1H and 31P NMR study of ADP and GDP spontaneous synthesis from 2-methylimidazole-activated AMP and GMP nucleotides and inorganic phosphate

Article abstract of DOI:10.1039/a900884e

¹H and ³¹P NMR data show that adenosine 5′-(2-methylimidazol-1-ylphosphonate) (2-MeImpA) and guanosine 5′-(2-methylimidazol-1-ylphosphonate) (2-MeImpG), known as possible prebiotic precursors of polynucleotides, produce the corresponding diphosphonucleotides in sodium phosphate solution at pD 7.6. Phosphate ions also enhance the hydrolysis of these molecules by activating a water molecular associated as a weak complex with the phosphoryl moiety of 2-MeImpA and 2-MeImpG. A kinetic study was done by quantitative ¹H NMR spectroscopy and mechanistic hypotheses were tested by semiempirical PM3 modelling.

Full text of DOI:10.1039/a900884e

View Article Online / Journal Homepage / Table of Contents for this issue
31
1H and P NMR study of ADP and GDP spontaneous synthesis
from 2-methylimidazole-activated AMP and GMP nucleotides and
inorganic phosphate
José Pereira* and Ana Cadete
CEQUP/Instituto de Ciências Biomédicas de Abel Salazar, University of Porto,
Largo Professor Abel Salazar, 2, 4099-003 PORTO, Portugal
Received (in Cambridge) 2nd February 1999, Accepted 19th March 1999
1H and 31P NMR data show that adenosine 5Ј-(2-methylimidazol-1-ylphosphonate) (2-MeImpA) and guanosine
5Ј-(2-methylimidazol-1-ylphosphonate) (2-MeImpG), known as possible prebiotic precursors of polynucleotides,
produce the corresponding diphosphonucleotides in sodium phosphate solution at pD 7.6. Phosphate ions also
enhance the hydrolysis of these molecules by activating a water molecular associated as a weak complex with
the phosphoryl moiety of 2-MeImpA and 2-MeImpG. A kinetic study was done by quantitative 1H NMR
spectroscopy and mechanistic hypotheses were tested by semiempirical PM3 modelling.
NH2
O
Introduction
N
N
Imidazole-activated nucleotides can be formed in simulated
prebiotic conditions and have been used as precursors for
in vitro synthesis of various kinds of oligoribonucleotides1–4
and oligodeoxyribonucleotides5 using as templates pre-existent
polynucleotides. Other lines of work study the effectiveness of
inorganic catalysts or templates, more likely to be present in
prebiotic conditions, like metal ions6 or montmorillonite clays
and other minerals.7 Detailed mechanistic and kinetic analysis
of nucleotide imidazolide hydrolysis and polymerisation is a
fundamental step to understand the action of these particularly
reactive oligonucleotide precursors. The general reaction of
polymerisation occurs through a nucleophilic substitution on
the phosphoryl moiety of the activated nucleotide by another
nucleotide hydroxy group, imidazole being the leaving group.
