Unique Catalysis and Regioselectivity Observed in the Poly(C)-Directed RNA Dimer Formation from 2-MeImpG: Kinetic Analysis as a Function of Monomer and Polymer Concentration

Article abstract of DOI:10.1021/jo991216q

Polycytidylate, poly(C), serves as a scaffold or template to direct and catalyze the synthesis of long oligoguanylates from guanosine 5′-phosphate 2-methylimidazolide, 2-MeImpG. In the absence of poly(C), small amounts of three isomeric dimers, i.e., the 2′-5′-, the 3′-5′-, and the pyrophosphate-linked, are formed slowly. In the presence of poly(C) oligomers that are primarily 3′-5′-linked are formed quickly and in high yield. Product analysis suggests that the oligomers are elongation products of the 3′-5′-linked dimer, abbreviated D. Assuming that D is formed slowly from two molecules of 2-MeImpG (Scheme 1) and elongates relatively fast, the initial rate of dimerization, d[D]/dt in M h^-1, was determined using two independent methods. The first method is based on the approximation that at the onset of the reaction the substrate is consumed only via hydrolysis and dimerization, and thus elongation can be neglected. The second, more accurate, method exploits the assertion that every oligomer was once a 3′-5′-linked dimer. Hence the concentration of D was obtained indirectly from the concentration of the oligomer products. These two methods gave comparable results. Experiments were run in aqueous solution in the presence of 1.0 M NaCl, 0.2 M MgCl₂ at pH 7.9 ± 0.1 and 23°C. Controls were run in the absence of poly(C) and in the presence of other polynucleotides. The kinetics were determined as a function of both monomer and polymer concentration the latter expressed in C equivalents. The kinetic data obtained in the presence of poly(C) confirmed an earlier conclusion regarding the remarkable effect of poly(C) on the formation of the 3′-5′-linked diguanylate. Initial dimerization rates were quantitatively correlated using a simple template-directed (TD) model that presumes cooperative binding (two association constants) of 2-MeImpG on poly(C) and reaction between adjacent template-bound molecules. The model allows for the estimation of the association constants and the intrinsic rate constant of dimerization, k₂*- Insights into the detailed mechanism are also gained from this analysis. The fact that the proposed model can successfully correlate kinetic data that vary by more than 5000-fold between the slowest and the fastest reaction adds confidence and suggests the suitability of this model for describing TD reactions in general. It is anticipated that similar analysis of other known TD reactions may lead to clues that will facilitate the design of more efficient polynucleotide-synthesizing systems.

Full text of DOI:10.1021/jo991216q

J . Org. Chem. 1999, 64, 8323-8333
8323
Un iqu e Ca ta lysis a n d Regioselectivity Obser ved in th e
P oly(C)-Dir ected RNA Dim er F or m a tion fr om 2-MeIm p G: Kin etic
An a lysis a s a F u n ction of Mon om er a n d P olym er Con cen tr a tion
Anastassia Kanavarioti,* Eldon E. Baird, T. Brian Hurley, J ulie A. Carruthers, and
Sumana Gangopadhyay1
Department of Chemistry and Biochemistry, University of California, Santa Cruz, California 95064
Received August 3, 1999
Polycytidylate, poly(C), serves as a scaffold or template to direct and catalyze the synthesis of long
oligoguanylates from guanosine 5′-phosphate 2-methylimidazolide, 2-MeImpG. In the absence of
poly(C), small amounts of three isomeric dimers, i.e., the 2′-5′-, the 3′-5′-, and the pyrophosphate-
linked, are formed slowly. In the presence of poly(C) oligomers that are primarily 3′-5′-linked are
formed quickly and in high yield. Product analysis suggests that the oligomers are elongation
products of the 3′-5′-linked dimer, abbreviated D. Assuming that D is formed slowly from two
molecules of 2-MeImpG (Scheme 1) and elongates relatively fast, the initial rate of dimerization,
d[D]/dt in M h-1, was determined using two independent methods. The first method is based on
the approximation that at the onset of the reaction the substrate is consumed only via hydrolysis
and dimerization, and thus elongation can be neglected. The second, more accurate, method exploits
the assertion that every oligomer was once a 3′-5′-linked dimer. Hence the concentration of D was
obtained indirectly from the concentration of the oligomer products. These two methods gave
comparable results. Experiments were run in aqueous solution in the presence of 1.0 M NaCl, 0.2
M MgCl2 at pH 7.9 ( 0.1 and 23 °C. Controls were run in the absence of poly(C) and in the presence
of other polynucleotides. The kinetics were determined as a function of both monomer and polymer
concentration the latter expressed in C equivalents. The kinetic data obtained in the presence of
poly(C) confirmed an earlier conclusion regarding the remarkable effect of poly(C) on the formation
of the 3′-5′-linked diguanylate. Initial dimerization rates were quantitatively correlated using a
simple template-directed (TD) model that presumes cooperative binding (two association constants)
of 2-MeImpG on poly(C) and reaction between adjacent template-bound molecules. The model allows
for the estimation of the association constants and the intrinsic rate constant of dimerization,
k2*. Insights into the detailed mechanism are also gained from this analysis. The fact that the
proposed model can successfully correlate kinetic data that vary by more than 5000-fold between
the slowest and the fastest reaction adds confidence and suggests the suitability of this model for
describing TD reactions in general. It is anticipated that similar analysis of other known TD
reactions may lead to clues that will facilitate the design of more efficient polynucleotide-synthesizing
systems.
In tr od u ction
synthesis by another method. TD ligations of unprotected
monomers may have been the method of choice, albeit
they are inefficient.4a In the context of prebiotic chemistry
and in an attempt to unravel simple self-replicating
systems, the nonenzymatic TD synthesis of polynucle-
otides and analogs thereof has received considerable
attention.5 TD polymerizations are initiated by a rela-
tively slow two-monomer ligation step (dimerization),
followed by, most likely, faster and perhaps more regi-
Template-directed (TD) enzyme-assisted synthesis of
polynucleotides is the method chosen by Nature to
replicate its gene pool. In the process of evolution, Nature
has developed a rather complex procedure for transmit-
ting information accurately and in a regulated fashion.
Recently TD nonenzymatic synthesis has found many
applications in the pharmaceutical industry for the large-
scale production of antisense polynucleotides.2 Such
synthesis is performed in an elegant and efficient way
by TD ligation3 of two oligomers, but requires oligomer
(4) (a) Hill, A. R., J r.; Orgel, L. E.; Wu, T. Origins Life Evol.
Biosphere 1993, 23, 285-290. (b) Inoue, T. and Orgel, L. E. J . Mol.
Biol. 1982, 162, 201-218. (c) Fakhrai, H.; Inoue, T.; Orgel, L. E.
Tetrahedron 1984, 40, 39-45. (d) Inoue, T.; Orgel, L. E. J . Am. Chem.
Soc. 1981, 103, 7666-7667.
(5) Inoue, T.; Orgel, L. E. Science 1983, 219, 859-862. J oyce, G. F.
Cold Spring Harbor Symp. Quant. Biol. 1987, 52, 41-51. Orgel, L. E.
Nature 1992, 358, 203-209. Kanavarioti, A. Origins Life Evol. Bio-
sphere 1994, 24, 479-495. Orgel, L. E. Acc. Chem. Res. 1995, 28, 109-
118. Bag, B. G.; von Kiedrowski, G. Pure Appl. Chem. 1996, 68, 2145-
2152. Ertem, G.; Ferris, J . P. J . Am. Chem. Soc. 1997, 119, 7197-
7201. J oyce, G. F.; Orgel, L. E. The RNA World; Gesteland, R. F.,
Atkins, J . F., Eds.; Cold Spring Harbor Lab. Press: Cold Spring
Harbor, 1993; pp 1-25. Luther, A.; Brandsche, R.; von Kiedrowski, G.
Nature 1998, 396, 245-248. Zhan, Z.-Y. J .; Lynn, D. G. J . Am. Chem.
Soc. 1997, 119, 12420-12421. Prakash, T. P.; Roberts, C.; Switzer, C.
Angew Chem., Int. Ed. Engl. 1997, 36, 1522-1523.
(1) On leave of absence from the Depart. of Chemistry, Gurudas
College, Calcutta-700054, India.
(2) Bennett, M. R. J . Drug Dev. Clin. Practice 1995, 7, 225-235.
Field, A. K. Antiviral Res. 1998, 37, 67-81.
(3) Rohatgi, R.; Bartel, D. P.; Szostak, J . W. J . Am. Chem. Soc. 1996,
118, 3332-3339. Rohatgi, R.; Bartel, D. P.; Szostak, J . W. J . Am. Chem.
Soc. 1996, 118, 3340-3344. Boli, M.; Micura, R.; Eschenmoser, A.
