’rHt. .JOURNAL OF BIOLOGICAL CHEMISTRY Vol 256. No 8, I s u e of April 25, pp. 4W7-401fi. 1981 Prrnted m U S A

Human Prolactin cDNA STRUCTURAL ANALYSIS AND EVOLUTIONARY COMPARISONS*

(Received for publication, September 8, 1980, and in revised form, December 1, 1980)

Nancy E. Cookeg, Doris Coit, John Shines, John D. Baxterj, and Joseph A. Martial 1) From the Howard Hughes Medical Institute Laboratories, Departments of Medicine, Biochemistry, and Biophysics and the Metabolic Research Unit, University of California, Sun Francisco, California 94143

Prolactin (Prl), growth hormone, and chorionic som- matomammotropin form a set (the “Prl set”) of hor- mones which is thought to have evolved from a common ancestral gene. This assumption is based on several lines of evidence: overlap in their biological and im- munological properties, similarities in their amino acid sequences, and homologies in the nucleic acid se- quences of their structural genes. In the current study we report the cloning, amplification in bacteria, and sequence analysis of DNA compIementary to Prl mRNA isolated from human pituitary Prl-secreting adenomas. The cloned DNA contains 914 bases, which includes the entire coding sequence of human prehl as well as portions of the 5- and 3’-untranslated regions of the mRNA. The amino acid sequence predicted by our data differs from a previously reported amino acid sequence in 8 positions. With the results of this study we can now compare in one species the nucleotide sequences of the structural gene coding for each of the hormones of the Prl set. The sequence divergence at replacement sites is used to establish an evolutionary clock for the Prl set of genes. Using this clock, we postulate that the chromosomal segregation of human Prl and human growth hormone occurred about 392 million years ago and that growth hormone and chorionic sommatomam- motropin underwent an intrachromosomal recombi- nation within the last 10 million years.

Prolactin, growth hormone, and chorionic somatomammo- tropin (placental lactogen) appear to have evolved from a single ancestral gene by duplication and sequence divergence. The common ancestry of these proteins was indicated initially by similarities in their amino acid sequences (Catt et al., 1967; Li et al., 1969; Sherwood, 1967; Niall et al., 1971) and by overlap in their immunological and biological properties (Niall et al., 1973). Comparisons of the mRNA sequences of rat GH’ (Seeburg et aZ., 1977a) and rat Prl (Cooke et al., 1980; Gubbins et al., 1979), human GH (Martial et al., 1979; Roskam and Rougeon, 1979) and human CS (Shine et al., 1977), and bovine GH (Miller et al., 1980a) and bovine Prl (Miller et al., 1980b;

* 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,

$ T o whom correspondence should be addressed. 8 Present address, Department of Genetics, Research School of

tralia. Biological Sciences, Australian National University, Canberra, Aus-

11 An Investigator of the Howard Hughes Medical Institute. I/ Recipient of National Institutes of Health Grant GM 25549.

’ The abbreviations used are: Prl, prolactin, GH, growth hormone; CS, chorionic somatomammotropin, h, human; r, rat.

Nilson et aL, 1980) have given considerable additional support to the hypothesis of common ancestry.

In this paper we report the structure of the hPrl cDNA. This provides the first opportunity to compare the coding gene sequence for Prl, GH, and CS within one species. We have applied an improved method for the calculation of di- vergence (Perler et al., 1980) between coding sequences of the Prl family to obtain a more accurate description of evolution- ary relationships within this family. The sequence data permit us to report the complete signal peptide structure and to revise the published amino acid sequence of hPrl. The revised amino acid structure has been partially substantiated by the amino acid sequence of hPrl from normal pituitaries.2 This cloned hPrl cDNA should be useful as a probe to map and isolate the corresponding hPrl genomic sequences, to deter- mine the chromosomal location of the Prl gene (Owerbach et al., 1980b), to synthesize hPrl in bacteria,:’ and to study the regulation of Prl gene expression. The latter is of particular interest, since the mechanisms controlling Prl gene expression may differ from those for GH. Whereas prolactin is mainly under negative dopaminergic regulation (Takahara et al., 1974; MacLeod, 1976; Ben-Jonathan et al., 1977), GH may be predominantly under positive control (Krulich et al., 1968; Frohman et al., 1971).

EXPERIMENTAL PROCEDURES

mRNA Preparation-Human pituitary Prl-secreting adenomas were collected by transphenoidal hypophysectomy, frozen immedi- ately, and stored in liquid nitrogen. When six tumors had been collected, RNA was prepared from each according to the method of Chirgwin et ul. (1979) with modification as described previously (Cooke et al., 1980). The RNA purification included enrichment by chromatography on oligo(dT)-cellulose (P-L Biochemicals, Type 7).

mRNA Translation and Immunoprecipitation-The polyadenyl- ated RNA was translated in an in vitro rabbit reticulocyte lysate translation system in the presence of [““Slmethionine (Amersham, 600-1300 Ci/mmol) as previously described (Pelham and Jackson, 1967). Human prePrl synthesized in the cell-free system was immu- noprecipitated with rabbit antiserum to hPrl (a gift of H. Friesen, Department of Physiology, University of Manitoba). Control immu- noprecipitation using rabbit antiserum to hGH (Antibodies, Inc., Davis, CA.) or normal rabbit serum was also carried out. Immune complexes were precipitated with protein A of the Cowan strain of Staphylococcus aureus (Kessler, 1975; Martial et al., 1977). ”‘S-la- beled proteins were electrophoresed on sodium dodecyl sulfate-lO%’ polyacrylamide slab gels (Laemmli, 1970) at 20 mA/gel, fixed, washed, vacuum dried, and exposed to x-ray Fim (Kodak, NS-2T) by standard methods (Cooke et al., 1980).

cDNA Synthesis and Construction of Recombinant Plasmids- The mRNA from four of the tumors (about 19 pg) was pooled and used as template for synthesis of cDNA as described previously (Seeburg et al., 1977a). Avian myeloblastosis reverse transcriptase was received from Research Resources, National Cancer Institute,

M. John, H. Friesen, and H. Niall, personal communication. Manuscript in preparation.

4007

4008 Human Prolactin cDNA

Bethesda, MD, and S-l nuclease from Miles Laboratories. The dou- ble-stranded cDNA was tailed with dCMP and the Pst I-cleaved pBR322 was tailed with dCMP (Roychoudhury et al., 1976) using deoxynucleotidyltransferase (P-L Biochemicals). The annealing of the tailed cDNA and plasmid were as described (Cooke et aI., 1980). Restriction enzymes were obtained from New England Biolabs, Bev- erlv. MA. and Bethesda Research Laboratories. Bethesda. MD.

