Differential discrimination of DNA polymerases for variants of the non-standard nucleobase pair between xanthosine and 2,4-diaminopyrimidine, two components of an expanded genetic alphabet
Differential discrimination of DNA polymerases for variants of the non-standard nucleobase pair between xanthosine and 2,4-diaminopyrimidine, two components of an expanded genetic alphabet
Michael J.
Lutz
,
Heike A.
Held
,
Michael
Hottiger
1
,
Ulrich
Hübscher
1
and
Steven A.
Benner*
Department of Chemistry, Swiss Federal Institute of Technology, Universitätsstraße 16 and CH-8092
Zürich
,
Switzerland
and
1
Department of Veterinary Biochemistry, University of Zürich, CH-8057
Zürich
,
Switzerland
Received December 4, 1995;
Revised and Accepted February 13, 1996
ABSTRACT
Mammalian DNA polymerases
[alpha]
and
[epsilon]
, the Klenow fragment of
Escherichia coli
DNA polymerase I and HIV-1 reverse transcriptase (RT) were examined for their ability to
incorporate components of an expanded genetic alphabet in different forms.
Experiments were performed with templates containing 2
'
-deoxyxanthosine (dX) or 2
'
-deoxy-7-deazaxanthosine (c
7
dX), both able to adopt a hydrogen bonding acceptor-donor-acceptor pattern on a purine nucleus (puADA). Thus these
heterocycles are able to form a non-standard nucleobase pair with 2,4-diaminopyrimidine (pyDAD) that fits the Watson-Crick geometry, but is joined by a non-standard hydrogen bonding pattern. HIV-1 RT incorporated d(pyDAD)TP opposite dX with a
high efficiency that was largely independent of pH. Specific incorporation
opposite c
7
dX was significantly lower and also independent of pH. Mammalian DNA polymerases
[alpha]
and
[epsilon]
from calf thymus and the Klenow fragment from
E.coli
DNA polymerase I failed to incorporate d(pyDAD)TP opposite c
7
dX.
INTRODUCTION
Nucleobases in oligonucleotide strands form Watson-Crick base pairs following two rules of complementarity: (i) a large
purine from one strand pairs with a small pyrimidine from the other; (ii)
hydrogen bond donors (NH groups) from one base are matched with hydrogen bond
acceptors (lone pairs of electrons on oxygen or nitrogen) from the other. In
DNA, for example, cytosine, implementing a hydrogen bond donor-acceptor-acceptor pattern (pyDAA), pairs as the small component with
guanine, a large component implementing the hydrogen bond acceptor-donor-donor pattern (puADD).
Some time ago we pointed out that standard nucleobases exploit only part of the
potential of the Watson-Crick formalism (
1
). When fully exploited the Watson-Crick formalism permits 12 independently replicatable nucleobases joined
in six base pairs by mutually independent hydrogen bonding patterns (Fig.
1
). Previous work in these and other laboratories has yielded implementations of
all six hydrogen bonding patterns (
2
-
6
). Further, individual RNA and DNA polymerases have been found that catalyze
template-directed incorporation of several non-standard base pairs into duplex DNA (
7
-
9
). However, DNA polymerases involved in DNA transactions in mammals have so far
rejected non-standard base pairs.
MATERIALS AND METHODS
Synthesis of non-standard nucleobases
2,4-Diamino-5-([beta]-D-ribofuranosyl)pyrimidine (pyDAD) was
synthesized using the route of Chu
et al
. (
3
). This compound was converted to the 2'-deoxygenated nucleoside analog as described by Piccirilli
et al
. (
4
). The triphosphate d(pyDAD)TP was synthesized by a published procedure (
15
). 5'-Dimethoxytrityl-2'-deoxyxanthosine with both heterocyclic ring
oxygens protected as
p
-nitrophenylethyl ethers was prepared by the procedure of Van Aerschot
et al.
(
16
) and converted to the phosphoramidite following a standard method (
17
). 2'-Deoxy-7-deazaxanthosine (c
7
dX) was synthesized as 7-deaza-2'-deoxy(4,4'-dimethoxytrityl)xanthosine-3'-H-phosph- onate as
recently described (
18
). Standard dNTPs were from Pharmacia.
Oligonucleotides
The oligonucleotide bearing 2'-deoxyxanthosine was prepared by solid phase synthesis (Applied
Biosystems) from the [beta]-cyanoethyl-protected phosphoramidite, purified by the trityl-on procedure, deprotected and purified again by HPLC (
19
). The oligonucleotide bearing c
7
dX was synthesized by Dr L.Arnold (Czech Academy of Chemistry and Biochemistry, Prague) using H-phosphonate technology.
