HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Submit a Letter to Editor Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lindstrom, W. M. Articles by Reich, N. O. Search for Related Content PubMed PubMed Citation Articles by Lindstrom, W. M., Jr. Articles by Reich, N. O. Social Bookmarking                      What's this?

J Biol Chem, Vol. 275, Issue 7, 4912-4919, February 18, 2000

Reconciling Structure and Function in HhaI DNA Cytosine-C-5 Methyltransferase^*

William M. Lindstrom Jr., James Flynn^§¶, and Norbert O. Reich^§

From the  Department of Chemistry and Biochemistry and ^§ Program in Biochemistry and Molecular Biology, University of California, Santa Barbara, California 93106

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Pre-steady state partitioning analysis of the HhaI DNA methyltransferase directly demonstrates the catalytic competence of the enzyme·DNA complex and the lack of catalytic competence of the enzyme·S-adenosyl-L-methionine (AdoMet) complex. The enzyme·AdoMet complex does form, albeit with a 50-fold decrease in affinity compared with the ternary enzyme·AdoMet·DNA complex. These findings reconcile the distinct binding orientations previously observed within the binary enzyme·AdoMet and ternary enzyme·S-adenosyl-L-homocysteine·DNA crystal structures. The affinity of the enzyme for DNA is increased 900-fold in the presence of its cofactor, and the preference for hemimethylated DNA is increased to 12-fold over unmethylated DNA. We suggest that this preference is partially due to the energetic cost of retaining a cavity in place of the 5-methyl moiety in the ternary complex with the unmethylated DNA, as revealed by the corresponding crystal structures. The hemi- and unmethylated substrates alter the fates and lifetimes of discrete enzyme·substrate intermediates during the catalytic cycle. Hemimethylated substrates partition toward product formation versus dissociation significantly more than unmethylated substrates. The mammalian DNA cytosine-C-5 methyltransferase Dnmt1 shows an even more pronounced partitioning toward product formation.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

DNA methyltransferases sequence-specifically modify DNA in a wide range of organisms (1). The enzymes require the methyl donor S-adenosyl-L-methionine (AdoMet)¹ to modify the exocyclic amine (for cytosine-N⁴-specific methyltransferases (EC 2.1.1.113) and adenine-specific methyltransferases (EC 2.1.1.72)) or the 5-carbon ((cytosine-5-)-methyltransferases (EC 2.1.1.37)) of their target base. The family of exocyclic amine methylating enzymes is expected to share a common chemical mechanism (2) as is the family of (cytosine-5-)-methyltransferases (3, 4). Methyltransferases from each family have been shown to stabilize the target nucleoside out of the DNA helix (5, 6), and nucleoside "flipping" is expected to be a common strategy employed by DNA methyltransferases (7).

HhaI DNA methyltransferase (M.HhaI) modifies the carbon at position 5 (C-5) of the inner cytosine in the double-stranded cognate sequence 5'-GCGC-3'. Nucleophilic attack on C-6 of the target cytosine by Cys⁸¹ forms a covalent intermediate and activates C-5 for nucleophilic attack on the methylsulfonium of AdoMet. Methyl group transfer is followed by proton abstraction from C-5, -elimination of the enzyme-DNA adduct, and product dissociation. M.HhaI follows a rapid equilibrium ordered mechanism with a substrate containing multiple cognate sites in which the enzyme binds DNA prior to the cofactor (8). The x-ray crystal structure of the enzyme bound to AdoMet indicates a two domain organization in which a DNA binding cleft separates catalytic and target-sequence recognition domains (9). That AdoMet was bound in the M.HhaI crystal structure was surprising in light of the proposed kinetic mechanism. The crystal structure of the ternary complex formed with enzyme, DNA, and cofactor product, donor S-adenosyl-L-homocysteine (AdoHcy), reveals the target cytosine flipped completely out of the DNA helix and into the active site of the enzyme (5).

The two domain organization of M.HhaI is common to all DNA methyltransferases (4, 9-12). The catalytic domain of the catechol O-methyltransferase is structurally similar to that of M.HhaI and other DNA methyltransferases (13). Furthermore, the AdoMet binding pocket revealed in M.HhaI crystal structures is expected to be common to protein, DNA, RNA, and small molecule AdoMet-dependent methyltransferases (14, 15). Enzyme-mediated nucleoside flipping, first observed in M.HhaI, occurs with other DNA methyltransferases (6, 16) and DNA repair enzymes (17, 18). M.HhaI thus embodies structural and functional characteristics common to DNA repair enzymes, AdoMet-dependent enzymes, and DNA methyltransferases.

In view of this rich structural context, we sought to provide complementary mechanistic insights. We characterized the energetic contributions toward the binding preference of the enzyme with hemimethylated substrates. This binding discrimination is enhanced with the cofactor. Inspection of the available cocrystal (19, 20) shows that some of this discrimination derives from the maintenance of a cavity present in the structure containing unmethylated DNA and occupied by the methyl group in the structure containing hemimethylated DNA. Similarly, previous crystallography studies identified two binding orientations for AdoMet (5, 9). Isotope partitioning and protein fluorescence studies show that the low affinity enzyme·AdoMet complex is not catalytically competent. Mechanistic studies show that the partitioning of enzyme·substrate intermediates is modulated by hemi- and unmethylated substrates.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Enzyme Expression and Purification-- M.HhaI was expressed from vector pHSHW-5 in Escherichia coli strain ER1727 (kindly provided by S. Kumar, New England Biolabs) Purification was according to Greene et al. (21). Protein concentration was determined by active site titration (22).

Oligonucleotide Synthesis and Purification-- Mechanistic analysis required the use of a single-site DNA substrate. The unmethylated and hemimethylated 30-mer substrates are shown below. The target base is the internal cytosine in the recognition sequence (underlined); the 5-methycytosine in the hemimethylated substrate is denoted M.

