OncJune3 6..6 - Nature

Oncogene (1999) 18, 3617 ± 3625 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc

Eect of transition metals on binding of p53 protein to supercoiled DNA and to consensus sequence in DNA fragments Emil PalecÏek*,1, Marie BraÂzdovaÂ1, Hana CÏernockaÂ1, Daniel Vlk1, VaÂclav BraÂzda1 and BorÏ ivoj VojteÏsÏ ek2 Institute of Biophysics, Academy of Sciences of the Czech Republic, KraÂlovopolskaÂ 135, 612 65 Brno, Czech Republic; 2Masaryk Memorial Cancer Institute, 656 53 Brno, Czech Republic

1

Recently we have shown that wild-type human p53 protein binds preferentially to supercoiled (sc) DNA in vitro in both the presence and absence of the p53 consensus sequence (p53CON). This binding produces a ladder of retarded bands on an agarose gel. Using immunoblotting with the antibody DO-1, we show that the bands obtained correspond to ethidium-stained DNA, suggesting that each band of the ladder contains a DNAp53 complex. The intensity and the number of these bands are decreased by physiological concentrations of zinc ions. At higher zinc concentrations, binding of p53 to scDNA is completely inhibited. The binding of additional zinc ions to p53 appears much weaker than the binding of the intrinsic zinc ion in the DNA binding site of the core domain. In contrast to previously published data suggesting that 100 mM zinc ions do not in¯uence p53 binding to p53CON in a DNA oligonucleotide, we show that 5 ± 20 mM zinc eciently inhibits binding of p53 to p53CON in DNA fragments. We also show that relatively low concentrations of dithiothreitol but not of 2-mercaptoethanol decrease the concentration of free zinc ions, thereby preventing their inhibitory eect on binding of p53 to DNA. Nickel and cobalt ions inhibit binding of p53 to scDNA and to its consensus sequence in linear DNA fragments less eciently than zinc; cobalt ions are least ecient, requiring 4100 mM Co2+ for full inhibition of p53 binding. Modulation of binding of p53 to DNA by physiological concentrations of zinc might represent a novel pathway that regulates p53 activity in vivo. Keywords: p53; DNA-p53 interaction; divalent cation; gel electrophoresis

sc

DNA;

Introduction Mutations in the p53 gene are one of the most frequent genetic alterations observed in human cancer (in about 50% of human malignancies). Speci®c binding of the wild-type tumor suppressor protein p53 to the consensus sequence (p53CON) is crucial for the ability of this protein to operate as a transcription factor (reviewed in Cox and Lane, 1995; Elledge and Lee, 1995; Soussi and May, 1996; Levine, 1997; Ko and Prives, 1996). p53CON consists of two copies of

*Correspondence: E PalecÏek Received 7 August 1998; revised 18 January 1999; accepted 19 January 1999

the sequence 5'-PuPuPuC(A/T)(T/A)GPyPyPy-3', separated by 0 ± 13 bp. Three-dimensional models of the tetramerization domain (Arrowsmith and Morin, 1996) and of the core DNA-binding domain (Arrowsmith and Morin, 1996; Cho et al., 1994) are now available. The latter model of the DNA-binding domain is based on the co-crystal structure of the DNA-p53 complex composed of residues 94 ± 312 of the human p53 protein and a 21-base pair DNA duplex containing a p53 consensus half-site. This structure contains three protein molecules but only one of them is bound to DNA in sequence-speci®c manner. Recently, a structural model for the complex of four p53 protein molecules with a functional response element has been reported (Nagaich et al., 1997a,b). This model (based on high resolution chemical footprinting, cross-linking and molecular modeling) suggests that when four subunits of the p53 DNA binding domain bind to the DNA response element, the DNA has to bend to relieve steric clashes among the subunits. p53 is a metalloprotein containing one zinc atom in the core domain (Arrowsmith and Morin, 1996; Cho et al., 1994). The ligands for the tetrahedrally coordinated Zn atom are Cys 176, Cys 238, Cys 242 and His 179. The conformation of the protein is critical for its biological activity and for its sequence speci®c binding to DNA. It has been suggested that the dual ability of p53 to function as a tumor suppressor or to promote cell proliferation is due to switching between two conformational states, wildtype and mutant conformations, respectively (Hainaut and Milner, 1993; Hainaut et al., 1995; Pavletich et al., 1993). It has been shown that the wild-type conformation can be aected by binding of metal ions (Coer and Knowles, 1994). A panel of monoclonal antibodies, whose epitopes map throughout the protein, was used to test puri®ed human wild-type p53 to assess whether divalent metal ions Co2+, Ni2+ and Zn2+ aect the conformation of the protein. It was found that micromolar concentrations of Zn2+ reduce or block accessibility of epitopes while Co2+ and Ni2+ show no eect under the same conditions. Maximum loss of epitope reactivity induced by Zn2+ was between the protein mid-region and C-terminus. These results obtained with puri®ed p53 in vitro diered from the observations reported for 35S-labeled murine p53 in a rabbit reticulocyte lysate where Zn2+ showed no eect at concentrations up to 0.5 mM (Hainaut and Milner, 1993). The ability of the puri®ed human p53 to bind speci®cally to p53CON was not aected by Zn2+ at physiological levels (Coer and Knowles, 1994).