Chem. Biol. 1997, 4, 309-320. Kuznetsova, S. A.; Merenkova, I. N.;
Kanevsky, I. E.; Shabarova, Z. A.; Blumenfeld, M. Antisense Nucleic
Acid Dev. 1999, 9, 95-100. Selvasekaran, J .; Turnbull, K. D. Nucleic
Acid Res. 1999, 27, 624-627. Xu, Y.; Kool, E. T. Nucleic Acid Res. 1999,
27, 875-881. Koppitz, M.; Nielsen, P. E.; Orgel, L. E. J . Am. Chem.
Soc. 1998, 120, 4563-69.
10.1021/jo991216q CCC: $18.00 © 1999 American Chemical Society
Published on Web 10/14/1999

8324 J . Org. Chem., Vol. 64, No. 22, 1999
Kanavarioti et al.
groups and analyzed with RPC5 chromatography (see below).
RPC5 chromatography resolves oligoguanylates according to
length and isomerism.4b Evaluation of dimer concentration
based on oligomer yields has been described earlier.6a In
contrast to RPC5, C18 chromatography resolves 2-MeImpG
from 5′GMP, i.e., the hydrolysis product as well as from the
dimerization products. Product distribution was obtained
directly from HPLC reports as the percent of the total HPLC
area corresponding to the initial substrate (Go). For example
(% 5′GMP) ) 100 (HPLC area of 5′GMP peak)/(total HPLC
area). It follows that [5′GMP] ) [G]o (% 5′GMP)/100 where [G]o
is the initial or formal concentration of substrate obtained by
weight. Percent total dimer yields (see Scheme 1), % Dall, are
reported in monomer equivalents and are uncorrected for
hypochromicity h.8 Hence [Dall] ) h [G]o (% Dall)/200.
Analysis with C18 chromatography was performed on a C18
Alltima (3.2 × 250 mm, 5 µm by Alltech) solvent minimizer
column run at 0.5 mL/min.9 Solvent A is 0.02 M KH2PO4 with
0.2% w/v trifluoroacetic acid (TFA) pH 2.5; solvent B: 30%
CH3CN in water v/v with 0.2% w/v TFA. 0 to 20% B in 10 min;
isocratic at 20% B for 4 min and then 20% to 32% B in 6 min;
2 min wash with 100% B. This chromatography was used for
the “faster” samples. A slightly modified gradient that exhib-
ited better resolution was used for the “slower” samples; 0 to
15% B in 10 min; isocratic at 15% B for 4 min and then 15%
to 45% B in 16 min; 2 min wash with 100% B. The order of
eluting guanosine derivatives with both of these gradients is
5′GMP, HEPES-pG, guanosine cyclic 3′-5′ monophosphate,
pG2′pG, G5′ppG, 2-MeImpG2′pG, 2-MeImpG, pG3′pG, and
2-MeImpG3′pG coelute, whereas oligomers longer than the
dimers elute later.
oselective, primer extension (elongation) steps. Hence the
overall efficiency of a TD polymerization depends criti-
cally on the yield and regioselectivity of the dimerization
step. Insights in TD mononucleotide dimerization may
facilitate the design of more efficient TD polynucleotide-
synthesizing systems and find application in the large-
scale synthesis of polynucleotide-based pharmaceuticals.
To the best of our knowledge the most efficient non-
enzymatic TD polymerization so far is the poly(C)-
directed oligoguanylate synthesis from 2-MeImpG or G
for simplicity.4bc,6 In the presence of poly(C) this reaction
yields quantitatively oligoriboguanylates,4b whereas in
the absence of poly(C) only a small percentage of dimers
is formed (Figure 1a).7 Enzymatic degradation of the
oligomers indicated that the majority of the linkages is
3′-5′,4b leading to the conclusion that, depending on the
temperature, the primer is an all 3′-5′-linked-dimer or
-trimer.4c Kinetic determinations at 23 °C in the range
0.005 M e [G] e 0.045 M at a constant 0.05 M poly(C)
concentration suggested substantial catalysis by poly(C)
and indicated that d[D]/dt exhibits a third-order depen-
dence on monomer concentration.6a To explain this strong
dependence, a model was proposed in which dimerization
occurs in long stacks, six or longer, of template-bound
monomers. However, the effect of poly(C) concentration
on dimerization was not investigated and thus the
proposed model was not fully tested.6a
Here we have extended the earlier kinetic studies by
including a dependence on poly(C) concentration, a larger
range of monomer concentration, and several controls.
The product distribution and the kinetics of dimerization
were determined in the presence/absence of poly(C) as
well as in the presence of other polymers, such as
polyinosinate, poly(I), polyuridylate, poly(U), and double-
stranded polycytidylate‚polyguanylate, poly(C)‚poly(G).
Perhaps not surprising it was established that these
polymers exhibit practically no effect on 2-MeImpG
dimerization, whereas the effect of poly(C) is dramatic
in comparison. Moreover, the extended set of data was
found to be quantitatively consistent with a TD mecha-
nism of dimerization and allowed an important refine-
ment of the earlier proposed cooperative model describing
such a mechanism.6a
Some of the experiments with samples devoid of polymer
were analyzed with a different C18 chromatography (pH 6.5
in contrast to the one described above at pH 2.5). This analysis
was performed with Hypersil C18 column 4.6 × 250 mm 5 µm
from Phenomenex used at 1.0 mL/min flow. Solvent A is 0.02
M KH2PO4 at pH 6.5, and solvent B is 30% CH3CN in water
(v/v). Gradient is 0 to 13% B in 10 min; isocratic at 13% B for
8 min. Order of elution and typical, retention times were
5′GMP, 4.2 min; pGpG (both isomers), 4.9 min; G5′ppG, 9.5
min; HEPES-pG, 9.8 min; 2-MeImpGpG (both isomers), 10.5
min; 2-MeImpG, 13.5 min. pGpG does not always resolve well
from 5′GMP. Both chromatographies, the one at pH 2.5 and
the one at pH 6.5, gave comparable results.
Resu lts
Dim er F or m a tion w ith or w ith ou t P olym er , Ex-
clu d in g P oly(C). All experiments were performed with
0.0005 M e [G] e 0.083 M in the presence of 1.2 M or
1.0 M NaCl, 0.2 M MgCl2, and 0.5 M HEPES buffer at
pH 7.90 ( 0.1 at 23 °C. Samples of 2-MeImpG in the
presence and in the absence of non-C polymers were
analyzed by C18 chromatography (see Experimental
Section) which allowed monitoring of 2-MeImpG disap-
pearance and product formation. In accord with the
reaction pathways proposed for 2-MeImpC and 2-Me-
ImpU investigated under identical conditions with the
ones reported here,7 we found that 2-MeImpG hydrolyzes
to 5′GMP, reacts with the buffer HEPES to form HEPES-
pG, and yields three isomeric dimers. These are the two
internucleotide linked dimers, pG2′pG and pG3′pG, and
Exp er im en ta l Section
Ma ter ia ls, Meth od s, a n d HP LC An a lysis. Acquisition of
materials, preparation of samples, pH measurements, and
product identification were done following already developed
methods.4b,6,7 2-MeImpG with ꢀ ) 12 000 at 253 nm was better
than 97% pure as tested by C18 chromatography. The poly-
mers poly(C), poly(I), poly(U), and poly(C)‚poly(G) were pur-
chased from Sigma. The potassium salt of poly(C) is about 100
to 300 units long. Analysis of samples was performed with high
performance liquid chromatography (HPLC) using a 1090 LC
from Hewlett Packard equipped with a diode array detector
set at 254 nm. Samples were incubated at 23 °C in HPLC vials
in the thermostated autosampler of the HPLC instrument, and
the analysis was run with C18 chromatography (see below).
Alternatively, samples were incubated in a Lauda bath at 23
( 0.1 °C, quenched in regular intervals by dilution and
addition of acidic EDTA. Then these samples were hydrolyzed
at pH 3 and 50 °C overnight in order to remove the 2-MeIm
(8) Hypochromicity h ) 1.16 (in neutral and acidic solutions) was
determined by enzymatic degradation of G5′ppG to 5′GMP with PDE
(snake venom phosphodiesterase from Crotalus durrisus from Boering-
er Mannheim) from the ratio of the areas between produced 5′GMP
and consumed G5′ppG. An internal standard, 5′AMP, which was not
degraded by the enzyme was included. It was presumed that h ) 1.16
is the same for all guanosine dimers.
(9) C18 packing elutes short oligoguanylates and oligocytidylates
(not shown here) but retains poly(C) and oligoguanylates longer than
the tetramer. Shorter than tetramer oligoguanylates are, most likely,
not quantitatively eluted either.
(6) (a) Kanavarioti, A.; Bernasconi, C. F.; Alberas, D. J .; Baird, E.
