THE JOURNAL OF BIOLOGICAL CHEMISTRY 8 1992 by The American Society for Biochemistq and Molecular Biology, Inc.

Vol. 267, No. 19, Issue of July 5, pp. 13714-13718,1992 Printed in U. S. A.

MyoD Binds to the Guanine Tetrad Nucleic Acid Structure* (Received for publication, November 13, 1991)

Kenneth WalshS and Antonio Gualbertoil From the Department of Physiology and Biophysics, Case Western Reserve University, Cleveland, Ohio 44106

A high affinity interaction between a protein and the guanine tetrad nucleic acid structure is described. Re- combinant MyoD, a transcription factor that can ini- tiate myogenesis, specifically bound to helical struc- tures formed by stacks of guanine residues in square planar arrays. The N-7 methylation of a set of consec- utive dG residues in a single-stranded probe of the creatine kinase enhancer interfered with the forma- tion of this nucleic acid structure and prevented pro- tein binding. Recombinant MyoD also bound to a gua- nine tetrad formed with a telomeric DNA probe, and it had a higher affinity for the four-stranded structure than for the double-stranded E-box-binding site. These data are the first report of a direct interaction between a protein and this nucleic acid conformation. The po- tential biological significance of this finding is dis- cussed.

The recognition of specific nucleic acid conformations by proteins is a central feature of transcription, translation, and replication. Recently, a complex nucleic acid structure was identified in studies analyzing the self-association of single- stranded DNA probes of telomeres, immunoglobulin switch regions, and other G-rich DNA sequences (1-5). Two related helical conformations, referred to as G-quartets and G4 DNA, have 4 guanine residues in a square planar array and a monovalent cation at the core of the helix. These four- stranded helices are stabilized by eight Hoogsteen hydrogen bonds/tetrad of guanine residues. This structure is kinetically slow to form, but it is extremely stable under physiological conditions. The half-life of unfolding for G-quartet structure formed from a Oxytricha telomere probe is from 4 to 18 h at 37 “C depending on the identity of the monovalent cation (6). Because runs of guanine residues are common in telomeres, gene regulatory sequences, and recombination hotspots it has been proposed that these stable guanine tetrad structures have roles in various nuclear processes.

In a previous study we determined that MyoD bound with apparent high affinity to single-stranded DNA probes of specific sequence, but the conformation of the nucleic acid was not determined (7). MyoD is a member of the family of helix-loop-helix proteins that share homology with the myc oncogene and function as transcription factors and regulators of cellular differentiation (8,9). The expression of MyoD, and

* This work was supported by National Institutes of Health Grants AR 40197 and HL 45345. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Established Investigator of the American Heart Association. TO whom correspondence should be addressed Dept. of Physiology and Biophysics, Case Western Reserve University, 2109 Adelbert Rd., Cleveland, OH 44106. Tel.: 216-368-3487; Fax: 216-368-5586.

Fellow of the Ministerio de Educacion y Ciencia, Spain.

the related proteins, myogenin, MRF4, and Myf-5, are typi- cally confined to skeletal muscle cell types; however, these proteins can heterodimerize with the E12 and E47 proteins which are ubiquitous. The entire family of helix-loop-helix proteins bind to sites in the major groove of B DNA with the core consensus sequence of CANNTG. This double-stranded DNA-binding site, referred to as an E-box, occurs in a nu- merous of muscle and non-muscle genes. The novel, “single- stranded” binding site for MyoD was identified in the non- coding strand of and E-box element from the muscle-specific creatine kinase enhancer (7). In the experiments reported here, we demonstrate that recombinant MyoD specifically binds to a low abundance, non-B DNA structure that is present in the single-stranded DNA pool, and not to randomly coiled nucleic acid. The properties of this nucleoprotein com- plex indicate that MyoD specifically recognizes a quadruple helix formed by the self-association of guanine residues. Fur- thermore, we show that this protein also binds to an G-quartet structure formed from the association of a Tetrahymena te- lomere probe. The affinity of recombinant MyoD for the guanine tetrad structure was significantly higher than for the E-box consensus sequence in B DNA. The identification of a high affinity protein interaction with guanine tetrads suggests that this nucleic acid structure may be important in biological systems.

