DNA-adenine methylation at certain GATC sites plays a pivotal role in bacterial and phage gene expression as well as bacterial virulence. We report here the crystal structures of the bacteriophage T4Dam DNA adenine methyltransferase (MTase) in a binary complex with the methyl-donor product S-adenosyl-L-homocysteine (AdoHcy) and in a ternary complex with a synthetic 12-bp DNA duplex and AdoHcy. T4Dam contains two domains: a seven-stranded catalytic domain that harbors the binding site for AdoHcy and a DNA binding domain consisting of a five-helix bundle and a β-hairpin that is conserved in the family of GATC-related MTase orthologs. Unexpectedly, the sequence-specific T4Dam bound to DNA in a nonspecific mode that contained two Dam monomers per synthetic duplex, even though the DNA contains a single GATC site. The ternary structure provides a rare snapshot of an enzyme poised for linear diffusion along the DNA.
DNA methyltransferases methylate target bases within specific nucleotide sequences. Three structures are described for bacteriophage T4 DNA-adenine methyltransferase (T4Dam) in ternary complexes with partially and fully specific DNA and a methyl-donor analog. We also report the effects of substitutions in the related Escherichia coli DNA methyltransferase (EcoDam), altering residues corresponding to those involved in specific interaction with the canonical GATC target sequence in T4Dam. We have identified two types of protein-DNA interactions: discriminatory contacts, which stabilize the transition state and accelerate methylation of the cognate site, and anti-discriminatory contacts, which do not significantly affect methylation of the cognate site but disfavor activity at noncognate sites. These structures illustrate the transition in enzyme-DNA interaction from nonspecific to specific interaction, suggesting that there is a temporal order for formation of specific contacts.
The phage T4Dam and EcoDam DNA-[adenine-N6] methyltransferases (MTases) methylate GATC palindromic sequences, while the BamHI DNA-[cytosine-N4] MTase methylates the GGATCC palindrome (which contains GATC) at the internal cytosine residue. We compared the ability of these enzymes to interact productively with defective duplexes in which individual elements were deleted on one chain. A sharp decrease in kcat was observed for all three enzymes if a particular element of structural symmetry was disrupted. For the BamHI MTase, integrity of the ATCC was critical, while an intact GAT sequence was necessary for the activity of T4Dam, and an intact GA was necessary for EcoDam. Theoretical alignment of the region of best contacts between the protein and DNA showed that in the case of a palindromic interaction site, a zone covering the 5′-symmetric residues is located in the major groove versus a zone of contact covering the 3′-symmetric residues in the minor groove. Our data fit a simple rule of thumb that the most important contacts are aligned around the methylation target base: if the target base is in the 5′ half of the palindrome, the interaction between the enzyme and the DNA occurs mainly in the major groove; if it is in the 3′ half, the interaction occurs mainly in the minor groove.
A nomenclature is described for restriction endonucleases, DNA methyltransferases, homing endonucleases and related genes and gene products. It provides explicit categories for the many different Type II enzymes now identified and provides a system for naming the putative genes found by sequence analysis of microbial genomes.
The fluorescence of 2-aminopurine (2A)-substituted duplexes
(contained in the GATC target site) was investigated by titration
with T4 Dam DNA-(N6-adenine)-methyltransferase.
With an unmethylated target (2A/A duplex) or
its methylated derivative (2A/mA duplex),
T4 Dam produced up to a 50-fold increase in fluorescence, consistent
with 2A being flipped out of the DNA helix. Though neither S-adenosyl-l-homocysteine nor
sinefungin had any significant effect, addition of substrate S-adenosyl-l-methionine (AdoMet) sharply reduced the Dam-induced
fluorescence with these complexes. In contrast, AdoMet had no effect on
the fluorescence increase produced with an 2A/2A double-substituted
duplex. Since the 2A/mA duplex cannot
be methylated, the AdoMet-induced decrease in fluorescence cannot
be due to methylation per se. We propose that T4
Dam alone randomly binds to the asymmetric 2A/A
and 2A/mA duplexes, and that AdoMet
induces an allosteric T4 Dam conformational change that promotes
reorientation of the enzyme to the strand containing the native
base. Thus, AdoMet increases enzyme binding-specificity, in addition
to serving as the methyl donor. The results of pre-steady-state
methylation kinetics are consistent with this model.