E. J . Am. Chem. Soc. 1993, 115, 8537-8546. (b) Kanavarioti, A.; Baird,
E. E. J . Mol. Evol. 1995, 41, 169-173. (c) Kanavarioti, A.; Bernasconi,
C. F.; Baird, E. E. J . Am. Chem. Soc. 1998, 120, 8575-8581.
(7) Kanavarioti, A. Origins Life Evol. Biosphere 1997, 27, 357-376.

Poly(C)-Directed RNA Dimer Formation
J . Org. Chem., Vol. 64, No. 22, 1999 8325
F igu r e 1. (a) HPLC profiles for the self-condensation of 0.066 M 2-MeImpG in the presence of 0.0073 M poly(U) after 6 days
(top) and after 0.35 h (bottom) of incubation at 23 °C. Similar HPLC profiles were obtained in the absence of a control polymer.
Y-axis: milliabsorbance units at 254 nm. Identification of peaks: 1, 5′GMP and HEPES-pG; 2, pG2′pG; 3, G5′ppG; 4, 2-MeImpG2′-
pG; 5, 2-MeImpG; 6, pG3′pG and 2-MeImpG3′pG. C18 Chromatography at pH 2.5 long run as described in the Experimental
Section. No other peaks are detected at later elution times. (b) HPLC profiles for the self-condensation of 0.060 M 2-MeImpG in
the presence of 0.010 M poly(C) after 7.3 h (top) and after 0.2 h (bottom) of incubation at 23 °C. Y-axis: milliabsorbance units at
254 nm. Identification of peaks: 1, 5′GMP; 2, HEPES-pG; 3, pG2′pG; 4, G5′ppG; 5, 2-MeImpG; 6, pG3′pG and 2-MeImpG3′pG. The
peaks seen eluting at later times correspond to oligoguanylates and are presumably products of pG3′pG. Chromatography as
described under part a. Retention times between the samples in Figure 1a and 1b vary because of a guard column that was used
in conjuction with the analytical column but only for the samples under part a. (c) HPLC profiles for the self-condensation of
guanosine derivatives in the presence of a C-template. Y-axis: milliabsorbance units at 254 nm. Top: 0.010 M d(pC)50 with 0.045
M 2-MeImpG after 2.5 h incubation. Bottom: 0.012 M poly(C) with 0.045 M ImpG after 10.5 h incubation. Identification of peaks:
1, G5′ppG; 2, Activated pG2′pG; 3, substrate; 4, activated pG3′pG. 2, 3, and 4, 2-MeIm derivatives for the top and Im derivatives
for the bottom profile. Experimental conditions and HPLC analysis the same as in Figure la. Retention times and yields of these
two profiles are not comparable.
the pyrophosphate-linked dimer, G5′ppG (Scheme 1 and
Figure 1a). The first two appear initially as the imidazo-
lide-activated derivatives, 2-MeImpG2′pG and 2-MeImpG3′-
pG, which hydrolyze slowly to pG2′pG and pG3′pG. In
contrast, G5′ppG has been detected only in the form of
the deactivated dimer, most likely because the corre-
sponding activated derivative, G5′p(2-MeIm)pG, hydro-
lyzes rapidly compared to our detection methods. Dimer-
ization products were detected with [G] > 0.005 M.
None of the tested control polymers had a substantial
effect on the rate or product distribution of the 2-MeImpG
reaction. Figure 2 shows percent substrate, percent
5′GMP, and total percent dimers, % Dall, as a function of
time as described in the Experimental Section. From such
plots, initial rates were obtained from the slope of the
line at 0 time using a second-order fit of the data
(Kaleidagraph, Abelbeck software). Initial rates were
determined for the formation of the hydrolysis product,

8326 J . Org. Chem., Vol. 64, No. 22, 1999
Kanavarioti et al.
Sch em e 1. Rep r esen ta tion of th e P r op osed P oly(C)-Dir ected 2-MeIm p G Dim er iza tion w ith k2* th e TD
In tr in sic Ra te con sta n t for Dim er iza tion (see Discu ssion ). Th e Th ir d Mon om er is Mech a n istica lly
Im p or ta n t Bu t Its P osition Is Ar bitr a r ily Sh ow n Her e Up str ea m
otide-linked dimers. These ratios stayed constant during
the first two days of reaction, and the averages are listed
in Table 1. They indicate that in solution the synthesis
of the 2′-5′-linked dimer is about four times more favored
than the synthesis of any of the other two dimers which
are formed about equally. At later times the product
distribution favors the pyrophosphate dimer, because of
an additional pathway. This pathway is reaction of the
substrate with the hydrolysis product 5′GMP that ac-
cumulates with time and yields G5′ppG as the major
product.7
Dim er F or m a t ion in t h e P r esen ce of P oly(C).
Product distribution in samples of 2-MeImpG was con-
veniently monitored by C18 chromatography every 37 to
45 min depending on the gradient. Samples without poly-
(C) exhibit a decrease from 95% to 75% substrate during
the first 15 h of incubation, whereas samples with poly-
(C) exhibit a similar decrease in 2 to 4 h. After 2 h of
incubation and in the presence of poly(C) the product
distribution includes 0.5% of the RNA dimer, about 6%
of oligomers, 0.1% of the 2′-5′-linked dimer, and no
detectable change in the pyrophosphate-linked dimer
which is an impurity (≈2%) of the substrate preparation
(Figure 1b). Longer incubation than 2 h shows that the
RNA dimer stays constant at 0.5% (plateau, see later),
whereas small amounts of the other two dimers are being
formed and large amounts of longer oligomers accumu-
late. On the basis of the notion that the oligoguanylate
products are more than 90% 3′-5′-linked,4b our observa-
tions can be most easily explained as follows. A poly(C)-
catalyzed process strongly accelerates the synthesis of
the RNA dimer, which does not accumulate beyond a
certain plateau value because it serves as a primer for
further extension. The small fraction of 2′-5′-linked and
pyrophosphate-linked dimers forms in solution rather
than on the template; their yield increases steadily
because no further reaction occurs with these dimers.
These observations show that the poly(C)/2-MeImpG
F igu r e 2. Reaction of 0.066 M 2-MeImpG in the presence of
0.0073 M poly(U). Percent product distribution as a function
of time: H ) hydrolysis products, (circles, left axis); Dall ) sum
of all dimers (squares, left axis); G ) unreacted 2-MeImpG
(triangles, right axis). The percent dimer yield is expressed in
monomer equivalents and is not corrected for hypochromicity
(see Experimental Section). The downward curvature indicates
that the rate of hydrolysis as well as dimer formation is slowed
down at later times. The lines are second-order fits to the data;
d(% H)/dt, d(% G)/dt, d(% Dall)/dt in h-1 are obtained from such
plots from the slopes at 0 time (see Results).
d(%H)/dt (H includes 5′GMP and HEPES-pG), the disap-
pearance of the substrate, d(% G)/dt, and the appearance
of the sum of all dimers, d(% Dall)/dt. These values are
listed in Table 1. The experimental error associated with
the determination of the slopes is estimated at 20% or
less.
Product ratios, 2′/3′, were obtained from the yields of
the two internucleotide-linked dimers 2′-5′-linked vs 3′-
5′-linked. Product ratios, p/(2′+3′), were obtained from
the yield of G5′ppG and the total yield of the internucle-

Poly(C)-Directed RNA Dimer Formation
J . Org. Chem., Vol. 64, No. 22, 1999 8327
Ta ble 1. In itia l Ra tes in th e Rea ction of 2-MeIm p G (G) in th e Absen ce/P r esen ce of Non tem p la tin g P olym er s a t 23 °C
a n d p H 7.90 ( 0.05, 0.5 M HEP ES, 1.0 M Na Cl, a n d 0.2 M MgCl2
f
[G]o, Ma
with
d(% H)/dt,b h-1
d(% G)/dt,c h-1
d(% Dall)/dt,d h-1
p/(2′+3′)e
2′/3′
0.0005
0.0007
0.0010
0.0053
0.0060
0.0101
0.0219
0.0310
0.0333
0.0363
0.0370
0.0450
0.0530
0.0690
0.0830
0.0660
0.0710
0.0720
0.0680
0.0690
0.0680
0.49
0.50
0.49
1.01
1.04
1.15
1.13
1.47
1.37
1.26
1.36
1.35
1.80
1.67
1.55
1.67
1.59
1.49
1.43
1.46
1.35
0.53
0.58
0.59
1.12
1.02
1.14
1.29
1.67
1.68
1.57
1.51
1.66
2.17
2.00
1.94
2.08
2.08
1.98
1.90
1.86
1.69
g
g
g
0.053
0.091
0.094
0.150
0.251
0.198
0.200
0.220
0.272
0.378
0.414
0.472
0.379
0.409
0.424
0.333
0.390
0.342
0.05 M KCl
0.05 M KCl
0.033 Mh
4.3
4.8
4.6
4.8
3.6
4.0
3.6
3.7
3.7
4.6
0.26
0.30
0.33
0.0073 Mh
0.0148 Mh
0.0076 Mi
0.0137 Mi
0.0061 Mj
0.0147 Mj
0.24
0.34
0.24
0.42
0.26
a
b
Formal 2-MeImpG concentration determined by weight. Initial slope of a plot of percent hydrolysis product, H, with time where H
includes both 5′GMP and HEPES-pG. c Initial slope of a plot of percent 2-MeImpG as a function of time. Initial slope of a plot of percent
d
total dimer, Dall, formed as a function of time. Experimental error of the slopes estimated at less than 20%. e Average ratio ((10%) of
pyrophosphate-linked vs internucleotide-linked dimers. f Average ratio ((10%) of 2′-5′ vs 3′-5′-linked dimers. Not detected. Poly(U).
g
h
i
j
Poly(C)‚poly(G). Poly(I).