%czn.sformation and Colony Screening. An aliquot (20%) of the double-stranded cDNA was annealed to the Pst I-cleaved and dGMP- tailed pBR322. Transformation of Escherichia coli x1776 with the recombinant plasmids was carried out in compliance with NIH guide- lines. There were 232 colonies, of which 69 were ampicillin-resistant. All of the colonies were replica-plated onto Whatman 540 paper (Grunstein and Hogness, 1975) and screened with a ‘“P-labeled cDNA probe (Cooke et al., 19801, made using mRNA from tumor 5 as template. Sixteen prominently hybridizing colonies were selected, plasmids were isolated from them, cleaved with Pst I, and analyzed by agarose gel electrophoresis and Southern hybridization (Southern, 1975) using the same cDNA probe. The colony that contained the largest hybridizing insert and that was selected for nucleotide se- quence analysis was both tetracycline- and ampicillin-resistant. After positive identification of the clone by sequence analysis, the hPrl- containing plasmid was isolated and transformed into E. coli RRl.

DNA Sequence Determination-After restriction enzyme mapping of the hPrl cDNA clone, appropriate restriction fragments were selected and labeled at their 5’-ends with T-4 polynucleotide kinase (P-L Biochemicals) and [y-‘r’P]ATP (ICN, 4200-4500 Ci/mmol) after removal of the 5’-phosphate groups with bacterial alkaline phospha- tase (Enzo Biochemicals. Inc., New York, NY), or at their 3’-ends with the Klenow fragment of DNA polymerase I (New England Biolabs) and the appropriate [a-‘r’P]dNTP (Amersham, - 2500 Ci/ mmol). The chemical cleavage technique of Maxam and Gilbert (1977) was used with one change. The adenine modification reaction contained 25 al of formic acid (Mallinckrodt) and 10 al of “2P-labeled DNA restriction fragments. Reaction was for 10 min at 19 “C!. The reaction was stopped by addition of 200 al of 0.3 M sodium acetate, 0.1 mM EDTA, and 25 pg/ml of tRNA. DNA fragments were separated on polyacrylamide thin gels (8 and 15%) containing 7 M urea (Sanger et al., 1977).

RESULTS

Isolation and Identification of hPrl-enriched mRNA- RNA was isolated from six human pituitary P&secreting adenomas; 100 to 270 ag of RNA were obtained from each tumor. The total RNA from each tumor was passed through oligo(dT)-cellulose to enrich for polyadenylated species. This yield was not quantitated in order to minimize losses. The relative abundance of hPr1 mRNA from each tumor was analyzed by cell-free translation, polyacrylamide gel electro- phoresis, and autoradiography of the protein products. A prominent band at about 26,000 daltons was seen in each translation that contained tumor mRNA (Fig. 1A). Other prominent bands synthesized in the translation were endoge- nous to the reticulocyte lysate system. The band at 26,000 daltons was specifically immunoprecipitated by antibody to hPr1 (Fig. 1B). This band was not precipitated by antisera to hGH or by normal rabbit serum, indicating that the predom- inant protein produced by the polyadenylated RNA from these pituitary tumors was human prePr1.

Double-stranded cDNA Synthesis and Cloning-The mRNA from four of the six tumors (1 through 4, Fig. 1 A) was pooled and used as template for the synthesis of a single- stranded DNA. Polyacrylamide gel electrophoresis of the DNA indicated that a high molecular weight species had been produced (Fig. 2A). When the single-stranded cDNA was cleaved with Hae III as described previously (Seeburg et al., 1977b), three prominent bands were produced (Fig. 2A). This suggested that the synthesized single-stranded cDNA re- flected an enrichment for a unique mRNA species in the tumors, presumably hPr1.

The second strand of DNA was synthesized using reverse transcriptase. Following this, the resulting hairpin loop was digested by S-l nuclease. Fig. 2B shows the Hae III digestion

12 3456 7 1 2 3 4 A B

kd

- -- --- ---- -46-

- 14.3 -

FIG. 1. Autoradiographs of SsS-labeled cell-free translation products after electrohporesis on 10% sodium dodecyl sulfate polyacrylamide gels. A, translation products of polyadenylated RNA isolated from six human pituitary prolactinomas (Z-6) and the background with no added RNA (7) are shown. The approximate locations of %-labeled molecular weight markers, expressed in kilo- daltons (kd) are indicated by arrows. B, immunoprecipitates of the translation products using antiserum to human Prl (Z), antiserum to hGH (Z), normal rabbit serum (3), and total translation products with no added antibody (4) are shown. Experiment B was performed with 6 times less material (cpm) than Experiment A.

li , I

- I

FIG. 2. Autoradiographs of 32P-labeled single- and double- stranded cDNA to hPr1 mRNA after electrophoresis on 5% nondenaturing polyacrylamide gels. A, single-stranded cDNA intact (1) and cleaved with Hae III (2) are shown. The arrow indicates the location of the predominant single-stranded cDNA product. B, double-stranded cDNA was cleaved with Hae III without (I) and with S-l nuclease digestion (2). Molecular size marker (3) is ‘r’P- labeled pM2 digested with HindIII. From top to bottom the sizes of these marker bands are: 2.2, 1.0, 0.5, 0.45, 0.25, and 0.1 kilobases.

of this double-stranded DNA. Comparison of lanes 1 and 2 (before and after S-l nuclease digestion) confirms that most of the DNA was double-stranded. The single-stranded back- ground decreased with S-l digestion, but the major bands were resistant to S-l digestion and their sizes are consistent with the Hae III restriction fragments of the final cloned cDNA (Fig. 3B).

Human Prolactin cDNA 4009

A strand of dCMPs was added to the 3"ends of the cDNA, Plasmid DNA was isolated from each of these colonies, and it was then annealed to pBR322 which had been previ- cleaved with Pst I, and analyzed by gel electrophoresis. The ously digested with Pst I and tailed with dGMPs. When a largest insert, which contained 959 base pairs (including syn- portion of this recombinant plasmid was transformed into the thetic dGMP/dCMP tails), was selected for nucleotide se- E. coli strain ~ 1 7 7 6 , 232 tetracycline-resistant colonies were quence analysis. produced. The 16 colonies hybridized prominently with a Nucleotide Sequence of hPrl cDNA-The sequencing strat- cDNA probe made from the mRNA of tumor 5 (Fig. 1A). egy used is shown in Fig. 3A. Both message and anti-message

A Pst I Avo I I EcoR I I Toq I EcoR I PVU I I Xbo I Pst I Taq I Avo I I Pst I

I 5' /G2 A 8 1 -/c 18/3'

I

1 I I I I I I I 1 0 100 200 300 400 500 600 700 800 900

""

