Peas te Menan Shouty fr comity and Maula Bs, ts Changes in Superhelicity Are Introduced into Closed Circular DNA by Binding of High Mobility Group Protein /Y* (Received for publication, October 8, 1994, and in revised form, December 27, 1994) Mark S. Nissen} and Raymond Reeves? From the ‘Department of Biochemistry and Biophysics and the §Department of Genetics and Cell Biology, Washington State University, Pullman, Washington 99164-4660 ‘Mammalian high mobility group HMG-VY chromatin proteins bind to the minor groove of A‘T-rich DNA sequences with high affinity both in vivo and in vitro. ‘Topoisomerase I-mediated relaxation assays, analyzed by ono- and two-dimensional agarose gel electrophoresis, in- dicate that binding of recombinant human HMG-LY to closed circular DNA introduces positive supercoils at low protein to nucleotide molar ratios and negative super- coils at higher ratios, This is interpreted to mean that HMG-LY binding initially causes bending of the DNA h lix followed by unwinding of the helix. In contrast, bind- ing of another minor groove binding ligand, netropsin, introduces positive supercoils only. An in vitro produced mutant HMG-LY protein lacking the negatively charged carboxyl-terminal domain binds A’T-rich DNA approxi- mately 1.4-fold better than the native protein, yet it is estimated to be 8-10-fold more effective at introducing negative supercoils. This finding suggests that the highly acidic C-terminal region of the HMG-LY protein may fune- tion as a regulatory domain influencing the amount of topological change induced in DNA substrates by binding of the protein. Footprinting of HMG-IY on negatively supercoiled A:T-rich DNA using diethylpyrocarbonate suggests that the protein is able to recognize, bind to, and alter the conformation of non-B-form DNA. ‘The HMG-VY proteins are small, nonhistone, chromosomal proteins of the “high mobility group” (HMG)." Members of this family include the isoform proteins HMG-1 and HMG-Y (1, 2) and the homologous protein HMG-IC) (8). These proteins are distinguished from other HMG (4) proteins by their ability to recognize and specifically interact with the minor groove of A-T-rich DNA in vitro (6-1). The sequences of HMG-LY respon- sible for the interaction with A-T-rich DNA have been identi fled (8), and results from two-dimensional "H-NMR studies support the predicted netropsin/distamycin-like structure of the DNA binding domains (9). In addition, the A-T minor groove binding ligands netropsin, distamycin, and Hoechst 39258 have been shown to compete with HMG-VY for binding to A‘Prich DNA, suggesting they posses a structure similar to the HMG-I/Y DNA binding domains (8, 10) and may bind DNA in a similar fashion. ‘This work was supported in part by National Institutes of Health Grant 5-R01-A126356 and National Science Foundation Grant DCB- 8904408 (both to RR.) The costs of publieation of this article were defrayed in part by the payment of page charges, This article must therefore he hereby marked “advertisement in accordance with 18 USC. Section 1734 solely to indicate thie fact. ‘To whom correspondence should be addressed, Tel; 600-995-1048, Fax; 600-995-0686. "the abbreviations used are: HMG, high mobility group nenhistone chromatin protein; MOPS, 4-morpholinepropaneeulfonc acid; DEPC, ‘ethyl pyrecarbonate, In vivo, HMG-VY has been immunolocalized to the A‘T-rich GiQ and C bands of mammalian metaphase chromosomes (11, 12), suggesting a structural role for the protein, and good evidence has been presented linking HMG-I/Y with activation of chromatin domains via displacement of histone Ht from scaffold attachment regions (13, 61). In addition to possible roles as a chromatin structural factor, recent reports indicate that HMG-Y also functions aa a general transcription factor (14-18). With respect to these observations, elevated in vivo levels of HMG-LY have been correlated with both neoplastic transformation (1-3, 20-28) and with metastatic tumor pro: gression (24-26). Furthermore, proteins otherwise unrelated to HMG-Y have been discovered that contain amino acid sequences with similarity to the DNA-binding domains. Exam- ples include HRX (ALL) in humans (27), D1 in Drosophila (62), ATBP-1 from pea (28), LAT1, NAT1, and 2 from soybean (29), PF1 from rice (30), and MIF2 (31) and datin (63) from Saccha: romyces cerevisiae, In this report, we extend our earlier studies of HMG-UY binding to naked A-T-rich DNA and to chicken mononucleo somes to include the effect of protein binding on the topological state of closed circular DNA. Topoisomerase T-mediated relax. ation assays of complexes of HMG-Y with supercoiled plasmid DNA reveal that HMG-L/Y induces positive supercoiling at low protein to nucleotide molar ratios in a manner similar to ne: tropsin, suggesting that bends are induced in the substrate at these protein concentrations. At higher ratios, however, nega- tive supercoils are induced, indicating that protein binding leads to underwinding of DNA at these concentrations, in con trast to the positive supercoils induced by binding of netropsin Additionally, we present footprinting evidence that HMG-1/Y is ‘capable of binding to and altering the secondary structure of ‘non-B-formn DNA. EXPERIMENTAL PROCEDURES Preparation of HMO-IY) Proteins and’ DNA—Recombinant hum HMG-T and HMG-Y were produced vsing the expression actor pET7C ‘and were purified by ion exchange chromatography as described previ ‘ualy (32) Protein concentrations were determined epectrophotometri- cally using ény ~ 7,000 and én ~ 65,000 itersmol-em for IMG Land HIMGLY, respectively (8), An additional HMG-I protein, SEOL, was also propared using the same methoda. This protein lacks the 17 glutamic Acidrich residues atthe carboxyl terminus of the intact HMG-T mole fle, and its construction haa been detailed previously (33). The con- struction of plasmid pBLT, which eontains the S00-base pair A:T-ich {untranslated region derived from the bovine interleakin-2 cDNA, bas been desribed (Plasmids pUC-18 and pBLT were purified by band- ‘ng on CsCl gradients, and concentrations were determined by abeorp- tion at 260 nm, Preparation of Topoisomerase I—Washed, packed, and frozen chicken erythrocytes were purchased as 10-ml aliquots from Lempire Biological, Pipersville, PA. Cells were simulteneously thavred and lysed in iceeald 10 mt MOPS, pH 7.2, 150 mu NaC, 0.5 mat phenylmethy sulfonyl fluoride and 0.2% Nonidet P-40. Nuclei were pelleted by cen teifugation at 2000 % g for 10 min, The supernatant was carefully removed, and the nuclear pellet was washed by resuspension in buffer 43554356 Fic, 1, Topoisomerase-mediated re- Inxation amays. Plasmids pBLT and UCIS were relaxed with topoizomerase I In the presence of dilferent DNA binding ligands In each of he four panel, the lft sat of lanes represents pBLT, and the ‘ight set represents pUCIB, In all eases, Tigand concentration increases from ff to right. The relative positions of form I and IEDNA are indicated. A, EMG at pro- tein to nucleotide molar ratios of 0, 0.27, 0.34, 081, 11, 16, and 22:5, HMG-Y the same protein to nucleotide ratios asin ASC, EtBr at 0, 005, 0.075, 0.1, 025,05, dnd 1 ygiml; B, neiropsin at 0, 0.0075, 0.01, 0025, 0.05, 0.1, and 0.2 molar ratio. of drug to nucleatide without Nonidet P-40 and subsequent centrifugation. Topoisomerase 1 was then isolated from washed nuclei, employing the procedure de- tailed by Pfale and Jackson (24). Fractions containing topoisomerase activity were pooled and split into aliquots. These aliquots were stared frozen at ~70°C. Enzyme activity was asayed using PUCIS, where one ‘nit wil relax 0.5 ug of supercoiled DNA in 0.5 h at 37°C Topoisomerase Fmediated Relaxation Assay—For a typical assay cither HMG-I or HDMG-Y protein, ethidium bromide (EtBr or netropsin ‘was added in varying ratios to2 of negatively supercoiled pBLT or UCI plasmid, and incubated at 29°C for 90 min in 200 ul of 20 mat Tris-Cl, pH 8, 60 mut NaCl, 50 pg/ml bovine serum albumin, 5% glye rol, and 1 mat dithiothreitol, Twenty unit of topolsomerase T were fadded, and the samples were incubated for 1h at 37°C. Samples were ‘then made 0.1% in SDS and digested with 01 ml Proteinase K for 30 min at 37°C followed by extensive extraction with phenolCHC1, and ‘ethanol procipitation of the DNA. The DNA was redissolved in TE: (10 tmat Tri, pH 16, 10 ma EDTA) and electrophoresed overnight on 1.5% ‘agarose-TAE gels at 0.5 Viem and 23°C. The gels were stained with EtBr and photographed, Twordimensional Chloroquine Gels—Aliquots of topoisomerase relaxed DNA prepared for one-dimensional analysis were loaded onto 115% agarose-TAE (40 mit Tris, pH83, 20 mM sodium aeetate, 2 mat EDTA) gels and electrophoresed for 3h at 6 Viem and 23°C. The gels wore then stained for 1 h in TAE containing 1.5 jgiml chloroquine phosphate, rotated 90° with respect tothe original direction of migra tion, and the electrophoresis was repeated in TAE supplemented with 1.5 nim chloroquine phosphate, The gels were stained with EXBr and photographed ‘Determination of Protein-DNA Diasociation Constants—Dissociation constants of HMG- and HMG-I(SE91) to superesiled plasmids were determined using a fluorescence competition assay employing the dye Hoechst 38258 as reported previously (8, 35). For these ascays, plasmid concentration was fixed at 0.33 nll, protein concentration was 10 ns, land dye concentrations ranged from 0 to 60 a Footprints of Supercoled DNA~Rootprinting of supercoiled DNA with diethyl pyrocarbonate (DEPC) was largely as described by MeLean (36), Various