MATERIALS AND METHODS

Preparation of Nucleic Acid Probes-The DNA oligonucleotides were prepared with an Applied Biosystems DNA synthesizer using the phosphoramidite method and purified by electrophoresis on a denaturing 20% polyacrylamide gel and Sep-Pak C18 cartridges (Waters Associates). The single-stranded CKMnc probe contains the enhancer sequences from positions -1140 to -1160 of the mouse creatine kinase gene. This probe has the sequence, 5”TCGAT- CAGGCAGCAGGTGTTGGGGGA-3’. The double stranded E-box probe has the CKMnc probe annealed to the reverse complement of this sequence, 5’-TCGATCCCCCAACACCTGCTGCCTGA-3’, to give rise to a duplex with Sal1 compatible ends. These sequences were annealed at 37 “C in a buffer of 10 mM Tris, pH 8 , l mM EDTA, and 100 mM NaCl for approximately 1 h, and the double-stranded probe was separated from residual single-stranded material by native poly- acrylamide gel electrophoresis. The Tetrahymena telomere probe, TE1, has the structure, 5”ACTGTCGTACTTGATATTGGG- GTTGGGG-3’, with two telomeric repeats, in bold type, 3’ of an unrelated sequence. This sequence is identical to the TE1 probe analyzed in the guanine tetrad study of Sen and Gilbert (10). Single- and double-stranded probes were radiolabeled with polynucleotide kinase and [Y-~*P]ATP (>4500 Ci/mmol, Amersham Corp.) for ap- proximately 2 h at 37 “C. Unreacted ATP was removed with Elutip- d columns (Schleicher & Schuell) according to the directions cf the manufacturer. Following ethanol precipitation, radiolabeled probes were resuspended in a solution of 10 mM Tris, pH 8, 1 mM EDTA, and 100 mM NaCl. Under these conditions of probe preparation approximately 2% of the CKMnc sequence occurred in the guanine tetrad conformation which is designated F*. The guanine tetrad was separated from the randomly coiled CKMnc probe by running the Elutip-d-purified probe on a native 5% polyacrylamide gel. Following autoradiography, the F and F* bands were excised and eluted into a solution of 10 mM Tris, pH 8, 1 mM EDTA, and 100 mM NaC1.

13714

MyoD-Guanine Tetrad Interaction 13715

Polyribo(G) (Pharmacia LKB Biotechnology Inc.) was diluted in a solution of 10 mM Tris, pH 8 , l mM EDTA, and 100 mM NaCI.

Preparation of Proteins-The bacterially produced, recombinant MyoD protein, with all but the first three amino acids of MyoD fused to glutathione S-transferase, was purified by chromatography on glutathione-agarose (11). The helix-loop-helix proteins, myogenin and E12, were prepared by in vitro translation from their correspond- ing RNAs (12). RNA was synthesized using linearized DNA template and T3 RNA polymerase for 2 h at 37 "C, and translation of the RNA was performed with a rabbit reticulocyte lysate (Promega). The E12- myogenin heterodimer was formed by mixing transcription products.

