THE JOURNAL OF B I O L ~ I C A L CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 269, No. 14, Issue of April 8, pp. 10341-10351, 1994 Printed in U.S.A.

Identification of a Novel Determinant for Basic Domain-Leucine Zipper DNA Binding Activity in the Acute-Phase Inducible Nuclear Factor-Interleukin-6 Transcription Factor*

(Received for publication, December 3, 1993, and in revised form, January 11, 1994)

Allan R. BrasierS§n and Amalendra Kumar§ From the SDiuision of Endocrinology, Departments of Medicine, $Human Biological Chemistry and Genetics, and the Sealy Center for Molecular Science, University of Texas Medical Branch, Galueston, Texas 77555-1060

Nuclear factor-interleukin-6 (NF-KG), a member of the CCAAT bodenhancer-binding protein (C/EBP) family, contains a basic domain-leucine zipper (bZIP) DNA binding motif. Controlled protease digestion was used to probe free and DNA-complexed NF-IL6 protein. Diges- tion with trypsin in the absence of DNA produced the leucine zipper domain (containing residues 303-345). In contrast, digestion of NF-IL6-DNA complexes produced a stable domain, spanning residues 266-345, termed the tryptic core domain (TCD). The NH,-terminal boundary of the TCD is longer than tryptic peptides reported from C/EBPa.DNA complexes. Digestion of NF-IL6 with endo- protease Asp-N produced a domain smaller than the TCD (NF-IL6 bZIP domains (NFBD) (272-345)), a domain identified either in the absence or the presence of DNA. Both recombinant peptides bind acute-phase response element DNA in a sequence-specific fashion. The equi- librium disassociation constant (K,) for the TCD was 36

8 m, whereas the Kd for NFBD (272-345) was 283 * 160 m. Moreover, in comparison with the TCD, NFBD (272- 345) formed unstable DNA complexes with a 15-fold faster off-rate. We conclude that the amino acids repre- sented between 266 and 272 termed the complex stabi- lizing subdomain, influences DNA complex formation in- dependent of DNA binding specificity, and may be one mechanism for heterogeneity of DNA interaction by CEBP family members.

The human DNA-binding protein termed nuclear factor- interleukin-6 (NF-IL6)’ is a C/EBP family member implicated in the regulation of hepatic acute-phase reactants (Brasier et al., 1990a; Poli et al . , 1990) and cytokine gene promoters (Akira et al., 1990). As a consequence of systemic injury or inflamma- tion, NF-IL6 is synthesized in the mammalian hepatocyte where it binds to distinct cis-regulatory elements in the pro- moters of acute-phase reactants and regulates their activity. The effect of NF-IL6 on transcription appears to be dependent

Grant R29 HL45500, G. D. Searle & Co., and the J o h n Sealy Memorial * This work was supported in part by National Institutes of Health

Endowment Fund. 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.

Fax: 409-772-8709. ll To whom correspondence should be addressed. “el.: 409-772-2824;

bZIP, basic domaideucine zipper; CIEBP, C U T bodenhancer-binding The abbreviations used are: NF-IL6, nuclear factor-interleukin-6;

protein; TCD, tryptic core domain; NFBD, NF-IL6 bZIP domains; CSSD, complex stabilizing subdomain; APRE, acute-phase response el- ement; IPTG, isopropyl-l-thio-6-D-galactopyranoside; Dm, dithiothrei- tol; PAGE, polyacrylamide gel electrophoresis; BSA, bovine serum al- bumin; WT, wild type; CD, circular dichroism.

on the promoter context in which it binds. Although NF-IL6 activates the cytokine IL-6 gene promoter (Akira et al., 1990), our work has demonstrated that NF-IL6 attenuates activity of the inducible enhancer of the angiotensinogen gene, termed the acute-phase response element (APRE), by displacing the potent NF-KB transactivator from an overlapping binding site (Brasier et al., 1990a). Because both activation and attenuation are dependent on DNA interaction, understanding how NF-IL6 binds to DNA is of central relevance to transcriptional events occurring during the hepatic acute-phase response.

NF-IL6 represents the human homolog of rat liver-enriched transcriptional activator protein (Descombes et al., 19901, interleukin-6-response element DNA-binding protein (Poli et al., 1990), q a c i d glycoprotein enhancer binding protein (Chang et al . , 19901, mouse C/EBPP (Cao et al . , 1991), or C/EBP-related protein 2 (Williams et al . , 1991) and, like its rodent counterparts, contains three in-frame initiator methio- nine residues producing NH,-terminal-deleted proteins of -38, 33, and 16 kDa in size (Descombes et a l . , 1990,1991). Compari- son of the predicted amino acid sequence of the 16-kDa COOH terminus of NF-IL6 with other proteins in National Biomedical Research Foundation data base identified an 84 amino acid region homologous to the DNA-binding COOH-terminal do- main of C/EBPa (Akira et al . , 1990; Landschulz, 1988a). Selec- tive expression in Escherichia coli of the 16-kDa COOH termi- nus, termed LIP for liver-enriched inhibitory protein, produces a DNA binding peptide that binds to identical sequences as the mature 33-kDa protein (Descombes et al., 1991). These data indicate that the 16-kDa COOH terminus of liver-enriched transcriptional activator protein contains the DNA-binding domain.

Detailed structural analysis of recombinant C/EBPa DNA binding peptides has resulted in the concept of the basic do- main-leucine zipper (bZIP) motif. The bZIP motif contains two functionally distinct elements. At the NH, terminus is a region rich in the basic amino acids, arginine and lysine, and at the COOH terminus is a 3 4 repeat of hydrophobic leucine resides (Landschulz et al., 1988b, 1989; Kouzarides and Ziff, 1988). Multiple lines of evidence indicate that the basic region of the bZIP protein motif is involved in DNA contact, including mu- tational analysis (Landschulz et al., 1989) and “domain swap- ping” experiments in which the basic region of C/EBP is re- placed with the homologous region of the yeast bZIP protein GCN4 (Agre et al., 1989; Johnson, 1993). Moreover, bromode- oxyuracil cross-linking experiments using GCN4 bZIP peptides identify atomic interactions between thymidine and a highly conserved asparginase residue AS^,^^) within the basic domain (Blatter et al., 1992). Finally, the crystal structure of the GCN4.DNA complex demonstrates interaction of basic domain amino acid side chains through hydrogen bonds to the DNA phosphodiester backbone and with bases in the major groove of the GCN4 binding site (Ellenberger et al., 1992). Importantly,

10341

10342 NF-IL6 Dyptic Core Domain

the underlined amino acids within the GCN4 sequence R 2 3 2 M - T E m S 2 4 3 participate in this interaction with DNA and are highly conserved in bZIP family members, al- though subtle differences in base contacts may exist between the bZIP peptides (Johnson, 1993).

The leucine zipper is an amphipathic coiled-coil dimerization domain (Landschultz et al., 1988b; Kouzarides and Ziff, 1988; O'Sheaet al., 1991; Vinson et al., 1993; Ellenbergeret al., 1992). Precise orientation and spacing between the basic region and leucine zipper is necessary for DNA binding. The only tolerated amino acid substitutions between these two motifs are those that produce an integral number of a-helical turns (Pu and Struhl, 1991). Moreover, alterations in the angle at which the two basic domains are fured in space alters DNA binding speci- ficity (Cuenoud and Schepartz, 1993). These data argue that the leucine zipper also participates in DNA binding by correctly positioning the two basic regions for contact with adjacent half- site DNA.

