The influence of locked nucleic acid residues on the thermodynamic properties of 2'-O-methyl RNA/RNA heteroduplexes.

The influence of locked nucleic acid (LNA) residues on the thermodynamic properties of 2'-O-methyl RNA/RNA heteroduplexes is reported. Optical melting studies indicate that LNA incorporated into an otherwise 2'-O-methyl RNA oligonucleotide usually, but not always, enhances the stabilities of complementary duplexes formed with RNA. Several trends are apparent, including: (i) a 3' terminal U LNA and 5' terminal LNAs are less stabilizing than interior and other 3' terminal LNAs; (ii) most of the stability enhancement is achieved when LNA nucleotides are separated by at least one 2'-O-methyl nucleotide; and (iii) the effects of LNA substitutions are approximately additive when the LNA nucleotides are separated by at least one 2'-O-methyl nucleotide. An equation is proposed to approximate the stabilities of complementary duplexes formed with RNA when at least one 2'-O-methyl nucleotide separates LNA nucleotides. The sequence dependence of 2'-O-methyl RNA/RNA duplexes appears to be similar to that of RNA/RNA duplexes, and preliminary nearest-neighbor free energy increments at 37 degrees C are presented for 2'-O-methyl RNA/RNA duplexes. Internal mismatches with LNA nucleotides significantly destabilize duplexes with RNA.

INTRODUCTION

Understanding the thermodynamics of nucleic acid duplexes is important for many reasons. For example, such knowledge facilitates design of ribozymes (1), antisense and RNAi oligonucleotides (2–9), diagnostic probes including those employed on microarrays (10–23) and structures useful for nanotechnology (24–27). Many modified residues have been developed for such applications. Examples include propynylated bases (28–30), peptide nucleic acids (5,31–33), N3′–P5′ phosphoramidates (34–38) and 2′-O-alkyl RNA (39–43). A modification that is particularly stabilizing in DNA and RNA duplexes (44–51) is a methyl bridge between the 2′ oxygen and 4′ carbon of ribose to form a ‘locked nucleic acid’ or LNA as shown in Figure 1. McTigue et al. (48) have shown that the enhanced stability due to a single LNA residue in a DNA duplex can be predicted from a nearest-neighbor model.

Figure 1

Covalent structures of sugars used.

Hybridization of oligonucleotides to RNA is important for applications, such as antisense therapeutics (4,8,21,46,52–54), diagnostics (32,33,42,55), profiling gene expression with microarrays (18–20,56), identifying bands by Northern blots of gels (57,58) and probing RNA structure (1,3,15,59–61). Oligonucleotides with 2′-O-alkyl modifications can be particularly useful for these applications because they are easily synthesized (39,43), chemically stable and bind relatively tightly to RNA (39–42). However, for many applications, it is desirable to modulate the binding affinity. For example, sequence independent duplex stabilities would benefit applications that involve multiplex detection, such as microarrays. Here, we show that introduction of LNA into 2′-O-methyl RNA oligonucleotides can increase stabilities of 2′-O-methyl RNA/RNA hybrid duplexes and that the enhancements in stability can usually be predicted with a simple model.

MATERIALS AND METHODS

General methods

High-performance liquid chromatography (HPLC) was performed on a Hewlett Packard series 1100 HPLC with a reverse-phase Supelco RP-18 column (4.6 × 250 mm). Mass spectra were obtained on an LC MS Hewlett Packard series 1100 MSD with API-ES detector or on an AMD 604/402. Thin-layer chromatography (TLC) was carried out on Merck 60 F254 TLC plates with the mixture 1-propanol/aqueous ammonia/water = 55:35:10 (v/v/v).

Synthesis and purification of oligonucleotides

Oligoribonucleotides were synthesized on an Applied Biosystems DNA/RNA synthesizer, using β-cyanoethyl phosphoramidite chemistry (62). For synthesis of standard RNA oligonucleotides, the commercially available phosphoramidites with 2′-O-tertbutyldimethylsilyl groups were used (Glen Research). For synthesis of 2′-O-methyl RNA oligonucleotides, the 3′-O-phosphoramidites of 2′-O-methylnucleotides were used (Glen Research and Proligo). The 3′-O-phosphoramidites of LNA nucleotides were synthesized according to the published procedures with some minor modifications (44,47,63). The details of deprotection and purification of oligoribonucleotides were described previously (64).

UV melting

Oligonucleotides were melted in buffer containing 100 mM NaCl, 20 mM sodium cacodylate, 0.5 mM Na2EDTA, pH 7.0. The relatively low NaCl concentration kept melting temperatures in the reasonable range even when there were multiple LNA substitutions. Oligonucleotide single-strand concentrations were calculated from absorbencies above 80°C and single-strand extinction coefficients were approximated by a nearest-neighbor model (65,66). It was assumed that 2′-O-methyl RNA and RNA strands with identical sequences have identical extinction coefficients. Absorbancy versus temperature melting curves were measured at 260 nm with a heating rate of 1°C/min from 0 to 90°C on a Beckman DU 640 spectrophotometer with a water cooled thermoprogrammer. Melting curves were analyzed and thermodynamic parameters were calculated from a two-state model with the program MeltWin 3.5 (67). For almost all sequences, the ΔH° derived from versus ln (CT/4) plots is within 15% of that derived from averaging the fits to individual melting curves, as expected if the two-state model is reasonable.

