Proc. NatI. Acad. Sci. USAVol. 84, pp. 4841-4845, July 1987Biochemistry

Metal-dependent folding of a single zinc finger from transcriptionfactor I11AALAN D. FRANKEL*, JEREMY M. BERGt, AND CARL 0. PABO*t*Department of Molecular Biology and Genetics, Johns Hopkins University School of Medicine, Baltimore, MD 21205; tDepartment of Chemistry, JohnsHopkins University, Baltimore, MD 21218; and tHoward Hughes Medical Institute, Johns Hopkins University School of Medicine, Baltimore, MD 21205

Communicated by Hamilton 0. Smith, March 30, 1987 (received for review February 20, 1987)

ABSTRACT A 30-amino acid peptide, which correspondsto the second "zinc finger" domain of transcription factorIIIA, has been synthesized and purified. This peptide folds inthe presence of zinc: adding Zn2+ significantly changes thecircular dichroism spectrum, and Zn + protects the peptidefrom tryptic digestion. The peptide also binds Co2+, and theabsorption spectrum of the Co21 complex suggests that atetrahedral binding site is formed by two cysteines and twohistidines. Experiments at higher temperatures (60-750C)suggest that these folded metal-peptide complexes are quitethermostable. The peptide shows some sequence-specific effectsin DNase and methylation protection experiments. However, itdoes not give a clear "footprint," and some effects are observedin the absence of added zinc.

Zinc binding domains appear to be important structuralelements in many nucleic acid binding proteins (1-8). Suchdomains were first noted in Xenopus transcription factor IIIA(TFIIIA) (1, 2). This protein contains nine tandem repeats ofa 30-amino acid motif, and each repeat has two invariantcysteines and two invariant histidines. When isolated,TFIIIA contains Zn2+ (8), with 7-11 ions in the ribonucleo-protein complex containing TFIIIA (1). Based on theseobservations, it was suggested that each repeat was a Zn2+binding domain (a "zinc finger") and that hydrophilic sidechains from these domains could make specific contacts withnucleic acids (1, 2). Sequences containing similar patterns ofcysteines and histidines have been found in a wide variety ofproteins involved in nucleic acid binding or gene regulation(3-7).There is substantial evidence that the nine domains in

TFIIIA are independent units and are stabilized by Zn2+.Partial proteolysis of TFIIIA yields fragments that differ inmolecular size by 3 kDa (1), a size difference consistent withthe removal of successive 30-amino acid domains. Thesequence of the TFIIIA gene shows that most of thesedomains are encoded by separate exons (9), suggesting thatthese units have structural significance (10). X-ray absorptionspectra of the ribonucleoprotein complex show that thecoordination sphere of the Zn2+ ions contains two cysteineand two histidine residues (11). Removal of Zn2+ from theprotein destroys site-specific DNA binding activity (8). Tosimplify physical and structural characterization ofthese zincdomains, we have prepared a peptide corresponding to onezinc finger from TFIIIA. This peptide folds in the presence ofZn2+. The peptide also binds DNA with some sequencespecificity, but binding shows no clear dependence on Zn2+.

MATERIALS AND METHODS

Preparation of DNA. Plasmid DNAs [ptrpLE' (16),pTFIIIA-2 (see below), and pXBS201 (19)] were purified

from Escherichia coli MM294 cells by the rapid alkalineextraction procedure of Birnboim and Doly (12) and equilib-rium centrifugation in CsCl/ethidium bromide gradients (13).The gene encoding TFIIIA-2 was synthesized in four frag-ments on an Applied Biosystems (Foster City, CA) model380A DNA synthesizer. These oligonucleotides, which were39-60 bases long, were purified on 15% polyacrylamide/ureagels.

Construction of the pTFIIIA-2 Plasmid. The TFIIIA-2 genewas constructed by phosphorylating two of the syntheticoligonucleotides, annealing with the other two oligonucleo-tides, and ligating. Duplex DNA of the correct size waspurified on a 1% agarose gel. This DNA was phosphorylatedand inserted into EcoRI-cleaved ptrpLE', and the ligationmixture was used to transfect CaCl2-treated MM294 cells(14). Transformants were selected on LB (Luria Broth) agarplates containing ampicillin at 100 ,ug/ml. Plasmid DNA wasprepared from several colonies, and the presence ofTFIIIA-2gene was confirmed by restriction endonuclease analysis.Several plasmids containing an insert of the proper size weresequenced by the method ofMaxam and Gilbert (15), and theplasmid containing the TFIIIA-2 gene was named pTFIIIA-2(Fig. 1). Phosphorylation reactions were performed using T4polynucleotide kinase, and ligation reactions were carried outat 14WC using T4 DNA ligase. Enzymes were purchased fromNew England Biolabs, Pharmacia P-L Biochemicals, andBoehringer Mannheim.

