Differences in the Kinetic Properties of BarnHI Endonuclease and Methylase with Linear DNA Substrates*

BamHI endonuclease and methylase were found to exhibit a kinetic preference for linear pBR322 DNA substrates containing the recognition site in a central location. The K,,, values for substrates having the recognition site in a terminal location were approximately %fold greater than those with a centrally located site. This phenomenon may be partially due to facilitated transfer of the enzymes to the recognition site over nonspecific flanking sequences. The exploitation of facilitated transfer by these enzymes has been inferred from studies demonstrating kinetic preferences for longer DNA substrates. The reaction rates of the endonuclease were 9-fold greater with full-length pBR322 DNA than with a 74-base pair derivative. The methylase exhibits a kinetic preference for longer sub- strates but only under conditions of comparatively higher DNA concentrations. In addition, the methylase has the property of increasing long chain preference with increasing salt concentrations up to 120 mM. In- creasing salt concentrations decreased the endonucle-ase’s preference for longer substrates. Nonspecific in- hibition studies revealed qualitative and quantitative differences between the two enzymes under catalytic conditions. These studies suggest that BamHI endonuclease and methylase interact with nonspecific DNA in different ways. The restriction-modification system of the duplex symmetrical sequence 5’-GGATCC-3’. The double-stranded cleavage between and could be readily fractionated and quantitated. BamHI methylase reactions were quenched with MgC1, to a final concentration of 25 mM. Twenty units of BamHI endonuclease were added per pg of DNA, and the mixture was incubated for 1 h at 37 "C. Controls were identical, except no methylase was present, and showed complete cleavage of DNA by BamHI and AuaI endonu- cleases at all salt concentrations and with all types of linear substrates used in these studies. Unit assays showed that the activity of BamHI and AuaI endonucleases was not affected in the presence of 35 PM AdoMet, BamHI methylase, and MgC1,. The initial velocity of BamHI methylation was linear with respect to enzyme concentration from 0.03 to 1.2 nM in the presence of saturating substrates (25 nM Form I pBR322 DNA and 35 p~ AdoMet). The velocity of endonuclease ChVage was linear with enzyme concentration from 0.03 to 1 nM in the presence of 25 nM pBR322 DNA. The velocity of cleavage and methylation was linear for at least 10 min in reactions having DNA/ enzyme concentration ratios of 4 or greater.

the reaction rates of certain restriction-modification enzymes (7,8). The nature of flanking sequences immediately adjacent to the recognition site appear to influence the kinetic behavior of EcoRI, BanHI, and PstI endonucleases and dam methylase (9)(10)(11)(12). However, nonspecific flanking DNA that is located over 1000 base pairs away from a recognition site also has large effects on the kinetic ability of EcoRI endonuclease to form specific enzyme-substrate complexes (13). These longrange effects have been theoretically explained by thermally driven facilitated diffusion mechanisms that can involve the linear transfer of a protein along the DNA contour length, intradomain dissociation-association events, and intersegment exchange of a protein between areas of the DNA polymer brought into close proximity by conformational fluxes (14). Such intramolecular-facilitated transfer mechanisms are expected to enhance specific binding rate constants of a protein since they are inherently faster than random sampling by three dimensional diffusion. Utilization of facilitated diffusion may be of adaptive value to restriction-modification enzymes, RNA polymerases in the targeting of promoters (15), and repressors (16,17) considering the great molar excess of nonspecific sequences and the small diffusion constants of large, solvated DNA molecules. In this article we report the effects on the reaction rates of BamHI endonuclease and methylase as the distance between the recognition site and a DNA terminus is changed. Some of the kinetic observations described can be qualitatively explained in terms of facilitated diffusion. However, the methylase exhibits unique behavior that differs from the endonuclease with respect to salt and substrate concentrations. Implications on the nonspecific interactions with DNA are discussed. Preliminary results on this work have been previously reported (18).

Materials
Bacterial alkaline phosphatase, T d polynucleotide kinase, Bal-31 nuclease, agarose, acrylamide, NACS' resins and restriction endonucleases (unless otherwise noted) were purchased from Bethesda Research Laboratories. [y3'P]ATP (2900 Ci/mmol) was purchased from New England Nuclear. AdoMet and ethidium bromide were from Sigma. Proteinase K was purchased from Boehringer Mannheim. All other reagents were of the highest available purity.