BASE PAIRS FIG. 3. Sequencing strategy and restriction map of hPrl untranslated regions of the corresponding Prl mRNA. B, the restric-

cDNA. A, restriction fragments were labeled at the site of the solid tion sites predicted by the DNA sequence were checked by restriction circles and DNA sequencing proceeded in the direction of the arrow- enzyme digestion followed by sizing on polyacrylamide gels. In some heads for the distance indicated by the length of the arrow. The 5'- instances restriction sites indicated in A were not copied onto B to end of the cDNA was sequenced in the 5' to 3' direction after labeling minimize crowding, although the digestions were performed. Both A at the Alu I site in pBR322 (indicated by dashed line). The dotted and B are drawn to scale as indicated by the calibration in base pairs bar indicates the coding region of the secreted hormone, the white at the bottom of the figure. bar indicates the signal peptide, and the solid lines indicate the

A G G A T 4 c

B G W T + C C GG+AT*CC

A G C A G C A G G A

G

FIG. 4. Autoradiographs of "P-la- beled DNA fragments generated by Maxam-Gilbert cleavage reactions. A, two representative 15% polyacryl- amide sequencing gels are shown in which the DNA was 3' labeled at the Aua I1 site that occurs between amino acids-19 and -18 of the signal peptide. The bottom band of the left sequence begins with the third nucleotide of amino acid-19 and proceeds in the 5'-direction on the coding strand, while the right sequence begins a t the second nucleotide of amino acid-18 and proceeds in the 3'- direction on the noncoding strand. B, a segment of an 8% polyacrylamide se- quencing gel generated from DNA la- beled at the Puu I1 site shows the fmt base of the codon for amino acid 79 and continues through amino acid 95.

40 10 Human Prolactin cDNA

strands were completely sequenced. Representative sequenc- ing gels are shown in Fig. 4. The presence of restriction endonuclease sites predicted from the nucleotide sequence was confirmed by restriction mapping with the indicated enzymes (Fig. 3B), and no missing or additional restriction sites were detected with the enzymes used. A characteristic restriction map of the cDNA insert is shown in Fig. 5.

The hPrl mRNA sequence and the derived amino acid sequence are shown in Fig. 6. If this assignment of amino acids is correct and if translation begins at the Met codon a t position

A 1 2 B 1 2 3 4 5 6 7 8

FIG. 5. Restriction digests of hPrl cDNA displayed on a 5% polyacrylamide gel stained with ethidium bromide. A, the plas- mid containing the hPrl cDNA cleaved with Pst I ( 1 ) is displayed next to the molecular size marker pBH322 cleaved with Hpa I1 (2). From top to bottom the marker fragments are: 622,527,403,309,242, 238,217.201,190,180,160,147,122, and 110 base pairs. B, the plasmid containing hPrl cDNA was cleaved with Pst I and the two hPrl cDNA fragments were isolated from a 5% polyacrylamide preparative gel. These two fragments were pooled and electrophoresed on a 5% polyacrylamide analytic gel without ( 1 ) and with further cleavage by Bg1 I1 (2), Bst E11 (3). Hha I (4, HindIII (5), Tac I, (6). or Bgl I (7). Hpa 11-cleaved pBR322 was used as size marker (8).

-28, then the primary translation product of hPrl mRNA is a protein of 25,880.03 daltons. The sequence of 227 amino acids includes the 28 amino acid signal peptide of hPrl and an additional amino acid, Ser, at position 86, which was not reported previously (Shome and Parlow, 1977). The sequence includes all of the 3"untranslated region, the AAUAAA seen in most eukaryotic mRNAs, and a part of the poly(dAMP) tail. Our DNA sequence, however, is lacking most of the 5'- untranslated region, possibly due to incomplete synthesis of the fist strand or to digestion of the cDNA corresponding to this region with S-1 nuclease following preparation of the second strand. By comparison, the 5"untranslated region in the analogous rPrl mRNA is a t least 51 base pairs in length (Cooke et al., 1980). The codon selection for hPrl is nonran- dom, as it is in rPrl. In contrast to rPrl, however, a 63% preference is shown for dGMP or dCMP in the third position of the codons.

The orientation of hPrl in the plasmid was determined by cleavage of the plasmid with HindIII. There is one HindIII site in the insert and one in the plasmid. The fragments predicted for an orientation of 3 to 5' (3801 and 1529 base pairs) were observed (data not shown). This places the hPrl cDNA within and in the same transcriptional orientation as the p-lactamase gene.

Comparisons of Prl, GH, a n d CS-The sequences of the related human polypeptide hormones Prl, GH, and CS were aligned to emphasize their homologies. As shown in Fig. 7, arbitrary gaps were introduced into the sequences to maximize their nucleotide and amino acid identity and also the number of conservative amino acid replacements (Dayhoff, 1978). Since CS and GH are highly homologous, only those positions at which CS differs from GH are indicated. No attempt was made to introduce gaps in the untranslated areas. Fig. 8 shows a similar alignment of hPrl and rPrl in which identical amino acids have been shaded to emphasize homologous areas. Since rat prePrl is only 225 amino acids in length, two amino acids shorter than human prePrl, a gap was introduced after amino

-2 8 -2 0 -10 met asn ile lys gly ser pro trp lys g ly ser leu leu leu leu leu Val ser am leu leu l eu cys gln ser Val ala pro

AAIY:AU;ARCWCAAAa;RUCG~U1;AAAmcrxl:ar,cvccvccvccvcax:~ARCCUGcvcCUGUGCCIY;Pa:C;UGCa3CCC

1 10 20 30

uuGCCCAu3ux1CCC(xx3cxx;axrca:a;A~cAG(u;prcalucGAcvccuGcGA(xx:cGcGa3Ga3CUGcuGczTuzy:Auccw leu pro ile cys pro gly gly ala ala arg cys gln V a l thr leu arg asp leu phe asp arg ala Val V a l leu ser his tyr ile h is

40 50 60 asn leu ser ser glu met phe ser glu phe asp lys arg tyr thr his gly arg gly phe ile thr lys a l a ile asn ser cys his t h r ARC(M:cuGucAGAAAuGu[I:Pa:GAAuu3GwAAAax;vwAcccw~ax;mcuu3AuvAa:ARGcGcAu3ARCIyx:ua3czTAN

70 80 90

u c u ~ a K I c G c ( u ; C C C G A A G p T A R G G A G c A A ~ c A A ~ ~ A w c A A A A A c G A u u u c u ; P a : c u ; A u A G a 3 P a : A u A u u G c a : I l c c ser ser leu ala thr pro glu asp lys glu gln ala gln gln met asn gln lys asp phe l eu ser leu ile V a l ser i le l eu arg ser