Substrate oligonucleotides were synthesized by Research Genetics (Huntsville, AL) and high pressure liquid chromatography purified on a Dynamax PureDNA column (Rainin Instrument Co.) according to the manufacturer's specifications. Oligonucleotides were stored in 10 mM Tris, pH 8.0, 1 mM EDTA. Concentrations were determined using calculated extinction coefficients (23). For gel mobility shift assays, DNA substrates were radiolabeled using [-³²P]ATP (Amersham Pharmacia Biotech) and T4 polynucleotide kinase (New England Biolabs). Unincorporated label was removed with Bio-Gel P-6 spin columns (Bio-Rad).

Cofactor Purification-- S-Adenosylmethionine, S-adenosylhomocysteine, and sinefungin were purchased from Sigma. AdoMet was further purified as described previously (22, 24). S-[methyl-³H]adenosylmethionine was purchased from Amersham Pharmacia Biotech. All cofactor dilutions were in 0.1 N HCl.

DNA Equilibrium Dissociation Constants-- For K_D^DNA determinations in the absence of cofactor, binding assays containing 1 nM ³²P-labeled DNA and M.HhaI from 16 to 1000 nM in MR buffer (100 mM Tris, pH 8.0, 10 mM EDTA, 200 µg/ml bovine serum albumin, and 10 mM dithiothreitol) were incubated at 37 °C for 10 min. For K_D^DNA determinations in the presence of cofactor analogues, binding assays containing 1.0 pM ³²P-labeled DNA, 20 µM AdoHcy or sinefungin, and M.HhaI from 1.0 to 200 pM in MR buffer were incubated at 37 °C for 10 min. The samples were loaded onto prerun, 12% nondenaturing polyacrylamide gels. Gels were run at 300 V for 90 min at room temperature. Gels were dried, exposed to film or image plates, and analyzed on a STORM 840 densitometer (Molecular Dynamics, Inc.). Densitometry was performed using either ImageQuant (Molecular Dynamics, Inc.) or National Institutes of Health Image software. Under assay conditions where [DNA] [M.HhaI], dissociation constants were derived from data fit to rectangular hyperbolic equations using KaleidaGraph (Synergy Software). Under assay conditions where [DNA]  [M.HhaI], data were fit to the system of equations: K_D^DNA = E · S/ES, E₀ = E + ES, and S₀ = S + ES (25, 26) using the program Scientist (MicroMath Software, Inc.). (E and S are the free enzyme and DNA concentrations, respectively; ES is the concentration of E·DNA complex; and E₀ and S₀ are the total concentrations of enzyme and DNA, respectively). S₀ was used as a fitting parameter with this system of equations. The resultant fitting generated S₀ values commensurate with the expected S₀ values.

AdoMet Equilibrium Dissociation Constant-- A Perkin-Elmer LS50B luminescence spectrometer was used for fluorescence measurements. Excitation and emission slit widths were 5.0 mm. A xenon lamp was used at an excitation wavelength of 280 nm. Emission spectra were recorded from 320 to 400 nm from a 3.0-ml stirred quartz cuvette at 22 °C containing M.HhaI (1 µM), 100 mM Tris, pH 8.0, 10 mM EDTA, 10 mM dithiothreitol. Spectra were recorded as the AdoMet concentration was varied from 0.30 to 103 µM. To determine K_D^AdoMet, the fluorescence intensity at the initial _max, F₀, was subtracted from the intensity, F, after addition of AdoMet. F  F₀ was plotted versus AdoMet concentration and fit to a rectangular hyperbola using KaleidaGraph.

Methyltransferase Assays-- Filter binding assays monitored the incorporation of tritium labeled methyl groups into DNA. Reaction buffer, protein dilution buffer, and processing of filters were as described previously (22).

Steady State Assays-- Single time point assays were started with the addition of DNA to the remaining reaction components. Samples were processed after 10 min at 37 °C. Final DNA concentrations were from 1 to 12 nM, AdoMet concentrations ranged from 25 to 400 nM, and the M.HhaI concentration was 0.2 nM. Data were analyzed using the programs of Cleland (27) and Scientist (MicroMath Software).

Burst Assays-- Time course assays were performed at 37 °C with saturating substrate concentrations. The final AdoMet concentration of 500 nM was 3-fold greater than K_m^AdoMet, and the final DNA concentration of 250 nM was 50-fold greater then K_m^DNA (K_m^DNA and K_D^DNA correspond to K_mA and K_ia, respectively, of Ref. 31). M.HhaI concentrations were 25 and 50 nM. Reactions were started by the addition of DNA to the preincubated mixtures of M.HhaI and AdoMet.

AdoMet Inhibition-- One experimental strategy was designed to determine the effect of high AdoMet concentrations on the initial velocity. Single time point assays were performed at 37 °C with DNA and enzyme concentrations of 100 and 0.2 nM, respectively, while the AdoMet concentration was varied from 24 nM to 450 µM. Another strategy was designed to determine the effect of high AdoMet concentrations on the free enzyme, as revealed by AdoMet-dependent changes in burst magnitude. Time course assays were performed with enzyme and DNA concentrations of 25 and 250 nM, respectively, and AdoMet concentrations from 0.5 nM to 100 µM.

Single-turnover Assays-- Time course assays were performed at 37 °C with AdoMet (5 µM), M.HhaI, (200 nM), and DNA (200 nM). Reactions were started by the addition of AdoMet to the preincubated M.HhaI·DNA complex in a KinTek Corp. RQF-3 apparatus. The reactions were quenched with 0.5% SDS solution. Quenched samples were spotted onto DE-81 filters and processed (22).

Cofactor Exchange Assay-- To analyze cofactor isotope partitioning (28, 29) by M.HhaI, 100 nM substrate DNA and 400 nM unlabeled AdoMet were added to 400 nM M.HhaI preincubated at 37 °C with 400 nM [methyl-³H]AdoMet. In these time course assays, the final concentrations of enzyme, DNA, and cofactor were 20, 95, and 400 nM, respectively. Control reactions included mixing DNA (at the final concentrations specified above) and [methyl-³H]AdoMet with M.HhaI and [methyl-³H]AdoMet at both high [methyl-³H]AdoMet specific activity and a [methyl-³H]AdoMet specific activity identical to that of the final reaction mixture described above.