Analyses of Protein-Nucleic Acid Interactions-Electrophoretic mobility shift assays were typically performed with 7 p~ to 1.2 nM DNA fragment and 0-2.25 nM recombinant MyoD homodimer or 4 pl of programmed rabbit reticulocyte lysate that was prepared as described above. The concentration of active MyoD was determined by measuring the concentration of the protein-nucleic acid complex at saturating levels of the duplex E-box probe from the creatine kinase enhancer. Standard binding reactions contained 10 mM Tris, pH 7.5, 30 mM KCI, 3 mM MgC12, 1 mM EDTA, 1 mM dithiothreitol, 5% glycerol, and 0.5% Nonidet P-40. Variations in the binding mixtures are indicated in the figure legends. Binding reactions were initiated by the addition of protein. Following a 30-min incubation at room temperature, binding mixes were loaded on a 5% polyacrylamide gel and electrophoresis was performed at 10-14 V cm-I in 22 mM Tris borate buffer with 0.5 mM EDTA. No changes in complex formation were detected by longer incubations prior to the electrophoresis step. Relative binding affinities was were determined by the electrophoretic mobility shift assay method. Constant amounts of purified CKMnc guanine tetrad or duplex E-box were incubated with increasing amounts of Glu-MyoD protein. Following electrophoresis and auto- radiography, the ratios of bound and free DNA probes were deter- mined by scanning the autoradiograms were with an LKB densitom- eter. Methylation interference footprinting was performed with par- tially methylated CKMnc probe. The radiolabeled CKMnc oligonucleotide was heated to 70 "C to eliminate pre-existing struc- tures and partially methylated with dimethyl sulfate under standard conditions. Scaled-up, electrophoretic mobility shift assays were per- formed under standard conditions with 12 ng of partially methylated CKMnc fragment in the presence or absence of 0.15 pmol of the MyoD fusion protein. The nucleoprotein complex and free DNA bands corresponding to the F and F* species were excised from the 5% non-denaturing gel and eluted into 20 mM Tris, pH 7.5, 1 mM EDTA, and 200 mM NaCI. This material was purified with Elutip-d columns, and the modified guanine residues were cleaved by incuba- tion with 1 M piperidine at 90 "C for 30 min. The samples were dried under vacuum, dissolved in formamide, and loaded onto a denaturing 20% polyacrylamide gel. Following electrophoresis the gel was dried and exposed to film at -80 "C.

RESULTS

Interactions of Helix-Loop-Helix Proteins with a Single- stranded DNA Probe-Helix-loop-helix proteins were tested for their ability to bind the noncoding strand of the E-box motif from the mouse creatine kinase enhancer (CKMnc) (Fig. 1). We analyzed proteins that were prepared by in vitro translation from a reticulocyte lysate (12) and the purified, bacterially produced MyoD protein which is fused to gluta- thione S-transferase (11). Incubation of the single-stranded CKMnc probe with unprogramed rabbit reticulocyte lysate gave rise to two complexes. These complexes were detected with all single- and double-stranded probes that were tested, and they were not sensitive to competition by a molar excess of non-labeled DNA (not shown). No additional complexes could be detected when the reticulocyte lysate was programed with myogenin RNA. However, a new complex with a slower electrophoretic mobility was detected when the translation lysate was programed with E12 in addition to myogenin. This new complex had a similar mobility as the MyoD-nucleopro- tein complex that was formed with the bacterially produced fusion protein. In separate experiments with the double- stranded E-box probe, Glu-MyoD and the myogenin/E12- programed lysate gave rise to specific complexes, but no

- F'

FIG. 1. Interaction of helix-loop-helix proteins with a sin- gle-stranded DNA probe. Binding reactions were performed with the CKMnc probe and no protein, 4 pl of lysate programed with myogenin RNA (MyoG), 4 pl of lysate programed with myogenin and E12 (MyoG + E12), 4 pl of nonprogramed lysate, or 7.5 fmol of Glu- MyoD fusion protein. Each reaction contained 20,000 cpm of single- stranded CKMnc probe, 10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 5% glycerol, and 0.5 pg of poly(dI-dC). poly(d1-dC) in 10 pl. The positions of the free fragment species, F and F*. are indicated.

specific complexes were detected with the lysate control or with myogenin alone (not shown).