100 110 I20 t r p am glu pro leu tyr his l eu V a l thr glu V a l arg gly mt gln glu a la pro qlu a la i le leu ser lys a la Val glu ile glu u X ; A w G A G c c U ( M ; W c W ~ G a 3 ~ G A A ~ ~ o c t r A u G c A A G A A c G c a X ; G p r , ~ A u 3 ~ c u G A A A G c u G u A G I y ; A u u G p r ,

130 140 150 glu gln thr lys arg leu leu glu gly met glu leu ile V a l ser gln V a l h i s pro glu thr lys glu asn glu ile tyr pro Val t r p ~cAA(u;AAAax;cuJ~GAG(xx3Au;GAGcvcAuA~wccIy;(;uucWca)GAApa3AAAGAAAw~Aucuzy:(xxIGucuGG

160 170 180 ser gly leu pro ser leu gln met ala asp glu glu ser arg leu ser ala tyr tyr asn l eu leu his cys l e u arg arg asp ser his u I ; a ; R p r c ~ u I : ~ c I y ; A u G ~ G w G A A G A G u u c c x : a K I u c u ~ u w u w A R C ( M ; ( M : c A c a ; A ~ ( x x : p a : G A u u 3 A c w

190 199 l y s ile asp a m tyr leu lys leu leu lys cys arg ile ile h is asn asn asn cys 02 ~ A u 3 G p T A w W ~ ~ ( M : c v c ~ ~ c a : A u 3 ~ ~ A R C A R C ~ a ; A U ~ ~ ~

FIG. 6. Nucleotide sequence and predicted amino acid sequence of the mRNA coding for hPrl. The amino acid sequence is deduced from the mHNA sequence using the genetic code. OC, the ochre terminator.

Human Prolactin cDNA 401 1

% m a n Prolactin

Human Ciwrimic Scmatammtropin Human Growth 90-e

Ala Ile Thr Asp Ser 50 70 80

H i s Asp Ser

Cys H i s Thr Ser Ser Leu Ala Thr Pro Glu Asp Lys Glu G l n Ala G l n G l n ,Xet Asn G l n Lys Asp Fhe Leu Ser Leu Ile ~cET~~u1)cuuGccIy3ccccGAAGpc~GAGcAAaxcAAcpGALR;AAucAAAAAGpcuuucu;pGccuGAuA

muxcuc~uu3ucAGAGIx3uAuuccGFcA@xuocApcpGGGAGGAAmcAAcFGAAAuL?cAlylcuAGAGcuGcuccGcAucux Ser Leu Cys Phe Ser Glu Ser Ile Pro !l%r Pro Ser Asn Arg Glu Glu Thr G l n Gln Lys Ser Asn Leu G l u Leu Leu A r g Ile Ser

50 m m Gpc 60 Au; m 70 Aw Phe Cys Asp ,Yet %hr

90 100 Asn

110 V a l Ser I le Leu A r g Ser Trp Asn Glu mo Leu Tyr H i s Leu V a l Thr Glu Val Arg Gly Met Gln Glu Ala Pro Glu Ala Ile Leu Ser Gu3Iyx:AuAuuGcGAu1)uxAAuGRGccucu;~CAUcuGGu3pIJGGAAcwLcGucGumcAAGAAGccaK;~GcuAuccuA~ cuGcuGcucAuccFGmuGGcuGGFGccc mcPGmcucpGGAGuGxuucGccAAcpGccuGmuIy3GasGcc

Ar9 Met Asn Tyr Asp ”hr

~mGuAGpGAuuGAGGFGcAApccAAAax;~cuAGpGaxmGpGcucAuAGu3pa:cAGGuucwczlLIGAAFa:AAAGAAAAu u3uGAcpccAAc~uAuGpccuccuA~WcuAGAGGAAaxAuccAAFa;cu;Au;GGGpGGcuGGAAGw~pGc@xcccpcu

UCG Gww cpc 120 GAc ax: ASP A r Y

Leu Leu Leu Ile Gln Ser Trp Leu Glu Pro Val G l n Phe Leu Arg Ser Val Phe Ala Asn Ser Leu Val Tyr Gl;r Ala UAUGwm 80 GFG 30 ax m m

Glu 120 13 0 140

Lys Ala Val Glu Ile Glu Glu Gln Thr Lys Arg Leu Leu Glu Gly Met Glu Leu I le Val Ser G l n Val His Pro Glu Thr Lys Glu Asn

Ser Asp Ser Asn Val “yr Asp Leu Leu Lys Asp Leu Glu Glu Gly Ile Gln Thr Leu Met Gly A r g Leu Glu Asp Gly Ser Pro Arg “,hr

Ser Asp Asp H i s

Glu Ile Tyr Pro Val Trp Ser Gly Leu Pro Ser Leu Gln Met Ala Asp Glu Glu Ser A r g Leu Ser Ala ‘Iyr Tyr Asn Leu Leu H i s Cys 150 100 170

GAG~uIu3~Gu3uxucGQ;Acwa3AuI:cuGcpGmGcuGAuGAAGAG~cGccuuucuocuuwuwAlu=cuGcuccpc~ GcGcpGALR~ApG~mIw3AGcAAGuuc~mApcu3AcpcAIy3GAuWccAcuAcucAAGApcuIy3GGGcoGcucuIy3ucc Gly Gln Ile Phe Lys Gln 3-u Q r Ser Lys Phe Asp Thr Asn Ser H i s Asl Asp Asp Ala Leu Leu Lys Asn Tyr Gly Leu Leu Tyr Cys

cu3 140 Leu

uuu Phe

UCG cw cu; 160 Ser H i s Leu

139 Leu Arg A r g Asp Ser H i s Lys Ile Asp Asn Tyr Leu Lys Leu Leu Lys Cys Arg Ile Ile H i s Asn Asn Asn Cys

180 190

cuAcccpGGGwucAcAuAAAALlcWAAuuAucu:AFGcuccuGApGuGccGAAu3ALlccpcApcApc~ucc”- ~ ~ ~ ~ ~ A u ; ~ ~ ~ ~ ~ - ~ ~ ~ A u c m ~ ~ c G c ~ ~ ~ - ~ p G c ~ ~ ~ ~

cc UAA

Phe Arg Lys Asp Met Asp Lys Val Glu ’Ihr Phe Leu A r g Ile V a l Gln Cys A r g Ser Val Glu Gly Ser CIS Gly Pr,e Tw 170 m 180 190 191