Substrate Exchange Assay-- To analyze the partitioning of substrate DNA from the M.HhaI·DNA complex, 0.3 nM 390-base pair poly(dI-dC:dI-dC) was incubated with 10 nM M.HhaI in MR buffer at 37 °C for 1.5 min. To start the reaction, a mixture containing 1 µM [methyl-³H]AdoMet plus 1.8 nM 1400-base pair poly(dI-dC:dI-dC) as a molecular competitor was added. Aliquots of the reaction mixture were removed over time and centrifuged through P-6 spin columns (Bio-Rad) to remove unincorporated AdoMet. The DNA was separated on a 6% polyacrylamide, 8 M urea gel run at 400 V for 1.5 h and visualized via fluorography (30) with Liquiscint (National Diagnostics). The gel was dried and exposed to Fuji XAR film for 1 week at 70 °C.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Steady State Parameters and Kinetic Mechanism for Single-site Substrates-- Intrinsic catalytic turnover constants (k_cat), Michaelis constants for AdoMet (K_m^AdoMet) and DNA (K_m^DNA), and specificity constants (k_cat/K_m^DNA) given in Tables I and III were obtained from double reciprocal analyses (Fig. 1). The rate equations for several kinetic mechanisms were fit to the steady state data for both the unmethylated and hemimethylated substrates. For the unmethylated substrate, the data are fit equally well by the equation describing a rapid equilibrium ordered mechanism and a steady state ordered mechanism (31). For the hemimethylated substrate, the data are best described by the equation for a steady state ordered mechanism. Both equations describe ordered, bisubstrate mechanisms in which DNA binds before productive AdoMet binding. The K_m^DNA term present in the equation for a steady state mechanism is a function of kinetic constants corresponding to the rates of substrate binding and product release. When the rate constant for the DNA dissociation step is small relative to the other rates of binding and release, the K_m^DNA term is reduced to zero, and the equation for the steady state mechanism reduces to the equation for a rapid equilibrium mechanism (31). Our results for single-site unmethylated substrate are generally consistent with the rapid equilibrium ordered kinetic mechanism proposed by Wu and Santi (8). However, the data shown in Fig. 1 and summarized in Table I support slightly different kinetic mechanisms for unmethylated versus hemimethylated substrates, as will be discussed.

View larger version (29K): [in this window] [in a new window]   Fig. 1.   Steady state kinetics of M. HhaI with unmethylated and hemimethylated 30-mer DNA substrate containing a single HhaI site. Ten-minute reactions were conducted at 37 °C and contained 0.2 nM M.HhaI, DNA, and AdoMet in MR buffer. A, 1/v versus 1/[DNA] at AdoMet concentrations of 25 (), 50 (), 100 (), 200 (), and 400 nM (). B, 1/v versus 1/[AdoMet] at DNA concentrations of 1 (), 2 (), 4 (), 8 (), and 12 nM (). C, 1/v versus 1/[hmDNA] at AdoMet concentrations of 50 (), 100 (), 200 (), 400 (), and 800 nM (). D, 1/v versus 1/[AdoMet] at hmDNA concentrations of 4 (), 8 (), 16 (), 32 (), and 64 nM (). The rate equation for the steady state ordered mechanism is fit to the data. Because lines were fit to the entire data set for each substrate, the fit of an individual line is less important than the fit of the set of lines to the entire data set.

Competency Analysis of the Enzyme·AdoMet Complex-- The kinetic mechanism proposed by Wu and Santi (8) and suggested by our results predicts either that the binary enzyme·AdoMet complex is not formed or that, if formed, it is not catalytically competent. These alternatives can be further probed using isotope-partitioning analysis (28, 29). Preincubation of M.HhaI with radiolabeled AdoMet followed by addition of DNA and radiolabeled AdoMet results in a burst of methylated product formation, followed by a slower rate of product formation. Preincubation of M.HhaI with radiolabeled AdoMet followed by addition of DNA and unlabeled AdoMet reduces the burst magnitude and steady state rate of product formation 20-fold, commensurate with the dilution of specific activity of the cofactor (Fig. 2A). If the enzyme·AdoMet complex does not form at all, then this apparent reduction of burst magnitude and steady state rate of product formation is expected due to the unlabeled AdoMet added to the reaction mixture with the substrate DNA. If the enzyme·AdoMet complex does form but in a catalytically unproductive manner, then this result is expected because the preformed complex must dissociate in order to bind DNA and then AdoMet in a catalytically productive manner. Under these circumstances, the unlabeled AdoMet added to the enzyme with the DNA causes the apparent reduction of burst magnitude and steady state rate of product formation. Although consistent with both predictions described above, the results shown in Fig. 2A argue against any mechanism that invokes a functional enzyme·AdoMet complex.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   A, isotope partitioning analysis of M. HhaI. Methyltransferase assays contained 20 nM M.HhaI, 95 nM DNA, and 400 nM AdoMet. , product formation after enzyme was preincubated at 37 °C with high specific activity [methyl-3H]AdoMet and reaction started with DNA and high specific activity [methyl-3H]AdoMet. , product formation after enzyme was preincubated with high specific activity [methyl-3H]AdoMet and reaction started with DNA and unlabeled AdoMet. , control experiment showing product formation after enzyme was preincubated with low specific activity [methyl-3H]AdoMet and reaction started with DNA and low specific activity [methyl-3H]AdoMet. B, molecular partitioning analysis of M.HhaI. Product formation over time as detected by fluorography after the reaction was started by the simultaneous addition of [methyl-3H]AdoMet, 1.8 nM 1400-base pair DNA, and 0.3 nM 390-base pair DNA (Combined, left), and after 0.3 nM 390-base pair DNA was preincubated with the enzyme prior to the addition of 1.8 nM 1400-base pair DNA and [methyl-3H]AdoMet (Partitioned, right). C, steady state burst analysis of M.HhaI. Product formation over time at 37 °C for methyltransferase assays containing 25 nM enzyme () and 50 nM enzyme (). DNA and AdoMet concentrations were 250 and 500 nM, respectively. The non-zero y-intercept indicates that a kinetic step after the chemical (methyl group transfer) step governs the overall rate of catalysis.

