DNAVolume 6, Number 4, 1987Mary Ann Lieber!, Inc., PublishersPp. 325-330

Isolation of Rat Testis cDNAs Encoding anInsulin-Like Growth Factor I Precursor

SAMUEL J. CASELLA,* ERIC P. SMITH,*4 JUDSON J. VAN WYK,* DAVID R. JOSEPH,ÎMARY A. HYNES.t E. COOKIE HOYT,t and P. KAY LUNDt

ABSTRACT

We have characterized rat testis cDNAs encoding insulin-like growth factor I (IGF-I) precursor to facilitatestudies of IGF-I expression in the male reproductive system. Two clones, P2 and P3, with inserts of 786 and1200 bp, respectively, were isolated from a Xgtll library of rat testis cDNAs. The longest open reading frameof cDNA P2 predicts a 153-amino-acid residue IGF-I precursor that has only 11 amino acid substitutionscompared with a human IGF-IA precursor encoded by a human liver mRNA. Three substitutions are withinthe predicted rat IGF-I sequence: a Pro for Asp in the B domain, an He for Ser in the C domain, and Thr forAla in the D domain. Only two substitutions distinguish the predicted rat sequence from a mouse liver IGF-IA precursor: Thr for Ala in the signal peptide and Ala for Ser in the D domain. P2 hybridizes with poly(A)+mRNAs of 7.5, 4.7, 1.7, and 1.2-0.9 kb in rat liver and testis. The other testis cDNA, P3, appears to repre-sent a partially processed rat IGF-I mRNA precursor. By comparing the sequence of cDNA P2 with that ofcDNA P3 and a 2.3-kb rat IGF-I genomic fragment, we predict exon splice sites within the codon for residue26 and between residues 86-87 of the rat IGF-I precursor. Both of the predicted splice sites align with exon-

intron junctions in the human IGF-I gene. We conclude, therefore, that IGF-I is synthesized as a precursor inthe rat testis and that the structure of IGF-I genes, mRNAs, and precursors are highly conserved across

species.

INTRODUCTION

HUMAN INSULIN-LIKE GROWTH FACTOR I (IGF-I), alsoknown as somatomedin-C, is a 70-amino-acid peptide

that has both mitogenic and differentiating actions (re-viewed by Van Wyk, 1984). There is recent evidence thatthis peptide growth factor has an important role in the re-

productive system (Ritzen, 1983; Bernier et al., 1986; Treset al., 1986). Because the rat is used extensively as a modelof IGF-I action, we have characterized the rat homolog ofIGF-I and the expression of IGF-I mRNAs in the rat testis.

Rubin et al. (1982) purified small quantities of rat IGF-Ifrom pooled sera of Wistar Furth rats that had been inocu-lated with growth hormone-secreting tumor cells. Quantita-tive amino acid analysis suggested that there were signifi-cant differences between human and rat IGF-I. Rat IGF-I

was apparently more basic than human IGF-I, even thoughpartial sequence analysis of the first 29 amino acids re-

vealed complete homology with the human peptide. Smithet al. (1987) characterized the IGF activity from the me-

dium of cultured Sertoli cells isolated from rat testes. Thepartially purified rat preparation paralleled human IGF-Iin a variety of assays, but, as with the preparation ofRubin, had a more basic isoelectric point.

We have utilized RNA blot hybridizations to demon-strate that IGF-I mRNAs are synthesized in the rat testisand have characterized cDNAs encoding IGF-I from a rattestis cDNA library to determine the structure of the rattestis IGF-I precursor. Comparison of the rat testis cDNAsequences and rat genomic DNA sequences also has pro-vided information about exon splice junctions within therat IGF-I gene.

Department of 'Pediatrics, tphysiology, and the t Laboratories for Reproductive Biology, School of Medicine, University of NorthCarolina at Chapel Hill, Chapel Hill, NC 27514.