[“let

FIG. 7. Comparison of the nucleotide sequences and pre- deduced from the mRNA sequences using the genetic code. The hCS dicted amino acid sequences of the mRNAs coding for human sequence is indicated only where it differs from the hGH sequence. Prl, GH, and CS. The mRNA sequence for hPrl is taken from Fig. The ochre terminator is indicated by OC, and the amber terminator 6, for hGH from Martial et al., (1979), and for hCS from Shine et al., by AM. The gaps were arbitrarily introduced to maximize homology. (1977) and Goodman et al., (1980). The amino acid sequences were

acid 8 of the rat sequence. Analysis of the homologies seen in us to predict the amino acid sequence of the human prepro- these alignments is presented under the “Discussion.” lactin signal peptide and to revise the known sequence of the

secreted protein. Comparison of the hPrl sequence to the other related human hormones, GH and CS, has permitted us

We have presented the nucleotide sequence of a 914 base to make some evolutionary speculations concerning the pro- pair-cloned DNA fragment containing the hPrl structural tein-coding areas of these three genes. gene sequence. The nucleotide sequence of this cDNA enabled RNA Isolation, cDNA Cloning, and Nucleotide Sequence

DISCUSSION

4012 Human Prolactin cDNA

QIZ rnm Wa" 13Lt A A A L I A A A R p L z l G A a x x u l p o l y (A)

FIG. 8. Comparison of the nucleotide sequences and predicted amino acid sequences of the mRNAs coding for human and rat Prl. The mHNA sequences are taken from Fig. 6 and Cooke et al. (1980). Shaded rectangles indicate identical amino acids. A single gap was introduced in the rat sequence to maximize afignment.

Analysis-Polyadenylated RNA was isolated from hPrl-se- creting pituitary adenomas. The cell-free translation products of this mRNA were highly enriched in prePrl. This enrichment of hPrl mRNA in pituitary tumors was remarkable in that it appeared to exceed that which we were able to produce previously in rats by estrogen implantation and hypothalamic ablation (Cooke et al., 1980). This assumes, of course, that

both Prl mRNAs were translated with equal efficiency in the cell-free system. In the translation of mRNA from human pituitary tumors, no hGH was seen either in the total trans- lation or by immunoprecipitation using hGH antiserum. Al- though it is possible that the presence of some preGH is obscured by the width of the prePrl signal or by the limited exposure of the gel to x-ray film, these data may indicate that

Human Prolactin cDNA 401 3

the tumor is not synthesizing GH. In earlier work the mRNA from a Prl-secreting tumor was translated and did contain a small amount of detectable GH mRNA, although it was not determined whether this GH mRNA originated in the tumor tissue or from adjacent normal pituitary tissue (Martial et al., 1979).

The mRNA from these human tumors, naturally enriched in Prl mRNA, yielded a cDNA highly enriched in Prl se- quences. Initial attempts to screen the bacterial colonies con- taining recombinant plasmids with a probe made from cloned rPrl cDNA (Cooke et al., 1980) were unsuccessful. Even under conditions of low stringency of hybridization, rPrl cDNA did not cross-hybridize with hPrl sequences. In retrospect, this result was unexpected, since the nucleotide sequence homol- ogy of rat and human Prl is 73%.

The accuracy of the sequence was documented by analyzing both strands of the hPrl cDNA completely and by verifying the presence of most of the predicted restriction sites. Errors in reverse transcriptase synthesis of the first DNA strand (Gopinathan et al., 1979; Fagan et al., 1980) cannot be ex- cluded.

The clone containing hPrl cDNA that was ultimately stud- ied was both tetracycline- and ampicillin-resistant (Villa-KO- maroff e ta l . , 1978). Resistance to both antibiotics was retained after retransformation of the hPrl-containing plasmid into the RR1 strain of E . coli. Since only one plasmid species was visible on gel electrophoresis, these results suggest that the ampicillin resistance was not due to co-transformation with a wild type plasmid carrying an intact p-lactamase gene. Re- striction endonuclease mapping showed that the cDNA was inserted in the transcriptional direction of p-lactamase. At- tempts to sequence from the Alu I site just 5’ to the Pst I site of the p-lactamase gene into the inserted cDNA suggested that the hPrl cDNA was not in the p-lactamase reading frame. However, determination of the exact length of the long 5’- dGMP tail was difficult due to base stacking and compressions on the gels. Immunoprecipitation with hPrl antiserum of a - 45,000-dalton protein from the extracts of E. coli infected with the hPrl recombinant plasmid suggests that a hybrid of p- lactamase and hPrl is produced by this clone.” Thus, the retention of ampicillin resistance by the hPrl recombinant plasmid might be explained if the fusion protein retained p- lactamase activity.

The signal peptide sequence of hPrl has not been previously reported. Our sequencing data (Fig. 4A) has shown that it is 28 amino acids in length. I t is identical with the signal peptide of rPrl in 16 positions (Fig. 8), including a hydrophobic block of 4 Leu residues. The secreted protein contains 199 amino acids, one more than previously reported by amino acid se- quencing techniques (Shome and Parlow, 1977). In addition, these sequences disagree at eight positions. At positions 82, 83, 85, 86, 105, 144, 162, and 163, we report Ser, Leu, Val, Ser (the additional amino acid) (Fig. 4B), Met, Asn, Glu, and Ser, while Shome and Parlow (1977) report Val, Ser, Leu, no amino acid, Asx, Asp, Ser, and Glu, respectively. Partial revision of the amino acid sequence of hPrl is in complete agreement with our results at positions 105, 144, 162, and 163; however the chymotryptic peptide at 81-89 was not isolated or rese- quenced, so the discrepancies at positions 82-86 were not resolved.’

The reported nucleotide sequence represents a single hPr1 molecule produced by a Prl-secreting pituitary adenoma, whereas the amino acid sequence represents the predominant Prl sequence in pools of normal human pituitaries. Conse- quently, the degree of identity between these sequences is striking and might suggest that both adenomatous and normal pituitaries can produce identical or nearly identical Prl mole-

cules. We cannot determine, however, if the molecule that we sequenced was produced by the adenoma or by the surround- ing normal pituitary tissue, although adenomatous tissue pre- dominated in the samples used.

