Cloning and expression of the chick anti-Müllerian hormone gene.

Müllerian duct regression in male embryos is due to early production by fetal Sertoli cells of anti-Müllerian hormone, a homodimeric protein of the transforming growth factor- beta superfamily. In mammals, both female Müllerian ducts develop into the uterus and Fallopian tubes, whereas in birds, the right oviduct does not develop. To gain insight into sex differentiation in birds, we have cloned the cDNA for chick anti-Müllerian hormone using antibodies raised against the partially purified protein. Expression cloning was required because of the lack of cross-hybridization between mammalian and chick anti-Müllerian hormone DNA. The chick DNA and protein are significantly longer, due to insertions that abolish nucleotide homology, except in the cDNA coding for the C-terminal, bioactive part of the protein. Nevertheless, the general structure of the gene, sequenced from the transcription initiation to the polyadenylation site, and the main features of the protein are conserved between the chick and mammals. The chick anti-Müllerian hormone gene is expressed at high levels in Sertoli cells of the embryonic testes and in lower amounts in both ovaries, higher levels being reached on the left side after 10 days of incubation.

Mü llerian duct regression in male embryos is due to early production by fetal Sertoli cells of anti-Mü llerian hormone, a homodimeric protein of the transforming growth factor-␤ superfamily. In mammals, both female Mü llerian ducts develop into the uterus and Fallopian tubes, whereas in birds, the right oviduct does not develop. To gain insight into sex differentiation in birds, we have cloned the cDNA for chick anti-Mü llerian hormone using antibodies raised against the partially purified protein. Expression cloning was required because of the lack of cross-hybridization between mammalian and chick anti-Mü llerian hormone DNA. The chick DNA and protein are significantly longer, due to insertions that abolish nucleotide homology, except in the cDNA coding for the C-terminal, bioactive part of the protein. Nevertheless, the general structure of the gene, sequenced from the transcription initiation to the polyadenylation site, and the main features of the protein are conserved between the chick and mammals. The chick anti-Mü llerian hormone gene is expressed at high levels in Sertoli cells of the embryonic testes and in lower amounts in both ovaries, higher levels being reached on the left side after 10 days of incubation.
Regression of Mü llerian ducts, the primordia for female genital ducts, is mediated in male embryos by anti-Mü llerian hormone (AMH), 1 also called Mü llerian inhibiting substance, a member of the transforming growth factor ␤ (TGF-␤) family (1,2). Mammalian AMHs are homodimers formed of N-and Oglycosylated protein chains linked by disulfide bonds. After removal of the signal peptide and dimerization, AMH undergoes an activating peptide cleavage at the target tissue level (3). The C-terminal fragment is the active moiety, but its bioactivity is strongly enhanced by the presence of the N-terminal fragment (4). In mammals, AMH is synthesized by Sertoli cells immediately after testicular differentiation, but AMH production by granulosa cells begins only after birth (5). Untoward exposure of fetal mammalian female reproductive organs to AMH results in Mü llerian regression and severe ovarian le-sions (6,7). In birds, the situation is somewhat different. Embryonic gonads of both sexes are endowed with anti-Mü llerian activity (8,9), but this does not affect the development of the female left Mü llerian duct; only the right one regresses in female embryos. The role of AMH in avian sex differentiation has not been investigated in depth because mammalian probes do not recognize chick AMH and because human AMH, the only recombinant hormone available at the present time, is inactive in the chicken (10). Purification of avian AMH from chick testicular tissue has been reported some time ago (11) but has not led to further molecular developments. To obtain tools appropriate for molecular analysis, we have successfully cloned the chick AMH gene (ckAMH), using an expression cloning approach.