Purification of TFIIIA-2. TFIIIA-2 was purified by growingE. coli MM294 cells containing the pTFIIIA-2 plasmid in M9medium containing ampicillin at 100 ptg/ml and tryptophan at100 Ag/ml. During late logarithmic phase, cells were centri-fuged, resuspended in medium lacking tryptophan, andgrown for 3-12 hr more at 370C. Removal of tryptophaninduces production of the fusion protein (Fig. 1), which canbe seen as granules by phase-contrast microscopy (16).

Cells were harvested and lysed, and the granules wereisolated by centrifugation as described (16). The crudeprotein pellet from an 8-liter culture was treated with 100 mgofcyanogen bromide in 40 ml of70% (vol/vol) formic acid for24 hr at room temperature (17) and lyophilized. To identifyTFIIIA-2, cyanogen bromide-treated samples from bothpTFIIIA-2 and ptrpLE' preparations were electrophoresedon a native 15% polyacrylamide gel (30 mM sodium acetate,pH 4.5) and stained with Coomassie blue. An additional bandwas observed in the pTFIIIA-2 preparation. To purifyTFIIIA-2, dried, cyanogen bromide-treated samples wereresuspended in water and centrifuged to remove insolublematerial. This crude mixture was loaded onto a Vydac C4HPLC column in 0.1% trifluoroacetic acid and eluted with anacetonitrile gradient (0-30%). The absorbance was moni-tored at 214 nm, peaks were collected, and TFIIIA-2 wasidentified by native gel electrophoresis. The purified peptidewas sequenced using an Applied Biosystems automated

Abbreviations: TFIIIA, transcription factor IIIA; DTNB, 5,5'-dithiobis(2-nitrobenzoic acid).

4841

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.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

0

Proc. Natl. Acad. Sci. USA 84 (1987)

trp P (TFIIIA-2k gene

FIG. 1. Plasmid encoding the TFIIIA-2 peptide. The gene en-coding TFIIIA-2 was inserted into the EcoRI site of ptrpLE'.Production of the resulting LE'-TFIIIA-2 fusion protein is controlledby the tryptophan promoter and operator. A methionine placed at theprotein-peptide junction allows TFIIIA-2 to be released with cyan-ogen bromide,. This plasmid confers ampicillin resistance.

peptide sequencer, and its concentration was determined byamino acid analysis. Approximately 1 mg of purified peptidewas pbtained per liter of cell culture.The number of free thiols in TFIIIA-2 was determined by

reaction with 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB)(18). Reactions were carried out in 100mM Tris HCl (pH 8.0)with DTNB at 0.2 mg/ml. When necessary, TFIIIA-2 wasseparated from excess dithiothreitol by chromatography on aPharmacia Superose 12 column in 1 mM acetic acid/50 mMNaCl before adding DTNB and buffer.

Tryptic Cleavage of TFIIIA. Proteolysis of TFIIIA-2 wascarried out at room temperature in 20 mM Tris IHCl (pH 7.0)with various ZnCl2 concentrations. Each sample contained2.5 ug of peptide. After equilibrating the peptide with bufferand ZnCl2 for 30 min, 20 ng of trypsin was added, andproteolysis proceeded for 3 min. Reactions were stopped byadding 10 ,g of phenylmethylsulfonyl fluoride, and thesamples were electrophoresed on a 20% polyacrylamide gelcontaining 30 mM Tris HCl (pH 7.0) at 12 V/cm for 3 hr.Reduced TFIIIA-2 was prepared by treating the peptide with10 mM dithiothreitol at 90°C for 30 min and lyophilizing. Intryptic digestions with oxidized TFIIIA-2, an equivalentamount of dithiothreitol was added to the reaction mixturesat the time of trypsin addition.

