Optimized DNA-targeting using triplex forming C5-alkynyl functionalized LNA

Abstract

Triplex forming oligonucleotides (TFOs) modified with C5-alkynyl functionalized LNA (locked nucleic acid) monomers display extraordinary thermal affinity toward double stranded DNA targets, excellent discrimination of Hoogsteen-mismatched targets, and high stability against 3′-exonucleases.

Substantial efforts have been invested to develop probes for use in the antigene technology, i.e., sequence-specific targeting of double stranded DNA (dsDNA).¹ The efforts are motivated by the prospect of developing fundamental research tools and new types of therapeutic and diagnostic agents.^2,3 (TC)-motif triplex forming oligonucleotides (TFOs), which bind to polypurine regions in dsDNA targets via Hoogsteen base pairs in the major groove, are the most widely studied DNA-targeting agents,¹ although promising alternatives are emerging.⁴ A drawback of the TFO-approach is the instability of triplexes at physiological pH due to electrostatic repulsion between the three strands and insufficient N3-protonation of cytosines resulting in weak Hoogsteen base pairing. Chemical modification of TFOs is therefore essential to increase triplex stability as well as to protect TFOs from enzymatic degradation.¹ Three classes of nucleotide building blocks have found particular use as TFO-modifications: (a) ribonucleotides and O2′-alkylated analogs hereof as they favorably preorganize the TFO strand for hybridization to dsDNA targets,⁵ (b) conformationally restricted nucleotides with RNA-like furanose puckers such as locked nucleic acid (LNA, monomer V, Fig. 1)⁶ or analogs thereof⁷ in what is an extension of the TFO preorganization concept, and (c) C5-alkynyl functionalized pyrimidine DNA monomers, as they promote enhanced base stacking interactions in triplexes.⁸ Interestingly, building blocks that combine C5-alkynyl functionalization and RNA-character induce greater triplex stabilization than the respective C5-alkynyl DNA and RNA TFO monomers by themselves.⁹ In light of this, we hypothesized that C5-alkynyl functionalized LNA monomers X and Y (Fig. 1) would synergistically integrate beneficial properties from two compound classes, i.e., (1) high thermal affinity toward dsDNA and partial resistance toward nucleases (LNA-component), and (2) resistance toward nucleases due to the presence of steric shields along with favorable contributions to binding affinity (C5-substituent).

Fig. 1.

Structures of LNA and novel C5-alkynyl functionalized LNA monomers.

The corresponding phosphoramidites¹⁰ of the C5-ethynyl and C5-propargylamine functionalized LNA monomers X and Y and the commercially available LNA thymine phosphoramidite were incorporated one, three or six times into a 15-mer (TC)-motif DNA TFO sequence (Table 1) that has previously been used to evaluate hybridization properties of LNA,^7d ENA^7d and 2′,4′-BNA^NC7e TFOs. TFOs were synthesized on automated DNA synthesizers (0.2 mmol scale) using standard conditions except for extended coupling (15 min, using 4,5-dicyanoimidazole as activator) and oxidation times (45 s).^† This resulted in stepwise coupling yields of >98% for the LNA monomers. 5-Methyldeoxycytidine rather than deoxycytidine monomers were used to ameliorate the normal pH dependence of (TC)-motif TFOs.^11‡ Following workup and purification, the composition and purityz (>90%) of all synthesized TFOs were verified by MALDI-TOF MS (Table S2)^† and HPLC, respectively.^†

Table 1.

T_m-values and association rate constants (k_on) for triplexes formed by C5-alkynyl functionalized LNA, conventional LNA or DNA TFOs with dsDNA target^a

			Tm [ΔTm/mod]/°C			kon/M–1 s–1
ON	Seq. 5′ → 3′	B =	V	X	Y	V	X	Y
B1	TTT TTmCTBT mCTmCTmCT		40.0 [+11.0]	42.5 [+13.5]	42.5 [+13.5]	3800	4400	ND
B2	TTT TBmC TTBmCTmC BmCT		48.5 [ +6.5]	54.0 [+8.3]	57.0 [+9.3]	3600	4000	3700
B3	TTT TBmC TBT mCBmC TmCT		53.5 [+8.2]	59.5 [+10.2]	60.0 [+10.3]	3900	4100	3900
B4	TTB TBmC BTBmCBmC BmCT		61.0 [+5.3]	67.5 [+6.4]	74.5 [+7.6]	4300	4400	5500

^aT_m's were determined as the first derivative of differential thermal denaturation curves (A₂₆₀vs. T)^6c† recorded in a pH 7.0 phosphate buffer solution containing 140 mM KCl using 1.0 μM of each strand. T_m-values are averages of at least two independent measurements within 1 °C. k_on's were measured in pH 7.2 HEPES buffer.^† k_on's are averages of at least three independent measurements. Data for DNA reference TFO T1 (B = T): T_m = 29.0 °C, k_on = 4400 M^–1 s^–1. Structures of monomers are shown in Fig. 1. ^mC denotes 5-methyldeoxycytidine monomers. dsDNA target sequence (TFO binding region underlined): 5′-GCT AAA AAG AAA GAG AGA TCG-3′, 3′-CGA TTT TTC TTT CTC TCT ACG-5′.