Parameter fitting

Free energy parameters for predicting stabilities of 2′-O-methyl RNA/RNA and 2′-O-methyl RNA–LNA/RNA duplexes with the Individual Nearest-Neighbor Hydrogen Bonding (INN-HB) model (64) were obtained by multiple linear regression with the program Analyse-it v.1.71 (Analyse-It Software, Ltd, Leeds, England; ) which expands Microsoft Excel. Analyse-It was also used to obtain parameters for enhancement of stabilities of 2′-O-methyl RNA/RNA duplexes by substitution of LNA nucleotides internally and/or at the 3′ end when the LNAs are separated by at least one 2′-O-methyl nucleotide. Results from versus ln (CT/4) plots were used as the data for the calculations.

RESULTS

Figures 2 and 3 show typical data from optical melting curves, and Table 1 lists the thermodynamic parameters for the helix to coil transition with either no or one LNA nucleotide in the primarily 2′-O-methyl strand of a hybrid with a Watson–Crick complementary RNA strand.

Figure 2

Representative UV melting curves 5′AMCMUMCMCMAM/5′UGGCAGU (1, 52 µM), 5′AMCMUMAMCMCMAM/5′UGGUAGU (2, 46 µM), 5′AMCMUMCMCMAM/5′UGGUAGU (3, 51 µM) and 5′AMUMCMAM/5′UGGUAGU (4, 51 µM). Note that sequence 1 has an ALC mismatch in the middle.

Figure 3

Reciprocal melting temperature versus ln (CT/4) for 5′AMCMUMCMCMAM/5′UGGCAGU (solid squares), 5′AMCMUMAMCMCMAM/5′UGGUAGU (solid circles), 5′AMCMUMCMCMAM/5′UGGUAGU (solid stars) and 5′AMUMCMAM/5′UGGUAGU (solid triangles) in 100 mM NaCl, 20 mM sodium cacodylate, 0.5 mM Na2EDTA, pH 7.0. The thermodynamics of duplex formation were obtained by fitting the data points to the equation, = (R/ΔH°) ln (CT/4) + ΔS°/ΔH°. Here, is the inverse melting temperature in kelvin, R is the gas constant, 1.987 cal K−1 mol−1, CT is the total oligonucleotide strand concentration, and both strands have the same concentration.

Table 1

Thermodynamic parameters of duplex formation between RNA and 2′-O-methyl oligoribonucleotides with and without a single LNA substitutiona

