChouS From the Human Genetics Branch, National Institute of Child Health and Human Deuelopment, National Institutes of Health, Bethesda, Maryland 20892

Alkaline phosphatase (ALP) in human choriocarci- noma cells (malignant trophoblasts) was characterized by its greater sensitivity to EDTA and L-leucine inhi- bition as compared with the placental isozyme. In ad- dition, both the fully processed and the nonglycosy- lated forms of choriocarcinoma ALP migrated faster than the corresponding placental enzyme in sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Choriocarcinoma cells express a 2.6-kilobase (kb) ALP mRNA unlike normal human placenta which expresses a 2.8-kb ALP mRNA. Administration of sodium butyr- ate to choriocarcinoma cells greatly increased the steady-state levels of the 2.6-kb choriocarcinoma ALP mRNA, which resulted in an increase in enzyme activ- ity and biosynthesis. S 1 nuclease analysis using probes derived from a placental ALP cDNA and ribonuclease protection assays employing probes derived from the germ cell ALP gene demonstrated that choriocarci- noma cells express the germ cell ALP gene. The germ cell ALP gene encodes the placental ALP-like isozyme that is primarily expressed in the thymus, testis, and germ cell tumors. The structures of the internal exons (11-X) of the germ cell ALP gene were determined previously based on their similarity to the placental ALP gene. However, the boundaries of exons I and XI (3’ exon) of the germ cell ALP gene were not defined due to sequence divergence between the two genes at the 5’ and 3’ regions. Ribonuclease protection and primer extension assays demonstrated that exon I of this gene is 119 base pairs in length and that germ cell ALP mRNA contains one major transcription initiation site. The isolation and characterization of germ cell ALP cDNA clones from a butyrate-treated choriocar- cinoma cDNA library showed that the germ cell ALP mRNA is 2487 bases in length and exon XI of this gene is 1135 base pairs long.

The existence of multiple alkaline phosphatase (ALP)’ genes in humans has been demonstrated by genetic and bio-

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession number($ 504948.

8C429, NIH, Bethesda, MD 20892. Tel.: 301-496-1094. $ To whom correspondence should be addressed Bldg. 10, Room

The abbreviations used are: ALP, alkaline phosphatase; SDS, sodium dodecyl sulfate; kb, kilobase(s); bp, base pair(s); Pipes, 1,4- piperazinediethanesulfonic acid.

chemical analyses (1-3). Recently, four ALP genes, the pla- cental (4), the placental ALP-like or germ cell (5, 6), the intestinal (7), and the tissue-nonspecific liver/bone/kidney- type (8), were isolated and characterized. Placental, placental ALP-like, and intestinal ALP genes are linked on the long arm of chromosome 2 (7, 9, 10). The liver/bone/kidney ALP gene is located on chromosome 1 (11). The structures of placental and intestinal ALP genes were established by se- quence analysis of cDNA clones along the corresponding genes ( 4 7 ) . Both genes are composed of 11 exons interrupted by 10 introns, and introns are located at analogous positions. Placental and intestinal ALPs have been purified and the purified enzymes extensively characterized (12-14). However, the placental ALP-like isozyme was identified primarily by physicochemical and immunological methods (15-17). The placental ALP-like isozyme is normally found in trace amounts in the testis (16) and the thymus (17). The level of enzyme is elevated in serum of patients with germ cell tumors of the testis, especially seminomas (18, 19).

In human, the placental ALP is a polymorphic enzyme that consists of 3 common and 15-20 rare variants (1, 3, 20-22). As a result, the existence of a distinct human placental ALP- like gene was in question even after two placental ALP genes (ALP-1 and ALP-2) were identified (6). Millan and Manes ( 5 ) sequenced the placental ALP-like (ALP-2) gene, synthe- sized a peptide that was unique to the ALP encoded by the gene, and finally generated antibodies that reacted specifically with the seminoma placental ALP-like isozyme (but not the placental ALP) thus establishing the presence of a placental ALP-like (germ cell ALP) gene in humans. However, dem- onstration of the existence of a functional gene requires the identification and characterization of its mRNA, and this has not been done for the germ cell ALP gene.

The germ cell ALP gene, whose structure was determined based on its similarity to the placental ALP gene (5), is also composed of 11 exons and 10 introns. In the absence of protein or cDNA sequence, the 5‘ and 3’ boundaries of the germ cell ALP gene were not defined due to sequence divergence from the placental ALP gene in these regions (5). However, ge- nomic sequence analysis predicts that the primary amino acid sequences of placental and germ cell ALPs are strongly ho- mologous (5). It has been demonstrated that placental and choriocarcinoma ALPs are synthesized as preproproteins con- taining both amino- and carboxyl-terminal signal peptides that are cleaved from the nascent protein during processing (23, 24). In placental ALP, the phosphatidylinositol-glycan moiety is covalently attached to aspartic acid residue 484 of the nascent protein which is then anchored to the plasma membranes (23,25). Thus, mature placental ALP is composed of 484 amino acid residues (23). A similar conclusion has also

12611

12612 Choriocarcinoma Alkaline Phosphatase

been made for the choriocarcinoma ALP based on the as- sumption that choriocarcinoma cells express the placental ALP gene (24).

ALP in choriocarcinoma cells is immunologically similar to the placental ALP, but it can be distinguished from the latter by its greater sensitivity to heat, EDTA, and the noncompet- itive inhibitor L-leucine (26,27). The placental ALP-like germ cell isozyme is also characterized by its sensitivity to EDTA and L-leucine inhibition (15-17), suggesting that choriocar- cinoma cell lines may express the germ cell ALP gene. Cho- riocarcinoma is a type of germ cell tumor which is of extraem- bryonic origin (28). However, the ALPs found in plasma samples of choriocarcinoma patients were mainly the placen- tal ALP type (29). Thus, it is uncertain if choriocarcinoma cells synthesize the germ cell ALP or a placental ALP variant.