The MgII ion has an important role in the process,8 further
activating the phosphoryl moiety. Depending on the conditions
(templates, catalysts, pH, temperature) several kinds and
lengths of oligonucleotides may result, even dimers like NppN
(pyrophosphoryl bridge between two nucleotides). The acti-
vated phosphoryl group is in fact sensitive to any nucleophile,
suggesting a method for the prebiotic synthesis of nucleotide
polyphosphates by reaction with inorganic phosphate or pyro-
phosphate in solution. In this work, the effect of phosphate
concentration on 2-methylimidazole-activated nucleotides
was studied in aqueous systems containing sodium phosphate
and adenosine 5Ј-(2-methylimidazol-1-ylphosphonate) (2-
MeImpA) or guanosine 5Ј(2-methylimidazol-1-ylphosphonate)
(2-MeImpG) (Scheme 1).†
N
HN
2
8
8
N
N
N
H2N
N
A
G
CH3
2
CH3
+
N
N
N
HN
4
5
2-Melm
O
H-2-Melm+
R1
5′
R2
P
O
CH2
α
1′
O–
H
H
H
H
OH
OH
2-MelmpA–
R1 = A, R2 = 2-Melm
R1 = A, R2 = 2-Melm+
R1 = G, R2 = 2-Melm
R1 = G, R2 = 2-Melm+
R1 = –NH2, R2 = 2-Melm
R1 = –NH2, R2 = 2-Melm+
H-2-MelmpA±
2-MelmpG–
H-2-MelmpG±
2-MelmpR–
H-2-MelmpR±
Scheme 1
Results and discussion
Analysis and interpretation of 2-MeImpG and 2-MeImpA
hydrolysis NMR spectra
NMR was chosen as the qualitative and quantitative
analytical technique in this study as an alternative to the com-
monly used HPLC because it provides an equally accurate and
more straightforward experimental method. It also has the
advantage of providing structural insight into the reaction
products. The pD value of 7.6 was chosen for the experiments
because it is low enough to avoid reaction of 2-MeImpN with
OHϪ and large enough to limit the reaction rate so that NMR
spectra acquisition times are negligible. It is also inside the
presumed pH values of prebiotic sea water.
Typical H and 31P spectra of 2-MeImpG and 2-MeImpA in
1
phosphate buffer are presented in Fig. 1 after some of the reac-
tion has occurred. The spectra taken in 0.2  TRIS buffer
instead of phosphate buffer are identical to the spectra in Fig.
1, but the signals assigned to ADP or GDP species are absent.
The chemical shifts presented in Fig. 1 are constant for all
experiments reported in this work with respect to buffer type,
buffer concentration and time and so correspond to a full
1H and 31P NMR spectral characterisation of 2-MeImpA and
2-MeImpG and their hydrolysis products, under the standard
conditions used in all experiments: 0.02  2-2-MeImpN, 0.5 
NaCl, pD 7.6, 308 K (see Experimental).
† In the text 2-MeImpA stands for 2-MeImpAϪ and H-2-MeImpA ,
2-MeImpG stands for 2-MeImpGϪ and H-2-MeImpG , 2-MeImpN
The assignment of the 2-MeImpG spectrum was per-
formed by comparing with spectra of GMP, guanosine and
stands for 2-MeImpA and 2-MeImpG. The word phosphate or the
2Ϫ
symbol Pi designate H2PO4Ϫ and HPO4
.
J. Chem. Soc., Perkin Trans. 2, 1999, 1143–1148
1143

View Article Online
Fig. 1 Chemical shifts are in ppm. (Top) Typical 2-MeImpA (left) and 2-MeImpG (right) 1H NMR spectra (aromatic region) in phosphate buffer,
showing reactants and products signals. Shown spectra are after approximately 4 h of reaction, [Pi] = 0.20  in 2-MeImpA solution and [Pi] = 0.40 
in 2-MeImpG solution; all other conditions are standard (see Experimental). The methyl signals Me2-MeIm (1.36 ppm), Me2-MeImpA (1.14 ppm) and
Me2-MeImpG (1.20 ppm) are not shown. (Bottom) Typical 31P NMR spectra for 2-MeImpA or 2-MeImpG samples in phosphate buffer. Shown
spectrum is for 2-MeImpG solution after approximately 13 h of reaction, [Pi] = 0.20 ; all other conditions are standard (see Experimental). When
0.2  TRIS buffer is used instead of phosphate buffer all the signals labelled ADP or GDP are not present; the other signal chemical shifts are the
same as for phosphate buffer solutions.
2-methylimidazole taken under the standard conditions used in
this work. The suspicion that GDP was present in 2-MeImpG
phosphate buffer solutions stems mainly from analysis of the
31P NMR spectrum, which shows two doublets that are not
present when TRIS buffer is used. These correspond to two
phosphorus nuclei that are mutually coupled, as can be proved
by a simple 31P–31P COSY experiment, with an apparent coup-
ling constant of 32 Hz. One of these doublets also shows an
unresolved fine structure which was attributed by a 1H–31P
heteronuclear correlation experiment to coupling with the H-5Ј
nuclei from the ribosyl moiety. This observation alone excludes
the possibility of GppG being one of the hydrolysis products
and strongly suggests the presence of GDP in solution, as well
as the expected GMP. In fact, the GDP 31P NMR spectrum was
found to match perfectly the two doublets observed in the
ously been reported and the kinetic and mechanistic aspects of
this reaction are the subjects of the next sections. No evidence
for nucleotide dimers or polymers was found in the spectra.
