Determinants of cofactor binding to DNA methyltransferases: Insights from a systematic series of structural variants of S-adenosylhomocysteine

Article abstract of DOI:10.1039/b415446k

S-Adenosylmethionine (AdoMet) is a commonly used cofactor, second only to ATP in the variety of reactions in which it participates. It is the methyl donor in the majority of methyl transfer reactions, including methylation of DNA, RNA, proteins and small molecules. Almost all structurally characterised methyltransferases share a conserved AdoMet-dependent methyltransferase fold, in which AdoMet is bound in the same orientation. Although potential interactions between the cofactor and methyltransferases have been inferred from crystal structures, there has not been a systematic study of the contributions of each functional group to binding. To explore the binding interaction we synthesised a series of seven analogues of the methyltransferase inhibitor S-adenosylhomocysteine (AdoHcy), each containing a single modification, and tested them for the ability to inhibit methylation by HhaI and HaeIII DNA methyltransferase. Comparison of the K_i values highlights the structural determinants for cofactor binding, and indicates which nucleoside and amino acid functional groups contribute significantly to AdoMet binding. An understanding of the binding of AdoHyc to methyltransferases will greatly assist the design of AdoMet inhibitors.

Full text of DOI:10.1039/b415446k

View Article Online / Journal Homepage / Table of Contents for this issue
A R T I C L E
Determinants of cofactor binding to DNA methyltransferases:
insights from a systematic series of structural variants of
S-adenosylhomocysteine†
Helen M. Cohen,a,b Andrew D. Griffiths,b Dan S. Tawfikc and David Loakes*b
a The MRC Centre for Protein Engineering, MRC Centre, Hills Road, Cambridge, UK CB2
2QH
b MRC Laboratory of Molecular Biology, MRC Centre, Hills Road, Cambridge, UK CB2 2QH.
E-mail: david.loakes@mrc-lmb.cam.ac.uk; Fax: +44 1223 412178; Tel: +44 1223 402206
c Department of Biological Chemistry, The Weizmann Institute of Science, Rehovot, 76 100,
Israel
Received 6th October 2004, Accepted 28th October 2004
First published as an Advance Article on the web 26th November 2004
S-Adenosylmethionine (AdoMet) is a commonly used cofactor, second only to ATP in the variety of reactions in
which it participates. It is the methyl donor in the majority of methyl transfer reactions, including methylation of
DNA, RNA, proteins and small molecules. Almost all structurally characterised methyltransferases share a conserved
AdoMet-dependent methyltransferase fold, in which AdoMet is bound in the same orientation. Although potential
interactions between the cofactor and methyltransferases have been inferred from crystal structures, there has not
been a systematic study of the contributions of each functional group to binding. To explore the binding interaction
we synthesised a series of seven analogues of the methyltransferase inhibitor S-adenosylhomocysteine (AdoHcy),
each containing a single modification, and tested them for the ability to inhibit methylation by HhaI and HaeIII
DNA methyltransferase. Comparison of the Ki values highlights the structural determinants for cofactor binding,
and indicates which nucleoside and amino acid functional groups contribute significantly to AdoMet binding. An
understanding of the binding of AdoHyc to methyltransferases will greatly assist the design of AdoMet inhibitors.
methyltransferases, involved in cell cycle regulation, and the
Introduction
erythromycin resistance methyltransferases, which methylate
Methyl transfer is a reaction central to cellular biochemistry,
prokaryotic rRNA.7,8 Analogues of the methylation target, or
with 140 different classes of methyltransferase alkylating such
mechanism-based inhibitors, are more specific than inhibitors
diverse substrates as DNA, RNA, proteins, lipids, polysaccha-
such as sinefungin. A multisubstrate adduct synthesised by
rides and small molecules.1 S-Adenosylmethionine (AdoMet)
Wahnon et al., for example, inhibits adenine N6, but not cytosine
contains a highly reactive methylthiol group and is the cofac-
C5, DNA methyltransferases.8
tor and methyl donor for the majority (over 90%) of these
Despite the structural conservation of the cofactor binding
enzymes (www.expasy.ch/enzyme).2 With enzymes catalysing
site there is little sequence identity between methyltransferases
the methylation of N, O or S the substrates contain a polar-
in different classes. Only three motifs are conserved amongst
isable nucleophile, which directly attacks the activated methyl
the AdoMet-dependent methyltransferase family, located in the
group, whereas enzymes catalysing methylation of carbon atoms
usually proceed via addition and subsequent elimination from
P loop, G loop and part of strand b4.9 Notably each of these
positions is involved in cofactor binding.
a thiol group in the enzyme active site. The structures of
There is often greater sequence conservation within methyl-
over 50 different AdoMet-dependent methyltransferases are
transferase classes and DNA and RNA methyltransferases
known and include five distinct AdoMet-binding folds.3 Despite
contain particularly strongly conserved motifs. Cytosine C5
the extensive chemical and structural diversity of methylation
DNA methyltransferases share ten well defined and highly con-
substrates, the great majority of known methyltransferases
served motifs.10 Three structures of C5 DNA methyltransferases
contain the Class I fold and share a common evolutionary
are known: two bacterial type II restriction modification sys-
origin.3,4 Currently the structures of 31 methyltransferases have
tem methyltransferases, M.HhaI (recognition sequence GCGC,
been solved in complex with AdoMet or the reaction product,
solved in complex with AdoHcy and DNA)10 and M.HaeIII
S-adenosylhomocysteine (AdoHcy) and the relative position of
AdoMet is almost identical in each structure.
Analogues of AdoMet that inhibit methyltransfer, such as
(recognition sequence GGCC, solved in complex with DNA),11
and human DNMT2, which shows homology to other prokary-
otic and eukaryotic cytosine C5 DNA methyltransferases (solved
the natural product sinefungin, have therapeutic properties
in complex with AdoMet).12 Seven residues are proposed to form
including antifungal, antiviral, antiparasitic and antitumour
hydrogen bonds between M.HhaI and AdoMet, with four of
activities.5,6 However the high conservation of the AdoMet-
these identical (six conserved) in M.HaeIII and three identical
binding domain means that these analogues often inhibit a wide
(seven conserved) in DNMT2.
range of methyltransferases, leading to in vivo toxicity. Previous
In order to characterise the functional groups that allow
studies have investigated the ability of AdoMet analogues to
cofactor recognition by the active site of cytosine C5 DNA
inhibit enzymes such as the essential bacterial adenine N6-
methyltransferases we measured the inhibition of M.HhaI
and M.HaeIII by AdoHcy analogues. AdoHcy is structurally
similar to AdoMet, binding to methyltransferases in the same
† Electronic supplementary information (ESI) available: Table S1:
orientation, and with a similar Kd to AdoMet, but it lacks
the reactive methyl group and acts as a competitive inhibitor
with respect to AdoMet.13 A series of analogues was prepared,
proposed hydrogen bonds between methyltransferase and cofactor based
on the crystal structure of each methyltransferase with bound AdoHcy.
See http://www.rsc.org/suppdata/ob/b4/b415446k/
1 5 2
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1
T h i s j o u r n a l i s
T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 5
©

View Article Online
Fig. 1 AdoHcy analogues synthesized. Arrows show the functional groups that have been modified in each case.
with single modifications to the adenine, ribose and methionine
moieties (Fig. 1).
yield (Scheme 1). This was reacted with the di-sodium salt of
L-homocysteine in aqueous DMF solution to give 10. This
compound partially lost the benzoyl group, but in a slower
reaction than was observed with the acetyl group. Finally,
removal of the isopropylidene protection was carried out in 5%
aqueous TFA to give 1. N6-benzoyl-AdoHcy was stable under
these conditions, and no AdoHcy was observed, as was found
with the N6-acetyl route.
Results
Synthesis of AdoHyc analogues
A series of seven AdoHcy analogues were synthesised in order to
probe the structural determinants of cofactor binding (Fig. 1).
Initially we chose the N6-acetyl derivative. This analogue has
the advantage of weakening the hydrogen bonding ability of
the amino group, but with minimal steric effects. However, it
was found that the acetyl group was both acid and base labile
and it was never possible to prepare the material free from
AdoHcy. We therefore chose to make the N6-benzoyl derivative
(1) which has greater stability, although the benzoyl group adds
steric bulk, but still weakens the hydrogen bonding capability
of the amino group. In addition, the inosine derivative (SIH, 2)
was prepared, which has an altered hydrogen bonding pattern.
