Efforts toward the expansion of the genetic alphabet: Information storage and replication with unnatural hydrophobic base pairs

Article abstract of DOI:10.1021/ja9940064

The faithful recognition of the interstrand hydrogen bonds between complementary nucleobases forms the foundation of the genetic code. The ability to replicate DNA containing a stable third base pair would allow for an expansion of the information content

Full text of DOI:10.1021/ja9940064

3274
J. Am. Chem. Soc. 2000, 122, 3274-3287
Efforts toward the Expansion of the Genetic Alphabet: Information
Storage and Replication with Unnatural Hydrophobic Base Pairs
Anthony K. Ogawa, Yiqin Wu, Dustin L. McMinn, Jianquan Liu, Peter G. Schultz,* and
Floyd E. Romesberg*
Contribution from the Department of Chemistry, The Scripps Research Institute,
10550 N. Torrey Pines Road, La Jolla, California 92037
ReceiVed NoVember 15, 1999
Abstract: The faithful recognition of the interstrand hydrogen bonds between complementary nucleobases
forms the foundation of the genetic code. The ability to replicate DNA containing a stable third base pair
would allow for an expansion of the information content of DNA by supplementing the existing two base
pairs of the genetic alphabet with a third. We report the optimization of unnatural nucleobases whose pairing
in duplex DNA is based on interbase hydrophobic interactions. We show that the stability and selectivity of
such unnatural base pairs may be comparable to, or even exceed, that of native pairs. We also demonstrate
that several unnatural base pairs are incorporated into DNA by Klenow fragment of Escherichia coli DNA
polymerase I with an efficiency equivalent to that of native DNA synthesis. Moreover, the unnatural bases are
orthogonal to the native bases, with correct pairing being favored by at least an order of magnitude relative to
mispairing.
1. Introduction
incorporation into DNA by a polymerase, kinetic selectivity
against enzymatic mispairing with a natural base (kinetic
orthogonality), and the efficient continued primer extension after
synthesis of the nascent unnatural base pair.
Biological information storage and replication is based on a
genetic alphabet encoded by the specific Watson-Crick hy-
drogen bonding (H-bonding) patterns of the adenine:thymine
(A:T) and guanine:cytosine (G:C) base pairs. The faithful
recognition of these H-bonds underscores virtually all nucleic
acid biochemistry, including DNA structure and replication.1
We are interested in increasing the information content of DNA
with unnatural nucleic acids, where the pairing of the unnatural
bases is driven by hydrophobic interactions instead of H-
bonding. We have recently shown that hydrophobic interactions
can thermodynamically compensate for the interstrand Watson-
Crick H-bonds in duplex DNA.2 Such bases would represent
an expanded genetic alphabet, provided a polymerase is capable
of efficient and high fidelity synthesis of DNA containing the
unnatural pair. The ability to replicate DNA containing a third
base pair would also allow for the enzymatic incorporation of
additional functional moieties into DNA (e.g., catalytic groups,
fluorophores, spin labels, etc.), allowing for increased func-
tionality in in vitro selection experiments. In addition, the
availability of a number of orthogonal base pairs would facilitate
hybridization or encoding experiments in cases where natural
sequences cross hybridize.
Previous efforts to generate orthogonal base pairs have relied
on H-bonding patterns which are not found with the canonical
Watson-Crick pairs. For example, the noncanonical H-bonding
base pairs formed between dκ and dX3-6 or d-iso-C and d-iso-
G7-11 adopt standard Watson-Crick geometry, but have an
H-bonding pattern unlike that found in the natural base pairs,
dG:dC and dA:dT. When incorporated into B-form DNA or
DNA-RNA hybrid duplexes, these unnatural base pairs display
moderate stability and selectivity.4,5,12,13 Moreover, no DNA
polymerase has been identified that can incorporate these base
pairs into DNA with high efficiency and fidelity both in the
template and as incoming triphosphate. In all cases, the modified
bases were not kinetically orthogonal and instead competitively
paired with natural bases during polymerase-catalyzed DNA
synthesis.3-5,9,11,14 Tautomeric isomerism, which would alter
(3) Lutz, M. J.; Held, H. A.; Hottiger, M.; Hu¨bscher, U.; Benner, S. A.
Nucleic Acids Res. 1996, 24, 1308-1313.
(4) Horlacher, J.; Hottiger, M.; Podust, V. N.; Hu¨bscher, U.; Benner, S.
A. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 6329-6333.
(5) Piccirilli, J. A.; Krauch, T.; Moroney, S. E.; Benner, S. A. Nature
1990, 343, 33-37.
(6) Bain, J. D.; Switzer, C.; Chamberlin, A. R.; Benner, S. A. Nature
1992, 356, 537-539.
(7) Sugiyama, H.; Ikeda, S.; Saito, I. J. Am. Chem. Soc. 1996, 118, 9994-
9995.
Essential thermodynamic requirements for third base pair
candidates include stable pairing in duplex DNA and thermo-
dynamic selectivity against mispairing with a natural base
(thermodynamic orthogonality). The stability and selectivity of
an unnatural base pair should be of the same order of magnitude
as those for native base pairs. Essential kinetic properties for
the third base pair candidates include efficient enzymatic
(8) Tor, Y.; Dervan, P. B. J. Am. Chem. Soc. 1993, 115, 4461-4467.
(9) Lutz, M. J.; Horlacher, J.; Benner, S. A. Bioorg. Med. Chem. Lett.
1998, 8, 499-504.
(10) Switzer, C.; Moroney, S. E.; Benner, S. A. J. Am. Chem. Soc. 1989,
111, 8322-8323.
* To whom correspondence should be adressed. Telephone: (858) 784-
7290. Fax: (858) 784-7472. E-mail: floyd@scripps.edu.
(1) Kornberg, A.; Baker, T. A. DNA Replication, 2nd ed.; W. H. Freeman
and Company: New York, 1992.
(2) McMinn, D. L.; Ogawa, A. K.; Wu, Y.; Liu, J.; Schultz, P. G.;
Romesberg, F. E. J. Am. Chem. Soc. 1999 121, 11585-11586.
(11) Switzer, C. Y.; Moroney, S. E.; Benner, S. A. Biochem. 1993, 32,
10489-10496.
(12) Horn, T.; Chang, C.-A.; Collins, M. L. Nucleosides Nucleotides
1995, 14, 1023-1026.
(13) Roberts, C.; Bandaru, R.; Switzer, C. J. Am. Chem. Soc. 1997, 119,
4640-4649.
10.1021/ja9940064 CCC: $19.00 © 2000 American Chemical Society
Published on Web 03/24/2000

Expansion of the Genetic Alphabet
J. Am. Chem. Soc., Vol. 122, No. 14, 2000 3275
Scheme 1a
a
Nucleobase is labeled R, and protecting groups are labeled R′,
R′′, and R′′′ (see Experimental Section).
formation (Scheme 1). In the first, chloroglycoside 821 served
as a general electrophile for aryl Grignard additions and nitrogen
nucleophiles.22 Although Grignard addition to 8 generally
favored the R-anomer, acid treatment of the corresponding
benzylic ethers in refluxing xylene resulted in epimerization at
C1′ to yield the â-anomer. This methodology, previously
employed by Kool and co-workers,22 was amenable to the large-
scale nucleoside synthesis. The nucleophilic addition of indoles
to 8, by contrast, afforded the desired â-anomer, exclusively,
which allowed the facile synthesis of azaindole nucleosides. The
syntheses of aryl nucleosides 1 and 2 proceeded from condensa-
tion of in situ generated Grignard reagents with 8, followed by
epimerization and methoxide-mediated deprotection (Scheme
2).22
For some nucleosides, Grignard formation or condensation
with 8 proved difficult. In these cases an alternate synthetic
strategy was used that involved alkylation of an aldehyde
substrate (15 and 20), followed by cyclization to the nucleoside
(Scheme 1).23,24 The seven-step synthetic route to 15 from
D-ribose followed literature precedent and did not require the
generation of any stereocenters beyond C1′.24 A separate, five-
step route route to C-nucleosides from aldehyde 20 again
followed from literature precedent,23 in which C3′ was generated
from Felkin-Anh addition of an allylzinc nucleophile to
isopropylidine-protected glyceraldehyde with high diastereose-
lectivity. Mesylation, followed by treatment with excess tri-
fluoroacetic acid, afforded a diastereomeric mixture of separable
C-nucleosides.
Figure 1.
H-bond donor and acceptor patterns, likely contributes to this
kinetic infidelity.13,15,16
As an alternate approach we have been evaluating hydro-
phobic bases that are incapable of tautomerization.2 Hydropho-
bicity should be a strong and selective force in the DNA duplex,
based on the known importance of the hydrophobic effect in
the folding and stability of proteins.17 Desolvation of a native
base when paired with an unnatural hydrophobic base should
significantly disfavor mispairing. Moreover, the structure of the
hydrophobic bases may be optimized to maximize interstrand
packing interactions. We report here a detailed examination and
optimization of interstrand hydrophobic packing in a series of
hydrophobic base pairs with the intent of establishing a system
of information storage, replication, and retrieval that is orthogo-
nal to the natural H-bond-based system.
2. Results
1.1. Base Pair Design Considerations. In an effort to design
two simple hydrophobic moieties that pack well when paired
opposite one another in duplex DNA, several substituted phenyl
rings were modeled as nucleobases18 in B-form DNA.19 Several
bases were identified that could be accommodated without
significant distortion of the helix. The trimethylphenyl nucleo-
side:dimethylphenyl nucleoside (TM:DM) pair was found to
have the lowest calculated strain energy and was therefore
chosen as the experimental starting point (Figure 1). To optimize
the stability and selectivity of the unnatural pair, a variety of
TM and DM derivatives were also synthesized. Modifications
included increased aromatic surface area, alkyl group substit-
uents, and the inclusion of minor groove H-bond acceptors
(Figure 1). All of these bases have been evaluated with respect
to their thermodynamic base pair stability in duplex DNA and
the kinetics of enzymatic DNA replication.
Naphthalene-substituted nucleosides 3 and 4 were derived
from the condensation of the aryllithium species onto aldehydes
15 and 20, respectively (Schemes 3 and 4).23,24 In situ cyclization
of the mesylate afforded a ∼1:1 mixture of diastereomers from
which 16 was isolated. Detritylation, followed by treatment with
DDQ, afforded nucleoside 3 in 55% yield. An alternative method
for deprotection of the PMB-group using trityl tetrafluoroborate
proceeded in comparable yield. The synthesis of the 2-meth-
ylnaphthalene nucleoside 4 involved the two-step mesylation-
acid treatment previously used by Hopkins and co-workers and
afforded 4 in 24% yield.
2.2. Base Pair Synthesis. The syntheses of nucleosides 1-720
involved two strategies for stereoselective glycosidic bond
Vorbruggen glycosylation methodology was used to synthe-
size nucleoside 23, as it generally resulted in N-glycosylation
to exclusively yield the â-anomer.25 However, in the case of
23, the reaction produced a 1:1 mixture of separable diastere-
(14) Lutz, M. J.; Horlacher, J.; Benner, S. A. Bioorg. Med. Chem. Lett.
1998, 8, 1149-1152.
(15) Roberts, C.; Chaput, J. C.; Switzer, C. Chem. Biol. 1997, 4, 899-
908.
(21) Takeshita, M.; Chang, C.-N.; Johnson, F.; Will, S.; Grollman, A.
P. J. Biol. Chem. 1987, 262, 10171-10179.
(22) Schweitzer, B. A.; Kool, E. T. J. Org. Chem. 1994, 59, 7238-
7242.
(23) Solomon, M. S.; Hopkins, P. B. J. Org. Chem. 1993, 58, 2232-
2243.
(24) Eaton, M. A. W.; Millican, T. A. J. Chem. Soc., Perkin Trans. 1
1988, 545-547.
(25) Niedballa, U.; Vorbruggen, H. J. Org. Chem. 1974, 39, 3654-3660.
(16) Robinson, H.; Gao, Y.-G.; Bauer, C.; Roberts, C.; Switzer, C.; Wang,
A. H.-J. Biochemistry 1998, 37, 10897-10905.
(17) Dill, K. A. Biochemistry 1990, 29, 7133-7155.
(18) These are obviously not basic, h., we refer to all of them as “bases”
or “nucleobases” by analogy to their natural counterparts.
(19) Modeling was done with Insight II and Discover from Biosym/
MSI of San Diego, CA.
(20) C1′-stereochemistry assigned from 2D-COSY/NOESY assignments.