Kinetic model
The NMR integrals are proportional to concentrations and not
to activities and so all derivations will be made using concen-
trations. In view of the NMR species analysis, eqns. (1) and (2)
k1Ј
2-MeImpN
2-MeIm ϩ NMP
(1)
(2)
k2Ј
2-MeImpN ϩ Pi
2-MeIm ϩ NDP
1
2-MeImpG 31P NMR spectrum in phosphate buffer. The H
are necessary to describe the 2-MeImpN system for the pH
value of 7.6 used in this work.
NMR spectrum of 2-MeImpG in phosphate buffer also shows
an unexpected signal for 8-H which was found to have the same
chemical shift as the 8-H signal in the GDP 1H NMR spectrum,
thus corroborating the conclusions drawn from the 31P NMR
spectral analysis.
To qualitatively analyse the 2-MeImpA reactions all the
experiments mentioned above for 2-MeImpG were reproduced.
To reach the assignments presented in Fig. 1 (left and bottom)
the AMP and ADP spectra were compared with the 2-MeImpA
spectra. The 2-MeImpA 31P NMR spectra in phosphate
buffer show exactly the same pattern of signals present in the
2-MeImpG 31P NMR spectra and so the reaction products are
AMP and ADP if the 2-MeImpA hydrolysis is performed in
phosphate buffer and only AMP if TRIS buffer is used instead.
The spontaneous phosphorylation of 2-methylimidazolide-
activated nucleotides from inorganic phosphate has not previ-
Phosphate concentration will always be considered in great
excess over 2-MeImpN concentration and therefore included in
the rate coefficients when necessary. By making the simplest
mechanistic assumptions the main rate equations applicable
to the degradation of 2-MeImpN are given in eqns. (3)–(5).
Ϫd[2MeImpN]/dt = kobs[2-MeImpN]
d[NMP]/dt = k1Ј[2-MeImpN]
d[NDP]/dt = k2Ј[2-MeImpN]
(3)
(4)
(5)
Based on eqns. (1)–(5), expressions for the time dependence of
[2-MeImpN], [2-MeIm], [NMP] and [NDP] can be derived
[eqns. (6)–(9)].
1144
J. Chem. Soc., Perkin Trans. 2, 1999, 1143–1148

View Article Online
Table 1 Observed rate coefficients and final ratio between NMP and NDP mole fractions in solution of 0.02  2-MeImpA and 2-MeImpG; pD 7.6,
308 K
2-MeImpA
0a
2-MeImpG
0a
[Pi]/
0.20
0.30
0.40
0.50
0.20
0.30
0.40
0.50
kobs/hϪ1
χNMP :χNDP
0.036
—
0.083
0.78
0.116
0.71
0.152
0.69
0.180
0.63
0.027
—
0.071
0.54
0.103
0.50
0.132
0.45
0.165
0.43
a 0.20  TRIS buffer.
Fig. 2 Time evolution of [2-MeImpG] (squares), [GMP] (circles) and
[GDP] (triangles) in 0.20  phosphate buffer as monitored by the inte-
grals of 8-H2-MeImpG, 8-HGMP and 8-HGDP NMR signals. Actual initial
concentration of 2-MeImpG was 0.019  due to time zero hydrolysis.
Fittings from eqns. (6), (8) and (9).
Fig. 3 Effect of phosphate concentration on observed rate coefficients.
Squares are for 2-MeImpA and circles are for 2-MeImpG (0.002 
solutions). Open symbols represent values obtained from mixed buffer
solutions with [TRIS]:[phosphate] equal to 0.20:0, 0.15:0.05,
0.10:0.10 and 0.05:0.15 .