DNA-pPBR322 DNA was purified from Escherichia coli HBIOI using base-acid extraction and chromatography on NACS-37 resin (20). Form I pBR322 DNA was linearized with either NdeI or EcoRI endonucleases. This produced full-length molecules with the BamHI cleavage site 1921 or 375 bases from the nearest end. The protein in all preparative restriction digests was removed by the addition of proteinase K to a concentration of 100 pg/ml (incubated for 0.5 h at 37 "C) and phenol-chloroform extraction. The DNA was ethanol precipitated in the presence of 2.5 M ammonium acetate and dried under vacuum. The DNA was subsequently purified by chromatography on NACS-57 resin. Restriction fragments 1000 base pairs or greater were isolated from agarose gels by electroelution. Smaller fragments were isolated from 8% polyacrylamide gels. The DNA fragments were filtered through Millex-G-50 0.22 Fm units (Millipore), concentrated, and purified on NACS-57 resin. DNA in all preparative gels was visualized at 366 nm after ethidium bromide staining. DNA fragments were also end-labeled with T, kinase, and the unincorporated ATP was removed on NACS-57 resin. Specific activities ranged from 9.6 X 10' to 1 x lo' cpm/pg DNA.
Protein and DNA Quuntitation-Protein was determined by the method of Bradford (21) using bovine gamma globulin as a standard. DNA was determined at 260 nm.
Quuntitation of Reaction Products-Densitometric analysis was performed by a modification of published procedures (23,24). Gels were stained in 5 pg/ml of ethidium bromide for 15 min and photographed for 15 s at 305 nm. Gels were exposed to ultraviolet light only during photography. Negatives were scanned at 595 nm in a Bio-Rad 1650 densitometer equipped with a Spectro Physics SP4270 integrator. Film response was linear from 1 to 400 ng of DNA. Film calibration was performed with EcoRI or H i d 1 1 digestion products of X-DNA (agarose) or HaeIII digestion products of phiX174RF DNA (polyacrylamide). No significant differences in fluorescence was detected between known amounts of various fragments of pBR322 DNA generated by combinations of BarnHI, EcoRI, AuaI, and NdeI endonucleases as judged by calibration curves. Quantitation was performed by calculating the percent of a given species of DNA in the total digest or by comparison to known amounts of restriction fragments run in parallel. Alternatively, radioactive cleavage products were excised from gels and counted in a Packard liquid scintillation spectrometer. Both methods of analysis were in good agreement.
Enzyme Assays-Ten-microliter aliquots from reactions containing BamHI endonuclease and full-length linear pBR322 DNA were placed at 85 "C for 1 min. Time course studies indicated that this temperature irreversibly deactivated catalytic amounts of the endonuclease within 10 s. BamHI endonuclease and methylase reactions containing EcoRI-linearized DNA were subjected to secondary digests with 5 units of AuaI endonuclease/0.3 pg of DNA at 45' for 0.5 h. This produced fragments that could be readily fractionated and quantitated. BamHI methylase reactions were quenched with MgC1, to a final concentration of 25 mM. Twenty units of BamHI endonuclease were added per pg of DNA, and the mixture was incubated for 1 h at 37 "C. Controls were identical, except no methylase was present, and showed complete cleavage of DNA by BamHI and AuaI endonucleases at all salt concentrations and with all types of linear substrates used in these studies. Unit assays showed that the activity of BamHI and AuaI endonucleases was not affected in the presence of 35 PM AdoMet, BamHI methylase, and MgC1,. The initial velocity of BamHI methylation was linear with respect to enzyme concentration from 0.03 to 1.