325

326 CASFLLAETAL.

METHODSIsolation of IGF-I cDNAs from a rat testiscDNA library

Complementary DNA was prepared from the testes of28-day Sprague Dawley rats and cloned into the bacterio-phage Xgt 11 as previously described (Joseph et al., 1985). Ahuman liver IGF-I cDNA (Jansen et al., 1983; designatedthe IGF-IA variant by Rotwein, 1986a) was kindly pro-vided by Dr. Martin Jansen and Dr. Leo Van den Brande(State University of Utrecht, The Netherlands). The "P-la-beled human IGF-I cDNA was used to screen 200,000 re-combinants by the filter hybridization method (Benton andDavis, 1977). Two positive clones, P2 and P3, with insertsof 786 bp and 1200 bp, respectively, were characterized bynucleotide sequence analysis and RNA blot hybridizationswith rat poly(A)* RNA.

series of M13mpl9 subclones containing overlapping frag-ments of the original cDNA. Subclones were amplified inE. coli JM101 and sequenced by the dideoxy terminationmethod (Sanger et al., 1977).

RESULTS

Nucleotide sequence of cDNA P2 encodinga rat IGF-I precursor

The sequence of cDNA P2 (Fig. 1) predicts a 5'-untrans-lated region of 181 bases preceding a putative initiatormethionine codon that is at the beginning of an open read-

TCTGCTTGCTAAATCTCACTGTCGCTGCTAAATTCAGAGCAGATAGAGCCTGCGCAATCG

RNA blot hybridizations of poly(A)* RNA

Poly(A)+ RNA was isolated from rat tissues and analyzedby RNA blot hybridization with human and rat IGF-IcDNA probes. Methods for poly(A)+ RNA isolation, hy-bridization, and probe labeling were described previously(Lund et al., 1986).

Isolation of a 2.3-kb rat IGF-I genomicDNA fragment

A rat genomic DNA library, prepared by cloning a par-tial Eco RI digest of rat DNA into the bacteriophage X vec-tor Charon 4A, was kindly provided by T. Sargent, R.Wallace, and J. Bonner (California Institute of Technol-ogy). A 32P-labeled human IGF-I cDNA probe was used toscreen 106 recombinants by the filter hybridization method(Benton and Davis, 1977). Four positive recombinants eachcontained a 2.3-kb fragment that hybridized with the hu-man IGF-I cDNA (data not shown) and was characterizedfurther by nucleotide sequence analyses.

AAATAAAGTCCTCAAAATTGAAATGTGACTTTGCTCTAACATCTCCCATCTCTCTGGATT

TCLTTTTGCCTCATTATTCCTGCCCACCAATTCATTTCCAGACTTTCTACTTCAGAAGCC

ATGEGGAAAATCAGCAGTCTTCCAACTCAATTATTTAAGATCTGCCTCTGTGACTTCTTG

AAGATAAAGATACACATClATCfTCGTCTTCACATCTCTTCTACCT.GCACTCTGCTTGCTC¡MetperSerSerHlsLeuPheTyrLeuAl aLeuCysLeuLeu

B Domain 360acctttaccagctcggccacagccIggaccagagaccctttccggggctgagctggtggacçœcilalcljThrPheThrSerSerAlaThrAlaplyProGluThrLeuCysGlyAlaGluLeuValAap

O 9r- 420GCTCTTCAATTCGTGTGTGGACCAAGGGGCTTTTACTTCAACAAGCCCACApGCTATGGCAlaLeuGlnPheValCysGlyProArgGlyPheTyrPheAsnLysProThrjGlyTyrGly

A 480TCCAGCATTCCGACCGCACCACAGAcdGGCATTGTGGATGACTGTTGCTTCCCGAGCTGTSerSerlleArgArgAlaProGlnThrlciylleValAspGluCysCysPheArgSerCys

GATCTGAGGAGGCTGGAGATGTACTGTGCTCCGJCTGAAGCCTACAAAGTCAGCTJCGTTCCAspLeuArgArgLeuGluMetTyrCysAlaProLeuLysProThrLysSerAla|ArgSer+ 600

ATCCGGGCCCAGCGCCACACTGACATGCCCAAGACTCAGAAGGAAGTACAC_TTGA_AGAACIleArgAlaClnArgHisThrAspMetProLysThrGlnLysGluValHlsLeuLysABn