[2-MeImpN] = [2-MeImpN]0 eϪtk
[2-MeIm] = [2-MeImpN]0(1 Ϫ eϪtk
(6)
(7)
(8)
(9)
obs
on kobs is obvious for both nucleotides. The standard deviation
associated with these values is smaller than 5% of the average
and reveals the unavoidable errors on integral measurements
mostly due to phase and baseline correction of spectra.
obs
)
[NMP] = (k1Ј/kobs)[2-MeImpN]0(1 Ϫ eϪtk
)
obs
Solutions with mixed TRIS and sodium phosphate buffer
were analysed to investigate the effect of buffers on the experi-
mental rate constants. In Fig. 3 the first four data points were
taken from solutions with mixed buffer and show a different
apparent linear slope than the last four data points that corre-
spond to phosphate buffer only solutions. TRIS seems to have
an increasing effect on kobs relative to the expected linear
extrapolation of the four last points to zero concentration and
so the inertness of this buffer when used for kinetic calculations
of this kind of compound on aqueous solution can be ques-
tioned. A possible mechanism for the influence of TRIS and
phosphate buffers on 2-MeImpN hydrolysis rate will be dis-
cussed in the next sections.
[NDP] = (k2Ј/kobs)[2-MeImpN]0(1 Ϫ eϪtk
)
obs
The coefficients k1Ј and k2Ј can be calculated indirectly from
kobs, [NMP] and [NDP] using eqns. (10) and (11).
k1Ј = kobs/(1 ϩ ([NDP]/[NMP]))
k2Ј = kobs/(1 ϩ ([NMP]/[NDP]))
(10)
(11)
NMR data, kobs calculation and the buffer effect
Fig. 2 shows a typical graph of species mole fraction evolution
with time. Some hydrolysis is noticeable at time zero and was
accounted for in the fittings by letting [2-MeImpN]0 be an
adjustable parameter in eqns. (6), (7) and (8). The long term
storage and occasional opening of the vacuum containers
where the 2-MeImpA and 2-MeImpG solids were stored may
have caused some water vapor to be adsorbed leading to some
hydrolysis.
The overall observation is that the reaction rates increase
with phosphate concentration and that 2-MeImpA reactions
are slightly faster. The different contributions to the rates can
be analysed using the final mole fractions of NMP and NDP in
solutions with varying concentrations of phosphate buffer.
Rate coefficients for NMP and NDP formation
Very good agreement was generally found between data
1
and the exponential functions for all the H nuclei, but the
The χNMP :χNDP values in Table 1 are the average of the
2-HAMP and 2-HADP integral ratio measured for 2-MeImpA
solutions and for 8-HGMP and 8-HGDP integral ratio measured
for 2-MeImpG solutions at the end of the reaction. The phos-
phate concentration dependence of k1Ј and k2Ј values, derived
from data in Table 1 and eqns. (10) and (11), is presented in Fig.
4 for k1Ј and Fig. 5 for k2Ј (logarithmic form).
kobs values obtained from the signals 4-H/5-H2-MeIm and
4-H2-MeImpG, which overlap at the base, had to be excluded from
the average kobs calculations. For 2-MeImpA solutions, only the
overlapping signal sets of 2-H and 8-H were not used. The 31P
signals could not be used for quantitative measurements due to
unacceptable spectral noise.