EXPERIMENTAL PROCEDURES
Obtention of Anti-chick AMH Rabbit Immunoglobulins-Chick AMH was obtained from 16-day-old White Leghorn chick embryo testes maintained in organ culture for 3 days. The hormone was partially purified from the culture medium by affinity chromatography on a Lens culinaris lectin column (Sigma) followed by a preparative polyacrylamide gel electrophoresis as described (9). Polyacrylamide gel fragments containing the AMH band were crushed, mixed with adjuvant, and injected thrice at monthly intervals to a female rabbit sacrificed 1 month after the last injection. Immunoglobulins were purified by adsorption on a protein A-Sepharose-4 Fast Flow column (Pharmacia Biotech Inc.).
Construction of a gt11 cDNA Expression Library-A cDNA library was constructed from 16-day-old chick embryo testis RNA, according to Young and Davis (12). Total RNA was extracted from 120 mg of frozen tissue (approximately 100 testes) as described (13). Poly(A) ϩ RNA was purified by oligo(dT)-cellulose binding. cDNA was synthesized by murine Moloney leukemia virus reverse transcriptase, using 4 g of poly(A) ϩ RNA and oligo(dT) [12][13][14][15][16][17][18] as first strand primer (Time Saver cDNA synthesis kit, Pharmacia). EcoRI-NotI cohesive-end adaptors were added to double-stranded cDNA of a size above 1 kb. 67 ng of the resulting cDNA were ligated to 1.3 g of dephosphorylated EcoRIdigested gt11 vector (Promega). Packaging of recombinant DNA (Gigapack II Gold packaging extract, Stratagene), infection of Y1090 Escherichia coli bacteria and plating were carried out as recommended. The resulting library contained 4.5 ϫ 10 6 recombinant clones. The amplified library was derived from 1.1 ϫ 10 6 recombinant phages.
Expression Screening of the cDNA Library-Screening of the expression library with anti-AMH rabbit polyclonal antibodies was performed according to a modification of the protocol of de Wet et al. (14), using Vectastain Elite kit (Vector, Burlingame, CA). Approximately 6 ϫ 10 5 clones from the nonamplified library were adsorbed onto Y1090 E. coli bacteria, plated at a density of 5 ϫ 10 4 plaque-forming units per 150-mm diameter Petri dish, grown at 42°C for 3 h, and induced with isopropyl-␤-thiogalactoside-soaked Hybond C nitrocellulose filters (Amersham), at 37°C for 3 h, as described (12). Extensive washing of the filters with 1 ϫ Tris-buffered saline (0.1 M Tris-HCl, pH 7.5, 0.15 M NaCl) was performed between each incubation step. The membranes were incubated at room temperature in the blocking solution (1 ϫ Tris-buffered saline, 2% glycine, 3% lowfat milk, 0.02% sodium azide) for 30 min and overnight at 4°C in a solution of anti-ckAMH rabbit antibodies diluted 1:1000 (12.8 g of protein/ml) in 1 ϫ Tris-buffered saline, 2% glycine, 3% bovine serum albumin, and 0.02% sodium azide. After a second incubation in the blocking solution for 15 min and in biotinylated goat anti-rabbit IgG antibodies for 30 min in the presence * 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 sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) X89248.
of an excess of normal goat serum, the filters were incubated with avidin-biotinylated peroxidase complex for 30 min. Peroxidase was revealed by reaction with oxygen peroxide and 4-chloro-1-naphtol (Vector). Positive clones are seen as fast appearing purple spots or rings on a background of more slowly staining plaques. Positive clones were further purified by three rounds of replating and screening.