3276 J. Am. Chem. Soc., Vol. 122, No. 14, 2000
Ogawa et al.
Scheme 2a
a
(a) Mg0, then 8; (b) NaOMe; (c) DMTr-Cl, NEt3; (d) Cl-P(NiPr2)(OCH2CH2CN), NEt3; (e) POCl3, 0 °C, then Bu3N-PPi.
omers (Scheme 5). Subsequent purification and deprotection
yielded 5 in 89% overall yield. Our synthesis of hydrophobic
purine analogue 7 again took advantage of the versatile
chloroglycoside 8. Following literature procedure,26 the addition
of the 7-azaindole sodium salt to 8 afforded a single diastere-
omer in 60% yield. Removal of the toluoyl groups with sodium
methoxide subsequently produced the desired nucleoside 7.
The assignment of â-stereochemistry at C1′ for each free
nucleoside was based on NOESY data, in which H1′ showed
otides, 5′-GCGATGXGTAGCG-3′ and 5′-CGCTACYCATCGC-
3′. The stability of the base pairs, as well as the mispairs with
natural bases, was determined in this sequence context by duplex
melting experiments. We use the terms “thermal stability” and
“thermodynamic stability” interchangeably to refer only to the
duplex stability as measured by the duplex melting temperature
(Tm ).
2.3.1. Stability and Thermodynamic Orthogonality of the
DM:TM Unnatural Hydrophobic Base Pair. Initially, we
evaluated the simple methyl-substituted phenyl bases, TM and
DM. Duplex DNA containing the TM:DM, TM:TM, and DM:
DM unnatural base pairs was less stable than that containing a
dA:dT pair by 4.0-6.4 °C (Table 1). Despite this decrease in
pairing stability, the hydrophobic bases displayed thermal
selectivity against mispairing with any of the native bases (Table
1). For both TM and DM, the mispairs with dG, dA, and dT
were all significantly less stable, and the mispair with dC was
the least stable. The thermodynamic preference for the TM:
TM, DM:DM and the TM:DM pairs relative to the most stable
mispairs ranged from 2.8 to 5.2 °C. For comparison, in the same
duplex the thermodynamic orthogonality of the dA:dT base pairs
relative to mispairs with G, A, T, and C was 3.8-10.5 °C.
1
cross-peaks with both H4′ and R-H2′.27 In addition, 1-D H
NMR splitting patterns for H1′ generally conformed to literature
precedent in which a doublet of doublets indicated â-stereo-
chemistry for C-nucleosides, and a pseudo-triplet indicated
likewise for N-nucleosides. In all cases, conversion of free
nucleosides to the corresponding phosphoramidites utilized
standard literature procedures.28 In addition, nucleosides were
converted to triphosphates following the methodology of Otvos
(Schemes 6 and 7).29
2.3. Stability of Unnatural Hydrophobic Base Pairs. To
evaluate the thermodynamic stability of these novel hydrophobic
bases they were incorporated into complementary oligonucle-
(26) Seela, F.; Gumbiowski, R. Heterocycyles 1989, 29, 795-805.
(27) Assignment of R-H2′ was based on a smaller corresponding cross-
peak to H3′.
2.3.2. Stability and Thermodynamic Orthogonality of
Unnatural Nucleobases with Increased Hydrophobic Surface
Area. In an effort to increase the overall stability of the
hydrophobic base pairs, while maintaining or increasing ther-
(28) Schweitzer, B. A.; Kool, E. T. J. Am. Chem. Soc. 1995, 117, 1863-
1872.
(29) Kovacs, T.; Otvos, L. Tetrahedron Lett. 1988, 29.

Expansion of the Genetic Alphabet
J. Am. Chem. Soc., Vol. 122, No. 14, 2000 3277
Scheme 3a
Scheme 5a
a
a
(a) nBuLi, -78 °C, then 15; (b) MsCl, NEt3, pyridine, 0 °C; (c)
(a) TMSNdC(OTMS)CH3, then 8, SnCl4; (b) NaOMe; (c) DMTr-
Cl, NEt3; (d) Cl-P(NiPr2)(OCH2CH2CN), NEt3; (e) POCl3, 0 °C, then
Bu3N-PPi.
AcOH; (d) DDQ; (e) DMTr-Cl, NEt3; (f) Cl-P(NiPr2)(OCH2CH2CN),
NEt3; (g) POCl3, 0 °C, then Bu3N-PPi.
Scheme 4a
Scheme 6a
a
(a) nBuLi, -78 °C, then 20; (b) MsCl, 0 °C; (c) TFA; (d) POCl3,
0 °C, then Bu3N-PPi; (e) DMTr-Cl, NEt3; (f) Cl-P(NiPr2)(OCH2CH2CN),
NEt3.
modynamic selectivity, we designed analogues of DM and TM
with increased aromatic surface area. These analogues, di-
methylnaphthyl nucleoside (DMN) and 2-methylnaphthyl nu-
cleoside (2MN), were designed to probe the effect of increased
aromatic surface area in different regions of the major groove.
Molecular modeling studies suggested that the aromatic ring
of DMN should project into the major groove toward the pairing
a
(a) ICl; (b) CuI, Pd(II)(PPh3)2Cl2, propyne, NEt3; (c) NaOMe; (d)
DMTr-Cl, NEt3; (e) Cl-P(NiPr2)(OCH2CH2CN), NEt3; (f) POCl3, 0 °C,
then Bu3N-PPi.
base, in a fashion similar to the methyl group of DM.30
Additionally, to prevent syn-anti isomerization around the
glycosidic bond, DMN has a methyl group at C2 that should

3278 J. Am. Chem. Soc., Vol. 122, No. 14, 2000
Ogawa et al.
Scheme 7a
suggested that there is a correlation between increased aromatic
surface area and duplex stability (Table 2). The 2MN:DM and
2MN:TM base pairs were more stable than the TM:DM and
TM:TM pairs by approximately 1.9 and 1.1 °C, respectively.
Unlike DMN, a comparison between the 2MN pairs with other
hydrophobic bases and mispairs with native bases showed that
2MN retained the previously observed degree of thermodynamic
selectivity (∆Tm ) 2-5 °C), while improving the overall duplex
stability.
2.3.3. Stability and Thermodynamic Orthogonality of
Unnatural Nucleobases with Increased Polarizability. The
design of the isocarbostyril nucleoside (ICS) paralleled that of
DMN, with the notable exception of an N-glycosidic linkage
and a minor groove carbonyl group provided by the endocyclic
amide moiety.30 Comparison of ICS with its carbocyclic
analogue DMN suggests that the increased polarity provided
by the amide functionality resulted in a marginal (opposite DM)
to significant (opposite TM and DMN) increase in duplex
stability (Table 3). The ICS:ICS self-pair had a Tm equal to
that of a dA:dT pair (59.3 °C for ICS:ICS vs 59.2 °C for dA:
dT). In comparison to the ICS self-pair, the hybrid base pair,
DMN:ICS, was slightly less stable (∆Tm ) -1 °C) and provided
a measure of the thermodynamic contribution of the amide. Due
to the stability of the ICS:dA mispair (55.1 °C), the thermo-
dynamic orthogonality of the unnatural base pairs between ICS
and DM and TM is compromised. However, the observed
thermodynamic selectivity favoring the ICS pair with itself or
with DMN was found to be 4.2 and 3.2 °C, respectively.
Our design of PICS capitalized on the stabilizing effect that
C5-hydrophobic substituents are known to have with dC and
dT.31 However, the C5-propynyl group of PICS was found to
have variable effects on duplex stability. When paired with the
DM-like ring structures (DM and DMN), the propynyl group
stabilized the resulting base pairs: the PICS:DM and the PICS:
DMN pairs were each 1.1 °C more stable than their corre-
sponding ICS pairs, ICS:DM, and ICS:DMN, respectively.
However, when paired with the TM-like ring structures (TM
and 2MN), the propynyl group destabilized the resulting base
pairs: the PICS:TM and PICS:2MN base pairs were 0.7 and
0.4 °C less stable than the ICS:TM and the ICS:2MN pairs,
respectively. The propynyl substituent led to increased stability
in the context of the PICS:PICS self-pair, which was 3.3 °C
more stable than the corresponding ICS:ICS self-pair. Signifi-
cantly, the PICS:PICS self-pair was more stable than both the
dA:dT and dG:dC base pairs (Tm ) PICS:PICS, 62.6 °C; dA:
dT, 59.2 °C; dG:dC, 61.8 °C). Moreover, the selectivity of the
a
(a) DMTr-Cl, NEt3; (b) Cl-P(NiPr2)(OCH2CH2CN), NEt3; (c)
POCl3, 0 °C, then Bu3N-PPi.
Table 1. Tm Values for Duplex Containing DM and TMa
5′-dGCGTACXCATGCG
3′-dCGCATGYGTACGC
X
Y
Tm (°C)
X
Y
Tm (°C)
TM
TM
DM
A
T
G
C
T
55.2
52.8
51.7
49.2
50.0
44.7
59.2
DM
TM
DM
A
T
G
C
C
53.8
53.7
48.5
48.2
48.7
43.8
61.8
A
G
a See text for experimental details.
Table 2. Tm Values for Duplex Containing DMN and 2MNa
5′-dGCGTACXCATGCG
3′-dCGCATGYGTACGC
X
Y
Tm (°C)
X
Y
Tm (°C)
DMN
TM
DM
DMN
2MN
ICS
7AI
A
T
G
C
55.0
54.5
56.5
54.5
58.3
55.8
53.2
51.5
53.7
49.2
2MN
TM
DM
2MN
ICS
7AI
A
T
G
C
56.3
54.7
55.0
55.4
54.9
52.3
50.2
52.5
47.2
PICS self-pair vs mispairing with the natural bases (∆Tm
)
a See text for experimental details.
7.1-11.2 °C) compares favorably to the selectivity of a dA:dT
base pair in the same sequence context (∆Tm ) 3.8-10.5 °C).2
The PICS:PICS self-pair is the first stable and thermodynami-
cally orthogonal unnatural base pair reported.
reside in the minor groove. The syn isomer should be destabi-
lized due to eclipsing interactions between the methyl group
and the ribose O4. In the case of 2MN, the aromatic ring is
expected to extend into the major groove, away from the pairing
base, in a mode more analogous to the TM-substitution pattern.
Both the DMN:DM and DMN:TM base pairs were more
stable than the DM:DM and DM:TM pairs by approximately
1 °C (Table 2). However, thermodynamic selectivity of the
unnatural base pairs containing DMN was compromised relative
to the monocyclic analogues. For example, the most stable
mispair, DMN:dG, was only 1.3 °C less stable than the DMN:
TM pair and only 0.8 °C less stable than the DMN:DM pair.
Analysis of the 2-methylnaphthyl (2MN) TM-analogue also
In addition to the bases derived from substituted phenyl rings,
we initiated studies based on a hydrophobic purine scaffold,
7-azaindole nucleoside (7AI).26 7AI is expected to pack in
duplex DNA with only its hydrophobic surface interacting with
its partner.30 In a fashion analogous to that of the amide moiety
in ICS and PICS, N7 in 7AI provides for nucleobase polariz-
ability as well as a minor groove H-bond acceptor. Thermody-
namic evaluation of 7AI paired with other hydrophobic bases,
and itself, yielded melting temperatures that ranged from 55 to
57.6 °C (see Table 3). Of particular interest was the 7AI:ICS
heteropair which was only slightly less stable than dA:dT,
providing an interesting lead for further optimization. Improve-
(30) Structural discussion of DNA containing unnatural base pairs is
based on analogy to native base pairs. Structural studies of DNA containing
unnatural base pairs is currently underway.
(31) Kool, E. T. Chem. ReV. 1997, 97, 1473-1487.