Metal effect on p53 binding E PalecÏek et al

3618

We have recently shown that wild-type human p53 protein (expressed in insect cells) binds strongly to negatively supercoiled (sc) DNA at native superhelix density (Palecek et al., 1997b). The presence of p53CON in scDNA is not necessary for this binding. Binding of p53 to scDNA results in full or partial relaxation of DNA visible on scanning force microscopy images. Agarose gel electrophoresis produces a series of discrete bands which resemble the topoisomer ladder yielded by topoisomerase I (Palecek, 1991; Yagil, 1991). In competition assays, p53 shows a preference for scDNA as compared to linear doublestranded DNA containing p53CON. On the other hand, thermally denatured DNA (regardless of the presence or absence of p53CON) is ecient in competing for the binding of p53 to scDNA. It appears probable that interactions involving both the core domain and the C-terminal domain may regulate the binding of p53 to scDNA. Strong binding of p53 to scDNA is a new phenomenon which may have interesting cell biological implications. In this paper, we study the in¯uence of some divalent metal ions on binding of p53 to DNA. We show that micromolar concentrations of Zn2+ efficiently inhibit binding of the p53 protein to supercoiled DNA and, in contradiction to the data in the literature (Coer and Knowles, 1994), also to p53CON in DNA fragments.

detected on the blot (Figure 1B, lane 8), suggesting a very strong binding of p53 to DNA. In the presence of zinc, a spot of the free protein appeared whose intensity increased with the concentration of zinc (Figure 1B, lanes 9 ± 12); at 20 mM Zn2+, the intensity of this spot corresponded closely to that of p53 in the absence of DNA. At lower p53/scDNA ratios a smaller number of bands were observed in the absence of zinc (Figure 1A, lane 3) and complete inhibition of p53 binding to scDNA was reached (at p53/scDNA=5) already at 10 mM Zn2+ (Figure 1A and B, lane 7). Similar results were obtained with the scDNA of pPGM1 containing a single p53CON in its molecule (not shown). If zinc was added after incubation of p53 with scDNA (instead of incubation in the presence of zinc as above), the eect of the metal ion was decreased but not eliminated (not shown) suggesting that some potential zinc-binding sites in the p53-DNA complex might not be accessible to zinc ions and/or

Results Micromolar concentrations of Zn2+ ions inhibit binding of p53 protein to scDNA Using agarose gel electrophoresis, we tested the eect of Zn2+ ions in the concentration range of 0.5 ± 20.0 mM on the binding of p53 to scDNA (pBluescript II SK7, not containing p53CON) at a p53/DNA ratio of 5 (Figure 1A, lanes 4 ± 7) or 10 (Figure 1A, lanes 9 ± 12). At p53/DNA=10 in the absence of Zn2+ ions, ethidium bromide staining of the gel revealed four wellresolved DNA bands in addition to a weak band 0 corresponding to the scDNA band observed in the absence of p53 (Figure 1A, lane 8). Incubation of p53 with scDNA in the presence of 500 nM Zn2+ (Figure 1A, lane 9) resulted in some fainting of bands 3 and 4 and increasing of the intensity of band 0. With increasing Zn2+ concentration, further fainting and disappearing of bands 1 ± 4 and increasing intensity of band 0 was observed (Figure 1A, lanes 10 ± 12), with a half-maximum eect at about 5 mM Zn2+ (Figure 1a, lane 10) and complete inhibition of p53 binding at 20 mM Zn2+ (Figure 1A, lane 12). Immunoblotting with the antibody DO-1, which binds to the N-terminal domain of p53 at amino acids 20 ± 25 (Stephen et al., 1995; Vojtesek et al., 1992), produced similar results (Figure 1B). A superimposition of the immunoblot on an ethidium-stained gel obtained in the absence of zinc (Figure 1C) showed four strong bands corresponding to ethidium bands 1 ± 4 and an additional weak band 5. No band on the blot corresponding to band 0 on the gel was observed (Figure 1B and C), in agreement with our assumption that band 0 re¯ects scDNA free of p53 protein (Palecek et al., 1997b). Almost no free p53 was

Figure 1 In¯uence of Zn2+ions on binding of human wild-type p53 to supercoiled (sc) DNA. (A) 0.2 mg of scDNA pBluescript was incubated with 0.12 mg of p53 (the molar ratio of p53 tetramer/DNA= 5/1) in lanes 3 ± 7 or with 0.24 mg of p53 (the molar ratio of p53 tetramer/DNA= 10/1) in lanes 8 ± 12 with or without zinc at various concentrations. c(DTT) was 17 mM when reacting 0.12 mg of p53 and 34 mM when reacting 0.24 mg of p53. The incubation was carried out in 50 mM KC1, 5 mM Tris-HC1 (pH 7.6), 0.01% Triton X-100 on ice for 30 min, and was stopped by adding the loading buer. Samples were loaded on a 1% agarose gel in 0.336TBE. Electrophoresis was run at 100 V, and 48C for 3 ± 4 h. DNA was stained with ethidium bromide. (B) Immunoblotting of the gel (shown in A) with the monoclonal antibody DO-1. (C) Superimposition of the ethidium-stained gel in A and the immunoblot in B showing lanes 1, 2 and 8 ± 12. Band 0 corresponds to scDNA without p53, bands 1 ± 5 correspond to scDNA-p53 complexes


that possible changes in p53 conformation (probably induced by zinc ions) might not be the same in free p53 as in p53 bound to scDNA. The eect of Zn2+ on p53 binding to scDNA is reverted by low concentrations of EDTA Inhibition of p53 binding to scDNA produced by 5 mM Zn2+ (Figure 2, lane 3) was almost completely prevented by 15 or 50 mM EDTA (Figure 2, lanes 5 and 6); in the presence of 5 mM EDTA, a small inhibitory eect of Zn2+ was still observed (Figure 2, lane 4). EDTA eectively reverted the eect of Zn2+ ions, and interestingly it made no dierence whether EDTA and Zn2+ were added simultaneously to the mixture (Figure 2, lanes 4 ± 6) or whether the addition of EDTA occurred after the incubation of the p53DNA complex with Zn2+ ions (Figure 2, lanes 7 ± 9).