Rapid Amplification of 5Ј-cDNA Ends (RACE) by PCR-The 5Ј-end of ckAMH cDNA was amplified and cloned using a modification of the RACE procedure of Frohman et al. (16). First strand cDNA was synthesized by reverse transcription of 10 g of total RNA from 16-day-old chick embryo testes, using 15 pmol of oligonucleotide a3, located in exon 2, as ckAMH-specific primer (Table I) and 800 units of murine Moloney leukemia virus reverse transcriptase (Life Technologies, Inc.), in the buffer provided with the enzyme with addition of 10 mM dithiothreitol, 1 mM each of the four dNTPs, and 20 units of RNasine (Promega), in a volume of 50 l for 1 h at 42°C. The enzyme was inactivated by heating at 95°C for 10 min. The RNA was hydrolyzed with RNase A (5 g) at 37°C for 20 min and freed from nucleotides and primers by centrifugation through a Centricon-100 spin filter (Amicon). The following steps were performed on half the volume of each sample. For poly(dG) tailing, cDNA was first denatured at 65°C for 5 min and placed on ice. Denatured DNA was incubated at 37°C for 15 min with 50 units of terminal deoxyribonucleotide transferase (Boehringer Mannheim) in 0.2 M potassium cacodylate, 25 mM Tris-HCl at pH 6.6, 5 mM dGTP, 0.25 mg/ml bovine serum albumin, and 0.75 mM CoCl 2 (20 l, final volume). Subsequent amplification of cDNAs was performed by 40 cycles of PCR as described (15), with primer a7 as nested ckAMH primer (50 pmol) and 20 pmol of a poly(dC) containing anchor primer (Table I), using 3.35 or 4.47 mM MgCl 2 . PCR products were drop-dialyzed on VMWP membranes (Millipore, Marlborough, MA) floating on distilled water and cloned into pGEM plasmid (TA-cloning system, Promega). The clones of interest were screened by hybridization with 32 P end-labeled oligonucleotide a8 ( Table I).
Cloning of the Introns-The position of introns was determined by comparing the size of the products obtained by PCR amplification of genomic DNA and cDNA. Intron 1 was amplified using primer pairs s9/a3 or s11/a3 and introns 2-4 using primer pair s5/a9 (Table I). PCR fragments were filled in using the Klenow fragment of E. coli DNA polymerase and cloned into SmaI-digested phosphatase-treated M13 mp18 phage as described previously (15).
Sequencing-Recombinant gt11 DNA was purified by the plate method (17), and both ends of the inserts were sequenced using Cir-cumVent thermal cycle sequencing kit (New England Biolabs, Beverly, MA), with 0.9 pmol of 32 P or 33 P 5Ј-labeled forward or reverse primers (Promega) and 12 fmol of recombinant DNA for 25 cycles (95°C, 30 s; 50°C, 30 s; 72°C, 1 min). For complete sequencing, cDNA inserts were subcloned into Bluescript KSII(ϩ) (Stratagene) at EcoRI or NotI sites. Plasmid DNA was purified using Wizard minipreps (Promega) or CsCl gradient centrifugation and sequenced by the dideoxynucleotide method (18) with Sequenase 2.0 (U. S. Biochemical Corp.) using the procedure indicated or the denaturation method of Hsiao (19). Whenever possible, at least two independent overlapping cDNA clones were sequenced in both directions. Regions of the sequence difficult to elucidate by double-stranded DNA sequencing were PCR-amplified and subcloned into phage M13 mp18; at least four clones were sequenced as single-stranded DNA. Sequences were analyzed on 6% Hydrolink long ranger (Bioprobe Systems, Montreuil, France), 8 M urea sequencing gels, with 0.6 ϫ TBE (54 mM Tris borate, 1.2 mM EDTA) as running and gel buffers.
Southern Blot Analysis-Southern blot analysis was performed on genomic DNA extracted from chick liver, human blood, and the blood of the European pond turtle, Emys orbicularis. DNA (6 g/lane), digested with 12 units of HindIII or BamHI restriction endonucleases, separated on 0.8% agarose gel, and transferred to Hybond-N membrane (Amersham) was hybridized, as described above, with a random-primed 32 Plabeled probe spanning all the cDNA. The probe was made from equimolecular amounts of the inserts of a RACE clone (position 1-393) and cDNA clone 81 (position 331-4197). The membrane was washed at 65°C in 5 ϫ SSC, 0.1% SDS for 25 min; 2 ϫ SSC, 0.1% SDS for 15 min; and twice in 1 ϫ SSC, 0.1% SDS for 10 min.
