Proc. Nadl. Acad. Sci. USAVol. 81, pp. 1346-1350, March 1984Biochemistry

Isolation and characterization of the rat tyrosineaminotransferase gene

(exon-intron structure/glucocorticoid control/cAMP control/regulatory DNA sequence)

TAKAHISA SHINOMIYA*, GERD SCHERER*t, WOLFGANG SCHMID*, HANSWALTER ZENTGRAFt,AND GUNTHER SCHUTZ**Institute of Cell and Tumor Biology, and tInstitute of Virus Research, German Cancer Research Center, Im Neuenheimer Feld 280, D-6900 Heidelberg,Federal Republic of Germany

Communicated by M. Lindauer, November 14, 1983

ABSTRACT Tyrosine aminotransferase (TAT; L-tyro-sine:2-oxoglutarate aminotransferase, EC 2.6.1.5) from ratliver is subject to glucocorticoid, cAMP, and developmentalcontrol. To study the underlying regulatory mechanisms, theTAT structural gene was isolated from a X bacteriophage ratDNA library. Heteroduplex analysis revealed that the 2.4-kilo-base-long TAT mRNA is encoded by a gene that extends over11 kilobases and is interrupted by 11 introns. To characterizethe presumptive control region, the DNA sequence around the5' end of the gene was determined and the start site of tran-scription was identified by nuclease S1 protection experi-ments. A short sequence homology in an equivalent positionrelative to the cap site was detected between TAT and trypto-phan oxygenase, another glucocorticoid-controlled gene fromrat liver. This sequence is related to the sequence 5' T-G-T-T-C-T 3' found in regions of the long terminal repeat of mousemammary tumor virus, which has been shown to interact withthe glucocorticoid receptor [Scheidereit, C., Geisse, J., West-phal, H. M. & Beato, M. (1983) Nature (London) 304, 749-7521.

A classical example for steroid hormone control is the induc-tion of the enzymatic activities of both tyrosine aminotrans-ferase (TAT; L-tyrosine:2-oxoglutarate aminotransferase,EC 2.6.1.5) and tryptophan oxygenase [TO; tryptophan 2,3-dioxygenase, L-tryptophan:oxygen 2,3-oxidoreductase (de-cyclizing), EC 1.13.11.11] in rat liver. The increase in activi-ty of these two gluconeogenic enzymes after glucocorticoidstimulation results from an increased rate of enzyme synthe-sis (1-3), which derives from an increase in the amount oftranslatable mRNA (4-6). Apart from glucocorticoid induc-tion, the activity of TAT is also inducible by cAMP (7, 8).Since both hormonal controls for TAT are also operative in anumber of established rat hepatoma cell lines, the hormonalinduction ofTAT in hepatoma cells has become an attractivemodel system to study regulated gene expression in mamma-lian cells (9).

In addition to hormonal modulation, both TAT and TOactivity are also under developmental control. While TATenzyme activity appears around birth (10), TO enzyme activ-ity can be detected around postnatal day 15 (11).By genetic and biochemical analysis of several albino le-

thal mutants of the mouse, a control region required forexpression and inducibility of mouse TAT has been assignedto the region of the albino locus on chromosome 7 (12). Inthese mutants, the activity of several liver enzymes is affect-ed. The basal level of TAT is severely decreased and theenzyme activity is no longer inducible by glucocorticoids,although the structural gene for TAT is still present (13).

To study the complex control mechanisms operating onthese two liver-specific genes at the molecular level, wehave isolated the TAT and TO genes from the rat genome..We have previously described the isolation and characteriza-tion of cDNA as well as genomic clones for TO (14). Withthese molecular clones as probes, some of us have shownthat glucocorticoid induction of TO occurs at the transcrip-tional level (15). Similar analysis has shown that glucocorti-coid and cAMP induction of TAT also result from transcrip-tional activation of the gene (unpublished data).

In this report, we describe the isolation of the TAT genefrom a rat phage library, using a recently isolated cDNAclone as probe (16). The structural features of the rat TATgene are presented. To study the mechanism of transcrip-tional activation and to enable the design of hybrid genes,the DNA sequence at the 5' end of the TAT gene was deter-mined. Comparison of this sequence with the DNA sequencein the 5' region of TO and with sequences in the long terminalrepeat of mouse mammary tumor virus DNA known to inter-act with the glucocorticoid receptor (17) revealed shortstretches of sequence homology.

