include "../../../../includes/header.php"; ?>
John A. Wendel and Steven S. Smith
City of Hope
1500 E. Duarte Rd.
Duarte, CA 91010-3000
This is a draft paper
for a talk at the
Fifth Foresight Conference on Molecular Nanotechnology.
The final version has been submitted
for publication in the special Conference issue of Nanotechnology.
This page uses the HTML <sup> and <sub> conventions for superscripts and subscripts. If "103" looks the same as "103" then your browser does not support superscripts. If "xi" looks the same as "xi" then your browser does not support subscripts. Failure to support superscripts or subscripts can lead to confusion in the following text, particularly in interpreting exponents.
Bacterial DNA methyltransferases offer an approach to addressable protein targeting in macromolecular assembly that permits the construction of ordered arrays of functional proteins or peptides. This approach uses the natural recognition specificity of the bacterial DNA cytosine methyltransferase to target fusion proteins to preselected sites on a DNA scaffold through the formation of a stable covalent attachment to a mechanism-based methyltransferase inhibitor. Addressable targets on DNA molecules can readily be created with 5FdC* at the enzyme recognition site. This is because the substitution of the targeted pyrimidine ring with fluorine at C5 generates a kinetic barrier to the methyltransfer step in the reaction and a thermodynamic barrier to the beta-elimination step that stall the enzyme after it forms a covalent bond with DNA. In this report we have studied the capacity of dU to substitute for 5FdC in addressable targeting. When dU is used in place of 5FdC as the base targeted by the enzyme for nucleophilic attack, it forms a mispair with dG. Like 5FdC, dU creates a kinetic barrier to completion of the reaction causing the enzyme to stall. Unlike 5FdC, dU does not create a thermodynamic barrier to beta-elimination and release after methyltransfer, however, the mispaired state of the dU in the dG:dU basepair appears to create a transition-state analog that cannot be easily released by the enzyme. The data show that dU can be used to substitute for 5FdC in the production of ordered nucleoprotein-based macromolecular assemblies.
|* Abbreviations Used: FdC, 5-fluorodeoxycytidine; dU, deoxyuridine; dG, deoxyguanosine; dC, deoxycytidine; mdC, 5-methyldeoxycytidine; mdU, 5-methyldeoxyuracil = dT, deoxythymidine.|
One of the problems in producing nanoscale assemblies and devices has been developing technologies for the control of position ( Eigler, et al., 1990, Merkle, 1993 ). Living systems routinely produce nanoscale biostructures that take advantage of precise spatial arrangements.The assembly of higher order biostructures in living systems generally depends on the exquisitely selective interactions between proteins, between nucleic acids, and between proteins and nucleic acids that have evolved over millenia. Chromatin, the complex aggregate of proteins and nucleic acids that forms the fundamenal material in chromosomes, is composed of precisely ordered functional proteins and DNA. In general, protein-DNA interactions in chromatin are reversible, so they can disassemble and reassemble during the cyclic processes that are required in cellular reproduction. However, some of the spatial arrangements between proteins on DNA appear to be set down as patterned biostructures that are copied precisely from one cell to the next and can remain intact for many years.
Considerable research in this area of molecular biology has focused on how such patterns are established and maintained since this is not only the central question of developmental biology, it is a key question in biological research on aging and cancer. One aspect of these studies has been research on the function of DNA methyltransferases and their biological role in this process. This research suggests that these molecules have been designed by evolution and natural selection to stabilize biostructures in chromatin by physically participating in the assembly of these structures (Smith, 1998). Whether or not this interpretation of the function of methyltransferases in biology turns out to be correct remains to been seen. However, the capacity of these enzymes for positionally controlled self-assembly can easily be utilized in the production of stable ordered arrays of functional proteins on DNA in vitro (Smith, et al., 1995, Smith, et al., 1997 and Smith, et al., 1997a).
Control of positioning for molecular components is achieved through the addressing capacity of these enzymes that is rooted in their sequence specificity and their mechanism of action. Of the more than 40 different bacterial methyltransferases and the several eukaryotic methyltransferases that have been cloned and carefully studied, all have well characterized DNA sequence specificities (McClelland, M. & Nelson, M., 1988). A subset of these, the cytosine-5 methyltransferases, form abortive covalent complexes between an active-site cysteine and 5-fluorocytosine in their DNA recognition sequences (Osterman, et al., 1988, Chen, et al., 1991, Smith, et al., 1992). This permits the cytosine methyltransferases to be converted into targeting devices by molecular cloning techniques that produce fusion proteins containing a functional enzyme, because a methyltransferase with a given sequence specificity will serve to uniquely target that address. For example, the cytosine methyltransferases like M·HhaI and M·MspI have distinct recognition specificities. When 5-fluorocytosine (F) is placed at the targeted cytosine in each recognition sequence (GFGC for M·HhaI and FCGG for M·MspI) the first recognition site becomes a unique address for M·HhaI and the second recognition site becomes a unique address (Figure 1) for M·MspI (Smith, et al., 1997, Smith, et al., 1997a). This permits precise control over the positioning of proteins or peptides fused to the methyltransferases (Smith, 1997a).