Despite the marked primary amino acid homology found within the basic domain of bZIP proteins, distinct bZIP proteins recognize unrelated DNA motifs. NF-IL6, for example, recog- nizes redundant binding sites loosely fitting within the consen- sus sequence 5'-TT/GNNGNAAT/G-3' (Akira et al., 1990), dis- tinct from the recognition sites of GCN4 and AP-1. Are these distinct binding specificities the consequence of uniquely folded basic domains or of an altered spatial register imposed by the leucine zipper?

To answer these questions, we found it necessary to define the minimal boundaries of the basic amino acid-rich domain of NF-IL6. We engineered T, promoter/polymerase-based expres- sion vectors that produce high levels of the recombinant COOH-terminal domain of NF-IL6. Using the homogeneous protein, we probed the DNA-binding region using controlled endoprotease digestion to identify solvent-accessible domains. When NF-IL6 is bound to DNA before trypsin digestion, a tryp- tic core domain (TCD) is produced. We identify residues at the NH, terminus of the TCD as important in controlling DNA complex stability, termed the complex stabilizing subdomain (CSSD). Potential mechanisms for CSSD influence on bZIP DNA binding activity are discussed.

MATERIALS AND METHODS

Restriction enzymes and T4 DNA ligase were obtained from Boeh- ringer Mannheim and used according to the manufacturer's recommen- dations. Taq DNA polymerase was obtained from Promega Biotechnol- ogy. IPTG (50% w/w dioxane) was obtained from U. S. Biochemical Corp. Q-Sepharose, S-Sepharose, Mono-S column, and resins were pur- chased from Pharmacia LKB Biotechnology Inc.

Oligonucleotides Polyacrylamide gel-purified oligonucleotides corresponding to the rat

angiotensinogen AF'RE (Brasier et al., 1990a) and the human IL-6 gene promoter (Akira et al., 1990) were annealed and used for DNA binding assays. The sequences used are tabulated below (underlines designate changes from wild type sequence, and asterisks indicate methylation interference contact points) for the angiotensinogen promoter se- quences:

* * GATCCACCACAGTTGGGATTTCCCAACCTGACCA APRE WT

GTGGTGTCAACCCTAAAGGGTTGGACTGGTCTAG

* * GATCCACCACAGTTGTGATTTCACAACCTGACCA APRE M6

GTGGTGTCAACACTAAAGTGTTGGACTGGTCTAG

GATCCACCACATGTTGGATTTCCGATACTGACCA AF'RE M2 GTGGTGTACAA_CCTAAAGGCTaGACTGGTCTAG

and for IL-6 promoter sequences:

* * GATCCGGACGTCACTTGCACAATCTTAATAA NF-ILGWT

GCCTGCAGTGAACGTGTTAGAATTATTCTAG

GATCCGGACGTCACACTACAAACTCTTAATAA NF-IL6M GCCTGCAGTGEATGTTTGAGAATTATTCTAG

Expression of NF-IL6 bZIP Domains (NFBDs)

For construction of the fusion protein NF-IL6(199-345), NF-IL6 cod- ing sequences were amplified in the polymerase chain reaction, using as

GGC TTC CCG TAG GCG CTG CGC GCT TAC CTC GGC TA-3' (bold an upstream primer, the sequence 5"AGGTTA ACC ATGGCG GCG

ATG indicates initiation codon, and underline corresponds to NcoI re- striction site). The downstream primer contained the sequence 5'-GCG GAA ITC AAG CTT CTAGCAGTG GCC GGA GGAGGC-3' (containing the underlined HindIII restriction site downstream of the stop codon in bold type). The polymerase chain reaction product was digested with the restriction endonucleases NcoI and HindIII, gel-purified, and li- gated into the pRSETB expression plasmid (Invitrogen, $an Diego, CA). In this expression vector the NF-IL6 coding sequences are placed down- stream of the polyhistidine leader sequence. Sequence identity for each expression vector was confirmed by dideoxy sequencing.

NFBDs were expressed in the T7 promoter/polymerase expression plasmid pET3d (Studier et al., 1990). NcoI-BarnH1 restriction frag- ments containing the indicated coding sequences were produced by polymerase chain reaction amplification using the following upstream primer oligonucleotides: 5'-GTG GAC AAG CCC ATG GAC GAG TAC AAGATC CGG CGC GA-3' for NFBD (272-345) (bold ATG indicates the

AGG ATT ACC ATG GCC GTG GAC AAG CAC AGC GAC GAC-3' for initiation codon, and the underline corresponds to the NcoI site); 5'-

AAG AAG-3' for NFBD (259-345). The downstream polymerase chain reaction primer used was 5'-AAG GCG GGG GGA TCC TAG CAG TGG CCG GAG GAG GCG AGC-3' (the underline indicates the BarnHI site, and bold type is the stop codon).

For recombinant protein production, the appropriate plasmid DNA was transformed into the T, expression host E. coli BL21(DE3) pLysS (Studier et al., 1990). E. coli were grown until A,,, reached 0.6-1.0 for small scale inductions and 4.0 for large scale fermentation. Recombi- nant protein was produced upon addition of 1 rn IPTG for 2-i h at 37 "C .

NFBD (266-345); 5'- CGC CCC TCC ATG GTC AAG AGC AAG GCC

Harvest and Purification of NF-IL6 Protein The cell pellet from 4 liters of bacterial culture was suspended in 250

ml of BufferA(20 rn Tris-HC1, pH 7.0,0.5 rn EDTA, 0.5 rn DTT, 1 m~ phenylmethylsulfonyl fluoride, and 1 pg/ml pepstatin A) containing 100 rn NaCl and subjected to freeze-thaw lysis. After centrifugation at 15,000 rpm for 30 min, the supernatant fraction was layered onto a Q-Sepharose fast flow column (40-ml bed column) attached in tandem to an S-Sepharose fast flow column (40-ml bed volume) equilibrated in Buffer A with 100 rn NaCl. The NFBDs come out in the Q-Sepharose flow-through fraction and bind avidly to the S-Sepharose resin. Elution of NFBDs was accomplished with a gradient of 0.1-1 M NaCl in Buffer A at 1.0 d m i n . Peak fractions containing NFBD were pooled and dialyzed against Buffer B (10 rn Tris-HC1, pH 8.0, 0.5 m~ EDTA, and 0.5 rn DTT) containing 100 rn NaCl and layered on a Mono-S column (HR 5/5) equilibrated with Buffer B containing 100 rn NaCI. After washing, NFBD was eluted with a gradient of 100-1000 m~ NaCl in buffer B at 1.0 ml/min.

Purified peptides were monitored by Coomassie Brilliant Blue stain- ing after separation by 20% SDS-polyacrylamide gel electrophoresis (Giulian, 1985). NH,-terminal sequence analysis was performed by elec- trotransfer of purified or proteolyzed proteins onto polyvinylidene di- fluoride membranes as described (Kumar et al., 1993). The concentra- tion of the proteins was determined using the protein assay kit from Bio-Rad (Bradford, 1976).

Domain Mapping Studies Domain mapping studies were carried out using the proteases tryp-

sin and Endoproteinase Asp-N as described below. 12ypsin-Digestion with trypsin was carried out in the absence or

presence of an equimolar amount of specific or nonspecific DNA. Typi- cally, 15 pg of polypeptide fragment was incubated in 25 rn Tris-HC1, pH 7.5, 50 m~ NaCl, 0.1 rn EDTA, and 0.1 rn DTT with trypsin in a total volume of 40 pl at S:E ratio 1O:l at 25 "C for varying times. 6.5 pl of reaction mixture was taken out at each time point and mixed with

NF-IL6 l lypt ic Core Domain 10343 SDS-PAGE sample buffer. Samples were applied on 20.1% SDS-PAGE (Giulian, 1985).