MATERIALS AND METHODSConstruction and Screening of the Rat Genomic Library.

High molecular weight DNA (100 ,g) from livers of twoadult male Wistar rats was partially digested with Sau3A andfractionated by electrophoresis on a 0.5% low-melting-tem-perature agarose gel. DNA fragments 15-19 kilobases (kb)long were recovered as described (18) and ligated to BamHI-cut DNA of the A phage vector EMBL3. This cloning vector,a derivative of the vector X1059 (18), allows recovery of thecloned DNA as Sal I fragments (19). The ligated DNA waspackaged in vitro yielding a total of 11 x 106 recombinantphages. To establish a library, 3 x 106 recombinants wereamplified, and 1 x 106 original recombinants were screeneddirectly (20) by using nick-translated pcTAT-3 DNA (16) asprobe. Five independent clones were isolated, one of which,XrTAT1, is described in this report.

Subcloning and DNA Sequence Analysis. The 2.7-kb Sal I-fragment of XrTAT1 was subcloned into the polylinker re-gion of plasmid pUC8 (21) yielding plasmid pUTATS2.7. De-letions removing segments of the 5' flanking region weregenerated essentially as described by Frischauf et al. (22). Inbrief, the plasmid was linearized at random by partial diges-tion with DNase I and ligated to BamHI linkers. After diges-tion with BamHI, which cuts once in pUC8, the DNA prepa-ration was fractionated on an agarose gel and DNA frag-ments of the desired lengths (5.4-3.5 kb) were circularizedwith T4 DNA ligase and cloned. Appropriate clones, identi-

Abbreviations: TAT, tyrosine aminotransferase; TO, tryptophanoxygenase; kb, kilobase(s).tTo whom reprint requests should be addressed.

1346

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad Sci USA 81 (1984) 1347

fied by restriction analysis, were sequenced (23). Aftercleavage at the single BamHI site, the DNA was labeled atthe 5' or 3' end and recut with EcoRI. Both strands werealways sequenced.Heteroduplex Analysis. Heteroduplex analysis was carried

out as described (14), except that enriched TAT mRNA wasobtained by hybridizing total rat liver RNA to pcTAT-3DNA immobilized to macroporous Sephacryl S-500 (24).

Nuclease Si Mapping. Nuclease Si analysis was carriedout essentially as described (14), except that RNA-DNA hy-brids were formed at 480C. The nuclease Si probe was ob-tained as follows: the subclone pUTATS2.7 was digestedwith Mst II, treated with calf intestine alkaline phosphatase,and 5'-end-labeled with [y-32P]ATP using T4 polynucleotidekinase. After recutting with EcoRI, the 1.7-kb EcoRI/Mst IIfragment, uniquely labeled at the Mst II 5' end at position+84, was isolated.

RESULTSIsolation and Characterization of a X Recombinant Phage

Containing the Complete TAT Structural Gene. RadiolabeledpcTAT-3 DNA, a cDNA clone containing approximatelyone-third of the TAT mRNA sequences (16) (Fig. iC) wasused to screen a phage library of rat liver DNA generated bypartial digestion with Sau3A. From 1 x 106 recombinantphages screened, five independent phages were isolated andshown by restriction endonuclease analysis and Southernblotting (not shown) to span a contiguous region of 27 kb.The restriction map of one genornic clone, XrTAT1, whichcontains the entire TAT structural gene, is shown in Fig. lA.This clone carries a 17.5-kb insert with 1.5 kb of 5'-flankingDNA, 11 kb of gene sequences, and 5 kb of 3'-flanking DNA(Fig. 1B).The structure of the TAT gene was determined by analysis

of heteroduplexes formed between XrTAT1 DNA and rat liv-er poly(A)+ RNA enriched for TAT mRNA. A represent-ative electron micrograph with corresponding interpretationis shown in Fig. 2.