Larger version (22K)
Figure 1. Schematic of a three factor linear addressing system. The recognition sites for M·HhaI and M·MspI are placed at preselected points along a linear duplex DNA molecule. Methylation sites within the recognition sequence that are targeted by the methyltransferase are replaced with 5FdC (F). Attack of the FdC moiety by the appropriate enzyme produces a covalently linked complex between the enzyme and its recognition site. (M): mdC.
The availability of DNA scaffolds of almost any topology from the connectivity of a cube to that of a truncated octahedron (Seeman, 1993, Zhang and Seeman, 1992; Zhang and Seeman, 1994) now makes it possible to envision such scaffolds decorated with a variety of functional peptides or proteins in precisely controlled ways. In this report we have studied the utility of dU in the trapping methyltransferases at addressable sites in DNA.
DNA synthesis: Oligodeoxynucleotides were synthesized on a Cyclone Plus DNA Synthesizer (Millipore, Marlborough, MA) using standard phosphoramidite chemistry. Precursor phosphoramidites were purchased from PerSeptive Biosystems (Farmington, MA) or from Glenn Research (Sterling, VA). They were then purified using Oligo-Pure cartridges (Hamilton, Reno, NV) according to the manufacturer's protocol. This was followed by 32P end-labelling, as previously described (Smith, et al. 1991).
Mobility Shift Assays: Duplexes were formed by combining equimolar (20 µM) amounts of complementary strands in annealing buffer (10 mM Tris-HCl pH 7.4, 1 mM EDTA, and 100 mM NaCl) then treating with 95°C for 5 minutes and 50°C for 60 minutes. The samples were allowed to cool to room temperature for 10 min. and were then stored on ice until needed.
Methyltransferase Purification: M·HhaI was obtained via purification from E. coli RR1 containing the pSP72 plasmid (Promega, Madison, WI) carrying the entire HhaI methyltransferase gene ( Smith, et al. , 1997a). Transcription was from the endogenous hsdM promoter.
Formation of Assemblies: M·HhaI was incubated with either a 30mer containing a single recognition site with dU (bold) at the target site:
5'TCACCAGAT GCCG GUGC GTGACCTGTAGTT3'
3'AGTGGTCTA CGGC CGMC CACTGGACATCAA5'
or a 60mer containing two recognition sites with dU (bold) at each target site :
5'TCACCAGAT GUGC TGTAGGTCGT GCTACCTGGT TCCACCAGAT GUGC GTGACCTGTAGTT3'
3'AGTGGTCTA CGMG ACATCCAGCA CGATGGACCA AGGTGGTCTA CGMG CACTGGACATCAA5'.
The reaction took place overnight at 37°C in a buffer containing 50 mM Tris-HCl pH 7.5, 10 mM EDTA, 5 mM 2-mercaptoethanol, 80 µM S-adenosyl-[L]-methionine (New England Biolabs, Beverly,MA), and 4 µM oligodeoxynucleotide duplex. The reaction mixture was separated on a 6-20% non-denaturing gradient polyacrylamide gel containing a 4% stacking gel (Smith, et al., 1997). The gel was then exposed directly to X-ray film (Kodak, Rochester, NY) for autoradiography. 32P end-labelled fragments of TaqI-digested phiX174 RFII (New England Biolabs, Beverly, MA) were used as molecular length markers for DNA.
Ab initio Methods: We performed ab initio calculations on a number of variants of cytosine and uracil structures which we believe accurately model key intermediates in the proposed reaction pathway. Ab initio geometries were calculated at the Hartree-Fock level of theory using the STO-3G basis set with SPARTAN 4.0 (Wavefunction, Irvine, CA) running on a network of Silicon Graphics workstations. Single-point ab initio orbital energies calculated with the 6-31G* basis set were used to construct electron density surfaces with color maps of frontier orbital values. Blue indicates a high value for the orbital, and red indicates a low value.