Endoproteinase Asp-N-Proteolytic cleavage of NF-IL6 peptides (10 pg20 p1) in 25 m Tris-HC1, pH 7.5.300 rn NaCl, 0.1 m~ EDTA, 0.1 m D'IT was carried out at 25 "C with Asp-N (1OO:l or 250:1, w/w) for indicated times. The reaction was terminated by the addition of SDS- PAGE sample buffer and electrophoresed on a 20.1% SDS-PAGE.

DNA Binding Assays Electrophoretic gel mobility shift assays were performed with the

double-stranded APRE M6 oligonucleotide radioactively labeled with [CY-~~PI~ATP by Klenow DNA polymerase to a specific activity of lo6 cpdpmol. 20,000 cpm were used in each binding reaction constituting 10 fmol of DNA probe in a total volume of 20 pl as described (Brasier et al., 1990a). Scatchard analysis was performed using the gel shift assay by incubation of increasing concentrations of input DNA with 50 ng of purified NFBD peptide. For this analysis, binding volume was in- creased to 48 pl and contained 1 mgml BSA as a nonspecific carrier. Bound and free DNA complexes were separated and quantitated by exposure of the gel to a Molecular Dynamics PhosphorImager screen. DNase I footprinting was performed using a DNA probe containing the rat angiotensinogen promoter labeled uniquely at nucleotide -526 (non- coding strand) or alternatively -615 (coding strand), using T4 polynucleotide kinase and [Y-~~PIATP as previously described (Brasier et al., 1990a; Ron et al., 1990a). Exonuclease 111 footprinting reactions were performed by incubating 5,000 cpm of uniquely end-labeled DNA fragment (1 h o l ) with equivalent amounts of BSA or NFBD peptide (100 ng), for 20 min a t room temperature in a total volume of 50 pl containing 50 m~ NaCl, 10 m Tris HCI, pH 8.0, 1 m D'IT (Wu, 1985). One pg of denatured herring sperm DNA was used as a nonspecific competitor in each reaction. After binding, samples were adjusted to a final concentration of 5 m~ MgCI,, and freshly diluted exonuclease I11 was added to each sample at the indicated concentration for a 10-min digestion a t 22 "C. Reactions were pheno1:chloroform-extracted, fol- lowed by ethanol precipitation of the aqueous phase, and resolved on a 9% polyacrylamide/8 M urea sequencing gel in 1 x TBE (Tris-borate- EDTA) buffer (Brasier et al., 1990a; Ron et al., 1990a).

Nitrocellulose filter binding assays were performed using the indi- cated homogenous fractions of NFBD peptides. Peptide was added to a known concentration of APRE M6 or NF-IL6 oligonucleotide in 5O-pl final volume of binding buffer containing 10 m~ Tris-HCI, pH 7.4,50 m KCI, 1 mM D'IT, 0.1 rn EDTA, (Riggs et al., 1970a, 1970b; Widen and Wilson, 1991). The precise concentration of binding sites was deter- mined as described (Riggs et al., 1970a). At saturation, typically 30% of the input label was bound to the nitrocellulose filter. The off-rate, k,, was determined by using 4 x 10"' M binding activity and 8 x 10"' M probe in the same binding buffer. Samples were allowed to bind for 20 min until equilibrium was reached and then a 1000-fold excess of oli- gonucleotide was added as a trap for disassociated protein. At the in- dicated times, samples were filtered, washed, and quantitated as above.

RESULTS

Overexpression of NF-IL6 (199345) COOH-terminal Domain-Coding sequences for residues 199-345 from the NF- IL6 COOH terminus, containing the last in-frame initiator me- thionine and the bZIP motif, were cloned into an expression vector designed to produce a fusion protein with multiple NH,- terminal histidine residues termed NF-IL6 (199-345). By den- sitometric analysis of total E . coli protein lysates after SDS- PAGE, the NF-IL6 (199-345) fusion protein constitutes 10- 30% of the total protein after 4 h of IPTG induction (not shown).

Although the predicted mass of the fusion protein is 20.2 kDa (4-kDa polyhistidine leader and 16.2-kDa NF-ILG), the appar- ent size of the expressed protein was -27 kDa. To exclude the possibility of potential cloning artifacts, the NH, terminus was determined to be MRGSH, exactly corresponding to the pre- dicted NH,-terminal sequence of the fusion protein. Moreover, we produced and sequenced fragments of NF-IL6 (199-345) by digestion with Endo-Asp-N, trypsin (see below), and V8 pro- teases, and all sequences were exactly the predicted sequence of the NF-IL6 protein. These data indicate the authenticity of the recombinant protein. The purified NF-IL6 protein repre- sented >90% of the Coomassie-stained material on an over- loaded SDS-PAGE (Fig. 1B, lane 1 ).

Bound -

Free

30-

21 5 - 14.3-

6.5-

3.4-

3c

- + NF-ILB(199-345) cds

B _ _ APRE M6 APRE M2

- 1 i

TIME: 0 5 15 30 60 5 15 30 60 90 5 15 30 60 90

shift analysis using 0.1 pg of NF-IL6 (199-345) protein binding to 10 FIG. 1.A, sequence-specific binding of NF-IL6 (199-345). Gel mobility

fmol of 32P-labeled APRE M6 DNA. In the indicated lanes, unlabeled APRE WT, M6, M2, NF-kB ( K ) , NF-IL6 WT ( N ) , or NF-IL6 mutant (NM) oligonucleotides were included in the binding reaction a t a 1000- fold molar excess as competitors. Bound and free complexes indicated a t left are resolved after nondenaturing polyacrylamide gel electrophoresis and visualized after autoradiographic exposure. B, identification of NF- IL6 TCD. NF-IL6 (199-345) protein alone or separately equimolar ra- tios of protein and APRE DNA were mixed and allowed to reach binding equilibrium. Trypsin was then added at 20:l S:E ratio, and timed ali- quots (indicated at the top of the gel) were resolved on a 20.1% SDS- PAGE and stained with Coomassie Blue. Migration of molecular weight markers is indicted on the left. On the right, the asterisk represents a 20-kDa APRE M6 DNA-dependent unstable fragment, the large hori- zontal arrow represents an 8.8-kDa tryptic core domain, and the small horizontal arrow represents a 5-kDa leucine zipper fragment

By using fractions containing homogeneous NF-IL6 (199- 345), sequence-specific DNA binding was demonstrated by com- petition in a gel shift assay using, as a radiolabeled probe, the NF-IL6 binding site from the rat angiotensinogen gene APRE M6 (Fig. IA). The APRE wild type (WT), APRE M6, and the interleukin-6 WT ( N ) oligonucleotides competed for DNA bind- ing, yet site mutations of these respective oligonucleotides (APRE M2, NF-IL6 mutant) and the unrelated NF-KB oligo- nucleotide ( K ) failed to compete. These data indicate that the 147 amino acid COOH-terminal peptide of NF-ILG contains sequence-specific DNA binding activity representative of the mature NF-ILG protein.