The calculated kobs values for phosphate buffer only solutions
are presented in Table 1 and were obtained by fitting eqns. (6) to
(9) to the NMR mole fraction data and averaging for the
selected signals. The marked effect of phosphate concentration
The k1Ј dependence on phosphate concentration was
unexpected in view of the proposed model [eqn. (1)] but it is
in fact a consequence of the observed relative increase of kobs
over χNDP :χNMP, that is, χNMP also responds positively to the
J. Chem. Soc., Perkin Trans. 2, 1999, 1143–1148
1145

View Article Online
Table 2 Calculated rate coefficients for eqns. (1) and (2)a
and the logarithmic form of eqn. (13); Table 2 summarises the
calculated values. Eqn. (12) was used with n equal to 1 because
the data are apparently linear. This linearity, observed for both
nucleotide derivatives, indicates a first order dependence on
phosphate concentration for the phosphate catalysed hydrolysis
reaction [embedded in eqn. (1)]. This suggests a process in
which a phosphate ion activates a water molecule by hydrogen
bonding leaving more net negative charge in the water oxygen
atom that attacks more effectively the positively charged phos-
phorus atom of the nucleotide. This might also be the mechan-
ism by which TRIS enhances the 2-MeImpN hydrolysis, acting
as a base for one of the hydrogen atoms of a solvation water
molecule.
k1 b/hϪ1
k1p b/MϪ1 hϪ1
k2 c/MϪ1 hϪ1 nc
2-MeImpA 0.014 0.003
0.11 0.01
0.21 0.01 0.94 0.02
2-MeImpG 0.009 0.001 0.081 0.003 0.23 0.01 0.99 0.02
a The error intervals presented are the linear fitting parameters standard
errors. b From fitting of eqn. (12), n set to 1. c From the fitting of eqn.
(13) logarithmic form.
The fitting of the logarithmic form of eqn. (13) to the k2Ј data
yielded a value of approximately 1 for n on both nucleotide
derivatives (Table 2), which is consistent with a nucleophile sub-
stitution of second order (SN2) for the mechanism associated
with eqn. (2).
The difference in kobs observed for 2-MeImpA and 2-
MeImpG is in fact a consequence of a slightly faster rate of
hydrolysis (catalysed and non-catalysed) for 2-MeImpA, the
rate coefficient for the formation of ADP and GDP being very
similar in both nucleotides. There weren’t expected to be large
differences between the reactivity of the two activated nucleo-
tides given the bond distance between the reaction location (the
nucleotide phosphoryl) and the nucleotide base. As will be dis-
cussed below, the dependence of k1 on pD can account for the
observed rate coefficient differences between 2-MeImpA and
2-MeImpG because its pKa values can be slightly different.
At pD 7.6 both the protonated and deprotonated forms of
2-MeImpN exist in solution in approximately equal amount
(pKa ~ 7.74)8 but the protonated form, H-2-MeImpG , reacts
much more rapidly with water, due to the positive charge on the
2-methylimidazole ring. The calculated rate coefficient k1
should then be about half of the maximum possible value.
Allowing for this, for the fact that these experiments were done
in D2O and not H2O, and also for experimental error, a value of
two times 0.009 hϪ1 for 2-MeImpG k1 is still less than the previ-
ously published value8 of 0.033 hϪ1. The k1 values calculated in
this work are in fact for zero concentration of buffer, which
could account for the difference observed. The rate coefficients
k1p and k2 have not previously been reported and also depend
on pD not only because of the 2-methylimidazole moiety
protonation but also because of phosphate protonation. Two
species of phosphate are in solution at pD 7.6 with different
Fig. 4 Calculation of k1 and k1p by fitting the eqn. (12) to k1Ј depend-
ence on phosphate concentration, n set to 1. Squares are for 2-MeImpA
and circles are for MeImpG (0.02  solutions).
2Ϫ
charges (HPO4 and H2PO4Ϫ) and probably with different
reactivities. Nevertheless, given the k2 values here determined
and the rate coefficients determined for template directed
dimerisation of 2-methylimidazole activated nucleotides,9
inorganic phosphate, if present in solution in equivalent
amount to the nucleotide and at pH values near 7.6, is able to
compete effectively with template directed polymerisation.
Fig. 5 Calculation of k2 by fitting the logarithmic form of eqn. (13) to
k2Ј dependence on phosphate concentration. Squares and solid line are
for 2-MeImpA, circles and dashed line are for 2-MeImpG (0.02 
solutions).