In the present study, we demonstrated that choriocarci- noma and placental ALP mRNAs differ in sequences at the 5' and 3' regions. Ribonuclease protection assays provided evidence that human choriocarcinoma cells express the germ cell ALP gene. This was confirmed by the isolation and characterization of cDNA clones encoding the germ cell ALP from a butyrate-treated choriocarcinoma cDNA library.

EXPERIMENTAL PROCEDURES

Cell Culture-JEG-3 choriocarcinoma cells were grown at 37 "C in a-modified minimal essential medium supplemented with strepto- mycin, penicillin, and 4% fetal bovine serum. Cells in midlogarithmic growth phase were used in this study.

Primary term placental cells were prepared according to the method of Kliman et al. (30) and were grown at 37 "C in a-modified minimal essential medium supplemented with streptomycin, penicil- lin, and 10% fetal bovine serum.

Alkaline Phosphatase Assay-Choriocarcinoma and placental ex- tracts prepared as described previously (26) were used for measure- ment of enzyme activity. ALP activity was measured by the release of p-nitrophenol from p-nitrophenyl phosphate at pH 10.7 and 37 "C (26). By definition, 1 unit of enzyme releases 1 pmol ofp-nitrophenol/ min. Enzyme activities between treatments were compared using Student's t test.

Biosynthesis of Alkaline Phosphatase-JEG-3 or primary placental cultures were labeled with ~-["S]methionine (100 pCi/ml; ICN Bio- chemicals, Inc., Lisle, IL) for 3 h in the presence or absence of tunicamycin (1 pg/ml). Polypeptides were isolated by immunoprecip- itation with rabbit antiserum to placental ALP (31) and analyzed by 10% SDS-polyacrylamide gel electrophoresis (32) and fluorography. Apparent molecular weights were determined using ["C]methionine- labeled protein standards (Amersham Corp.).

P'PIPhosphate Binding-ALPS were labeled by incubation with carrier-free H?'PO4 (Du Pont-New England Nuclear) at pH 5 follow- ing the procedure of Milstein (33). The 32P-labeled ALPs were isolated by immunoprecipitation with antiplacental ALP serum and analyzed by SDS-polyacrylamide gel electrophoresis.

Library Construction, Screening, and Characterization of cDNA and Genomic Clones-A human placental cDNA library in Xgtll (Dr. Frank Gonzalez, NIH) and a human leukocyte genomic library in XEMBL-3 (Clontech Laboratories, Inc., Palo Alto, CA) were screened with the 5' EcoRI-KpnI insert of the placental ALP cDNA clone TPAP8E (31). Poly(A)' RNA used to construct a cDNA library in Xgtl0 was isolated from JEG-3 choriocarcinoma cells that had been exposed to 3 mM sodium butyrate for 3 days. The double-stranded cDNA was synthesized using a cDNA synthesis kit obtained from Boehringer Mannheim and ligated to custom synthesized adaptors (5'-,GGTCGACG-3' and 5"AATTCGTCGACC-3'). cDNA larger than 1.5 kb was ligated to phosphorylated X g t l O arms (Stratagene, La Jolla, CA) and packaged in vitro using the Gigapack Gold System (Stratagene). The library was screened with a 42-base oligonucleotide probe corresponding to nucleotides 598-639 (exon I includes nucleo- tides 546-664) of the germ cell ALP gene.

The cDNA and genomic inserts from the positive clones were ligated into pGEM vectors (Promega Biotec, Madison, WI) for further characterization. Sequencing of cDNA and genomic clones was ac- complished by the Sanger dideoxy chain termination method (34) using [ C X - ~ ~ S ~ ~ A T P (400 Ci/mmol; Amersham COT.).

Nucleic Acid Hybridization Analysis-Total RNA was extracted by the guanidinium thiocyanate method of Chirgwin et al. (35), and poly(A)+ RNA was obtained by oligo(dT)-cellulose chromatography. RNA was separated by electrophoresis in 1.2% agarose gels containing 2.2 M formaldehyde (36), transferred to Zetabind membrane (AMF, Meriden, CT), and hybridized with a nick-translated 32P-labeled 568- bp SstI-SstI insert of a genomic clone containing nucleotides 2233- 2800 of the germ cell ALP gene. Hybridization in the presence of dextran sulfate and washing were as described previously (31).

S I Nuclease Analysis-Uniformly labeled single-stranded frag- ments generated by primer extension of a near full length placental ALP cDNA TPAP51A (in an M13mp18 vector) were used to analyze choriocarcinoma ALP mRNA. The probe used to examine the 5' sequence of the choriocarcinoma transcript was synthesized by primer extension of TPAP51A using a 20-base oligonucleotide primer cor- responding to nucleotides 126-145 or nucleotides 242-261 of TPAP51A and released by NarI (inside the M13 vector) digestion. The probe used to examine the 3' region of the choriocarcinoma transcript was synthesized by primer extension of TPAP5lA using a 20-base oligonucleotide primer corresponding to nucleotides 1509- 1528, nucleotides 1624-1643, or nucleotides 1809-1828 of TPAP5lA and released by SstI (nucleotide 1221) digestion. S1 nuclease mapping was performed by hybridizing poly(A)+ RNA of control, butyrate- treated JEG-3 choriocarcinoma cells, or placenta with the respective probe (lo6 cpm) in buffer containing 80% formamide, 0.4 M NaCI, 0.04 M Pipes (pH 6.4), and 1 mM EDTA at 52 "C for 18 h. The hybrids were digested with S1 nuclease (900 units, Boehringer Mannheim) for 30 min at 42 "C and electrophoresed on polyacrylamide-urea sequencing gels. Single base substitutions are normally not digested by S1 nuclease under the assay conditions used.

