THE JOURNAL OF BIOLWICAL CHEMISTRY Vol. 269, No. 33, Issue of August 19, pp. 21110-21116, 1994 Printed in U.S.A.

The Leucine Zipper Is Necessary for Stabilizing a Dimer of the Helix-Loop-Helix Transcription Factor USF But Not for Maintenance of an Elongated Conformation*

(Received for publication, March 17, 1994, and in revised form, May 13, 1994)

Emery H. Bresnick and Gary FelsenfeldS From the Laboratory of Molecular Biology, NIDDK, National Institutes of Health, Bethesda, Maryland 20892

The basic helix-loop-helix transcription factor USP3 binds to E box motifs on certain promoters and enhanc- ers, as well as the P-globin locus control region. We have used gel filtration chromatography, velocity centrifuga- tion, and chemical cross-linking methods to investigate the stoichiometry and shape of USF4S in solution and when boynd to DNA. USFa has a very large Stokes’ ra- dius (44 A) and a high frictional ratio (1.64), consistent with an asymmetric elongated oligomer. Under a variety of conditions, the only detectible USFa species in solu- tion and bound to DNA is a dimer. The carboxyl-terminal leucine zipper is absolutely essential for a stable dimer but not for the elongated conformation. We used a pro- tease footprinting assay to demonstrate that, when USFQ binds to DNA, a 45i-kDa U S P domain becomes resistant to cleavage with trypsin. This domain includes sequences that are not expected to interact with the DNA helix, suggesting that trypsin cleavage sites are masked by a conformational change. Our results show that the oligomerization state of USP3 does not change upon binding to DNA, and the helix-loop-helix oligomer- ization motif of USF43 is not itself sufficient to form a high affinity dimerization interface.

The P-globin locus control region (LCR)’ is a cluster of four erythroid-specific DNase I-hypersensitive sites at the 5’ end of the P-globin locus that are necessary for correct developmental and tissue-specific expression of globin genes (1-3). The human LCR was initially identified by its ability to confer copy num- ber-dependent and integration site-independent expression on a linked P-globin gene in transgenic mice (3). Another impor- tant feature of the LCR is that its transcriptional activation property can be shared by multiple promoters on a single chro- mosome (4).

Two of the hypersensitive sites, HS2 and HS3, also have LCR activity when linked individually to globin genes and inte- grated into the chromosome (5, 6). While it is known that nu- merous erythroid-specific and ubiquitous transcription factors bind to HS2 and HS3 (6-8), the detailed composition and stoi- chiometry of the protein components of these heteromeric com-

payment of page charges. This article must therefore be hereby marked * The costs of publication of this article were defrayed in part by the

“uduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Biology, NIDDK, NIH Bldg. 5 , Rm. 214, Bethesda, MD 20892. Tel.: $ To whom correspondence should be addressed: Lab. of Molecular

The abbreviations used are: LCR, locus control region; bHLHzip, basic helix-loop-helix leucine zipper; bp, base pair; BSA, bovine serum

site; IPTG, isopropyl thio galactoside; USF43, upstream stimulatory fac- albumin; DTT, dithiothreitol; HLH, helix-loop-helix; hss, hypersensitive

tor; ALZUSF43, USF43 mutant that lacks the leucine zipper; A l - 180USF43, USF43 mutant that lacks amino acids 1-180.

301-496-1898; Fax: 301-496-0201.

plexes are unresolved. It is clear, however, that no single factor is responsible for the LCR activity of HS2 (9), suggesting that a functional LCR requires multiple interacting components.

We recently purified one of the ubiquitous factors that binds with high affinity to an E box in HS2 and identified it as the bHLHzip transcription factor, USF43 (10). USF3 was initially purified from HeLa cells (11) and was shown to be necessary for optimal transcription from the adenovirus major late promoter (12-14). We observed a strong temperature dependence to USF43 DNA binding and suggested that this reflects either an intramolecular rearrangement or a change in the oligomeriza- tion state of USF43 (10). To assess these possibilities, and to begin to analyze the structure of the HS2 complex, we investi- gated the oligomerization state and shape of USP3 in solution and when bound to HS2.

bHLHzip proteins have been reported to exist as monomers, dimers, and tetramers in solution and as dimers or higher order oligomers when bound to DNA (15-18). The bHLHzip proteins contain a basic region that determines DNA binding specificity (191, and at least two motifs that mediate protein-protein in- teractions, the HLH and the leucine zipper (for review, see Ref. 20). One might expect that both of these regions would be involved in the formation and/or stabilization of protein oli- gomers. In the case of the bHLH proteins myogenin and MyoD, however, it has been reported that stable dimers and tetramers can form even though they lack leucine zippers (21, 22). This suggests that the HLH motif, alone, can be sufficient to form a high affinity dimerization interface.