Circular Dichroism of TFIIIA-2. Circular dichroism spectrawere measured using an Aviv model 6ODS spectropolari-meter. Spectra were recorded from 300 nm to 200 nm andaveraged over five scans. Samples were in 10 mM Tris HCl(pH 7.0, degassed with helium), and peptide concentrationsvaried from 20 ,ug/ml to 100 Ag/ml. Temperature wascontrolled using a water-jacketed cuvette holder and anexternal water bath. Reduced TFIIIA-2 was prepared bytreating the peptide with 250 mM dithiothreitol at 909C for 30min. The reduced peptide was purified by HPLC, lyophil-ized, and stored at -80°C.

Absorption Spectra with Cobalt. Absorption spectra weremeasured on a Bausch & Lomb Spectronic 2000 spectropho-tometer and on a Perkin-Elmer Lambda 3A spectrophotom-eter. Temperature was controlled using a water-jacketedcuvette holder and an external water bath. Reduced TFIIIA-2was prepared as described in the previous section. Lyophil-ized peptide was resuspended in 10 mM Tris HCl (pH 7.0,degassed with helium), and CoCl2 was added as indicated inthe figure legend. For experiments at 60°C, the buffer was

titrated to compensate for the temperature dependence ofTris ionization.DNase I and Dimethyl Sulfate Protection Experiments.

DNase I digestions were done at room temperature in 20 mMTris HCl, pH 7.0/20 mM NaCl/10 mM MgCl2/1 mM ZnCl2with various concentrations of TFIIIA-2. After equilibratingfor 30 min, 100 ng of 32P-labeled pXBS201 DNA was added,and, after another 30 min, 3 ng of DNase I was added. The

reaction was stopped after 1 min by adding 1 ml of 0.3 Mammonium acetate, 5 gg of tRNA, 70% (vol/vol) ethanol.DNA was ethanol precipitated twice, resuspended in loadingbuffer (15), and electrophoresed on an 8% polyacrylamide/urea sequencing gel.Dimethyl sulfate protection reactions were carried out at

room temperature in the same buffer as the DNase I reac-tions, except that MgC12 was not included. After equilibratingin buffer for 30 min and then with 100 gg of 32P-labeledpXBS201 DNA for another 30 min, 1 gl of 2% (vol/vol)dimethyl sulfate was added and allowed to react for 2 min.The stop mixture (15) was added, and the DNA was treatedto cleave at methylated guanines (15) and electrophoresed.Reduced'and oxidized samples of TFIIIA-2 were prepared

as for the partial proteolysis experiments 32P-labeledpXBS201 (label on the coding strand of the 5S rRNA gene)was prepared as described by Smith et al. (19).

RESULTSProduction and Purification of TFIIIA-2 Peptide. We pro-

duced a single zinc finger by synthesizing its gene, expressingit in E. coli as part of a fusion protein, and releasing thepeptide by cyanogen bromide cleavage. This peptide, namedTFIIIA-2, corresponds to the second of the nine repeats inTFIIIA. It contains residues 42-71 from the intact protein andhas the sequence:

Pro-Phe-Pro-y-Lys-Glu-Glu-Gly-y-Glu-Lys-Gly-Phe-Thr-Ser-Leu-His-His-Leu-Thr-Arg-His-Ser-Leu-Thr-His-Thr-Gly-Glu-Lys,

where underscores indicate conserved residues. This seconddomain contains all the conserved residues found when thenine repeats are aligned, including the two cysteines and twohistidines thought to be involved in coordinating the zinc ion,two highly conserved aromatic residues, and one highlyconserved leucine.We constructed a plasmid that would encode the TFIIIA-2

peptide fused to the carboxyl terminus of the trp LE' protein.The LE' protein is produced at very high levels in E. coli andhas been used in a similar way to make other peptides (16, 20).The plasmid pTFIIIA-2 is sketched in Fig. 1. The codon formethionine was placed at the protein-peptide junction so thatthe peptide could be released with cyanogen bromide. Allother codons were chosen on the basis of their preferentialusage in E. coli (21). This plasmid was used to produce thedesired peptide, which was cleaved from the fusion proteinand purified.The peptide, as initially isolated, contained an internal