ments (1-1274 SstI-SstI, 1275-2232 SstI-SstI, 2233-2800 SstI-SstI, Ribonuclease Protection Assay-Four germ cell ALP gene frag-

and 2801-4637 SstI-HindIII) subcloned into pGEM-2 vectors were used to generate riboprobes following the procedures supplied by Promega Biotec. Ribonuclease protection assays were performed es- sentially as described (37). Briefly, poly(A)+ RNA of control, butyrate- treated JEG-3 choriocarcinoma cells, or placenta was hybridized with the respective probes (5 X lo5 cpm) in buffer containing 80% form- amide, 0.4 M NaC1,0.04 M Pipes (pH 6.4), and 1 mM EDTA at 45 "c for 18 h. The hybrids were digested with RNases A (40 pg/ml) and T1 (2 pglml) for 60 min at 30 "C and then electrophoresed on polyacrylamide-urea sequencing gels. Single base substitutions are normally not digested by these ribonucleases under the assay condi- tions used.

54 (5'-ATGTCTGGAAGCAGTCGGAG-3',antisense) and nucleo- Primer Extension-Two primers corresponding to nucleotides 35-

tides 86-105 (5'-AGGGAGAGCTGTAGCCTCAG-3', antisense) of germ cell ALP mRNA were labeled at the 5'-OH end with [r-3'P] ATP using T4 polynucleotide kinase (Bethesda Research Laborato- ries). Butyrate-treated JEG-3 choriocarcinoma poly(A)+ RNA or yeast tRNA and lo6 cpm of primer in hybridization buffer (10 mM Pipes, pH 6.4, 400 mM NaC1, and 1 mM EDTA) were incubated at 42 "C overnight. The samples were ethanol precipitated and extended with avian myeloblastosis virus reverse transcriptase (20 units, Bio- Rad) for 60 min at 42 "C in 40 pl of a solution containing 50 mM Tris-HCI, pH 8,100 mM KCl, 10 mM MgCl', and 0.5 mM dNTP. The extended fragments were analyzed by polyacrylamide-urea sequenc- ing gels.

RESULTS

Characterization of Alkaline Phosphatase in Choriocarci- noma Cells-We have shown previously that choriocarcinoma ALP is similar to the placental ALP in its sensitivity to L- phenylalanine but that it is more sensitive to L-leucine and EDTA inhibition than the placental isozyme (26) (Table I). This suggests that choriocarcinoma cells may produce the placental ALP-like isozyme that is distinguished by its sen- sitivity to EDTA and L-leucine inhibition (15-17). ALP activ- ity is 10-20-fold lower in choriocarcinoma cells than in human term placenta, but enzyme activity and synthesis can be greatly stimulated by sodium butyrate (38). The ALP in butyrate-treated choriocarcinoma cells is equally sensitive to these agents, indicating that a similar ALP was induced by butyrate (Table I). In earlier studies, we examined ALP synthesis in choriocarcinoma cells and found that the fully

Choriocarcinoma Alkaline Phosphatase 12613

TABLE I Effects of L-leucine. bdwwlalanine. and EDTA on ALP actiuities

Choriocarcinoma Compounds Control/ Butyrate/ ALP Placental

ALP ALP ’?6 inhibition

L-Leucine (10 mM) 58 f 3” 79 f 2 15 f 5 L-Phenylalanine (10 mM) 67 f 5 70 f 6 60 f 2 EDTA (IO mM) 7 O f 5 7 4 + 2 1 4 f 4 Mean f S.E.

Synthesis

Phosphate Binding

‘ a 1 a P) c

67K \

65K -

I I

control Tunicamycin

- 93K

n 30K

FIG. 1. SDS-polyacrylamide gel electrophoresis of placen- tal and choriocarcinoma ALPs. Phosphate binding: partially pu- rified placental ( P U P ) or butyrate-treated JEG-3 choriocarcinoma ALP (JEGlbutyrate A P ) was labeled by phosphate binding as de- scribed under “Experimental Procedures.” Synthesis: JEG-3 chorio- carcinoma cells grown for 3 days in the presence of 3 mM sodium butyrate or primary placental cells were labeled with ~-[“sS]methio- nine in the presence or absence of tunicamycin (1 pg/ml) for 2 h. ALPS were isolated by immunoprecipitation with antiplacental ALP serum and analyzed by gel electrophoresis and fluorography. To demonstrate clearly that a difference in mobility occurs between these two phosphatases, placental and choriocarcinoma ALPS were pre- mixed (plac~nta + JEG/butyrate) before electrophoresis.

processed ALP monomer in control and butyrate-treated cells is a 65-kDa glycoprotein (38). However, the choriocarcinoma ALP monomer (65 kDa) labeled by [32P]phosphate migrated faster than the corresponding 32P-labeled placental ALP mon- omer (67 kDa) after electrophoresis in SDS-polyacrylamide gels (Fig. 1). Since ALPs are glycoproteins that may behave abnormally during gel electrophoresis, we compared the elec- trophoretic mobilities of the unglycosylated ALP monomers. Choriocarcinoma and primary placental cells were pulse la- beled with [“Slmethionine in the absence or presence of the

protein glycosylation inhibitor tunicamycin, and the newly synthesized ALPs were analyzed (Fig. 1). Tunicamycin inhib- its glycosylation via the dolichol pathway such that oligosac- charide side chains normally bonded by a glycosylamine link- age are not synthesized (39). Both the glycosylated (65 kDa) and nonglycosylated (58 kDa) forms of choriocarcinoma ALP monomers migrated faster than the corresponding placental ALP monomers of 67 and 59 kDa, respectively.