The effect of altering the nucleobase was examined by the use
of SIH and also the cytosine derivative (3). Modification of
the amino acid carboxyl group to alter the hydrogen bonding
behaviour and probe for steric effects was carried out by
conversion to the methylamide derivative 4. The remaining
modifications arise from the removal of functional groups,
namely the 2ꢀ-hydroxyl group of the ribose to give 5 (S2ꢀdAH),
and removal of either of the amino acid functionalities to
give 5ꢀ-(3-aminopropylthio)-5ꢀ-deoxyadenosine, 6, and 5ꢀ-(3-
carboxypropylthio)-5ꢀ-deoxyadenosine, 7, as described.14
Synthesis of AdoHcy and derivatives has been previously
described via coupling of 5ꢀ-O-tosyl15 or 5ꢀ-deoxy-5ꢀ-halo deriva-
tives of adenosine16 with the di-sodium salt of L-homocysteine.
Coupling of 5ꢀ-O-tosyl derivatives is generally carried out in
liquid ammonia after reduction of homocysteine, whereas 5ꢀ-
halo derivatives are coupled with the di-sodium salt of L-
homocysteine in water.17
Scheme 1 Synthesis of N6-benzoyl-AdoHyc. (a) NaI–acetone, 69%. (b)
Homocysteine, Na–NH3, 57%. (c) 5% aqueous TFA, 70%.
S-Inosinyl-L-homocysteine (SIH, 2) has been previously
described,18 prepared by coupling 5ꢀ-O-tosyl-isopropylidine
inosine and the di-sodium salt of L-homocysteine in liquid
ammonia. However, no characterisation of the product was
provided. As we had found that coupling of 5ꢀ-iodo deriva-
tives was more efficient, 5ꢀ-O-tosyl-isopropylidine inosine was
converted to 5ꢀ-deoxy-5ꢀ-iodo-isopropylidine inosine, and then
reacted with the di-sodium salt of L-homocysteine (Scheme 2).
However, the product, which was not identified, did not contain
homocysteine, and was possibly the cyclonucleoside. Repeating
the previously described synthesis,18 it became apparent from
the NMR spectra that the product was not the SIH derivative,
The first modification was the addition of a benzoyl group
to the adenine N6-amino group (analogue 1, Fig. 1). The N6-
benzoyl-isopropylidene adenosine derivative was converted to
its 5ꢀ-O-tosylate 8, and then reaction with sodium iodide in
acetone yielded the 5ꢀ-deoxy-5ꢀ-iodo derivative 9 in moderate
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1
1 5 3

View Article Online
Scheme 2 Synthesis of S-inosyl-AdoHcy. (a) DMAP, TEA, tert-butyldiphenylchlorosilane. (b) Homocysteine, Na–NH3, 53%. (c) 5% aqueous TFA,
69%.
and again may have been the cyclonucleoside; evidently the
inosine-6-oxo group required protection. Using a modification
of a procedure described by Grøtli et al.,19 isopropylidene
inosine was converted to its 5ꢀ-O-tosylate and then reacted
with tert-butyldiphenylchlorosilane. The authors stated that
the silyl protecting group is unstable, in particular to column
chromatography19 so the silylated inosine derivative was used
Scheme 3 Synthesis of 2ꢀ-deoxy-AdoHcy. (a) Homocysteine, Na–NH3,
without purification.
57%. (b) 5% aqueous TFA, 89%.
The silylated derivative was reacted with the di-sodium salt of
L-homocysteine in liquid ammonia. At the end of the reaction
the product was dissolved in aqueous ethanol and acidified
with acetic acid to remove the silyl protection. The product,
after reverse phase purification, was shown to be the desired
isopropylidene protected 11. Treatment with 5% aqueous TFA
removed the isopropylidene protection to give SIH, 2.
S-Cytidinyl-L-homocysteine (SCH, 3) has also been previ-
ously described,18 but the product was not characterised. In this
work SCH was prepared by an alternative route. Conversion
of cytidine to 5ꢀ-deoxy-5ꢀ-chlorocytidine was carried out by
reaction with thionyl chloride in hexamethylphosphoramide.16,20
This was then reacted with the di-sodium salt of L-homocysteine
in water to give, after purification by reverse phase chromatog-
raphy, SCH.
of AdoHcy in order to activate the carboxylic acid group. A
number of different protecting group strategies were employed,
primarily to avoid using acid labile protecting groups. However,
it was found that acid labile protecting groups could not be
avoided. The synthesis was carried out using the intermediate
10. Attempts to protect the homocysteine amino function with
fluorenylmethoxycarbonyloxy-succinimide in pyridine failed,
but using aqueous dioxane with sodium carbonate the desired
Fmoc derivative, 12, was prepared. This was then reacted with N-
hydroxysuccinimide with DCC to yield the active ester 13, which
was reacted with aqueous methylamine in DMF. This yielded the
desired methylamide derivative, 14, while removing the Fmoc
and benzoyl protecting groups in one step. Final removal of
the isopropylidene protecting group was carried out using 5%
aqueous TFA as before to give 4.
S-(2ꢀ-Deoxy-b-D-ribofuranosyladenosin-5ꢀ-yl)-L-homocyste-
ine (SdAH, 5) was prepared by an alternative route to that pre-
viously described (Scheme 3).21 N6-benzoyl-5ꢀ-O-tosyl-2ꢀdeoxy-
adenosine was converted to its 5ꢀ-iodo derivative 15, which
was then treated with the di-sodium salt of L-homocysteine in
aqueous DMF to give 16. Removal of the benzoyl group with
ammonia solution yielded 5.
The final modification we required was to the carboxylic acid
function, which was modified to the methylamide 4 (Scheme 4).
This required the protection of most of the functional groups
Kinetics and inhibition of methylation
The catalytic activities of M.HhaI and M.HaeIII were measured
by incorporation of radiolabelled methyl groups into synthetic
double stranded DNA substrates, Hhasub and Haesub, each
containing a single unmethylated recognition site, GCGC and
GGCC respectively. Steady state parameters, derived by fitting
Scheme 4 Synthesis of AdoHcy-methylamide. (a) Dioxane–H2O–Na2CO3–N-(9-Fmoc)succinimide, 94%. (b) N-Hydroxysuccinimide–DCCD, 72%.
(c) Aqueous MeNH2, 62%. (d) 5% aqueous TFA, 61%.
1 5 4
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1

View Article Online
the data to the Michaelis–Menten model (eqn. 1) are given in
Fig. 2 and Table 1.
KmAdoMet, 620 nM, and kcat, 0.0033 s−1, are comparable with those
measured in this study, but the higher KmDNA of 130 nM may have
been caused by the presence of the tag.
To determine the relative effect of each modification to
AdoHcy the analogues were tested for inhibition of M.HhaI
(Fig. 3) and M.HaeIII by fitting data to eqn. 2. All analogues
inhibit methylation by both enzymes, with Ki values ranging
from 8–3800 lM (Table 2), but none are as efficient inhibitors
as AdoHcy, which was determined to have a Ki of 15 2 nM
m0 = Vmax[S]0/([S]0 + Km)
(1)
AdoMet
Our measurement of Km
for M.HhaI, 110 nM, is
comparable to the published value of 161 nM, measured using
a similar unmethylated 30 base pair DNA substrate.22 Km
AdoMet
and KiAdoHcy of M.HhaI have also been determined using the
substrate poly(dG–dC), with values of 15 nM and 2.1 nM
respectively.13 We have previously measured kinetic constants
for M.HaeIII with an N-terminal FLAG tag:23 these values of
Fig. 2 Steady state kinetics of M.HaeIII. A. Plots of methyl group
incorporation by M.HaeIII versus time. Methyltransferase activity
of M.HaeIII was measured at 13 concentrations of AdoMet (three of
which are shown), in the range 25–2000 nM. Error bars are standard
AdoMet
error of the mean. B. Determination of the Km
of M.HaeIII. The
AdoMet
Km
was determined by fitting the data to the Michaelis–Menten
model (eqn. 1).