disulfide bond. Treatment with DTNB indicated there wereno free thiols in the purified peptide (18). A reduced form wasgenerated by treating the peptide with 250 mM dithiothreitolat 90°C for 30 min. A Pharmacia Superose 12 column was thenused to separate the peptide from dithiothreitol. Collectingthe peptide peak and immediately adding DTNB revealedthat there now were 2,0 mod of free thiol per mor of peptide.The elution times of the oxidized and reduced peptide wereidentical op the Superose 12 sizing colunn, and they also hadidentical mobilities on native 20% polyacrylamide gels. Theseresults suggest that the oxidized peptide -is a monomer withthe two cysteines forming an internal disulfide bond. To testother protocols for reduction of TFIIIA-2, we treated it withdithiothreitol at several concentrations and temperatures andchromatographed the peptide on a C4 reverse-phase HPLCcolumn. Eluting with acetonitrile resolved the reduced andoxidized peptides (Fig. 2). We found that incubating at 90°Cfor 30 min with a 5-fold molar excess of dithiothreitol gave80-90% reduction of the disulfide.

4842 Biochemistry: Frankel et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

0

Proc. Natl. Acad. Sci. USA 84 (1987) 4843

CY)

oxidized

-10 ;.*.,,/' ~A-201IV I

200 210 220 230

X (nm)

200 210 220 230 240X (nm)

FIG. 2. Reduction of TFIIIA-2. As initially isolated, TFIIIA-2(oxidized) contains an internal disulfide bond. Dithiothreitol (DTT)was used as a reducing agent, and the reduced and oxidized specieswere separated by HPLC. Conditions that give partial and completereduction are indicated. The peptide concentration was 0.2 mM inthese experiments.

Stability to Proteolytic Digestion. Since a stable, foldedstructure generally makes a polypeptide less accessible toproteolytic digestion, trypsin was used to monitor folding ofTFIIIA-2. These experiments (Fig. 3) compared trypticcleavage of the reduced and oxidized peptide at various Zn2+concentrations. ZnCl2 clearly inhibits digestion of the re-duced peptide (lanes 6-9). Control experiments revealed thatat least 20-fold higher ZnCl2 concentrations were needed toinhibit proteolysis of the oxidized peptide (lanes 2-5).

Circular Dichroism Spectra. Circular dichroism spectros-copy was also used to monitor peptide folding. As shown inFig. 4A, the circular dichroism spectrum changes markedlywhen zinc is added to reduced TFIIIA-2. In 50 ,gM ZnCl2 (aslight molar excess), there are the following two majorchanges: (i) a significant increase in negative molar ellipticitynear 220 nm and (ii) a dramatic reduction in the negative

1 2 3 4 5 6 7 8 9 10

FIG. 3. Cleavage of TFIIIA-2 by trypsin. Lane 1 of the native gelcontains untreated TFIIIA-2; lane 10 shows cleavage of the reducedpeptide in the absence of ZnC12. Trypsin concentrations were chosento give partial digestion ofTFIIIA-2 in the absence ofZnCl2 (lane 10).Various concentrations of ZnCl2 were added to oxidized (lanes 2-5)and reduced (lanes 6-9) peptide prior to proteolysis. ZnCl2 concen-trations were 10 mM (lanes 2 and 6); 2 mM (lanes 3 and 7); 0.5 mM(lanes 4 and 8); and 0.1 mM (lanes 5 and 9).

FIG. 4. Circular dichroism spectra of TFIIIA-2. (A) Spectra ofreduced TFIIIA-2 with no ZnC12 (---), with 50 ,M ZnCl2 (-) at 26TC,or with 50 ,uM ZnCl2 at 750C (.). (B) Spectra of oxidized TFIIIA-2with no ZnC12 (---), with 50 ,M ZnCl2 (-), and with 5 mM ZnCl2 (.)at 26TC. In all cases, peptide concentrations were 30 AM. Note thatthe pH of the buffer (10 mM Tris) shows a marked temperaturedependence, and pH changes may affect the experiments at highertemperatures.

molar ellipticity near 200 nm. Increasing the concentration ofZnCl2 (up to 5 mM) did not give further changes in thespectrum. Very similar spectra were obtained with CoCl2(data not shown). As a test for aggregation, spectra wererecorded at a series of peptide concentrations (6-30 uM) inthe presence of ZnCl2. No changes in molar ellipticity wereobserved as the peptide was diluted. To analyze the thermalstability of the folded structure, we recorded the circulardichroism spectra over a range oftemperatures. Remarkably,increasing the temperature to 750C only gave small changesin the spectrum (Fig. 4A). The original spectrum was recov-ered when the sample was cooled.The spectrum of the oxidized peptide (Fig. 4B) changes

very little in 50 /iM ZnCl2. Larger changes are seen as moreZnCl2 is added. However, even at the highest concentrationtested (5 mM) the spectral changes with the oxidized peptidewere less pronounced than with the reduced peptide, and thefinal spectrum with the oxidized peptide was significantlydifferent than that obtained with the reduced peptide.

