configuration of DNA - Europe PMC

Proc. Nati. Acad. Sci. USA Vol. 80, pp. 1862-1866, April 1983 Biochemistry

5S rRNA gene transcription factor HIlA alters the helical configuration of DNA (promoter sequences/linking number/DNA unwinding)

WANDA F. REYNOLDS AND JOEL M. GOTTESFELD Division of Cellular Biology, Research Institute of Scripps Clinic, La Jolla, California 92037

Communicated by James Bonner, January 3, 1983

ABSTRACT Relaxation of Xenopus 5S plasmid DNA (pXlo8) in the presence of transcription factor (TF) IIIA reduces the linking number of the DNA. Parallel experiments with plasmid pMB9 or cloned hepatitis B viral DNA indicate a degree of nonspecific unwinding by TF; however, 60% of the effect observed for pXlo8 is due to specific interaction of TF IIIA with the 5S rRNA gene internal promoter sequence. The extent of unwinding (0.2-0.4 helical turn per TF IIIA binding site) is not consistent with the complete denaturation of the 50-base-pair TF binding site; however, it is consistent with a change in helix rotation, denaturation of 2-4 nucleotides per binding site, or DNA wrapping about a protein core. We show that proteins other than TF IIIA (bovine serum albumin and RNase) have no effect on the linkdng number of DNA when present during relaxation and that the unwinding activity associated with TF is heat labile. These results suggest that TF IIIA may facilitate transcription by altering the helical configuration of 5S DNA.

gle-stranded nucleic acid. Because 5S RNA is analogous to the noncoding strand of the gene, this suggests that TF may act as a helix-destabilizing protein through preferential binding to one DNA strand. There are numerous reports of helix-destabilizing proteins that are involved in the processes of DNA replication or transcription (10, 11). In most cases, these proteins do not interact with specific DNA sequences but instead have a general affinity for single-stranded DNA. Recently Escherichia coli catabolite gene-activator protein (CAP), which binds near the promoter sequences of several operons, has been postulated to facilitate transcription by converting right-handed B-type DNA to the left-handed form (12). The predicted unwinding activity was not observed, however, in studies utilizing purified CAP and plasmid DNA containing the lac promoter (13). The 5S rRNA genes provide a developmentally regulated eukaryotic system for which a TF and its binding site have been identified. Moreover, the interaction of the TF with 5S RNA supplies a clue as to its mode of action. Therefore, we undertook experiments to determine whether, as a possible means of transcriptional activation, TF IIIA unwinds 5S DNA.

The 5S ribosomal RNA genes of Xenopus provide a well-defined system for investigating the protein-mediated activation of a eukaryotic gene. There are two major classes of 5S rRNA genes: the somatic type, present in 400 copies per haploid genome, and the oocyte type, present in 20,000 copies per haploid genome (for review and references, see ref. 1). Although the somatic-type genes are transcribed in most cell types, including oocytes, the oocyte-type genes are active only during oogenesis and early embryogenesis. Both are transcribed by RNA polymerase III to give rise to a 120-nucleotide-long transcript that does not undergo sequence excision or post-transcriptional modification. The 5S rRNA genes in oocyte chromatin are transcribed accurately by purified RNA polymerase III, whereas 5S DNA is not (2). However, 5S DNA will be correctly transcribed upon injection into Xenopus oocytes (3) or by using cell-free extracts of oocytes or somatic cells (4, 5). Such in vitro transcription systems have allowed the identification of certain regulatory components. By constructing cloned deletion mutants of 5S rRNA genes, Brown and co-workers (6, 7) defined a region within the 120-base-pair (bp) gene (from residues +50 to +83) that is essential for accurate initiation of transcription. A 37,000dalton protein, designated transcription factor (TF) IIIA, was shown by Engelke et al. (8) to bind this central promoter region (from residues +45 to +96) and to be required for 5S rRNA transcription. During oocyte maturation, TF associates with 5S RNA to form a 7S particle (9). This is thought to be a self-regulating device by which increasing concentrations of 5S RNA reduce the amount of TF available to bind the gene. The association of TF and 5S RNA indicates an affinity of the protein for sin-