Ta ble 2. Ra te Da ta for Dim er F or m a tion in th e P r esen ce of P oly(C) Deter m in ed by C18 Ch r om a togr a p h y
[T],a
M
[G]o,b M
rc
d(% H)/dtd h-1
d(% G)/dte h-1
d(% D)/dtf h-1
d[D]/dt (C),g M h-1
TD d[D]/dt,h M h-1
0.0050
0.0050
0.0080
0.0080
0.0080
0.0080
0.0140
0.0600
0.0550
0.0600
0.0550
0.0500
0.0450
0.0510
0.85
0.86
0.85
0.85
0.88
0.86
0.87
1.92
1.71
1.81
2.19
2.21
1.38
1.77
3.94
3.77
5.23
5.69
5.15
3.35
5.55i
2.02
2.06
3.42
3.50
2.94
1.97
3.78
7.03 × 10-4
6.57 × 10-4
1.19 × 10-3
1.12 × 10-3
8.53 × 10-4
5.14 × 10-4
1.12 × 10-3
6.87 × 10-4
6.43 × 10-4
1.17 × 10-3
1.10 × 10-3
8.40 × 10-4
5.05 × 10-4
1.11 × 10-3
a
Poly(C) concentration in cytidine equivalents. b Formal 2-MeImpG concentration determined by weight. c Factor r corrects for substrate
d
purity (see ref 14). Initial slope of a plot of percent hydrolysis product, H, with time where H includes both 5′GMP and HEPES-pG.
e Initial slope of a plot of percent 2-MeImpG as a function of time. f Calculated difference between d(% H)/dt and d(% G)/dt. Experimental
g
error of the slopes estimated at 15 to 20%. Initial rate of 3′-5′-linked dimer formation in the presence of poly(C) (see Results, eq 2′).
Calculated rate of TD 3′-5′-linked dimer formation (see Results). Experiments with higher rates of dimerization could not be analyzed
h
i
with this method.
system exhibits a substantially higher regioselectivity for
3′-5′-linkage formation compared to the other two link-
ages. A lower limit of 15-20 for preference of the
formation of the 3′-5′-linkage over the 2′-5′-linkage can
be estimated from our data, i.e., by comparing the sum
of the HPLC areas under peak 6 and the areas of later
eluting peaks with the area under peak 3 (Figure 1b).
Additional observations in similar systems suggest that
the remarkable catalysis and regioselectivity exhibited
in the poly(C)/2-MeImpG is unique for this system. For
example, product distributions determined under our
analytical conditions with systems d(pC)50/2-MeImpG10,11
and poly(C)/ImpG4d indicate that both internucleotide-
linked dimers are detected and formed in comparable
yields especially with the former system (Figure 1c: peak
2 represents the 2′-5′-linked and peak 4 the 3′-5′-linked).
More importantly, the small yield of the RNA dimer
detected with d(pC)50/2-MeImpG cannot be attributed to
consumption by oligomerization, because the latter reac-
tion is slow (Figure 1c, top). Interestingly, neither of the
above three TD reactions catalyzes the formation of the
pyrophosphate-linked dimer (product distribution as a
function of time, data not shown), which is a major
product in the reaction occurring in solution (Figure 1a
and Scheme 1, bottom).
Two methods were employed for the determination of
the rate of RNA dimer formation. The first method,
abbreviated C18 because samples were analyzed by C18
chromatography (see Experimental Section), was used
here for the first time and it is based on the approxima-
tion that at early reaction times the substrate is con-
sumed by the hydrolysis and the dimerization pathways
and that the elongation pathway can be neglected, i.e.,
-d(% G)/dt ) d(% H)/dt + d(% Dall)/dt. Since with poly-
(C) the dimer formed is practically only the RNA dimer
(Dall ) D), then d(% D)/dt can be determined from the
difference between d(% G)/dt and d(% H)/dt. Values for
d(% G)/dt and d(% H)/dt were determined as described
for samples without poly(C), and they are listed in Table
2 together with the calculated d(% D)/dt values.
The second method, abbreviated RPC5 because samples
were analyzed by RPC5 chromatography (see Experi-
mental Section), had been used earlier.6a This method is
based on the hypothesis that the observed concentration
(10) Chen, C. B.; Inoue, T.; Orgel, L. E. J . Mol. Biol. 1985, 181, 271-
279. Gangopadhyay, S.; Kanavarioti, A. Unpublished results.
(11) Kozlov, I. A.; Politis, P. K.; Van Aerschot, A.; Busson, R.;
Herdewijn, P.; Orgel, L. E. J . Am. Chem. Soc. 1999, 121, 2653-2656,
and references therein.

8328 J . Org. Chem., Vol. 64, No. 22, 1999
Kanavarioti et al.
Ta ble 3. Ra te Da ta for Dim er F or m a tion in th e P r esen ce of P oly(C) Deter m in ed by RP C5 Ch r om a togr a p h y
entry no.
[G]o, Ma
[T], Mb
rc
d[D]/dt (C),d M h-1
TD d[D]/dt,e M h-1
ref
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
0.005
0.008
0.010
0.015
0.020
0.030
0.040
0.045
0.010
0.015
0.020
0.030
0.040
0.020
0.005
0.020
0.040
0.020
0.008
0.045
0.020
0.050
0.050
0.050
0.050
0.050
0.050
0.050
0.050
0.030
0.030
0.030
0.030
0.030
0.020
0.010
0.010
0.010
0.035
0.025
0.005
0.002
0.88
0.88
0.91
0.92
0.91
0.90
0.87
0.88
0.92
0.91
0.92
0.92
0.90
0.92
0.89
0.93
0.92
0.91
0.87
0.93
0.93
1.0 × 10-6
2.9 × 10-6
9.0 × 10-6
2.9 × 10-5
7.8 × 10-5
2.4 × 10-4
5.3 × 10-4
8.1 × 10-4
8.7 × 10-6
3.2 × 10-5
6.2 × 10-5
2.2 × 10-4
4.6 × 10-4
5.7 × 10-5
3.3 × 10-7
5.2 × 10-5
3.4 × 10-4
6.2 × 10-5
3.3 × 10-6
4.0 × 10-4
2.0 × 10-5
8.8 × 10-7
2.6 × 10-6
8.5 × 10-6
2.8 × 10-5
7.6 × 10-5
2.4 × 10-4
5.2 × 10-4
8.0 × 10-4
8.2 × 10-6
3.1 × 10-5
6.0 × 10-5
2.2 × 10-4
4.5 × 10-4
5.5 × 10-5
2.1 × 10-7
5.0 × 10-5
3.3 × 10-4
6.0 × 10-5
3.0 × 10-6
3.9 × 10-4
1.8 × 10-5
f,g
f
f
f
f,g
f
f,g
f
g
g
g
g
g
Formal 2-MeImpG concentration. Poly(C) concentration in cytidine equivalents. c Factor r corrects for substrate purity (see ref 14).
a
b
Initial rate of 3′-5′-linked dimer formation in the presence of poly(C) (see Results and ref 6a). e Calculated rate of TD 3′-5′-linked dimer
d
formation (see Results). f Kinetic data from ref 6a. From ref 6b; unidentified, this work.
g
Discu ssion
Bim olecu la r Mech a n ism of 2-MeIm p G Dim er iza -
tion in Solu tion . Control experiments were performed
with 2-MeImpG alone as well as with 2-MeImpG in the
presence of polymers, such as poly(U), poly(I), and the
double-stranded poly(C)‚poly(G). The experiments with
poly(U) and poly(C)‚poly(G) were designed to test whether
or not 2-MeImpG dimerization is facilitated based on the
notion that guanosine forms a non-Watson-Crick base-
pair with U and two types of hydrogen-bonded triplets
with C‚G.12 The effect on the rate was tested at two
different concentrations of polymer. An additional reac-
tion was performed with 0.05 M KCl (instead of 0.05 M
polymer that is supplied as the potassium salt) at two
different monomer concentrations. None of these controls
exhibited noticeable differences compared to the results
obtained in the reaction of G alone. Therefore all these
controls were pooled together, listed in Table 1, and
plotted in Figures 4 and 5. It is likely that the scatter
seen in these figures is due to slight inhibition exhibited
by some of the polymers.
