Proc. Nati. Acad. Sci. USA Vol. 88, pp. 6901-6905, August 1991 Biochemistry The leucine zipper symmetrically positions the adjacent basic regions for specific DNA binding (protein-DNA interactions/bZIP domain/yeast GCN4/transcription factor) WILLIAM T. PU AND KEVIN STRUHL Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA 02115 Communicated by Thomas A. Steitz, May 13, 1991 (receivedfor review February 22, 1991) ABSTRACT The bZIP structural motif present in several that diverging a-helices from the leucine zipper continued eukaryotic transcription factors is defined by the leucine zip- into the adjacent basic regions. A more recently proposed per, a coiled-coil dimerization interface, and an adjacent basic "induced fork" model makes these same predictions (7) but region that directly interacts with DNA. To emi the differs from the scissors grip model with respect to whether functional importance ofthe highly conserved spacing between the invariant asparagine in the basic region directly contacts the leucine zipper and the basic region, we have analyzed the DNA or forms an "N-cap" that permits a sharp bend for DNA-binding ability of yeast GCN4 proteins containing amino wrapping around the DNA. The fact that the GCN4 bZIP acid insertions between these two subdomains. Proteins con- domain is almost entirely a-helical when bound to its target taining a surprisingly wide variety of seven-amino acid inser- site (7, 15) is consistent with both models. tions, but none containing two-, four-, or six-amino acid Here, we test critical aspects of the scissors grip and insertions, are functional. However, heterodimers between induced fork models by analyzing the DNA-binding proper- wild-type GCN4 and functional derivatives containing seven ties of GCN4 mutants in which the spacing between the amino acid insertions are unable to bind DNA. These obser- leucine zipper and basic region has been altered. We show vations provide strong experimental support for several aspects that although the spacing between these two subdomains is of the scissors grip and induced fork models for DNA-binding crucial for DNA binding, it can be altered by the insertion of by bZIP proteins. Specifically, they demonstrate that contin- an integral number of a-helical turns (seven amino acids) uous a-helices symmetrically diverging from the leucine zipper without affecting GCN4 function. Further, efficient DNA correctly position the two basic regions for specific binding to binding requires that the two basic regions ofthe GCN4 dimer abutting DNA half-sites. In addition, the results indicate that must be symmetrically related with respect to the leucine GCN4 homodimers are primarily responsible for transcrip- zipper. These observations strongly suggest that continuous tional activation in yeast cells. a-helices symmetrically diverging from the leucine zipper correctly position the two basic regions for specific binding to abutting DNA half-sites. The yeast GCN4 protein belongs to the class of eukaryotic transcription factors whose DNA-binding domains contain a bZIP structural motif (1). bZIP proteins bind as dimers to MATERIALS AND METHODS dyad-symmetric target sequences (2-5), indicative of pro- Construction of Insertion Mutants. The DNA molecules tein-DNA complexes in which two protein monomers inter- used in this paper were derived from YCp88-GCN4, a URA3 act with adjacent half-sites. The conserved bZIP structural centromeric vector that permits expression ofGCN4 in yeast motif that defines this class of proteins consists of a leucine cells from the DEDI promoter and in vitro from the bacte- zipper dimerization element (1) that structurally resembles a riophage SP6 promoter (16). To facilitate manipulations, the coiled coil (6, 7) and an adjacent basic region that determines GCN4 gene was modified without changing the encoded DNA-binding specificity (8). The leucine zipper and basic protein by using oligonucleotides to introduce restriction region are distinct structural subdomains because they can be sites throughout the bZIP domain, including the Xho I site interchanged between different family members to generate where all insertion mutations were made (Fig. 1). To generate chimeric proteins with predicted dimerization and DNA- libraries containing either LEX4 or LEX5 insertions between binding specificities (8-11). Moreover, synthetic leucine zip- the leucine zipper and basic region, degenerate double- per peptides form dimers of appropriate specificity (6, 12), stranded oligonucleotides (boxed positions were pro- and a synthetic basic region (dimerized via a disulfide bond) grammed to be 79o wild-type and 7% each other nucleotide; can specifically bind DNA, although with reduced affinity Fig. 1) were inserted in either