MATERIALS AND METHODS Isolation of TF IIIA. The protein, isolated by the procedure of Engelke et al. (8), was judged by NaDodSO4/acrylamide gel electrophoresis to be 30% pure. Activity was monitored by performing in vitro transcription assays with egg extracts supplemented with the TF preparation (8). Preparation of Plasmid DNA. pX1o8 consists of four repeating units of Xenopus oocyte-type 5S DNA inserted into the HindIII site of pMB9. This plasmid was provided by D. D. Brown (Carnegie Inst. of Washington, Baltimore, MD). Cloned hepatitis B viral (HBV) DNA inserted into pBR322 was a gift of F. Chisari (Scripps Clinic and Research Foundation). All plasmid DNAs were isolated by cesium chloride density gradient centrifugation. DNase I Protection Patterns. DNA fragments with a label at only one end of one strand were prepared by standard DNA sequence assay procedures (14). Plasmid pXloA3' +56 contains one copy of the oocyte-type 5S RNA gene, an intact A+T-rich spacer, and 5' flanking sequences but only 56 bp of 3' flanking sequence and no pseudogene sequence (15). Plasmid DNA was digested with EcoRI and end-labeled with [a-32P]dATP and dTTP and DNA polymerase I (16). The 3' end of the noncoding strand is labeled by this procedure. Alternatively, the 5' end of the coding strand was labeled with [y-32P]ATP and polynucleotide kinase after dephosphorylation with calf alkaline phosphatase (14). After secondary restriction with HindIII, labeled DNA fragments were recovered from

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

Abbreviations: TF, transcription factor; CAP, catabolite gene-activator protein; bp, base pair(s); HBV, hepatitis B virus.

1862

Biochemistry: Reynolds and Gottesfeld polyacrylamide gels (14). DNA was digested with DNase I in the presence and absence of TF IIIA as described by Engelke et al (8). Reactions were stopped after 90 sec at 220C with 0.5 vol of 1% NaDodSO4/45 mM EDTA, pH 8.0. After heating at 950C for 2-5 min, the, samples' were extracted with phenol, and the DNA was precipitated with ethanol. Dried. samples were redissolved in 95% formamide/10 mM NaOH, heated to 950C for 2 min, and subjected to electrophoresis in 8% polyacrylamide/8.3 M urea gels (17). Autoradiographic exposures were at -70'C with DuPont Cronex Lightning-Plus intensifying screens and Kodak XAR5 film. Restriction digests of the same' DNA fragments -served as length markers. Assay for Unwinding of DNA by TF I11A. Wheat germ topoisomerase I was obtained from Biotec (Madison, WI). One unit of the enzyme is sufficient to completely relax 1 Atg of supercoiled DNA in 10 min at 370C. Relaxation was carried out in 50 mM Tris HCI, pH 7.9/50 mM NaCl/1 mM EDTA/ 1 mM dithiothreitol/20% glycerol in a reaction volume of 50 /d. Relaxed DNA (1.5 tkg) was incubated with TF IIIA (0.51.5 jug) for 60 min at 21'C in 15 mM Tris HCI, pH 7.5/70 mM KCl/1 mM MgCl2/0.2 mM EDTA/1 mM dithiothreitol/20% glycerol. Topoisomerase (1.5.unit/fug of DNA}was then added, and the mixture was incubated ati370C for 45 min. The re-action was terminated by the addition of NaDodSO4 (2%) and the application of heat at 90'C for 5 min. Proteinase K (100 tug/ml) was added, and the mixture was incubated at 370C for 15 min, followed by phenol extraction and ethanol precipitation. The DNA was electrophoresed in 1% agarose gels in 40 mM Tris acetate, pH 7.9/20 mM sodium acetate/5 mM magnesium acetate/1 mM EDTA. Electrophoresis was carried out at 40C for 24-36 hr at 70 V. The gels were stained with ethidium bromide and photographed with UV illumination. Photographic negatives (Polaroid 655) were scanned with a Kontes densitometer. The area under the trace was excised, and the center of peak distribution was determined by weight. RESULTS We present evidence that 5S plasmid DNA relaxed in the presence of TF IIIA has a lower linking number than DNA relaxed in its absence. To measure unwinding activity, we used a sensitive electrophoretic assay that quantitates the change in linking number due to TF binding at the time of DNA relaxation (18). Relaxation of closed circular DNA with topoisomerase I produces a family of DNA molecules having zero or few superhelical turns because of the Boltzmann distribution in linking number in an equilibrium population of molecules. If a protein alters the helical twist Tw or the writhing number Wr of. the DNA at the time of relaxation, this will result in a change in the linking number L according to the relationship L = Tw + Wr. After inactivation of the topoisomerase and removal of the protein, this change in linking number can be determined by gel electrophoresis. TF IIIA was isolated by the method of Engelke et aL (8) to 30% purity (Fig. 1A). RNA polymerases, in this procedure, do not copurify with TF IIIA. This TF preparation was shown to supplement extracts of Xenopus eggs, which lack significant amounts of TF IIIA, such that in vitro transcription of 5S DNA was possible (not shown; see figure 2 of ref. 8). TF IIIA protected a region within the oocyte-type 5S RNA gene (residues 45-90) from DNase I digestion (Fig. 1B), as has been reported by Engelke et al (8). The regions of protection were not precisely the same on both strands: protection extended from nucleotides +45 to +84 on the noncoding strand and from nucleotides +48 to +90 on the coding strand. Furthermore, enhanced cleavages by DNase were caused by the TF