Our data confirm a previous report that helix-loop-helix proteins can bind to the single-stranded CKM probe (7). However, experiments with the in vitro translated proteins are complicated by the high level of nonspecific DNA binding activities in the reticulocyte lysate. We suspect these com- plexes arise from the interaction with an abundant, nonspe- cific nucleic acid-binding protein in the lysate. However, other investigators have provided evidence that one of these com- plexes may be due to the binding of the endogenous USF transcription protein (14). Due to uncertainties about the endogenous, nonspecific binding activities in the reticulocyte lysate, further experiments to determine the nature of the nucleoprotein complex were performed with the MyoD fusion protein purified from Escherichia coli.

MyoD Binds to a Low Abundance Structure in the Single- stranded DNA Pool-A band with a slower mobility than the bulk-free DNA fragment was detected in the autoradiograph when protein was left out of the electrophoretic mobility shift assay (Fig. 1). This species of DNA fragment, referred to as F*, comprised less than 2% of the total CKMnc probe. To test the possibility that the F* species is specifically recog- nized by MyoD, the level of recombinant protein was varied while the level of total CKMnc probe was kept constant in binding assays (Fig. 2). Increasing amounts of MyoD led to a concomitant decrease in the intensity of the F* band. When the F* species was depleted, additional increases in the con- centration of MyoD did not increase the intensity of the nucleoprotein complex. These data provided the first indica- tion that the recombinant MyoD protein recognizes a specific nucleic acid structure, represented by the F* band, and not randomly coiled, single-stranded DNA. The specificity of this interaction was further indicated by the inability of MyoD to bind five other unrelated single-stranded DNA probes or the reverse complement of the CKMnc probe (7).'

To investigate the interactions of MyoD with the F* and F structures of the CKMnc probe in greater detail, each DNA species was purified and analyzed in the electrophoretic mo- bility shift assay. The F and F* species were excised from a

K. Walsh and A. Gualberto, unpublished results.

13716 MyoD-Guanine Tetrad Interaction

[Glu-MyoD] 0 4 “ “ W 1

F’

FIG. 2. The formation of the MyoD-nucleoprotein complex correlates with the diminution of the F* species in the single- stranded DNA pool. Binding reactions were performed with 1.2 nM of the CKMnc probe and increasing levels of Glu-MyoD under standard conditions. The Glu-MyoD levels were: 0,0.025,0.075,0.25, 0.75, and 2.25 nM. The positions of the F and F* species of free fragments are indicated.

d

PB

F B b

a b c FIG. 3. MyoD preferentially binds to the F* structure that

occurs in the single-stranded DNA pool. Binding reactions were performed in the presence or absence of 1.1 nM Glu-MyoD and 1100 cpm of either the non-fractionated probe (a) , the purified F* species ( b ) , or the purified F species (c). The F* and F probes were prepared from the non-fractionated CKMnc probe by gel purification as de- scribed under “Materials and Methods.” Electrophoretic mobility shift assays were performed under standard conditions.

preparative native gel, and the DNA was eluted into a solution of 10 mM Tris, 1 mM EDTA, and 100 mM NaC1, and ethanol- precipitated. The gel-purified F* and F species were compared with the non-fractionated CKMnc probe in the electropho- retic mobility shift assay (Fig. 3). A constant amount of radioactivity was used for each lane of the gel. In the absence of protein, the purified F and F* species had electrophoretic mobilities that were identical to the F and F* bands from the non-fractionated CKMnc probe. These data indicate that the F* structure was stable during this isolation procedure and that it is not in rapid equilibrium with the F structure. However, the electrophoretic mobility of the F* species could be converted to that of the F species by a preincubation at 90 “C or by electrophoresis in 50% urea (not shown).

MyoD predominantly recognized the F* species of the CKMnc probe (Fig. 3). In the presence of MyoD, all of the F* species became complexed with the protein, but the gel- purified F species was bound to a much lesser extent. Under the conditions of this assay, MyoD gave rise to a multitiered pattern of complex formation. This pattern of complex for- mation was most prominent when the input DNA was the purified F* species rather than the non-fractionated CKMnc probe and appeared to depend on the amount of input F*. These patterns of complex formation may result from the formation of higher order, protein-nucleic acid interactions at the higher levels of F*.

