A Simple Cytosine-to-G-Clamp Nucleobase Substitution Enables Chiral γ-PNAs to Invade Mixed-Sequence Double Helical B-form DNA

Graphical Abstract

Watson-Crick recognition of double helical B-form DNA by chiral γ-PNAs: C→G-clamp nucleobase substitution provides the necessary energetics for chiral γ-PNAs to invade mixed-sequence B-DNA

Nature utilizes Watson-Crick base-pairings as a means to store and transmit genetic information because of their high fidelity. These specific A-T (or A-U) and G-C nucleobase interactions, in turn, provide chemists and biologists with general paradigm for designing molecules to bind to DNA and RNA. With knowledge of the sequence information, one can design oligonucleotides to bind to just about any parts of these biopolymeric targets simply by dialing-in the corresponding nucleobase sequence based on these digital base-pairing rules. Although conceptually simple, such principles in general can only be applied to the recognition of single-stranded DNA or RNA, but not the double-stranded form. The reason is that in double helical DNA (or RNA) not only are the nucleobases already occupied, they are buried within the double helix.¹ Such molecular encapsulation imposes a steep energetic barrier on the designer molecules. To establish binding not only would they need to be able to gain access to the designated nucleobase targets, which are blocked by the existing base-pairs, they would also need to be able to compete with the complementary DNA strand to prevent it from re-annealing with its partner—a task that has rarely been accomplished by any class of molecules.

To circumvent this challenge, most of the research effort to date has been focused on establishing principles for recognizing chemical groups in the minor and major groove instead because they are more readily accessible and energetically less demanding.² While impressive progress has been made on this front, especially in the development of triplex-forming oligonucleotides,^3–5 polyamides^6–8 and zinc-finger binding peptides,^9–13 the issues of sequence selection, specificity and/or target length still remain for many of these approaches.^13–16 In the last decade, however, several studies have shown that peptide nucleic acid (PNA), a particular class of nucleic acid mimics comprised of pseudopeptide backbone (Scheme 1), could invade double-stranded DNA (dsDNA).^17–21 This finding is significant because it demonstrates that the same Watson-Crick base-pairing principles that guide the recognition of single-stranded DNA and RNA can also be applied to dsDNA. Aside from the simplicity, this recognition strategy is general and could potentially be applied to any sequence or target length—just as in the recognition of single-stranded DNA or RNA. The downside to this approach, however, is that PNA can only recognize homopurine and homopyrimidine targets. Mixed-sequence PNAs do not have sufficient energetics to invade double helical B-DNA. Though double-duplex invasion strategy has been developed to try to overcome this energetic barrier,²¹ the issue of sequence selection still remains due to complication with self-quenching.²² In this Communication, we show that a simple nucleobase substitution, replacing cytosine with G-clamp (Scheme 1), provides the necessary energetics for chiral γ-PNAs to invade mixed-sequence B-DNA. Unlike the double-duplex invasion strategy which requires two strands of PNAs, only a single strand of γ-PNA is required for binding to B-DNA.

Scheme 1.

Chemical structures of PNA, γ-PNA, C-G and X-G base-pairs along with the sequence of the oligomers used in this study. Bold letters indicate γ-backbone modifications.

Recently, we showed that randomly-folded, single-stranded PNAs can be preorganized into a right-handed helix simply by installing L-alanine-derived, S-chiral center at the γ-position of the N-(2-aminoethyl) glycine backbone unit.²³ These helical γ-PNAs exhibit strong binding affinity and sequence selectivity for DNA and RNA and are capable of invading mixed-sequence dsDNA as demonstrated by in situ footprinting (data not shown)—however, the invasion complex is not sufficiently stable under prolonged electrophoresis, a necessary condition to separate the bound from the unbound DNA target.²⁴ This result indicates that the binding free energies of these γ-PNAs are already within the invasion threshold of B-DNA. The additional binding free energies needed to stabilize the invasion complex, therefore, may not be much more and such could potentially be attained by replacing a cytosine nucleobase with 9-(2-guanidinoethoxy) phenoxazine (G-clamp, X)—a cytosine analogue that can form five H-bonds with guanine in addition to providing extra base-stacking as the result of the expanded phenoxazine ring system.²⁵ Prior studies showed that a single C→X nucleobase substitution can stabilize a PNA-DNA duplex by as much as 23 °C.^26,27 This level of stabilization may be sufficient for γ-PNAs to invade and stabilize the B-DNA invasion complex. To test this hypothesis, we synthesized a series of dodecameric γ-PNA oligomers (PNA2 through 4) as shown in Scheme 1, where the cytosine nucleobase was systematically replaced with G-clamp, and characterized their conformation, thermal stability and DNA strand invasion capabilities using a combination of spectroscopic and biochemical techniques.