Expansion of the Genetic Alphabet
J. Am. Chem. Soc., Vol. 122, No. 14, 2000 3279
Table 3. Tm Values for Duplex Containing ICS, PICS, and 7AIa
Table 5. Steady-state Kinetic Constants for KF-Mediated
Synthesis of DNA with DMN and 2MN in the Templatea
5′-dGCGTACXCATGCG
3′-dCGCATGYGTACGC
5′-dTAATACGACTCACTATAGGGAGA
3′-dATTATGCTGAGTGATATCCCTCTXGTCA
X
Y
Tm (°C)
X
Y
Tm (°C)
X
Y
Tm (°C)
nucleoside
template (X) triphosphate
kcat
KM
(µM)
kcat/KM
ICS TM
56.8
PICS TM
DM
56.1
55.8
7AI TM
55.8
(min-1
)
(M-1 min-1
)
DM 54.7
7AI 56.5
ICS 59.3
A
T
G
C
DM 55.0
ICS 57.2
7AI 55.5
A
T
DMN
DM
TM
DMN
2MN
7AI
ICS
PICS
A
1.3 ( 0.2
18.0 ( 1.2
n.d.b
78 ( 25
25 ( 6
1.7 × 104
7.2 × 105
e1.0 × 103
2.6 × 107
4.1 × 106
7.1 × 105
9.1 × 105
1.3 × 105
e1.0 × 103
1.4 × 103
2.1 × 104
DMN 60.8
2MN
7AI
ICS
PICS
C
57.4
56.8
60.0
62.6
48.5
n.d.b
55.1
53.0
51.0
52.2
52.5
50.5
51.5
138 ( 2
5.4 ( 0.4
13 ( 2
24 ( 1
5.7 ( 2.2
29 ( 4
n.d.b
53 ( 2
G
17 ( 1
5.2 ( 0.5
3.64 ( 0.15
n.d.b
a See text for experimental details.
G
C
T
Table 4. Steady-State Kinetic Constants for KF-Mediated
0.24 ( 0.01 170 ( 18
1.5 ( 0.1 72 ( 5
Synthesis of DNA Containing DM and TM in the Templatea
5′-dTAATACGACTCACTATAGGGAGA
3′-dATTATGCTGAGTGATATCCCTCTXGTCA
2MN
DM
TM
DMN
2MN
7AI
ICS
PICS
A
1.66 ( 0.07 141 ( 15
1.2 × 104
7.8 × 106
4.2 × 106
4.4 × 107
9.9 × 106
1.2 × 107
1.8 × 107
3.5 × 106
e1.0 × 103
8 × 103
78 ( 5
11.7 ( 0.2
175 ( 10
138 ( 4
140 ( 5
86 ( 8
10 ( 3
2.8 ( 0.3
4.0 ( 0.5
14 ( 1
nucleoside
kcat/KM
template (X) triphosphate
k
cat (min-1
)
KM (µM) (M-1 min-1
)
DM
DM
TM
DMN
2MN
7AI
ICS
PICS
A
1.0 ( 0.1
29 ( 2
359 ( 88
21 ( 5
6 ( 1
5.9 ( 0.3
66 ( 4
47 ( 4
7.4 ( 1.5
75 ( 15
n.d.b
2.8 × 103
12 ( 1
1.4 × 106
5.0 × 104
2.2 × 107
3.2 × 104
3.6 × 105
5.1 × 105
1.5 × 104
<1.0 × 103
2.2 × 103
1.6 × 104
4.8 ( 1.8
30 ( 4
0.3 ( 0.1
131 ( 2
20.8 ( 0.5
17.1( 0.5
3.8 ( 0.2
1.1 ( 0.1
n.d.b
104 ( 4
G
C
T
n.d.b
n.d.b
0.25 ( 0.05
2.2 ( 0.2
30 ( 10
181 ( 31
1.2 × 104
a See text for experimental details. b Rates too slow for determination
of kcat and KM independently.
G
C
T
0.68 ( 0.03 307 ( 28
2.9 ( 0.5 182 ( 30
Table 6. Steady-state Kinetic Constants for KF-Mediated
TM
DM
TM
DMN
2MN
7AI
ICS
PICS
A
1.14 ( 0.04 315 ( 22
3.6 × 103
2.2 × 106
1.1 × 106
3.5 × 107
8.7 × 105
1.1 × 106
1.7 × 106
2.5 × 105
5.0 × 102
4.7 × 102
3.0 × 104
Synthesis of DNA with ICS, PICS, and 7AI in the Templatea
30.9 ( 1.8
3.3 ( 0.2
174 ( 10
27.8 ( 0.6
27.2 ( 0.8
6.8 ( 0.2
6.6 ( 0.2
14 ( 3
3.0 ( 0.5
5.0 ( 0.8
32 ( 2
5′-dTAATACGACTCACTATAGGGAGA
3′-dATTATGCTGAGTGATATCCCTCTXGTCA
nucleoside
template (X) triphosphate
kcat
KM
(µM)
kcat/KM
(min-1
)
(M-1 min-1
)
24 ( 3
3.9 ( 0.4
26 ( 5
ICS
DM
TM
DMN
2MN
7AI
ICS
PICS
A
0.92 ( 0.1
0.99 ( 0.03
n.d.b
358 ( 57
34 ( 4
2.6 × 103
2.9 × 104
e1.0 × 103
5.3 × 105
2.5 × 105
6.3 × 104
7.8 × 104
2.1 × 103
3.4 × 103
8.8 × 102
5.1 × 103
G
C
T
0.07 ( 0.01 140 ( 25
0.18 ( 0.04 381 ( 35
6.9 ( 0.6
n.d.b
46 ( 2
87 ( 10
45 ( 17
57 ( 6
33 ( 11
73 ( 10
83 ( 12
34 ( 19
234 ( 83
227 ( 42
11.3 ( 1.4
3.6 ( 0.1
2.6 ( 0.4
0.15 ( 0.01
0.28 ( 0.02
0.03 ( 0.01
1.2 ( 0.1
a See text for experimental details. b Rates too slow for determination
of kcat and KM independently.
G
C
T
ment in the heteropair stability would yield excellent thermal
selectivity given the modest Tm values for the mispairs between
7AI and the native bases (48.5-52.5 °C).
PICS
7AI
DM
TM
DMN
2MN
7AI
0.45 ( 0.08
40 ( 20
1.1 × 104
7.0 × 104
e1.0 × 103
4.7 × 105
1.7 × 105
3.0 × 104
2.4. Enzymatic Incorporation of Single Unnatural Nucle-
otides into DNA. The unnatural nucleobases were evaluated
as substrates for the exonuclease-deficient Klenow fragment of
E. coli DNA polymerase I (KF). Initial velocities were
determined during [γ-32P]primer extension reactions by literature
methods with varying concentrations of nucleoside triphos-
phates.32 The reactions were analyzed by polyacrylamide gel
electrophoresis and a PhosphorImager (Molecular Dynamics)
was used to quantify gel band intensities corresponding to the
extended primer. The measured velocities were plotted vs
[dNTP] and fit to the Michaelis-Menten equation. Unnatural
nucleobases were assayed both in template DNA and as in-
coming triphosphate nucleotides. Steady-state kinetic parameters
for single nucleotide incorporation are reported in Tables 4-7.
2.4.1. Enzymatic Synthesis of the DM:TM Unnatural Base
Pair. The rates of KF-dependent insertion of dDMTP and
dTMTP were approximately the same opposite either DM or
0.67 ( 0.03 9.6 ( 1.5
n.d.b
n.d.b
64 ( 7
23 ( 2
23 ( 3
30 ( 1
4.0 ( 0.12
0.7 ( 0.03
ICS
DM
TM
DMN
2MN
7AI
ICS
PICS
A
0.61 ( 0.03 346 ( 38
1.7 × 103
9.4 × 104
e1.0 × 103
4.7 × 106
2.2 × 105
3.8 × 105
3.9 × 105
5.9 × 103
1.6 × 103
5.5 × 103
1.6 × 103
3.1( 0.1
33 ( 4
n.d.b
n.d.b
98 ( 3
21 ( 3
40 ( 5
21 ( 3
7.5 ( 2.0
54 ( 10
49 ( 13
31 ( 11
8.7 ( 0.5
7.9 ( 0.3
2.9 ( 0.2
0.32( 0.02
0.08( 0.01
0.17 ( 0.02
G
C
T
0.23 ( 0.05 146 ( 40
a See text for experimental details. b Rates too slow for determination
of kcat and KM independently.
TM in the template. The triphosphate dTMTP bound to the
enzyme-DNA complex more tightly than dDMTP (KM(apparent)
(32) Goodman, M. F.; Creighton, S.; Bloom, L. B.; Petruska, J. Crit.
ReV. Biochem. Mol. Biol. 1993, 28, 83-126.

3280 J. Am. Chem. Soc., Vol. 122, No. 14, 2000
Ogawa et al.
Table 7. Steady-state Kinetic Constants for KF-Mediated
Synthesis of DNA with Natural Templates and Unnatural
Nucleoside Triphosphatesa
increased aromatic surface area on the recognition of hydro-
phobic bases by KF. A comparison of templates containing DM
and DMN reveals that increased aromatic surface area of bases
in the template increases the absolute rate for incorporation of
dDMTP (dDMTP is incorporated an order of magnitude more
efficiently opposite the bulkier template). However, the opposite
was true for dTMTP, which was incorporated with a slightly
reduced efficiency opposite DMN relative to DM in the
template. A comparison of dDMTP and dDMNTP incorporation
reactions reveals that increased aromatic surface area in the
triphosphate favors the insertion opposite TM in template by 3
orders of magnitude (103 M-1 min-1 and 106 M-1 min-1 for
dDMTP and dDMNTP, respectively) while not affecting the
efficiency for incorporation opposite DM in the template. It is
interesting to note that with the DM-like ring structures,
increased steric bulk in the template (DM to DMN) is favorable
for insertion of dDMTP and disfavorable for dTMTP, while
the opposite is true when the increase in steric bulk is in the
triphosphate. This behavior may result from the aromatic surface
area of DMN as well as the C2-methyl group.
5′-dTAATACGACTCACTATAGGGAGA
3′-dATTATGCTGAGTGATATCCCTCTXGTCA
nucleoside
template (X) triphosphate
kcat
KM
(µM)
kcat/KM
(min-1
)
(M-1 min-1
)
A
T
DM
TM
DMN
2MN
7AI
ICS
PICS
163 ( 7
3.5 ( 1.0
4.7 × 107
e1.0 × 103
2.7 × 105
e1.0 × 103
1.0 × 107
2.0 × 105
5.3 × 104
2.0 × 104
n.d.b
6.1 ( 0.4
n.d.b
n.d.b
23 ( 5
n.d.b
14 ( 2
40 ( 12
57 ( 16
20 ( 10
144 ( 4
7.9 ( 0.8
3.0 ( 0.3
0.40 ( 0.10
G
T
C
DM
TM
DMN
2MN
7AI
ICS
PICS
n.d.b
n.d.b
n.d.b
n.d.b
e1.0 × 103
e1.0 × 103
e1.0 × 103
e1.0 × 103
2.5 × 104
n.d.b
n.d.b
n.d.b
n.d.b
0.59 ( 0.02
24 ( 3
26 ( 3
0.43 ( 0.02
1.7 × 104
0.09 ( 0.01 9.3 ( 3.6
9.7 × 103
The rates for the KF-mediated synthesis of mispairs between
DMN and a natural base were nearly identical to those found
with DM, regardless of whether the unnatural base was present
in the template or as the triphosphate. No mispairs were
synthesized with rates greater than 104 M-1 min-1, with the
single exception of incorporation of dATP opposite DMN in
the template, which proceeded with an efficiency of 105 M-1
min-1 (10-fold faster than the incorporation of dATP opposite
DM in the template). The increased rate for insertion of dATP
opposite DMN relative to opposite DM in the template, resulted
from changes in both the apparent KM and kcat. The increased
efficiency of pairing dATP with DMN may result from a
specific interaction between adenine and the minor groove
methyl group of DMN, not present in DM (vide infra).
DM
TM
DMN
2MN
7AI
ICS
PICS
n.d.b
n.d.b
e1.0 × 103
2.0 × 103
e1.0 × 103
8.4 × 104
2.8 × 104
4.1 × 104
3.0 × 105
0.31 ( 0.05 151 ( 55
n.d.b
11.9 ( 0.3
3.9 ( 0.1
4.1 ( 0.2
n.d.b
142 ( 9
120 ( 7
101 ( 13
2.80 ( 0.10 9.4 ( 2.0
DM
TM
DMN
2MN
7AI
ICS
PICS
n.d.b
n.d.b
n.d.b
n.d.b
e1.0 × 103
e1.0 × 103
e1.0 × 103
1.7 × 104
1.5 × 104
1.8 × 104
4.0 × 105
n.d.b
n.d.b
1.2 ( 0.1
1.4 ( 0.1
1.5 ( 0.2
72 ( 12
96 ( 4
83 ( 27
1.05 ( 0.04 2.6 ( 0.7
a See text for experimental details. b Rates too slow for determination
of kcat and KM independently.
Unlike the DM-ring structure discussed above, an increase
in the surface area of the TM-like ring structure (TM to 2MN)
in template or triphosphate, resulted in more efficient synthesis
of the unnatural base pair. 2MN in the template strand directed
the incorporation of dDMTP and dTMTP with 3-fold increased
efficiency relative to TM in the template. The triphosphate
d2MNTP was inserted with greater than 10-fold increased
efficiency opposite DM and TM in the template, relative to
dTMTP. The increased rates resulted from both apparent kcat
and KM effects. (The data in Table 5 shows the same trend with
the base pairs between 2MN and all of the hydrophobic bases
described in this manuscript.) The increased kcat/KM for the 2MN
nucleobase makes the synthesis of the 2MN:DM, 2MN:TM,
and 2MN:2MN unnatural base pairs as efficient as the synthesis
of a native base pair.
) 20 vs 300 µM) and was also incorporated into the growing
strand more quickly (kcat(apparent) ) 30 vs 1 min-1). Tight binding
and fast turnover result in the efficient insertion of dTMTP
opposite either DM or TM in the template (106 M-1 min-1).
KF did not efficiently incorporate a natural triphosphate
opposite DM in the template (apparent kcat/KM < 104 M-1
min-1). However, dATP was incorporated with a catalytic
efficiency of 105 M-1 min-1 opposite TM in the template (10-
fold less efficiently than the incorporation of dTMTP opposite
DM or itself). The triphosphate of thymine was incorporated
opposite TM with a catalytic efficiency of 104 M-1 min-1, while
the triphosphates of both dG and dC were only inefficiently
incorporated opposite TM in the template.
Mispairs resulting from the insertion of dATP or dCTP
opposite 2MN were formed faster than opposite TM, by
approximately 1 order of magnitude. The increased aromatic
surface area of 2MN also resulted in the more efficient
incorporation of d2MNTP opposite natural bases in the template,
relative to dTMTP. The efficiency of incorporation of d2MNTP
opposite dA and dT in the template increased 30-40-fold,
predominantly from an increase in the apparent kcat. In fact, the
incorporation of d2MNTP opposite dA in the template is
virtually as efficient as the insertion of dTTP opposite dA in
the same sequence context (Table 7). Like the TM self-pair
described above, the 2MN:2MN self-pair was marginally
kinetically orthogonal, with the correct pair being synthesized
5-fold faster than all possible mispairs. However, the 2MN self-
pair is synthesized with a native-like catalytic efficiency (4.4
The increased orthogonality of DM relative to TM was also
evident when these unnatural bases were present as triphos-
phates. The triphosphate of TM was incorporated reasonably
efficiently only opposite dA in the template (105 M-1 min-1);
however, dDMTP was not incorporated opposite any natural
template. The relatively poor recognition of dDMTP and the
reasonable efficiency of KF mediated synthesis of the mispair
between TM and dA, in either template or triphosphate
combination, limits the kinetic orthogonality of a base pair
composed of DM and TM. The TM:TM self-pair is marginally
kinetically orthogonal, being synthesized at least 10-fold faster
than all possible mispairs.
2.4.2. Enzymatic Synthesis with Unnatural Nucleobases
with Increased Hydrophobic Surface. Enzymatic synthesis
with DMN and 2MN was examined, to evaluate the effect of