A L :i I I I1 1 I I

B

FIG. 2. Electron microscopy of a TAT heteroduplex. (A) Hybridformed between TAT mRNA and the DNA of XrTAT1. (far = 0.1,um.) (B) Interpretive drawing with RNA sequences represented by adashed line and DNA sequences by a solid line. The lengths of exonsA-L and introns 1-11 are given in Fig. 1B.

The resulting structural map of the TAT gene relative tothe restriction map is shown in Fig. 1. The direction of tran-scription was initially established by determining whichstrand of the cDNA insert in clone pcTAT-3 is protectedfrom nuclease S1 digestion by TAT mRNA (data not shown)

c cw -rI w w

I I I I I -ER

1021 176 224799 24 134

1 2 35'

A BC70 20710914 31 22

1152 614 632 820 511 417 72436691 65 72 84 51 42 64504 5 6 7 8 9 10 11

D E F G H I J K L3'

79 155 140 82 145 136 95 106 990

19 26 22 21 20 25 20 23 47

I-- -H1-IipcTAT-3 pcTAT-2

I .. I I I I0 1 2 3 4 5 6 7 8 9 10 11 kb

FIG. 1. Organization of the rat TAT gene. (A) Map of restriction sites in clone XrTAT1. Left and right arms of the EMBL3 vector areindicated by L and R, respectively. Fragments containing repetitive sequences are marked with a dot (e). (B) Location and sizes of the exons(A-L, open boxes) and introns (1-11) as determined by heteroduplex experiments, an example of which is shown in Fig. 2. Except for exon A(only 10 molecules), the sizes for each individual exon and intron were determined from 20-70 molecules. (C) Positions of the two cDNA clonespcTAT-2 and pcTAT-3. The position of pcTAT-3 is approximate, as indicated by the dashes. (D) Scale in-kb.

Intron

BExon

C

D

Biochemistry: Shinomiya et aL

1348 Biochemistry: Shinomiya et al.

G A T C._e~rn

Sma I Hind mSmam I < E z > Pst ISalHI Sal I

V e0 -,_4

FIG. 3. Construction of deletion mutants for sequence analysis.Plasmid pUTATS2.7 contains the 2.7-kb Sal I fragment of XrTAT1cloned in pUC8. Deletions were generated (22) by inserting BamHIlinkers at random and deleting the DNA segments between theBamHI site in the polylinker region of pUC8 and the inserted linker,as indicated by arrows. The map is not drawn to scale. The filledbox symbolizes exon A as determined by heteroduplex measure-ments, and the arrow shows the start site of transcription. Five dele-tion clones were selected to establish the sequence shown in Fig. 4.The following regions were analyzed: A2/8 (-600 to -418), A5/3(-436 to -173), A5/4 (-220 to -7), A7/1 (-11 to +151), and A4/6(+97 to +300).

and later confirmed by the nuclease S1 mapping experimentsdescribed below. As shown in Fig. 1, the TAT gene spans 11kb and has a total of 11 introns. The heteroduplex measure-ments give a total length for the 12 exons of 2300 base pairs,in good agreement with our earlier size estimate for the TATmRNA of about 2400 bases (16).

Restriction fragments carrying repetitive sequences, iden-tified by hybridization to 32P-labeled rat liver DNA, can befound both within and flanking the TAT gene (Fig. 1A).

Determination of the Sequence of the 5'-End Region and ofthe Start Site of Transcription of the TAT Gene. To measurethe distance of the first exon from the left arm of the EMBL3vector, heteroduplex experiments were carried out as de-

1 2 3 4 5

G

CA \C

G

A

T

C .+1A -

r~~~A-

T

A

Gw

3

FIG. 5. Localization of the TAT mRNA start site by nuclease S1mapping. Lanes G, A, T, and C: sequencing ladder starting from theMst II site at position +84. Note that the labeling of the sequencinglanes has been transposed so the mRNA sequence can be read di-rectly. Lane 1, nuclease Si-resistant products obtained by digestinga hybrid between total poly(A)+ RNA isolated from livers of dexa-methasone-induced animals and the Mst II probe fragment with 200units of nuclease S1 per ml. Lane 2, same as lane 1, but using 1600units of nuclease S1 per ml. Lane 3, same as lane 1, but with totalpoly(A)+ RNA isolated from dibutyryl cAMP-injected animals.Lane 4, same as lane 1, but with RNA isolated from adrenalecto-mized rats. Lane 5, same as lane 1, except that yeast tRNA was

incubated with the probe. Two micrograms of RNA was used in allcases. The putative transcription start site is indicated by an arrow

at the sequence of interest.