660A C AA GTA G AG G A_A G TG_C AG_G A AAÇAAG hÇÇ TAC AGA A TG TAG GAG GAG CCJC CÇ GAG GMThrSerArgGlySerAlaGlyAsnLysThrTyrArgMecEnd

Nucleotide sequencing of rat IGF-I cDNAs andthe 2.3-kb genomic DNA fragment

Bacteriophage DNA was isolated by precipitation withpolyethylene glycol, followed by differential centrifugationin cesium chloride (Maniatis et al. 1982). BacteriophageDNA was cleaved with Eco RI and electrophoresed on

agarose gels. Complementary DNA inserts of 786 bp (P2)and 1200 bp (P3), and a 2.3-kb fragment of rat genomicDNA were isolated by electroelution and subcloned intopBR322 (Sutcliffe, 1978) and M13mpl9 (Norander et al.,1983). DNA fragments cloned into pBR322 were sequencedby the method of Maxam and Gilbert (1977) and also wereisolated for use as probes in RNA blot hybridizations. TheM13mpl9 recombinants were transfected into E. coliJM101 and clones containing the complementary strands ofeach insert were subjected to rapid deletion subcloning(Dale et al., 1985) (Cyclone System, IBI) to generate a

CAGA AA_A T G_CC AC G TC A C_C G C_A AGA T Ç_C T T_T G CT G CT T GA£CAA CC T GCA AAA CATC G G_A780

ACACCJGCC^AAJATCAATA^TGJlGTJCA^TJi£C^Tjrj(^AGAGJ^TGj:GCJiTTJCCj:TC_AAjy 840

8 50A AA AA_C AA AAA CA AAA C

FIG. 1. Nucleotide sequence of P2 cDNA encoding ratIGF-I. The predicted amino acid sequence is shown belowthe DNA sequence. Two potential initiation methioninesare boxed. Large letters B, C, A, and D designate the do-mains of IGF-I. The E domain corresponds to a predictedcarboxy-terminal extension peptide of the rat IGF-I pre-cursor. The region homologous with P3 cDNA (solidunderline) extends to a predicted exon splice site (openarrow). Two base pairs (123, 285) that differed between thetwo clones are indicated. Another exon splice site (solidarrow) was predicted based on the homology of P2 cDNAwith a 2.3 rat IGF-I genomic fragment (dashed underline)that extends past the 3' end of P2 cDNA (hatched arrow).

RAT TESTIS IGF-I cDNA 327

ing frame of 459 bases. The open reading frame ends with a

termination codon (TAG) and 144 bases of 3'-untranslatedregion. Within the 153-amino-acid residue sequence pre-dicted by the open reading frame of cDNA P2, there are

two potential initiator methionine residues. Use of the sec-ond methionine codon as an initiator would lead to syn-thesis of a rat IGF-I precursor of 127 amino acid residues.The coding sequence for the 70-amino-acid residue ratIGF-I begins 49 codons into the open reading frame ofcDNA P2 and is followed by a coding sequence for a car-

boxy-terminal extension peptide (E domain) of 35 aminoacid residues.

Homology between rat, human, and mouseIGF-I mRNAs and precursors

The 153-amino-acid precursor encoded by the longestopen reading frame of cDNA P2 is remarkably similar tothe human IGF-IA precursor (Jansen et ai, 1983; Rotwein,1986a) with only 11 amino acid substitutions, as shown inFig. 2. Five substitutions occur between the two potentialinitiator methionine residues. Within the rat IGF-I se-

quence, compared with human IGF-I, there is a substitu-tion of a Pro for an Asp (residue 20) in the B domain, an

He for a Ser (residue 35) in the C domain, and a Thr for an

Ala (residue 67) in the D domain. The predicted rat IGF-Icarboxyl-terminal extension peptide (E domain) of 35amino acids also is highly homologous to the E domain ofthe human IGF-IA precursor with only three amino acidsubstitutions (Fig. 2). Comparison of the nucleotide se-

quences of the rat and human cDNAs reveals homologiesof 86%, 86%, 88%, and 79% within the 5'-untranslated re-

gion, putative 22-amino-acid signal peptide, coding se-

quence, and 3'-untranslated region, respectively.The 5'-segment of the mouse IGF-IA cDNA (Bell et al.;,

1986), containing 5'-untranslated region and an open read-ing frame encoding 20 amino acids, has poor homologywith both the rat and human cDNAs. The remaining por-tion of the mouse IGF-1 precursor, however, is almostidentical to the rat homolog. Only two amino acids differand one of these is a conservative substitution (Fig. 2).