The primer (5'-GCATGGATCCCACTGCACTCCAGGG-3') was synthesized by Microsynth (Windisch,
Switzerland) and purified by PAGE. Labelling of the primer at the 5'-end with Redivue [[gamma]-
32
P]ATP (Amersham) was performed using T4 polynucleotide kinase (Life
Technologies).
Nucleic acid substrates
The primer was annealed with a template (5'-ACCCCqCCCCCCCTGGAGTGCAGTGGGATCCATGC-3'), where q is either dX or c
7
dX, in 500 [mu]l total buffer containing 50 pmol template and 15 pmol labelled primer in 1.8 mM Tris-HCl, pH 7.0, 0.5 mM MgCl
2
, 23 mM NaCl by heating the mixture at 85oC for 15 min followed by subsequent slow cooling to room temperature over a
period of 1 h.
DNA polymerases
HIV-1 RT, overexpressed using the plasmid pJS3.7 in
Escherichia
coli
, was purified by a published procedure (
20
). Calf thymus DNA polymerases [alpha] and [epsilon] were purified according to the methods of Podust
et al
. (
21
) and Weiser
et al
. (
22
) respectively. Enzymatic activity was determined as described in these references. The Klenow fragment of DNA polymerase I was from Boehringer Mannheim.
Assays to detect incorporation of the bases
Incorporation of a non-standard base opposite the complementary non-standard base was performed in a total volume of 25 [mu]l using 0.15 pmol labelled and annealed primer and all required
dNTPs, at a final concentration of 5 [mu]M each. Reaction buffers contain the following: for HIV-1 RT, 50 mM Tris-HCl, pH 7.2 (unless otherwise stated), 5 mM MgCl
2
, 100 mM KCl, 1 mM DTT, 0.5 mM EDTA; for DNA polymerases [alpha] and [epsilon] from calf thymus, 50 mM Tris-HCl, pH 6.5, 1 mM DTT, 0.25 mg/ml BSA; for Klenow fragment
of DNA polymerase I, 50 mM Tris-HCl, pH 7.5, 1 mM DTT, 0.1 mg/ml BSA. The amount of enzyme used was 0.1 U
Klenow fragment and HIV-1 RT, 0.11 U DNA polymerase [alpha] and 0.04 U DNA polymerase [epsilon]. The reactions were started by adding the enzyme and
incubated for 15 min at 37oC and finally quenched by adding 5 [mu]l of a mixture of stop/loading dye (New England Biolabs), which
contains 0.3% xylene cyanol, 0.3% bromphenol blue and 0.37% Na EDTA, pH 8.0.
The samples were heated (20 min, 95oC) and aliquots (5 [mu]l) were loaded onto a 17% polyacrylamide gel containing 7 M urea.
Following electrophoresis (constant power 25 W) the gels were fixed (12% MeOH,
10% HOAc, diluted with water), dried and autoradiographed. Radioactivity was
quantified using a PhosphorImager (Molecular Dynamics), with 3 h exposures and
the ImageQuant program from Molecular Dynamics. To determine the amount of
specific formation of the non-standard base pairs the amount of full-length product was quantified, divided by the total amount of
radioactivity in the lane and expressed as a percentage. To correct for non-specific misincorporation of standard nucleobases opposite the non-standard nucleotides the amount of misincorporation of natural
dNTPs, determined in a control experiment, was subtracted.
RESULTS
No evidence could be obtained for incorporation of d(pyDAD)TP opposite c
7
dX in a template when Klenow fragment of DNA polymerase I from
E.coli
was incubated at pH 7.5. Oligonucleotide products indicating extension of the
primer past the non-standard base were found both in the presence and absence of d(pyDAD)TP.
It is possible that the Klenow fragment misincorporates dGTP opposite c
7
dX (Fig.
3
, lanes 5-7). However, the principal product is shorter than the full-length product by one base. Why this
n
- 1 product is formed is not known. It may arise from the DNA polymerase
skipping over the non-standard nucleobase or may be a response of the DNA polymerase to a
mismatch in the template-primer complex. Similar production of
n
- 1 product has been observed with other unsuccessful fill-in experiments using Klenow fragment (
4
,
9
). In any case, a quantitative analysis using a PhosphorImager shows that at
most 1% of the longest product is derived from specific incorporation of
d(pyDAD)TP opposite c
7
dX in the template under these conditions.