The codons used in hPrl mRNA are nonrandom (data not shown) as found in all eukaryotic mRNAs sequenced to date. The codon choices of hGH (Martial et al., 1979) and hCS (Shine et al., 1977) mRNAs are more similar to each other than either of them is to hPrl mRNA. It seems unlikely that differences in codon choice reflect evolutionary pressures to adapt to different tRNA populations. First, almost all of the codons are used in all of the genes of the Prl set. Second, GH and Prl genes are expressed in the same tissue in vivo and can be expressed by the same cell (the rat GHs cell line; Dannies and Tashjian, 1973). The differences in codon choice between GH and Prl contrast with the similarities in the codon choices of GH and CS, which are expressed by different tissues. An analogous situation exists in the globin gene family. All of the globins have a nonrandom and unique distribution of codon choices, yet are expressed either sequentially (y,/3) or simul- taneously (a&) in the same tissue. Consequently, no relation- ship between expression and tRNA availability has been shown in this family (Wilson et al., 1980). In the Prl set of genes there is a spectrum of preference for guanosine plus cytosine in the third position of the codons: bovine GH, 82% (Miller et al., 1980a); hCS, 80%; hGH, 76%; rGH, 74%; hPrl, 63%; and rPrl, 50% (Cooke et al., 1980). Codons that end with dUMP or dCMP and codons that end with dAMP or dGMP code for the same amino acid (Met and Trp are exceptions). Consequently, a bias in the use of dGMP and dCMP in the third position of codons cannot be explained on the basis of evolutionary constraints against amino acid subsiitution. This suggests that synonymous codons might be expected to be neutral in terms of evolution, and as discussed, this appears to be the case.

Molecular Evolution of the Prolactin Set of mRNAs-A direct comparison of the sequences of the Prl family of genes (Table I), aligned as shown in Figs. 7 and 8, shows a greater nucleotide than amino acid sequence homology in each case. In addition, very similar amino acid and nucleotide homolo- gies were obtained in the interspecies comparisons of Prl and GH (z.e. hPrl with rPrl, and hGH with rGH). The same parallelism in homologies was observed in the intraspecies comparisons of Prl and GH (i.e. rPrl with rGH, and hPrl with hGH). There is a very high degree of homology between hGH and hCS, as has been noted previously (Martial et al., 1979).

There can be two types of nucleotide differences between genes when they are compared codon by codon. The first is a difference that results in an amino acid replacement (replace- ment substitutions); the second results in a synonymous codon (silent substitutions). The percentage of silent substitutions in each comparison in Table I (two identical nucleotides/codon) is greater than the 25% that would be expected when compar- ing random sequences (Jukes and King, 1979). This may reflect the evolution of these genes from a common ancestor.

Surprisingly, the hGH and hCS cDNAs, which have the highest nucleotide sequence homology of the set (92%), also have the lowest ratio of silent to expressed single base substi- tutions, 0.45 (31%/69%, Table I). In contrast, the hPrl and rPrl genes with only 73% nucleotide homology have a silent to expressed ratio of 1.2 (54%/46%, Table I). These results are consistent with the hypothesis that positive selective influ- ences have caused rapid fixation of replacement substitutions in the genes of these recently diverged hormones (hGH and hCS).

To improve the estimates of divergence among the members of the Prl family of genes, the methods of Perler et al. (1980)

4014 Human Prolactin cDNA

TABLE I Amino acid and nucleotide sequence comparison within the Prl family

The results are derived from the sequence alignments in Figs. 7 and 8. Codons corresponding to gaps were excluded from the calculations, as were the termination codons. Onlv coding seauences have been comDared.

Amino acids Nucleotides ~~

Identical Conservative" differ- bases Comparison ences Identical codons Codons differing by one base

%J No. % No. o/n No. % No. Silent Expressed

%J No. %' No.

Conservative amino acid substitutions were calculated according to Dayhoff (1978) and include only the most common two categories of substitutions.

Cooke et al. (1980). 'Martial et al. (1979).

Seeburg et al. (1977a). 'Shine et al. (1977). 'Goodman et at. (1980).

and Efstratiadis et al. (1980) were used. This technique as- sumes that nucleotide substitutions occur in a Poisson distri- bution and that transitions (pyrimidine to pyrimidine or pu- rine to purine substitutions) and transversions (purine to pyrimide or pyrimidine to purine) are equally probable. It attempts to determine the mutation rate more accurately by considering all of the possible ways that codon differences could have arisen and assigns a probability to each of these. The possibility of multiple events a t a single site enables the divergence to be greater than 100%. The results of this correc- tion on both silent and replacement sites are displayed in Table 11.

To establish an evolutionary clock for the Prl family of genes, the corrected divergence at replacement sites for each pair of genes has been plotted against divergence time esti- mated from the fossil record (Fig. 9). Silent site divergences were similarly calculated and plotted. The divergence between human and rat hormones was assumed to occur during the mammalian radiation about 85 to 100 million years ago (Rom- ero-Herrera, et al., 1973; McKenna, 1969). The divergence between Prl- and GH-like genes (including CS) was assumed to occur about 400 million years ago, when fish and tetrapods diverged, because the existence of distinct Prl-like and GH- like hormones has been dated to this time (Acher, 1976). The straight line ( R ) drawn through the origin and the replacement sites serves as the molecular clock for the Prl gene family. Its slope gives the rate of fixation of replacement substitution in this family. The reciprocal of this slope corresponds to a unit evolutionary period (UEP), i.e. the time in millions of years for a 1% sequence divergence to arise between homologous sequences (Dickerson, 1971; Wilson et al., 1977) of 4.5 for the Prl family. Substantiation of this UEP will require additional points on the curve such as might be obtained from the examination of Prl-like genes from birds, reptiles, or fish.

Use of this UEP would predict that Prl and GH diverged about 392 million years ago, and Prl and CS about 399 million years ago; this results in a minimal revision of our previous estimate of 380 million years ago (Cooke et al., 1980), which was based on a Prl UEP derived from amino acid comparisons (Wilson et al., 1977).

Applying the evolutionary clock and UEP for the Prl family to the human GH and CS comparison, we arrive at a diver- gence time of 10 million years ago. This is very different from previous estimates of 56 million years ago (Martial et al.,

TABLE I1 Percentage of corrected divergence among mRNAs of the prolactin

family The percentage of corrected divergence for each sequence pair was

calculated as described under "Discusion." Codons corresponding to gaps were excluded from the calculations.

Comparison Replacement Silent sites sites

Human Prl/rat Prl 20.8 76.6 Human GH/rat GH 19.4" 71.0" Human Prl/human GH 87.0 194.0 Human Prl/human CS 88.7 240.0 Human GH/human CS 2.3 11.5

" Results are in agreement with data from Perler et al., 1980.

1979), and would date the divergence long after the mamma- lian radiation began, and long after the first presumed require- ment for specific placental hormones. This paradox might be explained as follows. A mechanism for correcting one gene against another during evolution has been postulated based on a structural analysis of the human fetal y-globin genes (Slightom et al., 1980). A homologous but unequal cross-over may occur between two such mutually correcting genes ("con- certed evolution," Zimmer et nl., 1980), and a calculated divergence time will date this correction event instead of the original gene duplication. Since the human GH and CS genes reside on chromosome 17 (Owerbach et ab, 1980a), and since a human GH gene has been shown to be linked to a CS gene on a single DNA fragment isolated from a human DNA library (Goodman et al., 1980), it might be postulated that the 10 million year divergence time dates an exchange of DNA between the GH and CS loci on sister chromatids during meiosis rather than the original gene duplication. Dating of the original gene duplication by comparison of these two sequences would be more difficult and perhaps impossible if this is true.