Note that, although the open reading frames of IGF-IcDNAs from all three species have multiple potential initia-tion methionine residues preceding the coding sequence,only the methionine residue at position -22 is conserved.

RatHumanMouse

Met-41

GlyLysIleSerSerLeuProThrGlnLeuPheLysIleCysLeuCys-Cys-Phe-

Met

R AspPheLeuLysIleLysIleHisIleMetH-Val-Met-ThrM ThrAlaProAla-

-21SerSerSerHlsLeuPheTyrLeuAlaLeu

-i B,DomoinR CysLeuLeuThrPheThrSerSerAlaThrAlalGlyProGluThrLeuCysGlyAlaGluH-

-Thr-20

R LeuValAspAlaLeuGlnPheValCysGlyProArgGlyPheTyrPheAsnLysProThrH-Asp-

^. 40GlyTyrGlySerSerlleArgArgAlaProGlnThrJGlylleValAspGluCysCysPhe-Ser-

60 D.R ArgSerCysAspLeuArgArgLeuGluMetTyrCysAlaPrcJLeuLysProThrLysSerH

R Ala|HM

-Ala--Ala

%* 80ArgSerlleArgAlaGlnArgHlsThrAspMetProLysThrGlnLysGluValHis

-Val-

100R LeuLysAsnThrSerArgGlySerAlaGlyAsnLysThrTyrArgMetEndH-Ala-Asn-M-

FIG. 2. Comparison of rat, human, and mouse IGF-I precursors. The predicted amino acid sequence of the rat testisIGF-I precursor is compared with that of the human IGF-IA variant (Jansen et al., 1983; Rotwein, 1986) and with mouse

IGF-IA precursor (Bell et al., 1986). Dashes indicate identical residues compared with the rat IGF-1 precursor. Conserva-tive substitutions are underlined.

328

Blot hybridizations of rat liver and testispoly(A)* RNA

Blot hybridizations of poly(A)+ RNA revealed that, likerat liver, the rat testis synthesizes four mRNAs (7.5, 4.7,1.7, and 1.2-0.9 kb) that hybridize with both the humanIGF-IA cDNA and the rat IGF-I cDNA P2 (Fig. 3).

Predicted exon splice sites in the rat IGF-I gene

As shown in Fig. 1, 300 bases of 5' sequence in cDNA P2were confirmed by sequencing the 5' end of the 1200-bpcDNA P3. Homology between P2 and P3, however, termi-nated within codon 26 (Asn) of the B domain, which is theexact location of an exon splice site within the humanIGF-I gene (Bell et al., 1985; de Pagter-Holthuizen et al.,1986; Rotwein et al., 1986b). The sequence of P3 adjacentto the point of divergence from P2 (TCA/gtgagtag) is ho-mologous with the consensus sequence CAG/gtgagt for an

exon-intron boundary (Mount, 1982) and has an in-framestop codon TAG that terminates the open reading frame ofP3. Preliminary nucleotide sequence data of P3, down-stream from regions of homology with cDNA P2, revealedno homology with known IGF-I coding sequences (data notshown). Furthermore, a 400-bp Pst I fragment of P3 (en-coding the 5'-untranslated region and B domain) hybrid-ized with mRNAs of the same size as cDNA P2, yet the re-

maining 800 bp of P3 (containing putative intron) did notrecognize any IGF-I mRNAs in rat testes or liver (Fig. 3).Both the 400-bp and 800-bp Pst I fragments of P3 do, how-ever, hybridize with the same rat IGF-I genomic fragmentsin Southern blots (data not shown) excluding the possibilitythat the 3' portion of cDNA P3 was generated by a cloningartifact.