Table 1 indicates that at very low concentration of
monomer the rate of substrate disappearance is equal to
the rate of 5′GMP appearance, i.e. d(% H)/dt ) d(% G)/
dt ) 0.53 × 10-2 h-1 (average from the first three entries),
whereas at the higher concentrations d(% H)/dt < d(%
G)/dt ) 2.04 × 10-2 h-1 (average from the values at 0.053,
0.069, and 0.083 M monomer). The observed increase in
the rate of substrate disappearance with initial monomer
concentration, [G]o, could, in principle, be attributed to
the onset of dimerization that becomes important with
the higher concentration of monomer. However, Figure
4 indicates hydrolysis is also accelerated and therefore
partially responsible for the increased substrate disap-
pearance. The increase in the rate of hydrolysis is
surprising. It is interpreted to indicate a change in the
thermodynamics of the system, perhaps by increasing
stacking interactions12 in the range 0.005 M < [2-Me-
F igu r e 3. Plot of the concentration of 3′-5′-linked dimer, [D],
as a function of time obtained from a reaction with 0.040 M
2-MeImpG and 0.010 M (triangles), 0.030 M (diamonds), and
0.050 M (circles) poly(C). The lines are second-order fits to the
data, and d[D]/dt (C) was obtained from the initial slopes of
these plots.
of oligomers is equal to the concentration of D, if D was
the only product of the reaction. Hence [D] ) [pG3′pG] +
Σ[(pG)i], i g 3 where (pG)i are the 3′-5′-linked oligomers
of length i. This hypothesis is valid at early reaction
times, when the concentration of the oligomers is much
less than the concentration of the template, and hence
synthesis of oligomers by ligation is negligible. The yield
of oligoguanylates as a function of time was determined
by RPC5 chromatography. Figure 3 illustrates represen-
tative plots of [D] as a function of time. The slight
downward curvature is attributed to a combination of two
factors: the depletion of monomer and to the fact that,
with time, substantial amounts of oligomers are formed
which form stable complexes with poly(C) and therefore
inhibit the dimerization process. The slope of the lines
at zero time provides initial rates, henceforth referred
to as d[D]/dt(C) to indicate the presence of poly(C); such
slopes are listed in Table 3.
(12) Saenger, W. Principles of Nucleic Acid Structure; Springer-
Verlag: New York, 1984; for hydrogen-bonding see p 120, for stacking
interactions see p 134.

Poly(C)-Directed RNA Dimer Formation
J . Org. Chem., Vol. 64, No. 22, 1999 8329
(see Experimental Section, h ) 1.16 is the hypochromicity
factor8). From eqs 1 and 2 it follows that d(% Dall)/dt )
200/h kdr2[G]o. The linear relationship between the
experimentally determined d(% Dall)/dt and [G]o (shown
in Figure 5) is consistent with the proposed bimolecular
mechanism of dimerization. The straight line drawn to
fit the data in Figure 5 has a slope equal to 5.36 M-1
h-1 15 From this slope and r ) 0.86,14 one calculates kd )
2 3
.
4.2 × 10-2 M-1 h-1. In eq 3 kd , kd , kdp are the bimolecular
rate constants for formation of the 2′-5′-, the 3′-5′-, and
the pyrophosphate bond, respectively. From kd ) 4.2 ×
10-2 M-1 h-1 and the average values p/(2′+3′) ) 0.28 and
2′/3′ ) 4.17 of the product ratios reported in Table 1, one
obtains kd ) 6.3 × 10-3 M-1 h-1
.
3
d[Dall]/dt ) kd(r[G]o)2
(1)
(2)
d[Dall]/dt ) h/200[G]od(% Dall)/dt
2
3
p
kd ) kd + kd + kd
(3)
F igu r e 4. Rate of disappearance of 2-MeImpG (G) and
appearance of the hydrolysis product (H) as a function of total
monomer concentration, [G]o. d(% G)/dt, large markers; d(%
H)/dt, small markers; without polymer, circles; with polymer,
squares. The lines are second-order fits through the data; no
parameters are obtained from these fitting curves.
Syn t h esis of t h e R NA Dim er on t h e Tem p la t e.
Several lines of evidence summarized in Results lead to
the conjecture that the synthesis of RNA dimer is, for
all practical purposes, exclusively catalyzed by poly(C);
estimated lower limit of TD 3′-5′:2′-5′ ≈ 15-20 (see
Results). Values of d(% D)/dt, obtained as the difference
d(% G)/dt - d(% H)/dt (see Results, C18 method), are
listed in Table 2. From d(% D)/dt the rate of dimer
formation in the presence of poly(C), d[D]/dt (C), was
calculated via eq 2′ which is another version of eq 2 and
is also listed in Table 2.
d[D]/dt (C) ) h/200[G]od(% D)/dt
(2′)
Values d[D]/dt (C) determined with the RPC5 method
(see Results, RPC5 method) were pooled together with
values obtained from earlier measurements conducted
under identical conditions with the ones used here; they
are all reported in Table 3. In both Tables, 2 and 3, the
rate of dimer formation that occurs exclusively on the
template, TD d[D]/dt, was obtained by subtracting out a
small contribution coming from the dimerization that
3
occurs in solution (d[D]/dt ) kd (r[G]o)2 see eqs 1 and 3).
Taken together the two tables suggest a complex depen-
dence of the TD rate on both monomer and polymer
concentration. What is more surprising is that the
reactivity of this TD reaction varies by more than 5000-
fold with a 12-fold change in monomer concentration.
This is seen by comparing the lowest rate obtained in
this study (2.1 × 10-7 M-1 h-1 with [G]o ) 0.005 M and
[T] ) 0.01 M, see entry 15 in Table 3 where [T] stands
for poly(C) concentration expressed in monomer equiva-
lents) with the highest rate (1.1 × l0-3 M-1 h-1, see
entries 3, 4, and 7 in Table 2 with 0.051 M e [G]o e 0.060
M and 0.008 M e [T] e 0.014 M). The strong dependence
of the rate on monomer concentration confirms earlier
F igu r e 5. Rate of total dimer formation, d(% Dall)/dt in h-1
in the absence of poly(C) as a function of total monomer
concentration, [G]o, in the range 0.005 M e [G]o e 0.083 M;
all values from Table 1 are included.
,
ImpG] < 0.05 M. This proposition relies on the notion
that guanosine derivatives favor stacking interactions.12,13
The constancy of the product ratios p/(2′+3′) and 2′/3′
(see Experimental Section and Table 1) confirms the
simultaneous formation of these three dimers from two
molecules of 2-MeImpG.7 The experimentally determined
values of d(% Dall)/dt (see Results) are reported in Table
1. They are in good agreement with the calculated
difference d(% G)/dt - d(% H)/dt (not shown). A bimo-
lecular mechanism of dimerization is described by the
rate law given in eq 1 where r corrects for substrate
purity14 and kd is the bimolecular rate constant for the
formation of all three dimers. Equation 2 describes the
relationship between the total dimer concentration and
the percent yield obtained directly from the HPLC reports
(14) Factor r is the ratio of [G]/[G]o where [G] is the average value
for the monomer concentration between the initial and the endpoint
of the interval ∆t, and [G]o is the formal concentration of the substrate
determined by weight. Substrate purity was better than 97%. In
actuality r corrects for both the purity of the substrate as well as the
fact that the activated monomer is consumed during incubation. For
the reactions in the absence of poly(C) r ) 0.86 ( 0.04.
(15) Another interpretation of the data in Figure 5 is a curvilinear
behavior with a plateau at [G]o > 0.05 M that would support the
stacking interactions proposed earlier. Both interpretations coincide
at [G]o < 0.05 M.
(13) Zimmerman, S. B. J . Mol. Biol. 1976, 106, 663-672.

8330 J . Org. Chem., Vol. 64, No. 22, 1999
Kanavarioti et al.
results at a constant 0.05 M poly(C).6a In order to
rationalize this dependence, it is helpful to review the
conclusions drawn from another TD dimerization.