Molecular modelling and reaction simulation
increasing concentration of phosphate. This observation was
interpreted as a result of phosphate catalysed 2-MeImp
hydrolysis and not caused by NDP species hydrolysis, which has
reasonable kinetic stability at pH 7.6. This required therefore
the definition of k1Ј as in eqn. (12), where k1 is the hydrolysis
Two different reactions are proposed for 2-MeImpN in phos-
phate buffer: the hydrolysis [eqn. (1)] and the 2-methylimidazole
substitution by phosphate [eqn. (2)]. We propose also that
hydrolysis can be accelerated by phosphate, or other bases, like
TRIS, through a mechanism in which a solvation water mole-
cule associated with P-α, structure 1 in Scheme 2, becomes
more nucleophilic when phosphate, for example, forms a hydro-
gen bond with one of the hydrogen atoms of the associated
water molecule.
Computational modelling of the reactions was performed to
validate the proposed mechanisms and to assess the importance
of protonation on the imidazole and on the phosphate ion.
Based on the experimental results described so far a semi-
empirical PM3 modelling of reagent association as “weak
complexes”, of which two examples are shown in Scheme 2,
k1Ј = k1 ϩ k1p[Pi]n
(12)
rate coefficient and k1p is the phosphate catalysed hydrolysis
rate coefficient. For k2Ј the phosphate dependence can be
described as in eqn. (13).
k2Ј = k2[Pi]n
(13)
Fig. 4 and 5 show in graphical form the k1Ј and k2Ј data used
to calculate k1, k1p and k2 by fitting eqn. (12), with n equal to 1,
1146
J. Chem. Soc., Perkin Trans. 2, 1999, 1143–1148

View Article Online
Scheme 2
was performed. The representation of exchangeable hydrogen
critical distances. Taking these distances as a crude indication
of reactivity, H-2-MeImpN and HPO42Ϫ are the most effective
reactants (longer critical distance) and should account for the
most part of the experimental rate coefficients.
1
2
atoms as H or H didn’t affect the results of the calculations
reported below. To speed up the calculations, and since the
results for the two nucleotide derivatives were qualitatively
equal and quantitatively very similar, a smaller model was used
in which the purine base moiety was replaced by an NH2 group,
molecules 2-MeImpRϪ or H-2-MeImpR (Scheme 1).
Conclusions
In the PM3 equilibrium geometries of H-2-MeImpR and 2-
MeImpRϪ the phosphorus atom has a clearly positive charge
but more so on H-2-MeImpR (electrostatic fit charge of about
ϩ2.5). In this molecule the electronic density in the P–N bond
(of approximately 2 Å in length) is less than in 2-MeImpRϪ,
consequently being easier to break.
For relatively simple systems like the ones studied in this work
NMR is a good qualitative and quantitative tool. Overlapping
of signals can, however, hinder the study of more complex sys-
tems, namely, systems where oligonucleotides are a product. In
that situation it would be difficult to distinguish between differ-
ent kinds and sizes of polymers. The choice of buffers is some-
what limited for NMR experiments since it is necessary to avoid
overlapping of buffer signals with the subject substance signals,
but direct influence of the buffer on the measured coefficient
rates is also an issue independent of the chosen analytical tech-
nique. The two buffers used in this work were shown to have a
non-negligible effect on the 2-MeImpA and 2-MeImpG
hydrolysis rates. Phosphate catalyses 2-MeImpN hydrolysis by
activating a solvation water molecule and TRIS also clearly
increases the 2-MeImpN hydrolysis rate coefficients (Fig. 3)
probably by the same mechanism, acting as a base.