Expansion of the Genetic Alphabet
J. Am. Chem. Soc., Vol. 122, No. 14, 2000 3281
× 107 M-1 min-1 for the self-pair and 4.6 × 107 M-1 min-1
for a dA:dT pair). This efficiency and orthogonality, albeit
marginal, make 2MN an interesting hydrophobic nucleobase
ring structure for further optimization.
106 M-1 min-1 vs 9.4 × 104 M-1 min-1); and d7AITP was
inserted opposite 2MN 10-fold more efficiently than opposite
TM in the template (9.9 × 106 vs 8.7 × 105 M-1 min-1).
The increased polarizability of dICSTP, relative to that of
dDMNTP, resulted in a significant increase in the efficiency of
incorporation opposite 7AI in the template. However, the same
increase in polarizabilty of the nucleobase in the template strand
resulted in less efficient insertion of d7AITP. The efficiency of
incorporation of d7AITP decreased from 4.1 × 106 to 2.5 ×
105 M-1 min-1 upon changing the unnatural base in the template
from DMN to ICS. Comparison of the rate data for ICS and
PICS shows that the propynyl group of PICS has no significant
effect on the synthesis of unnatural base pairs with 7AI.
When 7AI was present in the template, KF did not incorporate
native triphosphates with a catalytic efficiency greater than 6
× 103 M-1 min-1. The triphosphate, d7AITP, was inserted only
2.4.3. Enzymatic Synthesis with Unnatural Bases with
Increased Polaraizability. The ICS and PICS nucleobases
were designed to examine the effect of an N-glycosidic linkage
and the presence of a hydrophilic minor groove carbonyl group,
while retaining the hydrophobic packing surface found in DMN.
The triphosphates of both DM and TM were inserted roughly
an order of magnitude more slowly opposite ICS than they were
opposite DMN in the template, due to both kcat and KM effects.
A comparison of the rates for incorporation of dICSTP or
dDMNTP opposite TM in the template shows that the increase
in polarizability had little effectsboth bases were incorporated
with an efficiency of 106 M-1 min-1. However, in the case of
DM in the template, the insertion of dICSTP was significantly
more efficient than the insertion of dDMNTP.
slowly opposite dG, dC, and dT in the template (104 M-1 min-1
)
and only slightly faster opposite dA in the template (2.0 × 105
M-1 min-1).
The native triphosphates were incorporated by KF opposite
ICS in the template with reduced efficiency relative to DMN
in the template. Most dramatically, incorporation of dATP was
reduced 4000-fold relative to DMN in the template. However,
several unnatural triphosphates were incorporated with reason-
able efficiencies. For example, KF incorporated d2MNTP with
an efficiency of 5.3 × 105 M-1 min-1, which is more than two-
orders of magnitude faster than the most efficient insertion of
a native triphospahate, (dTTP, 5.1 × 103 M-1 min-1). As a
triphosphate, dICSTP was incorporated with greater efficiency
opposite the native bases in the template, relative to dDMNTP,
but never with a catalytic efficiency greater than 5 × 104 M-1
3. Discussion
Efforts to expand the genetic alphabet rely on the specific
thermodynamic and kinetic parameters of nucleosides containing
unnatural bases. The unnatural nucleobases should form stable
pairs in B-form DNA, with high selectivity relative to mispairing
with the native bases. Furthermore, a DNA polymerase must
be capable of efficiently synthesizing DNA containing the third
base pair, and must do so with high fidelity. We have evaluated
hydrophobicity, as opposed to H-bonding, as a driving force
for selective information storage and replication. Previous efforts
to expand the genetic alphabet by Benner and co-workers3,5,9,11,14
were predicated on the use of unnatural H-bonding nucleobases.
Such nucleobases were found to be marginally stable and
themodynamically orthogonal to the native bases when incor-
porated into duplex DNA.4,5,12,13 However, none of these bases
were found to be kinetically orthogonal. Significant mispairing
was believed to be caused, at least in part, by tautomeric forms
of the bases.3-5,9,11,14 Carbocyclic hydrophobic nucleobases
cannot adopt the tautomeric structures that would compromise
their orthogonality relative to the native bases. This is especially
important, considering our incomplete understanding of the
tautomeric equilibria of nucleobases in duplex DNA.
Kool and co-workers demonstrated that several hydrophobic
nucleobases are surprisingly good substrates for KF.33-35 These
hydrophobic bases were designed as shape-analogues of the
natural bases and were incorporated opposite the corresponding
native base with reasonable efficiency. However, it is unclear
to what extent the efficiency of incorporation was dependent
on shape-complementarity with a natural nucleobase (vide infra).
Kool also showed that pyrene paired opposite an abasic site in
duplex DNA results in only a relatively small destabilization
relative to a dA:dT pair.36 Moreover, a technique to sequence
abasic sites in DNA was developed, based on the fact that pyrene
is inserted efficiently and specifically opposite sites that lack
bases.36
min-1
.
PICS behaved similarly to ICS when in template. There were
small differences arising from the slightly tighter binding and
slightly smaller apparent kcat for the incorporation of an unnatural
triphosphates opposite PICS, relative to opposite ICS. Com-
parison of dICSTP and dPICSTP shows that the addition of
the propynyl group has very little effect on the efficiency of
incorporation, with small but compensating changes in the
apparent kcat and KM values. Like ICS, the insertion of dPICSTP
opposite 2MN is approximately as efficient as the insertion of
dTTP opposite dA.
With respect to the incorporation of a natural triphosphate,
PICS again behaved in a manner similar to that of ICS. No
natural triphosphate was incorporated opposite PICS with an
efficiency greater than 104 M-1 min-1. When present as a
triphosphate, dPICSTP was incorporated with an order of mag-
nitude greater catalytic efficiency opposite dC and dT, relative
to dICSTP. However, the rate of misincorporation of dPICSTP
opposite a natural template did not exceed 5 × 105 M-1 min-1
,
which is 3 orders of magnitude slower than the rate of insertion
of a correct, natural base. Therefore, these rates do not
compromise the orthogonality of the unnatural nucleobase.
The hydrophobic purine analogue, 7AI, was also examined
as a substrate for KF. The triphosphate of TM was incorporated
two-orders of magnitude more efficiently than DM opposite 7AI
in the template. The triphosphate of 7AI was inserted more
efficiently opposite TM relative to DM in all cases, by 1 to 2
orders of magnitude. The increased aromatic surface area of
DMN relative to DM did not result in more efficient synthesis
with the unnatural base present as either triphosphate or in the
template. However, in comparing 2MN to TM, the increased
aromatic surface area resulted in more efficient synthesis of the
unnatural base pair. The triphosphate of 2MN was inserted 50-
fold more efficiently opposite 7AI than was dTMTP (4.7 ×
We have reported the design, synthesis, and characterization
of the triphosphates and phosphoramidites of seven hydrophobic
nucleobases. These hydrophobic bases have been characterized
thermodynamically in duplex DNA and kinetically during DNA
(33) Morales, J. C.; Kool, E. T. Nat. Struct. Biol. 1998, 5, 950-954.
(34) Moran, S.; Ren, R. X.-F.; Rumney, S. I.; Kool, E. T. J. Am. Chem.
Soc. 1997, 119, 2056-2057.
(35) Moran, S.; Ren, R. X.-F.; Kool, E. T. Proc. Natl. Acad. Sci. U.S.A.
1997, 94, 10506-10511.
(36) Matray, T. J.; Kool, E. T. Nature 1999, 399, 704-708.