It was initially reported that USF43 purified from HeLa cells had a sedimentation coefficient of 3.6 S , consistent with the -40-kDa monomer observed on SDS-polyacrylamide gels (23). Subsequently, evidence for oligomerization of recombinant USP3 was obtained by cotranslating full-length and truncated USF43 polypeptides (24). The DNA complex formed with the cotranslated product had a mobility on nondenaturing gels that was intermediate between the complexes formed with wild- type or truncated USF43 alone. Taken together with experi- ments in which purified USF43 was cross-linked to form dimers (25), it is clear that USF43 can form oligomers in solution and when bound to DNA.

Mutational analysis of the DNA binding activity of USF43 showed that deletion of the leucine zipper resulted in a polypep- tide with very weak DNA binding activity (24). As the DNA complex formed with the mutant protein had an identical mo- bility on nondenaturing gels as the wild-type USF43-complex, it was concluded that the mutant and wild-type proteins bind to DNA with an identical stoichiometry. A different result was obtained with the A1-180USF43 mutant that lacks the amino- terminal 180 amino acids. In contrast to wild-type USF43, de- letion of the leucine zipper from Al-180USF43 had no effect on DNA binding. In certain cases, therefore, the leucine zipper seems to be dispensible for oligomerization and DNA binding.

This study was prompted by our observation that the Stokes

21110

Oligomerization State of USF 21111

radius of recombinant USF43, and USF43 from the K562 human erythroleukemia cell line, was much larger than would be pre- dicted for a globular dimer or tetramer with a 34-kDa monomer subunit. Here, we use gel filtration chromatography, velocity centrifugation, and chemical cross-linking methods to investi- gate the size and shape of USF43 in solution and when bound to DNA.

EXPERIMENTAL PROCEDURES Expression and Purification of Proteins-The expression vectors for

wild-type and leucine zipper-deleted USF43 were kindly provided by Dr. R. Roeder (26). Proteins were overexpressed and purified by a modifi- cation of the previously described method (27). Transformed bacteria were grown to an A,,, of 0.8. IPTG (0.5 mM) was added, and the culture was incubated for 3 h at 37 "C. Cells were harvested, resuspended in 50 mM Tris (pH 8.0), 2 mM EDTA, and 0.1 mg/ml lysozyme (15 ml of bufferhacterial pellet from a 500-ml culture), and incubated for 15 min a t 4 "C. Lysates were prepared by sonicating IPTG-induced bacteria three times for 20 s, and soluble proteins were obtained by centrifuga- tion for 15 min at 12,000 x g. Lysates were fractionated by ammonium sulfate precipitation (35% saturation), and the pellet was resuspended in 4 ml of lysis buffer and recentrifuged for 5 min at 12,000 x g. Am- monium sulfate-precipitated USF43 was purified by two passes over a Superdex 200 prep grade (Pharmacia LKB Biotechnology Inc.) gel fil- tration column which was equilibrated in 10 mM HEPES (pH 7.81, 600 mM KCl, 10% glycerol, 1 mM EDTA and 10 mM DTT. Fractions contain- ing USF43 were pooled and dialyzed against the same buffer containing 60 mM KCl. This material was further purified on an SP-Sepharose fast flow (Pharmacia) column. USF43 was eluted with a 50-500 mM KC1 gradient. Aprotinin, (10 pg/ml), leupeptin (10 pg/ml), and PMSF (0.1 m) were included in all buffers.

SDS-PAGE and Immunoblotting-SDS-polyacrylamide gel electro- phoresis was performed in slab gels according to Laemmli (28). Quan- titative immunoblots were performed as described previously (29) with an anti-human USF43 polyclonal antibody. The antibody was prepared by inoculating New Zealand White rabbits with the purified antigen according to standard protocols. Western blots were incubated with unfractionated a r ~ t i - U S F ~ ~ antisera (diluted to 0.1% in phosphate-buff- ered saline containing 2% BSA and 0.2% Tween 20) for at least 2 h a t 23 "C, followed by 1 h a t 23 "C with 0.2 pCi of lZ5I-protein A (DuPont- NEN). The relative intensity of bands was quantitated with a Phosphor- Imager (Molecular Dynamics).

Nuclear Extract Preparation-Nuclei were isolated from human K562 erythroleukemia cells, and extracts were prepared as described previously (10).

Analytical Gel Filtration Chromatography-A Superdex 200 prep grade column (31 x 0.7 cm) was packed according to the manufacturer's instructions at 4 "C and was equilibrated in 10 mM HEPES (pH 7.8),60 mM KCl, 10% glycerol, 1 mM MgCl,, 10 m~ D l T (DNA-binding buffer). The column was calibrated several times by applying protein standards (10 p1 each of a 10 mg/ml solution) and eluting with equilibration buffer. The following R, values were used for ferritin, catalase, BSA, ovalbu- min, chymotrypsinogen, and cytochrome c standards: R, = 61.0, 52.2, 35.5, 30.5, 20.9, and 16.4 A, respectively (30). The column was run at 4 "C with a flow rate of 0.20 mumin, and 17 drop (480 pl) fractions were collected. Standard proteins were detected by measuring the absorb- ance of eluted material at 280 nm with an on-line absorbance detector. Free USF43 was detected by quantitative immunoblotting as described below. The USF43-DNA complex and free DNA were detected by using 32P-labeled DNA and scintillation counting. The coincidence of the 32P radioactivity with the positions of free DNA or the USF43-DNA complex was confirmed by resolving samples on a nondenaturing polyacrylamide gel. The void volume (V,,) was determined by measuring the eluted volume (V,) with blue dextran. The V, for protein standards, and the V, were used to calculate K,, using the equation K,, = V, - V& - V,. K,, values were plotted against the corresponding Stokes radii (R,) to ob- tain a linear calibration plot that was used to determine the R, for unbound USF43, DNA-bound USF43, and free DNA. In some experi- ments, the column was equilibrated with the same buffer containing 0.15 or 0.4 M KC1. The eluted volumes for standard proteins were iden- tical under both conditions.