Formation of the Enzyme·AdoMet Complex as Detected by Native Protein Fluorescence-- Although the isotope partitioning results argue against a competent enzyme·AdoMet complex, they leave open the question of whether such a complex forms at all. For M.HhaI, the single tryptophan residue (Trp⁴¹), located in the AdoMet binding pocket (9), facilitates detection of the enzyme·AdoMet complex via steady state native fluorescence experiments. In agreement with the crystal structures, the native fluorescence upon excitation at 280 nm has an emission maximum of 358 nm, indicative of a solvent-exposed tryptophan residue (32). Upon addition of AdoMet, the tryptophan fluorescence intensity decreases. Furthermore, an emission maximum blue shift, evident with increasing AdoMet concentration, suggests that cofactor binding may shield the tryptophan from the aqueous solvent (32). The dissociation constant for AdoMet was determined to be 11.5 µM. Fig. 4B shows the relative fluorescence as a function of AdoMet concentration. The K_D^AdoMet determined from these data is much greater than the K_m^AdoMet determined in the steady state assays, indicating that AdoMet is more weakly bound in the binary complex than in the presence of DNA. The isotope partitioning results leave open the possibility that the enzyme·AdoMet complex does not form, but the native fluorescence result confirms the formation of the complex. Together, the results indicate the formation of a relatively weakly bound, catalytically inactive enzyme·AdoMet complex.

Detection of a Dead-end Binary Enzyme·AdoMet Complex-- The formation of a dead-end binary enzyme·AdoMet complex predicts lower levels of free, functional enzyme as AdoMet concentrations are increased. In a classical burst analysis or active site titration (i.e. Fig. 2C), the burst magnitude corresponds to the functional enzyme concentration (33). After preincubation of high concentrations of enzyme with varying AdoMet concentrations, excess DNA was used to start the methyltransferase reaction and obtain a kinetic burst analysis. AdoMet concentrations were chosen based on the K_D^AdoMet determined from the fluorescence titration results. The predicted decrease in burst magnitude with increasing AdoMet concentrations was not observed. Also unobserved was the similarly predicted substrate inhibition by AdoMet (Ref. 8 and data not shown). The most plausible explanation for these observations is that the enzyme·AdoMet complex associates and dissociates rapidly relative to the dead time of the experiment (several seconds, in this case). If so, the AdoMet would not alter the effective concentration of free enzyme available for catalysis, resulting in identical burst magnitudes.

Competency Analysis of the Enzyme·DNA Complex-- The kinetic mechanism proposed previously (8) and suggested by our results also predicts that the enzyme·DNA complex should be catalytically competent. The results of a "molecular" partitioning experiment (34), in which DNA products are distinguished by their lengths, is shown in Fig. 2B. The left side of the fluorogram shows the products of the methyltransferase assay after 1.8 nM 1400-base pair and 0.3 nM 390-base pair DNA were added simultaneously to start the reaction. The 6-fold excess of 1400-base pair substrate over the 390-base pair substrate resulted in the predominance of the 1400-base pair product. The right side of the fluorogram (Fig. 2B) shows the results of a similar experiment in which the 0.3 nM 390-base pair substrate was preincubated with the M.HhaI. To start the reaction, 1.8 nM 1400-base pair substrate and [methyl-³H]AdoMet were added. The presence of the relatively large amount of 390-base pair product indicates that the initially formed enzyme·DNA complex is catalytically active. Thus, the relative rate constants for the forward reaction versus the release of the DNA (i.e. the partitioning from this binary complex) largely favor the forward process.

Identification of the Rate-limiting Catalytic Step-- Correlating functional analysis with the available M.HhaI structures requires a quantitative understanding of the relative contributions of various steps in the catalytic cycle toward k_cat. Time course assays under steady state conditions reveal an initial burst and subsequent linear increase of product formation (Fig. 2C). The simplest interpretation of these data is that catalysis (k_cat) is rate-limited by a step in the catalytic cycle after the chemical step (22) and that methylation and other prior steps occur more rapidly.

Measurement of the Rate of Methyl-group Transfer-- Product formation over time under single-turnover conditions was fit to a single exponential for unmethylated and hemimethylated DNA (Fig. 3). The k_methylation values obtained from the single-exponential fits are roughly 2- and 3-fold greater than k_cat for unmethylated and hemimethylated DNA, respectively (Table I). This confirms that the chemical step does not determine the overall rate of catalysis. For both substrates, the amount of product detected is slightly more than the theoretically expected amount for a single methylation event relative to the amount of unmethylated substrate present in the reaction mixture. This could most simply be accounted for by an actual concentration of substrate in the reaction mixture higher than calculated. The ratio of product to unmethylated substrate is twice that of the ratio of product to hemimethylated substrate. This result is consistent with previous pre-steady state kinetics results for the M.EcoRI and suggests that methylation of each strand occurs from a unique binding orientation (29), with half of the hemimethylated binding events occurring in the unproductive orientation.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Single-turnover kinetic analysis of M. HhaI. Product formation over time at 37 °C for unmethylated DNA substrate () and hemimethylated DNA substrate (). Substrate concentrations were 200 nM. Enzyme and AdoMet concentrations were 5 µM. Reactions were quenched with 0.5% SDS.

Measurement of Equilibrium Dissociation Constants-- The available crystal structures of the hemi- and unmethylated DNA and M.HhaI (19, 20) can form the basis of understanding substrate discrimination if the functional analysis includes the corresponding equilibrium binding parameters. Dissociation constants were determined by gel mobility shift analysis. Fig. 4A shows a typical binding isotherm and autoradiogram of the gel mobility polyacrylamide gel electrophoresis. Results are summarized in Table II. Our experimental approach differs from previous M.HhaI gel mobility shift assays (35, 36) in that we varied enzyme concentration while maintaining a constant DNA concentration. In the absence of cofactor, M.HhaI binds hemimethylated DNA 7-fold more tightly than unmethylated DNA. In the presence of AdoHcy, M.HhaI binds unmethylated DNA 500-fold more tightly and hemimethylated DNA 900-fold more tightly. The discrimination in favor of hemimethylated DNA is 12-fold.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Equilibrium binding of M. HhaI·DNA complex. The graph shows bound DNA as a function of M.HhaI concentration. The gel from which the graphed data were derived is shown (inset). Binding assay contained 1 nM 32P-labeled unmethylated DNA and M.HhaI from 16.8 to 1000 nM in MR buffer. Mixtures were preincubated and electrophoresed as described under "Experimental Procedures." Other binding constants were determined similarly and results are summarized in Table II. B, fluorescence of M.HhaI as a function of [AdoMet]. M.HhaI contains a single tryptophan residue (Trp41) located in the AdoMet binding site. Enzyme (1 µM) was titrated with AdoMet from 0.3 to 103 µM and excited at 280 nm, and F  F0 at the max of 358 nm was plotted against [AdoMet] and fit to a rectangular hyperbola.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