In Situ Hybridization-In situ hybridization with a ckAMH digoxigenin-labeled riboprobe (DIG-RNA) was performed according to the procedure of Millar et al. (22). Testes and ovaries from 8-and 17-day-old chick embryos were removed, fixed in 4% paraformaldehyde or in Bouin's fluid (7.7% paraformaldehyde, 3.8% acetic acid (v/v) in saturated picric acid solution) for 5 h, processed into paraffin wax, and cut at 6 m. A Bluescript KSII(ϩ) plasmid vector containing the 821-bp 5Ј-terminal AccI fragment of clone 30 (position 2121-3108) was used to produce the probes. The vector was linearized by digestion with AccI or EcoRI and transcribed with T7 or T3 RNA polymerase to synthesize sense or antisense probes, respectively. Transcription was performed using the RNA digoxigenin labeling kit (Boehringer) as indicated. The sections were prehybridized at 55°C for 2 h and hybridized with sense or antisense digoxigenin-labeled probes overnight at 55°C. DIG-RNA was detected with anti-DIG antibody coupled to alkaline phosphatase (Boehringer) and revealed by reaction with nitro blue tetrazolium salt, 5-bromo-4-chloro-3-indolyl phosphate (Boehringer), and levamisole (Sigma) in the dark.

RESULTS
Cloning of Chicken AMH cDNA-18 positive clones were detected by screening 6 ϫ 10 5 independent clones from a gt11 expression cDNA library of 16-day-old chick embryo testicular tissue, using rabbit polyclonal antibodies raised against a partially purified preparation of ckAMH; both extremities of their inserts were sequenced. Comparison of the encoded protein sequences with protein data banks, using the BLASTP program, allowed the identification of one partial ckAMH cDNA clone, clone 13 (spanning nucleotides 2103-4190). It was used as hybridization probe to identify two additional ckAMH clones, clones 23 (spanning nucleotides 571-4179) and clone 30 (spanning nucleotides 2121-4200), which ends with a 21-nucleotide-long poly(A) tail. The other immunologically positive clones presented no homology with AMH; their selection may be due to immunological cross-reaction or to contaminants present in the ckAMH protein preparation used to raise antibodies. A second screening, performed on 1.2 ϫ 10 6 clones of the amplified gt11 library, using a 5Ј-terminal fragment of clone 23 as hybridization probe, recovered 120 positive clones. Clones 81 (spanning nucleotides 331-4197) and 165 (starting at nucleotide 185) harbored the longest inserts but still lacked an ATG initiation codon and a sequence coding for a signal peptide. The 5Ј-end of ckAMH cDNA was obtained by the technique of RACE, using total RNA from 16-day-old chick embryo testis. 47 positive clones were isolated, 19 had inserts corresponding to a cDNA fragment of about 400 bp, and all other inserts were smaller.
Nucleotide Sequence of the ckAMH Gene-The complete nucleotide sequence of the ckAMH gene is shown in Fig. 1. The sequence of the introns was determined on cloned PCR-amplified genomic DNA fragments. The 5Ј-end of the cDNA was determined on clones from two RACE experiments. The 19 large inserts started in a common region, while smaller inserts started at various downstream positions. Several 5Ј termini were found, corresponding to alternative transcription sites. These can be identified through the presence of a copy of the cap G, present in some of the reverse-transcribed RACE clones but not in DNA (23). The main initiation site is designated A1; 10 clones out of 13 begin with an extra G. Two minor initiation sites were also detected. One is located at Gϩ2 (2 clones out of 3 have an extra G). The second putative one is located 2 bases upstream from A1; in 2 clones A1 is preceded by a GGC sequence and in 1 clone only by GC. However, because genomic DNA has not been sequenced at that locus, it is not possible to determine whether the first G represents a copy of the cap G or whether it is an authentic part of the DNA. Multiple initiation sites have also been found in human AMH gene (24), rather more distally from the main initiation site than in the chicken gene.