Oligonucleotides (5′ to 3′)	RNA (5′ to 3′)	Average of curve fits				Tm−1 versus log (CT/4) plots
		−ΔH° (kcal/mol)	−ΔS° (eu)	−ΔG37° (kcal/mol)	Tmb (°C)	−ΔH° (kcal/mol)	−ΔS° (eu)	−ΔG37° (kcal/mol)	Tmb (°C)	ΔΔG37° (kcal/mol)c	ΔTmb (°C)
ACUACCA	UGGUAGU	58.5 ± 14.4	166.8 ± 45.7	6.80 ± 0.15	38.5	56.4 ± 6.6	160.0 ± 21.4	6.74 ± 0.23	38.2
						(53.8)	(146.6)	(8.27)	(50.0)
AMCMUMAMCMCMAM	UGGUAGU	51.5 ± 7.7	143.0 ± 24.3	7.18 ± 0.21	41.0	44.4 ± 1.3	120.2 ± 4.0	7.13 ± 0.01	41.3
						(53.8)	(146.6)	(8.27)	(50.0)
ALCMUMAMCMCMAM	UGGUAGU	54.1 ± 4.9	149.3 ± 15.8	7.80 ± 0.10	44.4	52.1 ± 2.0	143.2 ± 6.5	7.72 ± 0.03	44.2	−0.59	2.9
CMCMUMAMCMCMAM	UGGUAGG	65.5 ± 7.1	181.2 ± 21.8	9.27 ± 0.36	50.5	60.5 ± 3.5	165.9 ± 11.0	9.07 ± 0.13	50.6
						(59.5)	(160.3)	(9.74)	(57.2)
CLCMUMAMCMCMAM	UGGUAGG	66.5 ± 3.6	183.1 ± 10.8	9.74 ± 0.27	52.7	60.9 ± 5.6	165.9 ± 17.2	9.47 ± 0.28	52.7	−0.40	2.1
GMCMUMAMCMCMAM	UGGUAGC	62.3 ± 6.9	171.5 ± 21.2	9.10 ± 0.36	50.4	54.3 ± 2.2	146.7 ± 6.9	8.78 ± 0.07	50.4
						(61.0)	(164.6)	(9.90)	(57.6)
GLCMUMAMCMCMAM	UGGUAGC	60.5 ± 4.6	164.6 ± 13.7	9.45 ± 0.34	52.7	52.7 ± 1.7	140.4 ± 5.3	9.11 ± 0.06	53.0	−0.33	2.6
UMCMUMAMCMCMAM	UGGUAGA	50.7 ± 5.9	140.0 ± 18.4	7.25 ± 0.19	41.5	45.2 ± 1.6	122.5 ± 5.3	7.18 ± 0.03	41.5
						(54.8)	(149.6)	(8.38)	(50.4)
ULCMUMAMCMCMAM	UGGUAGA	55.7 ± 3.3	154.4 ± 10.0	7.84 ± 0.19	44.4	47.2 ± 3.0	127.5 ± 9.6	7.64 ± 0.08	44.5	−0.46	3.0
AMCMUMAMCMCMAL	UGGUAGU	60.7 ± 5.5	168.1 ± 17.3	8.54 ± 0.15	47.6	55.2 ± 2.6	150.9 ± 8.1	8.43 ± 0.06	48.0	−1.30	6.7
AMCMUMAMCMCMCM	GGGUAGU	62.9 ± 4.0	176.1 ± 12.6	8.30 ± 0.09	46.0	61.2 ± 2.3	170.4 ± 7.2	8.29 ± 0.03	46.2
						(60.4)	(163.0)	(9.87)	(57.6)
AMCMUMAMCMCMCL	GGGUAGU	71.3 ± 4.9	200.4 ± 15.2	9.17 ± 0.23	48.9	76.6 ± 7.0	216.8 ± 21.9	9.34 ± 0.27	48.8	−1.05	2.6
AMCMUMAMCMCMGM	CGGUAGU	56.8 ± 2.8	157.1 ± 8.9	8.11 ± 0.11	45.9	53.0 ± 2.5	145.3 ± 7.9	8.04 ± 0.05	46.1
						(57.7)	(157.0)	(8.97)	(53.2)
AMCMUMAMCMCMGL	CGGUAGU	61.4 ± 3.9	167.6 ± 11.8	9.44 ± 0.19	52.4	58.3 ± 1.3	158.0 ± 3.9	9.29 ± 0.04	52.4	−1.25	6.3
AMCMUMAMCMCMUM	AGGUAGU	59.6 ± 5.7	168.4 ± 18.9	7.41 ± 0.28	41.2	62.2 ± 7.3	176.8 ± 23.6	7.37 ± 0.26	41.2
						(53.8)	(146.9)	(8.24)	(49.8)
AMCMUMAMCMCMUL	AGGUAGU	59.2 ± 5.2	167.7 ± 17.2	7.19 ± 0.17	40.5	55.6 ± 2.6	155.8 ± 8.5	7.24 ± 0.04	41.0	+0.13	−0.2
AMCMUMAMCMGMUM	ACGUAGU	56.0 ± 5.3	157.7 ± 16.7	7.08 ± 0.16	40.1	50.9 ± 2.1	141.3 ± 6.9	7.05 ± 0.04	40.2
						(52.0)	(143.4)	(7.50)	(45.6)
AMCMUMAMCMGMUL	ACGUAGU	55.9 ± 3.5	157.6 ± 11.0	7.00 ± 0.17	39.6	51.1 ± 3.9	142.0 ± 12.8	7.03 ± 0.09	40.1	+0.02	−0.1
AMCLUMAMCMCMAM	UGGUAGU	58.3 ± 5.4	160.3 ± 16.9	8.57 ± 0.22	48.2	53.1 ± 2.9	143.9 ± 9.3	8.42 ± 0.09	48.4	−1.29	7.1
AMCMULAMCMCMAM	UGGUAGU	54.4 ± 4.3	147.4 ± 13.7	8.62 ± 0.11	49.4	57.4 ± 2.7	157.1 ± 8.6	8.66 ± 0.08	48.9	−1.53	7.6