In the presence of butyrate, the steady-state levels of ALP mRNA in choriocarcinoma cells were greatly increased and reached levels as high as those observed in term placenta (Fig. 2). We have shown previously that the ALP mRNA found in three choriocarcinoma cell lines (JEG-3, JAR, and BeWo) is a 2.6-kb polynucleotide but the placental ALP mRNA is a 2.8-kb polynucleotide (31, 40). Sodium butyrate stimulated primarily the 2.6-kb ALP mRNA, suggesting that malignant placental cells express a different or altered ALP mRNA.

Induction of Placental Alkaline Phosphatase Synthesis by Sodium Butyrate-Since the placental ALP-like isozyme in choriocarcinoma cells appears to be preferentially induced by butyrate, we examined the effects of butyrate on placental ALP gene expression in primary placental cultures that ex- press mainly the placental ALP mRNA. The results illus- trated in Fig. 3 demonstrate that placental ALP mRNA and synthesis in primary placental cells could be stimulated by butyrate. Although butyrate slightly inhibited the growth rate of choriocarcinoma cells, cell viability was maintained in the presence of butyrate. In contrast, this short chain fatty acid was very toxic to primary placental cells, and only a low concentration of butyrate (1 mM) could be used in this study. This may account for the low level (3-4-fold) of ALP stimu- lation observed (Fig. 3).

SI Nuclease Analysis-To examine sequence divergence between choriocarcinoma and placental ALP mRNAs, a series of S1 nuclease mapping experiments was performed using probes derived from a placental ALP cDNA (TPAP51A). Near full length cDNA (9,41) and genomic clones (4) encoding placental ALP have been isolated and characterized. However, the sequence in the 5“untranslated region of the cDNA reported by Kam et al. (9) differs greatly from the cDNA sequence reported by Millan (41) and the genomic sequence reported by Knoll et al. (4). The TPAP51A clone used in the

”Cn ”- ~

FIG. 2. Increase of ALP mRNA levels in JEC-3 choriocar- cinoma cells by sodium butyrate. Poly(A)’ RNA for human term placenta, control, and butyrate-treated (3 mM for 3 days) choriocar- cinoma cells was separated by electrophoresis on formaldehyde-aga- rose gels. RNA was hybridized to a nick-translated ”P-labeled probe containing nucleotides 2233-2800 (568 bp, containing exons VII, VIII, and IX) of the germ cell ALP gene as described under “Experimental Procedures.”

Choriocarcinoma Alkaline Phosphatase 12614 Synthesis

- 4 * 93K- rl)

68K-- 67K

46K-m L

i

a -2.8-kb

FIG. 3. Increase of placental ALP biosynthesis and mRNA levels in primary placental cells by sodium butyrate. Primary placental cells were grown in the absence or presence of 1 mM butyrate for 3 days. Cultures were either labeled with ~-[~‘S]methionine for 3 h (synthesis) or lysed with guanidium thiocyanate for RNA isolation (mRNA). ALPS were isolated by immunoprecipitation and analyzed by gel electrophoresis and fluorography. Total RNA (10 pgllane) was separated by electrophoresis on formaldehyde-agarose gels, trans- ferred to Zetabind membrane, and hybridized with a probe containing nucleotides 2233-2800 of the germ cell ALP gene as described under “Experimental Procedures.”

present study is a near full length placental ALP cDNA clone isolated in this laboratory which contains a 5‘ sequence identical to that reported by Millan.

Two uniformly labeled single-stranded DNA probes con- taining 230 bases of M13 vector sequence and nucleotides 1- 145 and 1-261 of TPAPBlA, respectively, were used to map the 5‘ region of the choriocarcinoma transcript (Fig. 4, A and B ) . Placental RNA yielded 145- and 261-base protected frag- ments with the 1-145 and 1-261-nucleotide probes, respec- tively. However, the butyrate-treated choriocarcinoma RNA yielded a major 74-base protected fragment with the 1-145- nucleotide probe and a major 190-base protected fragment with the 1-261-nucleotide probe. This analysis localized one region of divergence at the 5’ region; the homology between placental and choriocarcinoma ALP mRNAs begins at nu- cleotide 72 of the placental ALP mRNA. By S1 analysis, control choriocarcinoma RNA yielded similar protected frag- ments at greatly reduced concentrations, indicating that bu- tyrate increases expression of a preexisting ALP mRNA in these cells. Fully protected fragments were also detected in butyrate-treated choriocarcinoma mRNAs, suggesting that these cells also express a low level of the placental ALP mRNA (Fig. 4, A and B) .

A uniformly labeled probe containing nucleotides 1221- 1528 of TPAP51A was used to examine (by S1 nuclease analysis) the 3’-coding region of the choriocarcinoma tran- script (Fig. 4C). Both placental and choriocarcinoma RNAs yielded one protected fragment of 308 bases, indicating that both mRNAs contain similar sequences in that region. The 308-base probe was synthesized by primer extension of TPAP51A using a 20-base primer corresponding to nucleo- tides 1509-1528 of TPAP51A. The 308-base fragment, re- leased by SstI (nucleotide 1221) digestion and purified by gel electrophoresis, was copurified with a 360-base fragment that corresponds to nucleotides 862-1221 of TPAP51A. As a result,

the 360-base protected fragment was also detected in placental and choriocarcinoma RNAs (Fig. 4C). Our data indicate that the choriocarcinoma RNA also contains nucleotides 862-1221 of the placental ALP mRNA sequence.