Table 1 Steady state parameters of M.HhaI and M.HaeIII with un-
methylated 30-mer DNA. Rate constants have been previously measured
for M.HhaI using a similar DNA substrate.22 Here we verify the Km
AdoMet
for this enzyme (previously measured as 161 nM) and establish the steady
state parameters for M.HaeIII
M.HhaI
M.HaeIII
KmAdoMet/nM
KmDNA/nM
a110 20
b280 40
Fig. 3 Plots of M.HhaI methyltransferase activity versus inhibitor
concentration for AdoHcy and two analogues with modifications to the
methionine moiety, 5ꢀ-(3-aminopropylthio)-5ꢀ-deoxyadenosine, 6, and
5ꢀ-(3-carboxypropylthio)-5ꢀ-deoxyadenosine, 7. Inhibition of M.HhaI
and M.HaeIII by all AdoHcy analogues were measured in a similar
manner. Ki values listed in Table 2 were derived by fitting the data to
eqn. 2.
d4.0 0.7
c3.7 1.1
k
cat/s−1
d0.085 0.005
c0.0068 0.0006
a Measurements were taken at 0.5 lM DNA. b 1.25 lM DNA. c or 1 lM
AdoMet. d Measured previously.22
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1
1 5 5

View Article Online
Table 2 Inhibition constants of AdoHcy analogues for M.HhaI and M.HaeIII
M.HhaI
M.HaeIII
aInhibitor
Modification
Ki/lM
Ki/KiAdoHcy
DDG/kJ mol−1
Ki/lM
Ki/KiAdoHcy
DDG/kJ mol−1
AdoHcy
—
0.015 0.002
8.1 2.1
1.0
540
—
16
32
19
21
21
18
17
0.069 0.011
1.00
460
6600
4200
4300
1000
540
—
16
23
21
22
18
16
21
1
2
3
4
5
6
7
N6-benzoyl
Inosine
32
4
b3900 800
26 0000
1800
3900
2900
1100
850
b450 80
b280 60
290 46
69 10
Cytidine
Methylamide
2ꢀ-Deoxy
Decarboxy
Deamino
27
59
44
16
13
3
9
6
4
2
37
5
230 50
3400
a See Fig. 1. b These values are estimates only, as methylation proceeded at 35–41% of the uninhibited rate at the highest inhibitor concentrations
tested.
for M.HhaI and 68 11 nM for M.HaeIII.
3, the positions of all the potential hydrogen bond-forming
groups are altered. Despite the large structural differences
from AdoHcy, the cytidine derivative inhibits both M.HhaI
and M.HaeIII, albeit with higher Ki values than AdoHcy.
The Ki of analogue 2 for M.HhaI, 3.9 mM, is 260 000 fold
higher than that of AdoHcy, indicating poor competition with
AdoMet for the cofactor binding site. The Ki for analogue
2 for M.HaeIII is lower (0.45 mM), perhaps indicating some
difference between M.HhaI and M.HaeIII in the hydrophobic
pocket that binds the adenosine moiety.
m0 = Vmax[S]0/(Km(1 + [I]/Ki) + [S]0)
(2)
Effect of base modifications (analogues 1, 2 and 3)
Analogue 1 is modified at the N6 position of the adenine ring
by the addition of a benzoyl group. Hydrogen bonding between
N6 on the adenine ring and an acidic side chain in the loop
following strand b3 is a conserved feature of Class I AdoMet-
dependent methyltransferases (Asp 60 in M.HhaI, Asp 50
in M.HaeIII, Fig. 4), although in some cases serine or asparagine
replace the acidic side chain. Only three of the structurally
characterised type I AdoMet-dependent methyltransferases lack
any side-chain mediated interaction with N6 (Table S1, ESI†).
The addition of a benzoyl group at N6 disrupts hydrogen
bonds and acts as a steric probe. N6-benzoyl AdoHcy, 1,
inhibits M.HhaI and M.HaeIII with Ki of 8.1 lM and 32 lM
respectively (Table 1). These inhibition constants are 500 fold
higher than those of AdoHcy, representing changes in DG of
16 kJ mol−1. This is equivalent to the strength of a single
hydrogen bond (typical hydrogen bond energies lie in the range
of 12–38 kJ mol−1).24
Despite the addition of a large hydrophobic group, analogue
1 was the best inhibitor of both methyltransferases, indicating
that it is still able to compete with AdoMet for its binding site. In
the structure of M.HhaI, and other methyltransferases, the edge
of the AdoMet adenine ring is situated at the entrance to the
AdoMet binding site where it is solvent accessible, explaining
how the additional benzoyl group might be accommodated.
Analogues 2 and 3, S-inosinyl-L-homocysteine and S-
cytidinyl-L-homocysteine, contain alternative nucleobases.
There is no conserved interaction between methyltransferases
and the N1, N3, and N7 atoms of the adenine ring. Hydrogen
bonds to N1 and N3 are suggested in about half the structures,
with hydrogen bonds predicted more frequently to N1 than N3
(Table S1†) and hydrogen bonds to N7 only predicted in two
methyltransferase structures. Where such hydrogen bonds are
suggested they always involve main chain atoms, often using
the same residues that form hydrophobic interactions with the
adenine and ribose rings. The interaction with N1 is often
provided by the main chain amino group of a hydrophobic
residue immediately following the conserved Asp or Glu that
contacts N6. Similarly N3 is most frequently contacted by a
main chain amino group of the residue immediately following
or immediately preceding the residue that contacts the sugar
hydroxyl groups. In the structure of M.HhaI these residues are
Ile 61 and Trp 41 respectively. Ile 61 is conserved in M.HaeIII
whereas Trp 41 aligns with Tyr 30.
Effect of sugar modification (analogue 5)
Compound 5, 2ꢀ-deoxy-AdoHcy, lacks one of the hydroxyl
groups on the sugar ring. With only one exception (the mycolic
acid synthetase family) methyltransferases with the type I
fold have a conserved glutamate or aspartate residue (Glu 40
in M.HhaI, Glu 29 in M.HaeIII, Fig. 3) at the end of strand
b2 that forms a pair of hydrogen bonds to the 2ꢀ- and 3ꢀ-ribose
hydroxyl groups (Table S1†). This conserved pair of hydrogen
bonds cannot be formed with 5. The Ki of 5 for M.HhaI is
2900 fold higher than that of AdoHcy, allowing an estimate
of the strength of the interaction between M.HhaI and the
2ꢀ-OH of AdoHcy of 21 kJ mol−1. Similarly, the interaction
between M.HaeIII and the 2ꢀ-OH of AdoHcy is estimated as
18 kJ mol−1.
Effects of amino acid modifications (analogues 4, 6 and 7)
Interactions with the methionine moiety are not conserved
between methyltransferases in different sub-classes within E.C.
2.1.1.- and there is some variation even within the DNA cytosine
C5 methyltransferases. M.HhaI is thought to make a number of
side chain and main chain contacts with the cofactor carboxyl
group but only water-mediated contacts to the cofactor amino
group. Main chain contacts may be conserved in M.HaeIII, but
Ser 305, which makes a hydrogen bond to the AdoHcy carboxyl
group in the crystal structure of M.HhaI, is not conserved
(Table S1†).
Three of the analogues were modified at the amino acid
moiety; analogue 6 lacks the carboxyl group, analogue 7
lacks the amino group, and analogue 4 is modified at the
carboxyl group by conversion to the methylamide. Each of these
alterations caused a substantial increase in Ki compared to that
of AdoHcy, (values of Ki were 540 to 4300 fold higher), evidence
of the important role of these functional groups in cofactor
binding and recognition by both enzymes.
Discussion
Analogues 2 and 3 have substantial changes to the base.
The modifications to analogue 2 would disrupt the hydrogen
bonding interactions normally formed by the methyltransferase
to N6 and N1. In the case of the cytidine derivative, analogue
This study analyses the inhibition of two cytosine C5 DNA
methyltransferases, M.HhaI and M.HaeIII, by a series of
AdoHcy analogues (Fig. 1) to characterise quantitatively the
contribution of individual functional groups to affinity. All
1 5 6
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1

View Article Online
modifications to AdoHcy increase Ki, suggesting that multiple
protein–ligand interactions are involved in binding and recogni-
tion. We found evidence of significant interactions made by both
methyltransferases to the ribose 2ꢀ-hydroxyl group, the adenine
N6 amino group and the methionine amino and carboxyl groups
of AdoHcy. Additionally we found that a bulky substituent
on the adenine N6 amino group, or even replacement of the
adenine with hypoxanthine or cytosine, do not preclude binding
of AdoHcy analogue 1, 2 and 3 to these enzymes.
The dissociation constant (Kd) of four analogues of AdoHcy
for M.HhaI has previously been determined14 with a view to
understanding the role of the methionine moiety in protein–
cofactor interaction in DNA methyltransferases and other
methyltransferases. Two of these analogues were also tested in
this study (compounds 6 and 7).