Sucrose Gradient Ultracentrifugation-Sucrose gradients (2.6 ml, 520%) were formed in the gel filtration equilibration buffer. Samples (20 pl) were applied with three internal protein standards (8 pl of 10 mg/ml solutions of aldolase, ovalbumin, and cyt c; s ~ , , , ~ = 7.40,3.55, 1.90, respectively). Gradients were centrifuged for 16 h at 4 "C in a Beckman

TL100.2 rotor. Fractions (80 111) were collected from the top, and stand- ard proteins were detected by measuring the absorbance of fractions a t 280 nm. A sample of each fraction (10 pl) was assayed for the presence of USF43 protein by quantitative immunoblotting. The USF43-DNA com- plex and unbound DNA were detected by resolving a sample of each fraction on a 4.5% nondenaturing polyacrylamide gel and analyzing the gel with a PhosphorImager. The known sedimentation coefficients (sz0,,,) for the standard proteins (31) were plotted against the positions of these proteins in the gradient, and this calibration curve was used to calculate szo,w values for free USF43, DNA-bound USF43, and free DNA. The small corrections for differences in partial specific volume between the calibration proteins and the samples were omitted.

Calculation of Native Molecular Weight-The following equation was used to calculate the native molecular weight of free and DNA-bound USF43 from experimentally determined R, and szo,w values (32): sZn,=, = M(l-Vp)/6~qRfl~. The frictional ratio ( f l f , ) was calculated according to the following formula: f l f , = R,(~GM/~PN)"~. Values used for the partial specific volume of free DNA, free USF43 and the USF43-DNA complex were 0.520, 0.722, and 0.680, respectively (33, 34).

Glutaraldehyde Cr~ss-Zinking-USF~~ (200 ng) was incubated in DNA binding buffer containing 4 pg of poly(d1-dC), 10 mM DTT, with or without a stoichiometric equivalent of a 22-bp double-stranded USF43 DNA-binding site (10). Glutaraldehyde solutions were prepared by di- lution in water immediately prior to performing the cross-linking reac- tion. Reactions were incubated for 10 min at 23 "C and were terminated by adding 1 p1 of 1 M lysine. Samples were boiled in SDS-sample buffer and resolved on SDS-polyacrylamide gels with a &15% gradient of polyacrylamide.

Protease Sensitivity of Unbound and DNA-bound USP3-DNA bind- ing reactions contained USF43 (300 ng), 4 pg of poly(d1-dC) and 10 mM DTT with or without a 2-fold stoichiometric excess of a 22-bp double- stranded USF43-binding site in a final volume of 20 pl. Reactions were carried out for 10 min a t 23 "C. Aliquots of unbound or DNA-bound USF43 (4 pl) were incubated in 10-1.11 reactions with 0, 250, or 750 ng of TPCK-treated trypsin (Worthington) for 4 min at 23 "C. Reactions were terminated by adding 1 pl of TLCK (1 mg/ml), followed by boiling in an equal volume of SDS-PAGE sample buffer. Samples were resolved on 15% SDS-polyacrylamide gels, and proteins were detected by quantita- tive immunoblotting with anti-USF43 antiserum.

RESULTS AND DISCUSSION

USP3 Exists as an Elongated Dimer in Solution and When Bound to DNA-When we purified recombinant USF43 by gel filtration chromatography we discovered that the Stokes' ra- dius (R,) was much greater than would be predicted for a glob- ular monomer, dimer, or tetramer with a 34-kDa subunit. To investigate the basis for the large R,, and to ask if USF43 un- dergoes a change in oligomerization state upon binding to DNA, we analyzed free and DNA-bound USF43 by gel filtration chromatography and sucrose gradient centrifugation. The R, and the sZO,+ values can be used to calculate the native molec- ular weight of a macromolecule, as well as information about its shape.

Fig. 1 shows an SDS-polyacrylamide gel of the purified re- combinant proteins used in this study. The elution profile of the complex of USF43 with a 22-bp E box oligonucleotide (closed circles) on a Superdex 200 column is shown in Fig. 2 A . The complex eluted as a homogeneous species between the two marker proteins, ferritin and catalase. The free DNA (open circles) eluted as a much smaller species between ovalbumin and chymotrypsinogen. A representative standard curve is shown in Fig. 2C; the R, values for the USF43-DNA complex and the free DNA are 51 2 2.8 and 21 2 1.2 (mean 2 standard error, n = 41, respectively.