Proc. Nati Acad. Sci. USA 80 (1983)

A

1863

B 1

2. 3 4 M-F

.....

11.-MOF

L.

8-

TF--

31-j 41-

...

fh1 66-

-

105- * j

FIG. 1. Isolation of TF IIIA. (A) NaDodSO4/polyacrylamide gel electrophoresis of TF IIIA isolated by the method of Engelke et al. (8). (B) DNase protection assay of TF IIIA binding to 5S DNA. End-labeled DNA fiagments were prepared and digested either in the presence (lanes 1 and 3) or absence (lanes 2 and 4) of TF IIIA. TY IIIA (100-200 ng) was complexed with an equivalent amount of end-labeled DNA and digested with 10 ng of DNase I in a 40-1.l reaction volume. Results for DNA fragments labeled on the noncoding (lanes 1 and 2) and coding strands (lanes 3 and 4) are shown. Products of DNase digestion were analyzed on denaturing polyacrylamide gels; an autoradiogram is shown. Nucleotide positions within the 5S rRNA gene are given by the restriction digests of the same labeled DNA fragments.

but were found at different locations on the two strands. An enhanced cleavage was found at residue +45 on the coding strand, whereas enhanced sites were found at residues +54, +66, and +88 on the noncoding strand. Thus, TF IIIA interacted with 5S DNA in an asymetric manner. In the experiments described below, we used a plasmid, pXlo8, containing four tandem repeat units of the oocyte-type 5S rRNA gene. Each repeat unit contains one copy of the gene and one copy of a pseudogene sequence (1). The pseudogene consists of the first 101 bp of the gene, including the TF IIIA binding site. Form I supercoiled pXlo8 was relaxed by incubation with wheat germ topoisomerase I. The relaxed plasmid (Fig. 2, lane 1) was resolved by agarose gel electrophoresis into a series of bands whose densities followed a Gaussian distribution. Each band represents a closed circular DNA molecule that differs from the adjacent band by one superhelical turn (19). Relaxed pX1o8 was incubated with increasing amounts of TF IIIA. After incubation, the DNA was treated with a second round of topoisomerase I. This resulted (Fig. 2) in an increase in the superhelical density of the plasmid. Densitometric scans (Fig. 3) showed that the center of the band distribution (arrows) shifts by as much as 4.0 superhelical turns relative to DNA relaxed in the absence of TF; Lane 3 of Fig. 2 represents a saturating level of TF IIIA (0.6 ug of TF per

1864

Proc. Nad Acad. Sci. USA 80 (1983)

Biochemistry: Reynolds and Gottesfeld

12 3

4

5

6

A 1

2

B 3

1

4

2 I.-

.9

FIG. 2. Introduction of superhelical turns in pXlo8 relaxed in the presence of TF IIIA. Relaxed pXlo8 DNA (1.5 jig) was mixed with 0 (lane 1), 0.3{lane 2), and0O.6 (lane 3) j~g of TF IIIA and incubated with topoisomerase 1 (3 units) for 45 min at 370C in a reaction volume of 50 Relaxed pXlo8 (1.5 ug) was mixed with bovine serum albumin at j~4. 100 pg/ml (lane 4), RNase at 10 jig/ml (lane 5), and TF HIA (0.3 jig) that had been heated at 900C for 10 min (lane 6), followed by topoisomerase 1 (3 units) for 45 min at 3700. The positions of -nicked, closed circular DNA (form 1I), and linear DNA (form III) are indicated. 1.0 ag of DNA). This corresponds to the relative concentrations of TF IIIA and DNA required to give a DNase I protection pattern. When the TF preparation was heated to 900C for 10 min prior to incubation with pXlo8 (Fig. 2, lane 6), there was no shift in superhelical density. The presence of bovine serum albumin (100 /ig/ml) or RNase (10 gg/ml) during DNA relaxation produced no detectable shift in banding pattern (Fig. 2, lanes 4 and 5). Identical experiments were performed with pMB9, the vector sequence present in pXlo8. When relaxed pMB9 DNA was incubated with TF IIIA and then treated with a second round Of topoisomerase, a small increase in superhelical density (0.8 superhelical turn) was observed (Fig. 4A, lanes 1 and 2). A second control plasmid, containing HBV DNA sequences in-