In the previous study of M.HhaI compounds 6 and 7 were
found to have Kd values 78 and 17 fold higher than that of
AdoHcy respectively,14 whereas we observed that Ki values 1100
and 850 fold higher than that of AdoHcy. The conclusion of the
previous study is that the adenosyl moiety acts as a “molecular
anchor” for cofactor binding, whereas our results emphasise
the involvement of multiple groups on both the adenosyl and
methionine moieties in cofactor binding.
Kd values were determined in the absence of DNA, by
measuring the quenching of tryptophan fluorescence upon
AdoHcy binding.14 It has been previously reported that AdoHcy
binds with high affinity to the M.HhaI–DNA complex but not
to the free enzyme or the binary product complex (enzyme–
methylated DNA),13 in fact AdoMet is bound in a different
orientation in the binary M.HhaI–AdoMet complex compared
to the ternary complex of M.HhaI, AdoMet and DNA.25,26
Our values of Ki are measurements of the competition of each
analogue with AdoMet for occupation of the cofactor binding
site in the catalytically competent enzyme–DNA complex. It is
likely that our choice to measure Ki instead of Kd accounts for
the differences in our estimated DDG values.
Crystal structures of HhaI and HaeIII in complex with
AdoHcy indicate hydrogen bonds with each of the nucleobase,
sugar and amino acid moieties, as shown in Fig. 4. However,
we compared the data from 31 crystal structures of methyl-
transferases in complex with AdoHcy or analogues and found
that the only conserved hydrogen bonds are between acidic side
chains and the adenine N6 amino group and the ribose hydroxyls
(Table S1†). Hydrogen bonds to the nitrogens N1 and N3 in the
adenine ring were made by the main chain amino groups of
hydrophobic residues, and were only predicted in about half
the methyltransferases studied. Hydrogen bonds to N7 were
predicted in two structures. In contrast to the interactions with
the adenosine moiety, the residues making hydrogen bonds to
the methionine moiety are not conserved. Interactions with both
the amino and carboxyl functional groups are predicted in 27
of the structures, yet these hydrogen bonds are a mixture of
main chain, side chain and water mediated contacts made by
a wide variety of residues from two regions of the consensus
fold or from outside the consensus fold. Examination of the
phylogenetic data alone might suggest that the interactions with
the methionine moiety are less significant than those with other
parts of the molecule. Our results indicate that, despite the
lack of conservation of hydrogen bond-forming residues, the
methionine functional groups are an important determinant of
cofactor binding by both M.HhaI and M.HaeIII. We observe
that modification of either the amino or the carboxyl group of
AdoHcy leads to an increase in Ki comparable to the change
caused by modification of the 2ꢀ-hydroxyl or N6 amino groups.
Our findings suggest that recognition of AdoMet by M.HhaI
and M.HaeIII relies on multiple interactions with the
adenine, ribose and methionine moieties. This pattern of
methyltransferase-cofactor recognition is likely to be conserved
amongst other cytosine C5 DNA methyltransferases and may
also be conserved in other proteins with the type I AdoMet-
dependent methyltransferase fold. However the residues in-
volved in cofactor binding are highly variable, particularly
residues predicted to contact the methionine moiety, therefore
the strength of interactions with this moiety may differ signifi-
cantly. These AdoHcy analogues should prove useful tools for
further study of methyltransferase function.
Experimental
General methods
1H-NMR spectra were obtained on a Bruker DRX-300 spec-
trometer in d6-DMSO unless otherwise stated. Mass spectra
were recorded on a Bruker FTICR Bioapex II instrument.
Ultraviolet spectra were recorded on a Perkin Elmer Lambda 40
spectrophotometer in 10% aqueous methanol unless otherwise
stated. TLC was carried out on pre-coated F254 or Merck RP-
18 F254s silica plates, and column chromatography with Merck
Kieselgel 60, or Merck RP-18 silica. Reactions were worked up
as follows, unless otherwise stated. Reaction mixtures were evap-
orated to dryness and the products dissolved in chloroform and
washed with saturated aqueous sodium bicarbonate solution.
The organic fractions were combined and dried over anhydrous
sodium sulfate, filtered and then evaporated to dryness. HPLC
was carried out using a Phenomenex Luna 10 lm C-18 reverse
phase column. Unless otherwise stated, HPLC conditions are
buffer A, 0.1% aqueous TFA; buffer B, 0.1% aqueous TFA, 25%
MeCN. 0% to 100% buffer B over 20 minutes, 100% buffer B 10
minutes.
N6 -Benzoyl-(O-2ꢀ,O-3ꢀ -isopropylidene-5ꢀ -O-toluenesulfonyl)-
adenosine 8. To
a
solution of N6-benzoyl-O-2ꢀ,O-3ꢀ-iso-
propylidene-adenosine (11.5 g, 28 mmol) in pyridine (150 cm3)
was added toluene sulfonyl chloride (10.7 g, 56 mmol) and the
solution stirred at room temperature overnight. The reaction
was quenched with methanol, evaporated and worked up
to give a brown gum. This was chromatographed on silica
(CH2Cl2–(0–5%) MeOH) to give an off-white foam. Yield
14.89 g, 94%. Spectra were as described.27
N6-Benzoyl-(O-2ꢀ,O-3ꢀ -isopropylidene-5ꢀ -deoxy-5ꢀ -iodo)-ade-
nosine 9. A solution of 8 (14.5 g, 25.6 mmol) in acetone
(200 cm3) containing sodium iodide (9.6 g, 64 mmol) was heated
at reflux overnight. The solution was evaporated, dissolved
in ethyl acetate and washed (2 M Na2S2O3) and evaporated
to give a pale brown foam. This was chromatographed on
silica (CH2Cl2–(0–5%) MeOH) to give an off-white foam. Yield
9.23 g, 69%. dH 1.34, 1.55 (6H, 2 × s, isopropylidene CH3),
3.33–3.49 (2H, m, H5ꢀ, H5ꢀꢀ), 4.35–4.40 (1H, m, H4ꢀ), 5.04–5.07
(1H, m, H3ꢀ), 5.57–5.60 (1H, m, H2ꢀ), 6.37 (1H, d, J 2, H1ꢀ),
7.52–7.67 (3H, m, Ph–CH), 7.99–8.09 (2H, m, Ph–CH), 8.68
(1H, s, H2), 8.79 (1H, s, H8), 11.25 (1H, s, NH). UV kmax/nm 281
Fig. 4 Schematic diagram of the proposed hydrogen bonds be-
tween M.HhaI and AdoHcy. M.HhaI amino acids are written in
grey, with the equivalent residues in M.HaeIII below. Water-mediated
hydrogen bonds are omitted for clarity.
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1
1 5 7

View Article Online
(18 400), kmin/nm 245; pH 1, kmax/nm 290 (21 800), kmin/nm 240;
pH 12, kmax/nm 302 (12300), kmin/nm 258. m/z 544 (70%, M +
Na)+, 522 (15%, M + H)+, 394 (100%, M − I)+. Accurate mass
measurement on C20H20N5O4INa gives 544.04600, deviation
0.36 ppm.
S-(O-2ꢀ,O-3ꢀ-Isopropylidene-inosin-5ꢀ-yl)-L-homocysteine 11.
Prepared by a modified procedure described by Grøtli et al.19 To
a solution of (O-2ꢀ,O-3ꢀ-isopropylidene-5ꢀ-O-toluenesulfonyl)-
inosine28 (0.6 g, 1.1 mmol) in dichloromethane (25 cm3) was
added DMAP (66 mg, 0.54 mmol) and triethylamine (1.5 cm3,
11 mmol) followed by tert-butyldiphenylchlorosilane (0.71 cm3,
2.77 mmol) and the solution stirred at room temperature for
48 hours. TLC showed complete conversion. The reaction was
quenched with methanol, evaporated, and the product coevapo-
rated twice with toluene to give 6-O-(tert-butyldiphenylsilyl)-(O-
2ꢀ,O-3ꢀ-isopropylidene-5ꢀ-O-toluenesulfonyl)-inosine as a yellow
foam which was used without purification.
Homocystine (0.4 g, 1.5 mmol) in liquid ammonia (approxi-
mately 50 cm3) was converted to its di-sodium salt as described
above. Once the blue colour had dissipated the reaction was
cooled to −78 ◦C and the above silyl protected inosine derivative
(0.4 g, 0.5 mmol) added, and the solution stirred at −78 ◦C for
6 hours. The reaction was allowed to warm to room temperature.