MyoD Recognizes the Guanine Tetrad Structure-To deter- mine the nature of the protein nucleic acid interaction be- tween F* and MyoD, methylation interference footprinting was performed on the F* complex and on the pool of MyoD.

-F* complexes (Fig. 4A). The non-fractionated CKMnc probe was heated to eliminate the F* structure, treated with di- methyl sulfate to methylate the N-7 groups, and incubated for 2 h at 37 “C to reform the structure. This partially meth- ylated probe was used in a scaled-up electrophoretic mobility shift assay. The fast mobility F, slow mobility F*, and nucleo- protein complex bands were excised, and the DNA was cleaved with piperidine. This reagent cleaves at guanine residues that are methylated on the N-7 position. The methylation of each of 5 consecutive guanine residues at the 3‘ portion of the probe interfered with nucleoprotein complex formation and also interfered with the formation of the F* structure. These data indicate that guanine N-7 methylation prevents with the formation of the F* structure and this, in turn, blocks nucleo- protein complex formation with MyoD.

The methylation-cleavage patterns of F* and the MyoD-F* nucleoprotein complex are strikingly similar to the methyla- tion protection and interference patterns observed for the guanine tetrad nucleic acid structures which form by the self- association of telomeres, immunoglobulin switch regions, and other single-stranded, G-rich sequences (1-4, 15). These structures have 4 guanine residues in a square, planar array, and they are stabilized, in part, by G. G base pairing through the N-7 positions (Fig. 4B). Based on the methylation-cleav-

A G C F G P F m

B 0

3‘ A G G G G G

T T

G T G G A C

A G

G C

A G

C T 5 ‘

FIG. 4. Methylation interference footprinting identifies consecutive dG residues in the CKMnc probe that are essential for MyoD binding and for the formation of the F* structure. A, scaled-up electrophoretic mobility shift assays were performed with CKMnc probe that was partially methylated with dimethyl sulfate as described under “Materials and Methods.” The free DNA species, F and F*, and the protein DNA complex (C), were excised from the gel following autoradiography, treated with piperidine, and analyzed on a denaturing 20% polyacrylamide gel. Some of the partially methylated CKMnc probe not used in the binding assay was also treated with piperidine to create a G ladder ( G ) on the sequencing gel. The sequence of the CKMnc probe is shown, and the guanine residues with critical N-7 positions are in bold type. B, diagram of Hoogsteen hydrogen bonding in the guanine tetrad. The N-7 residues are indicated by cross-hatched lines.

MyoD-Guanine Tetrad Interaction 13717

age patterns shown in Fig. 4A and the slow electrophoretic mobility of the F* DNA species, we propose that the nucleic acid structure recognized by the recombinant MyoD protein is a guanine tetrad formed by the self-association of four single-stranded CKMnc molecules through the 5 consecutive guanine residues at the 3' portion of the probe.

To further test the hypothesis that recombinant MyoD binds to the guanine tetrad structure, a single-stranded probe corresponding to the Tetrahymena telomere, referred to as TE1, was tested for binding. This same DNA probe was previously demonstrated to form guanine tetrads, presumably through a dimer fold-back mechanism (10). The guanine tetrad, or F*, structure for TE1 was purified by native gel electrophoresis as described previously. This slow mobility, F* species of TE1 was compared with the F* species of the CKMnc probed for binding to recombinant MyoD in an electrophoretic mobility shift assay (Fig. 5) . Under the con- ditions of this assay, virtually all of the F* species of either DNA probe was bound by MyoD, although MyoD gave rise to different multitiered patterns of complex formation. The control protein, glutathione S-transferase, had no DNA bind- ing activity. The " F form of the Tetrahymena telomere probe, which has a fast electrophoretic mobility and is probably a monomer, was not bound by MyoD (not shown). In other experiments we tested the binding of MyoD to oligonucleo- tides that contained insertion mutations which interrupt the tracts of guanines. These insertions prevented the formation of the F* structure, and these oligonucleotides did not show any ability to bind MyoD (not shown). These data add further support to the hypothesis that MyoD recognizes the guanine tetrad nucleic acid structure.