The gel filtration properties of free USF43 in the absence of DNA were also analyzed. A quantitative immunoblot assay was used to detect free USF43 in the column fractions. The free USF43 eluted from the column as a large species (Fig. 2B) , in a manner that was almost indistinguishable from that of the USF43-DNA complex. The R, value for free USF43 is 44 2 6.0 (mean 2 standard error, n = 4). As the globular proteins ferritin and catalase have molecular masses of 450,000 and 240,000

21112 Oligomerization State of USF

116- 85 - '$ 56- 39 -

5 27- X

1 2 3 Fro. 1. SDS-PAGE analysis of wild-type and mutant USF'

polypeptides. Purified protein samples were resolved on a 12% SDS- polyacrylamide gel and stained with Coomassie Blue. Lane 1, molecular weight markers; lane 2, 4 pg of wild-type USFd3; lune 3, 4 pg of ALZUSP3.

Da, respectively, and the USP3 monomer is 33,517 kDa, the gel filtration results suggest that both free and DNA-bound USF43 exist as large oligomeric complexes or that their shapes deviate significantly from a prototypical globular protein.

We previously demonstrated that recombinant human USP3, and the USP3 present in nuclear extracts from K562 erythro- leukemia cells, formed identical complexes with an E box oli- gonucleotide. Fig. 3 shows a gel filtration analysis of the USP3 that is present in crude K562 nuclear extracts at a very low concentration. Identical to the purified recombinant protein, the endogenous K562 USF43, in a mixture of many other nuclear components, also exhibits a large R, (48.7 A). The large R,, there- fore, is not unique to the recombinant protein and does not re- flect the self-association of USF43 a t higher concentrations.

The s20,ur value for recombinant USF43 was measured by su- crose gradient centrifugation. Three standard proteins were included in each gradient, and the positions of these proteins were used to generate a standard curve. As shown in Fig. 4A, the position of the USF43-DNA complex (closed circles) in the gradient was similar to the position of the ovalbumin standard which had an szoS of 3.55. Fig. 4B shows the profile of free USP3 in a separate gradient. A representative standard curve is shown in Fig. 4C. The s20,w values for free USP3, the USP3- DNAcomplex, and free DNAwere 3.2 * 0.15,3.5 * 0.15, and 2.5 * 0.25 (mean * standard error, n = 3).

The R, and s20,w values were used to calculate the native molecular weight of free USP3 and the USP3-DNA complex (Table I). The calculated values of 57,320 and 64,360 Da for free and DNA-bound USP3, respectively, are consistent with the major species being a dimer in both cases; the data are incon- sistent with higher order oligomers such as tetramers. As the calculated value for the DNA complex is lower than the theo- retical value (79% of the theoretical dimer), the data do not unequivocally distinguish between monomers and dimers. The lack of monomers, however, was ruled out by the glutaralde- hyde cross-linking experiment described below. The R, and ac- tual M, values were used to calculate the frictional ratio ( f l f , ) . The f / f o values for free and DNA-bound USP3 were 1.64 and 1.82, respectively, consistent with elongated molecules that de- viate significantly from perfect spheres.

The Leucine Zipper Is Necessary for a Stable U S P Dimer But Not for an Elongated Conformation-To investigate the role of the leucine zipper in the elongated shape and oligomer- ization state of USF43, we examined the physical properties of a mutant USP3 polypeptide that lacks the carboxyl-terminal

2ooo t

C

c. v) 0.8-

C ' 0.6-

9)

.-

.- > ' - a 0.4- L

d 0.2 -

0.0; =. = ' = .z .' . ' ' ' . ' ' ' 5 10 15 20 25 30

Fraction Number I

IO

Fro. 2. Gel filtration chromatography of free and DNA-bound USPs. A, the complex of USP3 with a 22-bp oligomer. Complexes were formed between purified recombinant USP3 (200 ng) and a 32P-labeled 22-bp USP3 DNA-binding site. The USF43-DNA complexes (0) or free DNA (0) were chromatographed on a Superdex 200 analytical gel fil- tration column as described under "Experimental Procedures." Samples of each fraction were counted for 32P, and a representative column profile is shown in panel A. V,, void volume; F, ferritin; C , catalase; B, bovine serum albumin; 0, ovalbumin; Ch, chymotrypsinogen; Cy, cyto- chrome c. B, free USP3. Purified recombinant USF43 (200 ng) was chromatographed as described above. Samples of alternate fractions were assayed for the presence of USP3 protein by immunoblotting as described under "Experimental Procedures." C , calibration curve. The K,, values for the protein standards shown in panel A (0) were plotted against their respective R, values to yield a linear calibration plot. The open symbols show the positions of free DNA(O), free USF43 (A), and the USP3-DNA complex (0) from the experiment shown in panel A. The inset shows the R, values from four independent experiments (mean * standard error).

zipper. We overexpressed ALZUSP3 in bacteria as described previously (27) and purified the recombinant protein to appar- ent homogeneity (Fig. 1). Our initial attempts to analyze ALZUSP3 by gel filtration chromatography and sucrose gradi- ent centrifugation were confounded by its propensity to aggre- gate in low salt buffers. In the buffer containing 60 mM KC1, that was used for the analysis of the wild-type protein, all of the mutant protein eluted in the void volume of the Superdex 200 column. This problem was overcome by performing the chro- matography in the presence of at least 150 mM KCl.