A 2.3-kb rat IGF-I genomic DNA fragment, isolatedfrom a rat genomic library, contains a region that is identi-cal to the 3' end (bases 580-786) of cDNA P2 (Fig. 1). Wepredict that the homologous region of the genomic DNAfragment represents a portion of an exon encoding residues87-105 of the E domain and the 3'-untranslated region ofrat IGF-I mRNAs because the location of the 5' end of thisputative exon is identical to that of the corresponding exonin the human gene (Bell et al., 1985; de Pagter-Holthuizenet al., 1986; Rotwein et al., 1986b). Furthermore, the5'-flanking sequence ttatctttcttacttgcag/G is homologous tothe consensus sequence for an intron-exon boundary,(J.)„Ncag/G (Mount, 1982). The 3' extent of this exon can-

not be predicted because cDNA P2 is not full length and no

polyadenylation signal (Proudfoot and Brownlee, 1976)was found in the available genomic DNA sequence 3' to thetermination of cDNA P2. It is interesting, however (seeDiscussion), that this region of the genomic DNA fragmentcontains long tracts of adenine residues.

CASELLA ET AL.

FIG. 3. Autoradiogram of blot hybridizations of rat liverand testis poly(A)+ RNA with IGF-I cDNAs. In each case,10 pg of rat liver (L) poly(A)* or 20 pg of rat testis (T)poly(A)+ RNA were denatured in glyoxal and dimethyl sulf-oxide, electrophoresed in 1 % agarose, and blotted onto ni-trocellulose. Filters were hybridized (42°C; 50% form-amide) with 32P-labeled probes as indicated: A. HumanIGF-IA; B. rat testis IGF-I cDNA P2; C. 400-bp Pst Ifragment of rat testis cDNA P3 (containing coding se-

quence for signal and B domains of IGF-I precursor); D.800-bp Pst I fragment of cDNA P3 (containing no IGF-Icoding sequence). After hybridization, blots were washedfor 60 min each in 2 x SSC, 0.5% NaDodS04 at 42°C, 2 xSSC, 0.5% NaDodS04 at 65°C, and 0.1% NaDodS04 at60°C; then they were air-dried and exposed to Kodak XARfilm for 24 hr at -70°C with intensifying screens. Longerexposures of D revealed no hybridizing mRNAs. Molecularweights of hybridizing mRNAs were based on comparisonwith denatured Hind III fragments of X DNA.

i

RAT TESTIS IGF-I cDNA 329

DISCUSSION

A 786-bp cDNA (P2), containing coding sequences forrat IGF-I, was isolated from a rat testis cDNA library. Thepresence of a IGF-I cDNA within a testis cDNA library, to-

gether with our findings that this cDNA hybridizes withfour mRNAs (7.5, 4.7, 1.7, and 1.2-0.9 kb) in testispoly(A)* RNA, provides direct evidence that IGF-I is syn-thesized in rat testis as previously suggested (Ritzen, 1983;D'Ercole et al., 1984; Smith et ai, 1987).

The rat IGF-I cDNA P2 contains a 459-base open read-ing frame with two potential initiator methionine codonsthat indicate synthesis of a 153- or 127-amino-acid ratIGF-I precursor. Both of the putative initiation se-

quences, AAGCGATG(G) for the first Met codon andACATCATG(C) for the second, are homologous with theconsensus initiation sequence CC^CCATG(G) (Kozak,1984). The methionine at position -22 might be the mostlikely initiation site, however, because it is followed by a

stretch of 21 hydrophobic amino acids that is typical of a

signal sequence (Blobel and Dobberstein, 1975) and it isconserved in both the mouse and human prcursors. Cell-free translation studies will be necessary to establish defini-tively which of the two methionines in the rat IGF-I cDNAcorresponds to the initiator and to determine the size of thetranslation products of rat IGF-I mRNAs.