FIG. 3. Shift in the center of topoisomer band distribution as determined by densitometry: densitometric scans of Fig. 2, showing pXlo8 relaxed in the-presence (A)-and absence (B) of TF IIIA. The arrows in-dicate the median of the topoisomer distribution. The peak corresponding to linear (form III) DNA inA andB was deleted where signified by a broken line. The position of form II nicked circles is indicated by an arrow.

FIG. 4. Gel electrophoresis of plasmid DNA relaxed in the presence of TF IIIA. (A) pMB9 and cloned HBV DNA werenrelaxed in the presence of TF IIIA. Relaxed pMB9 (1.5 jtg) (lanes 1 and 2) or 1.5 pg of cloned HBV (lanes 3 and 4) was mixed with 0 (lanes 1 and 3) or 0.6 (lanes 2 and4) pig of TF IIIA, followed by topoisomerase I (3 units) at 3700 for 45 min. (B) Negative superhelical turns are introduced in the presence of TF HIlA and topoisomerase I. Relaxed pXlo8 (1.5 pug) was mixed with 0 (lane 1) or 0.15 (lane 2) ,gg of TF HIlA and incubated with topoisomerase 1 (3 units) in the presence of EDTA (5 mM) for 45 min at 37TC. Arrows indicate the median of the band distribution as determined by densitometry.

serted into pBR322, was chosen for its similarity in size to pXlo8. The presence of TF during relaxation of cloned HBV DNA resulted in a shift of 1.5 superhelical turns (Fig. 4A, lanes 3 and 4). Nonspecific binding by TF resulted in a reduction in linking number of 0.15-0.2 turn per kb for pMB9 (5.3 kb) or cloned HBV DNA (7.5 kb). This compares to a reduction of 0.5 turn per kb for the 5S DNA-containing plasmid, pXlo8 (8.2 kb). We have noted that the observed increase in superhelicity represents negative, rather than positive, supercoils as is consistent with an unwinding activity. By unwinding activity, we mean any alteration of the DNA at the time of relaxation that leads to a reduction in the linking number. If DNA is relaxed in the absence of Mg2" and subsequently is electrophoresed in low-temperature Mg2+-containing gels, the degree of helix rotation is increased, causing the relaxed DNA to acquire several negative superhelical turns. The Gaussian distribution of topoisomers is then well-resolved from the nicked position, centering at approximately five negative superhelical turns. Any increase in superhelicity due to the presence of TF during relaxation then represents additional negative supercoils, whereas a decrease in superhelicity would indicate positive supercoils. Using low concentrations of TFJIIIA, so as to elicit ashluIt of only 1-2 superhelical turns, we found an increase in superhelicity, indicating negative supercoils (Fig. 4B). DISCUSSION Our results indicate that TF HIlA -alters the helical configuration of 5S DNA in a manner consistent with unwinding activity. Relaxation of 5S plasmid DNA in the presence of TF causes the DNA to become more negatively supercoiled. Although the observed effect is largely specific for 5S DNA sequences, there is a significant degree of nonspecific unwinding. We showed that proteins other than TF, including bovine serum albumin and RNase, have no effect on the linking number of DNA when present during relaxation and that the unwinding activity associated with TF is heat labile. pXlo8, relaxed in the presence of TF IIIA, gained an average of four negative superhelical turns, indicating a factorinduced unwinding of four helical turns. Parallel experiments

