Results show that large movements in the pyrimidine base of the A...C mismatch cause loss of stacking, especially with its penultimate base, concomitant with a variable mismatch hydrogen
Trang 1acidÆDNA hybrid in the wild-type and mutated alleles
of Ki-ras proto-oncogene
Stereochemical rationale for the low affinity of PNA in the presence
of an A C mismatch
Thenmalarchelvi Rathinavelan and Narayanarao Yathindra
Department of Crystallography and Biophysics, University of Madras, Guindy Campus, Chennai, India
Institute of Bioinformatics and Applied Biotechnology, ITPB, Bangalore, India
Peptide nucleic acids (PNAs) stand out from the rest
of the nucleic acid mimetic, in that they consist of an
uncharged N-(2-aminoethyl) glycine (Fig 1) backbone
scaffold [1,2] These enable them to defy protease and
nuclease digestion, and therefore serve as promising
contenders as antigene and antisense agents [3–10]
PNA mediated transcription inhibition occurs either
by strand invasion or by conventional triplex forma-tion [1,2] In the former, PNA displaces one of the strands of the DNA duplex by forming Watson and Crick (WC) base pairs leading to a PNAÆDNA duplex (duplex invasion) or by forming WC and Hoogsteen
Keywords
enthalpy-entropy contribution; fluctuating
A .C mismatch hydrogen bond; mismatch
containing PNAÆDNA hybrid; point mutation
Correspondence
N Yathindra, Institute of Bioinformatics and
Applied Biotechnology, G-05, Tech Park
Mall, ITPB, Bangalore-560 066, India
Fax: +91 80 2841 2761
Tel: +91 80 2841 0029
E-mail: yathindra@ibab.ac.in
(Received 13 April 2005, revised 3 June
2005, accepted 14 June 2005)
doi:10.1111/j.1742-4658.2005.04817.x
The low affinity of peptide nucleic acid (PNA) to hybridize with DNA in the presence of a mismatch endows PNA with a high degree of discriminat-ory capacity that has been exploited in therapeutics for the selective inhibi-tion of the expression of point-mutated genes To obtain a structural basis for this intriguing property, molecular dynamics simulations are carried out on PNAÆDNA duplexes formed at the Ki-ras proto-oncogene, compri-sing the point-mutated (GAT), and the corresponding wild-type (GGT) codon 12 The designed PNA forms an A C mismatch with the wild-type sequence and a perfect A T pair with the point mutated sequence Results show that large movements in the pyrimidine base of the A C mismatch cause loss of stacking, especially with its penultimate base, concomitant with a variable mismatch hydrogen bond, including its occasional absence These, in turn, bring about dynamic water interactions in the vicinity of the mismatch Enthalpy loss and the disproportionate entropy gain associ-ated with these are implicassoci-ated as the factors contributing to the increase in free energy and diminished stability of PNAÆDNA duplex with the A C mismatch Absence of these in the isosequential DNA duplex, notwith-standing the A C mismatch, is attributed to the differences in topology of PNAÆDNA vis-a`-vis DNA duplexes It is speculated that similar effects might be responsible for the reduced stability observed in PNAÆDNA duplexes containing other base pair mismatches, and also in mismatch con-taining PNAÆRNA duplexes
Abbreviations
DD wt , DNA duplex with A .C mismatch; DD mut , DNA duplex with A .T pair; LNA, locked nucleic acid; MD, molecular dynamics; PD wt , PNAÆDNA duplex with A .C mismatch; PDmut, PNAÆDNA duplex with Watson and Crick A .T pairing; PNA, peptide nucleic acid; RMSD, root mean square deviation; T m , melting temperature; WC, Watson and Crick.