A detailed understanding of how enzymes distort DNA conformation and the effect it has on catalysis and specificity demands detailed structural and functional characterizations. DNA bending by enzymes and other proteins is well known. Stabilization of extrahelical nucleosides has been observed via x-ray crystallography in M.HhaI, the HaeIII DNA methyltransferase (4), uracil-DNA glycosylase (17), and T4 endonuclease V (18). The nucleoside flipping process has been observed via 2-aminopurine fluorescence in M.EcoRI (6, 37), M.TaqI, and M.HhaI (16), and uracil-DNA glycosylase (38). Crystal structures are available of M.HhaI bound to AdoMet (9, 39); covalently bound to DNA (5); noncovalently bound to unmethylated, fully methylated (19), and hemimethylated DNA (20); and both nucleoside analog-containing (40, 41) and base pair mismatch-containing DNA (42). M.HhaI represents a rich structural paradigm for DNA-modifying enzymes, yet many mechanistic issues remain unresolved. For example, the contributions of individual conformational changes such as nucleoside flipping, protein active site loop movement, and cofactor binding to the overall catalytic cycle are presently unknown. Similarly, the fate and lifetime of the covalent cysteinyl-cytosine and other potential intermediates are also unknown. Knowledge of these mechanistic details is required to understand such biologically relevant issues the discrimination between hemimethylated and unmethylated DNA sequences; cytosine deamination (43); and methyltransferase regulation, inhibition, and processivity. The wealth of M.HhaI structural information significantly motivated our detailed functional analysis and pursuit of a comprehensive understanding of DNA methyltransferases in terms of structure and function.

The steady state constants shown in Table I are similar to those reported for the large, multisite DNA (8). However, for our 30-base pair single-site substrate, the enzyme shows significantly weaker binding of AdoMet, as indicated by K_m^AdoMet. On a multisite substrate such as poly(dG-dC), a processive² enzyme is not required to dissociate after each methylation event. The enzyme is thus "bound" to the multisite DNA ready to bind AdoMet to a greater extent than the single-site substrate, from which the enzyme must dissociate between catalytic turnovers. The higher K_m^AdoMet (11-19-fold) observed with our single-site substrate may derive from this difference because the various DNA-bound enzyme forms bind AdoMet more tightly than the free enzyme. The catalytic turnover numbers are similar to those reported for other DNA methyltransferases (22, 44-47). This suggests that k_cat is limited by the same process for all DNA methyltransferases. Furthermore, for M.HhaI, the similar k_cat for the short and long substrates (Table I) suggests that catalysis is limited by the same process for each.

The burst experiment in Fig. 2C demonstrates that the turnover rate constant, k_cat, is partially determined by methylation as well as one or more slower product release steps that follow methylation. The methylation rate constants, as determined by rapid quench experiments under single-turnover conditions, are only 2- or 3-fold greater than k_cat for unmethylated and hemimethylated DNA, respectively (Fig. 3). M.HhaI is similar to the mammalian cytosine-C-5 DNA methyltransferase, Dnmt1, in this respect (48). This relationship between k_cat and k_methylation is in contrast to that of the adenine-specific M.EcoRI, in which k_methylation is limited by nucleoside flipping and is at least 1600-fold greater than k_cat (37). This large difference between the cytosine and adenine DNA methyltransferases may derive from either their distinct chemical mechanisms or differences in the flipping of the target nucleosides. Although C-5-cytosine methylation is a priori more difficult to catalyze, DNA cytosine-C-5 methyltransferases employ a covalent intermediate to activate the poorly nucleophilic C-5 center. There is no indication that such nucleophilic catalysis occurs in the chemical mechanism of adenine DNA methyltransferases. Alternatively, if k_methylation is limited by the rate of nucleoside flipping for M.HhaI as it is for M.EcoRI, then the much slower rates may derive from inherently greater energetic cost and slower kinetics for disrupting the GC base pair (49). Cytosine-N⁴-specific DNA methyltransferases provide a compelling test of these alternatives because these enzymes catalyze exocyclic amine methylation, as do the adenine-specific DNA methyltransferases, but presumably disrupt a GC base pair, as do the DNA cytosine-C-5 methyltransferases. Our unpublished results with M.BamHI support the latter alternative.³

The k_cat/K_m ratios shown in Table III indicate that the steady state discrimination of hemimethylated and unmethylated DNA is less than 2-fold in favor of unmethylated DNA. This relatively small effect is similar to other bacterial DNA methyltransferases and contrasts with Dnmt1, which shows a 10-20-fold preference for hemimethylated DNA (48). For M.HhaI, k_cat is greater for hemimethylated DNA, and k_methylation is greater for unmethylated DNA. This lack of correlation of k_cat and k_methylation is consistent with our observation that k_cat is dominated by steps after catalysis and not by k_methylation (Fig. 2C). Furthermore, because k_methylation is not rate-limiting, the slight preference for unmethylated DNA observed at the level of the specificity constants (k_cat/K_m) cannot be due to the different contributions of k_methylation but must be due to other terms in k_cat/K_m.