Identification of the NF-ZL6 TCD-Controlled trypsin prote- olysis of the NF-IL6 (199-345) was used to define the minimal bZIP DNA binding element. Trypsin was chosen to probe the

10344 NF-IL6 Dyptic Core Domain

bZIP element because 1) the NF-IL6 basic region contains nu- merous potential trypsin cleavage sites and 2) trypsin prote- olysis is sensitive to changes in protein secondary structure upon nucleic acid binding (Taniuchi et al., 1969; Shuman et al., 1990). In this assay, the NF-IL6 protein was incubated sepa- rately in the absence of DNA, the presence of nonspecific oli- gonucleotide DNA, or the presence of specific oligonucleotide DNA and then subsequently challenged with the endoprotein- ase trypsin. Tryptic-digested peptides were analyzed by SDS- PAGE (Fig. lB), and parallel digestions were transferred to polyvinylidene difluoride membranes for NH,-terminal micro- sequencing. In the absence of DNA, NF-IL6 (199-345) was cleaved rapidly to small peptides; only a -5-kDa peptide con- taining the leucine zipper could be transiently visualized. In contrast, proteolysis of the peptide-APRE M6 DNA mixture produced a trypsin-resistant 8.8-kDa peptide, stable for incu- bations up to 90 min. NH,-terminal sequencing indicated that the 8.8-kDa peptide had three NH, termini, Ala2=, LysZG, and Th?66, corresponding to three adjacent lysine residues in the basic amino acid-rich region of NF-IL6 (residues 262,264, 265; also Fig. 8). Based on their size and lack of cleavage sites at the COOH terminus, these peptides contain complete COOH-ter- minal extensions of the NF-IL6 protein. We designate the shortest of the peptides, residues 266-345, as the TCD. The TCD is also formed in the presence of the APRE M2 oligonucle- otide, indicating an interaction of NF-IL6 (199-345) with this site-directed mutation, which has no detectable binding in the gel shift (Fig. IA) or transactivation assays in vivo (not shown).

Overexpression of the NFBD-Residues 259-345 of the NF- IL6 bZIP domain were expressed as a nonfksion (native) pro- tein. This NH, terminus was chosen to include the three tryp- sin cleavage sites identified for NF-IL6 (199-345); this smaller peptide deletes the aberrantly migrating proline/alanine-rich domain in the NF-IL6 amino terminus and so allows accurate interpretation of the size of proteolytic domains. To illustrate sequence-specific DNA binding, Fig. 2A demonstrates that E. coli extracts expressing NFBD (259-345) produced a specific DNase-1 footprinting activity after IPTG induction over the rat angiotensinogen promoter APRE (nucleotides -528 to -557). Fig. 2B demonstrates that this footprinting activity could be competed using oligonucleotides containing the NF-IL6 binding site from the rat angiotensinogen promoter (APRE WT, APRE M6) and the human IL-6 gene promoter (NF WT) but not by oligonucleotides containing site-directed mutations. We con- clude that NFBD (259-345) binds specifically to the oligonucle- otide DNA binding sites in a manner indistinguishable from

Fig. 3A demonstrates that the TCD is produced after trypsin proteolysis of NFBD (259-345) when complexed to APRE DNA binding sites. In the absence of DNA, NFBD (259-345) was rapidly cleaved by trypsin to COOH-terminal leucine zipper peptides, as seen with NF-IL6 (199-345). In the presence of APRE M6 DNA, the NH, terminus of NFBD (259-345) was cleaved into peptides labeled A, B, and C (Fig. 3A) . The TCD (fragment A) was detected as early as 5 min and remained stable from proteolytic digestion for up to 120 min. Fragment A contains the same three NH, termini (Ala263, LYS'~~, and Th?66) as the NH, termini produced by proteolysis of NF-IL6 (199- 345). The 8-kDa peptide labeled fragment B contains the same NH, terminus as fragment A. Based on its size and identical NH, terminus, fragment B probably represents a COOH-ter- minal truncation at G1n333 (within the leucine zipper). Frag- ment C contains the NH, terminus Va1303.

To determine whether the TCD is unique to APRE binding sites, trypsin proteolysis of NFBD.DNA complexes was per- formed after binding to the NF-IL6 DNA binding site derived from the IL-6 gene. Fig. 3B demonstrates that the same three

NF-IL6 (199-345).

proteolytic fragments were produced as with the angiotensino- gen APRE (Fig. 3A) . We also note a transient stabilization of the TCD and leucine zipper during digestion in the presence of NF-IL6 mutation DNA (Fig. 3B) as was observed with APRE M2 DNA(Fig. 3A). As additional controls, we used an unrelated DNA sequences in the trypsin proteolysis assay (Fig. 3C). The angiotensinogen glucocorticoid response element and mutant (Brasier et al., 1990b) as well as single-stranded DNA were preincubated with NFBD (259-345). Trypsin digestion did not produce a TCD peptide but did partially stabilize the leucine zipper residues as compared with no DNA (although to a lesser extent than the APRE and NF-IL6 sequences (Fig. 3A). These data indicate that the TCD is protected from proteolysis pre- dominately by interaction with transcriptionally active DNA templates, to a lesser extent with the corresponding point mu- tations, and cannot be visualized with unrelated DNA.

Identification of NH,-terminal Deletions of the NF-ZL6 bZZP Region by Proteolysis-Endoprotease Asp-N was used to pro- duce NH,-terminal-deleted domains within the NFBD (259- 345) peptide (Fig. 4). Under the conditions used, Endoprotease Asp-N hydrolyzes peptide bonds NH, terminus to aspartic acid residues and was chosen because the NF-IL6 bZIP domain is enriched in aspartic acid residues. Although the digestion pat- tern by Endo Asp-N was not different in the absence or pres- ence of DNA (data not shown), we observed differential sensi- tivity to proteolysis in the basic region. For example, hydrolysis of NFBD at Asp268 could not be detected. Endo-Asp-N hydroly- sis of NFBD (259-345) first produces an NH,-terminal trunca- tion of 14 amino acids from the parent peptide (fragment A, residues 272-3451; further cleavage generates a 32 amino acid NH,-terminal deletion (fragment B, residues 290-345). DNA binding assays using either SDS-PAGE-eluted or native pep- tides prepared by column chromatography indicated that only fragment A (NFBD (272-34511, but not fragment B, bound to APRE DNA (data not shown).

Expression of NFBDs in E. coli.-To allow systematic study of the effects of progressive NH,-terminal truncation of the NF-IL6 bZIP element on DNA binding, expression vectors en- coding NFBD (266-345) and NFBD (272-345) were constructed to produce these peptides at high levels (Fig. 5 A ) .

Homogenous NFBD peptides were assayed in parallel for APRE DNA binding activity (Fig. 5B). Binding activity pro- duced by NFBD (259-345) and NFBD (266-345) peptides in the gel mobility shift assay were competed using APRE M6 but did not recognize APRE M2 oligonucleotide competitors. Although NFBD (272-345) recognizes APRE M6 with a lower affinity, this peptide still retains binding specificity because it is not competed by APRE M2 site mutation. That the DNase I foot- print patterns were also qualitatively similar indicates that the recombinant NFBDs interact with APRE binding site DNA in a sequence-specific fashion. In the DNase I footprint assay shown (Fig. 5C) , the borders of the footprinted nucleotides appear somewhat shorter for NFBD (266-345) and NFBD (272-345) than for NFBD (259-345).

Because the borders of the DNase I footprint assay are dif- ficult to precisely define, the processive exonuclease (Ex0 ZZZ) was used to resolve whether the NH,-terminal truncations of NFBDs affect the length of DNA that can be protected by the protein. Fig. 5D illustrates the results of Exo I11 footprinting assays on the coding strand as compared with the DNase I- footprinted region. All of the NFBD peptides terminate Exo I11 digestion at nucleotides -535, -536, and -537 on the coding strand and at nucleotides -545, -547, and -549 on the noncod- ing strand. We conclude that the recombinant NFBDs bind to DNA in a sequence-specific fashion over the same nucleotide region of the angiotensinogen APRE.