Absorption Spectrum of the Co2'-Peptide Complex. Cobaltwas used to probe the metal binding site in TFIIIA-2. Sincecobalt complexes have distinctive optical absorption spectra,cobalt substitution has been used in many studies of metal-loproteins (22). Fig. 5 shows the spectrum obtained when astoichiometric amount of CoCl2 was added to reducedTFIIIA-2. Adding a molar excess of Co2+ (6.7-fold) did notchange the spectrum. Several absorption bands were ob-served: There is a maximum near 635 nm with an extinctioncoefficient (E) of 400 M -cm- and a shoulder near 575 nm.These absorption bands are responsible for the blue color ofthe sample. The spectrum also has an absorption maximumnear 310 nm with a shoulder at 340 nm. Spectra recorded at60'C were essentially identical to those at room temperature(data not shown). In control experiments, cobalt was addedto the oxidized peptide. No spectral changes were observed,even with a 70-fold excess of CoCl2.

Competition experiments show that zinc readily displacescobalt from the complex. As shown in Fig. 5, adding Zn2+eliminates the absorption at 635 nm and at 575 nm andreduces absorption at 310 nm and at 340 nm. Zinc must bind

0E-5

Ea00

V240 250

C? E0 V

x E_ 0

a)-

250

Biochemistry: Frankel et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

0

Proc. Natl. Acad. Sci. USA 84 (1987)

0.2 0

U1)

coenOn0Uj)

0.1 .

0.0 1300 400 500 600 700 800

X (nm)

FIG. 5. Optical absorption spectra of TFIIIA-2 with Co2+. Solidline shows the spectrum of 147 MM reduced TFIIIA-2 with 150 ALMCoCl2. Additional CoCl2 was added (to 1 mM) with no changes in thespectrum. Dashed line shows the spectrum after 150 MM ZnCl2 wasadded to the sample containing 147 MM reduced peptide and 1 mMCoCl2. The spectra have been corrected by subtracting the spectraof reduced peptide and of CoCl2.

significantly more tightly than cobalt, since this solutioncontained a stoichiometric amount of ZnCl2 and a 6.7-foldexcess of CoCl2.DNA Binding. Does a single zinc finger show sequence-

specific binding? Intact TFIIIA binds to the internal controlregion of the 5S rRNA gene (23-26), and we used DNase- andmethylation-protection experiments to monitor peptide bind-ing to this region. DNase protection experiments with highconcentrations of TFIIIA-2 show sequence-dependent ef-fects (Fig. 6). It is interesting that methylation protectionexperiments also show effects in the same regions (data notshown). However, effects are seen with the oxidized peptideand in the absence of ZnCl2. No conditions gave a clearlydefined "footprint," and these DNA binding experiments aredifficult to interpret.

DISCUSSION

Our results show that a single zinc finger-a peptide con-taining just 30 amino acids-can fold in the presence of zinc.Our first evidence came from proteolytic digestion experi-ments. Adding ZnCl2 protected the reduced peptide fromtryptic digestion. The oxidized peptide (which contains aninternal disulfide bond) also was protected, but at least20-fold higher ZnCl2 concentrations were needed (Fig. 3). Wepresume that the peptide folds in the presence of zinc and that

1 2 34

-.0

S-,

e. .-: qw-

W

+99 FIG. 6. DNase I protection of pXBS201 byTFIIIA-2. Several concentrations of reducedTFIIIA-2 were incubated with DNA beforeadding DNase I. Lanes: 1, no peptide; 2, 1 rnMpeptide; 3, 100 MM peptide; 4, 10 AM peptide.All reaction mixtures contained 1 mM ZnCl2.Note that at the highest TFIIIA-2 concentra-tion (lane 2), overall inhibition of DNase I wasobserved; whereas at the lower peptide con-centrations, the overall level of cleavage wasenhanced. Arrows indicate some of the bandswhere the effect of the peptide appears most

+41 obvious. Brackets and numbers indicate theposition of the TFIIIA binding site in the 5SrRNA internal control region.

the folded conformation is less susceptible to proteolyticcleavage. Zinc also stabilizes the gene 32 protein and protectsit from tryptic digestion (27).