DISCUSSION
The standard model of nucleic acid structure, proposed in its original form over
four decades ago by Watson and Crick (
23
), invokes the stacking of hydrophobic nucleobases as a central determinant of
the stability of the double helix. In its simplest form this model suggests
that the less hydrophobic a nucleobase, the less likely it is to be accepted
into a duplex structure by a DNA polymerase. Naively, this implies that given
the choice between a more acidic nucleobase (in this example dX) and a less
acidic nucleobase (c
7
dX), both meeting the minimum hydrogen bonding requirements, the latter would be
more easily accepted than the former.
This is not the case. A variety of polymerases accept c
7
dX as a complement for (pyDAD)TP more poorly than dX; several do not accept it
at all. Further, incorporation of (pyDAD)TP opposite dX in a template is
largely independent of pH over the range 6.2-9.5. This pH range is expected to span the p
K
a
of dX in a template, as the p
K
a
of dX free in solution (
5
.
7
) is expected to be increased by 2 to 3 p
K
a
units when incorporated into a polyanionic oligonucleotide, according to the
observed shift with 7-methyl-2'-deoxyguanine and guanylic acid when embedded in a DNA
oligonucleotide (
24
,
25
). As the p
K
a
of the nucleobase can be further perturbed in the active site of a DNA
polymerase, the ionization state of dX in a template at the instant when the
molecular recognition event occurs is not easily known. However, it is clear
that the intrinsic acidicity of dX does not present an obvious impediment to
its serving as a partner in a Watson-Crick base pair.
Why is c
7
dX accepted less efficiently (or not at all) than its analog dX? Three
explanations might be considered.
(i) Substitution of N-7 in dX by a CH group in c
7
dX might create structural perturbations that might be invoked to explain this
discrimination against c
7
dX. For example, the conformation of the base or the sugar might be influenced
by this substitution.
(ii) Alternatively, the DNA polymerase might actually recognize the deprotonated
form of dX, a form that cannot be attained by c
7
dX due to its higher p
K
a
.
(iii) The DNA polymerase might itself interact with N-7 in a way that causes it to reject c
7
dX as foreign. This proposal suggests that the DNA polymerase is `scanning' the
major groove of duplex DNA.
Each of these possibilities raises interesting questions concerning the event by
which DNA polymerases recognize base pairs. Explanation (i) is problematical,
because structural differences induced by the N-7 substitution are expected to be subtle. Further, HIV-1 RT seems to be largely indifferent to subtle structural features
of the nucleobase. For example, it accepts both DNA and RNA as template, which
have quite different conformations.
Explanation (ii) is problematical considering the fact that incorporation of dX
is essentially pH independent. If the DNA polymerase indeed prefers a
deprotonated form of the nucleobase over the protonated form, one might expect
the efficiency of incorporation of dX to increase with increasing pH. This is
not the case. Further, if the relative p
K
a
values of dX and c
7
dX in the template are the same as the relative p
K
a
values of dX and c
7
dX free in solution and if the only impact of the substitution at position 7 is
the shift in p
K
a
then c
7
dX at pH 8.5 should behave the same as dX at pH 7.0, but it does not.
The remaining possibility is that the DNA polymerase is itself examining
structural features of the nucleobases, presumably in the major and minor
grooves, to discard `unnatural' structures. At one level this proposal is
reasonable. To enforce a Watson-Crick geometry the DNA polymerase must interact in some way with the
nucleobases, in either the major or minor groove. This interaction presumes a
direct contact between functionality on the bases and functionality in the
protein. This proposal is problematical, however, as different nucleobases
present different functionality in these grooves and DNA polymerase should have
no intrinsic preference for one nucleobase over another, once the nucleobase
has been accepted by the template.
Thus DNA polymerases, if they are to interact with the nucleobases to enforce a
Watson-Crick geometry, must do so by identifying features in the grooves of
duplex oligonucleotides that are constant for all four nucleobases. One such
feature exists. In the minor groove the lone pair of electrons on N-3 of both purines approximately overlap in space the lone pair of
electrons presented by the 2-position carbonyl oxygens of both thymine and cytosine. Thus it is
conceivable that a DNA polymerase might present a hydrogen bond donor to this
lone pair in all four bases, allowing it to control the geometry of the
incoming nucleobase without having a preference for one over the other. Several
years ago Steitz noted that such minor groove `scanning' might be used by DNA
polymerases to improve their fidelity (
26
). Furthermore, the recently published crystal structure of mammalian DNA
polymerase [beta] co-crystallized with template, primer and a triphosphate analog
identified three amino acid residues that make contacts with these lone pairs (
27
).