To examine sequence divergence between placental and choriocarcinoma ALP mRNAs in the 3‘ regions, two uni- formly labeled probes containing nucleotides 1221-1643 and 1221-1828 of TPAP51A were used in S1 nuclease analysis (Fig. 4, D and E) . Both mRNAs yielded one 423-base pro- tected fragment with the 1221-1643-nucleotide probe (Fig. 40). Since nucleotide 1643 is the 3‘ boundary of the coding region of placental ALP, our data indicate that placental and choriocarcinoma ALP mRNAs have similar sequences at the 3’ end of the coding region. However, the 1221-1828-nucleo- tide probe yielded a 608-base protected fragment with placen- tal RNA and a major 540-base protected fragment with the butyrate-treated choriocarcinoma RNA, indicating that se- quence homology between the two mRNAs was interrupted at nucleotide 1712 of the placental ALP mRNA. The presence of the 608-base fully protected fragment with butyrate-treated choriocarcinoma RNA again indicates that these cells express low levels of the placental ALP mRNA. A 540-base protected fragment was also observed with control choriocarcinoma RNA after longer exposure (data not shown).

The nucleotide sequences of placental and germ cell ALP mRNA predicted from cDNA or genomic sequence analysis indicate that the two ALP mRNAs differ significantly in sequences at the 5’- and 3”untranslated regions (5, 6). The predicted sequence divergences of these two ALP mRNAs are similar to those observed between choriocarcinoma and pla- cental ALP mRNAs. Thus, the S1 nuclease mapping experi- ments suggest that choriocarcinoma cells express the germ cell ALP gene.

Ribonuclease Protection Anulysk-To confirm that chorio- carcinoma cells express the germ cell ALP mRNA, a series of RNase protection assays was performed using probes derived from the germ cell ALP gene. The entire germ cell ALP gene was contained within the 4637-bp SstI-Hind111 fragment of the XEMBL-3 genomic clone GCAPGSHPO, which was iso- lated by screening a XEMBL-3 genomic library with the 5’ EcoRI-KpnI insert of the placental ALP cDNA clone TPAP8E (31). GCAPGSHZO contains 154 bp of additional 5’ sequence of the germ cell ALP gene sequence reported previ- ously (5). Riboprobes generated from the four GCAPGSHPO fragments (1-1274b, 1275-2232b, 2233-2800b, and 2801- 4637b) that contain the entire germ cell ALP gene were used to examine the ALP transcript in JEG-3 choriocarcinoma cells (Fig. 5).

Choriocarcinoma RNA yielded two protected fragments of 122 and 125 bases with riboprobe 1-1274b, which contains exons I, I1 (117 bases, nucleotides 759-875), and I11 (116 bases, nucleotides 987-1102) of the germ cell ALP gene (Fig. 5A). Since a DNA size marker was used in this analysis and RNA has a lower mobility (5-10%) than DNA of the same length, the protected fragments after correction were about 117 and 119 bases, respectively (Fig. 5A). The predicted bands of 117 and 116 bases could not be clearly separated by this gel and appeared as a single band of 122 bases (uncorrected) in length. Thus, our data indicate that choriocarcinoma cells express the germ cell ALP mRNA. This analysis also defined the size of exon I of the germ cell ALP gene as 119 bases (nucleotides 546-664). The predicted fragments protected by placental ALP mRNA (4, 9,41) are 117 (exon 11), 72, and 40 bases (exon 111), and 49 bases (exon I); the results in Fig. 5A agree with this prediction.

Riboprobe 1275-2232b, which contains exons IV (175 bases,

Choriocarcinoma Alkaline Phosphatase 12615

39b 1643b

s d 4 Y

- 0 Z 1-145b

- L m z 1-261b

A 1-145b I a 1

- e

- u) v)

e 2 1221-1528b

- Z 1221-1643b

- 5 1221-1828b

0 1-261b C 1221-1528b D 1221-1643b

I"I 1OOb

E 1221-1826b

I Q I

261 - 190 -

600 - 540 -

FIG. 4. S1 nuclease mapping of choriocarcinoma ALP mRNA. Poly(A)' RNA from human term placenta (5 pg), control (10 pg), or butyrate-treated (5 pg) choriocarcinoma cells was annealed to a uniformly labeled probe corresponding to nucleotides 1-145, 1-261, 1221-1528, 1221-1643, or 1221-1828 of a near full length placental ALP cDNA, TPAP51A. Human term placenta expresses primarily the placental ALP mRNA, and yeast tRNA (10 pg) was used as a control. The 1-145 and 1-261 probes released by NarI digestion contained 230 bases of M13 vector sequence. Nucleotide 1643 is the 3' boundary of the placental ALP-coding region. The hybrids were digested with S1 nuclease as described under "Experimental Procedures." The S1 nuclease-protected fragments were separated by electrophoresis on 8% polyacrylamide-urea sequencing gels. HaeIII digestion of 4x174 was used as a marker.

-310

,201

'271 , 234

, q94

.118

-72

nucleotides 1295-1469), V (173 bases, nucleotides 1547-1719), VI (135 bases, nucleotides 1967-2101), and the 5' region of VI1 (32 bases, nucleotides 2201-2232), and riboprobe 2233- 2800b, which contains the 3' region of exons VI1 (41 bases, nucleotides 2233-2273), VI11 (135 bases, nucleotides 2402- 2536), and IX (183 bases, nucleotides 2618-2800), were used to map the middle region of the ALP transcript. Choriocar- cinoma ALP mRNA yielded the predicted protected frag- ments with both probes (Fig. 5, B and C). Placental ALP mRNA yielded similar protected fragments using riboprobe 2233-2800b (Fig. 5C) but slightly different protected frag- ments (173, 167, and 135 bases) using riboprobe 1275-2232b

(Fig. 5B). The single base pair substitutions present in exon IV of the two genes may account for the observed difference.