AMCMUMALCMCMAM	UGGUAGU	58.9 ± 3.5	162.4 ± 11.2	8.53 ± 0.19	47.9	50.2 ± 1.8	134.8 ± 5.6	8.35 ± 0.05	48.7	−1.22	7.4
AMCMUMAMCLCMAM	UGGUAGU	58.8 ± 2.9	161.5 ± 8.9	8.75 ± 0,19	49.2	61.1 ± 3.8	168.8 ± 12.0	8.80 ± 0.13	49.0	−1.67	7.7
AMCMUMAMCMCLAM	UGGUAGU	58.6 ± 4.5	160.3 ± 13.9	8.89 ± 0.24	50.0	57.6 ± 3.7	157.3 ± 11.7	8.82 ± 0.13	49.9	−1.42	6.8
AMCMUMAMAMCMAM	UGUUAGU	50.1 ± 4.8	144.7 ± 15.9	5.23 ± 0.18	29.2	48.9 ± 3.7	140.9 ± 12.5	5.23 ± 0.17	29.0
						(47.2)	(133.0)	(5.94)	(35.9)
AMCMUMALAMCMAM	UGUUAGU	48.7 ± 5.2	135.6 ± 16.3	6.64 ± 0.15	37.7	42.2 ± 1.4	114.7 ± 4.7	6.65 ± 0.03	37.9	−1.42	8.9
AMCMUMAMGMCMAM	UGCUAGU	75.8 ± 25.1	217.1 ± 81.4	8.47 ± 0.35	45.2	75.2 ± 6.3	216.0 ± 20.1	8.22 ± 0.15	44.1
						(54.3)	(148.5)	(8.27)	(49.9)
AMCMUMALGMCMAM	UGCUAGU	58.4 ± 3.0	158.3 ± 9.3	9.30 ± 0.13	52.4	55.8 ± 1.3	150.4 ± 4.0	9.18 ± 0.05	52.5	−0.96	8.4
AMCMUMAMUMCMAM	UGAUAGU	53.6 ± 3.5	154.1 ± 11.3	5.76 ± 0.07	32.6	52.1 ± 2.9	149.2 ± 9.7	5.82 ± 0.10	32.8
						(50.8)	(143.7)	(6.22)	(37.7)
AMCMUMALUMCMAM	UGAUAGU	50.3 ± 3.4	139.3 ± 10.7	7.03 ± 0.14	40.1	45.5 ± 1.6	124.0 ± 5.2	7.02 ± 0.03	40.4	−1.20	7.6
AMCMUMCMAMCMAM	UGUGAGU	64.7 ± 11.7	185.0 ± 37.6	7.33 ± 0.20	40.9	63.6 ± 6.1	181.7 ± 20.0	7.22 ± 0.17	40.4
						(55.6)	(152.9)	(8.14)	(48.8)
AMCMUMCLAMCMAM	UGUGAGU	58.0 ± 4.6	159.0 ± 14.7	8.70 ± 0.08	49.0	56.3 ± 2.3	153.8 ± 7.4	8.65 ± 0.04	49.1	−1.43	8.7
AMCMUMCMCMCMAM	UGGGAGU	61.5 ± 6.8	166.8 ± 20.6	9.78 ± 0.42	54.3	55.9 ± 1.8	149.5 ± 5.7	9.51 ± 0.07	54.5
						(60.5)	(161.8)	(10.31)	(60.2)
AMCMUMCLCMCMAM	UGGGAGU	69.2 ± 2.7	184.9 ± 8.3	11.88 ± 0.18	63.0	66.4 ± 4.7	176.4 ± 14.1	11.68 ± 0.33	63.1	−2.17	8.6
AMCMUMCMGMCMAM	UGCGAGU	64.2 ± 20.4	176.7 ± 64.2	9.36 ± 0.56	51.3	56.0 ± 2.5	152.5 ± 7.9	9.04 ± 0.7	51.5
						(59.2)	(160.1)	(9.57)	(56.3)
AMCMUMCLGMCMAM	UGCGAGU	71.4 ± 2.5	193.1 ± 7.4	11.53 ± 0.20	60.4	65.0 ± 5.9	173.8 ± 17.9	11.13 ± 0.37	60.6	−2.09	9.1
AMCMUMCMUMCMAM	UGAGAGU	50.8 ± 7.5	138.9 ± 24.1	7.72 ± 0.13	44.5	53.0 ± 1.7	146.1 ± 5.6	7.66 ± 0.02	43.8
						(56.6)	(156.1)	(8.22)	(49.0)
AMCMUMCLUMCMAM	UGAGAGU	64.1 ± 3.3	167.4 ± 10.4	9.23 ± 0.01	51.3	58.1 ± 1.7	158.0 ± 5.3	9.14 ± 0.04	51.6	−1.48	7.8
AMCMUMGMAMCMAM	UGUCAGU	54.2 ± 5.9	150.7 ± 18.5	7.46 ± 0.17	42.4	48.2 ± 0.7	131.8 ± 2.3	7.36 ± 0.00	42.4
						(55.6)	(152.9)	(8.14)	(48.8)
AMCMUMGLAMCMAM	UGUCAGU	57.8 ± 5.3	157.0 ± 16.2	9.08 ± 0.27	51.3	52.6 ± 1.5	141.0 ± 4.8	8.86 ± 0.05	51.4	−1.50	9.0
AMCMUMGMCMCMAM	UGGCAGU	59.1 ± 5.0	160.3 ± 15.2	9.41 ± 0.26	52.9	58.7 ± 2.3	159.2 ± 7.1	9.34 ± 0.09	52.6
						(60.0)	(160.4)	(10.23)	(59.9)
AMCMUMGLCMCMAM	UGGCAGU	60.6 ± 2.2	161.1 ± 6.7	10.66 ± 0.20	59.6	55.5 ± 1.9	145.6 ± 5.9	10.35 ± 0.11	59.9	−1.01	7.3
AMCMUMGMGMCMAM	UGCCAGU	60.5 ± 4.6	163.3 ± 14.0	9.86 ± 0.27	55.0	59.6 ± 1.5	160.7 ± 4.5	9.76 ± 0.06	54.8