Properties of a mutant bacteriophage T2 DNA [N6-adenine] methyltransferase
(T2 Dam MTase) have been investigated for its potential utilization
in RecA-assisted restriction endonuclease (RARE) cleavage. Steady-state
kinetic analyses with oligonucleotide duplexes revealed that, compared
to wild-type T4 Dam, both wild-type T2 Dam and mutant T2 Dam P126S
had a 1.5-fold higher kcat in methylating
canonical GATC sites. Additionally, T2 Dam P126S showed increased efficiencies
in methylation of non-canonical GAY sites relative to the wild-type
enzymes. In agreement with these steady-state kinetic data, when
bacteriophage λ DNA was used as a substrate,
maximal protection from restriction nuclease cleavage in
vitro was achieved on the sequences GATC, GATN and GACY, while
protection of GACR sequences was less efficient. Collectively, our
data suggest that T2 Dam P126S can modify 28 recognition sequences.
The feasibility of using the mutant enzyme in RARE cleavage with BclI and EcoRV endonucleases has been
shown on phage λ DNA and with BclI
and DpnII endonucleases on yeast chromosomal DNA embedded
The DNA methyltransferase of bacteriophage T4 (T4 Dam MTase) recognizes the palindromic sequence GATC, and catalyzes transfer of the methyl group from S-adenosyl-l-methionine (AdoMet) to the N6-position of adenine [generating N6-methyladenine and S-adenosyl-l-homocysteine (AdoHcy)]. Pre-steady state kinetic analysis revealed that the methylation rate constant kmeth for unmethylated and hemimethylated substrates (0.56 and 0.47 s–1, respectively) was at least 20-fold larger than the overall reaction rate constant kcat (0.023 s–1). This indicates that the release of products is the rate-limiting step in the reaction. Destabilization of the target-base pair did not alter the methylation rate, indicating that the rate of target nucleoside flipping does not limit kmeth. Preformed T4 Dam MTase–DNA complexes are less efficient than preformed T4 Dam MTase–AdoMet complexes in the first round of catalysis. Thus, this data is consistent with a preferred route of reaction for T4 Dam MTase in which AdoMet is bound first; this preferred reaction route is not observed with the DNA-[C5-cytosine]-MTases.
The interaction of the phage T4 Dam DNA-[N6-adenine] methyltransferase with 24mer synthetic oligonucleotide duplexes having different purine base substitutions in the palindromic recognition sequence, GATC, was investigated by means of gel shift and methyl transfer assays. The substitutions were introduced in either the upper or lower strand: guanine by 7-deazaguanine (G-->D) or 2-aminopurine (G-->N) and target adenine by purine (A-->P) or 2-aminopurine (A-->N). The effects of each base modification on binding/methylation were approximately equivalent for both strands. G-->D and G-->N substitutions resulted in a sharp decrease in binary complex formation. This suggests that T4 Dam makes hydrogen bonds with either the N7- or O6-keto groups (or both) in forming the complex. In contrast, A-->P and A-->N substitutions were much more tolerant for complex formation. This confirms our earlier observations that the presence of intact 5'-G:C base pairs at both ends of the methylation site is critical, but that base substitutions within the central A:T base pairs show less inhibition of complex formation. Addition of T4 Dam to a complete substrate mixture resulted in a burst of [3H]methylated product. In all cases the substrate dependencies of bursts and methylation rates were proportional to each other. For the perfect 24mer k cat = 0.014/s and K m = 7.7 nM was obtained. In contrast to binary complex formation the two guanine substitutions exerted relatively minor effects on catalytic turnover (the k cat was reduced at most 2. 5-fold), while the two adenine substitutions showed stronger effects (5- to 15-fold reduction in k cat). The effects of base analog substitutions on K m(DNA) were more variable: A-->P (decreased); A-->N and G-->D (unchanged); G-->N (increased).
Late in its growth cycle, transcription of the phage Mu mom promoter (Pmom) is activated by the phage gene product, C, a site-specific DNA binding protein. In vitro transcription analyses showed that this activation does not require specific contacts between C and the carboxyl-terminal region of the α or ς70 subunit of Escherichia coli RNA polymerase. Unexpectedly, these results are in contrast to those known for another Mu-encoded transcriptional activator, Mor, which has a high degree of sequence identity with C and appears to interact with the carboxyl termini of both α and ς70.