To confirm that these γ-PNA oligomers adopted helical structures, we measured the CD spectra of PNA2 through 4 and compared to that of the unmodified PNA (PNA1). Consistent with our earlier result,²⁴ we did not observe noticeable CD signals for PNA1 in the 220–300 nm nucleobase absorption regions (Figure 1), indicating the lack of a helical structure. We ruled out the possibility of PNA existing as a recemix mixture with an equal proportion of a left- and right-handed helix, as previously suggested by MD simulations,²⁸ on the basis of multi-dimensional NMR analyses.²³ However, in the case of PNA2 through 4, we observed pronounced CD signals, with biphasic exciton coupling patterns characteristic of a right-handed PNA-DNA double helix.²⁹ Substituting C with X did not appear to have a significant effect on the overall conformation of the oligomers, as judged from the similarities in the CD profiles of PNA3 and PNA4 to that of PNA2. A small difference in the amplitudes in the 200–250 nm regions could be attributed to the difference in the absorption strength of the cytosine and G-clamp nucleobase, and/or variations in the backbone conformation as the result of these oligomers trying to accommodate the more sterically hindered G-clamp nucleobase. In addition to CD measurements, we have also attempted to record the melting transitions (T_ms) of the corresponding PNA-DNA hybrid duplexes for this particular series of oligomers, but to no avail because the T_ms were too high to be measured accurately by UV-spectroscopic technique. The T_m of the unsubstituted PNA2 with its complementary (dodecameric) DNA strand alone, at 5 μM strand concentration each in 10 mM NaPi buffer, is already in excess of 90 °C.²⁴ Because the binding affinities of these γ-PNAs are already exceptionally high, replacing C with X should enable them to bind to DNA with an even greater affinity—perhaps sufficient to invade and form a stable complex with double helical B-DNA.

Figure 1.

CD spectra of single-stranded PNA (PNA1) and γ-PNA (PNA2 through 4) oligomers at 5 μM strand concentration each in 10 mM NaPi buffer (pH 7.4), recorded at room temperature.

To determine whether these γ-PNA oligomers can invade dsDNA, we performed electrophoretic mobility shift assay. A 171-bp PCR fragment containing an internal binding site (Scheme 1S, Supporting Information) was incubated with different concentrations of γ-PNA oligomers in 10 mM sodium phosphate buffer (pH 7.4) at 37 °C for 2 hrs. The mixtures were then separated on non-denaturing polyacrylamide gel (PAGE) and stained with SYBR-Gold for visualization. Consistent with the earlier finding,²⁴ our result showed no evidence of binding for PNA2 (Figure 2, lanes 2 & 3). However, in the case of PNA3 and PNA4, we noticed a distinct slow-moving band, with the intensity gradually increased with increasing oligomer concentrations (Figure 2, compare lanes 5 with 4 and lanes 7 with 6). Formation of this complex appeared to be complete at a PNA:DNA ratio of 3:1 for PNA4, as judged from the nearly complete disappearance of the unbound DNA target. Under identical condition, only ~ 20% binding was complete for PNA3. In addition to the binding efficiency, the mobility of the shifted complex also appeared to be different for the two oligomers, lower for PNA4 as compared to that of PNA3. The difference in mobility could be attributed to the difference in the overall size and charge density of the complex, reflecting the difference in the number of the G-clamps present in the oligomers (two for PNA4 and one for PNA3). In this case, formation of the shifted band was only observed in the presence of a perfectly-matched target (Figure 3, lanes 1–3). Introduction of an inverted, single-base mismatch (Scheme 1S) completely abolished the binding (Figure 3, lanes 4–6). This result shows that PNA4 binding occurred in a sequence-specific manner.

Figure 2.

Gel-shift assay of 171bp DNA fragment containing a perfectly-matched target with various concentrations of γ-PNA oligomers. 0.4 μM of DNA (duplex concentration) was incubated with 0 (lane 1), 0.4 (lanes 2, 4 and 6) and 1.2 μM (lanes 3, 5 and 7) of each respective γ-PNA oligomer in 10 mM NaPi buffer (pH 7.4) at 37 °C for 2 hrs. The mixtures were separated on 10% non-denaturing PAGE for 3 hrs at 5V/cm and then stained with SYBR-Gold.

Figure 3.

Gel-shift assay of PNA4 with perfect-match (PM) and single-base mismatch (MM) DNA targets. Samples were prepared by incubating 0.4 μM of 171bp DNA fragment containing PM (lanes 1–3) and MM (lanes 4–6) targets with 0 (lanes 1 and 4), 0.4 (lanes 2 and 5) and 1.2 μM (lanes 3 and 6) of PNA4 in 10 mM NaPi buffer (pH 7.4) at 37 °C for 2 hrs. The samples were separated and stained under identical condition as before.