						(60.0)	(160.4)	(10.23)	(59.9)
AMCMUMGLGMCMAM	UGCCAGU	62.8 ± 2.3	164.5 ± 6.7	11.76 ± 0.25	65.2	63.7 ± 2.6	167.4 ± 7.7	11.82 ± 0.19	65.1	−2.06	10.3
AMCMUMGMUMCMAM	UGACAGU	57.5 ± 6.3	159.5 ± 19.6	7.98 ± 0.25	45.0	52.5 ± 2.6	144.1 ± 8.2	7.83 ± 0.06	44.9
						(55.6)	(152.9)	(8.14)	(48.8)
AMCMUMGLUMCMAM	UGACAGU	56.0 ± 6.1	150.9 ± 18.5	9.21 ± 0.34	52.6	51.3 ± 2.5	136.4 ± 7.9	8.97 ± 0.10	52.5	−1.14	7.6
AMCMUMUMAMCMAM	UGUAAGU	50.4 ± 5.1	144.6 ± 16.5	5.58 ± 0.10	31.3	46.7 ± 2.1	132.2 ± 7.1	5.65 ± 0.08	31.3
						(47.2)	(133.0)	(5.94)	(35.9)
AMCMUMULAMCMAM	UGUAAGU	50.3 ± 6.2	139.8 ± 19.6	6.94 ± 0.13	39.5	46.8 ± 1.5	128.8 ± 4.7	6.88 ± 0.02	39.3	−1.23	8.0
AMCMUMUMCMCMAM	UGGAAGU	50.2 ± 3.9	138.0 ± 12.2	7.42 ± 0.12	42.6	47.6 ± 1.3	129.6 ± 4.2	7.36 ± 0.01	42.5
						(53.9)	(148.1)	(7.98)	(48.2)
AMCMUMULCMCMAM	UGGAAGU	56.0 ± 3.2	152.2 ± 10.2	8.83 ± 0.14	50.3	50.5 ± 1.1	134.7 ± 3.6	8.68 ± 0.03	50.8	−1.32	8.3
AMCMUMUMGMCMAM	UGCAAGU	61.2 ± 3.3	172.4 ± 10.0	7.71 ± 0.15	43.1	52.7 ± 0.7	145.3 ± 2.3	7.59 ± 0.01	43.4
						(53.4)	(146.7)	(7.90)	(47.8)
AMCMUMULGMCMAM	UGCAAGU	61.0 ± 2.3	167.7 ± 7.4	9.00 ± 0.12	50.1	59.5 ± 1.6	163.1 ± 5.0	8.96 ± 0.05	50.2	−1.37	6.8
AMCMUMUMUMCMAM	UGAAAGU	46.1 ± 4.5	130.5 ± 14.8	5.59 ± 0.12	30.8	39.6 ± 1.0	109.2 ± 3.2	5.76 ± 0.04	31.1
						(47.4)	(134.5)	(5.65)	(34.1)
AMCMUMULUMCMAM	UGAAAGU	46.1 ± 5.3	126.5 ± 16.9	6.88 ± 0.16	39.4	39.9 ± 1.1	106.5 ± 3.4	6.85 ± 0.02	39.5	−1.09	8.4
AMCMAMUMCMCMAM	UGGAUGU	47.9 ± 6.1	132.1 ± 19.7	6.92 ± 0.07	39.5	43.5 ± 0.8	118.2 ± 2.7	6.86 ± 0.01	39.4
						(56.4)	(155.6)	(8.18)	(48.8)
AMCMAMULCMCMAM	UGGAUGU	50.4 ± 5.1	136.9 ± 16.4	7.96 ± 0.12	46.0	46.4 ± 2.1	124.2 ± 6.7	7.85 ± 0.05	46.1	−0.99	6.7
AMCMCMUMCMCMAM	UGGAGGU	59.7 ± 6.2	162.0 ± 19.4	9.49 ± 0.25	53.2	50.9 ± 2.3	134.6 ± 7.0	9.20 ± 0.09	54.1
						(60.5)	(161.8)	(10.31)	(60.2)
AMCMCMULCMCMAM	UGGAGGU	58.9 ± 5.1	156.5 ± 16.3	10.45 ± 0.08	59.1	50.2 ± 3.2	129.7 ± 9.9	10.02 ± 0.18	60.2	−0.82	6.1
AMCMGMUMCMCMAM	UGGACGU	56.9 ± 5.4	154.4 ± 16.8	9.02 ± 0.23	51.2	55.8 ± 2.3	151.1 ± 7.2	8.96 ± 0.08	51.1
						(58.7)	(158.3)	(9.57)	(56.5)
AMCMGMULCMCMAM	UGGACGU	55.4 ± 5.1	149.5 ± 15.6	9.05 ± 0.29	51.8	50.4 ± 0.8	134.2 ± 2.6	8.78 ± 0.04	51.5	+0.18	0.4
AMGMUMAMCMCMAM	UGGUACU	51.4 ± 8.1	141.6 ± 25.5	7.44 ± 0.20	42.6	44.5 ± 0.7	119.9 ± 2.3	7.33 ± 0.01	42.7
						(53.8)	(146.6)	(8.27)	(50.0)
AMGLUMAMCMCMAM	UGGUACU	52.6 ± 6.5	141.9 ± 20.2	8.55 ± 0.25	49.4	51.7 ± 1.8	139.3 ± 5.7	8.47 ± 0.05	49.1	−1.14	6.4
UMCMGMGMCMAM	UGCCGA	65.6 ± 18.0	185.2 ± 56.9	8.18 ± 0.46	45.0	65.5 ± 5.7	185.1 ± 18.2	8.06 ± 0.16	44.4
						(50.7)	(136.2)	(8.51)	(52.4)
UMCMGMGMCLAM	UGCCGA	60.9 ± 5.3	164.2 ± 15.9	10.00 ± 0.39	55.7	57.5 ± 7.2	153.8 ± 22.0	9.80 ± 0.43	55.7	−1.74	11.3