Since hPrl and hGH are located on separate chromosomes (6 and 17, respectively), they are less likely to exchange DNA in this manner. Therefore, the 392 million year divergence time between these two genes may more accurately reflect their initial duplication than does the 10 million year diver- gence between hGH and hCS.

The rate of silent site divergence in the Prl family may be described by a biphasic curve (Fig. 9) as has been reported for

Human Prolactin cDNA 4015

- a? - 200-

Yl z a t

0 I00 200 300 400 T I M E (MY1

FIG. 9. Curves for divergence of coding sequences among the Prl set of genes. The corrected percentage of divergences at silent (S) and replacement (R) sites for comparisons between hPrl and rPrl (O), hPrl and hGH (m), hGH and rGH (O), and hGH and hCS (0) are taken from Table 11. These are plotted against divergence times in millions of years (MY) which are minima1 estimates derived from the fossil record (see text for discussion and references). The solid lines (S and R ) were drawn through the points. Line R was passed through the origin. The reciprocal slope of R gives an estimate of the unit evolutionary period (UEP, the time in millions of years for a 1% sequence divergence to arise; Dickerson, 1971; Wilson et al., 1977) for replacement substitutions within the PrI set of genes. Line S does not extrapolate through the origin. In analogy with the preproinsulin (Perler et al., 1980) and globin (Efstratiadis et al., 1980) gene families, we presume that the silent sites fixed mutations with a higher rate (dashed line) for approximately 85 to 100 millions of years. The arrows indicate the derived time of last correction between hGH and hCS (see text).

preproinsulin (Perler et al., 1980) and globin (Efstratiadis et al., 1980) genes. The actual break point in the Prl curve cannot be determined without more data points. However, it is clear that the silent sites diverge rapidly a t fist. They may then continue to diverge a t a slower rate. The nature of the subset of rapidly diverging sites is not known. It should be borne in mind that the accuracy of both replacement and silent site curves is no better than the fossil record on which they are based.

There are several problems with this type of evolutionary analysis. First, the assumption that transitions and transver- sions are equally probable is not necessarily valid. For exam- ple, in certain human globin variants, adenine-guanine tran- sitions are observed more frequently than would be expected on a random basis (Fitch, 1971). Second, the divergence of two molecules does not necessarily coincide with and may antedate the divergence of species. Consequently, the diver- gence dates plotted are crude approximations. Third, rapid and slow periods have been detected in the evolution of the globins (Dickerson, 1971) and when rapid evolution is taking place, a protein may be in the process of assuming new functions. Because of such fluctuations, a linear evolutionary clock may not be accurate. The high rate of replacement substitutions compared to silent substitutions in the hGH and hCS divergence may be an example of such an acceleration of the clock. Another example may be seen in the rabbit ,&globin alleles (sequenced by Efstratiadis et al., 1977; Hardison et al., 1979; van Ooyen et al., 1979; and discussed by Perler et al., 1980) in which there are four nucleotide differences, each resulting in an amino acid change with no accompanying

silent replacements. These examples suggest that positive selection can rapidly fix replacement mutations without being accompanied by random silent or neutral mutations. Finally, in an unequal cross-over between two correcting genes, dating based on sequence homology will also reflect the extent of the correction. The derived divergence data may therefore be a composite of the dates of divergence and of last correction of the various portions of the genes.

The 3'-untranslated area of hPrl, when compared without gaps to the 3"untranslated area of rPrl, is identical in 43 of the 94 bases compared (54.3% divergence). The area contain- ing the sequence AAUAAA found in most eukaryotic mRNAs and its surrounding sequences is not available in the rPrl sequence for comparison. Homology in the 3"end of the adult /&globin family of genes (Efstratiadis et al., 1980) has been highest near this signal, so our 3'-end divergence may be an underestimate. One might speculate that some constraint on drift within this area exists for the Prl molecule. However, in general, no such constraints have been found (Efstratiadis et al., 1980). Comparing hGH and rGH directly without gaps, a 61.8% divergence is found. It is clear by inspection that the homology can be increased by creating appropriate gaps. Human Prl and GH have diverged 87.556, hPrl and hCS 87.2%, while hGH and hCS only 6.4% in the 3'-untranslated area. It is impossible to compare these results to the silent site replace- ments within the coding portions of the molecule, since the correction for multiple events within each site cannot be performed. However, compared to the nucleotide identity in Table I, it would appear that the 3"untranslated areas in each case (except in the comparison of hGH and hCS) have di- verged more than the coding areas. This is consistent with results found in comparisons of other eukaryotic genes (Heilig et al., 1980; van den Berg et al., 1978; Bell et al., 1980).

REFERENCES Acher, R. (1976) Ciba Found. Symp. 41,31-59 Bell, G. I., Pictet, R. L., Rutter, W. J., Cordell, B., Tischer, E., and

Goodman, H. M. (1980) Nature 284,26-32 Ben-Jonathan, N., Oliver, C., Weiner, H. J., Mical, R. S., and Porter,

J. C. (1977) Endocrinology 100,452-458 Catt, K. J., Moffat, B., and Niall, H. D. (1967) Science 157, 321 Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J., and Rutter, W. J.

(1979) Biochemistry 18, 5294-5299 Cooke, N. E., Coit, D., Weiner, R. I., Baxter, J. D., and Martial, J. A.

(1980) J. Biol. Chem. 255,6502-6510 Dannies, P. S., and Tashjian, A. J., Jr. (1973) in Tissue Culture:

Methods and Applications (Kruse, P. F., and Patterson, M. K., Jr., eds) pp. 561-569, Academic Press, New York

Dayhoff, M. 0. (1978) Atlas of Protein Sequence, Vol. 5, Suppl. 3, National Biomedical Research Foundation, Washington, D. C.

Dickerson, R. E. (1971) J. Mol. Evolution 1, 26-45 Efstratiadis, A., Kafatos, F. C., and Maniatis, T. (1977) Cell 10, 571-

585 Efstratiadis, A., Posakony, J. W., Maniatis, T., Lawn, R. M.,

O'Connell, C., Spritz, R. A., DeRiel, J. K., Forget, B., Weissman, S. M., Slightom, J. L., Blechl, A. E., Smithies, O., Baralle, F. E., Shoulders, C. C., and Proudfoot, N. J. (1980) Cell 21,653-668

Fagan, J. B., Pastan, I., and de Crombrugghe, B. (1980) Nucleic Acids Res. 8, 3055-3064

Fitch, W. M. (1971) Syst. 2001. 20, 46-50 Frohman, L. A., Maran, J. W., and Dhariwal, A. P. S. (1971) Endo-

crinology 88, 1483-1488 Gopinathan, K. P., Weymouth, L. A,, Kunkel, T. A,, and Loeb, L. A.