We replaced DTT with 2-mercaptoethanol (MEt) using dialysis against the p53 storage medium containing MEt instead of DTT; after 60 min of dialysis the concentration of DTT decreased below 1 mM as detected by constant current chronopotentiometric stripping analysis (see Materials and methods). The dependence on Zn2+ concentration obtained in the presence of 175 mM MEt Figure 3B, lanes 2 ± 5) was similar to that obtained in the presence of low concentrations of DTT (Figure 3A), while 175 mM DTT fully or partially suppressed the eect of zinc ions (Figure 3B, lanes 7 ± 10). We isolated p53 in the absence of DTT and tested the eect of various concentrations of either DTT or MEt on p53 binding in the presence of 10 mM Zn2+. This protein was less active as compared to that isolated in the presence of DTT due to partial oxidation of the former protein (M Fojta et al.,

DTT but not 2-mercaptoethanol decreases the inhibitory eect of Zn2+ on binding to scDNA p53 protein is usually isolated and stored in the presence of dithiothreitol (DTT) to keep the protein in its reduced state (Bullock et al., 1997; Parks et al., 1997). It has been shown that zinc-DTT complexes can be formed (Cornell and Crivaro, 1972; Hunt et al., 1985) but DTT at concentrations up to 40 mM induced no loss of the intrinsic zinc in p53 even during denaturation of the protein (Bullock et al., 1997). On the other hand, it can be anticipated that DTT will decrease the concentration of free Zn2+ ions. We isolated and stored p53 in the presence of 5 mM DTT at high protein concentration; the concentration of DTT during the incubation with DNA was however relatively low (5100 mM). Figure 3A shows the eect of DTT (50 mM ± 1 mM) on p53 binding to scDNA pBluescript II SK7 in the presence of 10 mM Zn2+. This concentration of Zn2+ resulted in complete inhibition of p53 binding to scDNA. Fifty mM DTT did not induce any change in this inhibition (Figure 3A, lane 4) while 100 mM DTT produced a weak band of the p53-DNA complex (Figure 3A, lane 5) suggesting a small decrease in the concentration of free Zn2+ ions; more bands were observed at higher DTT concentrations (Figure 3A, lanes 6 and 7) and with 5 mM DTT the inhibitory eect of 10 mM Zn2+ was eliminated (not shown).

Figure 2 Reversal of Zn2+-induced inhibition of p53 binding to scDNA by EDTA. 0.25 mg of scDNA pBluescript and 0.23 mg of p53 protein and 5 mM Zn2+ (lanes 4 ± 6) were incubated at the three concentrations of EDTA given in the ®gure, in lanes 7 ± 9 EDTA was added after 30 min of incubation. For details, see Figure 1

Figure 3 In¯uence of DTT and 2-mercaptoethanol on the binding of p53 to scDNA in the presence of Zn2+. (A) Reversal of Zn2+-induced inhibition of p53 binding to scDNA by DTT. 0.3 mg of scDNA pBluescript was incubated with 0.33 mg of p53 (the molar ratio of p53 tetramer/DNA= 10/1), in the presence of 10 mM Zn2+ and at the DTT concentrations shown in the ®gure. (B) Comparison of the eects of DTT and 2-mercaptoethanol. Incubation was performed either as in A (lane 6) or at p53/ DNA=5 (lanes 2 ± 5 and 7 ± 10) in 175 mM DTT (lanes 7 ± 10) or in 175 mM MEt (lanes 2 ± 5). Prior to incubation, the p53 protein was dialyzed to replace the DTT with MEt (lanes 2 ± 5). (C) Comparison of the eects of DTT and MEt on p53 protein isolated in the absence of DTT. 0.66 mg of p53 protein (isolated in the absence of DTT) and 0.3 mg of scDNA pBluescript (the molar ratio p53 tetramer/DNA= 20/1) were incubated without (lane 2) and with 10 mM Zn2+ (lane 3) and in the presence of DTT (lanes 4 ± 8) and MEt (lanes 9 ± 12) at concentrations shown in the ®gure. For details, see Figure 1