High Affinity Binding to Guanine Tetrads-In a previous study we showed that the single-stranded CKMnc probe and the double-stranded E-box probe were equally effective at competing the formation of a specific MyoD nucleoprotein complex (7). Since MyoD recognizes the low abundance, guanine tetrad structure that comprises a fraction of the total DNA pool, it follows that this protein has a higher affinity for guanine tetrad than for the double-stranded E-box probe. Binding assays were performed with a fixed level of DNA probe and variable levels of protein to directly compare the affinity of the MyoD for the guanine quartet structure relative to the duplex-binding site with the E-box sequence motif (Fig. 6). The double-stranded probe used in these experiments was synthesized to the high affinity E-box-binding site that occurs 3' of the tract of 13 consecutive dC - dG in the mouse creatine

Tetrahymena C K M ~ ~ telomere

FIG. 5. MyoD binds to the G-quartet conformation of the Tetrahymena telomere. Binding reactions were performed under standard conditions with approximately 30,000 cpm of the gel-puri- fied guanine tetrad structures of the TE1 or the CKMnc probes (referred to as I") with no protein, 1.5 nM Glu-MyoD, or the equiv- alent mass amount of the glutathione S-transferase control protein (Glu).

[Glu-MyoD] 0 d 0 d ___.

F' 1

I I. . . PO 9 0 - - F

Guanine tetrad Duplex E-box

FIG. 6. MyoD has a higher affinity for the guanine tetrad structure than for the duplex-binding site in the creatine kinase enhancer with the E-box motif. Binding reactions were performed under standard conditions with different concentrations of Glu-MyoD and fixed amounts of gel purified duplex E-box (27 PM) or CKMnc guanine tetrad (7 PM) probes. The concentration of Glu- MyoD was 0, 0.09, 0.19, 0.38, 0.76, or 1.5 nM.

Competitor: duplex E - ~ X PO'Y (G) 0 4 0 4

F b I FIG. 7. Polyribo(G) is a potent competitor of the MyoD-

nucleoprotein complex. Binding reactions were performed under standard conditions with 1.1 nM Glu-MyoD and 0.3 nM duplex E-box probe in the presence or absence of non-labeled duplex E-box probe or polyribo(G). Binding reactions contained either no competitor or competitor at 0.03, 0.3, 3.0, or 30 ng in a 10-pl reaction.

kinase enhancer (11). The guanine tetrad probe was formed from the single-stranded CKMnc DNA. The concentration of MyoD required to bind half of the guanine tetrad structure (F*) was 4-5-fold lower than the concentration of MyoD required to bind half of the duplex E-box probe. Under the conditions of this assay, 50% of the guanine tetrad structure was bound at a MyoD concentration of approximately 1 X 10"' M. The affinity of MyoD for the guanine tetrad is comparable to that of other proteins that recognize specific sequences in B-DNA. Similar dissociation constant values have also been reported for complexes between recombinant MyoD and the E-box element of the immunoglobulin K-light chain enhancer (13).

A guanine tetrad structure with four parallel strands was originally proposed for complexes of polyribo(G) (16). To test whether MyoD could also recognize potential guanine tetrad structures in ribonucleic acids, solutions of polyribo(G) were analyzed for their ability to compete for complex formation between MyoD and the duplex E-box probe (Fig. 7). On a mass basis, polyribo(G) was at least a 10-fold more potent competitor than the duplex E-box probe. In separate control experiments polyribo(G) did not effectively compete the bind- ing of the SRF and MAPF transcription factors to promoter- binding sites (not shown). These competition data provide

13718 MyoD-Guanine Tetrad Interaction

additional evidence for a specific, high affinity interaction between MyoD and G-strand structures, and suggest that ribonucleic acid may be a target site for this protein.