The 70-amino-acid rat IGF-I sequence predicted fromthe open reading frame of cDNA P2 shows remarkableconservation with the human IGF-I sequence with onlythree amino acid substitutions, one in each of the B, C, andD domains. The predicted amino acid sequence is in com-

plete agreement with the partial protein sequence reportedby Rubin et al. (1982). Cysteine residues are conserved be-tween human IGF-I and our predicted rat IGF-I sequence,which suggests conservation of disulfide bonds and muchof the tertiary structure of IGF-I across mammalian spe-cies. The substitution in the C domain supports the find-ings of Wilson and Hintz (1982) who, based on studies withan anti-peptide antibody directed against the human C do-main, concluded that the C domains of rat and humanIGF-I differed.

Our sequence data for rat IGF-I do not explain the sig-nificant difference in the isoelectric points of human IGF-Iand rat IGF-I (Rubin et ai, 1982; Smith et al., 1987). Onepossibility is that the three amino acid substitutions in ratIGF-I induce subtle changes in the tertiary structure thatshift the charge of the molecule. It is of interest, however,that our sequence does not entirely agree with the quantita-tive amino acid analysis of Rubin et al. (1982). Contrary tothe amino acid analysis data, the basic residues (6 Arg, 3Lys) were all conserved and no His residues were presentwithin the predicted IGF-I coding sequence. These consid-erations, along with the observations of four IGF-ImRNAs in rat testes and liver, raise the possibility thatthere are variant forms of rat IGF-I. Rotwein (1986a) hasdescribed two IGF-I mRNAs in human liver that encodedifferent carboxy-terminal extension peptides (E domains)and Gemmons and Shaw (1986) independently isolated a

form of IGF-I from human fibroblasts that had an aminoacid composition identical to the 1GF-IB precursor de-scribed by Rotwein. Recently, Bell et al. (1986) demon-strated that the mouse liver contains mRNAs encoding atleast two different precursors of IGF-I. The multiple IGF-ImRNAs in rat tissues may similarly represent differentprecursors or variant forms of IGF-I.

It is apparent that cDNA P2 is a truncated copy of one

of the hybridizing IGF-I mRNAs (7.5, 4.7, 1.7, and 1.2-0.9kb) present in rat testis. The absence of a polyadenylationsignal or poly(A) tract in cDNA P2 suggests that the 3'un-translated region of cDNA P2 is incomplete. Furthermore,no polyadenylation signal was present in the 80 bases of ge-nomic sequence downstream from the termination ofcDNA P2 (Fig. 1). We conclude, therefore, that the poly-adenylation signal and poly(A) addition site are further 3'in this genomic exon or represented on an additional exon,and further sequencing of rat IGF-I genomic DNA frag-ments will be necessary to distinguish between these possi-bilities. Within the genomic DNA sequence 3' to the termi-nation of cDNA P2, we found tracts of adenine residuesthat may be present in the rat IGF-I mRNA(s) upstreamfrom the poly(A) tail. A similar tract of adenine residueswas reported for the 3'-untranslated region of the humanIGF-IA cDNA and was also found in the corresponding ge-nomic exon (Bell et ai, 1985; de Pagter-Holthuizen et al,1986; Rotwein et ai, 1986b). Conservation of these A-richregions within the 3'-untranslated region of rat and humanIGF-I mRNAs and genomic fragments implies a functionalsignificance. Such A-rich sequences may influence the sta-bility of mRNAs encoding growth factors and oncogenes(Shaw and Kamen, 1986).

A second IGF-I cDNA, P3, was isolated from the testiscDNA library. This cDNA does not encode the completeIGF-I precursor sequence but contains coding sequencesfor putative signal domains and residues 1-25 of the B do-main. The sequence 3' to this region is a closed readingframe with no detectable homology with reported IGF-IcDNAs. We believe that this cDNA corresponds to a

partial copy of a IGF-I gene transcript with regions down-stream from the codon for residue 26 of the B domain cor-

responding to intron sequence. This is because the homol-ogy of P3 cDNA terminates at the exact location of an

exon splice site in the human IGF-1 gene and the 5' end ofthe putative intron is homologous to the consensus exon-