3619


3620

unpublished). We therefore used p53/scDNA=20 to obtain detectable p53 binding to scDNA on the gel (Figure 3C, lane 2). Under these conditions, 10 mM Zn2+ completely inhibited p53 binding to scDNA in the absence of DTT or MEt (Figure 3, lane 3) as well as in the presence of MEt (Figure 3C, lanes 9 ± 12) up to 5 mM MEt (not shown). In the presence of 100 mM DTT, the inhibitory eect of Zn2+ was completely reverted (Figure 3, lane 5) while higher DTT concentrations enhanced the binding activity of the partially oxidized p53 (Figure 3C, lanes 6 ± 8). We may thus conclude that higher concentrations of DTT should be avoided in experiments involving micromolar concentrations of zinc ions. Replacing DTT by MEt may be a good choice in these types of experiments. Cobalt and nickel inhibit p53 binding to scDNA less eciently than zinc We tested the in¯uence of other transition metal ions such as Co2+ and Ni2+ at concentrations from 5 ± 600 mM on p53 binding to scDNA (Figure 4). Only a partial inhibition of p53 binding was observed at 300 mM Co2+ or Ni2+ (Figure 4, lanes 5 and 9). Six hundred mM nickel ions fully inhibited p53 binding to scDNA (Figure 4, lane 10) while cobalt ions at the same concentration were less ecient (Figure 4, lane 6). This large dierence in the eciency of inhibition of p53 binding to scDNA between Zn2+ ions on the one hand and Co2+ or Ni2+ ions on the other roughly agrees with the eect of these divalent ions on the accessibility of epitopes of p53 observed by Coer and Knowles (1994). In spite of large changes in epitope accessibility induced by Zn2+ ions, these authors observed no changes in p53 binding to p53CON in a deoxyoligonucleotide. To verify this ®nding, we tested speci®c binding of p53 to p53CON in a linear DNA fragment. Sequence-speci®c binding of p53 protein to p53CON in a linear DNA fragment is inhibited by Zn2+4Ni2+4Co2+ Fragments from a PvuII digest of pPGM1 DNA (yielding a 2513 bp fragment and a 474 bp fragment containing the p53CON) were incubated with p53 in presence of Zn2+ ions in a concentration range 0 ± 20 mM (Figure 5). At p53/DNA=5 or 10 in the absence of Zn2+, the band of the 474 bp fragment weakened or

Figure 4 In¯uence of Co2+ and Ni2+ ions on the binding of p53 to scDNA. 0.2 mg of scDNA pBluescript and 0.24 mg of p53 (the molar ratio p53 tetramer/DNA= 10/1) were incubated in the presence of Co2+ and Ni2+ at concentrations shown in the ®gure. For details, see Figure 1

disappeared and a distinct retarded band designated as R and a much weaker band R1 were observed (Figure 5, lanes 2 and 7). Five mM Zn2+ completely blocked p53 binding to p53CON at p53/DNA=5 (at Zn/p53 monomer*50) (Figure 5, lane 5); at p53/DNA=10, a twofold higher concentration of zinc was necessary to obtain the same eect (Figure 5, lane 10). The eect of zinc ions was reverted by low concentrations of EDTA (not shown) as in the case of p53 binding to scDNA (Figure 2). Figure 6A shows that Co2+ and Ni2+ ions inhibit p53 binding to p53CON less eciently than Zn2+. To obtain full inhibition of p53 binding, 4100 mM Co2+ or 450 mM Ni2+ were necessary (Figure 6A, lanes 6, 7 and 10, 11). Superimposition of DO-1 immunoblot (Figure 6B) with the ethidium-stained gel is shown in Figure 6C. In the absence of zinc, a strong band and a much weaker band appeared on the blot corresponding to ethidium-stained bands R and R1, respectively. A strong spot of unbound p53 overlapped the band of the longer 2513 bp DNA fragment, suggesting that a large amount of p53 remained unbound; under the given conditions, a more accurate determination of free p53 was not possible. In the presence of 20 mM Zn2+, 100 mM Ni2+, or 300 mM Co2+, bands R and R1 completely disappeared both on the blot and on the ethidium-stained gel (Figure 6A ± C, lanes 3, 7 and 11). At lower metal concentrations, the intensity of bands R and R1 on the blot (Figure 6B, lanes 4 ± 6 and 8 ± 10) decreased with increasing metal concentration in agreement with the results of the ethidium-stained gel electrophoresis (Figure 6A). Eect of transition metal ions on non-speci®c binding of p53 to DNA In the absence of transition metals, p53 at p53/ DNA=10 induced a weak band N moving slower than the longer 2513 bp pPGM1 DNA fragment (not containing p53CON, Table 1) (Figure 6A, lane 2). A similar band was also obtained with the same fragment from DNA pBluescript (Figure 6D, lane 2). These bands might be due to non-speci®c binding of p53 to

Figure 5 Eect of Zn2+ on the binding of p53 to p53 CON in a DNA fragment. pPGM1 was cleaved by PvuII into two fragments of 474 bp (with the p53 consensus sequence) and 2513 bp (lacking the consensus sequence). The fragments (0.2 mg) were incubated with two dierent amounts of p53 (0.12 mg for a molar ratio of p53/DNA 5/1 and 0.24 mg for a molar ratio 10/1) at the Zn2+ concentrations given in the ®gure. R, retarded band due to the sequence-speci®c binding of p53 to the consensus sequence


the longer 2513 bp fragment which may oer a larger number of the potential non-speci®c binding sites than the shorter DNA fragment. Our search for half-sites of the p53CON with single-base mismatches showed two of these sites in the longer 2513 bp fragment (Table 1) and one in the shorter 448 bp pBluescript fragment

suggesting that the band N is probably not due to binding of p53 to these sites. At 100 mM Ni2+, the band N disappeared both in pPGM1 (Figure 6A, lane 7) and in pBluescript DNAs (Figure 6D, lane 6); fainting of this band was observed also with 300 mM Co2+ (Figure 6A, lane 11 and D, lane 8) as well as with Zn2+ ions