Trang 2base pairs leading to a PNAÆDNAÆPNA triplex (triplex
invasion) Duplex strand invasion mechanism has the
advantage of targeting any sequence in a DNA duplex,
without the stringent prerequisite of a polypurine tract
as in the conventional triplex-mediated transcription
repression
PNAÆDNA duplexes are more stable than their
iso-sequential DNA duplexes at moderate salt levels, as a
consequence of reduced electrostatic repulsion caused
by the conspicuous absence of phosphates in the PNA
strand [11,12] Another distinctive characteristic of
PNAÆDNA complex formation has been the high
degree of discrimination for sequence selectivity with
tion⁄ suppression of DNA target using PCR clamping [25–27], and selective suppression of replication [28] and gene expression [29,29a] by suitable choice of base sequence in PNA A case in point is its utility in selec-tive inhibition of gene expression in the mutational hotspots of ras oncogenes Normal ras proto-oncogenes express p21, an important signal transduc-tion protein, and a single mutatransduc-tion at one of the few critical positions of ras proto-oncogenes results in a single amino acid substitution in p21 [30] causing malignancy [31] One such point mutation, occurring
in codon 12 of the Ki-ras proto-oncogene, replaces GGT with GAT [32] (capped region in Scheme 1) in one of the alleles of pancreatic cells This leads to pan-creatic cancer [32], as Asp (GAT) replaces Gly (GGT)
in p21 A selective inhibition of the mutated Ki-ras proto-oncogene can be effected by designing a PNA so
as to form a mismatch (PDwt) with the wild-type allele (unmutated proto-oncogene), and a perfect WC base pair (PDmut) with the mutated allele (mutated proto-oncogene) The logic is that the former, in view of the mismatch, is rendered a less stable PNAÆDNA complex promoting normal expression, while the latter (mutated) forms a stable PNAÆDNA duplex (without mismatch) causing inhibition of gene expression Using this strategy, a differential proliferation effect of the wild-type (with A C mismatch), and mutated (with A T pair) alleles of Ki-ras proto-oncogene, has been reported recently [29,29a] Needless to say, a structure-based rationale is obligatory to comprehend the causa-tive factors for the destabilization of PNAÆDNA in the presence of mismatch compared to DNA duplex Inci-dentally, no structural information either from NMR, X-ray crystal structure or modelling is available for PNAÆDNA duplex with a mismatch It is in this con-text, molecular dynamics (MD) simulations have been carried out on PNAÆDNA and DNA duplexes, formed out of a sequence present in the Ki-ras proto-onco-gene, in the presence and absence of an A C mis-match Results reveal that enthalpic loss and the concomitant, but disproportionate entropic gain due to interrupted stacking, fluctuating nature of the hydro-gen bond and water organization in the vicinity of the mismatch might be the contributing factors for the increase in free energy and diminished stability of PNAÆDNA vis-a`-vis DNA duplex
Fig 1 Schematic representation of a section of peptide nucleic
acid (PNA) chain along with notations for the backbone and side
chain torsion angles: a(C6¢-N1¢-C2¢-C3¢), b(N1¢-C2¢-C3¢-N4¢),
c(C2¢-C3¢-N4¢-C5¢), d(C3¢-N4¢-C5¢-C6¢), e(N4¢-C5¢-C6¢-N1¢), n(C5¢-C6¢-N1¢-C2¢),
v1(C8¢-C7¢-N4¢-C3¢), v2(N9 ⁄ N1-C8¢-C7¢-N4¢) and v3(C4 ⁄ C2-N9 ⁄
N1-C8¢-C7¢) Planar peptide unit is enclosed in a rectangle Peptide
hydrogen atom alone is shown for clarity.