RESULTS
The reaction rates of BamHI endonuclease and methylase were examined with linear pBR322 DNA substrates having the BamHI cleavage site 375 or 1921 bases from the nearest end (EcoRI and NdeI linearized, respectively). Under standard reaction conditions and a DNA concentration of 12 nM the endonuclease and methylase exhibited faster reaction rates as the recognition site was located further from an end (Figs. 1, A and D). This kinetic preference was independent of enzyme concentrations in the range of 0.03 to 2 nM. Cleavage and methylation rates with PstI-and AuaI-linearized pBR322 DNA (1125 bases 5' and 1049 bases 3' to the BamHI cleavage site, respectively) were identical within experimental error and were intermediate between those rates obtained with the EcoRI-and NdeI-linearized substrates. Increasing concentrations of NaCl decreased the differences between cleavage rates (Fig. 1B) and any kinetic preference for a centrally located site was abolished at a NaCl concentration of 160 mM (Fig. 1C). In contrast, a difference of approximately 2-fold remained between methylation rates at NaCl concentrations in the range of 100 to 200 mM ( Fig. 1, E and F). In addition, the overall methylation rates were fastest at a NaCl concentration of 100 mM which is a result previously observed with supercoiled substrates (6).
Initial velocity kinetics were used to investigate the substrate concentration dependence of the observed reaction rate differences in the range of 1.5 to 35 nM DNA in the absence of added NaCl. Reciprocal plots revealed that the endonuclease and methylase have K , values for NdeI-linearized pBR322 DNA of 3.6 and 3.2 nM, respectively. The K, values for EcoRIlinearized DNA were 12.5 nM for the endonuclease and 11.2 nM for the methylase. Each enzyme had the same V,,, with both substrates. These results suggest that binding is more favorable for a centrally located recognition site. No change in the methylase kinetics with AdoMet were observed when either DNA was used as the nonvarying substrate.
The reaction rate differences described for the endonuclease and methylase at low salt concentrations could originate from one or more types of facilitated diffusion. If these enzymes can locate their recognition sites by a sliding mechanism, then an enhancement of this process would be expected to occur as the recognition site is located in a more central position. This would be a manifestation of the proportionately larger target area for nonspecific binding around the BamHI site. Implicit in this assumption is that the average scanning length of the enzymes is less than 3987 and greater than 375 base pairs (the longest and shortest distances between a DNA terminus and the BamHI cleavage site on EcoRI-linearized pBR322 DNA). A sliding mechanism has been implicated in the EcoRI endonuclease reaction as judged by increases in the specific association and dissociation rate constants with increasing DNA chain length up to approximately 4000 base pairs (13). The average scanning length was determined to be 1300 base pairs. In addition, the kinetics of cleavage were faster with longer substrates. This phenomenon, along with the increases in specific association rate constants with increasing substrate length, can be a result of the larger target area for nonspecific binding and the subsequent facilitated transfer of the enzyme to the recognition site.
In order to ascertain if facilitated diffusion is kinetically evident for BamHI, endonuclease and methylase fragments of pBR322 DNA were prepared with an attempt to place the BamHI site in a central location (Table I). Under standard reaction conditions and DNA concentrations of 1 nM, the initial rate of cleavage increased with increasing substrate length (Fig. 2). However, in the presence of 160 mM NaCl the   kinetic preference for longer substrates was markedly smaller as exemplified by a reduction in the velocity ratio of the 4362:74-base pair substrates from 9.0 to 1.9 (Fig. 2). The decrease in the kinetic preference for longer substrates at higher concentrations of salt is consistent with the concomitant decrease in the electrostatic component of nonspecific binding (13,14). The possibility of different inhibitory impurities in the DNA fragment preparations was tested by mixing equimolar amounts of the 74-and 4362-base pair substrates in the same reaction. A kinetic preference was again observed, with a ratio of long/short cleavage rates of 7.8 at 1 nM DNA. Furthermore, a kinetic preference for the longer substrate was observed over a 10-fold concentration range with a long/short ratio of approximately 5.0 at 12 nM DNA. NaCl concentrations of 160 mM eliminated long chain preference at both substrate concentrations in these mixing experiments.