Under a nitrogen atmosphere ethanol (20 cm3) was added
followed by water (20 cm3) and the solution acidified (HOAc)
and stirred at room temperature overnight to remove the silyl
protection. The solution was evaporated, dissolved in water and
purified on RP-C18 silica eluting with water and then a gradient
of 0–10% acetonitrile in water to elute the product as an off-
white foam. Yield 128 mg, 53%. dH (D2O) 1.24, 1.45 (6H, 2
× s, isopropylidene CH3), 1.85–2.00 (2H, m, CH2CH2S), 2.49
(2H, t, J 7.2, CH2S), 2.64–2.67 (2H, m, H5ꢀ, H5ꢀꢀ), 3.68 (1H,
t, J 5.7, NH2CHCO2H), 4.25 (1H, br. s, H4ꢀ), 4.90 (1H, br. s,
H3ꢀ), 5.24–5.27 (1H, m, H2ꢀ), 5.98 (1H, s, H1ꢀ), 7.99 (1H, s,
H2), 8.05 (1H, s, H8). UV kmax/nm 249 (11 600), kmin/nm 222;
pH 12 kmax/nm 254 (12 100), kmin/nm 232. m/z 448 (70%, M +
Na)+, 399 (10%, M − CO)+. Accurate mass measurement on
C17H23N5SO6Na gives 448.12670, deviation −0.01 ppm.
S-Inosin-5ꢀ-yl-L-homocysteine (SIH) 2. A solution of 11
(100 mg, 0.24 mmol) in 5% aqueous TFA (5 cm3) was stirred
at room temperature for 5 hours. The solution was applied to
an RP-C18 column and washed with water to neutrality and
the product eluted with 10% aqueous acetonitrile to give a white
foam. Yield 63 mg. 69%. dH (D2O) 1.89–2.06 (2H, m, CH2CH2S),
2.55 (2H, t, J 7.2, CH2S), 2.79–2.93 (2H, m, H5ꢀ, H5ꢀꢀ), 3.69 (1H,
t, J 5.8, NH2CHCO2H), 4.15–4.17 (1H, m, H4ꢀ), 4.25–4.27 (1H,
m, H3ꢀ), 4.58–4.68 (under H2O peak, H2ꢀ), 5.90 (1H, d, J 4,
H1ꢀ), 8.02 (1H, s, H2), 8.14 (1H, s, H8). UV (H2O) kmax/nm
249 (12 000), kmin/nm 223; pH 12 kmax/nm 254 (12 500), kmin/nm
231. m/z 408.09 (100%, M + Na)+, 386.09 (20%, M + H)+.
Accurate mass measurement on C14H19N5SO6Na gives 408.0957,
deviation 0.72 ppm.
S-(O-2ꢀ,O-3ꢀ-Isopropylidene-N6-benzoyladenosin-5ꢀ-yl)-L-homo-
cysteine 10. Homocystine (2 g, 7.4 mmol) was dissolved in
liquid ammonia (ca. 100 cm3) and freshly cut sodium was
added until the deep blue colour remained for 20 minutes. The
blue colour was quenched by the addition of small portions of
ammonium chloride, and the reaction allowed to warm to room
temperature. Under a nitrogen atmosphere the residue was
dissolved in dry ethanol (15 cm3) and filtered through Celite,
and washed with further portions of dry ethanol (3 × 5 cm3).
To the solution was then added dry ether and the white solid
was filtered and dried in vacuo over phosphorus pentoxide.
The product is highly hygroscopic and was used immediately,
assuming 100% conversion to L-homocysteine di-sodium salt.
The di-sodium salt of L-homocysteine from above was dis-
solved in DMF (20 cm3) and water (10 cm3), and to this was
added potassium iodide (10 mg) and (9)27 (6.50 g, 12.5 mmol),
and the solution heated at 100 ◦C for 45 minutes. The solution
was neutralized (acetic acid) and evaporated to dryness to give
a yellow–brown gum, which, by reverse phase TLC, showed
two products. The mixture was chromatographed (Merck RP-
18 silica, water to 25% acetonitrile in water gradient). The
first product eluted in 10–15% MeCN, was shown to be the
de-benzoylated product, S-(O-2ꢀ,O-3ꢀ-isopropylidene-adenosin-
5ꢀ-yl)-L-homocysteine. Yield 1.25 g, 24%. dH 1.30, 1.51 (6H, 2
× s, isopropylidene CH3), 1.78–1.99 (2H, m, CH2CH2S), 2.57–
2.61 (2H, m, CH2S), 2.63–2.82 (2H, m, H5ꢀ, H5ꢀꢀ), 3.30–3.34
(1H, m, COCHNH2), 4.22–4.26 (1H, m, H4ꢀ), 4.96–4.99 (1H,
m, H3ꢀ), 5.46–5.49 (1H, m, H2ꢀ), 6.14–6.15 (1H, m, H1ꢀ), 7.39
(2H, br. s, NH2), 8.17 (1H, s, H2), 8.35 (1H, s, H8). UV kmax/nm
259 (13 000), kmin/nm 227; pH 1 kmax/nm 257 (13 700), kmin/nm
228. m/z 425 (100%, M + H)+, 381 (10%, M − CO2)+. Accurate
mass measurement on C17H24N6SO5 gives 425.1603, deviation
−0.98 ppm.
The second product, eluting in 20–25% MeCN, was shown
to be the title product as a white foam. Yield 3.75 g, 57%. dH
(D2O) 1.34, 1.55 (6H, 2 × s, isopropylidene CH3), 1.73–1.99
(2H, m, CH2CH2S), 2.58–2.63 (2H, m, CH2S), 2.70–2.83 (2H,
m, H5ꢀ, H5ꢀꢀ), 3.25–3.29 (1H, m, COCHNH2), 4.28–4.30 (1H,
m, H4ꢀ), 5.02–5.04 (1H, m, H3ꢀ), 5.56–5.58 (1H, m, H2ꢀ), 6.29
(1H, d, J 1.9, H1ꢀ), 7.44–7.65 (3H, m, Ph–CH), 7.92–8.08 (2H,
m, Ph–CH) 8.69 (1H, s, H2), 8.79 (1H, s, H8). UV kmax/nm
281 (22 000), kmin/nm 246; pH 1, kmax/nm 288 (23 300), kmin/nm
263; pH 12, 303 (13 100), kmin/nm 260. m/z 529.19 (70%, M +
H)+, 439 (10%, M − COPh)+, 412 (10%, M − COPh − CO)+.
Accurate mass measurement on C24H29N6SO6 gives 529.18570,
deviation −2.330 ppm.
S-(N6-Benzoyladenosin-5ꢀ-yl)-L-homocysteine 1. A solution
of 10 (150 mg, 0.28 mmol) in 5% aqueous TFA (10 cm3) was
stirred at room temperature for 6 hours. The reaction was
lyophilised to give a white solid, which was chromatographed
(RP-C18) eluting with H2O then 0–15% MeCN in H2O. Yield
97 mg, 70%. dH (D2O) 1.75–2.04 (2H, m, CH2CH2S), 2.59 (2H,
t, J 7.7, CH2S), 2.79–2.94 (2H, m, H5ꢀ, H5ꢀꢀ), 3.27–3.33 (1H,
m, NH2CHCO2H), 4.03–4.08 (1H, m, H4ꢀ), 4.15–4.21 (1H, m,
H3ꢀ), 4.78–4.81 (1H, m, H2ꢀ), 5.78 (1H, br, OH), 6.01 (1H, d,
J 5.9, H1ꢀ), 7.51–7.66 (3H, m, Ph–H), 8.00–8.03 (2H, m, Ph–
H), 8.68 (1H, s, H2), 8.75 (1H, s, H8), 11.20 (1H, br, NH). UV
(H2O) kmax/nm 281 (19 100), kmin/nm 245; pH 1, kmax/nm 290
(23 600), kmin/nm 242; pH 12, kmax/nm 301 (12 800), kmin/nm
258. e260 (lM) 11.2, e280 (lM) 19. m/z 511.14 (50%, M +
Na)+, 489.17 (40%, M + H)+, 425 (5%, M − Ph)+, 399 (5%, M −
CoPh)+. Accurate mass measurement on C21H24N6SO6Na gives
511.13990, deviation 4.49 ppm.
S-Cytid-5ꢀ-yl-L-homocysteine (SCH) 3. The title compound
was prepared from 5ꢀ-chloro-5ꢀ-deoxycytidine according to
Ramalingam and Woodward.20 The title compound has been
reported,16,18,20 but no characterisation provided. dH (D2O) 1.95–
2.13 (2H, m, CH2CH2S), 2.61 (2H, t, J 7.6, CH2S), 2.76–2.95
(2H, m, H5ꢀ, H5ꢀꢀ), 3.71 (1H, t, J 6.5, NH2CHCO2H), 4.00–
4.09 (2H, m, H4ꢀ, H3ꢀ), 4.19–4.23 (1H, m, H2ꢀ), 5.72 (1H, d,
J 3.9, H1ꢀ), 5.92 (1H, d, J 7.6, H6), 7.58 (1H, d, J 7.6, H5).