Trang 3For convenience of discussion, and to be consistent
with the strategy of designing of PNA for gene
suppres-sion through PNAÆDNA duplex formation (see above),
the 15mer PNAÆDNA duplexes formed with an A C
mismatch (wild-type allele) and with WC A T pairing
(mutated allele) are referred to as PDwt and PDmut,
respectively (Scheme 1) Likewise, the corresponding
isosequential DNA duplexes are referred to as DDwt
(with an A C mismatch) and DDmut (with A T
pair), respectively Because base stacking and base
pair-ing interactions are the major sources of stabilization
of nucleic acid duplexes, their comparison, especially in
the vicinity of the mismatch in PDwt compared with
DDwt duplex may give clues towards deciphering the
origin of the destabilization and hence, diminution of
the melting temperature (Tm) in the former
Base stacking in the vicinity of A C mismatch
in PNAÆDNA and DNA duplexes
Intra strand base stacking at the AC(6–7) (Fig 2A) &
CC(7–8) steps (Fig 2B) of the DNA strand, and
GA(23–24) (Fig 2C) and AT(24–25) (Fig 2D) steps of
the PNA strand, flanking the A24 C7 mismatch in
PDwt (Scheme 1), and the corresponding AT(6–7) &
TC(7–8) steps of the DNA strand (Fig 2E,F), and
GA(23–24) & AT(24–25) (Fig 2G,H) steps of the
PNA strand in PDmut(Scheme 1) are monitored
Base stacking at the CC(7–8) step of PDwt (Fig 2B),
and the TC(7–8) step of PDmut (Fig 2F) of the DNA
strand show significant differences This is due to con-siderable movement of cytosine (C7) of the A24 C7 mismatch of PDwt, leading to large fluctuations in its interaction with the adjacent pyrimidine base (C8) This results in hardly any stacking between them Only occasionally, C5-H group of cytosine (C8) over-laps with C7 and, O2 of C7 overover-laps with C8 On the other hand, sustained stacking persists by way of par-tial overlap of T7 and C8 (Fig 2F) at the correspond-ing TC(7–8) step of PDmut A totally unstacked situation is seldom seen here indicating that occur-rence of an A24 C7 mismatch brings about signifi-cant reduction in adjacent base stacking in PDwt
compared to PDmut
On the other hand, stacking at the AC(6–7) step in the DNA strand of PDwt is retained during the entire simulation in spite of the large movement of C7 (Fig 2A) This occurs due to the coordinated move-ments of C7 and A6 which ensure stacking through-out Similarly, stacking persists at the corresponding AT(6–7) step in PDmut(Fig 2E) through interaction of A6 with either the six-member ring of T7 or through the methyl group and⁄ or O4 of T7 Thus, base stack-ing prevails at the AC(6–7) step (Fig 2A) of PDwt, and the AT(6–7) step of PDmut(Fig 2E) Likewise, the extent of intra strand base stacking at the GA(23–24) and AT(24–25) steps of the PNA strand in both PDwt
(Fig 2C,D) and PDmut (Fig 2G,H) is essentially sim-ilar Thus, A24 C7 mismatch leads to an almost complete loss of stacking only at the CC(7–8) step (PDwt), while the stacking is maintained in the other steps that flank the mismatch
Scheme 1 Sequences encompassing codon 12 (capped) of the Ki-ras proto-oncogene of wild type (wt) and mutated (mut) alleles Bold-italic regions in both wild-type and mutated sequences represent the PNAÆDNA duplex Mismatch (wild-type) and the corresponding ideal WC base pairs (mutated) are underlined The C- and N-termini of the PNA are considered as equivalent to 3¢ and 5¢ ends of a nucleic acid chain, respectively.
Trang 4It is clear from Fig 3A–D that presence of an A C
mismatch in DDwt, seemingly does not influence
adja-cent base stacking at the mismatch site Although
stacking at the CC(7–8) step is found to be only
margi-nal in DDwt during the first 220 ps, quite similar to
that seen in the PDwt, it is enhanced significantly
beyond 220 ps, so much so that almost a complete
overlap of adjacent pyrimidines is observed (Fig 3B) Stacking interactions at the neighbouring AC(6–7), GA(23–24) and AT(24–25) steps of the A24 C7 mis-match site are also maintained (Fig 3A,C,D) It is noteworthy that although stacking at the AC(6–7) step fluctuates, a complete loss of stacking is seldom found (Fig 4A) Stacking persists either through the overlap
A
B
C
D
E
F
G
H
Fig 3 Stereo diagram of adjacent bases at various steps flanking the A24 .C7 mis-match in DD wt : (A) AC(6–7); (B) CC(7–8); (C) GA(23–24) and (D) AT(24–25), and their equivalent steps in DDmut: (E) AT(6–7); (F) TC(7–8); (G) GA(23–24) and (H) AT(24–25) Notice that stacking prevails in all the steps, both in DDwtand DDmut C7 and A24 involved in A .C mismatch in DD wt and the equivalent T7 and A24 in DD mut are col-oured red Trajectories corresponding to every 20 ps are shown.
F
G
H
B
C
D
Fig 2 Stereo diagram of adjacent bases at various steps flanking the A24 .C7 mis-match in PDwt: (A) AC(6–7); (B) CC(7–8); (C) GA(23–24) and (D) AT(24–25), and their equivalent steps in PDmut: (E) AT(6–7); (F) TC(7–8); (G) GA(23–24) and (H) AT(24–25) Note the interruption of the stack at the CC(7–8) step (B) in PDwt, while base stack-ing prevails at all the steps in PDmut(E–H) Large movements of C7 at CC(7–8) step (B) are apparent C7 and A24 bases involved in
A .C mismatch in PDwtand, the equivalent T7 and A24 bases in PD mut are coloured red Trajectories corresponding to every
20 ps are shown.