DISCUSSION

A high affinity protein interaction with the guanine tetrad nucleic acid structure has been identified. Complexes were formed between recombinant MyoD and the helical structure formed from the self-association of guanine residues in a square planar array. Numerous lines of evidence indicate that the MyoD fusion protein specifically recognizes guanine te- trads. This protein bound to a low abundance structure that occurred in a pool of single-stranded CKMnc probe (Figs. 2 and 3), but no complex formation was detected with other single-stranded DNA probes to non-related sequences. The low abundance structure formed by the CKMnc DNA had a characteristically slow electrophoretic mobility which was not in rapid equilibrium with the bulk of randomly coiled, single- stranded probe. However, this slow mobility species could be eliminated by heating or by electrophoresis under denaturing conditions. The CKMnc probe has a stretch of 5 consecutive guanine which upon methylation at the N-7 position inter- fered with the binding of MyoD and also interfered with the formation of the nucleic acid structure (Fig. 4). These meth- ylation interference footprints are strikingly similar to the methylation/cleavage patterns of the guanine tetrad struc- tures formed by the self-association of telomere, IgG switch regions, and other single-stranded, G-rich probes (1-4, 15). To further test the possibility that MyoD was binding to a guanine tetrad structure, we analyzed a single-stranded, Te- trahymena telomere probe which had previously been shown to form the guanine tetrad structure (10). This telomeric guanine tetrad was bound by MyoD (Fig. 5). Since the telom- ere and CKMnc probes have little, if any, sequence similarity outside of the tracts of dG, it appears that the protein specif- ically recognizes the higher order nucleic acid structure that arises by the self-association of guanine residues. Finally, we demonstrated that polyribo(G), originally used for structural determinations of the guanine tetrad (16), was a potent com- petitor of the MyoD-nucleoprotein complex (Fig. 7). Collec- tively, these data indicate that the recombinant MyoD protein specifically recognizes guanine tetrads in DNA and RNA. This unusual nucleoprotein complex had a dissociation con- stant of approximately 1 X 10"' M, which is 4-5-fold more stable than the complex between MyoD and the E-box probe that has been described previously (Fig. 6).

Two related helical conformations have been proposed for the guanine tetrad structure (1-4). One conformation, referred to as "G4 DNA," has tetrads of parallel chains. This structure has 4-fold symmetry and all of the glycosyl bonds occur in the anti form. Another conformation, referred to as the "G- quartet," has tetrads with antiparallel chains and guanines in the syn and anti forms. The G-quartet has a 4-fold rotational symmetry within the base tetrad, but the sugar-phosphate backbone reduces the symmetry to 2-fold giving rise to two minor grooves and two major grooves. The conformation of the guanine tetrad formed by the CKMnc probe is not known. However, sequences with multiple separated tracts of guanine residues form G-quartets. The Tetrahymena telomere probe, with two dG tracts, can form the G-quartet structure by the dimerization of strands with fold-back geometries and gua- nines in the syn and anti conformations (10). Since MyoD

binds to this telomere probe (Fig. 5) it appears that the protein specifically recognizes the G-quartet conformation. It is also possible that MyoD does not distinguish between the G4 and G-quartet conformations because it primarily recognizes fea- tures of the purine tetrad. This latter hypothesis is unlikely, however, because the orientations of the sugar phosphate backbones and the glycosidic bond angles differ significantly between these two structures.