Recently we completed a kinetic study of the dimer-
ization of the deoxyguanosine 5′monophosphate 2-meth-
ylimidazolide, 2-MeImpdG, on poly(C) leading exclusively
to the synthesis of the pyrophosphate-linked dimer, dG5′-
ppdG.16 In the presence of poly(C), dG5′ppdG elongates
by incorporation of additional monomers and formation
of 3′-5′-internucleotide linkages to producedG5′ppdG-
(pdG)n.17 Accumulation of dG5′ppdG at the onset of
reaction was attributed to a slow elongation, most likely
due to the poorer nucleophilicity of deoxyribose compared
to the ribose. In contrast to the small yield of RNA dimer
detected here, the accumulation of the pyrophosphate
dimer allowed the kinetics to be determined directly from
the dimer yield.16 The highlights of that study were as
follows: First, the dependence of the rate of dG5′ppdG
formation as a function of 2-MeImpdG and poly(C)
concentration suggested a TD mechanism of dimeriza-
tion. Secondly, the kinetic behavior of the 2-MeImpdG/
poly(C) system as a function of monomer and polymer
concentration led to the formulation of a number of
principles governing TD chemistry. Thirdly, using a
simple cooperative model of template-monomer associa-
tion, the kinetic results were quantitatively correlated
and yielded binding constants for 2-MeImpdG on poly-
(C) and an intrinsic rate constant of TD dG5′ppdG
formation.16
of template-bound monomer, [G]tem, and total concentra-
tion of template sites, [T]. [G]f is the concentration of the
free monomer present in solution and θ is template
occupancy. The system is close to saturation when θ ≈ 1
and far from saturation when θ , 1. Another useful
parameter is the free monomer concentration at half
occupancy, [G]0.5θ, and it can be shown that [G]0.5θ
≈
1/Q.18,19
[G]o ) [G]tem + [G]f
θ ) [G]tem/[T]
(5)
(6)
When [G]f ) [G]0.50 then
[G]tem ) 0.5[T] and [G]o ) [G]0.5θ + [G]tem (7)
Some of the principles first established with the
2-MeImpdG/poly(C) system were exploited here to facili-
tate the experimental estimation of k2* and Q. (i) At a
constant [T] and as long as the system is close to
saturation, an increase in monomer concentration leaves
the rate of dimer formation unchanged. This is because
the additional monomer goes into solution. (ii) At a
constant [G]o and as long as the system stays close to
saturation, an increase in polymer concentration yields
a proportional increase in the rate of dimer formation.
This is because the additional template sites will be
occupied by G. In such case the rate of dimer formation
for a first-order TD reaction will be given by eq 8 where
k2* is the intrinsic rate constant of the process. (iii) When
[G]o . [T], departure from the anticipated proportionality
between TD d[D]/dt and [T] (eq 8) induced by a small
Coop er a tive Mod el for Tem p la te-Dir ected Dim er -
iza tion . The same model used for the above TD pyro-
phosphate synthesis is exploited here for the synthesis
of the RNA dimer. This model is summarized below. It
presumes that 2-MeImpG, G, binds cooperatively at C
sites of the template, T. Binding of G to an isolated site
on the template occurs with an association constant, q,
and binding of G adjacent to an occupied site occurs with
an association Q with Q > q and Q independent of the
length of G-bound molecules (eq 4). Q >> q implies that
the process is cooperative.18 As long as the relative
positioning of two molecules is right for internucleotide
bond formation, reaction can occur between any two
adjacent monomers within a stack of monomers hydrogen-
decrease in [G]o implies that θ < 1 and that [G]f ≈ [G]0.5θ
.
Under such conditions Q is estimated from eq 9 which is
derived from eq 7.
TD d[D]/dt ) k2*[G]tem and
when θ ) 1, TD d[D]/dt ) k2*[T] (8)
Q ) 1/[G]0.5θ ≈ 1/([G]o - 0.5[T])
(9)
Based on the above concepts, experiments were de-
signed in order to estimate k2* and Q values (Table 2).
For example, the first two entries in Table 2 at 0.055 M
and 0.060 M monomer with 0.005 M template exhibit
almost identical values of TD d[D]/dt (see principle i) and
provide an average TD d[D]/dt ) 6.65 × 10-4 h-1 which
yields k2* ) 0.13 h-1 from [T] ) 0.005 M and eq 8.
Similarly, the 3rd and 4th entries (at 0.008 M poly(C))
in Table 2 yield k2* ) 0.14 h-1, in excellent agreement
with the value of k2* obtained with 0.005 M poly(C) (see
principle ii). Furthermore, a small decrease in [G]o from
0.055 M to 0.050 M (5th entry in Table 2) slows down
the rate, indicating that the system is moving away from
saturation. The observation that with [G]o ) 0.045 M the
rate is half of the maximum obtained at saturation leads
to the conclusion that under these conditions the tem-
plate is approximately half occupied. These conditions
(6th entry in Table 1) yield an estimate for Q ≈ 1/0.041
) 24.4 M-1 from [T] ) 0.008 M and eq 9. In addition, it
is predicted from eq 8 and k2* ) 0.14 h-1 that a
“saturated” system with [T] ) 0.014 M should exhibit TD
d[D]/dt ) 1.96 × 10-3 h-1. On the basis of this calculated
rate, the conditions described by the 7th entry in Table
2 must be close to half-occupancy (from 1.96 × 10-3/1.11
2
3
m
bonded to the template. The rate constants k2 , k2 , k2
,
were defined as rate constants for formation of the dimer
from a stack containing two, three, or m monomers,
respectively. Initially, it is assumed that the rate of dimer
formation is independent of the length of the stack, so
2
that k2 ) k2*, where k2* is the intrinsic rate constant
for TD 3′-5′-internucleotide bond synthesis from two
3
molecules of 2-MeImpG. For statistical reasons k2
2k2*, k2 ) (m - 1)k2*, etc.
)
m
Equations 5-7 describe relationships between total or
formal monomer concentration, [G]o, the concentration
(16) Kanavarioti, A.; Gangopadhyay S. J . Org. Chem. 1999, 21,
7957-7964.
(17) Kozlov, I. A.; Orgel, L. E. Origins Life Evol. Biosphere 1999, in
press.
(18) Cantor, C. R.; Schimmel, P. R. Biophysical Chemistry; Freeman
and Co.: San Francisco, CA; Part III, 1980; p 864.

Poly(C)-Directed RNA Dimer Formation
J . Org. Chem., Vol. 64, No. 22, 1999 8331
F igu r e 6. Rate of dimer formation on the template, TD d[D]/
dt in M h-1, as a function of 2-MeImpG concentration, r[G]o,
at a constant poly(C) concentration, expressed in C equiva-
lents; squares at 0.008 M and circles at 0.005 M. TD d[D]/dt
values from Table 2. [G]o is the formal monomer concentration,
and r corrects for substrate purity.14 The lines are computer
simulations of a cooperative association model with q ) 2.2
M-1, Q ) 22.5 M-1 (eq 4) and an intrinsic rate constant for
TD dimerization k2* ) 0.18 h-1 assuming a TM3 mechanism
(see Discussion).
F igu r e 7. Rate of dimer formation on the template, TD d[D]/
dt in M h-1, as a function of 2-MeImpG concentration, r[G]o,
at a constant poly(C) concentration, expressed in C equiva-
lents; circles at 0.05 M, filled triangles at 0.03 M, and squares
at 0.01 M. TD d[D]/dt values from Table 3. [G]o is the formal
monomer concentration, and r corrects for substrate purity.14
The lines are computer simulations of a cooperative association
model with q ) 2.2 M-1, Q ) 22.5 M-1 (eq 4), and an intrinsic
rate constant for TD dimerization k2* ) 0.18 h-1 assuming a
TM3 mechanism (see Discussion).
× 10-3 ) 0.57) and provide Q ≈ 1/0.044 ) 22.7 M-1 via
eq 9 in good agreement with Q ≈ 24.4 M-1 obtained
above.