NFBD (272345) Binds DNA with Lower m n i t y -

t - I

""""

-526> -526> - "

FIG. 2. A, NF'BD (259-345) induction produces specific APRE DNA binding activity. E. coli BL21 (DE3) pLysS transformed with T, promoter/ polymerase expression vector encoding NFBD (259-345) were cultured in the absence (-) or presence (+) of 1 rn IPTG to induce protein expression. After a 2-h induction, bacteria were harvested and soluble protein extracted (see "Materials and Methods"). 0 (-), 1, 5, and 25 pg of extract was then incubated with an end-labeled DNA fragment containing APRE M6 sequences in the rat angiotensinogen (rAT) promoter and digested with DNase I. After resolution on a 9% polyacrylamide-urea sequencing gel the dried gel was exposed for autoradiography. As little as 1 pg of induced E. coli extract produced a DNase I footprint over the APRE sequences (nucleotide -557 to -528). B, sequence specificity of NFBD (259-345) DNase I footprinting activity. Crude E. coli extracts of induced NFBD (259-345) expression vectors were incubated with 50,000 cpm of end-labeled rAT promoter in the absence or presence of a 1000-fold excess ofAPRE WT (WT), APRE M6 ( M 6 ) , APRE M2 ( M 2 ) , NF-IL6 WT (NF), or NF-IL6 mutant ( N M ) unlabeled oligonucleotides and samples were processed for DNase I footprinting activity as described above. The footprinting activity is lost with coincubation of APRE WT, APRE M6, and NF-IL6 WT oligonucleotides.

Equilibrium disassociation constants (K,) were measured for the homogeneous peptides NFBD (266-345) and NFBD (272- 345) for binding to the APRE DNA using the method of Scatchard (Scatchard 1949) (Fig. 6). Saturation plots demon- strate that at lower DNA concentrations, a linear relationship existed between input DNA and bound DNA, but at higher concentrations apparent binding saturation had been reached. In three independent measurements of NFBD (266-345) the Kd was 36 2 8 n~ (X 2 S.D.), whereas the Kd of NFBD (272-345) was 283 -c 160 (X 2 S.D.), representing an -8-fold difference between the two peptides (p < 0.05, two-tailed t test). These results are consistent with the observations shown in Fig. 5B

that NFBD (272-345) binds to DNA with a lower affinity than NFBD (266-345).

NFBD (272345) Forms Unstable DNA Complexes-The ki- netics of DNA interaction by the recombinant NFBDs were quantitated by nitrocellulose filter binding experiments (Riggs 1970a, 1970b). On-rate measurements for all of the NFBD pep- tides were rapid and indistinguishable, with a t , of -5 s. The disassociation rate constants, k, were determined by chal- lenging bound protein-radiolabeled DNA complex with a 1000- fold molar excess of unlabeled DNA to trap disassociated pro- tein at the indicated times before filtration (Fig. 6B). DNA binding by the TCD and longer proteins was extremely stable,

10346

sis of NFBD (250345). NFBD (259-345) FIG. 3. Controlled trypsin proteoly-

was purified to homogeneity and incu- bated in the absence (-) or presence of oligonucleotide binding sites. After reach- ing binding equilibrium, endoprotease trypsin was added a t a 1O:l substrate: enzyme ratio, and a t the indicated times samples were quenched in 1 x SDS-PAGE sample buffer. Tryptic peptides were re- solved on a 20.1% SDS-PAGE and stained with Coomassie Blue. A, tryptic proteoly- sis without (-) or with APRE binding sites. No DNA (-1, APRE M6, or the non- specific competitor APRE M2 were sepa- rately preincubated with homogeneous NFBD(259-345) and subjected to tryptic proteolysis for 5-120-min incubations as indicated a t the top of the gel. M , molecu- lar weight markers; NF, undigested NFBD (259-345) peptide. In the absence of DNA, a 5-kDa peptide is produced (fragment C , VaI3O3). Fragment A, an 8.8- kDa peptide containing the NF-IL6 TCD is produced predominately in the pres- ence of APREM6 and to a lesser extent APREM2. Fragment A contains three NH, termini Ala2-, LYS*~', and Th?@. Fragment B in the APRE M6 digestion

fragment A. B , the NF-IL6 TCD is pro- lane contains the same NH, terminus as

duced on the interleukin-6 DNA binding site. Tryptic proteolysis of NFBD (259- 345) preincubated in the absence (-) or presence of either NF-IL6 WT or NF-IL6 mutant (Mut) binding sites for times (in min) indicated at the top of the gel. M , molecular weight markers; NF, undi- gested NFBD (259-345) peptide. Frag- ments A, B, and C are the same as de- scribed in Fig. 3 A . An identical pattern of tryptic peptides are produced on the NF- IL6 templates as APRE templates. C, trypsin proteolysis with a variety of unre- lated DNA fragments. Angiotensinogen glucocorticoid response element I double- stranded DNA and single-stranded DNA were used in the tryptic proteolytic assay. In all of these DNA samples, NFBD (259- 345) is digested only into fragment C (leucine zipper peptides), and no produc- tion of fragments A or B were observed.

46-

30-

21.5.

14.3.

6.5-

3.4-

30-

21.5-

14.3-

6.5-

3.4-

30-

21.5-

14.3-

6.5-

3.4-

B "" NF-IM WT NF-IM Mut

M NF 5 15 30 60 90 120 5 15 30 60 90 120 5 15 30 60 90 120

4 A * B

- C

C AGRE AGRE Mut Sinele-Stranded

M NF 5 15 30 60 90 120 5 IS 30 60 90 120 5 IS 30 60 90 120

4 c

46

30-

21.5-

14.3-

6.5-

3.4-

NF-IL6 l lyp t ic M 15 30 60 120 Minutes

CA

C B

Fragment A Dl‘ E Y K I R R E R

Fragment B: DY K V K M R N L E

259 I * v V

V K S M KKTVD KHSDE YKIRR ERNNI AVRKS RDKVK MRNLE ... FIG. 4. Endoprotease AspN proteolysis to identify additional

NH,-terminal deletions of NF-IL.6 bZIP region. NFBD (259-345) was digested with Endoprotease Asp-N at a 250:l substrate:enzyme ratio, and digests were visualized by SDS-PAGE with Coomassie Blue staining. Endo Asp-N digestion produced an 8-kDa peptide, fragment A, containing a 14 amino acid deletion spanning residues 272-345. Diges- tion for longer times produces 6-kDa fragment B, containing residues 290-345. At the bottom of thepanel, the NFBD (259-345) NH, terminus is depicted. The large arrow indicates the cleavage site at Asp272; smaller arrow, Aspm; asterisk at predicted site (Aspzm) that was not observed to be digested.

with half-lives of 37.5-45 min, resulting in a calculated k,, of -3 x s. Recombinant NFBD (272-3451, in contrast, had a half-life of DNA binding of 145 s and a calculated k,, of

Because recombinant NFBD (272-345) contains an initiator methionine residue that could influence its DNA binding prop- erties, NFBD (272-345) was produced and purified after pre- parative Endo-Asp-N proteolysis. This peptide then lacks the initiator methionine residue. The half-life of native NFBD (272-345) was even slightly faster than recombinant NFBD (272-345) at 80 s (k,, = 8.6 x s), indicating that the ini- tiator methionine has, if anything, a stabilizing role in DNA binding.