Wild-type and ALZUSP3 were chromatographed on the Su-

Oligomerization State of USF 21113

A

B 0

';\

1 ; 0.01 '= "= . ' . ' = * = 5 =' = ' '

0 5 10 15 20 25 30 Fraction Number

FIG. 3. USW in K562 nuclear extracts has an identical Stokes radius as recombinant USP'. K562 nuclear extract (40 pl) was chro- matographed on a Superdex 200 analytical gel filtration column as described under "Experimental Procedures." Aliquots of alternate frac- tions were analyzed for the presence of USF" protein by immunoblot- ting. The immunoblot is shown at the top, and the USF43 bands corre- spond to the fractions indicated on the plot. The inset shows theR, value that was calculated from this experiment.

L.

\

perdex 200 column, which was equilibrated in buffer contain- ing either 150 or 400 mM KCl. As shown in Fig. 5, ALZUSF43 has a signipcantly smaller R, than the wild-type protein (32 versus 44 A) in both low and high salt buffers. Under both conditions, the R, values for the wild-type protein were identi- cal to the values obtained in the experiments of Fig. 2 in which the buffer contained 60 mM Ksl. Analogous to the large R, of wild-type USF43, the R, of 32 A for the ALZUSF43 monomer is also disproportionately large given an actual molecular mass of 27,817 Da .

The R, and s ~ , , ~ values (Fig. 6) were used to calculate the native molecular weight for ALZUSP' (Table I). In contrast to wild-type USP3, which is a dimer, the calculated value of 24,080 is consistent with a monomer. The flf, value of 1.60 suggests that ALZUSF43 also exists in an elongated conformation.

Analysis of USP3 Stoichiometry by Chemical Cross-link- ing-As the molecular weights calculated from the gel filtra- tion and sedimentation experiments are slightly less than would be predicted for dimers, we used chemical cross-linking as an alternate assay to confirm our conclusions from the cal- culations. The cross-linking experiments were performed under buffer and protein concentration conditions that were identical to the gel filtration conditions. We treated free USF43, the USF43-DNA complex, or ALZUSF43 with increasing concentra- tions of glutaraldehyde and analyzed the size of the cross- linked products by quantitative immunoblotting. As shown in Fig. 7A, all of the USF43 monomer was cross-linked to form a -8O-kDa species, regardless of whether USP3 is first bound to DNA. The concentrations of glutaraldehyde required to cross- link all of the free and DNA-bound USF43 were similar. In contrast to wild-type USF43, only trace amounts of a putative dimer were observed with ALZUSF43 after treatment with very high concentrations of glutaraldehyde. We have used multiple approaches to show that USF43 exists as an elongated dimer in the presence or absence of DNA. The dimer is likely to be a very stable core unit. Thus, we cannot rule out the possibility that dimers interact weakly with each other to form tetramers or higher order structures. The oligomerization state of USF43 in

1.0-

.g 0.8 - .- c . 3 Q) 0.6

- " > . c d 0.4 - U ' a,

O.O' = .= ; . Ib 1: 2. =io= .= 25 =' =do I

0 ! . \

Fraction Number

USF 3.21 +/- 0.09 complex 3.54 +I- 0.15

O A " ' . ' . " " 5 10 15 20 25 Fraction Number

FIG. 4. Sucrose gradient centrifugation of free and DNA-bound USPS. A, the complex of USP3 with a 22-bp oligomer. Complexes were formed between purified recombinant USFJ3 (200 ng) and a 32P-labeled 22-bp USF43 DNA-binding site. The USP'-DNA complexes (0) or free DNA (0) were fractionated on a 5-20% sucrose gradient as described under "Experimental Procedures." Fraction 1 represents the top of the gradient. Aliquots of alternate fractions were assayed for the presence of the USF4'-DNA complex or free DNA by resolving on a nondenatur- ing polyacrylamide gel. B, free USF4$. Purified recombinant USY3 (200 ng) was analyzed as described above. Aliquots of alternate frac- tions were assayed for the presence of USF43 protein by immunoblot- ting. C, calibration curve. The spo,w values for protein standards (0) were plotted against the fraction number to obtain the linear calibra- tion plot. The open symbols show the positions of free DNA (0) and the USP3-DNA complex (0) from the experiment shown in panel A. The inset shows the s20,w values from three independent experiments (mean * standard error).

solution seems to differ from the bHLHzip proteins TFE-B and c-Myc, which are capable of forming stable tetramers (15, 16).