Figure 6 Eect of Co2+, Ni2+, and Zn2+ ions on (A ± C) sequence-speci®c binding of p53 protein to p53CON in a DNA fragment and on (D) non-speci®c binding to DNA fragments not containing p53CON. (A) Ethidium-stained gel of DNA fragments obtained by PvuII cleavage of pPGM1 DNA (molar ratio of p53 tetramer/DNA= 10/1). (B) Immunoblotting of the gel (shown in A) with the monoclonal antibody DO-1. (C) Superimposition of the ethidium-stained gel in A and the immunoblot in B. (D) Ethidiumstained gel showing the in¯uence of metals on non-speci®c binding of p53 to the longer DNA fragment (2513 bp) of pBluescript. R, R1 retarded bands due to the sequence-speci®c binding of p53 to the consensus sequence; N, retarded band due to non-speci®c p53 binding to the 2513 bp DNA fragment. For details, see Figures 1 and 5

Table 1

Half-sites with single-base mismatches (m) in pBluescript II SK7 and insertion of p53CON [AGACATGCCTAGACATGCCT] are shown

3621


3622

(not shown) at concentrations substantially higher than those necessary to remove band R (Figure 6A, lanes 2 and 3). These results suggest that the studied transition metals eciently inhibit not only the p53 binding to p53CON in linear DNA fragments but also the nonspeci®c binding. Our results showing inhibition of binding of p53 protein to p53CON in a linear DNA fragment (Figures 5 and 6) are at variance with the ®nding of Coer and Knowles (1994) that 100 mM Zn2+ has no eect on p53 binding to p53CON in a 30-mer oligonucleotide. The reason for this discrepancy is unclear. We cannot exclude the possibility that the presence of EDTA and possibly DTT might have in¯uenced the results reported by Coer and Knowles. They used the oligonucleotide in a solution containing 1 mM EDTA and isolated p53 protein in the presence of 5 mM EDTA (calculation of the EDTA and DTT concentrations during the DNA incubation with p53 in their paper is dicult). In the initial stage of this work (Vlk, 1997), we also used DNA dissolved in TE buer (see Materials and methods) and we observed inhibitory eects at concentrations higher than one order of magnitude as compared to our present experiments with DNA solutions not containing EDTA (Figures 1 ± 5). Discussion In this paper we show for the ®rst time that binding of the wild-type p53 to the consensus sequence and to supercoiled DNAs is inhibited in vitro by micromolar concentrations of zinc ions. Previously we showed that binding of p53 to scDNA results in formation of a ladder of bands on an agarose gel but we had no direct evidence that these bands corresponded to p53-DNA complexes (Palecek et al., 1997b). We show here that the ethidium-stained ladder of bands (Figure 1A) coincides with sharp bands on DO-1 immunoblot (Figure 1B and C). These results represent direct evidence for the presence of p53 protein in these discrete DNA bands, suggesting that the formation of the ladder is due to binding of p53 protein to DNA. We also show that the number and intensities of the bands are decreased by physiological concentrations of zinc both at the ethidium-stained gel and on the blot (Figure 1A and B, lanes 3 ± 12). Immunoblotting on p53 on agarose gels appears to be a useful technique not only in the studies of p53 binding to scDNA (Figure 1) but also in binding of p53 to p53CON (Figure 6). Protein-zinc binding Interactions of a transition metal with a binding site in a protein is connected with (a) chemical nature, charge and ionic size, i.e. hardness or softness of the metal and (b) simple electrostatic eects (Christianson, 1991). Under physiological conditions, zinc exists as divalent Zn2+, which has a completely ®lled d shell with 10 d electrons. In terms of hard-soft acid-base theory, zinc is regarded as a borderline acid which easily accommodates in its biological coordination polyhedra a variety of ligand types including nitrogen from histidine, sulfur from cysteine, and oxygen from

aspartate, glutamate and water. It binds to negatively charged residues (e.g. thiolates and carboxylates) and/ or to neutral dipolar residues (e.g. imidazoles and carbonyls). In proteins, the coordination number 4 is most commonly found, with zinc ion typically coordinated in tetrahedral or distorted tetrahedral fashion. Catalytic zinc is dominated by histidine ligands while the coordination polyhedron of structural zinc is dominated by cysteine thiolates; the metal ion is exposed to bulk solvent frequently binding a solvent molecule (Christianson, 1991; Vallee and Auld, 1990). Zn2+ is not redox active and undergoes ligand exchange relatively rapidly. The lack of redox activity appears crucial for the zinc role as a structural element in DNA binding. DNA-zinc binding In DNA, divalent zinc ions interact electrostatically with phosphates and bind to bases, particularly N-7 of adenine and guanine (reviewed in Klevickis and Grisham, 1996; Sabat and Lippert, 1996). Binding to the bases can destabilize DNA and cause local denaturation of the DNA double helix. It has been shown that some local DNA structures such as intramolecular triplexes (Beltran et al., 1993; Bernues et al., 1989), hairpins (Martinez-Balbas and Azorin, 1993) and homopurine duplexes (Ortiz-Lombardia et al., 1995) are stabilized by zinc. In the presence of zinc, formation of kinks in DNA molecules have been observed (Han et al., 1997a,b; Laundon and Grith, 1987). Using the newly developed magnetic a.c. mode of atomic force microscopy, Han et al. showed that small (168 bp) DNA circles distort their shape in the presence of 1 mM or 100 mM zinc ions from smoothly bent to abruptly kinked (Han et al., 1997a,b). All of the above eects of zinc were observed at Zn/ DNAP41 while we show the inhibitory eect of zinc on p53 binding to DNA at Zn/P551 (Figures 1 and 5). It therefore appears more probable that the eect of zinc on p53 binding to DNA is due to changes in the p53 protein rather than in DNA conformation. On the other hand, we can speculate that formation of kinks and/or other localized structural changes at speci®c DNA sequences might occur at Zn/P551 and interfere with binding of p53 to p53CON and to scDNA. In¯uence of Zn2+ ions on the binding of p53 to DNA The zinc ion in the core domain is required for the p53 sequence speci®c DNA binding. The crystal structure of the core domain of p53 showed two b-sheets of the b-sandwich packed together to form a hydrophobic core and two DNA binding loops tethered to one another and to a small helix via cysteine and histidine residues which coordinate a zinc ion (Arrowsmith and Morin, 1996; Cho et al., 1994). This p53 structure did not correspond to any known DNA-binding motif, thus providing the ®rst example of a transcription factor that uses loops arrangement to recognize DNA. In this paper we show that binding of p53 to p53CON (Figure 5) and to scDNA (Figure 1) is inhibited by low concentrations of Zn2+. Binding of p53 to scDNA is partially inhibited by as little as 2.5 Zn2+ ions per monomeric p53 (Figure 1A and B, lane 9)