Optimistic and pessimistic arguments regarding chemical
replication in prebiotic conditions have at this moment a
relative weight that is difficult to evaluate10 but for any hypoth-
esis to be valuable robustness is an important attribute. As was
mentioned in the Introduction the ribonucleoside 5Ј-(2-
methylimidazol-1-ylphosphonates) can polymerise in a tem-
plate, be it a previously available polyribonucleotide or an
inorganic surface like a montmorillonite clay, without the need
for a catalyst other than the template itself. But, as was shown
in this work, the hydrolyses of these activated nucleotides can
be easily enhanced by the presence in solution of phosphate
ions, and probably by other bases. The observed 2-methyl-
imidazole nucleophilic substitution by phosphate is also very
effective at pD 7.6. 2-MeImpA and 2-MeImpG are sensitive to
hydrolysis even in the solid state when stored in a vacuum
container with desiccant, as was reported here. Given all this
data and the presumed complexity of the prebiotic aqueous
environment it seems unlikely that polymerisation of these pre-
cursors could compete with other nucleophiles in solution for
substitution of the imidazolide moiety. The kinetically more
stable di- and triphosphornucleotides, that could have been
PM3 energy minimisations show that water can form a weak
complex with 2-MeImpRϪ and H-2-MeImpR , stabilised by a
dative bond to the P–N σ* MO and one or two hydrogen bonds
as in Scheme 2, complex 1. Hydrolysis, and the consequent P–N
bond breakage, can proceed via nucleophilic attack at one of
the hydrogen atoms on the associated water molecule. This kind
of relayed attack on the phosphoryl group was simulated by
2Ϫ
Ϫ
associating this complex with a HPO4 or H2PO4 ion as
shown in 1. These “complexes” were then subjected to geom-
etry optimisation (energy minimisation) without constraints
and with several initial distances along the O–H–O direction
between the reactants. Above some critical distance the energy
minimisation process physically separates the reactants but
below that distance the energy minimisation procedure repro-
duces the actual reaction, breaking the P–N bond and generat-
ing the ribose 1-amino 5-phosphate.
Ϫ
2Ϫ
The attack of H2PO4 and HPO4 on H-2-MeImpR and
2-MeImpRϪ was simulated in a similar way. Weak complexes
were built in which the phosphate attacks the phosphoryl group
along a straight line that contains the atoms (N–P)–(O–P), as
in 2, Scheme 2. Products were 2-methylimidazole and ribose
1-amino 5-diphosphate.
These critical distances, obtained in the simulated gas phase,
lie between 1.8 to 5.5 Å and relate to transition states inasmuch
as the selected intrinsic reaction co-ordinates are concerned.
Two such possible transition structures determined in this work
in the gas phase are shown in Scheme 2. Water solvation tends
to stabilise localised charge (ions) and to shield the reactants
interaction, thus increasing the energy barrier and slowing
down the reactions. However, the proposed reaction mechan-
isms are plausible also in aqueous solution, possibly with lower
J. Chem. Soc., Perkin Trans. 2, 1999, 1143–1148
1147

View Article Online
readily synthesized from the imidazolide nucleotides and
inorganic phosphate in prebiotic conditions, would require,
unlike the 2-methylimidazole-activated nucleotides, some kind
of catalyst for template directed polymerisation (probably
“inorganic”, like metal ions or mineral) but seem better candi-
dates for the prebiotic job of synthesizing the first polyribo-
nucleotides because of their kinetical stability to nucleophile
substitution.
and/or TRIS (up to 0.2 ) in D2O. Buffer pD was adjusted
with concentrated KOD and/or DNO3 to 7.6 (pD = pHmeasured ϩ
0.4)12 using a Crison Micro pH 2002 pHmeter and an Ingold
U402 microelectrode calibrated with pH concentration buffers
4 and 7. Samples were prepared by dissolving in the NMR tube
the required amount of 2-MeImpN in 0.5 ml of buffer to give a
concentration of 0.020 0.001 . Spectra were taken with 2 to
3 h intervals at 308 K (35 ЊC) until 2-MeImpN 1H signals were
no longer detected. Typically, 10 to 16 spectra were obtained;
the time attributed to each spectrum was its acquisition finish-
ing time. Mole fractions for each time were calculated based on
the relative integrals of each nuclei signal set. Kinetic functions
were fitted to the mole fraction data using linear and non-linear
least-squares analysis.
Experimental
Reagents
Nucleotides, KOD and NaOD (conc.) and reagents for
2-MeImpN synthesis were from Sigma; D2O (99.5% D), NaCl
and buffers were from Merck.