Guanine-rich DNA segments occur frequently in eucaryotic genomes. Numerous studies have shown that runs of G resi- dues in duplex DNA can adopt unusual conformations in vitro and, perhaps, in vivo (17-19). Guanine tetrads have been proposed to occur at IgG switch regions during meiotic pairing ( l ) , within the coding region of the c-Ha-ras gene (20), and at the single-stranded overhangs of telomeres (2-4). Despite the finding that many sequences can form guanine tetrads in vitro, a stable complex between a protein and this nucleic acid conformation has not been previously identified. The G- quartet at the Oxytrichu telomere is not recognized by telom- erase or other telomeric proteins, but this structure may function to inhibit the telomerase and act as a negative regulator of elongation in vivo (6, 21). The results of our experiments suggest the possibility of direct interactions be- tween proteins and guanine tetrads in vivo. This unusual conformation most likely occurs in ribonucleic acid which can adopt a multitude of structures (22). The guanine tetrad is very stable under physiological conditions, and it may con- tribute to RNA tertiary structure and its recognition by proteins. The identification of this novel nucleoprotein com- plex adds support to the hypothesis that the guanine tetrad structure is biologically significant and suggests a new mech- anism by which proteins can interact with nucleic acid.

Acknowledgments-We thank Bruce Wentworth and Nadia Rosen- thal for preparing the MyoD-glutathione S-transferase and glutathi- one S-transferase used for these experiments. We thank Paul Schim- me1 for advice.

REFERENCES

2. Williamson, J. R., Raghuraman, M. K., and Cech, T. R. (1989) Cell 5 9 , 1. Sen, D., and Gilbert, W. (1988) Nature 334,364-366

n m -nnn 3. Sundquist, W. I., and Klug, A. (1989) Nature 342,825-829 4. Panyutin, I. G., Kovalsky, 0. I., Budowsky, E. I., Dickerson, R. E., Rikhirev,

M. E., and Lipanov, A. A. (1990) Proc. Natl. Acad. Sei. U. S . A. 87,867-

-"

Q7n 5. Jin, R., Breslauer, K. J., Jones, R. A,, and Graffney, B. L. (1990) Science

6. Raghuraman, M. K., and Cech, T. R. (1990) Nucleic. Acids Res. 18,4543-

7. Santoro, I. M., Yi, T.-M., and Walsh, K. (1991). Mol. Cell. Biol. 1 1 , 1944-

8. Olson, E. N. (1990) Genes & Deu. 4 , 1454-1461 9. Weintraub, H., Davis, R., Tapscott, S., Thayer, M., Krause, M., Benezra,

R., Blackwell, T. K., Turner, D., Rupp, R., Hollenberg, S., Zhuang, Y., 10. Sen, D., and Gilbert, W. (1990) Nature 344,410-414

and Lassar, A. (1991) Science 251 , 761-766

11. Lassar, A. B., Buskin, J. N., Lockshon, D., Davies, R. L., Apone, S.,

12. Chakraborty, T., Brennan, T. J., Li, L., Edmondson, D., and Olson, E.

13. Sun, X.-H., and Baltimore, D. (1991) Cell 6 4 , 459-470; correction (1991)

14. Blackwood, E. M., and Eisenman, R. N. (1991) Science 251,1211-1217 15. Henderson, E., Moore, M., and Malcolm, B. A. (1990) Biochemistry 2 9 ,

16. Zimmerman, S. B., Cohen, G. H., and Davies, D. R. (1975) J. Mol. Biol.

17. Larsen, A,, and Weintraub, H. (1982) Cell 29,609-622 18. Nickol J. M. and Felsenfeld, G. (1983) Cell 35,467-477 19. Usdin,'K., aAd Furano, A. V. (1988). Proc. Natl. Acad. Sci. U. S. A. 8 5 ,

" I "

250,543-546

4552

1953

Hauschka, S. D., and Weintraub, H. (1989) Cell 58,823-831

(1991) Mol. Cell. Biol. 11 , 3633-3641

Cell 66 ,423

732-737

92,181-192

AAlG-AA3n 20. Smith, S. S., Baker, D. J., and Jardines, L. A. (1989) Biochem. Biophys.

21. Zahler, A. M. Williamson, J. R., Cech, T. R., and Prescott, D. M. (1991)

22. Kim, J., Cheong, C., and Moore, P. B. (1991) Nature 351 , 331-332

"-" l""

Res. Commun. 160 , 1397-1402

Nature 3 5 0 , 718-720