while full inhibition requires an excess of about 100 Zn2+ per p53 monomer (Figure 1A and B, lane 12). This inhibition correlates roughly with changes in the accessibility of some epitopes of p53, as observed by Coer and Knowles (1994). These authors used a number of antibodies, including PAb1801, PAb1620, PAb240, PAb121 and PAb421 (mapping between Nand C-terminus), to study the in¯uence of Zn2+, Co2+, and Ni2+ on the accessibility of epitopes of puri®ed human wild-type p53 protein (expressed in insect cells). At concentrations as low as 10 ± 20 mM, zinc ion strongly reduced the accessibility of previously exposed epitopes mapping to the core domain and to the C-terminal domain, including the conformationdependent and denaturation sensitive PAb1620. The eect of zinc was reverted by submillimolar concentrations of EDTA. PAb1801 mapping to the N-terminal domain was much less in¯uenced by the tested metal ions. The latter ®nding corresponds to our results showing no changes in accessibility of the DO-1 epitope at the zinc concentrations used in this paper (B Vojtesek, M Brazdova, H Cernocka and E Palecek, unpublished). Compared to the zinc strongly bound in the DNA-binding domain to Cys 176, Cys 238, Cys 242 and His 179, binding of additional Zn2+ ions to p53 appears rather weak since the latter zinc ions can be removed by EDTA (Figure 2) and DTT (Figure 3) at relatively low concentrations; such low concentrations of the chelating agents are not able to remove the strongly bound zinc from the p53 DNA-binding domain (Bullock et al., 1997; Hainaut and Milner, 1993; Pavletich et al., 1993). A number of putative zinc binding sites can be considered among the amino acid residues of the core domain including Cys 124, Cys 135, Cys 141, Cys 182, Cys 229, Cys 275, Cys 277, His 115, His 178, His 193, His 214, His 233, His 296 and His 297. In addition, His 365, His 368 and His 380 are located in the C-terminal domain. It cannot be excluded that zinc might weakly bind to other sites, including the phosphates which are involved in the transition of latent p53 into the active DNA-binding form (Hupp and Lane, 1994). Eects of metals in vivo Among the three transition metals studied in this paper (Figures 1 and 4 ± 6), zinc appears to be the best candidate to play a signi®cant role in regulation of binding of p53 to DNA in vivo. Physiological concentrations of nickel, cobalt and zinc ions in blood plasma were reported as 0, 2 and 20 mM, respectively (Glusker, 1991). Serum concentrations of zinc are altered in patients with lung, hepatic and prostate carcinomas, as well as in Hodgkin's disease and malignant melanoma (Ebadi and Iversen, 1994). Recently it has been shown that exposure of cultured cells to the membrane-permeable chelator N,N,N',N'tetrakis(2-pyridylmethyl)ethylenediamine induces an accumulation of wild-type p53 in an immunologically `mutant' form with decreased ability for binding to p53CON in a double-stranded oligonucleotide (Verhaegh et al., 1998). The possibility exists that changes in zinc level have an information-carrying role in vivo (Berg and Shi, 1996; Nagel and Valle,

1995; Verhaegh et al., 1998) and that they might also be involved in the regulation of p53 activity. Such regulation processes may depend on the interactions of Zn2+ with much simpler binding sites than found in the p53 DNA binding domain (Berg and Shi, 1996), including properly displayed single cysteine and histidine residues. The range of zinc concentrations optimal for the DNA binding activity of p53 might be rather narrow (Figures 1 and 5) and both the removal of the intrinsic zinc from the p53 molecule and an excess of zinc ions in the environment may down-regulate the DNA binding activity of p53 (Verhaegh et al., 1998). The eects of metals deserve further studies both in vitro and in vivo; the putative eects of metals in vivo may have important implications for cancer therapy.