aSolutions are 100 mM NaCl, 20 mM sodium cacodylate and 0.5 mM Na2EDTA, pH 7. Values in parentheses are thermodynamics predicted for RNA/RNA duplexes in 1 M NaCl (64).

bCalculated for 10−4 M oligomer concentration. The ΔTm is the difference in Tm due to the LNA substitution.

cThe is the difference in due to the LNA substitution.

Single LNA substitutions at the 5′ end of heptamer duplexes have little effect on stability

The effects of single LNA substitutions at the 5′ end of the 2′-O-methyl strand were studied in duplexes of the form, 5′NCMUMAMCMCMAM/3′r(QGAUGGU), where superscript M denotes a 2′-O-methyl sugar, N is A, C, G, or U with a 2′-O-methyl or LNA sugar, r denotes ribose sugars, and Q is the Watson–Crick complement to N. As summarized in Table 1, 5′ terminal LNA substitutions make duplex stability more favorable by 0.3–0.6 kcal/mol at 37°C with an average enhancement of 0.45 kcal/mol. Thus, 5′ terminal LNA substitutions increase the binding constant for duplex formation by ∼2-fold at 37°C.

The effects of single LNA substitutions at the 3′ ends of heptamer duplexes is idiosyncratic

The effects of single LNA substitutions at the 3′ end of the 2′-O-methyl strand was studied in duplexes of the form, 5′AMCMUMAMCMCMN/3′r(UGAUGGQ) and in 5′AMCMUMAMCMGMUM/L/3′r(UGAUGCA) (Table 1). If N is A, C or G, then LNA substitutions have similar effects. On average, an LNA substitution makes duplex stability more favorable by 1.2 kcal/mol at 37°C. In the two sequences with a 3′ terminal LNA U on the 2′-O-methyl strand, duplex stability is, however, affected little, averaging a destabilization of 0.08 kcal/mol at 37°C. In both cases, the terminal U is preceded by a GC pair, but both orientations of the GC pair give similar destabilization upon LNA substitution at the 3′ terminal U.

Single LNA substitutions in the interior of AMCMUMAMCMCMAM enhance the stability of the duplex formed with its complementary RNA by ∼1.4 kcal/mol

The effect of interior position on the free energy increment for a single LNA substitution for a 2′-O-methyl RNA was studied for the duplex 5′AMCMUMAMCMCMAM/3′r(UGAUGGU). As summarized in Table 1, a single interior LNA substitution makes duplex stability more favorable by 1.2–1.7 kcal/mol at 37°C, with an average of 1.4 kcal/mol. This corresponds to roughly a 10-fold increase in binding constant. Thus, interior and 3′ terminal LNA substitutions usually improve binding more than 5′ terminal LNA substitutions.

The effects of single LNA substitutions as a function of sequence context in the middle of heptamer duplexes

The effects from single LNA substitutions in the middle of heptamer duplexes were studied as a function of the LNA base and the 3′ adjacent 2′-O-methyl nucleotide in duplexes of the form, 5′AMCMUMNVMCMAM/3′r(UGAQWGU), where VM is 2′-O-methyl A, C, G or U, and W is the ribose Watson–Crick complement to V. The effects on the for duplex formation of substituting LNA for 2′-O-methyl at position N are summarized in Table 1. For 13 of 16 sequences, the LNA substitution makes duplex stability more favorable by 1.0–1.5 kcal/mol at 37°C, with an average enhancement of 1.3 kcal/mol. The enhancement for the other three sequences averages 2.1 kcal/mol at 37°C.