Trang 5of the amino group of C7 with A6 (Fig 4B) or
through the overlap of the six-member ring of C7 with
A6 (Fig 4C) Thus, unlike in PDwt (Fig 2B),
uninter-rupted stacking prevails at all the steps of DDwt
Interestingly, the extent of stacking at the TC(7–8)
step of DDmut with A T pair (Fig 3F), is similar to
that at the CC(7–8) step of DDwt (Fig 3B)
Further-more, it is evident that although the exact mode of
stacking interactions at the AT(6–7) (Fig 3E), GA(23–
24) (Fig 3G) and AT(24–25) (Fig 3H) steps in DDmut
appear to be different from the equivalent steps in
DDwt(Fig 3A,C,D), the degree or extent of stacking is
comparable This suggests that the stacking interaction
persists in the adjacent steps of both DDwtand DDmut
On the other hand, as noted above, stacking is
inter-rupted in PNAÆDNA duplex with an A C mismatch
Variation of A C mismatch hydrogen bond
in PNAÆDNA and DNA duplexes Fluctuations in the position of C7 of the A24 C7 mismatch in PDwt discussed above are also found to influence the nature of A24 C7 mismatch hydrogen bond It is found that hydrogen bond fluctuates between N6(A24) N3(C7) and⁄ or N6(A24) O2(C7) (Fig 5) As C7 approaches A24, it engages in N6(A24) N3(C7) hydrogen bonding, and when C7 moves away from A24 along the major groove, the other possible hydrogen bonding schemes emerge (Fig 5A–C) Extreme movement of C7 away from A24 can even result in the absence of both the hydro-gen bonds (Fig 5D) These are apparent in Fig 5E
MD simulations extended up to 4 ns further substanti-ates the variable nature of the hydrogen bond (Fig 5) These clearly indicate the absence of a stable hydrogen bond for the A24 C7 mismatch in PDwt
In sharp contrast, a stable N1(A24) .N4(C7) hydro-gen bond (Fig 6B) prevails in DDwt, although the ini-tial N6(A24) N3(C7) hydrogen bond (Fig 6A) lasts for a short duration (200 ps) (Fig 6C–F) The transi-tion to the favoured hydrogen bond occurs as a result
of movement of A24 rather than C7 (of the DNA strand) as found in PDwt and persists till the end of
4 ns dynamics Further, A C mismatch hydrogen bond in DDwt is different from that found in PDwt
(Fig 5) An earlier MD simulation (just over 100 ps) based on NMR data on DNA duplex, pointing to the Ki-ras proto-oncogene having an A23 C8 mismatch (Scheme 1) instead of A24 C7 as in the present study, has indicated the possibility of all the three schemes for A23 C8 mismatch hydrogen bonding [33], but without preference for any one of them However, it is found here that A24 C7 favours N1(A24) .N4(C7) hydrogen bond
In any case, the present analysis clearly points to the greater changeability and destabilization of the A C mismatch hydrogen bond in PNAÆDNA than in DNA duplex As expected, these bring forth significant varia-bleness in the water interactions surrounding the mis-match
Water interaction in the vicinity of A C mismatch
Figure 7A–L depicts the nature of water interaction in the neighbourhood of A24 C7 mismatch in PDwt Water interaction along the minor groove side of A24 C7 mismatch is conserved to the extent that either N1(A24) or O2(C7) or both, are involved in interaction with water This is true irrespective of
Fig 4 Stacking interactions seen at AC(6–7) step of DD wt Note
the prevalence of stacking interaction (B and C) almost throughout
dynamics (see also text) despite the fluctuations Complete loss of
stacking interaction is seldom seen (A).