ND = not determined.

Binding to a central target region in a 21-mer dsDNA was first characterized by non-denaturing PAGE using equimolar quantities of TFOs and dsDNA target in a pH 7.2 HEPES buffer containing Mg²⁺. Conventional LNA and C5-ethynyl functionalized LNA TFOs form stable triplexes with dsDNA targets under these conditions (V- and X-series, Fig. S1)^† as a single band of lower mobility than the corresponding dsDNA target was observed in all instances. Triplex formation between C5-propargylamine functionalized LNA TFOs and dsDNA target is incomplete as a band (~20–40% intensity) of identical mobility as the corresponding dsDNA target is observed in addition to the triplex band (Y-series, Fig. S2).^† Triplex formation was, however, driven toward completion using a two-fold excess of C5-propargylamine functionalized LNA TFOs (Fig. S2).^†

After qualitatively establishing triplex formation, the thermal affinity of C5-functionalized LNA TFOs toward the dsDNA target was determined by UV thermal denaturation experiments at pH 7.0 in the absence of Mg²⁺-ions and compared relative to reference DNA and LNA TFOs (Table 1). The thermal denaturation curves of the studied triplexes either exhibit (a) biphasic profiles when thermal denaturation temperatures (T_m's) of triplexes differed substantially from the T_m of the dsDNA target (T_m = 56.0 °C), or (b) monophasic profiles when denaturation curves for triplexes and dsDNA targets overlapped substantially (Fig. 2, Fig. S3 and S4).^† Similar observations in identical sequence contexts have been made with other LNA analogs.^7b,d To facilitate determination of triplex T_m's, differential thermal denaturation curves^6c were obtained by ‘subtracting’ the denaturation curve of the dsDNA target from the ‘raw’ triplex denaturation curve (Fig. S3).^†

Fig. 2.

Representative thermal denaturation profiles of triplexes between reference DNA T1 or TFOs from the B4-series, and complementary dsDNA target. For sequences and conditions see Table 1.

Singly modified C5-alkynyl functionalized LNA TFOs exhibit dramatically increased thermal affinity toward dsDNA targets relative to reference DNA TFO T1 (ΔT_m ≈ +13.5 °C, B1-series, Table 1). Additional modification of TFOs with X/Y monomers results in progressively more stabilized triplexes although less pronounced ΔT_m/mod-values are observed (e.g., compare data of B1-, B2- and B4-series, Table 1). C5-Propargylamine functionalized LNAs (Y-series) stabilize triplexes slightly more efficiently than the corresponding C5-ethynyl functionalized LNAs (X-series), which likely results from reduced electrostatic repulsion between the triplex strands due to partial protonation of the propargylamino substituent, or from improved π-stacking with neighboring nucleobases.^8b Consequentially, C5-propargylamine functionalized LNA TFO Y4 containing six Y monomers as next-nearest neighbours forms exceedingly stable triplexes with dsDNA targets (T_m = 74.5 °C, Table 1).

Importantly, the observed stabilization of triplexes by C5-alkynyl LNA TFOs compares favourably with the corresponding parent LNA TFOs (V-series) as additional increases in T_m-values of between +1.1 °C and +2.8 °C per modification are observed (compare T_m-data for X-, Y-and V-series, Table 1). In fact, direct comparison with previously reported hybridization data for ENA^7d and 2′,4′-BNA^NC7e TFOs in the same sequence, strongly suggests that C5-alkynyl LNA TFOs stabilize triplexes more efficiently than these ‘gold standard’ TFO building blocks. Thus, incorporation of C5-alkynyl functionalized LNA monomers into DNA TFOs is a simple,^† yet powerful approach to improve thermal affinity of TFOs toward dsDNA targets.