can be calculated from eqs 11 and 12. In eq 12 [T‚G2],
[T‚G3], [T‚G4], and [T‚Gm] are the concentrations of the
various stacks.6a,20
Com p u ter Sim u la tion of Mon om er Distr ibu tion
on th e Tem p la te. Figure 6 illustrates the kinetic data
obtained at 0.008 M and 0.005 M poly(C) (Table 2). The
slowest rate with 0.005 M poly(C) corresponds to the rate
listed as 20th entry in Table 3. The rates are plotted vs
r[G]o, to correct for substrate purity.14 Similarly, Figure
7 illustrates the effect of changing monomer concentra-
tion (r[G]o) at three different poly(C) concentrations, i.e.,
with 0.01 M (entries 15-17 in Table 3), 0.03 M (entries
9-13) and 0.05 M poly(C) (entries 1-8). It is worth
noticing that under these conditions, highly “unsatur-
ated” as will be shown later, the dependence of the rate
on poly(C) concentration is negligible between 0.05 M and
0.03 M and weak between these two families of data and
the family at 0.01 M poly(C). The rates here were plotted
as the log of the rate reflecting the large range in
reactivity. The lines in both figures are the result of a
computer simulation of a specific version (see below) of
the mechanism presented in eq 4. This simulation was
initially based on the values of Q and k2* obtained as
described above which were then further optimized to fit
all the data. The optimized values are Q ) 22.5 ( 1.5
M-1 and k2*) 0.18 ( 0.02 h-1. Together with these two
the value of q ) 2.2 ( 0.2 M-1 was used.19
2
3
4
k2 ) k2*; k2 ) 2k2*; k2 ) 3k2*;
m
k2 ) (m - 1)k2* (10)
TD d[D]/dt ) k2*FD (11)
FD ) [T‚G2] + 2[T‚G3] + 3[T‚G4] + ... + (m - 1)
[T‚Gm] (12)
Assuming M2 mechanism is operating and using eqs
11 and 12 rates, TD d[D]/dt were calculated for 0.005 M
e [G]o e 0.060 M and 0.002 M e [T] e 0.050 M (not
shown). While these calculated rates fit well data with
relatively high monomer concentration, they were ap-
proximately one order of magnitude faster than the
experimentally determined rates at low monomer con-
centration, i.e., most of the data in Table 3. Thus
mechanism M2 was discounted. The next mechanism that
we attempted to fit the data with was the M3 mechanism
(20) Calculations were executed using Microsoft Excel on a Power
Mac computer. The following is based on the description given in ref
6a. The concentration of T‚Gj species can be calculated from eqs 16
and 17 where [G]f is the free monomer concentration in solution, [T]f
is the concentration of template sites that are free and whose next-
neighbor sites on both sides are unoccupied, and [T] is the total
concentration of template sites (eq 18). [T]f is found by multiple
iteration carried out with Microsoft Excel’s Solver. In practice, calcula-
tion was initiated by a chosen [G]f, iteration provided the correct [T]f,
and then [G]o was obtained. For the simulation TD d[D]/dt was
obtained as TD d[D]/dt ) k2*FD and was plotted vs [G]o with r ) l.
In its simplest version the model assumes that various
template-bound monomer stacks have the same intrinsic
reactivity k2*. However, since in stacks containing more
than two monomers bond formation can occur between
any two neighbors, statistical corrections were applied.6a
The relationship between each rate constant and k2* for
this mechanism, so called “M2 mechanism”, is given by
eq 10. On the basis of this mechanism values TD d[D]/dt
[T‚G] ) q[T]f[G]f
(16)
(17)
(18)
[T‚Gj] ) q[T]f[G]fQj-1[G]fj-1
[T]f ) [T] - [G]tem - 2
[T‚Gj]
(19) Kanavarioti, A.; Hurley, T. B.; Baird, E. E. J . Mol. Evol. 1995,
41, 161-168, and references therein.
∑
j)1

8332 J . Org. Chem., Vol. 64, No. 22, 1999
Kanavarioti et al.
described by eqs 13 to 15. This mechanism implies that
the reaction is assisted by at least one next-neighbor
molecule. The fit with this mechanism is excellent, and
it is the one with which the lines were created in both
Figures 6 and 7.
Com par ison s with Ear lier Con clu sion s an d Oth er
System s. The value Q ) 22.5 M-1 for the association
constant at a site adjacent to an occupied site is a factor
of 9 lower than the value Q ) 180 M-1 obtained from the
binding isotherm for poly(C)/2-MeImpG under the exact
conditions used here.19 The binding isotherm was deter-
mined via hypochromicity measurements of poly(C)/2-
MeImpG mixtures. The disparity in Q determined from
binding and from kinetic studies may indicate that there
are a number of ways to construct a poly(C)/2-MeImpG
duplex, but that not all duplexes lead to dimerization.
Nevertheless, since Q ) 22.5 M-1 was obtained directly
from the kinetic data of this system, its usage here for
the evaluation of the model is justified. Interestingly, the
value q ) 2.2 M-1 obtained from the binding isotherm19
is the one that gives the best fit with the kinetic data in
this study (see Figures 6 and 7). An attempt to fit the
data in Tables 2 and 3 with values of q ) 2.2 M-1 and Q
2
3
4
m
k2 ) 0; k2 ) k2*; k2 ) 2k2*; k2 ) (m - 2)k2* (13)
TD d[D]/dt ) k2*FD
(14)
FD ) [T‚G3] + 2[T‚G4] + 3[T‚G5] + ... +
(m - 2)[T‚Gm] (15)
2
3
The corresponding M4 mechanism (with k2 ) k2 ) 0,
4
m
k2 ) k2* and k2 ) (m - 3)k2*) was tested and found
unsatisfactory because the calculated rates were about
1 order of magnitude lower than the experimental rates
obtained at low [G]o. The observation that all mechanisms
fit the data at the high concentrations is due to the fact
that at high [G]o the template is quite full and, conse-
quently, the concentration of the small stacks, i.e., [T‚
G2], [T‚G3], [T‚G4], is negligible. Therefore experiments
at low occupancy, i.e., with predominantly short stacks,
provide a good test for the elucidation of the specific
mechanism. Other mechanistic possibilities, such as a
) 180 M-1, instead of q ) 2.2 M-1 and Q ) 22.5 M-1
,
was unsuccessful. This is because the first set of con-
stants predicts a factor of two lower rates for the 0.03 M
family compared to the 0.05 M family which is not borne
out by the data. The earlier study of the kinetics of
dimerization in the range 0.005 M < [G] < 0.045 M at
0.05 M poly(C), used values q ) 2.2 M-1 and Q ) 180
M-1 to determine product distribution of the various
stacks.6a Based on this distribution, the conclusion was
drawn that the process of dimerization is optimal in a
stack of six or more template-bound molecules with k2*
) 0.018 h-1. Although the basic features of the model
(eq 4) are intact, the revised Q ) 22.5 M-1 leads to the
conclusion that the process of dimerization is optimal in
a stack of only three or more template-bound molecules
with a 10-fold faster intrinsic rate constant of dimeriza-
tion (k2* ) 0.18 h-1). Consequently, the dependence of
the rate on template concentration together with the
evaluation of Q directly from the kinetics led to an
important refinement of the mechanistic and kinetic
interpretations of the original model.22
Reaction of 2-MeImpdG with guanosine 5′-phosphoryl-
morpholinamide (mor-pG) in the presence of poly(C) leads
to the formation of the two isomeric dimers mor-pG3′pdG
and mor-pG2′pdG.23 The dimerization kinetics in this
system led to the conclusion that the template catalyzes
dimerization as well as increases regioselectivity toward
the RNA bond formation; the effect was small (about
4-fold) in both cases. The corresponding intrinsic rate for
TD RNA bond formation was determined kd* ) 4.1 × 10-3
2
3
m
mechanism with k2 < k2 ) k2* and k2 ) (m - 2)k2*,
although probable, were not tested because of the in-
creased number of unknown parameters involved. Thus
the proposed M3 mechanism is the simplest mechanism
consistent with the data (see Scheme 1).
In sigh ts in Tem p la te-Dir ected Ch em istr y. In ad-
dition to the principles mentioned earlier,16 the present
study led to following insights. (i) The best experimental
conditions to obtain reliable kinetic data are the ones
with [G]o ≈ [T]. This is because when [G]o ≈ [T] it takes
a lot of monomer to occupy the template. Thus small
changes in [G]o do not alter the occupancy dramatically
and with it the rates, in contrast to the situation when
[G]o . [T]. (ii) It is seen that an approximate doubling of
poly(C) concentration leaves the rates of the process
unchanged (0.05 M/0.03 M, see Figure 7). This negligible
dependence of the dimerization rate on [T] can be
mistakenly interpreted as absence of catalysis. This
phenomenon occurs because with highly “unsaturated”
systems, such as the ones with 0.05 M and 0.03 M
template, the increase in template concentration is
counterbalanced by a comparable decrease in occupancy
leading to negligible rate effects. (iii) The dimerization
of 2-MeImpG in the absence of poly(C) was shown to be
consistent with a bimolecular mechanism. The low
solubility of 2-MeImpG did not allow experimentation
under higher concentrations, i.e., [G]o > 0.1 M, where one
may anticipate a switch from second-order to first-order
kinetics consistent with extensive stacking.12,13,15 In the
presence of poly(C) the kinetic behavior is more complex.