As shown in the experiments of Figs. 5-7, an intact leucine zipper is absolutely required for the formation andor mainte- nance of stable dimers. Thus, the HLH dimerization motif is apparently inadequate to form a high affinity dimerization in- terface, in contrast to the bHLH proteins which are believed to exist as multimers in the absence of DNA. It is likely that subtle differences in the amino acid composition of the two helices can have major effects on the stability of dimers. I t should be interesting to systematically mutagenize amino acids that form the dimerization interface to ask if mutants can be

21114 Oligomerization State of USF TABLE I

Stoichiomety of free and DNA-bound USF The Stokes radius, spOtw, frictional ratio and partial specific volume

values were determined as described under “Experimental Procedures.” The theoretical molecular weight values for wild-type USF and the USF-DNAcomplex assume a homodimer and a homodimer bound to one molecule of DNA, respectively. The theoretical molecular weight value for ALZUSF assumes a monomer. The measured molecular weight for a given sample was divided by the respective theoretical molecular weight to yield the percentage value indicated below.

s m w theoretlcal measured f’fo Mr. M,

Wild-type USF 44 3.2 0.72 67,030 57,320 (86%) 1.64 USF-DNAcomplex 51 3.5 0.68 81,590 64,360 (79%) 1.82 ALZUSF 32 1.8 0.72 27,820 22,900 (83%) 1.60

A

”-

0.6

0.4

0.2

1 ; i i : . . . .

. . . . i :

i : . . . . . . . . . . . . . .

/ \{*

Fraction Number FIG. 5. Gel filtration chromatography of a mutant U S P

polypeptide that lacks the leucine zipper. A, immunoblots. Recom- binant wild-type (WT) or leucine zipper-deleted (ALZ) USP3 polypep- tides were chromatographed on an analytical Superdex 200 gel filtra- tion column in the presence of 150 mM KC1 as described under “Experimental Procedures.” Samples of alternate fractions were as- sayed for the presence of USF43 protein by immunoblotting. B and C, elution profiles of wild-type (0) and leucine zipper-depleted (0) USP3 in the presence of 0.15 M ( B ) and 0.40 M (C) KCl. The insets show the R, values that were calculated from this experiment.

isolated that differ in dimer stability, and if the strength of the dimerization interface correlates with biological efficacy.

Identification of a I)ypsin-resistant Domain of U S p - A s the USF43 dimer is the only detectible species under a variety of conditions, i t is unlikely that U S P undergoes a change in oligomerization state upon binding to DNA. It is possible, how- ever, that the dimer undergoes a conformational change as it binds to DNA. Such a conformational transition might not change the overall shape of USF, and, therefore, the hydrody-

‘ 0 5 10 15 20 25 30

-wT

-nu

Fraction Number

U S P polypeptide that lacks the leucine zipper. A, Immunoblots. FIG. 6. Sucrose gradient centrifugation analysis of a mutant

Wild-type USF43 and ALZUSF43 were fractionated on a 5-20% sucrose gradient in the presence of 150 mM KC1 as described under “Experi- mental Procedures”. Aliquots of alternate fractions were assayed for the presence of wild-type (0) or mutant (0) USF43 protein by immunoblot- ting. B, Gradient profiles. The positions of the internal marker proteins in the gradient are indicated at the top. Cyt, cytochrome C (1.90 SI; Ov, ovalbumin (3.55 S);AI, aldolase (7.40 S). The inset shows the s20.w values calculated from this experiment.

A. Wild-Type USF Unbound Complex ,

Glutaraldehyde (Yo) o .001 .01 . I o ,001 .01 . I “ ”

7

S 4 3 - 1 p- e- 1-lx

B. A Leu Zipper USF

29-

1 2 3 4 5 6 7 8

Glutaraldehyde (%) o .001 .01 .I 1 Gri+lx r‘ 43 -

29-

1 2 3 4 5 FIG. 7. Glutaraldehyde cross-linking of free U S P , DNA-bound

USPs and ALZUSI?. A, Complexes were formed between purified recombinant USF43 and a 22 bp USF4’ binding site oligonucleotide. Free and DNA-bound USP3 were treated with the indicated concentrations of glutaraldehyde and then analyzed by immunoblotting as described under “Experimental Procedures”. B, The leucine zipper-deleted mu- tant of USP3 was cross-linked and analyzed as described above.

namic experiments would not necessarily detect the transition. Thus, we employed a more sensitive assay, protease footprint- ing, to investigate this possibility. The underlying assumption of this technique is that if a protein exists in multiple confor- mations, each conformation will have a characteristic sensitiv-