To evaluate the influence of C5-alkynyl functionalized LNA monomers on association kinetics of triplex formation, the association rate constant k_on was determined for TFOs by fitting a second order rate equation to the A₂₆₀ decay curve¹³ obtained by mixing pre-annealed dsDNA target with equimolar quantities of TFOs (Fig. S6).^† Generally little variation in the k_on-values between DNA, LNA or C5-functionalized LNA TFOs was observed (~4000 M^–1 s^–1, Table 1), which verifies that LNA-type TFOs primarily influence triplex dissociation kinetics.^6a As an exception hereto, densely modified C5-propargylamine functionalized LNA Y4 exhibits faster triplex association kinetics which most likely is linked to the presence of partially positively charged ‘patches’ of the C5-propargylamine substituent.^9b

Increased affinity of TFOs toward dsDNA targets should preferably be accompanied by concomitant improvements in mismatch discrimination to avoid off-targets effects.^5b Accordingly, the ability of singly modified C5-alkynyl LNA TFOs (B1-series) to discriminate dsDNA targets with a single Hoogsteen base pair mismatch opposite of the modification was evaluated in a thermal denaturation buffer containing 10 mM Mg²⁺ (see footnote Table 2), which is known to accelerate and stabilize triplex formation.^14§,^†

Table 2.

T_m-values of matched and mismatched triplexes in the presence of Mg²⁺-ions^a

	Tm[ΔTm]/°C
ON	A : T (match)	T : A	C : G	G : C
T1	37.0	10.0 [–27.0]	19.5 [–17.5]	13.5 [–23.5]
V1	49.0	10.0 [–39.0]	28.5 [–20.5]	24.0 [–25.0]
X1	51.5	10.0 [–41.5]	24.5 [–27.0]	27.5 [–24.0]
Y1	50.0	10.0 [–40.0]	26.5 [–23.5]	27.5 [–22.5]

^aT_m-values (ΔT_m = change in T_m-value relative to matched triplex). Conditions and TFO-sequences (B1-series) as described in Table 1 except for addition of 10 mM MgCl₂. dsDNA targets = 5′-GCT AAA AAG AMA GAG AGA TCG-3′ :3′-CGA TTT TTC TM′T CTC TCT ACG-5′, where M : M′ = A : T, T : A, C : G and G : C, respectively.

As anticipated, less stable triplexes are generally formed with mismatched dsDNA targets. Introduction of C5-functionalized LNA monomers into TFOs greatly improves discrimination of Hoogsteen TA- and CG-mismatches compared to reference TFOs T1 and V1, while discrimination of GC-mismatches is largely unaffected (Table 2).

Consequentially, the most demanding mismatched dsDNA targets are more efficiently discriminated by C5-functionalized LNA TFOs (ΔT_m = –24.0 °C and –22.5 °C for X1 and Y1, respectively) than by reference TFOs (ΔT_m = –17.5 °C and –20.5 °C, for T1 and V1, respectively). This data demonstrates that C5-alkynyl functionalized LNAs exhibit increased thermal affinity and specificity relative to parent LNAs.

Finally, the resistance of singly modified C5-alkynyl functionalized LNA TFOs (B1-series) against degradation by the 3′-exonuclease snake venom phosphodiesterase (SVPDE) was evaluated, by following the increase in absorbance (hyperchromicity) at 260 nm¹⁵ of TFOs in the presence of SVPDE (Fig. 3). As expected, DNA TFO T1 is rapidly degraded (>85% cleavage after 1 h) whereas singly modified LNA V1 confers some protection against SVPDE-mediated degradation (~70% cleavage after 1 h). C5-Ethynyl and, in particular, C5-propargylamine functionalized LNA TFOs are considerably more resistant against nucleolytic degradation (X1 ~60%, Y1 ~35% after 1 h). This suggests that the C5-substituents act as a steric shield, to ensure greater biostability.

Fig. 3.

Exonuclease (SVPDE) degradation of singly modified C5-functionalized LNA and reference TFOs. Nuclease degradation studies performed in Tris-buffer (50 mM Tris-HCl, 10 mM MgCl₂, pH9.0) at 37 °C using 2 μM of TFO strands and 0.43 μg of SVPDE (snake venom phosphordiesterase). For sequences of T1–V1–X1–Y1,see Table 1.

To conclude, C5-alkynyl functionalized LNAs exhibit unsurpassed TFO-hybridization properties, which are likely to enable efficient DNA-targeting under physiological conditions. Studies along these lines are ongoing.

We appreciate financial support from Idaho NSF EPSCoR, the BANTech Center at the Univ. of Idaho, a University of Idaho Research Office and Research Council Seed Grant, and National Institutes of Health [grant number P20 RR016448 from the COBRE Program of the National Center for Research Resources]. MEØ acknowledges a scholarship from the College of Graduate Studies. We thank the EBI Murdock Mass Spectrometry center for mass spectrometric analyses.