Transcription of the bacteriophage Mu mom operon is strongly repressed by the host OxyR protein in dam - but not dam + cells. In this work we show that the extent of mom modification is sensitive to the relative levels of the Dam and OxyR proteins and OxyR appears to modulate the level of mom expression even in dam + cells. In vitro studies demonstrated that OxyR is capable of binding hemimethylated P mom , although its affinity is reduced slightly compared with unmethylated DNA. Thus, OxyR modulation of mom expression in dam + cells can be attributed to its ability to bind hemimethylated P mom DNA, the product of DNA replication.
The DNA-[N 6-adenine]-methyltransferase (Dam MTase) of phage T4 catalyzes methyl group transfer from S-adenosyl-l-methionine (AdoMet) to the N6-position of adenine in the palindromic sequence, GATC. We have used a gel shift assay to monitor complex formation between T4 Dam and various synthetic duplex oligonucleotides, either native or modified/defective. The results are summarized as follows. (i) T4 Dam bound with approximately 100-fold higher affinity to a 20mer specific (GATC-containing) duplex containing the canonical palindromic methylation sequence, GATC, than to a non-specific duplex containing another palindrome, GTAC. (ii) Compared with the unmethylated duplex, the hemimethylated 20mer specific duplex had a slightly increased ( approximately 2-fold) ability to form complexes with T4 Dam. (iii) No stable complex was formed with a synthetic 12mer specific (GATC-containing) duplex, although T4 Dam can methylate it. This indicates that there is no relation between formation of a catalytically competent 12mer-Dam complex and one stable to gel electrophoresis. (iv) Formation of a stable complex did not require that both strands be contiguous or completely complementary. Absence of a single internucleotide phosphate strongly reduced complex formation only when missing between the T and C residues. This suggests that if T4 Dam makes critical contact(s) with a backbone phosphate(s), then the one between T and C is the only likely candidate. Having only one half of the recognition site intact on one strand was sufficient for stable complex formation provided that the 5'G.C base-pairs be present at both ends of the palindromic, GATC. Since absence of either a G or C abolished T4 Dam binding, we conclude that both strands are recognized by T4 Dam.
The bacteriophage T2 and T4 dam genes code for a DNA (N6-adenine)methyltransferase (MTase). Nonglucosylated, hydroxymethylcytosine-containing T2gt- virion DNA has a higher level of methylation than T4gt- virion DNA does. To investigate the basis for this difference, we compared the intracellular enzyme levels following phage infection as well as the in vitro intrinsic methylation capabilities of purified T2 and T4 Dam MTases. Results from Western blotting (immunoblotting) showed that the same amounts of MTase protein were produced after infection with T2 and T4. Kinetic analyses with purified homogeneous enzymes showed that the two MTases had similar Km values for the methyl donor, S-adenosyl-L-methionine, and for substrate DNA. In contrast, they had different k(cat) values (twofold higher for T2 Dam MTase). We suggest that this difference can account for the ability of T2 Dam to methylate viral DNA in vivo to a higher level than does T4 Dam. Since the T2 and T4 MTases differ at only three amino acid residues (at positions 20 [T4, Ser; T2, Pro], 26 [T4, Asn; T2, Asp], and 188 [T4, Asp; T2, Glu]), we have produced hybrid proteins to determine which residue(s) is responsible for increased catalytic activity. The results of these analyses showed that the residues at positions 20 and 26 are responsible for the different k(cat) values of the two MTases for both canonical and noncanonical sites. Moreover, a single substitution of either residue 20 or 26 was sufficient to increase the k(cat) of T4 Dam.