Although the 2-fold steady state discrimination for unmethylated DNA is small, we were compelled to examine its molecular underpinnings because of the wealth of structural information available for M.HhaI. Enzymes frequently exploit binding interactions to affect specificity. In addition to complementing the structural analysis, thermodynamic constants are not complicated by kinetic terms beyond those involved in complex formation. The E·AdoMet, the unmethylated ternary, and hemimethylated ternary complexes for which thermodynamic constants have been determined (Table II) correspond directly to available crystal structures. The structures of the E·DNA·AdoHcy complexes show DNA bound in a central cleft dividing target-recognition and catalytic domains of the two-domain protein. Two "recognition" loops from the target recognition domain interact with the major groove of the DNA. Across the helix, the target nucleoside is extracted from the DNA double helix, placing its base into the active site pocket. The active site loop, containing the active site nucleophile (Cys⁸¹), is seen in alternate conformations in the ternary and binary complexes. AdoHcy is bound in the ternary complexes, whereas in the binary complex, AdoMet is bound in a different orientation in the absence of DNA (5, 9, 39).

Dissociation constants for the binary M.HhaI·DNA complexes derived from gel mobility shift analyses (Fig. 4A) indicate an approximately 7-fold binding preference for hemimethylated DNA (Table II). AdoHcy, one of the two reaction products, increases the stability of the M.HhaI·^hmDNA complex 900-fold (Table II). The corresponding M.HhaI·DNA complex is stabilized 500-fold by AdoHcy, resulting in a 12-fold preference for the hemimethylated DNA substrate. (Similar results were obtained with the AdoMet analog sinefungin.) Binary crystal structures of the E·DNA complexes, which in comparison to the ternary structures might provide a structural explanation for the stabilization by AdoHcy, are unavailable. However, ¹⁹F NMR and gel mobility studies indicate that upon binding, AdoHcy causes a detectable change in the orientation of the extrahelical cytosine and in the conformation of the protein·DNA complex (50). Our binding results support these prior findings and reveal by inspection of the dissociation constants that AdoHcy increases the binding energy by approximately 4 kcal/mol.

The crystal structures of M.HhaI·AdoHcy·DNA and M.HhaI·AdoHcy·^hmDNA complexes show no gross structural differences to explain how M.HhaI preferentially binds ^hmDNA. The two structures have an -carbon root-mean-square deviation of approximately 0.6 Å. The observed van der Waals contact between the carboxylate of Glu²³⁹ and the methyl group of the hemimethylated DNA revealed in the M.HhaI·^hmDNA·AdoHcy complex may partially explain the binding preference (20). Additionally, the absence of the C-5-methyl group in the M.HhaI·AdoMet·DNA complex may be considered a "cavity-creating" perturbation with respect to the M.HhaI·^hmDNA·AdoHcy structure with an associated cost in binding energy that contributes to the observed differences in K_D^DNA. Proteins pay an energetic cost for the presence of cavities in the core (51). Although partially solvent-exposed, the volume occupied by the C-5-methyl group in the M.HhaI·^hmDNA·AdoHcy structure is not compensated for by protein conformational changes in the M.HhaI·DNA·AdoHcy crystal structure (Fig. 5). Of the residues that define this cavity, an ordered water that is hydrogen bonded to O of Glu²³⁹ is conspicuous because of its presence in both structures, its proximity to the cavity, and lack of compensatory movement.

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Structural contributions to binding discrimination. Each M.HhaI recognition site contains two cytosines that can undergo methylation (5'-GCGC-3'/5'GCGC3'). The nontarget cytosine (green) in the unmethylated M.HhaI·DNA·AdoHcy crystal structure is shown (left) with the proximal structural elements (19). The equivalent image of the M.HhaI·hmDNA·AdoHcy structure is also shown (right) (20). The black dots represent the van der Waal surfaces of the cytosine C-5 (left) and 5-methyl moiety (right). The Gln239 and Glu237 side chains are colored pink. The guanosine to which each cytosine is base paired is shown in red. The ordered water molecule found in both structures is shown in blue. The cavity surface (gray) was generated with the program SURFNET (56). Glu239 was previously implicated in binding discrimination (20). The larger cavity volume seen in the unmethylated DNA complex (left) results from the absence of the methyl moiety and the lack of compensatory repacking of the proximal structural elements.

Inspection of the M.HhaI·AdoMet and M.HhaI·DNA·AdoHcy crystal structures (Fig. 6) indicates that the region normally bound by the cofactor is quite accessible. Thus, there is no a priori reason why the cofactor is precluded from binding prior to DNA. This would result in a random kinetic mechanism, which is not observed with our substrates or with poly(dG-dC). However, M.HhaI could formally have a random mechanism with a large preference for the initial addition of DNA (31). The most compelling evidence against a random mechanism is our isotope partitioning experiment, which clearly argues against the catalytic competence of the M.HhaI·AdoMet complex (Fig. 2A). The binary M.HhaI·AdoMet structure (Ref. 9 and Fig. 6) and our fluorescence results (Fig. 4B) show that M.HhaI does bind AdoMet in the absence of DNA. However, comparison of K_D^AdoMet and K_m^AdoMet (Tables I and II) indicate that the binding affinity is much weaker than in the presence of DNA. Furthermore, the binding orientations within the binary and ternary complexes are fundamentally distinct. AdoMet has been observed in the productive orientation in binary structures of M.HhaI and M.TaqI, although both were crystallized in the presence of DNA (11, 39). Thus, DNA is required for AdoMet binding in the productive orientation. (Not all DNA methyltransferases require bound DNA for productive AdoMet binding, however: M.EcoRI proceeds by way of a mechanism in which productive AdoMet binding can precede DNA binding (22)). The combined functional data show that in the absence of DNA, M.HhaI binds AdoMet nonproductively and that bound DNA is required for productive AdoMet binding.

View larger version (48K): [in this window] [in a new window]   Fig. 6.   Cofactor binding orientation. The binary M.HhaI·AdoMet complex (9) (left) and the ternary M.HhaI·DNA·AdoHcy complex (19) (right) are shown. Amino acid residues Trp41 and Phe18 (red) are visible through the calculated molecular surface of the enzyme. The extrahelical cytosine of the ternary complex is shown (blue). The cofactors (yellow) are seen in different orientations relative to the enzyme. In the putative nonproductive binding orientation of the binary complex, the adenosyl moiety is inserted into the binding pocket. In the ternary complex, however, it is the methionine moiety in the binding pocket. Among a variety of protein contacts, the purine ring stacks with the indole ring of Trp41 and is perpendicular to the plane of the phenyl group of Phe18. The image was generated with GRASP (57).