orientation at the Xho I site. (13). These were obtained by converting CGACTCGAG- The spacing between the leucine zipper and basic region GiATAAAGT ACTCGAGT or AGGCTCGAGITCL (defined by the distance between the N-terminal leucine in TCCACTlllATCCTCGAG (underlined residues indicate po- the zipper and the conserved pair of C-terminal arginines in sitions of degeneracy) to double-stranded DNA by mutually the basic region) is seven amino acids in all known bZIP primed synthesis (17), followed by cleavage with Xho I. proteins. Moreover, insertions of five amino acids between Derivatives containing two- or four-amino acid inserts were the leucine zipper and the basic region oftranscription factor constructed by inserting Xho I linkers. The insertion mutants C/EBP abolish DNA binding (8). Largely on the basis of the are named by the position and length of the additional conserved spacing between the subdomains, it was proposed residues; e.g., 2521n7-1 means the first isolate of a seven- that the leucine zipper correctly and symmetrically positions residue insertion at position 252. the two basic regions for specific DNA binding to adjacent Phenotypic Analysis. Libraries encoding the LEX4 or LEX5 half-sites (14). Further, this "scissors grip" model suggested insertion proteins were introduced into the Saccharomyces cerevisiae gcn4 deletion strain KY803 by selecting for uracil- The publication costs of this article were defrayed in part by page charge independent transformants, and the resulting strains were payment. This article must therefore be hereby marked "advertisement" assayed for GCN4 function by growth in 20-80 mM amino- in accordance with 18 U.S.C. §1734 solely to indicate this fact. triazole as described previously (16). Plasmid DNAs were 6901 Downloaded by guest on September 27, 2021 6902 Biochemistry: Pu and Struhl Proc. Natl. Acad Sci. USA 88 (1991) I 88 147 221 281 a IzI~zI- ACTIVATION DNA BINDING XhoI CGTGCTCGTAAACTGCAGCGTATGAAGCAGCTCGAGGATAAIAGTGGAAGAACTC ArgA1aArgLysLeuGlnArgMetLysGlnLeuGl uAspLysValGl uGi uLe 250 255 2K b TCGArC TCGAA AAT G TATTTCAC C¢TATTTCACCTTC~tGAGCT 79 % WT 79% w-r C DPAALKRARNTEAARRSRARKLQRMKQ LEDKVEE LLSKNYH LENEVAR LKKLVGER Nlame Phenotyp 253In2-1 MKQ LEV EDKVEE L 253In2-2 MKQ LDL EDKVEE L 253In4-1 MKQ LDLEV EDKVEE L FIG. 1. Construction and characterization of 252In6-1 MKQ LEANVA LEDKVEE L insertion mutants. (a) Structure of GCN4, showing 252In6-2 ... LEDRVE 252In6-3 LERHCI the locations of the activation and DNA-binding (bZIP) domains (numbers refer to amino acid resi- 252In6-4 ... LEFQF( 252In6-5 ... LEAQVV dues), the DNA sequence of the junction between 252In6-6 ... LECHFI the basic region (double bar) and leucine zipper (single bar), and the amino acid sequence (italics). 252In7-1 MKQ LEFFRFI LEDKVEE L ++ The GCN4 allele encodes the wild-type protein but 252In7-2 LTLLECI .. +++ has been modified to introduce restriction sites, 252In7-3 ... LBFNHFI ... ++ 252In7-4 ... LRIFYNI including the indicated Xho I site, where all inser- ++ 252In7-6 ... LEDRGDE ... tion mutations were made. (b) Degenerate double- 252In7-7 ... LEDVERE ... ++ stranded oligonucleotides [boxed positions were +++ 252In7-8 ... LELFQFI ... programmed to be 79% wild-type (WT) and 7% each 252In7-9 ... LBDEVVE ... ++ other nucleotide] were inserted in either orientation 252In7-10 ... LEFLIFI ... ++ at the Xho I site, thus generating either LEX4 (left) 252In7-11 ... LEFFQFE ... ++ or LEX5 (right) insertions between the leucine 252In7-12 ... LEDIVRE ... ++ ++ and basic region. (c) Insertion mutants used 252In7-14 ... LBSYLI ... zipper 252In7-15 ... LESYHFI ... ++ in this study. Listed below the sequence (standard 252In7-16 ... LEFFHFI ... ++ one-letter symbols) of wild-type GCN4 are the ++ 252In7-17 L.DNVQG ... names (e.g., 2521n7-1 means the first isolate of a 252In7-18 ... LEYVNI ... seven-residue insertion at position 252), inserted sequences (bold), and phenotypes of the insertion 252In7-20 ... LEHFNCI ... mutants. The phenotype refers to the ability of 252In7-21 ... LEFFICI ... 252In7-22 ... LEFFPFI ... these mutants to complement a gcn4 deletion strain 252In7-27 ... LBFCPFI ... (see Materials and Methods). recovered from individual yeast strains by transformation Glutaraldehyde Crosslinking. Equal amounts of in vitro into Escherichia coli, and the sequences of the inserted synthesized, I'S-labeled proteins were incubated for 10 min regions were determined. These DNAs were also reintro- at room temperature in 20 mM potassium phosphate at pH duced into KY803 to confirm their growth phenotypes, which 7.0/100 mM KCl/3 mM MgCl2/1 mM EDTA, in either the are designated as follows: + + +, indistinguishable from wild presence or the absence of 6 ,uM GCN4p, the 58-residue type; + +, slightly