Molecular modelling
Synthesis of 2-MeImpA and 2-MeImpG
Semiempirical PM3 molecular modelling was done with PC
Spartan Plus (Wavefunction, Inc.), ver. 1.4, using H–H repul-
sion correction, running in a 300 MHz Pentium II PC with
Windows NT 4.0 Workstation OS.
Preparation of the activated nucleotides was done by a pro-
cedure previously described.2 The products were dried in a Petri
dish for at least a week under vacuum with desiccant and stored
also under these conditions in a small capped flask; yield was
better than 95% in weight. However, the 1H NMR spectra
revealed unaccounted singlets at 0.68 and 1.48 ppm (chemical
shift reference: tert-butyl alcohol) that do not belong to the
products. The intensity and chemical shifts of these signals
didn’t change with time. The 31P NMR spectra didn’t show
any unexpected signals. The chemical shifts observed for
2-MeImpA and 2-MeImpG spectra taken in the standard con-
ditions for this work (see below) are presented in Fig. 1.
Acknowledgements
The authors would like to thank CEQUP and ICBAS for
financial support and Professor Anake Kijjoa for fruitful
discussions.
References
1 T. Inoue and L. E. Orgel, J. Mol. Biol., 1982, 162, 201.
2 G. F. Joyce, T. Inoue and L. E. Orgel, J. Mol. Biol., 1984, 176, 279.
3 L. E. Orgel, Nature, 1992, 358, 203.
NMR
4 D. Sievers and G. von Kiedrowski, Nature, 1994, 369, 221.
5 T. Wu and L. E. Orgel, J. Am. Chem. Soc., 1992, 114, 317; 5496;
7963.
6 H. Sawai and M. Ohno, Chem. Pharm. Bull., 1981, 29, 2237; Bull.
Chem. Soc. Jpn., 1985, 58, 361; H. Sawai, K. Higa and K. Kuroda,
J. Chem. Soc., Perkin Trans. 1, 1992, 505.
7 J. P. Ferris and G. Ertem, Origins Life Evol. Biosphere, 1992, 22, 369;
K. Kawamura and J. P. Ferris, J. Am. Chem. Soc., 1994, 116, 7564;
P. Ding, K. Kawamura and J. P. Ferris, Origins Life Evol. Biosphere,
1996, 26, 151.
8 A. Kanavarioti, C. F. Bernasconi, D. L. Doodokyan and D. S.
Alberas, J. Am. Chem. Soc., 1989, 111, 7247.
Spectra were taken on a Bruker AC200 spectrometer in D2O
solution using 5 mm NMR tubes. Chemical shift references
were tert-butyl alcohol for 1H spectra (internal) and 65% phos-
phoric acid for 31P spectra (external). The 1H spectra were
acquired at 200 MHz over 13 min with 15 ppm sweep width, 60Њ
pulse width, 8 dummy scans and 160 scans. The 31P spectra were
acquired at 81 MHz over 27 min with 37 ppm sweep width, 70Њ
pulse width, 4 dummy scans and 512 scans.
Homonuclear (1H–1H and 31P–31P) and heteronuclear (1H–
31P) chemical shift correlation experiments were done after
complete hydrolysis of the samples using H–H-COSY and H–
X-COSY pulse sequences.11
9 A. Kanavarioti, C. F. Bernasconi, D. L. Doodokyan and E. E.
Baird, J. Am. Chem. Soc., 1993, 115, 8537.
10 J. Ferris, Nature, 1994, 369, 184.
11 W. P. Awe, E. Bartholdi and R. R. Ernst, J. Chem. Phys., 1976, 64,
2229; A. Bax and G. Morris, J. Magn. Reson., 1981, 42, 501.
12 P. K. Glasoe and F. A. Long, J. Phys. Chem., 1960, 64, 188.
Sample preparation and standard experimental conditions for
kinetic measurements
Buffer stock solutions were prepared by dissolving the required
amounts of NaCl (0.5 ), Na2HPO4/NaH2PO4 (up to 0.5 )
Paper 9/00884E
1148
J. Chem. Soc., Perkin Trans. 2, 1999, 1143–1148