Our results with hemi- and unmethylated substrates provide insights into the relationship between the affinity of the enzyme for its substrates, enzyme·substrate intermediate lifetimes, and the observed kinetic mechanisms. The double reciprocal plots for hemi- and unmethylated substrates (Fig. 1) are clearly distinct. Although both data sets are well fit with the equation for a steady state mechanism, the data for the unmethylated substrate can also be fit with the equation for a rapid equilibrium mechanism (31). The difference between these mechanisms in the context of M.HhaI centers on the processing of various enzyme·substrate intermediates, including enzyme·DNA, enzyme·DNA^flipped, enzyme'·DNA^flipped (i.e. after active site loop movement), enzyme'·DNA^flipped·AdoMet, etc. A rapid equilibrium ordered Bi Bi mechanism involves fast interconversion steps between enzyme forms prior to a slower forward step. In contrast, the "steady state mechanism" (i.e. ordered Bi Bi) requires that enzyme·substrate intermediates partition largely in the forward direction, and enzyme-product complexes may account for a significant portion of the total enzyme concentration. Inspection of the corresponding rate equations⁴ shows that the relationship between K_D^DNA and K_m^DNA correlates with the degree to which a steady state mechanism approaches a rapid equilibrium mechanism. When K_D^DNA is large relative to K_m^DNA, the pattern of double reciprocal plots will be consistent with a rapid equilibrium mechanism. Conversely, when K_D^DNA is small relative to K_m^DNA, the resulting double reciprocal patterns intersect further to the left of the vertical axis and are more indicative of a steady state mechanism. For M.HhaI and unmethylated DNA, the relative values of K_D^DNA and K_m^DNA (100 versus 4 nM), like the double reciprocal plots, indicate a mechanism that boarders between rapid equilibrium and steady state, partially satisfying rapid equilibrium assumptions. For hemimethylated DNA, the relative values of K_D^DNA and K_m^DNA (15 versus 7 nM) and double reciprocal plots are consistent with a mechanism having more steady state character.

The results of our analysis of M.HhaI and the effect of the methylation status of substrates can be seen as part of a larger mechanistic continuum that includes the mammalian DNA cytosine-C-5 methyltransferase Dnmt1. This enzyme modifies cytosine within CpG dinucleotides and is essential for mammalian viability (52). For Dnmt1, K_D^DNA is 10-fold lower than K_m^DNA. The double reciprocal plots intersect to the left of the vertical axis so far as to appear nearly parallel (34). Thus, Dnmt1 partitions forward from the individual enzyme·substrate intermediates even more than M.HhaI does with hemimethylated substrates. This must result in part from slower reverse steps because Dnmt1 is overall a slower enzyme than M.HhaI.

Rapid equilibrium and steady state mechanisms are distinguished by the partitioning forward, toward catalysis or back, toward dissociation of each enzyme·substrate intermediate. For M.HhaI, hemimethylated substrate demonstrates a greater forward partitioning, "commitment to catalysis" (53), or "stickiness" (54) than the unmethylated substrate. This substrate-induced modulation of kinetic mechanism derives from a single methyl group on the cytosine. The insight we have gained into how this small decoration of DNA alters the manner in which M.HhaI processes its substrate can be applied to DNA methyltransferases and to other enzymes and proteins that bind DNA. Examples include DNA replication and transcriptional regulation in bacteria, gene silencing in eukaryotes (including genomic imprinting and X-inactivation (55)), restriction/modification systems, and DNA mismatch repair.

	ACKNOWLEDGEMENTS

We thank Prof. John Perona and Dr. Nancy Horton for critical review of the manuscript.

	FOOTNOTES

^* This work was supported by National Institutes of Health Grant GM 463333 and National Science Foundation Grant MCB-9603567 (to N. O. R.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^¶ Present address: EpiGenX Pharmaceuticals, 2124 Bath St., Santa Barbara, CA 93105.

To whom correspondence should be addressed. Tel.: 805-893-8368; Fax: 805-893-4120; E-mail: reich@chem.ucsb.edu.

² Unpublished observations.

³ W. M. Lindstrom, Jr., E. G. Malygin and N. O. Reich, manuscript in preparation.

⁴ We refer specifically to equations VI-59 and IX-89 of Ref. 31 for the rapid equilibrium and steady state mechanisms, respectively.

	ABBREVIATIONS

The abbreviations used are: AdoMet, S-adenosyl-L-methionine; AdoHcy, S-adenosyl-L-homocysteine; M.HhaI, HhaI DNA methyltransferase; ^hmDNA, hemimethylated DNA.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				B. Youngblood, E. Bonnist, D. T.F. Dryden, A. C. Jones, and N. O. Reich Differential stabilization of reaction intermediates: specificity checkpoints for M.EcoRI revealed by transient fluorescence and fluorescence lifetime studies Nucleic Acids Res., May 1, 2008; 36(9): 2917 - 2925. [Abstract] [Full Text] [PDF]

				A. A. Evdokimov, B. Sclavi, V. V. Zinoviev, E. G. Malygin, S. Hattman, and M. Buckle Study of Bacteriophage T4-encoded Dam DNA (Adenine-N6)-methyltransferase Binding with Substrates by Rapid Laser UV Cross-linking J. Biol. Chem., September 7, 2007; 282(36): 26067 - 26076. [Abstract] [Full Text] [PDF]

				R. A. Estabrook and N. Reich Observing an Induced-fit Mechanism during Sequence-specific DNA Methylation J. Biol. Chem., December 1, 2006; 281(48): 37205 - 37214. [Abstract] [Full Text] [PDF]

				E. Merkiene and S. Klimasauskas Probing a rate-limiting step by mutational perturbation of AdoMet binding in the HhaI methyltransferase Nucleic Acids Res., January 13, 2005; 33(1): 307 - 315. [Abstract] [Full Text] [PDF]

				N. Mashhoon, M. Carroll, C. Pruss, J. Eberhard, S. Ishikawa, R. A. Estabrook, and N. Reich Functional Characterization of Escherichia coli DNA Adenine Methyltransferase, a Novel Target for Antibiotics J. Biol. Chem., December 10, 2004; 279(50): 52075 - 52081. [Abstract] [Full Text] [PDF]