Riboprobe 2801-4637b, which contains exons IX (9 bases, nucleotides 2801-2809), X (117 bases, nucleotides 3029-3145), and XI, yielded two major protected fragments of 1400 and 117 bases (corrected) and two minor bands of 750 and 470 bases with choriocarcinoma RNA (Fig. 50) . The 117-base band maps exon X, and the 1400-base band maps exon XI. Control choriocarcinoma ALP mRNAs yielded similar pro- tected fragments after longer exposure (data not shown). Since the size of the large fragment cannot be determined reliably by this analysis, the 3' boundary of exon XI of the

12616

FIG. 5. Ribonuclease protection analysis of choriocarcinoma ALP mRNA. Poly(A)’ RNA from human term placenta (2 pg), control (10 pg) , or butyrate-treated (5 pg) choriocarcinoma cells was annealed to a uniformly labeled riboprobe corresponding to nucleotides 1-1274, 1275-2232,2233-2800, or 2801- 4637 of the germ cell ALP gene. Human term placenta expresses primarily the placental ALP mRNA, and yeast tRNA (10 pg) was used as a control. Exons (black boxes) contained in each probe were indicated. The hybrids were di- gested with RNases A and TI as de- scribed under “Experimental Proce- dures.” The ribonuclease-protected frag- ments were separated by electrophoresis on polyacrylamide-urea sequencing gels. The HaeIII digestion of $X174 was used as a marker.

Choriocarcinoma Alkaline Phosphatase

1-1274b 1 -

I I1 111

119b I l 7 b 116b 1275-2232b

4 1 0 1350 183b 2801-4637b

1353 \

1078.

872: 603

310 281: 271- 234 - 194 -

118-

72 -

$117 119

116

72

49

40

germ cell ALP gene could not be defined. The two minor bands of 750 and 470 bases may represent ALP polymorphism in choriocarcinoma cells. Placental RNA yielded two pro- tected fragments of 368 (corrected) and 117 bases (corrected) (Fig. 5D), which were the sizes of the predicted protected fragments for exons X and XI of the placental ALP gene.

Isolation and Characterization of cDNA Clones Encoding the Germ Cell Alkaline Phosphatase-To demonstrate posi- tively that choriocarcinoma cells express the germ cell ALP gene and to determine the structure of its 3’ exon, we con- structed a cDNA library using poly(A)+ RNA from butyrate- treated JEG-3 choriocarcinoma cells and screened the library with an oligonucleotide probe corresponding to nucleotides 598-639 of the germ cell ALP gene, a sequence that maps inside its first exon (nucleotides 546-644). The longest cDNA clone, GCAP8, contains nucleotides 54-2487 of germ cell ALP mRNA. The restriction map of GCAP8 and the nucleotide

- 183

- 135

- 41

- 1400

- 750

- 470

. 368

117

sequence of germ cell ALP mRNA (2487-bases) are illustrated in Fig. 6. Exon XI is 1135 bp in length (corresponding to nucleotides 3268-4402 of the germ cell ALP gene) and con- tains 839 bases of 3’-untranslated sequence, including one polyadenylation signal (AATAAA) at nucleotides 2463-2468 (Fig. 6). The nucleotide sequence of choriocarcinoma ALP mRNA differs slightly from the sequence predicted from the germ cell ALP gene reported by Millan and Manes (5), and the deduced amino acid sequences of the two mature proteins differ at 5 amino acid residues (Table 11). Amino acid residues in positions 38, 133, 297, 361, and 479 of choriocarcinoma ALP are isoleucine, methionine, arginine, leucine, and argi- nine, whereas the corresponding amino acid residues in the germ cell ALP of Millan and Manes are methionine, valine, leucine, valine, and proline. These amino acid substitutions may represent allelic polymorphism in germ cell ALPS.

Localization of the Cap Site of the Germ Cell Alkaline Phos-

Choriocarcinoma Alkaline Phosphatase 12617

112 I

126 39

340 77

454 115

568 153

682 191

796 229

910 167

1014 305

1138 343

1252 381

I366 419

1480 457

1594 495

1726 513

1876

1026

1177

1317

2477

2487

phatase rnRNA-The precise position of the germ cell ALP mRNA cap site was established by primer extension experi- ments using two 5"labeled 20-base oligonucleotide probes corresponding to nucleotides 35-54 and nucleotides 86-105 of germ cell ALP mRNA (Fig. 6). The probes were annealed to poly(A)+ RNA from butyrate-treated choriocarcinoma cells, and the extended end-labeled products were sized on a se- quencing gel (Fig. 7). One major extended fragment of 54 bases and one minor extended fragment of 55 bases were obtained with the oligonucleotide probe corresponding to nucleotides 35-54. One extended fragment of 105 bases with the oligonucleotide probe corresponding to nucleotides 86- 105 was also obtained. These results indicate that germ cell ALP mRNA has one major 5' terminus. These data agree with the results obtained with the ribonuclease protection

analysis and demonstrate that exon I of the germ cell ALP gene is 119 bp.

DISCUSSION

The placental ALP-like germ cell isozyme was discovered by Nakayama and co-workers in 1970 (15) and has subse- quently been found in trace amounts in the testis and thymus (16, 17) and in elevated amounts in germ cell tumors (18, 19). However, due to the high allelic polymorphism observed in human placental ALP, the existence of the fourth human ALP gene was not confirmed until the germ cell ALP gene was isolated and characterized in several laboratories (5, 6, this study). The structure of the germ cell ALP gene was initially determined based on its sequence similarity to the placental ALP cDNA (5). As a result, Millan and Manes (5)

12618 Choriocarcinoma Alkaline Phosphatase

TABLE I1 Amino acid substitutions in placental and germ cell ALPs

Positions Placental ALP ~ ~ ~ ~ $ ! ~ ~ ~ a Germ cell ALP"