Trang 6the presence or absence of N6(A24) .N3(C7) and⁄ or
N6(A24) O2(C7) hydrogen bond On the other hand,
water interaction along the major groove is influenced
by the nature of the mismatch hydrogen bond When
N3(C7) is not involved in hydrogen bond with
N6(A24), it engages itself in a variety of interactions
with water along the major groove side as shown
N6(A24) N3(C7) and N6(A24) O2(C7) hydrogen
bonds due to the displacement of the C7 towards the
major groove, N3(C7) and O2(C7), both are engaged
in interaction with water (Fig 7D) These are
demon-strative of significant fluctuations in the water structure
in the vicinity of A24 C7 mismatch in PDwt In con-trast, such variation is not observed in DDwt due to the strong preference for N1(A24) N4(C7) hydrogen bond (Fig 7M–T) As a result, N6(A24) and N4(C7) are involved in a variety of water interaction on the major groove side (Fig 7N–T) Similarly, N3(A24), N3(C7) and O2(C7) are also engaged in water interac-tion most of the time (Fig 7N–T) Thus, it is apparent that water interaction associated with the atoms parti-cipating in A24 C7 mismatch does not show fluctu-ation as in the case of PDwt
Fig 5 Interaction between the A24 (blue) and C 7 mismatch bases in PD wt (A–D) and variation of N6(A24) .N3(C7) and N6(A24) .O2(C7) hydrogen bond distances (F & H), and angles (G & I) over 4 ns dyna-mics Large movement of C7 and the asso-ciated variable hydrogen bonding pattern for A24 .C7 mismatch are clear from the superposition (E).
Trang 7Conformation of the PNA strand in PNAÆDNA
duplexes
Like in DNA, backbone conformation of the PNA
scaffold is governed by six backbone torsion angles,
a(C6¢-N1¢-C2¢-C3¢), b(N1¢-C2¢-C3¢-N4¢),
c(C2¢-C3¢-N4¢-C5¢), d(C3¢-N4¢-C5¢-C6¢), e(N4¢-C5¢-C6¢-N1¢) and
f(C5¢-C6¢-N1¢-C2¢) These are found to be confined to
the trans⁄ gauche–, gauche+, gauche+, gauche+, near
cis and trans range of conformations, respectively, in
both PDmut and PDwt (Fig S1A–F of Supplementary
material) The side chain torsion angles,
v1(C8¢-C7¢-N4¢-C3¢), v2(N9⁄ N1-C8¢-C7¢-N4¢) and v3(C4⁄
C2-N9⁄ N1-C8¢-C7¢) favour the cis, trans ⁄ gauche– and
gauche+ conformations, respectively (Fig S1G–I of
Supplementary material) It is noteworthy that both backbone, as well as side chain, conformations of the PNA strand observed in the present study generally fall in the same range of conformational angles seen in the crystal structures of PNAÆDNA duplex [34] and (PNA)2ÆDNA triplex [35] These are also broadly similar to the results obtained from earlier MD simu-lations on PNAÆDNA complexes [36,37] Some differ-ences seen from the NMR structure may be due to under-determination of the backbone structure by NMR as acknowledged by the authors [38] Inciden-tally, a designed PNA analogue with b¼ gauche+
region, similar to that observed in the current investi-gation, readily forms complex with both DNA and RNA [39–41]
A
B
Fig 6 Hydrogen bonding schemes (A and
B) observed for the A24 .C7 mismatch in
DDwt Variation of hydrogen bond distances
(C and E) and angles (D and F) over 4 ns
dynamics Note the strong preference for
N1(A24) .N4(C7) hydrogen bond beyond
220 ps (C and D).
Trang 8Base backbone hydrogen bonds and a(N1¢-C2¢)
and e(C5¢-C6¢) correlation in the PNA strand
Interestingly, a near-neighbour bond correlation between
the torsion angles a(N1¢-C2¢) and e(C5¢-C6¢) associated
with the peptide unit is recognized It is observed that
whenever a undergoes a conformation change from the
most preferred trans⁄ gauche–to the gauche+
conforma-tion, a concomitant change occurs in e from a cis to a
trans conformation (Fig 8) as found earlier [42] These
ensure stacking as well as the WC hydrogen bond
(Fig 9) Other transitions lead to a totally unstacked situation (data not shown) Further, the (gauche+, trans) conformational state for (a,e) promotes an intramolecu-lar O6¢ H-N2 (G) hydrogen bond between guanine and the carbonyl of the peptide (Fig 10I) However, this is not possible for the (trans⁄ gauche–, cis) conforma-tion as O6¢ orients towards the solvent with the amide (N1¢) hydrogen pointing inside the helix On the other hand, this facilitates in the formation of hydrogen bond with N3 of purines and O2 of pyrimidines either directly (Fig 10A,C,E,G) or through water molecules
Fig 7 Interaction of water (orange) with A24 .C7 mismatch in PD wt (A–L) and DD wt
(M–T) during the dynamics Variation in hydration pattern in PD wt (A–L) depending
on the A24 .C7 mismatch hydrogen bond-ing is readily apparent.