scribed but including single strands of an unrelated EMBL3clone. This clone can form hybrid molecules with cloneXrTAT1 via the common vector arms, while TAT mRNA can

- 600 TCTGTGCMTGAAMACTTATiAAATMIIGTACGCAMATTCTGCATTTCTAM AGCTATCCGCATACTTATCTATTTTGTGMAATCAGtGAAAITTI

- 500 fATATGrrTTMCCCTTGGAATGCGGTTGAAMTTMGTGGATATTACCGTTGTCCGTAGCAAATCCCACATATGGTAGAT-tGTATTCAACCAAACTCTTA

- L40 TCATTGCTTATATCCAGGAGTGN3ACTAGA3TGCACAGAGAAAGGGGTACCAGAAGGACCCTCGTTTAAATGTATIICTGTACTGTTAACACAAIMICA

- 300 TAGTCACCCCAACTGCCCTCCACCATTTTCTCAAACATCTCACCAMATGACTAGAAAGAGTTAACAGGrATr

- 200 CCAGATACTTGATGTAN3GACAMTCCCAGATTGGJAAGGTGGCCCAGGGTTGGGGTGAGAAACAGCAGAGTGGGGGGTGGGGTATGGGGGTAGGTCCGGG

- 100 GG TTAGTTCTCACTCTCGMGGCTTCGGGCCCMCGCCCATTGGCTGAAA TC3GGGtCAGGACTGCACCTrGAGCTCTO TCAGTTCTGA

+ 1 ATCATCAGAGGACTTGGTAGGAW3PCTTATAGGTMTGGCTGCTGGGAGGCTGCTTTGCCTGGACTTG;GAAGGTAATCCCTGAGGTTCTCTACACTGTGMst II

+ 101 ATGGTGATGGGATGCTCCTGGCTTTCTTi-CTG-ATAGTCTGACTT-MlCTGCCCAGGGCTGAGATGCTTATCTGACATGr-,CTTTACTTTAACTrGTATCA

+ 201 ATATTCATATCTCTCTCTCTCTCTCTCTCTCTCTCTCTCTCTCTCTCTCtCTCTCTCTCATAGATTGMCATAGGTCCTGrCTCTGCTAGCMAGTGCTTFIG. 4. Nucleotide sequence of the 5'-end region of the rat TAT gene. DNA sequences around exon A of the TAT gene as determined by

deletion sequence analysis (see Fig. 3). Position + 1 indicates the presumed cap site (see Fig. 5). The 3' end of exon A has not been mapped, butis definitely beyond the Mst II site. "TATA"- and CCAAT-like sequences are boxed. The sequence homologous to a sequence found at position-101 to -110 in the rat TO gene (14) is indicated by asterisks. Sequences similar to the hexanucleotide sequence T-G-T-T-C-T found in regionsof the long terminal repeat of MMTV interacting with the glucocorticoid receptor (17) are overlined.

Proc. NatL Acad ScL USA 81 (1984)

Proc. NatL. Acad. ScL USA 81 (1984) 1349

still hybridize to the insert region in XrTAT1 DNA. Thisanalysis allowed us (data not shown) to map the beginning ofthe first exon at about 1.5 kb from the left vector arm, orfrom the Sal I site at the arm-insert junction (Fig. LA).

In addition, run-off experiments using an in vitro tran-scription system revealed a start site for an RNA polymeraseII transcript 580 ± 10 base pairs upstream of the leftmostEcoRI site, again -1.5 kb from the Sal I site at the left arm-insert junction (data not shown).