15 GMGAG) Gln(CAG) Gln(CAG) 38 Ile(ATC) Ile(ATC) Met(ATG) 67 Ile(ATA) Thr(ACC) Thr(ACC) 68 Pro(CCC) Phe(TTC) Phe(TTC) 84 Asn(AAT) Ser(AGT) Ser(AGT)

133 Met(ATG) Met(ATG) Val(GTG) 241 Arg(CGC) His(CAC) His(CAC) 254 Met(ATG) Leu(CTG) Leu(CTG) 297 Arg(CGC) Arg(CGC) Leu(CTC) 361 Val(GTC) Leu(CTC) Val(GTC) 429 Glu(GAA) Gly(GGA) Gly(GGA) 479 Pro(CCC) Arg(CGC) Pro(CCC)

a Amino acid sequences deduced from the germ cell ALP gene reported by Millan and Manes (5).

were unable to define the boundary of exon I and had inad- vertently determined that exon XI has a length of 800 bp. Furthermore, in the absence of cDNA or mRNA data, it was uncertain whether this gene was expressed in humans. By isolating germ cell ALP cDNA clones from a cDNA library constructed with poly(A)+ RNA from butyrate-treated chorio- carcinoma cells, we not only defined the 5' and 3' boundaries of this gene but also established unequivocally the existence of the germ cell ALP gene in humans.

The structures of germ cell, placental, and intestinal ALP genes are similar; each is composed of 11 exons interrupted by 10 intervening sequences. Sequence analysis of germ cell ALP cDNA clones along the corresponding gene demonstrates that exons I and XI (3' exon) of the germ cell ALP gene are 119 and 1135 bp, respectively. The latter is 204 bp shorter than the corresponding exon in the placental ALP gene (1339 bp), which accounts for the smaller size of the germ cell ALP mRNA (2.6 kb) as compared with the placental ALP mRNA (2.8 kb). There appears to be one major transcription initia- tion site for the germ cell ALP mRNA differing from that observed for the placental ALP mRNA, which contains a t least three major transcription initiation sites (4). Placental (4) and intestinal (7) ALP genes contain several polyadenyl- ation signals, and the various cell line intestinal ALP mRNA species result from the differential use of a t least three of the four polyadenylation signals present in the intestinal ALP gene (7). In contrast, the germ cell ALP gene contains only one polyadenylation signal, and only one mRNA species has been demonstrated to date. The functional significance of these findings remains to be elucidated.

It has been demonstrated that sodium butyrate greatly increases ALP mRNA and activity in the extraembryonic germ cell tumor (choriocarcinoma) cells. In addition, chorio- carcinoma cells express the germ cell ALP gene either in the absence or presence of butyrate. This supports the theory that germ cell ALP gene is expressed primarily in tissues or cells of germ cell origin (18, 19). Although placental ALP mRNA could not be detected in control choriocarcinoma cells, low levels of placental ALP mRNA were found in butyrate- treated choriocarcinoma cells, suggesting that expression of the placental ALP gene is also stimulated by this fatty acid. Moreover, butyrate was capable of increasing both placental ALP mRNA and synthesis in primary placental cells, which express primarily the placental ALP gene. Therefore, both forms of ALP are responsive to butyrate induction, and cho- riocarcinoma cells in culture express only low levels of the placental ALP gene. Our results demonstrate that the malig- nant transformation of human placenta suppresses expression of the placental ALP gene but activates expression of the

I 0 , ' 0 , ' e c m a

; 1078 872 '603

-310 -28 1 '27 1 ,234

-194

-1 18

-105

- 72

/ 55 ' 54

FIG. 7. Mapping of the transcription initiation site of germ cell ALP mRNA. Poly(A)+ RNA of hutyrate-treated choriocarci- noma cells was annealed to a "P-labeled 20-base primer correspond- ing to nucleotides 35-54 (primer A ) or nucleotides 86-105 (primer B ) of the germ cell ALP mRNA. The primer was extended by reverse transcriptase, and the extended products were analyzed by electro- phoresis on polyacrylamide-urea sequencing gels. The HaeIII diges- tion of 4x174 and the sequencing ladder were used as markers.

ALP gene characteristic of germ cells. Choriocarcinoma cells provide an excellent model to study the molecular mechanism responsible for the repression of placental ALP gene expres- sion and induction of germ cell ALP gene expression following malignant transformation.

The amino acid sequences of mature placental and germ cell ALPs as deduced from cDNA sequences (Fig. 6 and Refs.

Choriocarcinoma A

9 and 41) and the amino- and carboxyl-terminal analysis of mature proteins (13, 23, 24) indicate that both ALPs are composed of 484 amino acid residues. Placental ALP differs from the choriocarcinoma germ cell ALP in 9 amino acid residues and differs from the germ cell ALP of Millan and Manes (5) in 10 amino acid residues (Table 11). The chorio- carcinoma germ cell isozyme contains 2 less charged amino acid residues (glutamic acid) than the placental ALP, which may account for its faster electrophoretic mobility on SDS- polyacrylamide gels. If the two germ cell ALPs represent allelic variants, the amino acid substitutions in positions 15, 67, 68, 84, 241, 254, and 429 may account for the increased sensitivity of germ cell ALP to L-leucine and EDTA. Site- directed mutagenesis analysis should allow us to identify the amino acid residue(s) responsible for the observed differences in physicochemical properties between the placental and germ cell ALPs.

Acknowledgments-We thank Drs. C. Plouzek, J. Kasik, and C. Roberts for critical reading of the manuscript.

REFERENCES

1. Mulivor, R. A., Plotkin, L. I., and Harris, H. (1978) Ann. Hum.

2. Stigbrand, T., Millan, J. L., and Fishman, W. H. (1982) Isoen-

3. Harris, H. (1982) Harvey Lect. 7 6 , 95-124 4. Knoll, R. J., Rothblum, K. N., and Longley, M. (1988) J. Biol.

Chem. 263,12020-12027 5. Millan, J. L., and Manes, T. (1988) Proc. Natl. Acad. Sci. U. S. A .