NFBD (272345) Is Not Protected by DNA in the lkypsin Proteolysis Assay-Comparison of the stability of NFBDs in the trypsin proteolysis assay indicates that the Endo Asp-N do- main, NFBD (272-345), behaves markedly differently than the longer NFBDs. The TCD, like NFBD (259-3451, was protected from proteolysis by preincubation with APRE M6 or NF-IL6 WT oligonucleotide binding sites (Fig. 7). In marked contrast, NFBD (272-345) was rapidly degraded by trypsin into leucine zipper peptides only, even in the presence of the specific APRE M6 oligonucleotide at concentrations far above the Kd for DNA binding (Fig. 7B 1.

5 x 10-3 s.

Core Domain 10347

DISCUSSION

By using controlled tryptic digestions of free and DNA-com- plexed NF-IL6 peptides, we have identified the minimal bZIP domain that exhibits indistinguishable DNA binding properties as the mature protein; this TCD spans amino acids 266-345 (Fig. 8). Endo-Asp-N proteolysis produces NFBD (272-345) in the absence or presence of DNA. NFBD (272345) forms un- stable, low affinity DNA complexes.

Production of the TCD is a specific indicator of protein-DNA interaction, because it is dependent on the presence of duplex DNA binding sites and is most stable in the presence of tran- scriptionally active DNA templates. Resistance of the DNA- complexed TCD to trypsin digestion is probably the conse- quence of steric interference of DNA to limit the protease access to the basic lysine and arginine residues or the consequence of conformational changes of the TCD upon DNA binding. Circu- lar dichroism (CD) measurements of the bZIP peptides of C/EBPa, GCN4, and AP-1, in the absence of DNA, are charac- teristic of proteins containing partial a-helical content (O’Neil et al., 1991; Pate1 et al., 1990; Weiss et al., 1990). Upon addition of cognate DNA binding sites, the shape and magnitude of the mean residue ellipticity at 222 nm is increased, indicating that the a-helical content of the peptide complexed to DNA is dra- matically increased. For C/EBPa, this DNA-dependent confor- mational change, in part, results in resistance to tryptic diges- tion (Shuman et al., 1990). Whether conformational changes are soley responsible for changes in tryptic digestion patterns has not been resolved. Irrespective of its mechanisms, this sen- sitive tryptic proteolysis assay indicates that NF-IL6 interac- tion with DNA is not just a simple binary “all-or-none” event. NF-IL6 interacts with the rAT APRE and IL-6 point mutants, as attested by the transient production of the TCD, but NF-IL6 does not stably bind in a manner sufficient for detection in conventional gel shift or DNase I footprint assays. A second event, such as a conformational change on the DNA, may be necessary for stable complex formation.

Controlled trypsin proteolysis also provides additional infor- mation about the structure of the NF-IL6 peptide complexed to DNA. For DNA-complexed NFBD (259-3451, containing 22 pre- dicted trypsin cleavage sites, the earliest time point of trypsin incubation produces digestion at Alaza, L y P , and Th?%, in- dicating that the peptide NH, terminus is not tightly folded and is freely solvent-accessible. Subsequent cleavage sites in the DNA-complexed peptide are detected between the basic region- leucine zipper boundary (Lys302) and within the leucine zipper itself ( L Y S ~ ~ ~ ) . Thus, the functional activities in the bZIP ele- ment corresponding to DNA binding (basic region) and dimer- ization (leucine zipper) can be identified as distinct domains by this proteolytic assay.

The tryptic proteolytic fragments we produce of NFBD.DNA complexes are markedly different than those reported for C/EBPa.DNA complexes (Shuman et al., 1990). Trypsin prote- olysis of C/EBPa.DNA complexes produces preferential cleav- age at G1uZw and AsnZg2, downstream of two highly conserved arginine residues and corresponding to a predicted NH,-termi- nal cap for the basic domain a helix (Shuman et al., 1990). Although these amino acids are absolutely conserved within the NF-IL6 basic region (see Fig. €9, we do not observe stable proteolytic peptides with these NH, termini. We note that the E. coli expressed C/EBPa peptides used in this report corre- spond to NH,-terminal deletions shorter than the TCD; it will be of interest to test longer C/EBPa peptides to determine its domain structure. Our observations on DNA-complexed NF- IL6 are also consistent with recently reported crystallographic data of the homologous bZIP peptide GCN4 (Ellenberger et al., 1992). In the GCN4.DNA crystal, the peptide monomer is one

10348

46-

30-

21.5- 14.3-

6.5-

1 2 3 4 - ""

NF-IL6 l lyp t ic Core Domain

B COMPETITOR: APRE M6

.. 1 4 16 64 .. 1 4 16 €4 APRE M2

C

4- NFBD25S345

-NFBD 272-345 -NFBD 26-

D

COMPETITOR: APRE M6 APRE M2 . - 1 4 1 6 6 4 - 1 4 16 €4

COMPETITOR: APRE M6 APRE M2 - 1 4 1664.- 1 4 16 64

"""- .. .

""" .- -" " - _" "_ -. _" " -

" . .

-61 5

FIG. 5. A, expression of NFBDs. Aphotograph of a 20.1% SDS-PAGE gel stained with Coomassie Blue after electrophoresis of crudeE. coli lysates expressing nonfusion NFBDs is shown. Lune 1, molecular weight standards; lune 2, NFBD (259345); lane 3, NFBD (266-345); lune 4, NFBD

NF-IL6 Dyptic Core Domain 10349

FIG. 6. A, comparison of equilibrium disassociation constants of NFBD pep- tides. Gel mobility shift assays of a con- stant amount NFBD (266-345) or NFBD (272-345) in the presence of increasing concentrations of APRE M6 DNA were performed to quantitate separately the bound and free fractions. Top left, satura- tion curves for NFBD (266-345) binding. Top right, Scatchard plot for the same ex- periment. The methods of least squares was used to derive the linear regression B/F = 2.66 - 0.0362(B) (RZ = 0.907), result- ing in an estimated Kd of 28 m. In three independent experiments, the Kd of NFBD (266-345) binding was measured to be 36 f 8 I ~ M ( x f S.D.). Bottom left,

NFBD (272-345) binding was more shal- saturation curve for NFBD (272-345).

was less evident than the same plot for low, and the curve inflection at saturation

NFBD (266-345), even at 700 m DNA concentration, indicating a qualitative difference in DNA binding. Bottom right, Scatchard analysis of NFBD (272-345). Least squares regression identified the linear relationship B/F = 1.769 -.00452(B) (R2 = 0.945) and an estimated Kd of 221 m. The flattened slope of the NFBD (272- 345) Scatchard plot resulted in a greater scatter in the Kd measurements. In three independent experiments, the NFBD (272-345) Kd was measured to be 283 f 160 m ( x * S.D.) ( p < 0.05 by two-tailed t test). B, NFBD (272-345) forms unstable DNA complexes. Off-rates of NFBD (266- 345) and NFBD (272-345) were compared with full-length NF-IL6 expressed in E. coli using nitrocellulose binding assay (see “Materials and Methods”). The lines plotted represent the linear regression equations derived for each peptide, with each point representing the mean of four

tein. M-NFBD (272-345) represents data independent time courses for each pro-

from recombinant protein and NFBD (272-345) data from the Endo-Asp-N do- main produced by preparative proteolysis of NFBD (259-345). Deletion of six amino acids between 266 and 272 of the NF-IL6 bZIP domain destabilizes DNA com- plexes.