Transcription of the bacteriophage Mu mom operon requires transactivation by the phage-encoded C protein. DNase I footprinting showed that in the absence of C, Escherichia coli RNA polymerase E(sigma)70 (RNAP) binds to the mom promoter (Pmom) region at a site, P2 (from -64 to -11 with respect to the transcription start site), on the top (non-transcribed) strand. This is slightly upstream from, but overlapping P1 (-49 to +16), the functional binding site for rightward transcription. Host DNA-[N6-adenine] methyltransferase (Dam) methylation of three GATCs immediately upstream of the C binding site is required to prevent binding of the E.coli OxyR protein, which represses mom transcription in dam- strains. OxyR, known to induce DNA bending, is normally in a reduced conformation in vivo, but is converted to an oxidized state under standard in vitro conditions. Using DNase I footprinting, we provide evidence supporting the proposal that the oxidized and reduced forms of OxyR interact differently with their target DNA sequences in vitro. A mutant form, OxyR-C199S, was shown to be able to repress mom expression in vivo in a dam- host. In vitro DNase I footprinting showed that OxyR-C199S protected Pmom from -104 to -46 on the top strand and produced a protection pattern characteristic of reduced wild-type OxyR. Prebinding of OxyR-C199S completely blocked RNAP binding to P2 (in the absence of C), whereas it only slightly decreased binding of C to its target site (-55 to -28, as defined by DNase I footprinting). In contrast, OxyR-C199S strongly inhibited C-activated recruitment of RNAP to P1. These results indicate that OxyR repression is mediated subsequent to binding by C. Mutations have been isolated that relieve the dependence on C activation and have the same transcription start site as the C-activated wild-type promoter. One such mutant, tin7, has a single base change at -14, which changes a T6 run to T3GT2. OxyR-C199S partially inhibited RNAP binding to the tin7 promoter in vitro, even though the OxyR and RNAP-P1 binding sites probably do not overlap, and in vivo expression of tin7 was reduced 5- to 10-fold in dam- cells. These results suggest that OxyR can repress tin7.
The phage Mu gene C encodes a 16.5-kDa site-specific DNA-binding protein that functions as a trans-activator of the four phage "late" operons, including mom. We have overexpressed and purified C and used it for DNase I footprinting and transcription analyses in vitro. The footprinting results are summarized as follows. (i) As shown previously (V. Balke, V. Nagaraja, T. Gindlesperger, and S. Hattman, Nucleic Acids Res. 12:2777-2784, 1992) in vivo, Escherichia coli RNA polymerase (RNAP) bound the wild-type (wt) mom promoter at a site slightly upstream from the functionally active site bound on the C-independent tin7 mutant promoter. (ii) In the presence of C, however, RNAP bound the wt promoter at the same site as tin7. (iii) C and RNAP were both bound by the mom promoter at overlapping sites, indicating that they were probably on different faces of the DNA helix. The minicircle system of Choy and Adhya (H. E. Choy and S. Adhya, Proc. Natl. Acad. Sci. USA 90:472-476, 1993) was used to compare transcription in vitro from the wt and tin7 promoters. This analysis showed the following. (i) Few full-length transcripts were observed from the wt promoter in the absence of C, but addition of increasing amounts of C greatly stimulated transcription. (ii) RNA was transcribed from the tin7 promoter in the absence of C, but addition of C had a small stimulatory effect. (iii) Transcription from linearized minicircles or restriction fragment templates was greatly reduced (although still stimulated by C) with both the wt and tin7 promoters. These results show that C alone is capable of activating rightward transcription in vitro by promoting RNAP binding at a functionally active site. Additionally, DNA topology plays an important role in transcriptional activation in vitro.
Comparison of the deduced amino acid sequences of DNA-[N6-adenine]-methyltransferases has revealed several conserved regions. All of these enzymes contain a DPPY [or closely related] motif. By site-directed mutagenesis of a cloned T4 dam gene, we have altered the first proline residue in this motif [located in conserved region IV of the T4 Dam-MTase] to alanine or threonine. The mutant enzymic forms, P172A and P172T, were overproduced and purified. Kinetic studies showed that compared to the wild-type [wt] the two mutant enzymic forms had: (i) an increased [5 and 20-fold, respectively] Km for substrate, S-adenosyl-methionine [AdoMet]; (ii) a slightly reduced [2 and 4-fold lower] kcat; (iii) a strongly reduced kcat/KmAdoMet [10 and 100-fold]; and (iv) almost the same Km for substrate DNA. Equilibrium dialysis studies showed that the mutant enzymes had a reduced [4 and 9-fold lower] Ka for AdoMet. Taken together these data indicate that the P172A and P172T alterations resulted primarily in a reduced affinity for AdoMet. This suggests that the DPPY-motif is important for AdoMet-binding, and that region IV contains or is part of an AdoMet-binding site.