The dependence on the 5′ nearest-neighbor nucleotide of effects from substituting UL for UM was studied in duplexes of the form, 5′AMCMVCMCMAM/3′r(UGWAGGU) (Table 1). For three of the four sequences, the range for enhancement of duplex stability is 0.8–1.3 kcal/mol and the average is 1.0 kcal/mol. The exception is for the nearest neighbor 5′GMUL/3′r(CA), where LNA substitution destabilizes the duplex by 0.2 kcal/mol.

The results for single LNA substitutions in other contexts provide additional insight into the dependence of free energy increments on the 5′ nearest neighbor 2′-O-methyl nucleotide. In particular, the Watson–Crick complementary duplexes with 5′AMCAMCMCMAM and 5′AMCMUMCCMAM strands have a 5′CLUM/3′r(GA) nearest neighbor that is preceded by AM and UM, respectively. The LNA substitution for 2′-O-methyl in these cases results in duplex formation being more stable by 1.29 and 1.48 kcal/mol, respectively, at 37°C. Similarly, the duplexes with 5′AMGAMCMCMAM and 5′AMCMUMGCMAM each have a 5′GLUM/3′r(CA) nearest neighbor that is preceded by AM and UM, respectively. In both cases, the LNA substitution enhances duplex stability by 1.14 kcal/mol at 37°C. Thus, for seven duplexes, the enhanced stability from an LNA substitution is relatively independent of the nearest-neighbor nucleotide 5′ to the LNA. The one exception is for the nearest neighbor 5′GMUL/3′r(CA). Interestingly, this nearest-neighbor combination is also destabilized by LNA substitution at a 3′ terminal U (Table 1). Evidently, an LNA substitution in the middle of a 2′-O-methyl strand usually affects heteroduplex stability with an RNA strand by about the same amount as an LNA substitution at a 3′ terminus.

The effects of LNA substitutions are approximately additive when LNA nucleotides are spaced by at least one 2′-O-methyl nucleotide

Table 2 contains thermodynamic parameters measured for duplexes having more than one LNA substitution and Table 3 compares the stabilities at 37°C with those predicted from four simple models. The first model, labeled ‘additivity’, predicts the for duplex formation in the 5′ACUACCA/3′UGAUGGU series by adding the free energy increments measured for single LNA substitutions in the same context to the for duplex formation in the absence of LNA nucleotides. The second model predicts the (kcal/mol) for duplex formation with the following equation as deduced from fitting the data in Tables 1 and 2 excluding sequences with adjacent LNAs. Here, (2′-O-MeRNA/RNA) is the free energy change at 37°C for duplex formation in the absence of any LNA nucleotides, n5′tL is the number of 5′ terminal LNAs, niAL/UL and niGL/CL are the number of internal LNAs in AU and GC pairs, respectively, n3′tU and n3′tAL/CL/GL are the number of 3′ terminal LNAs that are U or not U, respectively. Both methods that use experimental data for (2′-O-MeRNA/RNA) provide reasonable predictions that are within 1 kcal/mol of the measured value (Table 3). Two other methods that use nearest-neighbor models to approximate (2′-O-MeRNA/RNA) provide somewhat less accurate, but still reasonable predictions as described below. The duplex with the worst prediction, 5′GMULUMCLGMGL/3′CAAGCC has a 5′GMUL/3′CA nearest neighbor, consistent with this motif being unusually unstable by ∼1.2 kcal/mol. Thus, it is likely that the of Equation 1 should be made less favorable by 1.2 kcal/mol for every internal 5′GMUL/3′CA nearest neighbor in a duplex. Evidently, the effects of multiple LNA substitutions are approximately additive when the LNAs are spaced by at least 1 nt.

Table 2

Thermodynamic parameters of duplex formation between RNA and 2′-O-methyl oligoribonucleotides with and without multiple LNA substitutionsa