Enzyme Activity Assay: Activity was determined using trichloroacetic acid-precipitable radioactivity retained on glass fiber filters as described (Smith et al., 1992). The M·HhaI assay was assayed in a buffer containing 50 mM Tris-HCl (pH 7.5), 10 mM EDTA, 5 mM 2-mercaptoethanol, 6µM S-adenosyl-L-[3H-methyl]methionine 15 Ci/mmole (Amersham, Arlington Heights, IL) and 5 µM oligodeoxynucleotide duplex 30-mer for one hour at 37°C.
Molecular Modeling: Molecular models of the assembly formed between the 60mer containing dU at the target site were constructed in BIOGRAF 3.21 (Molecular Simulations, San Diego, CA). The initial conformation of M·HhaI was that determined from the crystalline protein complexed with 5-fluorocytosine at its target site (Klimasauskas, et al., 1994). In the DNA, the four nucleotides at the target site were taken from the crystal structure data, except that 5FdC was converted directly to dU. The remainder of the duplex 60mer was constructed using the software's DNA builder. The structure was minimized in molecular mechanics to 0.1 (kcal/mole)/Å, and rendered using standard visualization tools available in the program. The model assumes linear DNA outside of the target sequences with the pitch of 10.0 bp/turn (i.e., the helical twist of 36.0°/bp derived from fiber diffraction). With this assumption, the C termini of the two enzymes assume the 180° dihedral angle, measured down the helical axis, shown in the model. Although the DNA is not bent by the enzyme, unwinding of the helix within the DNA binding site has been observed in the M·HhaI DNA complex. This unwinding could result in a twist angle of 31.6°/bp for those base pairs in the binding site. Since it is also possible that the twist angles for the DNA outside the binding sites could be as low as 34.3°/bp based on solution measurements, the true dihedral angle could be anywhere from 80° to 180° .
In the dC reaction catalyzed by DNA (cytosine-5)-methyltransferases (Figure 2), binding of the DNA substrate initiates a series of conformational changes that result in the opening of a salt bridge present between an arginine and a glutamic acid in the apoenzyme. Binding of the components of the salt bridge to the deoxycytosine moiety in the DNA substrate is facilitated by the capture of a proton in a hydrogen bond formed between an oxygen on glutamic acid and N3 of the targeted cytosine. Nucleophilic attack at the C6 position of the cytosine moiety by a cysteine residue at the active site rapidly produces a resonance-stabilized carbanion that is activated to attack a methyl group present on bound S-Adenosylmethionine co-factor, through C5 of cytosine. The resulting dihydrocytosine intermediate rapidly undergoes beta-elimination to produce free enzyme, S-adenosylhomocysteine and DNA in which the targeted cytosine moiety has been converted to 5-methylcytosine.
Figure 2: Attack of Deoxycytosine by DNA (Cytosine-5)Methyltransferases. Top: Chemical Depiction of Intermediates. In this model of the chemical reaction, the break-up of the glutamic acid-arginine salt bridge is stabilized by the capture of a proton poised between the ionized carboxyl oxygen of the glutamic acid and N3 of cytosine. This activates the ring for nucleophilic attack by a cysteine residue also at the active site. In the model shown here, the captured proton is assumed to be derived from the cysteine sulfhydryl increasing its nucleophilicity. Once nucleophilic attack occurs, the resonance-stabilized carbanion attacks the methyl group on S-adenosylmethionine to generate a dihydrocytosine intermediate. Resonance stabilization is enhanced by the proximity of the positively charged guanidinium group on arginine which forms a hydrogen bond with O2 of the intermediate, and is in a position to ligate negative charge on O2 generated through resonance. This intermediate undergoes beta-elimination to generate a 5-methylcytosine product and active enzyme. A tightly bound water molecule on the enzyme surface appears to be the proton acceptor in the beta-elimination step. Bottom: Quantum Chemical Depiction of Models of the Intermediates. Ab initio geometries were calculated at the Hartree-Fock level of theory using the STO-3G basis set with Spartan 4.0 (Wavefunction, Irvine, CA) running on a network of Silicon Graphics workstations. Single-point ab initio orbital energies calculated with the 6-31G* basis set were used to construct electron density surfaces with color maps of frontier orbital values. Blue indicates a high value for the orbital and red indicates a low value. High values for the Lowest Unoccupied Molecular Orbital (LUMO) are seen over C6 and C4 for 1-methylcytosine, indicating that these are potential acceptor sites for nucleophilic attack.The 1-methyl-6-sulfhydryl-cytosine carbanion was used to model the intermediate formed by the enzyme after nucleophilic attack at C6. High values of the Highest Occupied Molecular Orbital are confined to C5. This highly reactive orbital is then poised for attack on the methyl of S-adenosylmethionine.