				R. A. Estabrook, R. Lipson, B. Hopkins, and N. Reich The Coupling of Tight DNA Binding and Base Flipping: IDENTIFICATION OF A CONSERVED STRUCTURAL MOTIF IN BASE FLIPPING ENZYMES J. Biol. Chem., July 23, 2004; 279(30): 31419 - 31428. [Abstract] [Full Text] [PDF]

				H. M. Cohen, D. S. Tawfik, and A. D. Griffiths Altering the sequence specificity of HaeIII methyltransferase by directed evolution using in vitro compartmentalization Protein Eng. Des. Sel., January 1, 2004; 17(1): 3 - 11. [Abstract] [Full Text] [PDF]

				E. G. Malygin, W. M. Lindstrom Jr., V. V. Zinoviev, A. A. Evdokimov, S. L. Schlagman, N. O. Reich, and S. Hattman Bacteriophage T4Dam (DNA-(Adenine-N6)-methyltransferase): EVIDENCE FOR TWO DISTINCT STAGES OF METHYLATION UNDER SINGLE TURNOVER CONDITIONS J. Biol. Chem., October 24, 2003; 278(43): 41749 - 41755. [Abstract] [Full Text] [PDF]

				C. B. Thomas, R. D. Scavetta, R. I. Gumport, and M. E. A. Churchill Structures of Liganded and Unliganded RsrI N6-Adenine DNA Methyltransferase: A DISTINCT ORIENTATION FOR ACTIVE COFACTOR BINDING J. Biol. Chem., July 3, 2003; 278(28): 26094 - 26101. [Abstract] [Full Text] [PDF]

				E. G. Malygin, V. V. Zinoviev, A. A. Evdokimov, W. M. Lindstrom Jr., Norbert. O. Reich, and S. Hattman DNA (Cytosine-N4-)- and -(Adenine-N6-)-methyltransferases Have Different Kinetic Mechanisms but the Same Reaction Route. A COMPARISON OF M.BamHI AND T4 Dam J. Biol. Chem., April 25, 2003; 278(18): 15713 - 15719. [Abstract] [Full Text] [PDF]

				S. Bheemanaik, S. Chandrashekaran, V. Nagaraja, and D. N. Rao Kinetic and Catalytic Properties of Dimeric KpnI DNA Methyltransferase J. Biol. Chem., February 28, 2003; 278(10): 7863 - 7874. [Abstract] [Full Text] [PDF]

				N. Huang, N. K. Banavali, and A. D. MacKerell Jr. Protein-facilitated base flipping in DNA by cytosine-5-methyltransferase PNAS, January 7, 2003; 100(1): 68 - 73. [Abstract] [Full Text] [PDF]

				U. T. Sankpal and D. N. Rao Mutational analysis of conserved residues in HhaI DNA methyltransferase Nucleic Acids Res., June 15, 2002; 30(12): 2628 - 2638. [Abstract] [Full Text] [PDF]

				T. Yokochi and K. D. Robertson Preferential Methylation of Unmethylated DNA by Mammalian de Novo DNA Methyltransferase Dnmt3a J. Biol. Chem., March 29, 2002; 277(14): 11735 - 11745. [Abstract] [Full Text] [PDF]

				C. P. Swaminathan, U. T. Sankpal, D. N. Rao, and A. Surolia Water-assisted Dual Mode Cofactor Recognition by HhaI DNA Methyltransferase J. Biol. Chem., February 1, 2002; 277(6): 4042 - 4049. [Abstract] [Full Text] [PDF]

				X. Cheng and R. J. Roberts AdoMet-dependent methylation, DNA methyltransferases and base flipping Nucleic Acids Res., September 15, 2001; 29(18): 3784 - 3795. [Abstract] [Full Text] [PDF]

				E. G. Malygin, A. A. Evdokimov, V. V. Zinoviev, L. G. Ovechkina, W. M. Lindstrom, N. O. Reich, S. L. Schlagman, and S. Hattman A dual role for substrate S-adenosyl-L-methionine in the methylation reaction with bacteriophage T4 Dam DNA-[N6-adenine]-methyltransferase Nucleic Acids Res., June 1, 2001; 29(11): 2361 - 2369. [Abstract] [Full Text] [PDF]

				E. G. Malygin, W. M. Lindstrom Jr, S. L. Schlagman, S. Hattman, and N. O. Reich Pre-steady state kinetics of bacteriophage T4 Dam DNA-[N6-adenine] methyltransferase: interaction with native (GATC) or modified sites Nucleic Acids Res., November 1, 2000; 28(21): 4207 - 4211. [Abstract] [Full Text] [PDF]

				S. S. Szegedi, N. O. Reich, and R. I. Gumport Substrate binding in vitro and kinetics of RsrI [N6-adenine] DNA methyltransferase Nucleic Acids Res., October 15, 2000; 28(20): 3962 - 3971. [Abstract] [Full Text] [PDF]

				G. Vilkaitis, E. Merkiene, S. Serva, E. Weinhold, and S. Klimasauskas The Mechanism of DNA Cytosine-5 Methylation. KINETIC AND MUTATIONAL DISSECTION OF HhaI METHYLTRANSFERASE J. Biol. Chem., June 8, 2001; 276(24): 20924 - 20934. [Abstract] [Full Text] [PDF]

				A. A. Evdokimov, V. V. Zinoviev, E. G. Malygin, S. L. Schlagman, and S. Hattman Bacteriophage T4 Dam DNA-[N6-adenine]Methyltransferase. KINETIC EVIDENCE FOR A CATALYTICALLY ESSENTIAL CONFORMATIONAL CHANGE IN THE TERNARY COMPLEX J. Biol. Chem., January 4, 2002; 277(1): 279 - 286. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Submit a Letter to Editor Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lindstrom, W. M. Articles by Reich, N. O. Search for Related Content PubMed PubMed Citation Articles by Lindstrom, W. M., Jr. Articles by Reich, N. O. Social Bookmarking                      What's this?