85,3024-3028 6. Knoll, B. J., Rothblum, K. N., and Longley, M. (1987) Gene

(Amst.) 60,267-276 7. Henthorn, P. S., Raducha, M., Kadesch, T., Weiss, M. J., and

Harris, H. (1988) J. Biol. Chem. 263 , 12011-12019 8. Weiss, M. J., Ray, K., Henthorn, P. S., Lamb, B., Kadesch, T.,

and Harris, H. (1988) J. Biol. Chem. 2 6 3 , 12002-12010 9. Kam, W., Clauser, E., Kim, Y. S., Kan, Y. W., and Rutter, W. J.

(1985) Proc. Natl. Acad. Sci. U. S. A. 82,8715-8719 10. Martin, D., Tucker, D. F., Gorman, P., Sheer, D., Spurr, N. K.,

and Trowsdale, J. (1987) Ann. Hum. Genet. 5 1 , 145-152 11. Swallow, D. M., Povey, S., Parker, M., Andrews, P. W., Harris,

H., Pym, B., and Goodfellow, P. (1986) Ann. Hum. Genet. 50,

12. Sussman, H. H., Small, P. A., Jr., and Cotlove, E. (1968) J. Biol.

13. Ezra, E., Blacher, R., and Udenfriend, S. (1983) Biochem. Bio-

14. Hua, J.-C., Berger, J., Pan, Y.-C., Ulmes, J. D., and Udenfriend,

Genet. 4 2 , 1-13

zymes 6,93-117

229-235

Chem. 2 4 3 , 160-166

phys. Res. Commun. 116, 1076-1083

lkaline Phosphatase 12619

S. (1986) Proc. Natl. Acad. Sci. U. S. A . 8 3 , 2368-2372 15. Nakayama, T., Yoshida, M., and Kitamura, M. (1970) Clin. Chim.

Acta 30,546-548 16. Chang, C. H., Angellis, D., and Fishman, W. H. (1980) Cancer

Res. 4 0 , 1506-1510 17. Goldstein, D. J., Rogers, C., and Harris, H. (1982) Clin. Chim.

Acta 125,63-75 18. Wahren, B., Holmgren, P. A,, and Stigbrand, T. (1979) Int. J .

Cancer 24 , 749-753 19. Horwich, A., Tucker, D. F., and Peckham, M. J. (1985) Br. J .

Cancer 5 1 , 625-629 20. Harris, H., Hopkinson, D. A., and Roboson, E. B. (1974) Ann.

Hum. Genet. 37,237-253 21. Millan, J . L., Beckman, G., Jeppsson, A., and Stigbrand, T. (1982)

Hum. Genet. 60,145-149 22. Henthorn, P. S., Knoll, B. J., Raducha, M., Rothblum, K. N.,

Slaughter, C., Weiss, M., Lafferty, M. A., Fisher, T., and Harris, H. (1986) Proc. Natl. Acad. Sci. U. S. A . 8 3 , 5597-5601

23. Micanovic, R., Bailey, C. A., Brink, L., Gerber, L., Pan, Y.-C. E., Hulmes, J. D., and Udenfriend, S. (1988) Proc. Natl. Acad. Sci.

24. Ogata, S., Hayashi, Y., Takami, N., and Ikehara, Y. (1988) J.

25. Low, M. G., and Saltiel, A. R. (1988) Science 2 3 9 , 268-275 26. Sakiyama, T., Robinson, J . C., and Chou, J . Y. (1978) Arch.

Biochem. Biophys. 191, 782-791 27. Neuwald, P. D., and Brooks, M. (1981) Cancer Res. 4 1 , 1682-

1689 28. Novak, E. R., and Woodruff, J. D. (1979) Gynecologic and Obstet-

ric Pathology, pp. 476-485, W. B. Saunders Co., Philadelphia 29. Mclaughlin, P. J., and Johnson, P. M. (1984) in Human Alkaline

Phosphatases (Stigbrand, T., and Fishman, W. H., eds) pp. 67- 75, Alan R. Liss, Inc., New York

30. Kliman, H. J., Nestler, J. E., Sermasi, E., Sanger, J. M., and Strauss, J. F., I11 (1986) Endocrinology 118 , 1567-1582

31. Chou, J . Y., and Takahashi, S. (1987) Biochemistry 26 , 3596- 3602

32. Laemmli, U. K. (1970) Nature 227,680-685 33. Milstein, C. (1964) Biochem. J. 92,410-421 34. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl.

Acad. Sci. U. S. A. 74,5463-5467 35. Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J., and Rutter,

W. J. (1979) Biochemistry 18 , 5294-5299 36. Lehrach, H., Diamond, D., Wozney, J. M., and Boedtker, H.

(1977) Biochemistry 16 , 4743-4751 37. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman,

J. G., Smith, J. A., and Struhl, K. (1987) Current Protocols in Molecular Biology, John Wiley & Sons, New York

U. S. A. 8 5 , 1398-1402

Biol. Chem. 263,10489-10494

38. Ito, F., and Chou, J. Y. (1984) J. Biol. C k m . 2 5 9 , 2526-2530 39. Short, J. M., Wynshaw-Boris, A., Short, H. P., and Hanson, R.

40. Ovitt, C. E., Strauss, A. W., Alpers, D. H., Chou, J. Y., and Boime,

41. Millan, J . L. (1986) J. Biol. Chem. 2 6 1 , 3112-3115

W. (1986) J. Biol. Chem. 261,9721-9726

I. (1986) Proc. Natl. Acad. Sci. U. S. A . 8 3 , 3781-3785