The results reported here are inconsistent with the scanning proposal in its
broadest form, as a lone pair of electrons at position O-2 in pyrimidines is not an absolute requirement for recognition by DNA
polymerases. The pyDAD nucleobase lacks the exoxyclic oxygen and would not be
accepted by any polymerase if the lone pair were an absolute specificity
determinant. As we have shown here and elsewhere (
9
), pyDAD is accepted by many polymerases, either in the template or as a
triphosphate. Further, in its protonated form dX also lacks the lone pair of
electrons at N-3 and yet is also accepted by DNA polymerases, although the possibility
remains that the polymerase is accepting the N-3 deprotonated form of the nucleobase, which carries the lone pair.
Explanation (iii) requires, however, a new type of scanning, in the major
groove. This scanning is also problematical, as no functional group is
consistently presented to the major groove by the standard nucleobases. For
example, thymine presents a hydrophobic methyl group to one region of the major
groove, cytosine presents a hydrogen atom and both purines present a hydrogen
bond acceptor, a lone pair of electrons on N-7. These functionalities are different and it is difficult to imagine a
DNA polymerase making a contact with this region of the major groove without
causing it to favor one of the standard bases over any other in a way that
would diminish faithful reproduction of information in the template.
The disfavoring of c
7
dX is more perplexing in the light of former results showing that Klenow
fragment and Taq DNA polymerase both accept 7-deaza-dGTP (
28
,
29
), as well as dGTP substituted at the N-7 position with either a methyl group or cyanoborane (
24
,
30
). Because Klenow fragment rejects d(pyDAD) as the triphosphate, both opposite
dX and c
7
dX, its rejection of c
7
dX is more difficult to interpret. Nevertheless, Klenow fragment does not seem
to require a lone pair of electrons on N-7 in the major groove for all purines.
These data suggest a paradox in the `model' for the selectivity of polymerases.
The selectivity of individual polymerases (such as Klenow fragment) with
respect to variants of non-standard nucleobases seems to be unrelated to their selectivity with
respect to analogous variants of the standard nucleobases. There is no simple
structural explanation for this fact. Further, even though crystal structures
compellingly argue that all polymerases are related by common ancestry (
31
), it is clear that the details of the molecular recognition process diverge
greatly with their sequences. There is not likely to be a general model
describing DNA polymerase specificity generally; each polymerase will need to
be described individually, with more work both in solution with non-standard nucleobases and other nucleotide analogs and in the crystal.
ACKNOWLEDGEMENTS
We thank Drs J.Opitz and A.Roughton for helpful discussions. We are indebted to
Dr L.Arnold from the Czech Academy of Organic Chemistry and Biochemistry in
Prague for oligonucleotide synthesis. MJL was supported by a scholarship from
the German Academic Exchange Service (DAAD) in the program HSP II/AUFE and MH
and UH were supported by the Swiss National Science Foundation (grants 3135-36713.92 and 31.37146.93).
19 Horlacher,J. (1995) Dissertation, ETH Zürich no. 11084.
20 Hafkemeyer,P., Ferrari,E., Brecher,J. and Hübscher,U. (1991) Proc. Natl. Acad. Sci. USA, 88, 5262-5266.MEDLINE Abstract
21 Podust,V.N., Mikhailov,V., Georgaki,A. and Hübscher,U. (1992) Chromosoma, 102, 133-141.
22 Weiser,T., Gassmann,M., Thömmes,P., Ferrari,E., Hafkemeyer,P. and Hübscher,U. (1991) J. Biol. Chem., 266, 10420-10428.MEDLINE Abstract
23 Watson,J.D. and Crick,F.H.C. (1953) Nature, 171, 964-967.
24 Hendler,S., Furer,E. and Srinivasan,P.R. (1970) Biochemistry, 9, 4141-4135.MEDLINE Abstract
25 Pochon,F. and Michelson,A.M. (1965) Proc. Natl. Acad. Sci. USA, 53, 1425-1430.MEDLINE Abstract
26 Steitz,T. (1987) In Burnett,R.M. and Vogel,H.J. (eds), Biological Organization: Macromolecular Interactions at High Resolution. Academic Press, New York, NY, pp.45-55.