Materials and methods In the initial stage of this work, we used scDNA dissolved in the usual TE buer. When we learned that the concentrations of Zn2+ aecting the binding of p53 to scDNA are of the same order of magnitude as the EDTA concentration in the sample, we completely eliminated EDTA from the samples by using scDNA dissolved in Tris buer. Prolonged storage of scDNA in this buer resulted in increased DNA nicking (Figure 1A); the amount of nicked molecules was however not suciently large to interfere with the experiments. It was shown (Jacob et al., 1998) that high concentrations of Tris (above 100 mM) may compete for zinc ions with proteins. With 5 mM Tris used in our experiments we did not observe such an eect. A comparison of the in¯uence of zinc on binding of p53 to scDNA in 5 mM Tris, HEPES or MOPS buers did not show any decrease of the zinc eect by the Tris buer. Stock solutions of (analytical grade) ZnC12, CoC12 and NiC12 at a concentration of 0.5 M in 0.05 M HC1 were used. DNA Synthetic oligonucleotide ds 5'-AGACATGCCTAGACATGCCT-3' (MWG, Germany) containing the p53 consensus sequence was used. Supercoiled plasmid DNAs pPGM1 (pBluescript II SK7 containing a 20-mer [AGACATGCCTAGACATGCCT] p53 consensus binding site (ElDeiry et al., 1992) cloned into the HindIII site (Palecek et al., 1997b)) and pBluescript II SK7 were used. Fragments from the same plasmids were generated by PvuII digestion. To remove EDTA from the DNA solution or after enzymatic cleaving, DNA was ethanol-precipitated and dissolved in water or 5 mM Tris-HC1, pH 7.5. Puri®cation of p53 Human p53 protein was expressed, isolated and puri®ed as previously described (Palecek et al., 1997b). The concentration of p53 was determined with a protein dye-binding procedure according to Bradford using bovine serum albumin as a standard (Bradford, 1976). Mundt et al. (1997) obtained p53 fractions on a heparin column distinguishable by their speci®c DNA binding activities to p53CON in a DNA fragment and by their recognition by PAb421. Quite recently, we used the same approach to characterize fractions by their ability to bind scDNA; our preliminary results show that these fractions also dier in their binding activities to scDNA. Further details are being prepared for publication elsewhere. In this paper, we used fractions 22 and 23 (with 0.4 ± 0.5 M KC1 in the mobile phase) and the corresponding fraction WD4 isolated in the absence of DTT.

3623


3624

DNA binding assay DNA (usually 0.3 mg), the metal solution and p53 were mixed at dierent ratios (the p53/DNA ratio was calculated from the concentrations of DNA molecules and p53 tetramers) in 20 ml of the DNA binding buer (5 mM TrisHC1, pH 7.5, 50 mM KC1 and 0.01% Triton X-100). Samples were incubated at 08C for 30 min (if not stated otherwise) and loaded on a 1% agarose gel containing 0.336TBE buer, pH 8.0. Agarose electrophoresis was performed for 3 ± 4 h at 100 V and 48C. Gels were stained with ethidium bromide, photographed, and subsequently immunoblotted.

Determination of DTT by constant current chronopotentiometric stripping analysis (CPSA) At neutral pH, DTT produces a CPSA peak at about 70.5 V (against the Ag/AgC1/3 M KC1 reference electrode) which can be utilized for the determination of submicromolar concentrations of DTT (L Havran, M Tomschik and E Palecek, unpublished). This technique can also be applied for the analysis of DNA and proteins (Palecek et al., 1997a; Tomschik et al., 1998). Measurements were performed with an AUTOLAB (EcoChemie, Utrecht, The Netherlands) connected to a Metrohm 647 VA Stand (mercury electrode) in HMDE mode (drop surface area 0.4 mm2) at a constant current of 71 mA using 0.2 M sodium phosphate as a background electrolyte.

Blotting with immunodetection of p53 Gels were blotted onto a nitrocellulose transfer membrane Protran R (Schleicher and Schuell, Germany) in 206SSC at 5 inches Hg on a vacuum blotting system (BioRad, USA) (Sambrook et al., 1989). The membrane was blocked with 5% milk in PBS and p53 was detected with 1 mg/ml primary antibody DO-1 supernatant diluted 1 : 5 ± 1 : 15 and secondary antibody anti-mouse IgG (alkaline phosphatase conjugate) diluted 1 : 1000. P53 was visualized with BCIP (5-bromo-4chloro-3-indolylphosphate) and NBT (nitrobluetetrazolium chloride) (Harlow and Lane, 1988).

Acknowledgments The authors are grateful to Drs M Fojta and M Sheard for stimulating discussions and critical reading of the manuscript. This work was supported by a grant of the Grant Agency of the Czech Republic Nos. 204/96/1680 and 301/ 99/0692, by Volkswagen Stiftung to E Palecek and Dr T M Jovin and by a grant of the Grant Agency of the Academy of Sciences CR A5004803 to E Palecek, B Vojtesek was partially supported by the grants of IGA MH CR Nos.: 3477-3, 4783-3 and MSMT No. VS 96154.