Oligomerization State of USF 21115

A DNA - USF GATA Trypsin (pg) o .25.75.25.75 .25 .75

"- 200 -

B 1 2 3 4 5 6 7 I 1

GGSQRSIAPBTHPYSPKSEAPRITRDE I 172 198

FIG. 8. A trypsin-resistant domain of DNA-bound USF43. A, Di- gestion of free and DNA-bound USF'::' with trypsin. Free USF'::' or USF"' incubated with a stoichiometric equivalent of a USF'" or GATA-1 oligo- nucleotide were digested with trypsin as described under "Experimen- tal Procedures". Samples were resolved on an SDS-polyacrylamide gel, and USF'" was detected by immunoblotting. B, Identification of the amino terminus of the trypsin-resistant USF":' polypeptide. The dia- gram depicts the domain structure of USF":'. stippled box - basic region; open boxes - helical regions; solid box - loop region; 4 vertical lines - leucine residues. The DNA-bound USF'$ fragment that is resistant to trypsin cleavage was isolated from an Immobilon P membrane and the amino-terminal sequence was determined. The underlined sequence (THPYSP) was obtained which is adjacent to R'", a potential trypsin cleavage site.

ity to site-specific proteases. We asked if the accessibility of free and DNA-bound USF"' to trypsin differs.

USF43 was incubated with or without the USF or GATA-1 oligonucleotides and subjected to a short incubation with tryp- sin. As shown in Fig. 8A, the unbound protein (lanes 2 and 3 ) , and the protein incubated with the GATA-1 oligonucleotide (lanes 6 and 71, were digested extensively with trypsin. In contrast, when USFd3 was first bound to a specific oligonucle- otide and then subjected to proteolysis with trypsin, a -15-kDa fragment of USFd3 was observed. The 15-kDa frag- ment was strongly resistant to proteolysis a t additional tryp- tic cleavage sites. The trypsin-resistant fragment was not ob- served when USF43 was incubated with a non-cognate (GATA-1) DNA-binding site. Two additional tryptic fragments migrate as a doublet just below the 15-kDa fragment (see lane 2). As the yield of these fragments varies considerably be- tween samples and is not reproducible, they appear to be un- stable proteolytic intermediates.

To determine the precise site of cleavage that yields the = 15- kDa product, we scaled up the digestion reaction and isolated the fragment by excising it from the Immobilon P membrane. The membrane-bound fragment was subjected to amino-termi- nal sequencing. As shown in Fig. 8B, the underlined sequence (THPYS) was obtained, which was adjacent to a potential tryp- sin digestion site, kg-'*'. Based upon the apparent molecular weight determined from SDS-PAGE, it is likely that the resis- tant fragment extends from threonine 182 to the carboxyl ter- minus of USF.

The 1-180 region is very susceptible to proteolysis, consist- ent with this region being on the surface of the protein and available for interaction with other proteins. Digestion of free USF43 with a broad range of trypsin concentrations (1/100 to 1 .250, t ryp~in/USF~~ (weightlweight)) results in extensive di-

gestion of the intact polypeptide without any stable proteolytic fragments (data not shown). The DNA-binding-induced resist- ance to trypsin cleavage could result from either a direct steric effect of the DNA to block potential cleavage sites or a confor- mational change that masks cleavage sites. As we would expect that the leucine zipper does not contact the DNA helix, and, in fact, probably extends away from the helix (35, 361, and this region is part of the trypsin-resistant domain, it is unlikely that steric blockage by DNA can explain our results. This argument is further strengthened by the fact that the portions of the trypsin-resistant domain on the amino-terminal and carboxyl- terminal sides of the basic region contain multiple potential trypsin cleavage sites and no stable proteolytic products are observed in the absence of a cognate DNA-binding site.

It seems likely that an intramolecular rearrangement occurs as USFd3 binds to DNA, and this results in the masking of trypsin cleavage sites. The rearrangement could be analogous to the disorder to order transition observed for bZip proteins as they bind to DNA(37,38). The acquisition of a-helical structure upon DNA binding of USF was recently demonstrated by cir- cular dichroism, consistent with such a transition (39). One might speculate that the conformational change is itself subject to regulation and is an essential step in DNA binding and transcriptional activation.

Acknowledgments-We are grateful to Dr. R. G. Roeder for gener- ously providing the USF expression vectors and Pat Spinella for performing the amino-terminal sequencing. We thank Dr. Rudolf0 Ghirlando for assistance in data analysis and David Clark, Robert Martin, Jim Omichinski, Moshe Sadowski, and Chuck Vinson for criti- cally reading the manuscript.

REFERENCES 1.

2.

3.

4.

5. 6.

7.

8.

10. 9.

11.

13. 12.

14.

15.

16.

17.

18. 19.

20. 21. 22.

23.

24.

25.

26.

27.

Tuan, D., Soloman, W., Li, Q., and London, I. M. (1985) Proc. Natl. Acad. Sci. U. S. A. 82.6384-6388

Forrester, W. S., Takegawa. S., Papayannopoulou, T., Stamatoyannopoulos, G..