intron boundary sequence (Mount, 1982). Our findingsthat the putative intron portion of the cDNA do not hy-bridize to any major mRNA species in rat liver and testis,argue against an alternate interpretation that this mRNA isa functional mRNA splicing variant that would be trans-lated into a truncated IGF-I precursor containing onlyputative signal and B domain sequences. If we assume,based on these considerations, that the 3' end of P3 cDNAis indeed intron, the lack of hybridization of this portion ofthe cDNA to the 7.5-kb and 4.7-kb IGF-I mRNAs also pro-vides evidence that these large-molecular-weight mRNAsdo not represent primary unprocessed transcripts of theIGF-I gene.

330 CASELLA ET AL.

ACKNOWLEDGMENTS

This work was supported by National Institutes ofHealth Grants 1R29 DK38542-01 (S.J.C.), 5 ROÍAM01022-31 (J.J.V.W.), and 5P30 HD18968-03 (Labora-tories for Reproductive Biology).

REFERENCES

BELL, G.I., GERHARD, D.S., FONG, N.M., SANCHEZ-PESCADOR, R., and RALL, L.B. (1985). Isolation of the hu-man insulin-like growth factor genes: Insulin-like growth factorII and insulin genes are contiguous. Proc. Nati. Acad. Sei. USA82, 6450-6454.

BELL, G.I., STEMP1EN, M.M., FONG, N.M., and RALL, L.B.(1986). Sequences of liver cDNAs encoding two different mouseinsulin-like growth factor I precursors. Nucleic Acids Res. 14,7873-7882.

BENTON, W.D., and DAVIS, R.W. (1977). Screening Xgt recom-binant clones by hybridization to single plaques in situ. Science196, 180-182.

BERNIER, M., CHATELAIN, P., MATHER, J.P., and SAEZ,J.M. (1986). Regulation of gonadotropin receptors, gonado-tropin responsiveness, and cell multiplication by somato-medin-C and insulin in cultured pig Leydig cells. J. Cell. Phys-iol. 129, 257-263.

BLOBEL, G., and DOBBERSTEIN, B. (1975). Transfer of pro-teins across membranes: I. Presence of proteolytically processedand unprocessed nascent immunoglobulin light chains on mem-

brane-bound ribosomes of murine myeloma. J. Cell Biol. 67,835-851.

CLEMMONS, D.R., and SHAW, D.S. (1986). Purification andbiological properties of fibroblast somatomedin. J. Biol. Chem.261, 10293-10298.

DALE, R.M.K., McCLURE, B.A., and HOUCHINS, J.P.(1985). A rapid single-stranded cloning strategy for producing a

sequential series of overlapping clones for use in DNA sequenc-ing: Application to sequencing the corn mitochondrial 18 SrDNA. Plasmid 13, 31-40.

DE PAGTER-HOLTHUIZEN, P., VAN SCHAIK, F.M.A.,VERDUIJN, G.M., VAN OMMEN, G.J.B., BOUMA, B.N.,JANSEN, M., and SUSSENBACH, J.S. (1986). Organizationof the human genes for insulin-like growth factors I and II.FEBS Lett. 195, 179-184.

D'ERCOLE, A.J., STILES, A.D., and UNDERWOOD, L.E.(1984). Tissue concentrations of somatomedin-C: Further evi-dence for multiple sites of synthesis and paracrine or automnemechanisms of action. Proc. Nati. Acad. Sri. USA 81, 935-939.

JANSEN, M., VAN SCHAIK, F.M.A., RICKER, A.T., BUL-LOCK, B., WOODS, D.E., GABBAY, K.H., NUSSBAUM,A.L., SUSSENBACH, J.S., and VAN DEN BRANDE, J.L.(1983). Sequence of cDNA encoding human insulin-like growthfactor 1 precursor. Nature 306, 609-611.

JOSEPH, D.R., HALL, S.H., and FRENCH, F.S. (1985). Identi-fication of complementary DNA clones that encode rat andro-gen binding protein. Andrology 6, 392-395.