The 4200 bp of the ckAMH gene are organized in five exons (Fig. 1); the position of the introns, relative to the coding sequence, is conserved by comparison with mammalian genes (Fig. 2). The complete cDNA contains 2834 nucleotides. As all AMH cDNAs, ckAMH cDNA is rich in G and C. The overall GC content is 62.5%, rising to 68.5% in the part coding for the protein C-terminal domain. As shown in Fig. 1, ten punctual nucleotide variations were observed in cDNA, often in several independent clones. Three out of the nine variations located in the coding sequence lead to amino acid changes.
Predicted Protein Sequence-The open reading frame encodes a protein of 644 amino acid residues with a calculated  (Table II). Fig. 2 shows the alignment of ckAMH with various mammalian sequences. The four mammalian protein sequences are closer to each other than they are to ckAMH. Overall, chick and mammalian AMH proteins have 27% amino acid identity. The identity is 49% for the C terminus and 23% for the N terminus. The homology, taking into account similar residues, is 63% for the C terminus and 38% for the N terminus (Fig. 2). Despite this relatively low degree of amino acid conservation, 11 of the 13 cysteine residues of ckAMH are also present in mammalian AMHs.
A cleavage site for the signal sequence is predicted between Ala-20 and Leu-21 (25). A consensus sequence for monobasic cleavage (26), 106 amino acids upstream from the C terminus, is identical to the proteolytic site where full-length human AMH dimers are cleaved into N-and C-terminal domains (3). Chick AMH has four potential N-linked glycosylation sites. The glycosylation site common to the four mammalian AMHs and effectively glycosylated in human and bovine proteins (27), Asn-416, is also conserved in ckAMH. Another potential glycosylation site, Asn-537, just precedes the site of cleavage between N-and C-terminal domains and is not found in mammalian proteins.
Blot Hybridization-Southern blot analysis (Fig. 3) indicates the existence of a single ckAMH gene in the chicken genome. With human or turtle (Emys orbicularis) genomic DNAs, the ckAMH probe displayed no hybridization, even at low stringency, indicating a low degree of homology in the AMH genes between these species.
Northern blot analysis by hybridization of gonadal RNA with a probe corresponding to parts of exons 4 and 5 is shown on Fig.  4. Hybridization was repeated with a probe corresponding to exons 2 and 3 with identical results (not shown). No hybridization was observed with heart tissues. The main band corre-sponds to an mRNA species of about 2.8 kb. This size implies the existence of a very short poly(A) tail, since cDNA is already 2,834 bp long. Two more slowly migrating minor bands, at 4.5 and 6.5 kb, have intensities always correlated with that of the 2.8-kb band and may correspond to aggregates. The size of these bands is too great to be explained by differences in the length of poly(A) tails, as found for rat AMH mRNA (28).
Chick AMH mRNA expression in the testis peaks at 10 days of embryonic life and decreases thereafter; only a relatively small amount is still present in adult life. In female embryos, ckAMH mRNA is present in both gonads at much lower levels than in males. Both ovaries express the same amount of transcript between 8 and 10 days, but thereafter levels are higher in the left gonad. The maximum, reached on both sides at 17 days, is lower than in testicular tissue at the same age. In the adult hen, the left ovary still expresses AMH at a moderate level while the vestigial right ovary could not be studied.