The kinetic behavior of BamHI methylase with substrates of various lengths differed from the endonuclease under similar reaction conditions. The velocity of methylation decreased in the range of 74 to 1,000 base pairs but increased in the range 1,500-4,362 base pairs (Fig. 3A). Mixing experiments with selected fragments revealed similar kinetic preferences: 74:1,500 = 4.3; 74:4,362 = 2.6; 4,362:1,000 = 2.8. It is tempting to explain the decrease in methylation rates between the 74-and 1,000-base pair fragments (Fig. 3A) in terms of diffusion-controlled mechanisms. In this context, the decrease in the free diffusion constants with increasing substrate length could kinetically dominate initial association events between enzyme and DNA in the absence of facilitated diffusion mechanisms. The native molecular weight of BamHI methylase is approximately 65,000 (6). Assuming a spherical shape, the diffusion constant is in the vicinity of 6 X cm2/s. Estimates of the free diffusion constants for the DNA fragments from 74 to 1,000 base pairs are in the approximate range of 3 X to 5 X lo-' c m ' /~.~ Changes in reaction Estimates for the translational frictional coefficient of the 74base pair fragment were based on a rigid rod configuration while those for the 1000-base pair fragment were based on a random coil. In both calculations a nondraining model was assumed. kinetics might be observed with 2-or &fold differences between the diffusion constants of the enzyme and DNA. However, larger differences would probably be masked in these studies because of the comparatively fast diffusion rate of the enzyme. Therefore, it seems unlikely that the decrease in methylation rates are due to substrate diffusion alone.
The increases in methylation velocity with substrates 1500 base pairs and longer (Fig. 3A) suggest that facilitated diffusion is a contributing factor. At NaCl concentrations of 200 mM and a DNA concentration of 1 nM, these increases in reaction rates were eliminated without changing the trend of velocity decreases from 74 t o 1000 base pairs (Fig. 3B). At DNA concentrations of 12 nM, in the absence of added salt, a comparatively small methylation preference for fragments larger than 74 base pairs was observed (Fig. 3C). An increase in the NaCl concentration to 100 mM increased the preference for longer substrates at these higher DNA concentrations (Fig. 3C). This preference was also observed at NaCl concentrations as high as 200 mM. The salt dependency of methylase long chain preference a t high DNA concentrations was further studied in reactions containing equimolar concentrations of the 4362-and 376-base pair substrates (Fig. 30). An increasing preference for the longer substrate was observed with increasing the NaCl concentrations up to 120 mM.
The relative nonspecific binding affinities of these enzymes was examined by inhibition experiments. The initial velocities of cleavage and methylation with the 74-base pair substrate were assessed in the presence of different concentrations of a NACS purified 3986-base pair fragment of pBR322DNA lacking the BamHI site. In the range of 0.1-1 p~ nonspecific base pairs, no significant inhibition of methylase activity was observed. In contrast, endonudease activity was inhibited 20% in the same concentration range (Fig. 4). Significant inhibition of methylase activity did not occur until the nonspecific nucleotide concentration was greater than 1.5 p~. In the range of 2-75 p~ base pairs, the rapid increase in methylase inhibition was considerably greater than that of the endonuclease. The endonuclease exhibited a biphasic inhibition pattern.

DISCUSSION
The reaction rates of BamHI endonuclease and methylase are affected by the proximity of their recognition site to the end of a linear DNA molecule. These effects appear to be correlated with the distance, and not the direction (5' or 3'), of the nearest DNA terminus. This suggests that the nature of the intervening flanking sequences between the recognition site and a terminus is not responsible for the variation in reaction rates. The dependence of the reaction rates on the relative location of the recognition site can be theoretically explained in terms of facilitated transfer of the enzymes via nonspecific sequences. This assumption requires that the enzymes have an average scanning length. The average scanning length is a function of the salt concentration since it affects the nonspecific binding constant (14). The kinetic preference for a centrally located site may exist because there is a functionally greater extension of nonspecific target area around the BamHI palindrome in NdeI-linearized pBR322 DNA (Fig. 1). Therefore, a greater proportion of enzyme molecules would be expected to come into the vicinity of the recognition site from both flanks. Recent studies have shown that EcoRI endonuclease has a kinetic preference for centrally located recognition sites on linear DNA (25). The authors demonstrated that facilitated transfer could account for this effect. They suggest that the kinetic preference can be explained by an increased frequency of entry to the recognition site from the proportionately larger area of nonspecific target sequences.