Therefore, the 2.7-kb Sal I fragment from the left end ofthe cloned insert (Fig. 1A) was subcloned into pUC8, and theregion of interest was analyzed (23) using the deletion se-quencing strategy of Frischauf et al. (22). The constructionof the deletion mutants is schematically shown in Fig. 3, withthe resulting nucleotide sequence presented in Fig. 4. In-spection of the DNA sequence reveals the sequence T-A-T-T-T-C-A, an atypical variant of the T-A-T-A-A-A-A consen-sus sequence characteristic for RNA polymerase II promot-er sites (25), located between nucleotides -31 and -25relative to the +1 start site defined below. In addition, be-tween nucleotides -77 and -69 is a sequence related to the"CCAAT"-box sequence found in a number of eukaryoticgenes at a corresponding distance from the cap site (26).The start site of transcription of the TAT gene was deter-

mined by nuclease S1 mapping (27). A 1.7-kb EcoRI/Mst IIfragment, 5'-end labeled in the coding strand of the TATgeneat the Mst II site at position +84 (Fig. 4), was used as aprobe. Hybridization to total poly(A)+ RNA from livers ofdexamethasone-stimulated rats yields two protected DNAfragments two nucleotides apart (Fig. 5, lane 1). The sametwo protected fragments are observed with RNA isolatedfrom livers of dibutyryl cAMP-injected or adrenalectomizedanimals, respectively (Fig. 5, lanes 3 and 4). In this experi-ment, the RNA samples isolated from the induced animalsshow a 5- to 10-fold increase in TAT mRNA, in confirmationof our earlier RNA blotting data (16). Note that uninducedand induced RNA samples originate from the same start site.Because the upper of the two protected fragments be-

comes weaker with the addition of increasing higheramounts of nuclease S1 (compare lanes 1 and 2 of Fig. 5),and because most eukaryotic mRNAs initiate with an ade-nine residue preceded by a cytosine residue (25), we definethe adenine residue labeled + 1 in Figs. 4 and 5 as the cap siteof the TAT mRNA, taking into account that nuclease S1-generated fragments are retarded by 1.5 nucleotides relativeto chemically cleaved DNA on a sequencing gel (28).

DISCUSSIONWe have isolated and characterized the structural gene en-coding rat TAT. This gene extends over 11 kb and is inter-rupted by 11 introns. It is thus larger by a factor of 4.6 com-pared to the 2400-base-long mRNA that it encodes. The larg-est exon of the TATgene is a 1.0-kb segment located at the 3'end. It very likely encodes mainly untranslated sequences,because only about 1400 nucleotides of the 2400-base-longTAT mRNA are needed to encode the 52,000-dalton TATprotein.Because the restriction maps of the five independently iso-

lated TAT clones from our rat liver genomic library overlapto form a unique map, we conclude that the rat TAT gene is asingle copy gene. This conclusion is corroborated by restric-tion analysis of TAT clones isolated from other phage librar-ies during the initial phase of this work. Two different cloneswere isolated from the rat liver DNA library of Sargent et al.(29), and five different clones were isolated from a rat HTCcell DNA library obtained by K. Yamamoto. While none ofthese clones contained the 5'-end region of the TAT gene,their restriction maps and the heteroduplex measurementsmade on two HTC clones agree with the data for the clonesfrom our own library. The only difference was a length het-

erogeneity, observed in intron 3, in which clone XrTAT1 has200 base pairs less sequence information than the corre-sponding HTC clone. This difference is probably due to thegain or loss of a repetitive sequence element (unpublisheddata). Furthermore, genomic blots of rat liver DNA probedwith various nonrepetitive restriction fragments derivedfrom XrTAT1 did not reveal evidence for a second TAT gene(data not shown).Unambiguous proof that clone XrTAT1 contains the au-

thentic rat TAT gene comes from the observation that stabletransformants of mouse LTK- cells, obtained by cotransferof XrTAT1 DNA together with the chicken thymidine kinasegene, do express TAT enzyme activity (unpublished data).To allow sequence comparisons with other glucocorticoid-

controlled genes and the design of chimeric genes for func-tional studies, a region of 800 bases around the 5' end of theTAT gene has been analyzed, and the start site of transcrip-tion has been determined by nuclease S1 mapping. It is notclear whether the two closely spaced nuclease-resistant frag-ments observed in this experiment (Fig. 5) result from in-complete nuclease S1 digestion or if they reflect a true start-site microheterogeneity as observed in a number of otherwell-characterized systems (30, 31).The 3' end of the first exon has not yet been determined,