(1979) Nature 278, 857-859 Goodman, H. M., DeNoto, F., Fiddes, J. C., Hallewell, R. A,, Page, G.

S. Smith, S., and Tischer, E. (1980) in Mobilization and Reassem- bly of Genetic Information (Scott, W. A,, Worner, R., Joseph, D. R., and Schultz, J., eds) pp. 155-179, Academic Press, New York

Grunstein, M., and Hogness, D. S. (1975) Proc. Natl. Acad. Sci. U. S. A. 72, 3961-3965

Gubbins, E. J., Maurer, H. A., Hartley, J. L., and Donelson, J. E. (1979) Nucleic Acids Res. 6, 915-930

4016 Human Prolactin cDNA

Hardison, R. C., Butler, E. T., 111, Lacy, E., Maniatis, T., Rosenthal,

Heilig, R., Perrin, F., Gannon, F., Mandel, J. L., and Chambon, P.

Jukes, T. H., and King, J. L. (1979) Nature 281, 605-606 Kessler, S. W. (1975) J. Zmmunol. 115, 1617-1624 Krulich, L., Dhariwal, A. P. S., and McCann, S. M. (1968) Endocri-

Laemmli, U. K. (1970) Nature 227,680-685 Li, C. H., Dixon, J . S., Lo, T. B., Pankov, Y. M., and Schmidt, K. D.

(1969) Nature 224,695 MacLeod, R. M. (1976) in Frontiers in Neuroendocrinology (Martini,

L., and Ganong, W. F., eds) Vol. IV, pp. 169-194, Raven Press, Pubs., New York

Martial, J . A., Baxter, J . D., Goodman, H. M., and Seeburg, P. H. (1977) Proc. Natl. Acad. Sci. U. S. A . 74, 1816-1820

Martial, J. A., Hallewell, R. A., Baxter, J., and Goodman, H. (1979) Science 205,602-607

Maxam, A. M., and Gilbert, W. (1977) Proc. Natl. Acad. Sci. U. S. A .

McKenna, M. C. (1969) Ann. N . Y. Acad. Sci. 167,217-240 Miller, W. L., Martial, J. A., and Baxter, J . D. (1980a) J. Biol. Chem.

Miller, W. L., Thirion, J. P., and Martial, J. A. (1980b) Endocrinology,

Niall, H. D., Hogan, M. L., Sayer, R., Rosenblum, I. Y., and Green- wood, F. C. (1971) Proc. Natl. Acad. Sei. U. S. A . 68, 866-869

Niall, H. D., Hogan, M. L., Tregar, G. W., Segre, G. V., Hwang, P., and Friesen, H. (1973) Recent Prog. Horm. Res. 29, 387-416

Nilson, J . H., Thomason, A. R., Horowitz, S., Sasavage, N. L., Blenis, J., Albers, R., Saker, W., and Rottman, F. M. (1980) Nucleic Acids Res. 8, 1561-1573

Owerbach, D., Rutter, W. J., Martial, J. A., Baxter, J . D., and Shows, T . B. (1980a) Science 209,289-292

Owerbach, D., Rutter, W. J., Cooke, N. E., Martial, J. A., and Shows, T. B. Science (1980b), in press

Pelham, H. R. B., and Jackson, R. J . (1976) Eur. J. Biochem. 67,

N., and Efstratiadis, A. (1979) Cell 18, 1285-1290

(1980) Cell 20,625-627

nology 83, 783-790

74,560-564

255, 7521-7524

107,851-854

247-256

Perler, F., Efstratiadis, A., Lomedico, P., Gilbert, W., Kolodner, R.,

Romero-Herrera, A. E., Lehman, H., Joysey, K. A,, and Friday, A. E.

Roskam, W. G., and Rougeon, F. (1979) Nucleic Acids Res. 7, 305-

Roychoudhury, R., Jay, E., and Wu, R. (1976) Nucleic Acids Res. 3,

Sanger, F., Nickien, S., and Coulson, A. R. (1977) Proc. Natl. Acad.

Seeburg, P. H., Shine, J., Martial, J. A., Baxter, J. D., and Goodman,

Seeburg, P. H., Shine, J., Martial, J. A., UlLrich, A., Baxter, J. D., and

Sherwood, L. M. (1967) Proc. Natl. Acad. Sci. U. S. A . 58,2307-2314 Shine, J., Seeburg, P. H., Martial, J. A., Baxter, J. D., and Goodman,

Shome, B., and Parlow, A. F. (1977) J. Clin. Endocrinol. Metab. 45,

Slightom, J. L., Blechl, A. E., and Smithies, 0. (1980) Cell 21, 627-

Southern, E. M. (1975) J. Mol. Biol. 98,503-517 Takahara, J., Arimura, A., and Schally, A. V. (1974) Endocrinology

95,462-465 van den Berg, J., van Ooyen, A., Mantei, N., Schambock, A,, Grosveld,

G., Flavell, R. A., and Weissman, C. (1978) Nature 276, 37-44 van Ooyen, A,, van den Berg, J., Mantei, N., and Weissman, C. (1979)

Science 206,337-338 Villa-Komaroff, L., Efstratiadis, A., Broome, S., Lomedico, P., Tizard,

R., Naber, S. P., Chick, W. L., and Gilbert, W. (1978) Proc. Natl. Acad. Sci. U. S. A. 75, 3727-3731

Wilson, A. C., Carlson, S. S., and White, T. J. (1977) Annu. Rev. Biochem. 46,573-639

Wilson, J. T., Wilson, L. B., Reddy, V. B., Cavallesco, C., Ghosh, P. K., deRiel, J. K., Forget, B. G., and Weissman, S. M. (1980) J. Biol. Chem. 255,2807-2815

Zimmer, E. A., Martin, S. L., Beverly, S. M., Kan, Y. W., and Wilson, A. C. (1980) Proc. Natl. Acad. Sci. U. 5'. A. 77,2158-2162

and Dodgson, J. (1980) Cell 20,555-566

(1973) Nature 246,389-395

320

863-877

Sci. U. S. A . 74,5463-5467

H. M. (1977a) Nature 270,486-494

Goodman, H. M. (1977b) Cell 12, 157-165

H. M. (1977) Nature 270,494-499

1112-1115

638