When FdC is placed at the target site (Figure 3) the reaction proceeds in the same fashion. However, the methyltransfer step is slowed dramatically by the presence of the FdC moiety (see below) and the beta-elimination step is blocked by the presence of the fluorine atom at C5.
Figure 3: Attack of 5-Fluorodeoxycytosine by DNA (Cytosine-5)Methyltransferases.Top: Chemical Depiction of Intermediates. Enzymatic attack on 5FdC proceeds as in Figure 1 for attack on cytosine. The resonance-stabilized carbanion of 5FdC can attack the methyl group on S-adenosylmethionine to generate a dihydrocytosine intermediate at a very slow rate. In the dC reaction, this intermediate would undergo beta-elimination to generate a mdC product and active enzyme. 5FdC, blocks beta-elimination because the fluorine at C5 cannot be abstracted. Bottom: Quantum Chemical Depiction of Models of the Intermediates. Ab initio geometries were calculated at the Hartree-Fock level of theory using the STO-3G basis set with Spartan 4.0 (Wavefunction, Irvine, CA) running on a network of Silicon Graphics workstations. Single-point ab initio orbital energies calculated with the 6-31G* basis set were used to construct electron density surfaces with color maps of frontier orbital values. Blue indicates a high value for the orbital and red indicates a low value. High values for the Lowest Unoccupied Molecular Orbital (LUMO) are seen over C6 and C4 in the 1-methyl-5-fluorocytosine used to model the intermediate, indicating that these are potential acceptor sites for nucleophilic attack as with cytosine. The 1-methyl-6-sulfhydryl-carbanion derivative of 5-fluorocytosine was used to model the intermediate formed by the enzyme after nucleophilic attack at C6. High values of the Highest Occupied Molecular Orbital are again confined to C5. This highly reactive orbital is then poised for attack on the methyl of S-adenosylmethionine.
When dU is placed at the target site (Figure 4), the progress of the reaction is disturbed in several ways by the presence of the mispaired target base. First, the reorientation of the salt bridge is not expected to draw a proton into a hydrogen bond between O4 and the oxygen until nucleophic attack occurs at C6. The formation of the resonance stabilized O4 enol dramatically reduces the capacity of the system to attack the methyl on S-Adenosylmethionine. Like the reaction with FdC, the reaction with dU is slowed by the presence of the additional electronegative atom on the ring (i.e. O4). However, unlike the reaction with FdC, the beta-elimination step is not blocked.
Figure 4. Attack of Deoxyuracil by DNA (Cytosine-5) Methyltransferases.Top: Chemical Depiction of Intermediates. The initial nucleophilic attack of uracil by the enzyme is not favored because the resulting intemediate must develop negative charge on O4 as the negatively charged carboxyl group on glutamic acid forms a hydrogen bond with N3. Decay of the system through protonation of O4 (perhaps by the cysteine sulfhydryl on the enzyme) can produce a resonance stabililized carbanion much like that produced from 5-fluorocytosine. Attack of the methyl group on S-adenosyl-[L]-methionine by the carbanion followed by beta-elimination would produce 5-methyluracil (thymidine). Bottom:Quantum Chemical Depiction of Models of the Intermediates. High values for the Lowest Unoccupied Molecular Orbital (LUMO) are seen over C6 and C4 1-methyl-deoxyuracil, indicating that these are potential acceptor sites for nucleophilic attack as with cytosine. The 1-methyl-6-sulfhydryl-4-ol-carbanion derivative of deoxyuracil was used to model the intermediate formed by the enzyme after nucleophilic attack at C6. High values of the Highest Occupied Molecular Orbital are again confined to C5. Once methyl-transfer takes place, the high values of the Highest Occupied Molecular Orbital (HOMO) move to the sulfur atom in (1-methyl-6-sulfhydryl-5-methyl-deoxyuracil).
The nature of the chemical barrier to completion of the reaction is best measured by the energy of the highest occupied orbital of the carbanion (Figure 5).
Figure 5. Electronic Trapping of the Methyltransferase. The observed rates of enzyme catalyzed methyltransfer observed for FdC and dU are plotted relative to that observed for dC as a function of the calculated ab initio energy of the Highest Occupied Molecular Orbital E(HOMO) of the enzyme activated intermediate.
For dC this corresponds to -0.11278 hartree, in FdC this is reduced to -0.11813, and for the attacking dU enol it is reduced to -0.11939. At 4 µM oligodeoxynucleotide, the rate of methyltransfer is slowed by 70 to 200 fold depending on the oligodeoxynucleotide employed. Clearly both dU and FdC are powerful mechanism-based inhibitors of the methyltransferase reaction. Even so, the existence of a measurable rate of methyltransfer (2.8 fmole/min) to dU under these conditions suggests that the complexes should decay slowly. Since the complexes formed with dU appear to have a stability similar to the stability of complexes formed by FdC (Klimasauskas and Roberts, 1995, Yang, et al., 1995) there must be other factors that contribute to the stability.