but must lie beyond the Mst II site used for nuclease S1 map-ping. There are four potential G-T splice-donor sequenceswithin the next 50 bases following the Mst II site, with theDNA sequence around the G-T at position + 104 having thebest homology to the splice-donor consensus sequence (25,32). If this site is actually used, the first exon will just termi-nate with an ATG codon. Note also that there is a dinucleo-tide repeat within the first intron, with 25 consecutive T-Cresidues starting at position +210.We have used computer analysis to search for sequence

homologies in the 5'-flanking regions of the TAT and the TOgene (14), because both genes are under glucocorticoid con-trol in rat liver. We have previously noted a sequence at po-sition -101 to -110 in the TO gene, which is similar to asequence present at position -102 to -109 in the long termi-nal repeat of mouse mammary tumor virus (MMTV) (14). Inthe TAT sequence shown in Fig. 4, we find a single sequencewith a fit of 8/10 nucleotides to this TO sequence, located atposition -83 to -92, again at a similar position relative tothe RNA start site. By nuclease protection experiments, tworegions in the long terminal repeat of MMTV have recentlybeen identified to interact with the glucocorticoid receptorprotein, and the hexanucleotide sequence T-G-T-T-C-T hasbeen found to be present four times within these regions (17;K. Yamamoto, personal communication). Interestingly, theshort TAT-TO homology described above is closely relatedto this T-G-T-T-C-T sequence, with the first thymine re-placed by an adenine residue. We note, however, that se-quences deviating by one nucleotide from the T-G-T-T-C-Tsequence motif of MMTV do occur at a total of six positionsin the 5'-flanking region of the TAT gene, indicated by over-lining in Fig. 3. There is no perfectly matching hexanucleo-tide T-G-T-T-C-T in our TAT sequence. We do not find anyregion in the TAT gene having more than 60% homology tothe 24-nucleotide-long sequence reported by Cochet et al.(33), which is shared by several glucocorticoid regulatedgenes.

Besides being controlled by glucocorticoids, the expres-sion of the TAT gene is also regulated by cAMP (7, 8). Wehave recently shown that the increase in TAT mRNA activi-ty in rat liver after cAMP stimulation results from a rapidincrease in the transcription rate of the gene (unpublisheddata). We therefore compared the 5'-flanking sequences ofthe TAT gene with corresponding sequences from the ratprolactin gene (34), which is expressed in the pituitary andalso rapidly responds to cAMP by increasing its nuclear tran-

Biochemistry: Shinomiya et aL

1350 Biochemistry: Shinomiya et al.

scription rate (35, 36). While we observe a number of shortsequence homologies, it is at the moment impossible to as-sess their biological significance. Gene transfer studies withchimeric genes containing the TAT promoter region may al-low us to probe the importance of these and of the othersequences discussed above in the various controls operatingon the TAT gene.

We thank T. Sargent and K. Yamamoto for their phage libraries,V. Pirrotta for gifts of packaging extracts, C. M. Strange for help inthe initial phase of this work, S. Hashimoto for doing the run-offexperiments, and S. Suhai for help with computer assisted sequenceanalysis. We are grateful to M. Jantzen, R. Miksicek, and R.Renkawitz for helpful comments on the manuscript and J. Purkertand M. Cole for typing. This work was supported by a grant from theHumboldt-Stiftung to T.S. and grants from the Deutsche For-schungsgemeinschaft (Schu 51/4-1 and Schu 51/5-3) and the Fondsder Chemischen Industrie to G. Schutz.

1. Feigelson, P. & Greengard, 0. (1962) J. Biol. Chem. 237, 3714-3717.

2. Schimke, K. T., Sweeney, E. A. & Berlin, C. M. (1965) J.Biol. Chem. 240, 322-331.

3. Kenney, F. T. (1962) J. Biol. Chem. 237, 3495-3498.4. Schutz, G., Beato, M. & Feigelson, P. (1973) Proc. Natl.

Acad. Sci. USA 70, 1218-1221.5. Schutz, G., Killewich, L., Chen, G. & Feigelson, P. (1975)

Proc. Natl. Acad. Sci. USA 72, 1017-1020.6. Rowekamp, W., Hofer, E. & Sekeris, C. E. (1976) Eur. J. Bio-

chem. 70, 259-268.7. Wicks, W. D., Kenney, F. T. & Lee, K. L. (1969) J. Biol.