The differences in K, values for the NdeI-and EcoRIlinearized substrates is supportive evidence for the occurrence of facilitated transfer since this process ultimately affects specific binding. However, it has been suggested that the difference in K, values is too large if a sliding mechanism was solely responsible for the differences in reaction rates. The maximum kinetic difference generated by a sliding model (14) can be calculated by the relationship: to 10-fold below initial substrate concentrations. Both enzymes are inhibited in a competitive manner by phiX174 RF DNA (contains no BamHI site) in the range of 1-10 nM as judged by initial velocity kinetics using Form I pBR322 DNA as substrate. This suggests that association with substrate DNA is rate limiting since the added DNA slows the arrival of enzyme in a manner directly proportional to inhibitor. The competitive effect is also observed at a substrate DNA concentratiop of 12 nM. Theoretical models of facilitated diffusion, and their application, require that once the protein is bound to DNA the complex can be considered as a closed system. This is because facilitated diffusion is an intramolecular phenomenon. The BamHI system probably satisfies this requirement in several respects. Rates of enzyme-DNA dissociation in these systems are low, on the order of minutes and the turnover number of both enzymes are small, approximately 1.3 min", at 25 nM DNA. The time frame in which catalysis occurs after specific association would not be expected to vary significantly since the BamHI sites in all fragments are identical with respect to flanking sequences within at least 35 base pairs. Variability in the stability of the specific complex might occur. It was found that the half-life of specific complexes with EcoRI endonuclease dramatically decreased with increasing substrate length, but the overall specific binding equilibrium remained unaffected (13). These compensatory changes between the specific association and dissociation rate constants might not apply to the BamHI system, particularly not to the methylase. Differences in the overall binding affinities between large DNAs and fragments 200 base pairs and smaller were found to occur with lac repressor (17). These possibilities could be involved in the kinetic profiles observed with the methylase at several DNA concentrations. Binding of AdoMet might also be affected by chain length but did not appear to be altered (on the basis of reaction rates) by using either the NdeI-or EcoRI-linearized substrates. Differences between the reaction rates with these two substrates were still apparent at concentrations of 1 nM, yielding a velocity ratio for NdeIIEcoRI-linearized DNAs on the order of 1.8. This result indicates that the differences in the kinetic properties between these DNA molecules are preserved over a 10-fold concentration range. The increase in the velocity of endonuclease cleavage with increasing substrate length is suggestive of facilitated transfer. This could be due to the greater extension of nonspecific target area in longer fragments. At a NaCl concentration of 160 mM long chain preference and cleavage rate differences between the NdeI-and EcoRI-linearized substrates are negligible (Figs. 2 and IC). These observations are consistent with a decrease in the electrostatic component of nonspecific binding and the subsequent decrease in facilitated transfer (13,14). While this manuscript was in preparation, Ehbrecht et al. reported a kinetic preference of BamHI endonuclease for longer substrates (26) which has confirmed our own results. The nonlinear profile of cleavage and methylation rates with different-sized substrates (Figs. 2 and 3C) indicates that there is a limit to the range of facilitated transfer. Extrapolation of this data gives an approximation of 1330 base pairs as an upper limit to the scanning length. This value is strikingly similar to the average scanning length determined for EcoRI endonuclease at an ionic strength of 0.08 (13). However, the scanning length of BamHI endonuclease may be different if it was determined at this higher ionic strength.
BamHI endonuclease and methylase show complex inhibition profiles with the 74-base pair substrate (Fig. 4). The biphasic inhibition pattern observed with the endonuclease may indicate that the enzyme possesses two binding sites with different affinities for non-specific DNA. Saturation of the stronger site results in 20% inhibition of activity while saturation of the weaker site almost totally inhibits activity. One of these sites might be the active site and both sites would be interactive assuming this interpretation is valid. The methylase might bind two non-specific DNA fragments cooperatively as judged by the abrupt increase in inhibition at 2 PM base pairs. These competition curves indicate that the enzymes nonspecific association with DNA are qualitatively different. This may account for some of the differences seen in the previous kinetic experiments. We are currently investigating the effects on inhibition when the length of the nonspecific DNA is varied.