3282 J. Am. Chem. Soc., Vol. 122, No. 14, 2000
Ogawa et al.
synthesis with KF. Generally, they have been found to be both
thermodynamically and kinetically orthogonal to the natural
bases.
problem altogether, without compromising the ability to store
increased information. The PICS self-pair is particularly at-
tractive and is discussed further below.
3.1. Stability and Orthogonality of Unnatural Hydropho-
bic Bases. To date there has not been a report of an unnatural
nucleobase without H-bonds that forms a base pair as stable as
native base pairs.2 A variety of the unnatural bases described
in this study, the pairing of which is based exclusively on
hydrophobicity, form base pairs that are virtually as stable as
native pairs. Several are actually more stable than native pairs.
This demonstrates that interstrand hydrophobic packing interac-
tions may effectively compensate for the complete removal of
Watson-Crick H-bonds. The initially designed methyl-substi-
tuted phenyl bases, DM and TM, generally formed base pairs
that were significantly less stable than native base pairs. The
TM:TM base pair was the most stable, but was still significantly
less stable than a dA:dT pair (55.2 vs 59.2 °C). Increasing the
aromatic surface area of DM and TM, resulting in DMN and
2MN, respectively, resulted in an increase in unnatural base
pair stability in all cases. The inclusion of an amide linkage in
ICS, did not significantly affect the stability of the base pair
with DM (54.5 and 54.7 °C for DMN:DM and ICS:DM,
respectively) but did stabilize the pair with TM (55.0 and 56.8
°C for DMN:TM and ICS:TM, respectively). The propinyl
group of PICS stabilized base pairs with DM-like ring structures
(DM and DMN), but destabilized base pairs with TM-like ring
structure nucleobases (TM and 2MN). The hydrophobic purine
analogue 7AI paired with greater stability opposite TM relative
to opposite DM (55.8 and 55.0 °C, respectively).
The unnatural hydrophobic bases are also generally found to
be thermodynamically orthogonal to the natural nucleobases.
This is demonstrated by the consistent increased stability of the
hydrophobic base pairs relative to that of the mispairs with
native bases. In fact, the most stable unnatural pairs were
virtually as thermodynamically selective as the native bases.
This presumably results, at least in part, from the desolvation
of the native bases that occurs upon their pairing with a
hydrophobic base; waters of solvation are lost and are not
replaced by compensating Watson-Crick H-bonds. This des-
olvation model is supported by the consistent decreased stability
of the mispairs between the hydrophobic nucleobase and
cytosine. Cytosine and guanine are the most hydrophilic of the
natural nucleobases, but the hydrophilicity of guanine may be
partly compensated for by its increased aromatic surface area.37
Cytosine does not have a large aromatic surface area, and the
instability of the hydrophobic base mispairs with cytosine may
be a result of desolvation in conjunction with poor base-stacking.
3.2. Kinetic Efficiency and Orthogonality of Unnatural
Hydrophobic Bases as Polymerase Substrates. Klenow frag-
ment of DNA Pol I is able to efficiently recognize a large
number of unnatural hydrophobic bases and incorporate them
into DNA. This is remarkable, considering that the ring
structures and functionalities of the unnatural bases are signifi-
cantly different from those of the native bases. The TM-like
ring structure was especially well-suited for KF-catalyzed DNA
synthesis. When in template, TM directed the incorporation of
five of the seven unnatural nucleobases, dTMTP, dDMNTP,
d2MNTP, dICSTP, and dPICSTP with efficiencies (apparent
kcat/KM) greater than or equal to 106 M-1 min-1. The nucleoside
triphosphate, dTMTP, was efficiently inserted opposite DM,
TM, and 2MN with rates greater than or equal to 106 M-1
min-1. These rates are only 10-fold lower than those for the
synthesis of native DNA.
Opposite all seven unnatural bases in the template, d2MNTP
was the most efficiently incorporated unnatural triphosphate.
Indeed, d2MNTP was incorporated opposite DM, TM, DMN,
or 2MN with efficiencies equal to those characteristic of native
base pair synthesis (107 M-1 min-1). The incorporation of
d2MNTP opposite 7AI is marginally less efficient (106 M-1
min-1), and its incorporation opposite the isocarbostyril tem-
plates, ICS and PICS, is reduced 100-fold (105 M-1 min-1).
When in the template 2MN directs the incorporation of ICS,
PICS, and 7AI with efficiencies that are within a factor of 5 of
that for dA:dT synthesis. There is a segregation of the unnatural
bases with respect to their KF-mediated pairing with 2MN,
depending on whether the 2MN is present in the template or as
the nucleoside triphosphate. When 2MN is present in the
template, the pairs with ICS, PICS, and 7AI are most efficiently
synthesized. However, when 2MN is present as the triphosphate,
the pairs with DM, TM, and DMN are more efficiently
synthesized.
The base pairs composed of hydrophobic bases other than
2MN or TM were generally poorer substrates for KF. The only
base pair synthesized with an efficiency greater than 106 M-1
min-1, which did not involve 2MN or TM, was the insertion
of d7AITP opposite DMN.
Kinetic orthogonality is evident from the increased efficiency
with which the unnatural hydrophobic nucleobases are paired
opposite other unnatural hydrophobic bases, relative to the
natural bases. With any natural base in the template, efficiencies
of unnatural hydrophobic base insertion were typically less than
or equal to 104 M-1 min-1. Mispairs resulting from the insertion
of an unnatural triphosphate are therefore formed at least 3
orders of magnitude less efficiently than native base pairs,
allowing for good kinetic orthogonality with the unnatural bases
as triphosphates. However, there are five exceptions where
mispairs between native and hydrophobic bases were synthe-
The most important result concerning the stability of the
unnatural base pairs reported in this manuscript is that H-bonds
are not required for the stable and selective pairing of bases in
duplex DNA. The base pairs ICS:DMN, ICS:ICS, PICS:DMN,
and PICS:ICS and PICS:PICS are each at least as stable and
selective as a dA:dT base pair in the same sequence context.
Despite the demonstrated selectivity possessed by the unnatural
bases against mispairing with the native bases, pairing between
the unnatural bases is not as selective as would be desirable for
an unnatural heteropair. Our efforts to understand the hydro-
phobic determinants of selectivity are only in their infancy, and
in fact, preliminary results indicate that judicious placement of
a methyl group on either the ICS or 7AI ring structures allows
for dramatically increased selectivity among the hydrophobic
bases. However, the use of hydrophobic self-pairs avoids the
sized with catalytic efficiencies greater than 104 M-1 min-1
;
dTMTP and d7AITP were inserted opposite dA, and dPICSTP
was inserted opposite dC or dT, with catalytic efficiencies of
approximately 104 M-1 min-1. The insertion of d2MNTP
opposite dA was catalyzed with a remarkable efficiency of 107
M-1 min-1
.
Insertion of a native triphosphate opposite an unnatural
hydrophobic bases was generally found to be less efficient than
insertion of a hydrophobic base opposite another hydrophobic
base. Unlike the case where the native base is in the template,
when the native base is present as the nucleoside triphosphate,
(37) Shih, P.; Pedersen, L. G.; Gibbs, P. R.; Wolfenden, R. J. Mol. Biol.
1998, 280, 421-430.

Expansion of the Genetic Alphabet
J. Am. Chem. Soc., Vol. 122, No. 14, 2000 3283
there is an approximate correlation between the hydrophobicity
of the native base and the efficiency of its insertion opposite
the unnatural base in the template. Recall that this same trend
was observed in the stability of the mispairs between hydro-
phobic and native bases. Correspondingly, the triphosphate of
adenine, the most hydrophobic of the natural nucleobases,37
tends to be inserted opposite the hydrophobic bases with the
greatest efficiency (up to 105-106 M-1 min-1), followed by
dTTP, dGTP, and dCTP. The most efficient insertion of dATP
occurred opposite 2MN, DMN, and TM with efficiencies of
105-106 M-1 min-1. However, hydrophobic bases are generally
incorporated at least 1 order of magnitude more efficiently than
are natural triphosphates opposite hydrophobic bases in the
template.
Figure 2.
The diversity of unnatural hydrophobic bases with which KF
is capable of efficiently synthesizing DNA is remarkable. In
addition to the efficient synthesis of unnatural DNA containing
the TM and the 2MN unnatural nucleobases, a variety of
reactions catalyzed by KF are remarkably efficient. The insertion
of d7AITP opposite DMN was catalyzed by KF with an
efficiency only 10-fold reduced relative to that for native
synthesis (4.1 × 106 M-1 min-1). The facile incorporation of
dATP opposite TM, DMN and 2MN is also fast (105-106 M-1
min-1). The presence of a templating minor groove methyl
group and a hydrophobic region of a purine are common to
each of these unnatural bases. A comparison of DMN and ICS
shows that the specific interactions between the incoming dATP
and the templating minor groove methyl group, as opposed to
minor groove amide carbonyl group, results in a 60-fold
reduction in the efficiency of dATP incorporation. It may seem
surprising that the substitution would be so detrimental to the
enzymatic incorporation of dATP, considering that thymine
possesses an equivalent carbonyl group at the analogous
position. However, it is possible that a component of the
transition state stabilization that favors dATP insertion opposite
DMN, relative to ICS, may result from specific hydrophobic
interactions between the minor groove methyl group of the
unnatural base and adenine. Adenine is the only natural
nucleobase that has a hydrophobic C-H group which is capable
of interstand hydrophobic packing in the minor groove.
The hypothesis of a specific interaction, favoring dATP
incorporation, should not be confused with the “A-rule” which
hypothesizes that polymerases tend to insert dATP opposite a
non-instructive site.38,39 However, as argued by Kool and co-
workers, the high efficiency of the incorporation makes the
A-rule less tenable in this case.35 The hypothesis of a specific
hydrophobic interaction should also not be taken as evidence
supporting a shape-complementarity mechanism for polymerase
specificity.34,35 In the majority of the data described in this
manuscript, such shape-complementarity appears to play a less
important role than hydrophobicity. For example, TM could
be viewed as a shape mimic of T, and KF does exhibit a higher
selectivity for the insertion of dTMTP opposite A (100-fold),
relative to the insertion of dTMTP opposite other natural bases.
However, d7AITP is inserted with virtually the same efficiency
as that with dTMTP opposite adenine, and dICSTP is inserted
only 5-fold less efficiently. There is also only a slight preference
(less than 10-fold) for dATP insertion opposite TM, relative to
the insertion of the other natural triphosphates. Moreover, dATP
is inserted opposite DMN in the template with only 5-fold
reduced efficiency relative to its insertion opposite TM in the
template. Other unnatural hydrophobic bases, which have no
Figure 3.
shape resemblance to adenine, were incorporated more ef-
ficiently opposite TM than was dATP. Therefore, although
shape may contribute to KF specificity, it is apparent that
hydrophobicity plays a more dominant role.
3.3. Unnatural Base Pairs. On the basis of the above
analysis, a number of combinations of these unnatural hydro-
phobic bases were found to possess thermodynamic and kinetic
properties that make them interesting leads for the development
of third base pair.
3.3.1. The ICS:7AI Unnatural Base Pair. Two candidate
bases discussed above are ICS and the related PICS. These
nucleobases have been found to be thermodynamically orthogo-
nal to the native nucleobases. In duplex DNA they form stable
base pairs with a variety of other hydrophobic bases, and some
of these pairs have a stability and selectivity that rivals or even
exceeds the stability of native DNA. During KF-mediated DNA
synthesis, the nucleobases most efficiently paired with ICS and
PICS were all hydrophobic. The incorporation of native
triphosphates opposite ICS all proceeded with very low ef-
ficiencies, and as a triphosphate, dICSTP insertion opposite a
native base in the template was not competitive with the
incorporation of a native triphosphate.
The unnatural nucleobase 7AI was found to be an attractive
partner found for ICS (Figure 2). As a pair in duplex DNA,
ICS:7AI was only slightly less stable than a dA:dT pair.
Misinsertion of native bases opposite 7AI were all also
unfavorable, all proceeding with a catalytic efficiency less than
6 × 103 M-1 min-1. d7AITP incorporation opposite native bases
was not competitive with native DNA synthesis, and in template,
7AI directed the incorporation of a native base with catalytic
efficiencies less than 7 × 103 M-1 min-1. The catalytic
efficiencies for DNA synthesis were 4 × 105 and 3 × 105 M-1
min-1 for pairing dICSTP opposite 7AI and d7AITP opposite
ICS, respectively. Moreover, preliminary results indicate that,
after synthesis of the unnatural base pair, continued DNA
synthesis is reasonably efficient. The major fidelity problem
for the ICS:7AI base pair arises from the synthesis of ICS:
ICS and 7AI:7AI. However, these base pairs are not extended
by KF (data not shown).
3.3.2. The PICS:PICS Self-Pair. The use of a hydrophobic
base that self-pairs eliminates half of the possible thermody-
namic and kinetic mispairs, thus facilitating orthogonality to
the native bases. As reported previously,2 the PICS:PICS self-
base pair (Figure 3) stabilizes the DNA duplex used in this study
relative to a dA:dT or dG:dC base pair by 3.4 and 0.8 °C,
respectively. The specificity of the stabilizing interactions is
demonstrated by the decreased stability of the mispairs of PICS
with native bases. The mispairs between PICS and the native
(38) Drew, H. R.; Travers, A. A. Cell 1988, 37.
(39) Randall, S. K.; Eritja, R.; Kalplan, B. E.; Petruska, J.; Goodman,
M. F. J. Biol. Chem. 1987, 262, 6864-6870.