UV (H2O) kmax/nm 270 (7900), kmin/nm 250; pH 1 kmax/nm 279
(11 250), kmin/nm 240; pH 12 kmax/nm 272 (7800), kmin/nm 251.
m/z 383 (100%, M + Na)+. Accurate mass measurement on
C13H20N4SO6Na 383.09980, deviation −0.93 ppm.
N6-Benzoyl-(2-deoxy-b-D-ribofuranosyl-5-deoxy-5-iodo)-adeno-
sine 15. A solution of N6-benzoyl-(2-deoxy-b-D-ribofuranosyl-
5-toluenesulfonyl)-adenosine29 (0.49 g, 0.96 mmol) in acetone
(50 cm3) containing sodium iodide (0.72 g, 4.8 mmol) was
heated at reflux overnight. The solution was evaporated,
dissolved in ethyl acetate and washed (2 M Na2S2O3) and
evaporated to a pale brown foam. This was chromatographed
on silica (CH2Cl2–(0–5%) MeOH) to give an off-white foam.
Yield 0.32 g, 71%. dH 2.82–2.90 (2H, m, H2ꢀ, H2ꢀꢀ), 4.00–4.02
1 5 8
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1

View Article Online
(1H, m, H4ꢀ), 4.20–4.33 (2H, m, H5ꢀ, H5ꢀꢀ), 4.47 (1H, br. s, H3ꢀ),
5.56 (1H, d, J 3.7, OH), 6.43 (1H, t, J 6.5, H1ꢀ), 7.53–7.93 (5H,
m, Ph–H), 8.53 (1H, s, H2), 8.66 (1H, s, H), 811.23 (1H, s, NH).
m/z 466 (M + H)+.
(2H, m, CH2S), 2.76–2.79 (2H, m, H5ꢀ, H5ꢀꢀ), 3.60 (4H, s,
2 × CH2CO), 4.11–4.31 (5H, m, NHCHCO2H, CO·OCH2,
CO·OCH2CH, H4ꢀ), 5.01–5.03 (1H, m, H3ꢀ), 5.54–5.56 (1H, m,
H2ꢀ), 6.30 (1H, s, H1ꢀ), 7.29–8.05 (13H, m, Ph–CH, Fmoc-CH),
8.67 (1H, s, H2), 8.78 (1H, s, H8), 11.2 (1H, s, NH). UV kmax/nm
270 (36 200), 209 (69 000), kmin/nm 245; pH 1, kmax/nm 281
(36 600), kmin/nm 245; pH 12, kmax/nm 300 (21 100), 255 (31 300),
kmin/nm 285, 240. m/z 870 (100%, M + Na)+, 848 30%, M +
H)+. Accurate mass measurement on C43H41N7SO10Na gives
870.2533, deviation −2.60 ppm.
S-(O-2ꢀ,O-3ꢀ -Isopropylideneadenosin-5ꢀ -yl)-L-homocysteine
methylamide 14. To a solution of 13 (0.5 g, 0.59 mmol) in
DMF (20 cm3) was added methylamine (40% solution in H2O,
10 cm3) and the solution stirred at room temperature overnight.
The solution was evaporated and the product chromatographed
on silica (CH2Cl2–(0–20%) MeOH) to give a white foam. Yield
0.16 g, 62%. dH 1.32, 1.52 (6H, 2 × s, isopropylidene CH3),
1.45–1.89 (2H, m, CH2CH2S), 2.13–2.29 (2H, m, CH2S), 2.55
(3H, br. d, J 3.5, CH3NH), 2.65–2.81 (2H, m, H5ꢀ, H5ꢀꢀ),
3.10–3.14 (1H, m, COCHNH2), 4.20–4.25 (1H, m, H4ꢀ),
4.97–5.00 (1H, m, H3ꢀ), 5.47–5.50 (1H, m, H2ꢀ), 6.14–6.16 (1H,
m, H1ꢀ), 7.35 (2H, s, NH2), 7.77 (2H, br, NH2), 8.16 (1H, s, H2),
8.32 (1H, s, H8). UV kmax/nm 260 (10 100), kmin/nm 227; pH 1,
kmax/nm 258 (10 000), kmin/nm 227. m/z 460.2 (100%, M +
Na)+, 438 (80%, M + H)+. Accurate mass measurement on
C18H27N7SO4Na gives 460.17280, deviation −3.31 ppm.
S-Adenosin-5ꢀ-yl-L-homocysteine methylamide 4. A solution
of 14 (140 mg, 0.32 mmol) in 5% aqueous TFA (5 cm3) was stirred
at room temperature overnight. The solution was evaporated,
and the product chromatographed (RP-C18) eluting with H2O
then 0–5% MeCN in H2O to give a white powder. Yield 77 mg,
61%. dH (D2O) 1.88–1.95 (2H, m, CH2CH2S), 2.39–2.54 (2H,
m, CH2S), 2.55 (3H, s, CH3N), 2.75–2.90 (2H, m, H5ꢀ, H5ꢀꢀ),
3.84 (1H, t, J 6.8, COCHNH2), 4.13–4.18 (1H, m, H4ꢀ), 4.23–
4.26 (1H, m, H3ꢀ), 4.63–4.68 (1H, m, H2ꢀ), 5.86 (1H, d, J 4.6,
H1ꢀ), 7.98 (1H, s, H2), 8.13 (1H, s, H8). UV (H2O) kmax/nm 260
(11 200), kmin/nm 228. m/z 398.16 (80%, M + H)+. Accurate
mass measurement on C15H23N7SO4 gives 398.1610, deviation
−0.13 ppm.
S-(N6-Benzoyl-2ꢀ -deoxy-b-D-ribofuranosyladenosin-5ꢀ -yl)-L-
homocysteine 16. Homocystine (0.25 g, 0.09 mmol) was
converted to its disodium salt as described above. The salt
was added to a solution of 15 (0.4 g, 0.086 mmol) in DMF
◦
(10 cm3) and water (10 cm3) and the solution stirred at 100 C
for 1 hour. The solution was cooled, neutralised (HOAc) and
evaporated, and the residue chromatographed (RP-C18, H2O)
and the product eluted in a gradient of 0–10% MeCN in H2O
to give a white powder. Yield 0.23 g, 57%. dH (D2O) 1.85–2.04
(2H, m, H2ꢀ, H2ꢀꢀ), 2.30–2.59 (4H, m, SCH2CH2), 2.61–2.79
(2H, m, H5ꢀ, H5ꢀꢀ), 3.62–3.66 (1H, m, NCH), 4.02 (1H, br. s,
H4ꢀ), 4.29 (1H, br. s, OH), 4.40 (1H, br. s, H3ꢀ), 6.22 (1H,
t, J 5.7, H1ꢀ), 7.11–7.31 (3H, m, Ph–H), 7.53–7.56 (2H, m,
Ph–H), 8.26 (1H, s, H2), 8.34 (1H, s, H8). UV kmax/nm 281
(21 000), kmin/nm 245; pH 1, kmax/nm 290 (25 400), 253 (10 800),
kmin/nm 261, 240; pH 12, kmax/nm 301 (13 900), kmin/nm 258.
m/z 473.2 (100%, M + H)+, 408 (10%, M − Ph)+. Accurate
mass measurement on C21H25N6SO5 gives 473.16010, deviation
−1.30 ppm.
S-(2ꢀ-Deoxy-b-D-ribofuranosyladenosin-5ꢀ-yl)-L-homocysteine
5. A solution of 16 (130 mg, 0.28 mmol) in concentrated
ammonia solution (10 cm3) was stirred at room temperature
overnight. The solution was evaporated and the product
chromatographed (RP-C18, H2O) and the product eluted in
a gradient of 0–10% MeCN in H2O to give a white powder.
1
Yield 90 mg, 89%. The H-NMR spectrum was as previously
described.21 UV (H2O) kmax/nm 260 (15 200), kmin/nm 227; pH 1
kmax/nm 258 (14 900), kmin/nm 229; pH 12 kmax/nm 260 (14 300),
kmin/nm 234. m/z 369.1 (40%, M + H)+, 391.1 (100%, M +
Na)+. Accurate mass measurement on C14H20N6SO4Na gives
391.11620, deviation −0.71 ppm.