Bacteriophage T4 codes for a DNA-[N6-adenine] methyltransferase (Dam) which recognizes primarily the sequence GATC in both cytosine- and hydroxymethylcytosine-containing DNA. Hypermethylating mutants, damh, exhibit a relaxation in sequence specificity, that is, they are readily able to methylate non-canonical sites. We have determined that the damh mutation produces a single amino acid change (Pro126 to Ser126) in a region of homology (III) shared by three DNA-adenine methyltransferases; viz, T4 Dam, Escherichia coli Dam, and the DpnII modification enzyme of Streptococcus pneumoniae. We also describe another mutant, damc, which methylates GATC in cytosine-containing DNA, but not in hydroxymethylcytosine-containing DNA. This mutation also alters a single amino acid (Phe127 to Val127). These results implicate homology region III as a domain involved in DNA sequence recognition. The effect of several different amino acids at residue 126 was examined by creating a polypeptide chain terminating codon at that position and comparing the methylation capability of partially purified enzymes produced in the presence of various suppressors. No enzyme activity is detected when phenylalanine, glutamic acid, or histidine is inserted at position 126. However, insertion of alanine, cysteine, or glycine at residue 126 produces enzymatic activity similar to Damh.
Bacteriophage T2 codes for a DNA-(adenine-N6)methyltransferase (Dam), which is able to methylate both cytosine- and hydroxymethylcytosine-containing DNAs to a greater extent than the corresponding methyltransferase encoded by bacteriophage T4. We have cloned and sequenced the T2 dam gene and compared it with the T4 dam gene. In the Dam coding region, there are 22 nucleotide differences, 4 of which result in three coding differences (2 are in the same codon). Two of the amino acid alterations are located in a region of homology that is shared by T2 and T4 Dam, Escherichia coli Dam, and the modification enzyme of Streptococcus pneumoniae, all of which methylate the sequence 5' GATC 3'. The T2 dam and T4 dam promoters are not identical and appear to have slightly different efficiencies; when fused to the E. coli lacZ gene, the T4 promoter produces about twofold more beta-galactosidase activity than does the T2 promoter. In our first attempt to isolate T2 dam, a truncated gene was cloned on a 1.67-kilobase XbaI fragment. This construct produces a chimeric protein composed of the first 163 amino acids of T2 Dam followed by 83 amino acids coded by the pUC18 vector. Surprisingly, the chimera has Dam activity, but only on cytosine-containing DNA. Genetic and physical analyses place the T2 dam gene at the same respective map location as the T4 dam gene. However, relative to T4, T2 contains an insertion of 536 base pairs 5' to the dam gene. Southern blot hybridization and computer analysis failed to reveal any homology between this insert and either T4 or E. coli DNA.
The T4 dam+ gene has been cloned (S. L. Schlagman and S. Hattman, Gene 22:139-156, 1983) and transferred into an Escherichia coli dam-host. In this host, the T4 Dam DNA methyltransferase methylates mainly, if not exclusively, the sequence 5'-GATC-3'; this sequence specificity is the same as that of the E. coli Dam enzyme. Expression of the cloned T4 dam+ gene suppresses almost all the phenotypic traits associated with E. coli dam mutants, with the exception of hypermutability. In wild-type hosts, 20- to 500-fold overproduction of the E. coli Dam methylase by plasmids containing the cloned E. coli dam+ gene results in a hypermutability phenotype (G.E. Herman and P. Modrich, J. Bacteriol. 145:644-646, 1981; M.G. Marinus, A. Poteete, and J.A. Arraj, Gene 28:123-125, 1984). In contrast, the same high level of T4 Dam methylase activity, produced by plasmids containing the cloned T4 dam+ gene, does not result in hypermutability. To account for these results we propose that the E. coli Dam methylase may be directly involved in the process of methylation-instructed mismatch repair and that the T4 Dam methylase is unable to substitute for the E. coli enzyme.
We compared the known DNA nucleotide and encoded amino acid sequences of the Escherichia coli and bacteriophage T4 dam (DNA-adenine methyltransferase) genes. Despite the absence of any DNA sequence homology, there were four regions (11 to 33 residues long) of amino acid sequence homology containing 45 to 64% identity. These results suggest that the genes for these two enzymes have a common evolutionary origin.
We examined the DNA of Saccharomyces cerevisiae by both HpaII-MspI restriction enzyme digestion and high-performance liquid chromatography analysis for the possible presence of 5-methylcytosine. Both of these methods failed to detect cytosine methylation within this yeast DNA; i.e., there is less than 1 5-methylcytosine per 3,100 to 6,000 cytosine residues.