A 60 -

Saturation plots:

50 -

40 -

$ 3 0 - m

20 -

10 -

-

0 I I I I I I I I I

0 20 40 60 80 100 120 140 160 180 200 Total (nM)

300

250

g 200

150

100

50

0

B

NFBD (272-345)

/”- /J

I I I I I 1 I

0 100 200 300 400 500 600 700 800

4 Scatchard plots:

NFBD (266-345) 1 y = 2.66 - 0.362(~)

Y Kd = 28 nM . ,

0- 0 10 20 30 40 50 60

3

3 2 Y

0

3

. Li3

1

0

T

i

Bound (nM) I

NFBD (272-345)

y = 1.769 - O.O0452(x) 1

\ Kd = 221 nM

I I I I I I 0 50 100 150 200 250 300

Total (nM) Bound (nM)

6 2.2

2.1 - Off-rate determination

2.0

1.9

8 1.8

8 8

c

1.7

-I 1.6

1.5

1.4 1 1 p-NFBD(272-345)

ANFBD (24-345)

NFBD (266-345) 7 ‘

1.3 I ’ ’ I I I I I I

0 10 20 30 40 50 60 70

Time (minutes)

(272-345). The expressed proteins are indicated by the arrows; by densitometric analysis, expression levels varied from 10-30% of the total protein within the E. coli extract. NH,-terminal sequence of NFBD (259-345) was v59KS; NFBD (266-345) was A2@VD; NFBD (272-345) was MDZ7,EY. B, gel shift assays using NH,-terminal-deleted NFBDs. An autoradiogram of a gel mobility shift analysis using homogeneous NFBD peptides binding toAPREM6 DNAin the absence or presence of unlabeled competitor DNA. 50 ng of NFBD (259-345) indicated as NFBD, NFBD (266-345) indicated as NFBDAK, and NFBD (272-345) indicated as NFBDAD14 were competed with increasing amounts of APRE M6 or the site mutation APRE M2. Increasing amounts of competitor DNA from 1 to 64 pmol were added simultaneously with 10 fmol of probe to the NFBD peptide, corresponding to a 100-6400-fold molar excess. NFBD (272345) competes with a lower aflinity than the longer NFBD peptides for APRE M6 binding. C, DNase I footprint assays using NH,-terminal-deleted NFBDs. Recombinant NFBD peptides were used in DNase I footprint assays using rAT promoter sequences containing the APRE M6. A photograph of an autoradiogram is shown. Lep. h?ne is a Maxam-Gilbert “G” reaction, second lune is DNase I ladder in the absence of protein. Proteins indicated are: NFBD, NFBD (residues 259-345); NFB K, NFBD (residues 266-345); NFB D, NFBD (residues 272-345). The indicated APRE sequences spanning nucleotide -557 to -528 are protected by all of the recombinant NFBDs. D, exonuclease I11 footprint assays of the NFBD peptides on the coding strand of the rAT promoter. Parallel reactions used 1 fmol of probe with 100 ng of control BSAor NFBD peptide. Lanes I and 2, DNase I digestion of the rAT coding strand in the absence (-) or presence (+) of NFBD (259-345) lane 3, Maxam-Gilbert “G” ladder; lane 4, unmodified probe. The rAT coding strand was then incubated separately with either BSA, NFBD (259-3451, NFBD D, or NFBD (272-345) and digested with 30-120 units of exonuclease I11 (Ex0 IZI). Indistinguishable minor and major Exo I11 stops (arrows) are produced by the NH,-terminal truncated NFBD (272-345), as by NFBD (259-345).

10350 NF-IL6 l lyp t ic Core Domain smooth continuous a helix between amino acids 226-281 in the GCN4 protein. In the absence of sharp bends in the homologous basic region of the NF-IL6 protein, there would be no prefer- ential tryptic cleavage sites at the residues homologous to

A

DNA NF-116 APRE M6 APRE M2

TlME(MIN): 5 15 60 120 5 15 60 120 5 15 60 120

B DNA: - APRE M6 APRE M2

nME(M’N): 0 5 15 30 60 90 5 15 30 60 90 5 1 5 30 Bo 90 ~~ ~

C

FIG. 7. Tryptic proteolytic assay of recombinant NFBDs. A, the TCD is protected from tryptic digestion by NF-IL6 binding sites. 450 pmol of dimeric TCD, NFBD (266-3451, were preincubated with 450 pmol of either NF-IL6 WT, APRE M6, or APRE M2 DNA and subjected to timed tryptic protease digestion using 1O:l S:E in 29 pl as indicated at the top of the gel. After analysis by SDS-PAGE and Coomassie stain- ing of tryptic peptides, we observe that identical tryptic fragments were produced as for NFBD (259-345) (Figs. 3, A X ) . Fragment B represents a COOH-terminal deletion at GIn3”. Fragment C begins a t Val”’. B, NFBD (272-345) is sensitive to tryptic digestion in the APRE.DNA complex. 550 pmol of dimeric NFBD (272-345) were preincubated with 550 pmol of duplex DNA binding sites and subjected to tryptic digestion a t 1O:l S:E in 34 p1 for times indicated at top. Only fragment C, corre- sponding to Val”’ of the leucine zipper domain, was protected. Protec- tion of the basic domain was not observed in the presence of DNA, probably due to its rapid off-rate from the liganded complex.

C/EBPa G1uZ9O and AsnZg2 (NF-IL6 residues G1uZ7’ and Asn2*l). Within the NH,-terminal basic region of NF-IL6, two func-

tionally distinct subdomains can be identified, an NH,-termi- nal region that confers DNA-complex stability and a COOH- terminal region essential for DNA binding specificity (Fig. 8). The CSSD is a novel bZIP DNA-binding determinant localized to 6 amino acids between residues 266-272. Evidence that pep- tides lacking the CSSD form unstable DNA complexes comes from equilibrium disassociation constant, nitrocellulose filter off-rate, and tryptic digestion assays.

Evidence that the CSSD is contained within an independ- ently folded subdomain comes from the controlled protease di- gestions. Either in the presence or absence of DNA, Endo Asp-N fails to cleave at aspartic acid Asp268 within the CSSD; simi- larly, V8 protease does not cleave at predicted sites adjacent to the CSSD (Figs. 4 and 8 and data not shown). The resistance of this domain to cleavage may be the consequence of a tightly folded secondary structure and, with both proteases, is inde- pendent of the presence of DNA. The residues contained within the CSSD (TVDKLS) are predicted to be an a-helix contiguous with the basic domain a-helix using the secondary structure prediction rules of Gamier et al. (1978). Thus, the CSSD may be contained within a stably folded a-helical domain.

The CSSD may affect DNA-protein interaction through sev- eral potential mechanisms. The most likely mechanism is that the CSSD is involved directly in DNA contact and, through these interactions, stabilize the bound complex. Although the exonuclease I11 footprinting data demonstrate that the nucle- otides protected by NFBD (2723451, lacking the CSSD, are identical to those protected by NFBD (266-345), containing the CSSD, Exo I11 has strand-displacing activity. By using this assay, we cannot exclude the presence of either nonspecific phosphate backbone or low affinity DNA interactions. Alterna- tively, the CSSD could regulate complex stability by altering the conformation of the bZIP domain. Spectroscopic determi- nations of NFBD protein conformations, containing or lacking the CSSD, will be informative to differentiate between these two mechanisms.