KOZAK, M., (1984). Compilation and analysis of sequences up-stream from the translational start site in eukaryotic mRNAs.Nucleic Acid Res. 12, 857-872.

LUND, P.K., MOATS-STAATS, B.M., HYNES, M.A., SIM-

MONS, J.G., JANSEN, M., D'ERCOLE, A.J., and VANWYK, J.J. (1986). Somatomedin-C/insulin-like growth factor-Iand insulin-like growth factor-11 mRNAs in rat fetal and adulttissues. J. Biol. Chem. 261, 14539-14544.

MANIAT1S, T., FRITSCH, E.F., and SAMBROOK, J. (1982).Molecular Cloning: A Laboratory Manual. (Cold SpringHarbor Laboratory, Cold Spring Harbor, New York).

MAXAM, A.M., and GILBERT, W. (1977). A new method forsequencing DNA. Proc. Nati. Acad. Sei. USA 74, 560-564.

MOUNT, S.M. (1982). A catalogue of splice junction sequences.Nucleic Acids Res. 10, 459-472.

NORANDER, J., KEMPE, T., and MESSING, J. (1983). Con-struction of improved M13 vectors using oligodeoxynucleotide-directed mutagenesis. Gene 26, 101-106.

PROUDFOOT, N.J., and BROWNLEE, G.G. (1976). 3' Non-coding region sequences in eukaryotic messenger RNA. Nature263, 211-214.

RITZEN, E.M. (1983). Chemical messengers between Sertoli cellsand neighbouring cells. J. Steroid Biochem. 19, 499-504.

ROTWEIN, P. (1986a). Two insulin-like growth factor I messen-

ger RNAs are expressed in human liver. Proc. Nati. Acad. Sei.USA 83, 77-81.

ROTWEIN, P., POLLOCK, K.M., DIDIER, D.K., and KRIVI,G.G. (1986b). Organization and sequence of the human insulin-like growth factor I gene. J. Biol. Chem. 261, 4828-4832.

RUBIN, J.S., MARIZ, I., JACOBS, J.W., DAUGHADAY,W.H., and BRADSHAW, R.A. (1982). Isolation and partial se-

quence analysis of rat basic somatomedin. Endocrinology 110,734-740.

SANGER, F., NICKLEN, S., and COULSON, A.R. (1977). DNAsequencing with chain-terminating inhibitors. Proc. Nati. Acad.Sei. USA 74, 5463-5467.

SHAW, G., and KAMEN, R. (1986). A conserved AU sequencefrom the 3' untranslated region of gm-csf mRNA mediates se-

lective mRNA degradation. Cell 46, 659-667.SMITH, E.P., SVOBODA, M.E., VAN WYK, J.J., KIERSZEN-

BAUM, A.L., and TRES, L.L. (1987). Partial characterizationof a somatomedin-like peptide from the medium of cultured ratSertoli cells. Endocrinology 120, 186-193.

SUTCLIFFE, J.G. (1978). pBR322 restriction map marked fromthe DNA sequence: Accurate DNA size markers up to 4361 nu-cleotide pairs long. Nucleic Acids Res. 5, 2721-2728.

TRES, L.L., SMITH, E.P., VAN WYK, J.J., and KIERSZEN-BAUM, A.L. (1986). Immunoreactive sites and accumulationof somatomedin-C in rat Sertoli-spermatogenic cell co-cultures.Exp. Cell Res. 162, 33-50.

WILSON, D.M., and HINTZ, R.L. (1982). Inter-species compari-son of somatomedm structure using immunological probes. J.Endocrinology 95, 59-64.

VAN WYK, J.J. (1984). The somatomedins: Biological actionsand physiologic control mechanisms. In Hormonal Proteins andPeptides, 12, C.H. Li, Ed. (Academic Press, New York) pp. 81-125.

Address reprint requests to:Dr. Samuel J. Casella

Depl. of Pediatrics509 Clinical Sciences Bldg. 229H

University of N. Carolina at Chapel HillChapel Hill, NC 27514

Received for publication March 9, 1987, and in revised form April24, 1987.