Chem. 244, 6008-6013.8. Noguchi, T., Diesterhaft, M. & Granner, D. (1978) J. Biol.

Chem. 253, 1332-1335.9. Tomkins, G. M., Gelehrter, T. D., Granner, D., Martin, D.,

Jr., Samuels, H. H. & Thompson, E. B. (1969) Science 166,1474-1480.

10. Greengard, 0. (1970) in Biochemical Actions ofHormones, ed.Litwack, G. (Academic, New York), Vol. 1, pp. 53-87.

11. Franz, J. M. & Knox, W. E. (1967) Biochemistry 6, 3464-3471.

12. Gluecksohn-Waelsch, S. (1979) Cell 16, 225-237.13. Cori, C. F., Gluecksohn-Waelsch, S., Klinger, H. P., Pick, C.,

Schlagman, S. L., Teicher, L. S. & Wang-Chang, H.-F. (1981)Proc. Natl. Acad. Sci. USA 78, 479-483.

14. Schmid, W., Scherer, G., Danesch, U., Zentgraf, H., Mat-thias, P., Strange, C. M., Rowekamp, W. & Schutz, G. (1982)EMBO J. 1, 1287-1293.

15. Danesch, U., Hashimoto, S., Renkawitz, R. & Schutz, G.(1983) J. Biol. Chem. 258, 4750-4753.

16. Scherer, G., Schmid, W., Strange, C. M., Rowekamp, W. &Schutz, G. (1982) Proc. Natl. Acad. Sci. USA 79, 7205-7208.

17. Scheidereit, C., Geisse, S., Westphal, H. M. & Beato, M.(1983) Nature (London) 304, 749-752.

18. Karn, J., Brenner, S., Barnett, L. & Cesareni, G. (1980) Proc.Natl. Acad. Sci. USA 77, 5172-5176.

19. Frischauf, A. M., Lehrach, H., Poustka, A. & Murray, N.(1983) J. Mol. Biol. 170, 827-842.

20. Benton, W. D. & Davis, R. W. (1977) Science 196, 180-182.21. Vieira, J. & Messing, J. (1982) Gene 19, 259-268.22. Frischauf, A. M., Garoff, H. & Lehrach, H. (1980) Nucleic

Acids Res. 8, 5541-5549.23. Maxam, A. M. & Gilbert, W. (1980) Methods Enzymol. 65,

499-560.24. Bunemann, H., Westhoff, P. & Herrmann, R. G. (1982) Nucle-

ic Acids Res. 10, 7163-7180.25. Breathnach, R. & Chambon, P. (1981) Annu. Rev. Biochem.

50, 349-383.26. Benoist, C., O'Hare, K., Breathnach, R. & Chambon, P.

(1980) Nucleic Acids Res. 8, 127-142.27. Weaver, R. F. & Weissmann, C. (1979) Nucleic Acids Res. 5,

1175-1193.28. Tomizawa, J., Ohmori, H. & Bird, R. (1977) Proc. Nail. Acad.

Sci. USA 74, 1865-1869.29. Sargent, T. D., Wu, J.-R., Sala-Trepat, J. M., Wallace, R. B.,

Reyes, A. A. & Bonner, J. (1979) Proc. Natl. Acad. Sci. USA76, 3256-3260.

30. Grez, M., Land, H., Giesecke, K., Schutz, G., Jung, A. &Sippel, A. E. (1981) Cell 25, 743-752.

31. Malek, L. T., Eschenfeldt, W. E., Munns, T. W. & Rhoads,R. E. (1981) Nucleic Acids Res. 9, 1657-1673.

32. Mount, L. M. (1982) Nucleic Acids Res. 10, 459-472.33. Cochet, M., Chang, A. C. Y. & Cohen, S. N. (1982) Nature

(London) 297, 335-339.34. Cooke, N. E. & Baxter, J. D. (1982) Nature (London) 297,

603-606.35. Maurer, R. A. (1981) Nature (London) 294, 94-97.36. Murdoch, G. H., Rosenfeld, M. G. & Evans, R. M. (1982) Sci-

ence 218, 1315-1317.

Proc. NatL Acad ScL USA 81 (1984)