amounts of HMG-I.AE91) were mixed with 1 ug of super ‘oiled pBLT in 45 nl of TE. Following 8 S0-min incubation at 28 °C, 5 ul ‘of BEPC was added, and incubation continued for an additional 15 min. ‘The samples were then precipitated with sodium acetate and ethanol and redissolved in water. Aer addition of the appropriate buffer, the DNA was digested with BeoRI, and the ends ware laboled by standard ‘methods (37) using [a2*PATP and Klenow polymerase. The labeled 500-base pair fragment was then released from the vector by digestion ‘vith Sal isolated by agarose gel electrophoresis, and purified from the ‘agarose with Geneclean (Bio 101, La Jolla, CA). The purified DNA was finaly cleaved by treatment with 10% piperidine for 15 min at 90°C land the cleavage products were visualized by autoradiography follow- ing electrophoresis on a 6% sequencing gel. Autoradiographs. were seanned with an LKB Ultrosean XL. laser densitometer RESULTS, Topoisomerase-mediated Relaxation Assays—The effect of various DNA binding ligands on the superhelicity of two plas- ‘mid substrates was investigated by agarose gel electrophoresis, HMG-I/Y Binding Induces Positive and Negative Supercoils In these experiments, ligand-DNA complexes were treated ‘with topoisomerase I to relax the superhelical stress in the DNA followed by removal of the ligand and subsequent elee- trophoretic resolution of the resulting topoisomers. Topoi somerase I removes any superhelical stress that is not con- strained by ligand-DNA interactions. Consequently, the topoisomers resolved following removal of the ligand reflect changes in DNA secondary and tertiary structure due to inter- action with the ligand. Plasmid pBLT was chosen as a test DNA substrate because it contains a 300-base pair A-T-rich insert whose interaction with HMG-UY has been previously characterized (8, 38), The A-T-rich insert was derived from the S/-untranslated region of the bovine interleukin-2 cDNA and hhas an A:T base content of 71.4 mol % discounting the poly(A) tail. Plasmid pUC18 was used for sake of comparison since BLT was derived from it. Fig. 1 shows the one-dimensional electrophoretic results typ- ical for a topoisomerase assay, Lane I of each panel shows the plasmid relaxed with topoisomerase Iin the absence of protein. ‘The relative positions of relaxed and supercoiled plasmid are indicated in the figure. In panel A, the two plasmids were relaxed with topoisomerase in the presence of molar ratios of HMG-I to nucleotide that varied from 0 to 2.2. It should be noted that in these assays, the plasmids are initially super- ‘coiled when HMG-L/Y is bound and that topoisomerase is added subsequently. Consequently, non-B-form DNA conformations, which might be formed as a result of supercoiling, could also serve as recognition sights for protein binding. It can be clearly seen that the presence of increasing amounts of HMG-I protein results in a change in the distribution of the topoisomers formed following relaxation of either plasmid and subsequent removal ofthe protein. Panel B of the figure shows results of an ‘assay performed with HMG-Y, Comparison of panels A and B indicates that increasing concentrations of both HMG-I and HMG-Y produce identical distributions of topoisomers in either BLT or pUC1B, Interestingly, HMG-I and HMG-Y produced ‘qualitatively similar distributions of topoisomers in both the AvTrrich pBLT plasmid and its parent vector, pUCIS. We in- terpret this to mean that HMG-LY, in addition to its ability to bind A‘T-rich sequences and influence the secondary and ter- tiary structure of the plasmid, may also recognize and bind structures induced in substrate DNA as a result of supercoil- ing. It should also be noted that the distribution of isomers over the course of the titration is different for the two plasmids, suggesting that the A’T-rich insert of supercoiled pBLT may have some intrinsic structure or that HMG-I/Y interacts differ- cently with itHMG-I/¥ Binding Induces Positive and Negative Supercoils (2.2, Two-dimensional chloro- quine gels Aliquots of samples shown in Fig. were run on 1.8% agarose gels, the gels were treated with 15 pgm! chioro- ‘sine nd rum in a second dimension. The migration direction of each dimension is Indicated. In each of the four panels, the Upper set of bands represents pBLT and the lower set of bande represents pUCIB, Ligand concentration increases from left ta right. The relative postions of form T find 11 DNA are indicated, & HMGT at protein to nucletide molar ratios of 037, 11, 16, and 22; B, HMG-Y at pro: tein to nucleotide molar ratios of 0, 027, 081, 1.1, and 1.6; C, netropein at 0, 0.028, (0.05, 0.1; and 0:2 drug to nucleotide molar atic, D, BtBr at 0, 0.075, 0.1, 025, and 05 ni ‘The same experiment was carried out using the