To further demonstrate that the observed binding occurred through a strand invasion mechanism, we performed diethyl pyrocarbonate (DEPC) probing assay. DEPC is an acylating reagent known to react selectively with adenines and to a lesser extent with guanines within a single-stranded or a perturbed region of a DNA duplex, which can be revealed in the form of strand cleavage following piperidine treatment.^30,31 Sequence-specific strand invasion of dsDNA by PNA4 is expected to result in a local displacement of the homologous DNA strand, which can be readily detected by DEPC assay. Consistent with the strand invasion mechanism, DEPC-treatment of the bound DNA complex, with the 3’-end of the homologous DNA strand labeled with P-32, revealed selective cleavage at the adenine residues on the homologous DNA strand across from the binding site (Figure 4). Strand cleavage was observed even at a PNA:DNA ratio of 0.5:1, but less intense as compared to that at a 1:1 ratio (Figure 4, lanes 2 and 3). Although there are four adenines within the expected locally-displaced DNA strand, only the two middle adenine residues showed significant cleavage pattern, while very little, if any, was observed for the two residues localized near the termini. The difference in the cleavage intensity could be attributed to the difference in the degree of interaction between these adenine residues and their neighboring nucleobases. Because they are located within the melted junction of the DNA duplex, the two flanking adenine residues could still be interacting with the adjacent nucleobases, or at least to a greater extent than those residing in the middle of the looped-out strand. This may explain why only the middle adenines are more susceptible to DEPC-treatment than those near the termini. Similar to the data obtained from the gel-shift assay, strand cleavage was only observed in the presence of the perfect-match target (Figure 4, compare lanes 1–3 with lanes 4–6). These results, taken together, show that PNA4 binding occurred in a sequence-specific manner through a strand invasion mechanism.

Figure 4.

DEPC-treatment following incubation of PNA4 with PM (lanes 1–3) and MM (lanes 4–6) DNA targets. The homologous DNA strand was 3’-labeled with P-32. The samples were prepared by incubating 10,000 cpm of the labeled and 0.4 μM of the cold (unlabeled) DNA with 0, 0.2 and 0.4 μM of PNA4 in 10 mM NaPi buffer at 37 °C for 2 hrs, followed by DEPC-treatment. The samples were separated on 10% denaturing PAGE, and the cleavage patterns were visualized by autoradiography.

Next, we investigated the effect of ionic strengths on PNA4 binding since salt has been shown to have a profound effect on the efficiency of DNA strand invasion.^19,32,33 To assess this effect, we incubated PNA4 with the perfect-match DNA target at 3:1 (PNA:DNA) ratio in buffers containing 10 mM NaPi and various concentrations of MgCl₂, including a simulated physiological salt concentration (150 mM KCl and 2 mM MgCl₂).³⁴ The mixtures were separated by non-denaturing PAGE and stained with SYBR-Gold as before. Inspection of Figure 5 reveals that while the binding efficiency of PNA4 with its target gradually decreased with increasing Mg²⁺ concentrations, a significant fraction (~60 %) of the DNA target was found in the bound state after 2 hrs of incubation at 37 °C in 10 mM NaPi and 2 mM MgCl₂—corresponding to less than a 2-fold decrease in the binding efficiency as compared to that in the original buffer with just 10 mM NaPi (Figure 5, compare lanes 6 with 2). Although the amount of the complex formed under physiological salt concentration was significantly less (only ~ 10%) than that in the 10 mM NaPi baseline, strand invasion still took place under this condition as evident from the presence of the retarded band (Figure 5, compare lanes 7 with 2). This result indicates that it is feasible to invade double helical B-DNA under physiological temperature and ionic strength—given that sufficient binding free energy is provided.

Figure 5.

Binding of PNA4 with PM DNA target at 3:1 ratio in buffer containing 10 mM NaPi and various concentrations of MgCl₂ (lanes 3–6), along with a simulated physiological salt concentration (lane 7). The samples were incubated at 37 °C for 2 hrs prior to separation and staining as before.

In summary, we have shown that γ-PNAs containing G-clamp in place of cytosine nucleobase can invade mixed-sequence B-DNA. In this case, only a single strand of γ-PNA is required and the invasion occurs through direct Watson-Crick base-pairings. In 10 mM NaPi buffer and at 37 °C, the invasion of PNA4 into DNA was complete at a 3:1 PNA:DNA ratio within 2 hrs of incubation. We attributed this relatively fast invasion kinetics to the preorganized structure of γ-PNAs and the stability of the invasion complex to the enhanced binding affinity of the X-G base-pair. Although the efficiency was lower, strand invasion still took place at elevated ionic strengths. This study shows that it is feasible to target mixed-sequence B-DNA with γ-PNAs not only at relatively low ionic strengths but also at physiological salt concentration. However, presently, it is not clear whether the reduction in the invasion efficiency is predominantly the result of a slower rate of base-pair ‘breathing’, or an increase in the thermodynamic stability of the dsDNA since both are interrelated and significantly affected by ionic strengths. Prior study with homopyrimidine PNA seems to suggest that it is the former.³⁵ While the initial searching steps for the two systems may be similar, the overall invasion efficiency is likely to be different—faster for the mixed-sequence γ-PNA since only a single strand is required for binding and stabilization, as compared to two strands for homopyrimidine PNA. Additional studies will be required to tease out these various contributions. Our ability to fine-tune the energetics of γ-PNAs without the need to change the nucleobase sequence or the oligomer size, by replacing C with X (or other nucleobases with their more thermodynamically stable synthetic analogues), should enable us to address this question in the future—which an important first step toward developing a more effective oligonucleotide platform for targeting double-stranded B-DNA on the basis of Watson-Crick pairing.