The kinetic behavior of the methylase at low DNA concentrations (Fig. 3, A and B ) cannot be fully explained at this time. The salt sensitivity of the methylation rates with DNA fragments 1500 base pairs and longer might be indicative of facilitated transfer. The gradual increase in methylation velocities with these fragments might be involved with the ability of longer DNA molecules to form domains. Longer molecules would also be expected to have greater segmental density. Consequently, increases in uncorrelated translocation processes such as intersegment transfer of protein or intradomain macroscopic dissociation-reassociation events can occur. Such translocation mechanisms are also decreased by increases in salt concentrations (14). Intersegment transfer requires the protein to be transiently bound at two nonspecific sites before exchange occurs. This criterion may be met by the methylase since the nonspecific inhibition studies (Fig. 4) suggest the possibility of two or more nonspecific binding sites. If these mechanisms are occurring, they are not accelerating reaction rates efficiently enough to overcome the methylation velocities observed with the 74-and 376-base pair substrates (Fig. 3A). The trend of velocity decreases over the range of the smaller fragments at low concentrations may be a reflection of different overall binding affinities or different modes of inhibitory, nonspecific binding that is not readily overcome by high NaCl concentrations.
Nonspecific, flanking DNA can slow down a reaction by providing competitive binding sites, or it can be the medium in which facilitated transfer occurs. Theoretical results predict a range of nonspecific binding constants that promote optimal increases in specific association rate constants (14). These predictions have been verified by experiments with lac repressor in which increases in KC1 concentrations to approximately 100 mM promoted increases in the specific association rate constants with long DNA to maximum values (17). Increases in KC1 past this point drastically lowered association constants below their original values in the low salt case. Since salt strongly affects the nonspecific binding constant, these results suggest that a marginal decrease in nonspecific binding affinity allows for a greater amount of productive interactions that lead to sliding as opposed to the protein being trapped in an unproductive, nonspecific binding mode. A similar situation might be occurring with BamHI methylase and could explain its enhanced long chain preference at NaCl concentrations of 100-120 mM (Fig. 3, C and D). Fig. 3 0 indicates residual long chain preference at NaCl concentrations as high as 200 mM. This comparatively refractory behavior to high salt concentrations at high DNA concentrations is consistent with the persistance of the differences between methylation reaction rates observed with the NdeIand EcoRI-linearized substrates at 200 mM NaCl (Fig. 1F). It is difficult to assess the range of these effects since NaCl concentrations exceeding 220 mM cause severe inhibition of methylase activity. The persistance of the differences in methylation rates between the full-length substrates could be a consequence of residual nonspecific binding at high DNA concentrations. The effects of salt on the enzyme's conformation (which may affect specific and nonspecific binding) and on its catalytic mechanism could create another dimension of kinetic consequences that have to be evaluated. Preliminary initial velocity studies using Form I pBR322 DNA indicate an increase in Vmax with NaCl concentrations up to 100 mM. Changes in catalytic efficiency at various concentrations of NaCl may contribute to kinetics observed in these experiments. Assuming that the catalytic mechanism does not significantly change with the length of nonspecific flanking DNA, salt-induced V,,, effects should remain the same in studies utilizing different-sized fragments at the same concentration of NaCl. As stated previously, the size ofthe difference between the K,,, values for the NdeI-and EcoRI-linearized substrates suggests that sliding may not be totally responsible for the variation in reaction rates. Other factors that contribute to this phenomenon may be relatively insensitive to high salt concentrations. Nonspecific binding by the methylase may also involve other interactions in addition to electrostatic forces. Hydrophobic interactions are stabilized by salt and could be partially involved with nonspecific binding.
Collectively, these data imply that BarnHI endonuclease and methylase differ in their nonspecific interactions with DNA. However, on the basis of initial velocity kinetics both enzymes share approximately the same binding affinity for the BamHI site. The presence of AdoMet may also affect the nonspecific interactions of the methylase. On the basis of the studies presented here, it seems reasonable to conclude that facilitated diffusion processes are at least partially responsible for the variations in reaction rates of BarnHI endonuclease and methylase with changes in the position of the recognition site and substrate length. A more complete appreciation of the experimental results awaits further kinetic studies with various sized DNA fragments, the effects of AdoMet on specific and nonspecific complex formation, and substrate binding experiments. Investigations into other restrietion-modification systems may demonstrate the occukende of facilitated difusion mechanisms. The exploitation of facilitated transfer by restriction-modification enzymes could be a general phenomenon used to expedite sequence-specific interactions with DNA.