Biochemistry: Reynolds and Gottesfeld with non-55-DNA indicate a certain amount of nonspecific interaction by TF IIIA:pMB9 (5.3 kb) and cloned HBV DNA (7.5 kb) gained an average of 0.8 and 1.5 negative superhelical turn(s), respectively. This corresponds to a nonspecific unwinding by TF IIIA of 0.15-0.2 helical turn per kb of DNA. Because pXlo8 contains 8&2 kb, we estimate a nonspecific unwinding of 1.6 helical turns, which represents 40% of the observed effect. Discounting this nonspecific contribution, the remaining unwinding effect of 2.4 helical turns may be attributed to unwinding of TF binding sequences. pX1o8 contains eight TF binding sites, each consisting of 50 bp, producing a total of 400 bp. This indicates a sequence-specific unwinding of six helical turns per kb, 30-fold greater than the nonspecific effect. Although the affinity of TF for its binding sequence is much greater than for non-5S-DNA, the higher proportion (95%) of nonspecific sequences in pXlo8 produces a significant background unwinding. pX1o8 contains four copies of the oocyte-type 5S rRNA gene repeat unit. Each repeating unit contains, in addition to the 5S rRNA gene, a pseudogene sequence consisting of a direct repeat of 101 bp of the gene with 85% homology (1). It has been demonstrated by Wormington et aL (20) that the pseudogene competes equally with the authentic gene for TF binding; therefore, we assume eight TF binding sites in pX1o8. The available binding sites are apparently saturated at a TF concentration between 0.3 and 0.6 Ag/lAg of pX1o8 (Fig. 2). Although the average increase in superhelical density at this point is four turns, the band distribution is broad. A small proportion of the- DNA molecules gain as many as seven turns, whereas some undergo no change in superhelicity. This broad banding pattern may indicate a nonhomogeneous population of DNA molecules with various numbers of binding sites occupied. Therefore, the average increase in superhelicity may not represent the full unwinding potential. For this reason, it is difficult to estimate the amount of unwinding per 5S rRNA gene. However, based on a specific unwinding of 2.0-3.0 helical turns for pXlo8 and eight binding sites with full occupancy, we obtain a minimum estimate of 0.2-0.4 helical turn unwound per binding site. We cannot exclude unwinding by proteins other than TF IIIA in our preparation. However, because the observed effect is largely specific for 5S DNA sequences, this does not seem likely. It remains a possibility that other factors that copurify with TF IIIA work in conjunction with TF IIIA to unwind 5S DNA. In preliminary experiments, we attempted to determine the relative affinity of TF IIIA for supercoiled and relaxed 5S DNA. Competition filter binding assays in which TF IIIA was incubated with labeled linear pXlo8 in the presence of increasing amounts of unlabeled supercoiled pXlo8 indicate a 2- to 5-fold higher affinity of TF for supercoiled over linear 5S DNA (data not shown). This is consistent with our results because the free energy available in supercoiled DNA aids any unwinding process. The observed increase in negative superhelical turns, which occurs when TF IIIA is present during relaxation, could be brought about in several ways: it could result from melting or separation of DNA strands, by wrapping the DNA about a protein core in a left-handed manner, or by reducing the pitch of the DNA helix. In each of these instances, the net result would be a gain in negative superhelicity. Our results cannot differentiate between these possibilities. However, there is indirect evidence that is supportive of helical unwinding rather than left-handed wrapping. In the maturing oocyte, TF is found complexed with 5S RNA as a 7S storage particle (9), indicating an affinity for single-stranded nucleic acid. Considering that 5S RNA is analogous in sequence to the noncoding strand of

Proc. Nati Acad. Sci. USA 80 (1983)