Grosveld, F., Blom van Assendelft, G., Greaves, D., and Kollias, G. (1987) Cell and Groudine. M. (1987) Nucleic Acids Res. 15, 10159-10177

Bresnick, E. H. and Felsenfeld, G. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 51,975-985

Talbot, D., and Grosveld, F. (1991) EMBO J. IO, 1391-1398 Philipson, S., Talbot, D., Fraser, P., and Grosveld. F. (1990) EMEO J. 9,2159- 2167

Talbot, D., Philipsen, S., Fraser, P., and Grosveld, F. (1990) EMEOJ. 9,2169- 2178

Caterina, J. J., Ryan, T. M., Pawlik, K. M., Palmiter, R. D.. Brinster, R. L., Behringer, R. L., and Townes, T. M. (1991)Proc. Natl. Acad. Sci. U. S. A. 88,

Ellis, J., Talbot, D., Dillon, N., and Grosveld, F. (1993) EMBO J. 12, 127-134 1626-1630

Rresnick, E. H., and Felsenfeld, G. (1993) J. Biol. Chem. 268, 18824-18834 Sawadogo, M., Van Dyke, M. W., Gregor, P. D., and Roeder, R. G. (198R)J. Biol.

Sawadogo, &I., and Roeder, R. G. (1985) Cell 43, 165-175 Carthew, R. W., Chodosh, L. A., and Sharp, P. A. (1985) Cell 43, 439448 Miyamoto, N. G., Moncollin, V., Egly, J. N., and Chamhon, P. (1985) EMBO J.

Fisher, D. E., Carr, C. S., Parent, L. A,, and Sharp, P. A. (1991) Genes & Deu.

Dang, C. V., McGuire, M., Buckmire, M., and Lee, W. M. F. (1989) Nature 337,

Hu, Y. F., Luscher, B., Admon.. A,, Mermod, N., and Tjian, R. (1990) Genes &

Peterson, C. L., and Calame, K. (1989) Mol. Cell. Biol. 9, 776-786 Murre. C., McCaw, P. S., Vaessin, H., Caudy, M., Jan, L. Y., Jan, Y. N., Cahrera,

C. V., Buskin, J. N., Hauschka, S. D.. Lassar, A. B., Weintrauh, H., and Baltimore, D. (1989) Cell 58, 537-544

1314-1317

Chem. 263, 11985-11993

4,3563-3570

5,2342-2352

664-666

Deu. 4, 1741-1752

Kadesch, T. (1993) Cell Growth Difi 4.49-55 Farmer, K., Catala, F., and Wright, W. E. (1992) J. Biol. Chem. 267,5631-5636 Anthony-Cahill, S. L., Benfield, P. A,, Fairman, R., Wasserman, %. R., Brenner,

S. L., Altenbach, C., Hubbell, W. L., Stafford, W. F., 111, and Degrado, W. F.

Chodosh, L. A,, Carthew, R. W., and Sharp, P. (1986) Mol. Cell. Biol. 6,4723- (1992) Science 255,979-983

4733 Gregor, P. D., Sawadogo, M., and Roeder, R. G. (1990) Genes & Deu. 4, 1730-

1740 Pognonec, P., Kato, H., and Roeder, R. G. (1992) J. Biol. Chem. 267, 24563- 24567

Kirschhaum, B. J., Pognonec, P., and Roeder, R. G. (1992) Mol. Cell. Biol. 12, 5094-5101

Pognonec, P., Kato. H.. Sumimoto, H., Kretzschmar, M., and Roeder, R. G.

21116 Oligomerization State of USF

28. Laemmli, U. K. (1970) Nature 227,680-685 29. Bresnick, E. H., Dalman, F. C., Sanchez, E. R., and Pratt, W. B. (1989) J. Biol.

30. Gel Filtration: Theory and Practice (1984) Pharmacia LKB Biotechnology Inc.,

31. Sober, H. A. (1970) Handbook of Biochemistry, 2nd ed., pp. C10427, The

32. Siegel, L. M., and Monty, K. J. (1965) Biochem. Biophys. Acta 45, 346-362 33. Zamyatnin, A. A. (1972) Prog. Biophys. Mol. Biol. 24, 109-123

(1991) Nucleic Acids Res. 19,665M650

Chem. 264,4992-4997

pp. 25-27, Uppsala

Chemical Rubber Go., Cleveland

34. Eisenberg, H. (1989) Landolt-Bornstein New Series: Biophysics-Nucleic Acids,

35. Ferre-DAmare, A. R., Prendergast, G. C. , Ziff, E. B., and Burley, S. K (1993)

36. Vinson, C . R. and Garcia, C . K. (1992) New Biologist 4, 396403 37. Patel, L., Abate, C., and Curran, T. (1990) Nature 572574 38. Weiss, M. A,, Ellenberger, T., Wobbe, C. R., Lee, J. P., and Struhl, K. (1990)

39. Feme-DAmare, A. R., Pognonec, P., Roeder, R. G., and Burley, S. K. (1994)

pp. 257-276, Springer Verlag, Berlin

Nature 363,3845

Nature 347,575-578

EMBO J . 13,180-189