A-T-binding antibiotic netropsin as the DNA binding ligand. Netropsin was chosen for two reasons. First, it has been suggested that the DNA binding domain of HMG-I may mimic the structure of netropsin (8, 39) and that the two molecules may interact with DNA in a similar fashion. Second, it has also been demon- strated that netropsin introduces positive supercoils into DNA when subjected to the sort of assay employed here (40, 41). As shown in Fig. 1C, relaxation of the plasmid in the presence of netropsin followed by removal of the antibiotic results in changes in the superhelical density of the DNA that are qual- itatively similar to those produced by HMG-L/Y. In order to help determine the sign of the supercoils introduced by HMG-UY, topological “standards” were prepared by topoisomerase. ‘mediated relaxation of plasmid DNA in the presence of the interealating dye EtBr. The binding of EtBr to DNA has been extensively characterized, and itis well known that intereal tion of EtBr causes unwinding of the helix (42, 43). It is also known that relaxation of circular DNA in the presence of EtBr followed by removal of the dye results in negatively supercoiled DNA; an effect opposite of that produced by netropsin (40). Fig. LD demonstrates the results of relaxation of pBLT and pUCI8 in the presence of increasing concentrations of EtBr. Determination of the Sign of Supercoiling—Aliquots of the samples shown in Fig. 1 were electrophoresed on two- dimensional agarose gels in the presence of 1.5 ug/ml of chlo- roquine in the second dimension. Chloroquine interealates into DNA and, under the conditions employed here, will retard the ‘migration of negatively supercoiled DNA while it increases the ‘migration of relaxed and positively supercoiled DNA (44). This effect is seen in Fig. 2. In the figure, negatively supercoiled topoisomers tend to the left of the figure (the origin of the second dimension), while positively supercoiled topoisomers tend to the right (the direction of migration in the second dimension). Fig. 2, A and B, shows two-dimensional gels of PBLT and pUC1S relaxed in the presence of HMG-I and HMG-Y, respectively. In both cases, it is seen that with increas- ing protein concentration, the plasmids initially become posi- tively supercoiled, that is, the migration rate ofthe topoisomers increases. However, as the protein concentration inereases, the 4357 plasmids become negatively supercoiled as demonstrated by the reduced mobility of the ensemble of topoisomers, In con- trast, pBLT relaxed in the presence of netropsin shows the opposite behavior (Fig. 2C). In this ease, migration of the to- poisomers is inereased, indicating that the DNA is positively supercoiled; an observation in agreement with that reported previously (40, 41). Is interesting to note that pUC18 exhibits both positive and negative supercoils at a netropsin concentra: tion that produces all positive supereoils in pBLT, again sug: gesting that the ligand interacts differently with the A-T-rich BLT DNA. Fig. 2D shows the effect of EtBr on the distribution of topoisomers. It is evident that EtBr induces formation of negative supercoils in this assay. Overall, the distribution of topoisomers produced at low protein to nucleotide molar ratios resembles that produced by netropsin, while at higher protein to nucleotide molar ratios, the distribution resembles that pro- ‘duced by ethidium. Effect of Removal of HMG-I Carboxyl Terminus—Topoi- somerase-mediated relaxation assays were performed using AF91, an HMG-I protein that lacks the 17-glutamic acid-rich C-terminal residues found in the full-length protein (38). Re- sults of a typical one-dimensional assay are shown in Fig. 3A. ‘The overall pattern of topoisomers produced by relaxation of DNA in the presence of the truncated protein is similar to that, produced by the full-length protein (Fig. 1A). It is important to note, however, that the AEQ1 to nucleotide molar ratios in Fig. 3 are approximately 10-fold less than the HMG-I to nucleotide ratios in Fig. 1. The sign of supercoiling induced by SE91 was determined by analysis on two-dimensional chloroquine gels, (Fig. 3B), and the sign of the induced supercoils was deter- ‘mined to be positive at relatively low protein to nucleotide molar ratios, with negative supercoils being induced at higher protein to nucleotide molar ratios. We investigated the relative affinity of both HMG-I and ‘AE91 for supercoiled DNA using an assay based on the compe- tition of binding of a luorescent dye, Hoechst 89258, to A:T. rich HMG-I binding sites (8), For this determination, super coiled plasmid DNA at a fixed concentration was titrated with Hoechst 39258 in the absence or in the presence, of 10 nw HMG-l or AE91 in buffer containing 50 mai NaCl. In Table I, it