Purified nuclear DNA from two mealybug species was analyzed for its 5-methylcytosine (m5C) content by reversed-phase high-pressure liquid chromatography. We observed that the percent m5C (percentage of cytosines which are methylated) varied between the two species, between males and females of the same species, and between lines with and without supernumerary B chromosomes. This is the first case of a sex-specific difference in overall DNA methylation level. In contrast to a recent report (Deobagkar et al., J. Biosci. [India] 4:513-526, 1982), we found no other modified bases in the DNA. Overall, the percent m5C in Pseudococcus obscurus was two to three times higher than in Pseudococcus calceolariae. In both species, the percent m5C in males was higher than in females, although only in P. calceolariae was the difference statistically significant (0.68 +/- 0.02 versus 0.44 +/- 0.04). The high m5C content in males was correlated with the presence of a paternally derived, genetically inactive set of chromosomes which is facultatively heterochromatic. The presence of constitutive heterochromatin, however, was associated with a lower m5C content. Thus, for example, the percent m5C in females of a P. obscurus line with heterochromatic B chromosomes (1.09 +/- 0.04) was significantly lower than that of a related line lacking such chromosomes (1.26 +/- 0.06). Our findings are discussed with respect to the possible relationship between DNA methylation and heterochromatization.
Mycoplasma virus L2 is subject to host-specific restriction and modification in Acholeplasma laidlawii strains JA1 and K2. We have examined the DNAs from both host cells and viruses propagated on these strains with respect to susceptibility to cleavage by restriction endonucleases and for DNA base modifications. We show that, in strain K2 and L2 virus grown on K2 cells, cytosine in the sequence GATC is methylated to 5-methylcytosine and, although strain K2 and L2 viruses grown on K2 contain N6-methyladenine in their DNA, adenine in the sequence GATC is not methylated. In contrast to K2, strain JA1 and L2 virus grown on JA1 cells contain no detectable methylated bases. It is not known which of the methylated bases in K2 is the basis for the K2 restriction-modification system operative on L2 virus.
The sequence specificity of the Tetrahymena DNA-adenine methylase was determined by nearest-neighbor analyses of in vivo and in vitro methylated DNA. In vivo all four common bases were found to the 5' side of N6-methyladenine, but only thymidine was 3'. Homologous DNA already methylated in vivo and heterologous Micrococcus luteus DNA were methylated in vitro by a partially purified DNA-adenine methylase activity isolated from Tetrahymena macronuclei. The in vitro-methylated sequence differed from the in vivo sequence in that both thymidine and cytosine were 3' nearest neighbors of N6-methyladenine.
Deoxyribonucleic acid (DNA) of the transcriptionally active macronucleus of Tetrahymena thermophila is methylated at the N6 position of adenine to produce methyladenine (MeAde); approximately 1 in every 125 adenine residues (0.8 mol%) is methylated. Transcriptionally inert micronuclear DNA is not methylated (< or = 0.01 mol% MeAde; M. A. Gorovsky, S. Hattman, and G. L. Pleger, J. Cell Biol. 56:697-701, 1973). There is no detectable cytosine methylation in macronuclei in Tetrahymena DNA (< or = 0.01 mol% 5-methylcytosine). MeAde-containing DNA sequences in macronuclei are preferentially digested by both staphylococcal nuclease and pancreatic deoxyribonuclease I. In contrast, there is no preferential release of MeAde during digestion of purified DNA. These results indicate that MeAde residues are predominantly located in "linker DNA" and perhaps have a function in transcription. Pulse-chase studies showed that labeled MeAde remains preferentially in linker DNA during subsequent rounds of DNA replication; i.e., there is little, if any, movement of nucleosomes during chromatin replication. This implies that nucleosomes may be phased with respect to DNA sequence.
Bacteriophage Mu DNA was labeled after induction in the presence of [8-3H]adenine. Purified DNA was enzymatically digested, and the 3H-labeled dinucleotides were isolated. Approximately 15 to 20% of the adenine residues were modified to a new form, Ax, as observed previously (S. Hattman, J. Virol. 32:468-475, 1979) in bulk DNA. Paper electrophoretic analysis revealed that only two dinucleotide species contain Ax, namely, (Ax,C) and (Ax,G). The observation that only C and G are the nearest neighbors of Ax is consistent with the proposal of Kahmann and Kamp (R. Kahmann and D. Kamp, J. Mol. Biol., in press) that modification of Mu DNA occurs at the A residue within the pentanucleotide sequence, 5'...(CG)-A-(GC)-N-Py...3'.