References Arrowsmith CH and Morin P. (1996). Oncogene, 12, 1379 ± 1385. Beltran R, Martinez-Balbas A, Bernues J, Bowater R and Azorin F. (1993). J. Mol. Biol., 230, 966 ± 978. Berg JM and Shi Y. (1996). Science, 271, 1081 ± 1085. Bernues J, Beltran R, Casasnovas JM and Azorin F. (1989). EMBO J., 8, 2087 ± 2094. Bradford MM. (1976). Anal. Biochem., 72, 248 ± 254. Bullock AN, Henckel J, Dedecker BS, Johnson CM, Nikolova PV, Proctor MR, Lane DP and Fersht AR. (1997). Proc. Natl. Acad. Sci. USA, 94, 14338 ± 14342. Cho Y, Gorina S, Jerey PD and Pavletich NP. (1994). Science, 265, 346 ± 355. Christianson DW. (1991). In: Advances in protein chemistry,Vol. 42. An®nsen CB, Edsall JT and Richards FM. (eds). Academic Press: San Diego, pp. 281 ± 353. Coer AI and Knowles PP. (1994). Biochim. Biophys. Acta, 1209, 279 ± 285. Cornell NW and Crivaro KE. (1972). Anal. Biochem., 47, 203 ± 208. Cox LS and Lane DP. (1995). BioEssays, 17, 501 ± 508. Ebadi M and Iversen PL. (1994). Gen. Pharmac., 25, 1297. El-Deiry WS, Kern SE, Pietenpol, JA, Kinzler KW and Vogelstein B. (1992). Nature Genet, 1, 45 ± 49. Elledge RM and Lee W-H. (1995). BioEssays, 17, 923 ± 930. Glusker JP. (1991). In: Advances in protein chemistry, Vol. 42. An®nsen CB, Edsall JT, Richards FM and Eisenberg DS. (Eds). Academic Press: San Diego, pp. 1 ± 75. Hainaut P and Milner J. (1993). Cancer Res., 53, 1739 ± 1742. Hainaut P, Rolley N, Davies M and Milner J. (1995). Oncogene, 10, 27 ± 32. Han W, Dlakic M, Zhu YJ, Lindsay SM and Harrington RE. (1997a). Proc. Natl. Acad. Sci. USA, 94, 10565. Han W, Lindsay SM, Dlakic M and Harrington RE. (1997b). Nature, 386, 563. Harlow E and Lane D. (1988). Antibodies. A Laboratory Manual, 497 ± 501. Cold Spring Harbor Laboratory. Hunt JB, Neece SH and Ginsburg A. (1985). Anal. Biochem., 146, 150 ± 157. Hupp TR and Lane DP. (1994). Curr. Biol., 4, 865 ± 875.

Jacob C, Maret W and Vallee BL. (1998). Proc. Natl. Acad. Sci. USA, 95, 3489. Klevickis C and Grisham CM. (1996). In: Metal ions in biological systems. Interactions of metal ions with nucleotides, nucleic acids and their constituents. Vol. 32. Sigel A and Sigel H. (eds). Marcel Dekker: New York, pp. 1 ± 26. Ko LJ and Prives C. (1996). Genes Dev., 10, 1054 ± 1072. Laundon C and Grith JD. (1987). Biochemistry, 26, 3759. Levine AJ. (1997). Cell, 88, 323 ± 331. Martinez-Balbas A and Azorin F. (1993). Nucleic Acids Res., 21, 2557 ± 2562. Mundt M, Hupp T, Fritsche M, Merkle C, Hansen S, Lane D and Groner B. (1997). Oncogene, 15, 237 ± 244. Nagaich AK, Appella E and Harrington RE. (1997a). J. Biol. Chem., 272, 14842 ± 14849. Nagaich AK, Zhurkin VB, Sakamoto H, Gorin AA, Clore GM, Gronenborn AM, Appella E and Harrington RE. (1997b). J. Biol. Chem., 272, 14830 ± 14841. Nagel WW and Valle BL. (1995). Proc. Natl. Acad. Sci. USA, 92, 579 ± 583. Ortiz-Lombardia M, Eritja R, Azorin F, Kypr J, Tejralova I and Vorlickova M. (1995). Biochemistry, 34, 14408 ± 14415. Palecek E. (1991). Crit. Rev. Biochem. Mol. Biol., 26, 151 ± 226. Palecek E, Tomschik M, Stankova V and Havran L. (1997a). Electroanalysis, 9, 990 ± 997. Palecek E, Vlk D, Stankova V, Brazda V, Vojtesek B, Hupp TR, Schaper A and Jovin TM. (1997b). Oncogene, 15, 2201 ± 2209. Parks D, Bolinger R and Mann K. (1997). Nucleic Acids Res., 25, 1289 ± 1295. Pavletich NP, Chambers KA and Pabo CO. (1993). Gene Dev., 7, 2556 ± 2564. Sabat M and Lippert B. (1996). In: Metal ions in biological systems. Probing of nucleic acids by metal ion complexes of small molecules. Vol. 33. Marcel Dekker: New York, pp. 143 ± 176.


Sambrook J, Fritsch EF and Maniatis T. (1989). Molecular Cloning. A Laboratory Manual, 2nd edn, 9.36 ± 9.37. Cold Spring Harbor Laboratory Press. Soussi T and May P. (1996). J. Mol. Biol., 260, 623 ± 637. Stephen CW, Helminen P and Lane DP. (1995). J. Mol. Biol., 248, 58 ± 78. Tomschik M, Havran L, Fojta M and Palecek E. (1998). Electroanalysis, 10, 403 ± 406. Vallee BL and Auld DS. (1990). Biochemistry, 29, 5647 ± 5659.

Verhaegh GW, Parat M-O, Richard M-J and Hainaut P. (1998). Mol. Carcinog., 21, 205 ± 214. Vlk D. (1997). PhD-Thesis, Institute of Biophysics, Academy of Sciences of the Czech Republic, Brno. Vojtesek B, Bartek J, Midgley CA and Lane DP. (1992). Immunological Methods, 151, 237 ± 244. Yagil G. (1991). Crit. Rev. Biochem. Mol. Biol., 26, 475 ± 559.

3625