N-a-Fmoc-S-(O-2ꢀ,O-3ꢀ -isopropylidene-N6-benzoyladenosin-
5ꢀ-yl)-L-homocysteine 12. To a solution of 10 (2.50 g, 4.7 mmol)
in an ice-cold solution of dioxane (25 cm3) and water (25 cm3)
containing sodium carbonate (4.3 g, 15 mmol) was added a
solution of N-(9-fluorenylmethoxycarbonyloxy)-succinimide
(3.35 g, 9.9 mmol) in dioxane (20 cm3) dropwise over 30
minutes. The solution was then stirred at room temperature for
2 hours. The reaction was then neutralised (acetic acid) and
concentrated and then worked up as described to give a yellow
foam. This was chromatographed on silica (CH2Cl2–(0–5%)
MeOH) to give an off-white foam. Yield 3.33 g, 94%. dH
1.31, 1.51 (6H, 2 × s, isopropylidene CH3), 1.80–1.90 (2H,
m, CH2CH2S), 2.31–2.34 (2H, m, CH2S), 2.56–2.81 (2H, m,
H5ꢀ, H5ꢀꢀ), 3.84–3.87 (1H, m, NHCHCO2H), 4.19–4.29 (4H,
m, CO.OCH2, CO.OCH2CH, H4ꢀ), 5.00–5.02 (1H, m, H3ꢀ),
5.54–5.56 (1H, m, H2ꢀ), 6.28 (1H, s, H1ꢀ), 7.26–8.06 (13H, m,
Ph–CH, Fmoc-CH), 8.66 (1H, s, H2), 8.77 (1H, s, H8), 12.2
(1H, br. s, NH). UV kmax/nm 266 (30 200), kmin/nm 234; pH 12,
kmax/nm 255 (28 100), kmin/nm 240. m/z 773.23 (80%, M +
Na)+. Accurate mass measurement on C39H38N6SO8Na gives
773.23720, deviation 0.28 ppm.
DNA substrates. Haesub, a synthetic 30 base pair DNA
substrate for M.HaeIII which contains a single unmethylated
recognition site (GGCC), was prepared by annealing comple-
mentary oligonucleotides Haesub1, 5ꢀ-TTCGAGAAGCTGA-
GGCCGGCGTACCTGGAG-3ꢀ, and Haesub2, 5ꢀ-CTCCAG-
GTACGCCGGCCTCAGCTTCTCGAA-3ꢀ, as described.30
Hhasub, a similar 30 base pair substrate containing a sin-
gle M.HhaI recognition site (GCGC), was prepared by anneal-
ing Hhasub1, 5ꢀ-TTCGAGAAGCTGAGCGCGGCGTACC-
TGGAG-3ꢀ, and Hhasub2, 5ꢀ-CTCCAGGTACGCCGCGCTC-
AGCTTCTCGAA-3ꢀ, as above.
For electrophoretic mobility shift assays a substrate with a
single hemimethylated HaeIII site was prepared by annealing
complementary oligonucleotides Gsub (5ꢀ-TTCGAGTGAC-
TGAGGCCGGTGTACCTGGAG-3ꢀ) and Gsub-meth (5ꢀ-CT-
CCAGGTACACCGGMCTCAGTCACTCGAA-3ꢀ), where M
represents C5 methylcytosine. The substrate was 32P end labeled.
N-a-Fmoc-S-(O-2ꢀ,O-3ꢀ -isopropylidene-N6-benzoyladenosin-
5ꢀ-yl)-L-homocysteine N-hydroxysuccinimide ester 13. To
a solution of the above acid (12) (2.80 g, 3.73 mmol) in
dichloromethane (50 cm3) was added N-hydroxysuccinimide
(0.86 g, 7.47 mmol), followed by dicyclohexylcarbodiimide
(2.31 g, 11.2 mmol) and the solution stirred at room temperature
overnight. The solid was filtered through Celite and the liquors
worked up as described to give a white foam, which was
chromatographed on silica (CH2Cl2–(0–2%) MeOH) to give
a white foam. Yield 2.29 g, 72%. dH 1.33, 1.54 (6H, 2 × s,
isopropylidene CH3), 1.84 (2H, br. s, CH2CH2S), 2.49–2.58
Methyltransferase assays. Filter binding assays monitored
the incorporation of tritium labelled methyl groups into
DNA. DNA and S-adenosyl-L-[methyl-3H]methionine, (20 Ci
mmol−1; Amersham-Pharmacia) were added to M.HhaI reac-
tion buffer (50 mM Tris–HCl (pH 7.5), 10 mM EDTA, 5 mM
b-mercaptoethanol) or M.HaeIII reaction buffer (50 mM NaCl,
50 mM Tris–HCl (pH 8.5), 10 mM dithiothreitol). Assays were
performed at 37 ◦C, started by the addition of methyltransferase,
and quenched by the addition of unlabelled AdoMet to a final
concentration of 80 lM. Quenched reactions were either spotted
onto DE81 filters (Whatman) and processed as described,30 or
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1
1 5 9

View Article Online
spotted onto Multiscreen DE plates, and washed three times
with 200 ll ammonium carbonate (0.2 M), and once with 100 ll
ethanol, before punching out individual filters for counting. The
background level of radioactivity was determined by performing
blank reactions (without enzyme) and was subtracted from all
readings. Enzymes were obtained from New England Biolabs
at 10 units ll−1 (one unit is the amount of enzyme required to
protect 1 lg k DNA in one hour at 37 ◦C in a 10 ll reaction
mixture) and diluted in the appropriate methylation reaction
buffer, supplemented with 1 mg ml−1 bovine serum albumin.
The concentration of M.HaeIII active sites was estimated by
electrophoretic mobility shift assay as described previously.31
Reactions contained 1–150 nM DNA, 2–5 nM enzyme, 200 lM
AdoHcy and 10 mM EDTA in M.HaeIII reaction buffer.
Reactions were incubated at 37 ◦C for 30 minutes before addition
of glycerol to 5% of the final volume and electrophoresis
using 10% TBE gels (Novex, Invitrogen). Gels were fixed and
dried according to the manufacturer’s instructions, exposed for
18 hours and scanned using a phosphorimager (Typhoon 8600,
Molecular Dynamics). Results were analysed using Imagequant
software (Molecular Dynamics).
same way, using final concentrations of 0.067 units ll−1 M.HhaI,
1.25 lM Hhasub, 200 nM AdoMet and 10 nM–15 lM AdoHcy
analogue. Reactions were quenched after eight minutes.
Plots of reaction velocity versus inhibitor concentration were
fit to the curve in eqn. 2 using the Levenberg–Marquardt
algorithm, where [I] is the concentration of AdoHcy or ana-
logue. Inhibition is assumed to be competitive with AdoMet,
as observed for inhibition of M.HhaI by AdoHcy (13), DDG
was calculated from the inhibition constants for AdoHcy and
analogues using eqn. 3.
DDG = −RTln(Kianalogue/ KiAdoHcy
)
(3)
Acknowledgements
We wish to acknowledge the use of the EPSRC’s Chemical
Database Service at Daresbury.32
References
1 S-Adenosylmethionine-dependent methyltransferases: structures and
functions, X. Cheng and R. M. Blumenthal, eds., World Scientific,
Singapore, 1999.
2 G. L. Cantoni, Biological methylation: selected aspects, Annu. Rev.
Biochem., 1975, 44, 435–451.
3 H. L. Schubert, R. M. Blumenthal and X. Cheng, Many paths to
methyltransfer: a chronicle of convergence, Trends Biochem. Sci.,
2003, 28, 329–335.
4 J. L. Martin and F. M. McMillan, SAM (dependent) I AM: the S-
adenosylmethionine-dependent methyltransferase fold, Curr. Opin.
Struct. Biol., 2002, 12, 783–793.
Determination of KmAdoMet for M.HaeIII. Time course assays
were started by the addition of M.HaeIII to a final concentration
of 5.9 nM. The concentration of Haesub, 1.25 lM, was more
than a hundred fold higher than KmDNA, and the concentration
of AdoMet ranged from 25 to 2000 nM. Aliquots of the
reaction mixture were removed at two minute intervals and
quenched as described. Under these conditions the incorpo-
ration of methyl-3H into DNA was linear for at least ten
minutes (Fig. 2A). A plot of reaction velocity versus AdoMet
concentration was fit to the Michaelis–Menten model (eqn. 1),
using the Levenberg–Marquardt algorithm, as implemented in
Kaleidagraph (Synergy Software, Reading, PA, USA). [S]0 is
the initial concentration of AdoMet and Km is the apparent
Michaelis–Menten constant of M.HaeIII for AdoMet.