One possibility is that there is a significant kinetic barrier established by the requirement for removal of the stabilizing proton at O4. In fact the system could stall if the relatively stable enol form of the intermediate were to form after methyltransfer. On the other hand, recent work with the human enzyme shows that it can stall at dC when the cytosine is mispaired. That inhibition stems from the fundamental catalytic imperative in enzymology: that an enzyme have a higher affinity for the transition state of a reaction than it has for either its reactants or its products (Haldane, 1930, Pauling, 1948).
For the dC methyltransferase reaction, the formation of the dihydrocytosine intermediate requires that the base unstack from the helix (Baker, et al., 1988) because the aromaticity of the ring is destroyed by the introduction of sp3 carbons at C5 and C6 at the same time that the geometry of the ring is changed from the planar conformation required for normal stacking in DNA to the non-planar conformation required for catalysis to proceed. This prediction has been confirmed by numerous experiments (Baker, et al., 1988, Smith, et al., 1991, Laayoun, et al., 1994) and was essentially proven by crystallographic work on the structure of the FdC complex formed between M·HhaI and FdC containing DNA (Klimasauskas, et al., 1994), which showed that the target base is rotated into a completely extrahelical conformation in order to reach the catalytic site on the enzyme surface.
Thus, any weakly stacked or extrahelical cytosine is a transition state analog for the catalysis (Smith, 1994) for which the enzyme is expected to have a very high affinity. The mispaired base appears to present a structural trap to the enzyme since it is locked in a structure resembling the transition state by the structure of the mispaired DNA substrate itself. In effect, the enzyme is expected to rapidly bind and attack the substrate but it is not expected to be able to release the methylated product (if it forms) because it retains a high affinity for the unstacked transtion-state analog. The displacement of the mispaired 5-methyldeoxyuracil product (mdU) in the dG:mdU mispair relative to the normally paired dG:mdC product is depicted in Figure 6.
Figure 6. Structural Trapping of the Methyltransferase. The crystal structures of the dG:dC substrate for the methyltransferase (Drew et al., 1981) and that of its normal dG:mdC product (Timsit and Moras, 1995) are compared with the structure of the mispaired dG:mdU product (Hunter, et al., 1987) expected after methylation of dU. Notice the strong displacement of the mdU moiety into the major groove of the helix, and the weaker displacement of the dG moiety into the minor groove necessitated by the abberrant hydrogen bonding scheme in the mispair.
Each of these considerations suggests that methyltransferases stalled at dU should be as stable as methyltransferases bound at FdC. The experiment depicted in Figure 7 shows that a stable assembly can be formed with a DNA 60mer containing two dU residues at targeted sites in the 60mer spaced 35 nt apart and M·HhaI. The yield and stability of the complex are comparable to that of the complex formed between M·HhaI and this same 60mer sequence containing FdC at the sites occupied by the dU in this oligodeoxynucleotide duplex (Smith, et al., 1997). A molecular model of the biostructure formed with dU is depicted in Figure 8.
Figure 7. Formation of Enzyme-DNA Complexes with dU-Containing Oligodeoxynucleotides. (a) A standard curve of M·MspI and M·HhaI molecules complexed with duplex 64mers or duplex 60mers containing an FdC residue at the enzymatic target site. Bacterial enzymes linked to DNAs of known lengths (Smith, et al., 1997), were constructed as described in Materials and Methods. (b) Electrophoretic separation of complexes formed between M·HhaI and dU-containing oligonucleotides. Markers in (a) identify the complexes as diagramed in the center of the figure. Markers to the right of panel (b) show the positions of duplex DNA molecules of the indicated lengths.
Larger version (80K)
Figure 8. Molecular Model of the Assembly Formed between a Duplex DNA 60mer Containing dU Moieties at Target Sites Spaced 35nt Apart. The model shown was constructed and rendered as described in Materials and Methods. In this rendering the C-termini of the two M·HhaI molecules (white) lie roughly on a vertical line that bisects the image.
In this report we have used a simple technology demonstration system to show that both FdC and dU can be used to address methyltransferases to preselected sites on DNA. Our data suggest that other mechanism-based inhibitors of these enzymes will also be useful in this application, and when coupled with previous work, suggests that a variety of useful biostructures and devices can be constructed with this technology.
*Supported by grant N00014-94-1-1116 from the Office of Naval Research
|Last Modified by authors: 02:01am PDT, October 04, 1997|