In Situ Hybridization-Using a digoxigenin-labeled ckAMH riboprobe, chick AMH mRNA was detected by in situ hybridization in the testis at 8 and 17 days of embryonic life. AMH expression was restricted to the cytoplasm of perigerminal cells, the future Sertoli cells (Fig. 5). The mRNA was undetectable in the left ovary at 8 days but was present at 17 days in the compact zone located between the cortex and the medulla lacunary region, with some clusters of strongly positive cells (Fig.  5). The right ovary was negative at all times examined. DISCUSSION Antibodies raised against a partially purified preparation of chick AMH have allowed us to isolate the cDNA coding for chick AMH. Although, as shown on Figs. 1 and 2, the gross structure of the gene and the amino acid sequence of the protein bioactive domain are conserved between chick and mammals, the chick gene and protein are longer and diverge significantly from mammalian ones. Divergences affect essentially the untranslated regions, which show no homology, and the N terminus (Table II). The overall low nucleotide conservation explains the absence of cross-hybridization by Southern AMH genes and proteins Sequence data are from the references given in the legend of Fig. 2. The number of amino acid residues in the proteins and N-terminal domains and the calculated relative molecular weight (M r ) do not account for signal peptide processing. Experimental values for relative molecular weight were determined by electrophoresis of glycosylated proteins under reducing conditions (11,49). Amino acid conservation corresponds to the upper and lower values of the percentages of amino acids conserved between pairs of AMH protein sequences. The N-terminal domain is coded by exons 1-4 and the 5Ј part of exon 5; the C-terminal domain by the 3Ј part of exon 5. analysis and validates a posteriori the cloning strategy, favoring expression cloning over hybridization with mammalian probes. The evolutionary variation of AMH appears much greater than for TGF-␤ (29). This suggests that selective pressure for conservation of the sequence is much lower for AMH than for TGF-␤.
Sex determination mechanisms differ significantly between mammals and birds. In mammals, the testis-determining gene, SRY (30), present in the male heterogametic sex, induces testicular development. In its absence, XX individuals develop as FIG. 4. Expression of ckAMH mRNA in different tissues, studied by Northern blot hybridization. Top, hybridization of chick total RNA (20 g per sample) with a 32 P-labeled ckAMH cDNA probe (position 2121-3108). Film exposure at Ϫ70°C was 5 h for embryonic testes and 70 h for ovaries, heart, and adult testis. Bottom, stripped blot, rehybridized with a 32 P-labeled rabbit ribosomal oligonucleotide probe (same exposure for all the samples). females. In birds, the female is heterogametic, with males having two copies of a large Z chromosome and females having one Z and one smaller W sex chromosome (31), devoid of any detectable sex-specific SOX gene (32, 33). Hormonal manipulations such as left ovariectomy (34), testicular grafts (35), or administration of aromatase inhibitors (36) result in complete sex reversal and spermatogenesis in genetic females. These authors suggest that AMH, by inhibiting aromatase transcription in embryonic gonads, might play a role in physiological sex determination in birds, a hypothesis that will be tested as soon as recombinant chick AMH becomes available.
Study of AMH expression sheds some light on the molecular basis of Mü llerian duct development in the chick embryo. In males, bilateral regression of Mü llerian ducts, between 8 and 13 days, coincides with high expression of AMH by testes during that period. In females, the right Mü llerian duct and ovary regress, but the left Mü llerian duct develops normally, despite the fact that the adjacent ovary exhibits some anti-Mü llerian activity prior to hatching when measured using rat fetal Mü llerian ducts (8) or ovaries (9) as target organs. At 8 days, the time at which chick Mü llerian ducts are sensitive to AMH (10,37), Northern analysis (Fig. 4) confirms the expression of AMH by both ovaries, albeit at lower levels than in testes at similar ages. The surprising maintenance of the left chick Mü llerian duct in the face of early AMH exposure has been attributed to protection by estrogens produced in abundance by the chick embryonic ovary (38) and acting through nuclear estrogen receptors, which are present in higher amounts on the left side (39). Estrogen pretreatment of mice (40) or chick (41) embryos leads to Mü llerian duct insensitivity to AMH.
The putative avian AMH receptor differs physiologically from the mammalian one, since the latter responds to both avian and mammalian AMHs while chick embryonic reproductive organs respond only to the homospecific hormone (9,37,42). Mammalian AMH binds to a serine-threonine kinase membrane receptor, belonging to the type II TGF-␤ receptor family (20,43,44). Cloning of the chick AMH receptor and study of its interaction with recombinant chick AMH will be significant steps toward the understanding of hormone-mediated sex differentiation in the chick.