5 A. K. Ghosh and W. Liu, Total synthesis of (+)-sinefungin, J. Org.
Chem., 1997, 62, 2299.
6 A. K. Ghosh and W. M. Liu, Total synthesis of (+)-sinefungin, J. Org.
Chem., 1996, 61, 6175–6182.
7 S. Hanessian and P. W. M. Sgarbi, Design and synthesis of mimics of
S-adenosyl-L-homocysteine as potential inhibitors of erythromycin
methyltransferases, Bioorg. Med. Chem. Lett., 2000, 10, 433–437.
8 D. C. Wahnon, V. K. Shier and S. J. Benkovic, Mechanism-based
inhibition of an essential bacterial adenine DNA methyltransferase:
Rationally designed antibiotics, J. Am. Chem. Soc., 2001, 123, 976–
977.
9 R. M. Kagan and S. Clarke, Widespread occurrence of 3 se-
quence motifs in diverse S-adenosylmethionine-dependent methyl-
transferases suggests a common structure for these enzymes, Arch.
Biochem. Biophys., 1994, 310, 417–427.
10 S. Kumar, X. D. Cheng, S. Klimasauskas, S. Mi and J. Posfai, The
DNA (cytosine-5) methyltransferases, Nucleic Acids Res., 1994, 22,
1–10.
11 K. M. Reinisch, L. Chen, G. L. Verdine and W. N. Lipscomb, The
crystal-structure of HaeIII methyltransferase covalently complexed
to DNA - an extrahelical cytosine and rearranged base-pairing, Cell,
1995, 82, 143–153.
Determination of KmDNA and Kcat for M.HaeIII. Time course
assays were started by the addition of M.HaeIII to a final con-
centration of 0.4 nM. The AdoMet concentration of 1 lM was
over three fold greater than Km, and the final DNA concentration
ranged from 1 to 30 nM. Aliquots of the reaction mixture were
removed at one minute intervals for six minutes and quenched as
described. Under these conditions the incorporation of methyl-
3H into DNA was linear. A plot of reaction velocity versus
AdoMet concentration was fit to the Michaelis–Menten model
(eqn. 1), as above, where [S]0 is the initial concentration of DNA
and Km is the apparent Michaelis–Menten constant of M.HaeIII
for the DNA substrate Haesub.
AdoMet
12 A. P. Dong, J. A. Yoder, X. Zhang, L. Zhou and T. H. Bestor,
Structure of human DNMT2, an enigmatic DNA methyltransferase
homolog that displays denaturant-resistant binding to DNA, Nucleic
Acids Res., 2001, 29, 439–448.
Determination of Km
for M.HhaI. Time course assays
were started by the addition of M.HhaI to a final concentration
of 0.067 units ll−1. The concentration of Hhasub was 0.5 lM,
a hundred fold higher than KmDNA, and the concentration of
AdoMet ranged from 40 to 1000 nM. Four aliquots of the
reaction mixture were removed at two minute intervals and
quenched as described. Under these conditions the incorpo-
ration of methyl-3H into DNA was linear. A plot of reaction
velocity versus AdoMet concentration was fit to the Michaelis–
Menten model (eqn. 1), as above. [S]0 is the initial concentration
of AdoMet and Km is the apparent Michaelis–Menten constant
of M.HhaI for AdoMet.
13 J. C. Wu and D. V. Santi, Kinetic and catalytic mechanism of HhaI
methyltransferase, J. Biol. Chem., 1987, 262, 4778–4786.
14 M. Pignot, G. Pljevaljcic and E. Weinhold, Efficient synthesis of S-
adenosyl-L-homocysteine natural product analogues and their use to
elucidate the structural determinant for cofactor binding of the DNA
methyltransferase M HhaI, Eur. J. Org. Chem., 2000, 3, 549–555.
15 J. Baddiley and G. A. Jamieson, Synthesis of S-(5ꢀ-deoxyadenosine-
5ꢀ)-homocysteine, a product from enzymatic methylations involving
“active methionine”, J. Am. Chem. Soc., 1955, 1, 1085–1089.
16 R. T. Borchardt, J. A. Huber and Y. S. Wu, Potential inhibitors
of S-adenosylmethionine-dependent methyltransferases. 4. Further
modifications of the amino acid and base portions of S-adenosyl-L-
homocysteine, J. Med. Chem., 1976, 19, 1094–1103.
Determination of Ki for AdoHcy and analogues. The initial
rate of methyltransfer by M.HaeIII was measured at a range of
concentrations of the eight inhibitors using single time point
assays. Final concentrations of 5.9 nM M.HaeIII, 1.25 lM
Haesub, 200 nM AdoMet and 10 nM–15 lM AdoHcy analogue
were used. Reactions were started with the addition of enzyme,
quenched after ten minutes and processed as above. The inhibi-
tion of M.HhaI by AdoHcy and analogues was measured in the
17 K. Ramalingam and R. W. Woodward, Preparation of S-(N6,N6-
dimethyladenosyl)-L-methionine, Carbohydr. Res., 1985, 142, 123–
126.
18 R. T. Borchardt, J. A. Huber and Y. S. Wu, Potential inhibitors of S-
adenosylmethionine-dependent methyltransferases. 2. Modification
of the base portion of S-adenosylhomocysteine, J. Med. Chem., 1974,
17, 868–873.
1 6 0
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1

View Article Online
19 M. Grøtli, M. Douglas, B. Beijer, R. G. Garcia and R. Eritja,
Protection of the guanine residue during synthesis of 2ꢀ-O-alkyl-
guanosine derivatives, J. Chem. Soc., Perkin Trans. 1, 1997, 18, 2779–
2788.
20 K. Ramalingam and R. W. Woodward, An improved synthesis of S-
adenosylhomocysteine and related compounds, J. Org. Chem., 1984,
49, 1291–1293.
26 S. Klimasauskas, S. Kumar, R. J. Roberts and X. D. Cheng, HhaI
methyltransferase flips its target base out of the DNA helix, Cell,
1994, 76, 357–369.
27 A. R. Maguire, W. D. Meng, S. M. Roberts and A. J. Willetts,
Synthetic approaches towards nucleocidin and selected analogs -
anti-HIV activity in 4ꢀ-fluorinated nucleoside derivatives, J. Chem.
Soc., Perkin Trans. 1, 1993, 15, 1795–1808.
21 R. T. Borchardt and Y. S. Wu, Potential inhibitors of S-
adenosylmethionine-dependent methyltransferases. 3. Modifications
of the sugar portion of S-adenosylhomocysteine, J. Med. Chem.,
1975, 18, 300–304.
22 W. M. Lindstrom, J. Flynn and N. O. Reich, Reconciling structure
and function in HhaI DNA cytosine-C-5 methyltransferase, J. Biol.
Chem., 2000, 275, 4912–4919.
23 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., 2004, 17, 3–
11.
28 A. Hampton, M. Bayer, V. S. Gupta and S. Y. Chu, Synthesis of
5ꢀ-substituted derivatives of inosine, J. Med. Chem., 1968, 11, 1229–
1232.
29 A. Romieu, D. Gasparutto, D. Molko and J. Cadet, Site-specific in-
troduction of (5ꢀ S)-5ꢀ,8-cyclo-2ꢀ-deoxyadenosine into oligodeoxyri-
bonucleotides, J. Org. Chem., 1998, 63, 5245–5249.
30 H. M. Cohen, D. S. Tawfik and A. D. Griffiths, Promiscuous methy-
lation of non-canonical DNA sites by HaeIII methyltransferase,
Nucleic Acids Res., 2002, 30, 3880–3885.
31 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.,
2002, 276, 20924–20934.
24 A. Fersht, Structure and mechanism in protein science, Freeman, San
Francisco, 1999.
25 X. Cheng, S. Kumar, S. Klimasauskas and R. J. Roberts, Crystal-
structure of the HhaI DNA methyltransferase, Cold Spring Harbor
Symp. Quant. Biol., 1993, 58, 331–338.
32 D. A. Fletcher, R. F. McMeeking and D. Parkin, The United
Kingdom Chemical Database Service, J. Chem. Inf. Comput. Sci.,
1996, 36, 746–749.
O r g . B i o m o l . C h e m . , 2 0 0 5 , 3 , 1 5 2 – 1 6 1
1 6 1