Trang 9(Fig 10B,D,F,H) Interestingly, water mediated N3 N1¢
[34] and O2 N1¢ [34,35] interactions are found in the
minor groove side of the PNAÆDNA duplex [34] and
(PNA)2ÆDNA triplex [35] crystal structures
PNAÆDNA duplex structure
Average structure of the central 11mer of PDwt over
2.5 ns dynamics is shown in Fig 11A Root mean
square deviation (RMSD) of the entire trajectory (2.5 ns) with respect to the average structure lies in the range 0.7–2.5 A˚ for both PDwtand PDmut
Average value of helical twist corresponding to the central 11mer is 27 in both PDwtand PDmutleading
to a 13-fold duplex This is similar to that observed in
an NMR study of a PNAÆDNA hybrid [38] Average value of rise, slide, X-displacement and propeller twist correspond to values around 3.3 A˚, )1.2 A˚, )3.7 A˚ and )10.7, respectively, for PDwt, and 3.3 A˚,)1.3 A˚, )4.0 A˚ and )10.6, respectively, for PDmut Average widths of minor and major grooves are around 9.5 A˚ and 25 A˚ in both PDwtand PDmut
Sugar puckers in DNA strands favour the C2¢ endo conformation in both PDwt and PDmut Interestingly, C7 involved in A C mismatch seems to favour the C4¢ exo sugar pucker, although C2¢ endo is seen dur-ing the dynamics (data not shown)
In general, stacking interaction is nearly similar in both PDmutand PDwt(data not shown) except at steps
on either side of the mismatched A C hydrogen bond
DNA duplex structure Average structure of the central 11mer of DDwt over
2 ns dynamics is shown in Fig 11B RMSD of the entire trajectory (2 ns) corresponding to DDwt and
DDmut varies from 1.2–2.1 A˚ and 1.0–2.8 A˚, respect-ively, with respect to their average structure Even
Fig 8 Correlation between the backbone torsion angles, a(N1¢-C2¢)
and e(C5¢-C6¢) in PD wt (red) and PDmut (black) Notice the
prefer-ence for (a,e) (trans ⁄ gauche – , near cis) compared to (a,e)
(gauche+, trans) conformation.
Fig 9 Stereo plots illustrating the stacking interaction at the GC step when (a,e) (trans ⁄ gauche –
, near cis) (A and B) and (a,e) (gauche+, trans) (C and D) conformations O6¢(G) .N2(G) hydrogen bond (C and D) is shown in dotted line (see also Fig 10 and text) Hydrogens at C2¢, C3¢, C5¢ and C8¢ are not shown for clarity.
Trang 10though RMSD is rather large for DDmut during the
first 500 ps of the dynamics, it stabilizes later
RMSD of DDmut and DDwt falls between 1.0 and
2.0 A˚, beyond 500 ps representing the equilibrium
state
The overall conformation of the helix is of B type Average value of the major groove width is 16.8 A˚ for
DDwtand 18.0 A˚ for DDmut, while the average widths
of the minor groove are 11.4 A˚ and 11.3 A˚ for DDwt and DDmut, respectively
Fig 10 Dependence of backbone .base hydrogen bond interactions in PNA on a and
e correlation Note that hydrogen bond between N1¢ (backbone) and base (O2 ⁄ N3) may be direct (A,C,E,G) or through water (B,D,F,H) when (a,e) (trans ⁄ gauche – , near cis) and N1¢ .N1¢ repeat is compact (5.5 A˚) Direct hydrogen bond between O6¢ and N2 (I) is seen for (a,e) (gauche + , trans) when N1¢ .N1¢ repeat is extended (6.5 A˚) Hydro-gens at C2¢, C3¢ and C5¢ are not shown for clarity.