Circular dichroism spectra show that TFIIIA-2 folds in thepresence of zinc. Without zinc, the spectrum of the reducedpeptide resembles that of a random coil; adding stoichiomet-ric amounts of zinc gives a very different spectrum (Fig. 4A).We believe this is a more ordered structure, but the observedchanges do not allow a simple assignment of secondarystructure. (We presume that the spectral changes reflectchanges in the peptide structure, since zinc and cobalt gavesimilar circular dichroism spectra.) Stoichiometric amountsof zinc have relatively little effect on the circular dichroismspectrum of the oxidized peptide (Fig. 4B). At 20- to 100-foldhigher zinc concentrations, there are significant changes inthe spectrum of oxidized TFIIIA-2. These results, along withthe trypsin digestions, suggest that the oxidized peptidebecomes partially structured at high ZnCl2 concentrations.To determine the coordination state ofthe bound metal, we

studied a cobalt-peptide complex. Cobalt can frequently besubstituted for zinc, and the optical absorption spectrum isvery sensitive to the nature of the coordination site. Thespectrum observed for the complex between reducedTFIIIA-2 and a stoichiometric amount of Co2+ (Fig. 5) clearlyshows that the metal binding site is tetrahedral. The inten-sities ofthe d-dtransitions (e.g., E = 400 M-1cm-l at 635 nm)rule out higher coordination numbers (22). The spectrum isentirely consistent with the structure proposed for the zincsites in TFIIIA (1, 2, 11). The charge-transfer bands near 310nm and 340 nm indicate that cysteinate ligands are involved.This spectrum is quite similar to the spectrum for theCo2"-substituted gene 32 protein (27) and to the "bluehybrid" of liver alcohol dehydrogenase (28). Competitionexperiments show that Zn2+ binds to the same site withhigher affinity since it readily displaces Co2+ from thecomplex (Fig. 5).

In contrast to the reduced peptide, no spectral changeswere observed when CoCl2 was added to the oxidized peptide(even with a 70-fold excess of CoCl2). This clearly demon-strates that the metal binding site in the reduced peptide isquite different from any metal binding sites in the oxidizedpeptide. Obviously, since the oxidized peptide has an internaldisulfide bond, the cysteines are not available to act as metalligands. The complex observed with the oxidized peptide athigh ZnCl2 concentrations (in the tryptic digestions andcircular dichroism experiments) must result from binding toa low-affinity site.The folded structure, observed in the presence of metals,

appears to be quite thermostable. Circular dichroism spectraof the zinc-peptide complex were recorded from 4°C to 75°C.There is remarkably little change in the spectrum over thistemperature range, and the small changes observed arereversible. The cobalt-peptide complex also is thermostable.Absorption spectra recorded at 60°C are essentially identicalwith spectra recorded at room temperature.Our DNA binding experiments are difficult to interpret.

The effects seen in DNA protection experiments are subtle(Fig. 6). There never is a clear footprint and changes arenoticed both inside and outside of the control region recog-nized by intact TFIIIA. Some effects are observed even in theabsence of added zinc. However, the results are intriguingbecause the effects are reproducible and because there is aclear relationship between the DNase protection experimentsand the methylation protection experiments (data notshown). It is not surprising that this isolated zinc finger showsvery little specificity in DNA binding. Intact TFIIIA, whichbinds tightly to the internal control region of the 5S rRNAgene, contains nine zinc fingers. Tight binding and site-specific recognition typically require a set of cooperativecontacts. For example, repressors and restriction enzymes

4844 Biochemistry: Frankel et al.

N1%

-

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

0

Proc. Natl. Acad. Sci. USA 84 (1987) 4845

bind as dimers, making extensive contacts along an entireface of the protein (29, 30).As discussed above, many observations have suggested

that zinc fingers are independent structural units. Our resultsdirectly demonstrate that an isolated zinc finger domain canfold in the presence of metal ions. TFIIIA-2 and relatedpeptides should facilitate physical and structural studies ofthe zinc finger motif and its role in nucleic acid recognition.