3284 J. Am. Chem. Soc., Vol. 122, No. 14, 2000
Ogawa et al.
otide using a Cary 200 Bio UV-visible spectrometer. Measurements
were taken over a range of 16-80 °C at 0.5 °C/min intervals. Melting
temperatures were obtained from the derivative method utilizing the
Cary Win UV thermal application software.
bases are all more than 7.1 °C destabilized, a thermodynamic
selectivity equivalent to that found with the native bases. The
kinetic selectivity of PICS for self-pairing is evident from the
20-1000-fold more efficient incorporation of dPICSTP opposite
PICS, when compared to the rate for incorporation of the natural
triphosphates opposite PICS. Moreover, dPICSTP incorporation
is not competitive (within a factor of 103) with the insertion of
any natural nucleoside triphosphate opposite its natural Watson-
Crick partner. Therefore, faithful replication of DNA, containing
native pairs in addition to the self-pair of PICS, is favored over
all possible mispairs by at least a factor 20.
3.3.3. Base Pairs Containing 2MN. Despite the reasonable
orthogonality demonstrated by ICS and PICS, the absolute
efficiency of KF-mediated synthesis of DNA containing either
unnatural base is several orders of magnitude reduced from the
efficiency for the synthesis of DNA containing only native bases.
However, several unnatural base pairs involving 2MN are
synthesized with efficiencies approaching, and in some cases,
virtually identical to those characteristic of native bases.
However, the utility of this base is compromised by the
efficiency with which its triphosphate is incorporated opposite
dA. As discused above, this may result from specific interactions
between adenine and the minor groove methyl group. If the
interactions between 2MN and adenine could be selectively
destabilized, 2MN would be very attractive nucleobase. Experi-
ments directed toward this end are in progress.
General Polymerase Kinetic Assay Protocol. Primer was 5′-end
32P-labeled with polynucleotide kinase. Primer-template duplexes were
annealed by mixing in the reaction buffer, heating to 90 °C, and slow
cooling to room temperature. Assay conditions include: 40 nM
template-primer duplex, 0.11-1.34 nM enzyme, 50 mM Tris buffer
(pH 7.5), 10 mM MgCl2, 1 mM DTT, and 50 µg/mL BSA. The
reactions were initiated by adding the DNA-enzyme mixture to an
equal volume (5 µL) of a 2× dNTP stock solution, incubated at room
temperature for 1-10 min, and quenched by the addition of 20 µL of
loading buffer (95% formamide, 20 mM EDTA). The reaction mixture,
(5 µL) was then analyzed by 15% polyacrylamide gel electrophoresis.
Radioactivity was quantified using a PhosphorImager (Molecular
Dynamics), with overnight exposures and the ImageQuant program.
The kinetic data were fit to the Michaelis-Menten equation using the
program Kaleidograph (Synergy Software). The data presented are
averages of duplicates or triplicates.
Compound 9. The data are presented for comparison to literature
values.40 1H NMR (400 MHz, CDCl3) δ 7.98 (4H, m), 7.20-7.30 (5H,
m), 6.91 (1H, s), 5.62 (1H, m), 5.39 (1H, dd, J ) 10.9, 5.0 Hz), 4.73
(1H, dd, J ) 11.7, 3.8 Hz), 4.67 (1H, dd, J ) 11.8, 3.6 Hz), 4.51 (1H,
m), 2.52 (1H, dd, J ) 13.8, 5.0 Hz), 2.43 (3H, s), 2.40 (3H, s), 2.28
(3H, s), 2.19 (3H, s), 2.15 (1H, m), 2.04 (3H, s). 13C NMR (150 MHz,
CDCl3) δ 166.4, 166.2, 144.1, 143.8, 136.0, 135.6, 134.4, 131.7, 131.6,
129.7, 129.2, 129.1, 127.1, 127.0, 126.2, 82.5, 77.6, 64.7, 40.6, 21.7,
21.7, 19.2, 18.5. HRMS calcd for C30H33O5Na (MH+): 473.2328;
found: 473.2339.
3.4. Efforts to Expand the Genetic Alphabet. Hydrophobic-
ity is a strong and selective force in nucleic acid biochemistry.
Hydrophobic bases are expected to avoid the issues of tau-
tomerization that have been an obstacle in past efforts to identify
new base pairs.13,15,16 Here we have summarized our initial
efforts to optimize interstrand hydrophobic packing to control
information storage and replication. Unnatural nucleobases have
been found which use hydrophobicity to drive correct pairing
in duplex DNA as well as during enzymatic DNA synthesis.
The stability and selectivity of the unnatural base pairs is
comparable with native pairs. Several unnatural base pairs are
synthesized by KF with a kinetic efficiency equivalent to that
of native DNA synthesis. Moreover, the unnatural bases are
orthogonal to the native bases, with at least an order of
magnitude selectivity favoring correct pairing. We are currently
examining derivatives of these nucleobases reported above, with
the goal of optimizing both function and orthogonality. Struc-
tural studies are also in progress to identify any perturbations
of the duplex which result from the unnatural bases. Polymerase-
mediated extension of primers containing the unnatural base is
also critical for the efficient synthesis of DNA containing three
base pairs, and experiments addressing this issue are also in
progress.
Compound 1. The data are presented for comparison to literature
values.40 1H NMR (500 MHz, CD3OD) δ 7.25 (1H, s), 6.86 (1H, s),
5.27 (1H, dd, J ) 10.5, 5.5 Hz), 4.29 (1H, m), 3.90 (1H, m), 3.71 (1H,
dd, J ) 11.8, 5.5 Hz), 3.67 (1H, dd, J ) 11.8, 5.5 Hz), 2.23 (3H, s),
2.19 (3H, s), 2.17 (3H, s), 2.16 (1H, m), 1.80 (1H, ddd, J ) 13.0,
10.5, 6.0 Hz). 13C NMR (150 MHz, CD3OD) δ 138.1, 136.2, 135.0,
133.0, 132.4, 127.3, 88.7, 78.3, 74.4, 64.0, 43.4, 19.4, 19.3, 18.6.
Compound 10. To a solution of nucleoside 1 (60 mg, 0.254 mmol)
in pyridine (1.5 mL) and CH2Cl2 (1.5 mL) was added triethylamine
(300 µL, 2.152 mmol), followed by DMTr-Cl (115 mg, 0.339 mmol).
After stirring at room temperature for 30 min, the reaction mixture
was concentrated and purified by column chromatography on silica
gel (50-80% ethyl acetate in hexane) to afford the tritylated nucleoside
(127 mg), which was dissolved in CH2Cl2 (2.5 mL). To the purified
nucleoside was added a catalytic amount of DMAP (∼2 mg), followed
by triethylamine (250 µL, 1.794 mmol) and 2-cyanoethyl diisopropyl-
aminochlorophosphoramidite (105 µL, 0.471 mmol). After 15 min, the
reaction mixture was partitioned between CH2Cl2 (20 mL) and saturated
aqueous NaHCO3 (20 mL). The layers were separated, and the aqueous
layer was extracted with 2 × 20 mL CH2Cl2. The combined organics
were dried over Na2SO4, filtered, and concentrated. Purification by
column chromatography on silica gel (10-30% ethyl acetate in 5%
triethylamine/hexane) afforded phosphoramidite 10 (151 mg, 80.5%
1
over two steps). H NMR (400 MHz, CDCl3) δ 7.25-7.54 (10H, m),
6.94 (1H, s), 6.85 (4H, m), 5.33 (1H, m), 4.55 (1H, m), 4.22 (1H, m),
3.25-4.85 (10H, m), 2.30-2.65 (5H, m), 2.29 (3H, m), 2.23 (3H, s),
2.17 (3H, s), 2.06 (1H, m), 1.05-1.20 (12H, m). 31P NMR (140 MHz,
CDCl3) δ 148.9, 148.2. HRMS calcd for C44H55N2O6PCs (MCs+):
871.2852; found: 871.2833.
Compound 11. Proton sponge (12 mg, 0.057 mmol) and nucleoside
1 (9 mg, 0.038 mmol) were dissolved in trimethyl phosphate (0.19
mL) and cooled to 0 °C. POCl3 (4 µL, 0.045 mmol) was added, and
the lavender slurry was stirred at 0 °C for 2 h. Tributylamine (56 µL,
0.235 mmol) was added, followed by a 0.4 M solution of tributylam-
monium pyrophosphate (33 mg) in DMF. After 1 min, the reaction
was quenched by addition of 1 M aqueous triethylammonium bicarbon-
ate (4 mL). The resulting crude solution was lyophilized; purification
via reverse phase (C18) HPLC (4-30% CH3CN in 0.1M TEA-
bicarbonate, pH 7.5) afforded triphosphate 11 as a white solid.
4. Experimental Section
General Methods. All reactions were carried out in oven-dried
glassware under inert atmosphere unless otherwise stated. All solvents
were dried over 4 Å molecular sieves except dichloromethane (distilled
from CaH2), tetrahydrofuran (distilled from sodium and potassium
metal), and diethyl ether (distilled from LiAlH4). Oligonucleotides were
synthesized using an Applied Biosystems Inc. 392 DNA/RNA synthe-
sizer. DNA synthesis reagents were purchased from Glen Research,
Sterling, VA. All other reagents were purchased from Sigma-Aldrich.
High-resolution mass spectroscopic data were obtained from the
facilities at The Scripps Research Institute and the University of
California-Riverside.
General Melting Temperature Procedure. Duplex oligonucleotide
denaturation temperature measurements were made in buffer containing
10 mM PIPES, 10 mM MgCl2, 100 mM NaCl, and 3 µM oligonucle-
(40) Schweitzer, B. A.; Kool, E. T. J. Org. Chem. 1995, 60, 8326.