A surprising difference in the conservation of the CSSD can be observed among the cloned C/EBP family members. The amino acids contained within the CSSD are highly conserved with the rodent NF-IL6 homologs, liver-enriched transcrip- tional activator protein, and C/EBPP (Descombes et al., 1990; Cao et al., 1991). CEBPP, the mouse homolog of NF-IL6, is highly conserved in the CSSD. The CSSD of C/EBPa is also 100% conserved considering conservative amino acid substitu- tions (4 of 6 exact amino acids). Although NF-IL6p and its murine homolog C/EBPG isolated from 3T3-Ll adipocytes share

U

CSSD

I I LEUCINE ZIPPER

I I

ENDO-ASP-N

I I

TRYPTIC CORE DOMAIN (TCD)

FIG. 8. Sequence of 8.8-kDa NF-ILG tryptic core peptide bZIP domain. Controlled trypsin proteolysis of recombinant NF-IL6 has defined a minimal bZIP DNA-binding domain. The NF-IL6 TCD spans amino acid residues T h P - c y ~ ~ ‘ ~ and produces DNA binding activity with identical sequence specificity and equilibrium binding constants as the mature protein. The three major DNA-dependent trypsin digestion sites of the basic domain are indicated by the lurge vertical arrows. Additionally, tryptic digestion of protein-DNA complexes produces a fragment with an amino terminus at Val”O”. This result identities the boundary of the COOH-terminal leucine zipper domain, indicated by the medium vertical arrow. Endo-Asp-N digests only six additional amino acids from the trypsin core peptide at a cleavage site indicated by the small vertical arrow. Although NFBD (272-345) interacts with DNA in a sequence-specific fashion, the disassociation rate constant is markedly affected. This short sequence of amino acids spanning 266-272, termed the CSSD, are involved in stabilizing the half-life of the DNA-protein complex.

NF-IL6 Tbyptic Core Domain 10351

87% exact amino acid identity in the TCD, the sequences of these proteins diverge within the CSSD (Cao et al., 1991; Ki- noshita et al., 1992). NF-IL6P and CEBPG contain 2 proline and 2 glycine residues; both of these amino acids are unfavor- able for a-helix formation. Moreover, Cao et al. (1991) observed that C/EBPG has a 10-fold lower disassociation constant than for CEBPP. This lower binding constant may be due to disrup- tion of the predicted a-helical structure of the NF-IL6 CSSD and may be one mechanism for heterogeneity in DNA binding by the C/EBP bZIP protein family.

The definition, by partial endoprotease digestion, of a mini- mal NF-IL6 tryptic core domain has important implications for structure-function studies of this bZIP protein. Structural stud- ies of shorter NH,-terminal-deleted subdomains may not yield information relevant to the mature peptide. Determination of the conformation of the TCD and that of NFBD (272-345) by spectroscopic means will provide important clues as to the role of the CSSD on the conformation of the basic domain. These results will contribute new information on the interaction of the acute-phase inducible NF-IL6 transcription factor with binding sites in inducible gene promoters.

Acknowledgments-We thank Shizuo Akira for the gift of CMV(+)NF-

vectors and bacterial hosts, Johnny W. Peterson for help with large IL6 plasmid, William H. Studier for T, promoter/polymerase expression

ing, Samuel H. Wilson for helpful discussions, David Konkel for review scale bacterial fermentation, Stefan Serabyn for protein microsequenc-

of the manuscript, and Becky Soliz for expert secretarial assistance.

REFERENCES

Agre, P., Johnson, P. F., and McKnight, S . L. (1989) Science 246,922-926 Akira, S . , Isshiki, H., Sugita, T., Tanabe, O., Kinoshita, S., Nishio, Y., Nakajima, T.,

Blatter, E. E., Ebright, Y. W., and Ebright, R. H. (1992) Nature 369, 65M52 Bradford, M. M. (1976) Anal. Biochem. 72, 24S254 Brasier, A. R., Ron, D., Tate, J. E., and Habener, J. F. (1990a) EMBO J. 9, 3933-

Brasier, A. R., Ron, D., Tate, J. E., and Habener, J. F. (1990b) Mol. Endocrinol. 4,

Hirano, T., and Kishimoto, T. (1990) EMBO J. 9, 1897-1906

3944

Cao, Z., Umek, R. M., and McKnight, S . L. (1991) Genes & Deu. 6,1538-1552 Chang, C. J., Chen, T. T., Lei, H. Y., Chen, D. S . , and Lee, S.-C. (1990) Mol. Cell Biol.

Cuenoud, B., and Schepartz, A. (1993) Science 269,510-513 Descombes, P., and Schibler, U. (1991) Cell 67, 569-579 Descombes, P., Chojkier, M., Lichtateiner, S. , Falvey, E., and Schibler, U. (1990)

Ellenberger, T. E., Brandl, C. J., Struhl, K., and Harrison, S . C. (1992) Cell 71,

Gamier, J., Osguthorpe, D. J., and Robson, B. (1978) J. Mol. Bid. 120,97-120 Giulian, G. G., Shanahan, M. F., Graham, J. M., and Moss, R. L. (1985) Fed. Proc.

Johnson, P. F. (1993) Mol. Cell. B i d . 13, 69194930 Kinoshita, S. , Akira, S. , and Kishimoto, T. (1992) hoc. Natl. Acad. Sci., 89, 1473-

Kouzarides, T., and Ziff, E. (1988) Nature 336, 646-651 Kumar, A., Kim, H.-R., Sobol, R. W., Becerra, S . P., Lee, B.-J., Hatfield, D. L.,

Landschulz, W. H., Johnson, P. F., Adashi, E. Y., Graves, B. J., and McKnight, S . L.

Landschulz, W. H., Johnson, P. F., and McKnight, S . L. (1988b) Science 240,1759-

Landschulz, W. H., Johnson, l? F., and McKnight, S . L. (1989) Science 243, 1681-

O'Neil, K. T., Shuman, J. D., Ampe, C., and DeGrado, W. F. (1991) Biochemistry 30,

OShea, E. K, Klemm, J. D., Kim, P. S. , and Alber T. (1991) Science 264,539-544 Patel, L., Abate, C., and Curran, T. (1990) Nature 347,572-575 Poli, V., Mancini, F. P., and Cortese, R. (1990) Cell 63, 643453 P u , W. T., and Struhl, K (1991) P m . Natl. Acad. Sci. U. S. A. 88, 69014905 Riggs, A. D., Suzuki, H., and Bourgeois, S . (1970a) J. Mol. B i d . 4 8 , 6 7 4 3 Riggs, A. D., Bourgeois, S., and Cohn, M. (1970b) J. Mol. Biol. 63,401417 Ron, D., Brasier, A. R., Wright, K A,, Tate, J. E., and Habener, J. F. (1990a) Mol.

Scatchard, G. (1949) Ann. N. Y Acad. Sci. 61, "72 Shuman, J. D., Vinson, C. R., and McKnight, S . L. (1990) Science 249, 771-774 Studier, F. W., Rosenberg, A. H., Dunn, J. J., and Dubendofl, J. W. (1990) Methods

Taniuchi, H., Moravek, L., and Anfinsen, C. B. (1969) J. Bid. Chem. 244, 4600-

Vinson, C. R., Hai, T., and Boyd, S . M. (1993) Genes & Deu. 7, 1047-1058 Weiss, M. A., Ellenberger, T., Wobbe, C. R., Lee, J. P., Harrison, S. C., and Struhl,

Widen, S . G., and Wilson, S . H. (1991) Biochemistry 30, 6296-6305 Williams, S . C., Cantwell, C. A., and Johnson, P. F. (1991) Genes & Deu. 6, 1553-

Wu, C. (1985) Nature 317, 84-87

1921-1933

10,66424653

Genes & Deu. 4, 1541-1551

1223-1237

44,686

1476

Suhadolnik, R. J., and Wilson, S . H. (1993) Biochemistry, 32,746G7474

(1988a) Genes & Deu. 2, 78-00

1764

1688

903C9034

Cell. B i d . 10, 1023-1032

Enzymol. lSS,60-89

4606

K. (1990) Nature 347,575578

1567