1865

the 5S rRNA gene, this suggests a preferential binding by TF to that strand, leading to helix destabilization. It is possible that such localized destabilization is sufficient to promote polymerase binding. Furthermore, although TF protects the 50-bp binding site from DNase I digestion, there are certain sites within and flanking this protected region that display enhanced cleavage by the enzyme (8). This may indicate torsional stress at these sites due to factor binding and could represent regions of localized melting. Preliminary experiments with S1 nuclease, which cuts single-stranded regions of DNA, have not provided consistent evidence for DNA melting. Further, attempts to localize unpaired bases by reaction with dimethyl sulfate have not proved successful. Although the possibility remains that certain specific base pairs are broken by TF IIIA, our results appear to be more consistent with a model in which TF interacts with DNA to bring about a change in the winding angle without melting the DNA. In this regard, a transition from the B form (10.4 bases per helix turn; ref. 21) to the A form (=I1-12 bases per turn; ref. 22) for the 50bp TF IIIA binding region would result in a linking number change of -2 to -3 for pXlo8. It is conceivable that such a change in helical repeat is a prerequisite for transcription, as DNA-RNA hybrids are of the A form. The magnitude of the change in linking number that we observe is not consistent with the formation of a 5S RNA-like DNA secondary structure (cruciform) within the 5S rRNA gene. There are a number of proteins involved in DNA replication or transcription that interact nonspecifically with DNA to bring about helix destabilization (for a review, see refs. 10 and 11); however, recent reports concerning sequence-specific DNA binding proteins have failed to demonstrate significant degrees of unwinding. Examples include Escherichia coli CAP and T antigen, which binds specifically to sequences at the origin of replication in Simian virus 40 (23), although in the latter case there is conflicting evidence (24). Our results suggest that a eukaryotic sequence-specific DNA binding protein promotes transcriptional initiation by altering the helical configuration of DNA, either by unwinding the helix or by wrapping in a left-handed manner. Because the 5S rRNA genes are transcribed by RNA polymerase III, their regulatory mechanisms differ somewhat from those of genes transcribed by RNA polymerase II, most notably by the inclusion of a promoter region within the 5S rRNA gene coding sequence. It is nevertheless possible that certain fundamental aspects of transcriptional control, relevant to the 5S system, will be common to other eukaryotic sequences. We wish to thank Drs. D. Bazett-Jones, C. Von Beroldingen, and L. Millstein for many helpful discussions and J. C. Wang for helpful criticisms. W. F. R. is supported by a postdoctoral fellowship from the National Institutes of Health. This work was supported by a grant from the

National Institutes of Health (GM-26453).

1. Korn, L. J. (1982) Nature (London) 295; 101-105. 2. Parker, C. S. & Roeder, R. G. (1977) Proc. NatL Acad. Sci. USA 74, 44-48. 3. Brown, D. D. & Gurdon, J. B. (1978) Proc. Nati Acad. Sci. USA 75, 2849-2853. 4. Birkenmeier, E. H., Brown, D. D. & Jordan, E. (1978) Cell 15, 1077-1086. 5. Weil, P. A., Segall, J., Harris, B., Ng, S. Y. & Roeder, R. G. (1979) J. BioL Chem. 254, 6163-6173. 6. Sakonju, S., Bogenhagen, D. F. & Brown, D. D. (1980) Cell 19, 13-25. 7. Bogenhagen, D. F., Sakonju, S. & Brown, D. D. (1980) Cell 19, 27-35. 8. Engelke, D. R., Ng, S.-Y., Shastry, B. S. & Roeder, R. G. (1980) Cell 19, 717-728.

1866

Biochemistry: Reynolds and Gottesfeld

9. Pelham, H. R. B. & Brown, D. D. (1980) Proc. Natl Acad. Sci. USA 77, 4170-4174. 10. Champoux, J. J. (1978) Annu. Rev. 'Biochem. 47, 449-479. 11. Geider, K. & Hoffnann-Berling, H. (1981) Annu. Rev. Biochem. 50, 233-260. 12. McKay, D. B.. & Steitz, T. A. (1981) Nature (London) 290, 744749. 13. Kolb, A. .& Buc, H. (1982) Nucleic Acids Res. 10, 473-485. 14. Maxam, A. & Gilbert, W. (1980)-Methods EnzymooL 65, 499-560. 15. Wormington, W. M., Bogenhagen, D. F., Jordan, E. & Brown,

D. D. (1981) Cell 24, 809-817.

16. -Smith, S.. & Thomas, C. A. (1981) Gene 13, 395-408.

Proc. Natd Acad. Sci. USA 80 (1983) 17. Sanger, F. & Coulson, A. R. (1978) FEBS Lett. 87, 107-110. 18. Wang, J. C., Jacobsen, J. H. & Saucier, J.-M. (1977) Nucleic Acids Res. 4, 1225-1241. 19. Keller, W. (1975) Proc. NatL Acad. Sci. USA 72, 4876-4880. 20. Wormingten, W. M., Bogenhagen, D. F., Jordan, Es & ,Brown, D. D. (1981) Cell 24, 809-817. 21. Peck, L.-J. & Wang, J. C. (1981) Nature (London) 292, 375-378. .22. Arnott, S.. & Hukins, D. W. L. (1972) .Biochem. Biophys. Res. Commun. 47, 1504-.1509. 23. Myers, R. M., Kligman, .M.& Tjian, R. (1981) J. BioL Chem. 256, 10156-10160. 24. Giacherio, D. & Hager, L. P. (1980)J. BioL Chem.-255, 8963-8966.