Expansion of the Genetic Alphabet
J. Am. Chem. Soc., Vol. 122, No. 14, 2000 3285
Compound 12. To a suspension of magnesium metal (52 mg, 2.167
mmol) in THF (2 mL) was added 5-bromo-m-xylene (293 µL, 2.164
mmol). The resulting suspension was heated to ∼50 °C. After 1 h,
400 µL of this solution was added to chloroglycoside 8 (126 mg, 0.324
mmol) in THF (1 mL). After 14 h, an additional 100 µL of the
aforementioned Grignard solution was added. After 15 h, the reaction
was partitioned between ethyl acetate (10 mL) and saturated aqueous
NH4Cl (10 mL). The layers were separated, and the aqueous layer was
extracted with ethyl acetate (2 × 15 mL). The combined organics were
dried over Na2SO4, filtered, and concentrated. Purification by column
chromatography on silica gel (5-15% ethyl acetate in hexane) afforded
nucleoside 12 (13 mg, 9%) and its R-anomer (48 mg, 32%). 1H NMR
(400 MHz, CDCl3) δ 7.99 (4H, m), 7.28 (2H, d, J ) 8.1 Hz), 7.24
(2H, d, J ) 8.1 Hz), 7.02 (2H, s), 6.92 (1H, s), 5.62 (1H, m), 5.20
(1H, dd, J ) 10.9, 5.0 Hz), 4.71 (1H, dd, J ) 11.8, 4.0 Hz), 4.66 (1H,
dd, J ) 11.8, 3.6 Hz), 4.53 (1H, m), 2.50 (1H, m), 2.44 (3H, s), 2.41
(3H, s), 2.26 (7H, m). HRMS calcd for C29H31O5 (MH+): 459.2171;
found: 459.2179.
Purification via column chromatography on silica gel (10-20% ethyl
acetate in hexane) afforded the desired S-diastereomer (712 mg, 48%;
the R-diastereomer was isolated in 42% yield), which was dissolved
in pyridine (9 mL) and cooled to 0 °C. Triethylamine (570 µL, 4.09
mmol) was added, followed by the slow dropwise addition of mesyl
chloride (90 µL, 1.17 mmol). The reaction mixture was warmed slowly
to room temperature over 1.5 h, at which time the reaction was
quenched by addition of saturated aqueous NaHCO3 (∼2 mL).
Concentration, followed by purification by column chromatography on
silica gel (5-15% ethyl acetate in hexane), afforded the protected
1
nucleoside 16 (176 mg, 17% over two steps). H NMR (400 MHz,
CDCl3) δ 8.17 (1H, m), 8.07 (1H, m), 7.83 (1H, s), 7.65 (2H, d, J )
7.9 Hz), 7.31-7.59 (11H, m), 6.92 (6H, m), 5.74 (1H, dd, J ) 10.6,
5.2 Hz), 4.58 (2H, s), 4.39 (1H, m), 4.32 (1H, m), 3.86 (3H, s), 3.84
(6H, s), 3.58 (1H, dd, J ) 9.9, 4.5 Hz), 3.47 (1H, dd, J ) 9.9, 3.7 Hz),
2.72 (3H, s), 2.67 (3H, s), 2.53 (1H, dd, J ) 13.1, 5.3 Hz), 2.06 (1H,
m). 13C NMR (100 MHz, CDCl3) δ 159.1, 158.4, 145.0, 136.2, 136.1,
132.7, 132.3, 132.0, 130.1, 130.1, 129.2, 128.2, 127.9, 127.7, 126.7,
125.4, 125.0, 124.5, 124.4, 124.1, 113.8, 113.0, 86.1, 84.1, 81.1, 77.5,
70.7, 64.3, 55.1, 55.0, 40.7, 19.4, 13.7. HRMS calcd for C46H46O6Cs
(MCs+): 827.2349; found: 827.2378.
Compound 17. To a stirred solution of nucleoside 16 (190 mg, 0.274
mmol) in acetic acid (6 mL) and methanol (1 mL) was added
trifluoroacetic acid (10 drops). After stirring at room temperature for
20 min, the orange reaction mixture was concentrated. Purification via
column chromatography on silica gel (20-50% ethyl acetate in hexane)
afforded nucleoside 17 (93 mg, 86%). 1H NMR (400 MHz, CDCl3) δ
8.09 (1H, m), 8.01 (1H, m), 7.54 (2H, m), 7.32 (2H, d, J ) 8.5 Hz),
6.92 (2H, d, J ) 8.6 Hz), 5.61 (1H, dd, J ) 10.6, 5.4 Hz), 4.53 (2H,
m), 4.21 (1H, m), 4.16 (1H, m), 3.91 (1H, dd, J ) 11.7, 3.7 Hz), 3.82
(6H, s), 3.80 (1H, m), 2.43 (1H, dd, J ) 13.4, 5.4 Hz), 1.89 (1H, ddd,
J ) 13.4, 10.6, 6.7 Hz). 13C NMR (100 MHz, CDCl3) δ 159.3, 135.1,
132.7, 132.4, 132.0, 130.0, 129.3, 128.5, 125.6, 125.2, 124.6, 124.5,
123.4, 113.9, 113.8, 85.1, 80.4, 77.6, 63.6, 55.2, 40.4, 19.6, 13.8. HRMS
calcd for C25H28O4Na (MNa+): 415.1885; found: 415.1882.
Compound 2. To a solution of nucleoside 12 in CH3OH (10 mL)
was added 1 M NaOMe (2 mL). After 45 min, the excess NaOMe was
quenched with NH4Cl (∼100 mg). The resulting slurry was concen-
trated, and purified by column chromatography on silica gel (1-5%
1
CH3OH in CH2Cl2) to afford nucleoside 2 (105 mg, 82%). H NMR
(400 MHz, CD3OD) δ 6.98 (2H, s), 6.78 (1H, s), 5.03 (1H, dd, J )
10.6, 5.3 Hz), 4.29 (1H, m), 3.92 (1H, ddd, J ) 7.6, 5.2, 2.4 Hz), 3.66
(1H, m), 2.27 (6H, s), 2.14 (1H, ddd, J ) 13.1, 5.4, 1.6 Hz), 1.91 (1H,
ddd, J ) 13.1, 10.6, 5.9 Hz). 13C NMR (150 MHz, CDCl3) δ 140.8,
138.1, 129.5, 129.3, 123.8, 123.4, 87.2, 80.1, 63.4, 43.9, 21.3. HRMS
calcd for C13H19O3 (MH+): 223.1334; found: 223.1326.
Compound 13. To a solution of nucleoside 1 (90 mg, 0.405 mmol)
in pyridine (2 mL) and CH2Cl2 (2 mL) was added triethylamine (300
µL, 2.152 mmol), followed by DMTr-Cl (175 mg, 0.516 mmol) in two
portions over 30 min. After stirring at room temperature for 2 h, the
reaction mixture was concentrated and purified by column chroma-
tography on silica gel (50-80% ethyl acetate in hexane) to afford the
tritylated nucleoside (185 mg), which was dissolved in CH2Cl2 (3.5
mL). A catalytic amount of DMAP (∼2 mg) was added, followed by
triethylamine (350 µL, 2.512 mmol) and 2-cyanoethyl diisopropyl-
aminochloro phosphoramidite (160 µL, 0.718 mmol). After 30 min,
the reaction mixture was partitioned between ethyl acetate (20 mL)
and saturated aqueous NaHCO3 (20 mL). The layers were separated,
and the aqueous layer was extracted with 2 × 20 mL ethyl acetate.
The combined organics were dried over Na2SO4, filtered, and concen-
trated. Purification by column chromatography on silica gel (10-30%
ethyl acetate in 5% triethylamine/hexane) afforded phosphoramidite
13 (238 mg, 81% over two steps). 1H NMR (400 MHz, CDCl3) δ 7.20-
7.55 (9H, m), 7.09 (2H, s), 6.93 (1H, s), 6.83 (4H, m), 5.12 (1H, m),
4.25 (1H, m), 4.14 (1H, m), 3.60-3.90 (8H, m), 3.20-3.40 (2H, m),
2.40-2.80 (5H, m), 2.30 (6H, m), 2.07 (1H, m), 1.03-1.29 (12H, m).
31P NMR (140 MHz, CDCl3) δ 148.5, 148.3.
Compound 3. To stirred a solution of 17 (8 mg, 0.020 mmol) in
CH2Cl2 (1 mL) and H2O (1 drop) was added DDQ (7 mg, 0.031 mmol).
After 1 h, the reaction mixture was partitioned between ethyl acetate
(15 mL) and saturated aqueous NaHCO3 (15 mL). The layers were
separated, and the aqueous layer was extracted with 2 × 15 mL ethyl
acetate. The combined organics were dried over Na2SO4, filtered, and
concentrated. Purification by column chromatography on silica gel (1-
1
5% CH3OH in CH2Cl2) afforded nucleoside 3 (3 mg, 55%). H NMR
(400 MHz, CDCl3) δ 8.07 (1H, m), 7.98 (1H, m), 7.50 (2H, m), 7.45
(1H, s), 5.24 (1H, dd, J ) 11.5, 1.7 Hz), 4.23 (1H, m), 3.96 (2H, m),
3.81 (1H, m), 2.67 (3H, s), 2.62 (3H, s), 2.10 (1H, m), 1.88 (1H, ddd,
J ) 14.2, 11.6, 2.3 Hz). 13C NMR (150 MHz, CDCl3) δ 135.5, 132.8,
132.7, 132.1, 127.6, 125.7, 125.3, 124.7, 124.5, 123.8, 76.0, 70.9, 69.2,
68.0, 36.5, 19.5, 13.8. HRMS calcd For C13H24NO3 (MNH4+): 290.1756;
found: 290.1764.
Compound 14. Proton sponge (23 mg, 0.107 mmol) and nucleoside
2 (16 mg, 0.072 mmol) were dissolved in trimethyl phosphate (0.36
mL) and cooled to 0 °C. POCl3 (8 µL, 0.090 mmol) was added
dropwise, and the lavender slurry was stirred at 0 °C for 2 h.
Tributylamine (105 µL, 0.441 mmol) was added, followed by a solution
of tributylammonium pyrophosphate (62 mg) in DMF (0.8 mL). After
1 min, the reaction was quenched via addition of 1 M triethylammonium
bicarbonate (7 mL). The resulting crude solution was lyophilized, and
purification via reverse phase (C18) HPLC (4-30% CH3CN in 0.1 M
TEA-bicarbonate, pH 7.5) afforded triphosphate 14 as a white solid.
Compound 16. To a stirred solution of 2-bromo-1,4-dimethylnaph-
thylene (640 mg, 2.72 mmol) in THF (15 mL) at -78 °C was added
nBuLi (2 mL, 2 M in cyclohexane) slowly dropwise. The lime-green
mixture was stirred for 15 min, at which time a solution of aldehyde
15 (1.214 g, 1.801 mmol) in THF (5 mL) was added dropwise down
the side of the flask. After 1 h, the bath was removed, and the reaction
mixture was warmed to room temperature. After a total of 2.5 h, the
reaction mixture was partitioned between ethyl acetate (100 mL) and
saturated aqueous NaHCO3 (100 mL). The layers were separated, and
the aqueous layer was extracted with 2 × 100 mL ethyl acetate. The
combined organics were dried over Na2SO4, filtered, and concentrated.
Compound 18. To a solution of nucleoside 3 (28 mg, 0.103 mmol)
in pyridine (1 mL) was added triethylamine (75 µL, 0.535 mmol),
followed by DMTr-Cl (70 mg, 0.206 mmol). After stirring at room
temperature for 30 min, the reaction mixture was concentrated;
purification via column chromatography on silica gel (50-80% ethyl
acetate in hexane) afforded the tritylated nucleoside (45 mg), which
was dissolved in CH2Cl2 (1 mL). A catalytic amount of DMAP (∼2
mg) was added, followed by triethylamine (65 µL, 0.470 mmol), and
2-cyanoethyl diisopropylaminochloro phosphoramidite (35 µL, 0.157
mmol). After 15 min, the reaction mixture was partitioned between
CH2Cl2 (20 mL) and saturated aqueous NaHCO3 (20 mL). The layers
were separated, and the aqueous layer was extracted with 2 × 20 mL
CH2Cl2. The combined organics were dried over Na2SO4, filtered, and
concentrated. Purification by column chromatography on silica gel (10-
30% ethyl acetate in 5% triethylamine/hexane) afforded phosphor-
amidite 18 (51 mg, 64% over two steps). 1H NMR (400 MHz, CDCl3)
δ 8.09 (1H, m), 7.98 (1H, m), 7.75 (1H, d, J ) 1.8 Hz), 7.53 (4H, m),
7.43 (4H, m), 7.20-7.30 (3H, m), 6.83 (4H, m), 5.66 (1H, m), 4.56
(1H, m), 4.26 (1H, m), 3.83 (1H, m), 3.78 (3H, s), 3.77 (3H, s), 3.45-
3.70 (4H, m), 3.32 (1H, m), 2.64 (3H, s), 2.62 (1H, m), 2.56 (3H, s),
2.46 (1H, dd, J ) 6.6, 6.5 Hz), 2.42 (1H, m), 2.08 (1H, m), 1.19 (9H,