Specifically, the rate of dimer formation exhibited a third
order dependence at very low monomer/high template
concentrations, followed by a second order and then a
first-order dependence (linear part in Figure 6) with
increasing occupancy, followed by a zero order depen-
dence at saturated conditions ([G]o > 0.05 M, plateau in
Figure 6 with 0.005 M poly(C)). This rich kinetic behavior
is a composite of the kinetics exhibited under conditions
close and far from saturation in conjuction with the
mechanistic prerequisite that efficient dimerization in the
poly(C)/2-MeImpG system takes place within a stack of
three or more monomers.21
(21) A critical evaluation of the numerical values of the constants
obtained from the TD model is in order. First, it should be recognized
that eq 4 represents the simplest possible model to describe a TD
reactive system. The equilibria involved in this model are described
by only two association constants, q and Q, and by a mechanism that
limits the reactivity of two adjacent molecules to be either equal to
zero or equal to the intrinsic rate constant k*. On first inspection, the
task is to fit a set of data with four “parameters”. In actuality this is
not so. It is seen that Q and k* can be estimated directly from
specifically designed experiments with systems close to saturation
following certain rules.16 Moreover, the value of q and the specific
mechanism are both derived from the kinetics with systems at very
low template occupancy. Therefore, if the experiments cover a large
range of reactivity from highly unsaturated to saturated systems, then
the obtained numerical values for q, Q, k* and the choice of mechanism
are, within the simple model, well evaluated.
(22) The argument can be made that the model for TD oligoguany-
late elongation6c should be reevaluated by using the constants q ) 2.2
M-1 and Q ) 22.5 M-1, instead of q ) 2.2 M-1 and Q ) 180 M-1
.
However, we have no evidence, kinetic or otherwise, that Q ) 22.5
M-1 derived from the dimerization kinetics is a more reliable constant
for the elongation reaction and hence a reevaluation will be postponed
until a better understanding of the binding process is in place.
(23) Kanavarioti, A. J . Org. Chem. 1998, 63, 6830-6838.

Poly(C)-Directed RNA Dimer Formation
J . Org. Chem., Vol. 64, No. 22, 1999 8333
h-1. In view of k2* ) 0.18 h-1 from 2-MeImpG/poly(C),
the low kd* value determined with 2-MeImpG/mor-pG/
poly(C) allows one of two conclusions to be drawn. Either
the conditions under which k2* ) 4.1 × 10-3 h-1 obtained
in the 2-MeImpG/mor-pG/poly(C) were not “saturated”
as presumed, or the configuration within the stack of
mor-pG/2-MeImpdG is much less suited for reaction
compared to the one in a stack of three 2-MeImpG
molecules. The fact that 2-MeImpG has two reactive sites
vs one each for 2-MeImpdG and mor-pG can only account
for approximately 4-fold out of an approximately 40-fold
different reactivity. Insights may come from molecular
dynamics simulations (MD). Preliminary MD studies of
a duplex between (pC)10 and 10 2-MeImpG molecules
suggest (data not shown) that 2-MeIm moieties of neigh-
boring monomers are partially stacked on each other and
form a third strand that wraps around the duplex formed
by the polymer and the stacked guanosine monomers.24
The third strand formed by the 2-MeIm moieties may
then be responsible for a more rigid structure leading to
a highly selective and efficient dimerization and oligo-
merization with poly(C)/2-MeImpG. The absence of such
a stabilizing factor may lead to reduced catalysis and
regioselectivity in the poly(C)/2-MeImpdG/mor-pG23 and
poly(C)/ImpG systems.4d,10
Association constants q ) 7 M-1 and Q ) 9.5 M-1 in
the dimerization reaction with the 2-MeImpdG/poly(C)
system were determined in a similar way as here.
Comparison between these two sets of constants indicates
that the deoxyribo derivative associates more strongly
at an isolated site of the template and more weakly at a
site adjacent to an occupied one compared to the ribo
derivative. The conclusion that Q of 2-MeImpdG is
smaller than Q of 2-MeImpG is reasonable, since it is
known that the affinity decreases in the order RNA‚RNA
> RNA‚DNA > DNA‚DNA.12 The reasons for the reversal
in the corresponding values of q are unclear. Evidently,
poly(C) catalyzes dimerization with high efficiency, re-
gioselectivity, and specificity (dG5′ppdG, from 2-MeImpdG
and pG3′pG from 2-MeImpG). The value of k2* ) 0.18 h-1
determined here for TD RNA dimer synthesis should be
compared with the corresponding kp* ) 0.055 h-1 for
pyrophosphate formation in the 2-MeImpdG/poly(C) sys-
tem. This comparison indicates that in the presence of
the template the known higher reactivity25 of pyrophos-
phate vs internucleotide bond synthesis is reversed,
suggesting that the poly(C)/2-MeImpG complex is better
suited for 3′-5′- than either 2′-5′- or pyrophosphate-bond
formation. The recent discovery that hexitol nucleic acids,
which also exhibit an A structure like RNA, are also
efficient templates for polynucleotide synthesis with
2-MeIm-activated nucleotides points to the importance
of the A structure among functional templates.11
h-1 in solution and d[D]/dt (C) ) 9 × 10-3 M-1 h-1 on the
template. These conditions yield a 150-fold catalysis and
represent the minimum; the maximum is better than
5200 (Discussion and Table 3). In addition, the rate of
hydrolysis is 5.3 × 10-3 h-1 for both systems compared
with the rate of dimerization on a saturated template (k2*
) 0.18 h-1) renders the TD dimerization under optimal
conditions 34 times more efficient than hydrolysis. With
such a highly efficient 3′-5′ dimerization, it is not surpris-
ing that elongation to form oligoguanylates is also
efficient.4b,6 Hence the conjecture is made that dimeriza-
tion represents a critical step in a TD polymerization and,
perhaps, one that indicates whether or not the specific
system under the conditions tested is well suited for TD
chemistry.
Con clu sion s
The rate of 2-MeImpG dimerization to form the RNA-
linked dimer has been obtained as a function of monomer
and of poly(C) concentration. Rate comparisons suggest
that poly(C) selectively catalyzes the synthesis of the 3′-
5′-linked ribodiguanylate up to 5200-fold over the solution
reaction. The dramatic effect of poly(C) on 2-MeImpG
dimerization is contrasted with the undetectable effect
of other polynucleotides that can not exploit Watson-
Crick base-pairing. Dimerization in solution exhibits a
simple second-order dependence on monomer concentra-
tion. In the presence of poly(C) and at very low template
occupancy, the rate of RNA-dimer formation exhibits a
third-order dependence on 2-MeImpG concentration fol-
lowed by a second-order, then a first-order with increas-
ing occupancy, and finally a zero-order dependence under
saturated conditions. This behavior is quantitatively
consistent with a TD mechanism of dimerization where
efficient dimerization occurs within stacks consisting of
three or more, but not two, monomers. The surprisingly
small dependence of the rate of dimer formation on
template concentration, when [G] ≈ [T], can be attributed
to extremely low template occupancy. Rates of dimeriza-
tion have been fitted to a simple cooperative model for
poly(C)/2-MeImpG complexation and afforded an associa-
tion constant at an isolated site q ) 2.2 M-1 and adjacent
to an occupied site Q ) 22.5 M-1, and an intrinsic rate
constant of TD dimerization k2* ) 0.18 h-1. This surpris-
ingly high rate constant suggests that strong catalysis
and practically exclusive regioselectivity in the formation
of the RNA dimer is exhibited within the duplex con-
structed from poly(C) and 2-MeImpG stacks. A better
understanding of the factors involved in such catalysis
could facilitate the design of efficient TD, nonenzymatic,
polynucleotide syntheses with unprotected reactive mono-
mers in water.
Ack n ow led gm en t. This research was supported by
the Exobiology Program of the National Aeronautics and
Space Administration (Grant No. NCC 2-534). We thank
Dr. L. E. Orgel from the Salk Institute for discussions
and for providing the RPC5 packing. We also thank Dr.
Mark Fonda from the Planetary Biology Branch of
NASA/Ames Research Center and Prof. C. F. Bernasconi
from the Chemistry and Biochemistry Department of
the University of California at Santa Cruz for providing
the facilities and Prof. C. F. Bernasconi for helping us
with the kinetic analysis and for reviewing the manu-
script.
To appreciate the catalytic effect of the template on
the dimerization, it may be fair to compare the rate of
dimerization under conditions that are optimal for dimer-
ization in solution (0.1 M 2-MeImpG) and on the template
(0.05 M template/0.1 M 2-MeImpG). These conditions are
based on the upper solubility limits of monomer and
polymer in water. One calculates that the rate of 3′-5′-
linked dimer formation will be d[D]/dt ) 6.3 × 10-5 M-1
(24) Cieplak, P.; Kanavarioti, A. Unpublished results.
(25) Rodriguez, L.; Orgel L. E. J . Mol. Evol. 1991, 32, 101-104.
Kanavarioti, A.; Rosenbach, M. T. J . Org. Chem. 1991, 56, 1513-1521.
Dolinnaya, N. G.; Tsytovich, A. V.; Sergeev, V. N.; Oretskaya, T. S.;
Shabarova, Z. A. Nucl. Acids Res. 1991, 19, 3073-3080.
J O991216Q