This research has been supported by Grant GM-31471 from theNational Institutes of Health and by the Howard Hughes MedicalInstitute. We thank Peter Kim for use of his spectropolarimeter,Clark Riley for sequencing the peptide, P. Shenbagamurthi for aminoacid analysis, Tim Whitcombe for technical assistance, and JoeColeman for helpful comments about the circular dichroism exper-iments. The pXBS201 plasmid was kindly provided by Don Brown,and the ptrpLE' plasmid was provided by Cetus Corporation. We aregrateful to Harriet Fowler and Pam Hill for expert typing of thismanuscript.

1. Miller, J., McLachlan, A. D. & Klug, A. (1985) EMBO J. 4,1609-1614.

2. Brown, R. S., Sander, C. & Argos, P. (1985) FEBS Lett. 186,271-274.

3. Berg, J. M. (1986) Science 232, 485-487.4. Rosenberg, U. B., Schrcder, C., Preiss, A., Kienlin, A., C6td,

S., Riede, I. & Jackle, H. (1986) Nature (London) 319,336-339.

5. Hartshorne, T. A., Blumberg, H. & Young, E. T. (1986)Nature (London) 320, 283-287.

6. Vincent, A., Colat, H. V. & Rosbash, M. (1985) J. Mol. Biol.186, 146-166.

7. Schuh, R., Aicher, W., Gaul, V., C6td, S., Preiss, A., Maier,D., Seifert, E., Nauber, U., Schrdder, C., Kemler, R. &Jackle, H. (1986) Cell 47, 1025-1032.

8. Hanas, J. S., Hazuda, D. J., Bogenhagen, D. F., Wu,F. Y.-H. & Wu, C.-W. (1983) J. Biol. Chem. 258, 14120-14125.

9. Tso, J. Y., Van Den Berg, D. J. & Korn, L. J. (1986) NucleicAcids Res. 14, 2187-2200.

10. Gilbert, W. (1978) Nature (London) 271, 501.11. Diakun, G. P., Fairall, L. & Klug, A. (1986) Nature (London)

324, 698-699.12. Birnboim, H. C. & Doly, J. (1979) Nucleic Acids Res. 7,

1513-1523.13. Clewell, D. B. & Helinski, D. R. (1970) Biochemistry 9,

4428-4440.14. Mandel, M. & Higa, A. (1970) J. Mol. Biol. 53, 159-162.15. Maxam, A. & Gilbert, W. (1980) Methods Enzymol. 65, 499-560.16. Kleid, D. G., Yansura, D., Small, B., Dowbenko, D., Moore,

D. M., Grubman, M. J., McKercher, P. D., Morgan, D. O.,Robertson, B. H. & Bachrach, H. L. (1981) Science 214,1125-1129.

17. Gross, E. (1967) Methods Enzymol. 77, 238-255.18. Means, G. E. & Feeney, R. E. (1971) Chemical Modification

ofProteins (Holden-Day, San Francisco).19. Smith, D. R., Jackson, I. J. & Brown, D. D. (1984) Cell 37,

645-652.20. Frankel, A. D. & Pabo, C. 0. (1985) J. Cell. Biochem. 9B, 132.21. Grosjean, H. & Fiers, W. (1982) Gene 18, 199-209.22. Bertini, I. & Luchinat, C. (1984) Adv. Inorg. Biochem. 6, 71-111.23. Sakonju, S., Bogenhagen, D. F. & Brown, D. D. (1980) Cell

19, 13-25.24. Bogenhagen, D. F., Sakonju, S. & Brown, D. D. (1980) Cell

19, 27-35.25. Engelke, D. R., Ng, S.-Y., Shastry, B. S. & Roeder, R. G.

(1980) Cell 19, 717-728.26. Sakonju, S., Brown, D. D., Engelke, D., Ng, S.-Y., Shastry,

B. S. & Roeder, R. G. (1981) Cell 23, 665-669.27. Giedroc, D. P., Keating, K. M., Williams, K. R., Konigsberg,

W. H. & Coleman, J. E. (1986) Proc. Natl. Acad. Sci. USA 83,8452-8456.

28. Maret, W., Andersson, I., Dietrich, H., Schneider-Bernlohr,H., Einarsson, R. & Zeppezauer, M. (1979) Eur. J. Biochem.98, 501-509.

29. Pabo, C. 0. & Sauer, R. T. (1984) Annu. Rev. Biochem. 53,293-321.

30. McClarin, J. A., Frederick, C. A., Wang, B.-C., Greene, P.,Boyer, H. W., Grable, J. & Rosenberg, J. M. (1986) Science234, 1526-1541.

Biochemistry: Frankel et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

0