3286 J. Am. Chem. Soc., Vol. 122, No. 14, 2000
Ogawa et al.
m), 1.08 (3H, d, J ) 6.8 Hz). 31P NMR (140 MHz, CDCl3) δ 148.9,
148.3. HRMS calcd for C47H55N2O6PCs (MCs+): 907.2852; found:
907.2837.
transferred to ethyl acetate (15 mL) and extracted with saturated
NaHCO3 (10 mL) and brine (10 mL), dried over Na2SO4 and
concentrated in vacuo. Flash chromatography (25% ethyl acetate in
1
hexane, 2% NEt3) yielded 120 mg (78%) of 22 as a white foam. H
Compound 19. Proton sponge (22 mg, 0.105 mmol) and nucleoside
3 (19 mg, 0.070 mmol) were dissolved in trimethyl phosphate (0.34
mL) and cooled to 0 °C. POCl3 (7 µL, 0.077 mmol) was added
dropwise, and the lavender slurry was stirred at 0 °C for 2 h.
Tributylamine (90 µL, 0.378 mmol) was added, followed by a 0.5 M
solution of tributylammonium pyrophosphate in DMF (0.35 mL). After
1 min, the reaction was quenched via addition of 1 M triethylammonium
bicarbonate (7 mL). The resulting crude solution was lyophilized;
purification via reverse phase (C18) HPLC (4-35% CH3CN in 0.1 M
TEA-bicarbonate, pH 7.5) afforded triphosphate 19 as a white solid.
Compound 4. To a solution of 2-bromo-3-methylnaphthalene (0.595
g, 2.70 mmol) in THF (13.3 mL) at -78 °C was added nBuLi (2.0 M
solution in cyclohexane) dropwise over 10 min. After 15 min, a solution
of aldehyde 20 was added dropwise down the side of the flask over 5
min. The reaction was stirred for 30 min and then brought to room
temperature during which time the reaction color turned from deep
green to brown. The reaction was partitioned between ethyl acetate
(100 mL) and saturated aqueous NaHCO3 (100 mL). The layers were
separated, and the aqueous layer was extracted with 2 × 100 mL ethyl
acetate. The combined organics were dried over Na2SO4, filtered, and
concentrated. The material was subjected to a short plug of silica gel
(20% ethyl acetate in hexane) to afford a crude mixture of both
diasteromers. The crude product was dissolved in CH2Cl2 (32 mL) and
the solution cooled to 0 °C. To the solution was added triethylamine
(0.241 mL) followed by methanesulfonyl chloride (0.120 mL). After
30 min, the reaction was quenched with saturated aqueous NaHCO3
(30 mL). The organic layer was dried over Na2SO4, filtered, and
concentrated. To the crude product was added a 4:1 solution of
trifluoroacetic acid:methanol (56 mL). After 20 min of stirring, the
reaction was concentrated. Residual trifluoroacetic acid was neutralized
with minimal saturated aqueous NaHCO3 and extracted with dichlo-
romethane (4 × 20 mL). The combined organics were dried over Na2-
SO4, filtered, and concentrated. Column chromatography (3% MeOH
in CH2Cl2) afforded the desired â-anomer 4 (0.108 g, 0.4181 mmol)
NMR (400 MHz, CDCl3) δ 8.16 (1H, d, J ) 5.6 Hz), 7.75 (1H, d, J )
7.9 Hz), 7.66 (1H, d, J ) 7.9 Hz), 7.62 (1H, s), 7.56 (2H, m), 7.45-
7.20 (9H, m), 6.83 (4H, m), 5.49 (1H, m), 4.56 (1H, m), 4.28 (1H, m),
3.84 (1H, m), 3.78 (6H, s), 3.7-3.3 (5H, m), 2.70-2.30 (3H, m), 2.03
(1H, m), 1.18 (9H, m), 1.09 (3H, d, J ) 9.2 Hz). ESMS calcd for
C46H53N2O6P (MH)): 761.3; found: 761.
Compound 23. To isocarbostyril (0.50 g, 3.4 mmol) was added
acetonitrile (11 mL) and bis(trimethylsilyl)acetamide (0.85 mL, 3.4
mmol). After 30 min, an additional 12 mL of acetonitrile was added to
the reaction solution followed by addition of 8 (1.1 g, 2.8 mmol). The
reaction was brought to 0 °C and SnCl4 (0.059 mL, 0.69 mmol) was
added dropwise. After 30 min, complete dissolution of the ribofurano-
side had occurred. Ethyl acetate (250 mL) and the resulting solution
was successively extracted with saturated NaHCO3 (3 × 100 mL) and
brine (1 × 100 mL). The organics were dried over Na2SO4 and solvents
were removed in vacuo. Flash chromatography (EtOAc:hexanes; 1:4)
yielded 0.39 g (23%) of â-isomer 23 (faster migrating anomer) as a
1
white foam. H NMR (400 MHz, CDCl3) δ 8.40 (1H, m), 7.98 (2H,
m), 7.92 (2H, m), 7.63 (1H, ddd, J ) 8.9, 7.3, 1.3 Hz), 7.47 (2H, m),
7.44 (1H, d, J ) 7.6 Hz), 7.20-7.28 (4H, m), 6.86 (1H, dd, J ) 8.6,
5.5 Hz), 6.43 (1H, d, J ) 7.7 Hz), 5.63 (1H, m), 4.72 (2H, m), 4.59
(1H, m), 2.87 (1H, m), 2.43 (3H, s), 2.39 (3H, s), 2.32 (1H, m). HRMS
(MALDI) calcd for C30H27NO6Na (MNa+): 520.1736; found: 520.1748.
Compound 5. To a solution of 23 (205 mg, 0.412 mmol) in methanol
(4 mL) was added 1 M NaOMe (1 mL). After 30 min, the reaction
was quenched by addition of NH4Cl (∼100 mg) and concentrated.
Purification via column chromatography on silica gel (1-10% CH3-
1
OH:CH2Cl2) afforded 96 mg (89%) of 5 as a white foam. H NMR
(400 MHz, CD3OD) δ 8.28 (1H, dd, J ) 8.1, 0.6 Hz), 7.75 (1H, d, J
) 7.6 Hz), 7.68 (1H, m), 7.59 (1H, d, J ) 7.7 Hz), 7.50 (1H, m), 6.69
(1H, ddd, J ) 6.7, 4.2, 2.7 Hz), 4.42 (1H, ddd, J ) 6.5, 3.3, 3.2 Hz),
3.99 (1H, dd, J ) 7.3, 3.7 Hz), 3.83 (1H, dd, J ) 12.0, 3.5 Hz), 3.77
(1H, dd J ) 12.0, 4.1 Hz), 2.42 (1H, ddd, J ) 13.5, 6.1, 3.4 Hz), 2.18
(1H, ddd, J ) 13.6, 7.0, 6.7 Hz). 13C NMR (100 MHz, CD3OD) δ
163.5, 138.6, 134.0, 128.3, 128.0, 127.9, 127.3, 126.5, 108.0, 88.9,
88.5, 72.4, 63.0, 42.0. HRMS (MALDI) calcd for C14H15NO4Na
(MNa+): 284.0899; found: 284.0897.
Compound 24. To a solution of nucleoside 5 (85 mg, 0.325 mmol)
in pyridine (3.5 mL) was added triethylamine (230 µL, 1.649 mmol),
followed by DMTr-Cl (229 mg, 0.678 mmol). After stirring for 20
min at room temperature, the reaction mixture was concentrated.
Purification via column chromatography on silica gel (50-80% ethyl
acetate in hexane) afforded 151 mg of tritylated product, which was
dissolved in CH2Cl2 (4 mL). A catalytic amount of DMAP (∼2 mg)
was added, followed by triethylamine (224 µL, 1.607 mmol) and
2-cyanoethyl diisopropylaminochloro phosphoramidite (120 µL, 0.536
mmol). After 15 min, the reaction mixture was partitioned between
CH2Cl2 (20 mL) and saturated aqueous NaHCO3 (20 mL). The layers
were separated, and the aqueous layer was extracted with 2 × 20 mL
CH2Cl2. The combined organics were dried over Na2SO4, filtered, and
concentrated. Purification via column chromatography on silica gel
(15-50% ethyl acetate in 5% triethylamine/hexane) afforded phos-
phoramidite 24 (151 mg, 61% over two steps). 31P NMR (140 MHz,
CDCl3) δ 149.8, 149.0. HRMS calcd for C44H50N2O7PCs (MCs+):
896.2441; found: 896.2412.
Compound 25. Proton sponge (28 mg, 0.132 mmol) and nucleoside
5 (23 mg, 0.088 mmol) were dissolved in trimethyl phosphate (0.9
mL) and cooled to 0 °C. POCl3 (9 µL, 0.101 mmol) was added
dropwise, and the lavender slurry was stirred at 0 °C for 2 h.
Tributylamine (130 µL, 0.546 mmol) was added, followed by a solution
of tributylammonium pyrophosphate (71 mg) in DMF (1 mL). After 1
min, the reaction was quenched via addition of 1 M triethylammonium
bicarbonate (15 mL). The resulting crude solution was lyophilized, and
purification via reverse phase (C18) HPLC (4-35% CH3CN in 0.1 M
TEA-bicarbonate, pH 7.5) afforded triphosphate 25 as a white solid.
31P NMR (140 MHz, 50mM Tris, 2mM EDTA, pH 7.5 in D2O) δ -7.62
(d, J ) 16.8), -10.61 (d, J ) 18.2 Hz), -22.04 (t, J ) 16.8 Hz).
1
in 24% yield over 3 steps. H NMR (600 MHz, MeOH) δ 8.03 (1H,
s), 7.78 (1H, d, J ) 7.1 Hz), 7.71 (1H, J ) 7.1 Hz), 7.59 (1H, s), 7.37
(2H, m), 5.44 (1H, dd, J ) 10.1, 5.3 Hz), 4.35 (1H, m), 3.98 (1H, m),
3.76 (2H, m), 2.47 (3H, s), 2.36 (1H, ddd, J ) 13.15, 5.7, 1.8 Hz),
1.90 (1H, m). 13C NMR (150 MHz, CDCl3) δ 139.8, 133.8, 133.6,
133.1, 128.6, 128.2, 127.3, 126.1, 125.6, 124.1, 88.2, 78.2, 73.8, 63.5,
42.8, 19.1. HRMS calcd for C16H18O3 (MNa+): 281.1154; found:
281.1158.
Compound 21. Proton sponge (13 mg, 0.062 mmol) and nucleoside
3 (10 mg, 0.038 mmol) were dissolved in trimethyl phosphate (0.2
mL) and cooled to 0 °C. POCl3 (4 µL, 0.044 mmol) was added
dropwise, and the lavender slurry was stirred at 0 °C for 2 h.
Tributylamine (60 µL, 0.252 mmol) was added, followed by a 0.5 M
solution of tributylammonium pyrophosphate in DMF (0.16 mL). After
1 min, the reaction was quenched via addition of 1 M triethylammonium
bicarbonate (7 mL). The resulting crude solution was lyophilized, and
purification via reverse phase (C18) HPLC (4-35% CH3CN in 0.1 M
TEA-bicarbonate, pH 7.5) afforded triphosphate 21 as a white solid.
31P NMR (140 MHz, 50mM Tris, 2mM EDTA, pH 7.5 in D2O) δ
-5.77 (d, J ) 18.5), -10.4. (d, J ) 17.1 Hz), -21.94 (t, J ) 49.0
Hz).
Compound 22. To 4 (0.093 g, 0.36 mmol), azeotroped from pyridine
(2 × 0.1 mL), was added pyridine (1.5 mL). To the resulting solution
was added a solution of DMTr-Cl (0.18 g, 0.54 mmol) in pyridine (0.67
mL) over 20 min. The reaction solution was stirred at room temperature
for an additional 20 min, at which point ethyl acetate (20 mL) was
added. The organics were extracted with saturated NaHCO3 (5 mL)
and brine (5 mL), dried over Na2SO4, and concentrated in vacuo. Flash
chromatography on silica gel (50-80% ethyl acetate in hexane) yielded
130 mg (66%) of 5′-protected nucleoside as a white foam. To the
protected nucleoside in CH2Cl2 (1.8 mL) was added triethylamine (0.11
mL, 0.80 mmol). The solution was cooled to 0 °C, and cyanoethyl
diisopropylchlorophosphoramidite (0.068 mL, 0.26 mmol) was added.
The reaction was allowed to reach rt over 15 min. The solution was

Expansion of the Genetic Alphabet
J. Am. Chem. Soc., Vol. 122, No. 14, 2000 3287
Compound 26. To a solution of nucleoside 7 (101 mg, 0.431 mmol)
in pyridine (4 mL) was added triethylamine (400 µL, 2.875 mmol),
followed by DMTr-Cl (310 mg, 0.915 mmol). After stirring at room
temperature for 15 min, the reaction mixture was concentrated;
purification by column chromatography on silica gel (30-50% ethyl
acetate in hexane) afforded the tritylated nucleoside (162 mg), which
was dissolved in CH2Cl2 (3.5 mL). A catalytic amount of DMAP (∼2
mg) was added, followed by triethylamine (250 µL, 1.797 mmol) and
2-cyanoethyl diisopropylaminochloro phosphoramidite (130 µL, 0.582
mmol). After 15 min, the reaction mixture was partitioned between
CH2Cl2 (20 mL) and saturated aqueous NaHCO3 (20 mL). The layers
were separated, and the aqueous layer was extracted with 1 × 30 mL
CH2Cl2. The combined organics were dried over Na2SO4, filtered, and
concentrated. Purification via column chromatography on silica gel
(10-30% ethyl acetate in 5% triethylamine/hexane) afforded phos-
Compound 27. Proton sponge (15 mg, 0.071 mmol) and nucleoside
7 (11 mg, 0.045 mmol) were dissolved in trimethyl phosphate (0.22
mL) and cooled to 0 °C. POCl3 (5 µL, 0.054 mmol) was added
dropwise, and the lavender slurry was stirred at 0 °C for 2 h.
Tributylamine (55 µL, 0.225 mmol) was added, followed by a solution
of tributylammonium pyrophosphate (40 mg) in DMF (0.5 mL). After
1 min, the reaction was quenched via addition of 1 M triethylammonium
bicarbonate (5 mL). The resulting crude solution was lyophilized, and
purification via reverse phase (C18) HPLC (4-35% CH3CN in 0.1 M
TEA-bicarbonate, pH 7.5) afforded triphosphate 27 as a white solid.
Acknowledgment. Funding was provided by the National
Institutes of Health (GM 60005 to F.E.R.) and the Skaggs
Institute for Chemical Biology (F.E.R. and P.G.S.) and a
National Institutes of Health postdoctoral fellowship (F32
GM19833-01, A.K.O.).
1
phoramidite 26 (194 mg, 61% over two steps). H NMR (400 MHz,
CDCl3) δ 8.30 (1H, m), 7.87 (1H, d, J ) 7.8 Hz), 7.15-7.50 (9H, m),
7.07 (1H, dd, J ) 7.8, 4.7 Hz), 6.91 (1H, m), 6.78 (4H, m), 6.44 (1H,
dd, J ) 3.6, 1.8 Hz), 4.76 (1H, m), 4.24 (1H, m), 3.55-3.85 (12H,
m), 3.38 (1H, m), 3.28 (1H, m), 2.72 (1H, m), 2.60 (1H, t, J ) 6.4
Supporting Information Available: Experimental proce-
dures and characterizations (PDF). This material is available
free of charge via the Internet at http://pubs.acs.org.
Hz), 2.58 (1H, m), 2.44 (1H, t, J ) 6.4 Hz), 1.09-1.20 (12H, m). 31
NMR (140 MHz, CDCl3) δ 149.1, 148.9.
P
JA9940064