Oligonucleotides (5′ to 3′)	RNA (5′ to 3′)	Average of curve fits				Tm−1 versus log (CT/4) plots
		−ΔH° (kcal/mol)	−ΔS° (eu)	−ΔG37° (kcal/mol)	Tmb (°C)	−ΔH° (kcal/mol)	−ΔS° (eu)	−ΔG37° (kcal/mol)	Tmb (°C)
ALCMULAMCLCMAL	UGGUAGU	60.7 ± 7.0	157.5 ± 21.0	11.85 ± 0.53	66.8	59.5 ± 4.9	154.0 ± 14.7	11.68 ± 0.40	66.4
AMCLUMALCMCLAM	UGGUAGU	66.3 ± 3.9	176.5 ± 11.7	11.55 ± 0.31	62.4	65.0 ± 2.5	172.7 ± 7.4	11.45 ± 0.16	62.4
AMCLUMAMCLCMAM	UGGUAGU	65.5 ± 7.6	179.3 ± 23.2	9.89 ± 0.35	53.8	77.5 ± 2.7	216.5 ± 8.6	10.36 ± 0.10	53.0
ALCMUMAMCLCMAM	UGGUAGU	60.7 ± 7.8	165.2 ± 23.7	9.42 ± 0.54	52.5	53.0 ± 1.6	141.7 ± 4.9	9.06 ± 0.06	52.5
AMCMULAMCMCLAM	UGGUAGU	55.4 ± 1.6	145.5 ± 5.0	10.24 ± 0.09	59.3	53.1 ± 1.3	138.7 ± 3.9	10.12 ± 0.07	59.5
ALCMUMALCMCMAL	UGGUAGU	56.7 ± 7.2	150.0 ± 21.8	10.12 ± 0.45	58.0	53.6 ± 2.5	141.1 ± 7.7	9.87 ± 0.13	57.6
ALCLULALCLCLAL	UGGUAGU	57.5 ± 3.2	146.8 ± 9.3	11.98 ± 0.35	69.5	51.0 ± 1.8	127.7 ± 5.4	11.40 ± 0.16	69.8
GMCMUMAMCMUMGM	CAGUAGC	63.9 ± 2.6	178.6 ± 8.2	8.49 ± 0.09	46.8	66.2 ± 2.6	186.0 ± 8.2	8.51 ± 0.06	46.6
						(61.8)	(169.6)	(9.17)	(53.2)
GMCLUMALCMULGM	CAGUAGC	81.3 ± 4.0	221.8 ± 12.2	12.52 ± 0.27	61.6	74.6 ± 4.9	201.6 ± 14.7	12.03 ± 0.34	61.7
GMCLUMAMCMULGM	CAGUAGC	67.3 ± 7.8	182.7 ± 23.8	10.59 ± 0.42	56.9	64.6 ± 5.5	174.7 ± 16.9	10.44 ± 0.28	57.0
GLCLULALCLULGL	CAGUAGC	66.3 ± 5.7	171.7 ± 16.5	13.10 ± 0.58	71.1	58.6 ± 2.5	149.2 ± 7.3	12.37 ± 0.22	71.3
GMCMAMUMGMGM	CCAUGC	60.6 ± 1.1	172.4 ± 3.4	7.13 ± 0.16	40.1	51.9 ± 5.1	144.1 ± 16.6	7.19 ± 0.14	41.0
						(54.9)	(151.6)	(7.91)	(47.5)
GMCLAMULGMGM	CCAUGC	57.1 ± 2.8	151.4 ± 8.7	10.13 ± 0.18	57.8	61.1 ± 5.4	163.7 ± 16.5	10.30 ± 0.31	57.4
GMGMCMAMUMGM	CAUGCC	60.7 ± 9.7	172.3 ± 31.5	7.29 ± 0.18	40.9	65.2 ± 4.1	187.1 ± 13.1	7.19 ± 0.07	40.2
						(54.9)	(151.6)	(7.91)	(47.5)
GMGMCLAMULGM	CAUGCC	66.5 ± 4.9	180.9 ± 14.6	10.39 ± 0.33	56.1	68.8 ± 4.0	187.8 ± 12.4	10.49 ± 0.20	55.9
CMGMGMCMAMUM	AUGCCG	45.6 ± 2.6	124.4 ± 8.3	6.97 ± 0.08	40.0	45.3 ± 1.9	123.5 ± 6.0	6.98 ± 0.02	40.1
						(51.4)	(140.8)	(7.71)	(47.0)
CMGLGMCLAMUM	AUGCCG	47.6 ± 6.1	121.0 ± 18.3	10.10 ± 0.41	62.1	50.6 ± 5.5	130.2 ± 16.8	10.22 ± 0.35	61.4
UMUMCMGMGMCM	GCCGAA	59.3 ± 6.6	166.1 ± 20.6	7.83 ± 0.23	43.9	53.5 ± 8.0	147.3 ± 25.3	7.76 ± 0.33	44.3
						(50.8)	(138.8)	(7.78)	(47.6)
ULUMCLGMGMCM	GCCGAA	58.8 ± 8.4	155.6 ± 25.2	10.51 ± 0.59	59.5	55.7 ± 1.6	146.5 ± 4.9	10.26 ± 0.07	59.2
GMUMUMCMGMGM	CCGAAC	55.2 ± 4.9	155.5 ± 16.3	6.96 ± 0.17	39.4	54.6 ± 3.8	153.8 ± 12.5	6.95 ± 0.10	39.4
						(51.1)	(142.0)	(7.05)	(42.9)
GMULUMCLGMGL	CCGAAC	65.3 ± 6.1	170.6 ± 18.0	12.38 ± 0.50	67.5	61.7 ± 2.3	160.2 ± 6.7	12.00 ± 0.19	67.4
CMGMGMCMAM	UGCCG	47.1 ± 8.8	131.6 ± 29.0	6.25 ± 0.19	35.1	46.0 ± 3.0	128.4 ± 9.8	6.19 ± 0.09	34.7
						(42.0)	(114.2)	(6.61)	(40.8)
CMGLGMCLAM	UGCCG	46.5 ± 5.3	118.8 ± 16.0	9.61 ± 0.36	59.0	43.1 ± 5.3	108.8 ± 16.0	9.36 ± 0.42	58.8
UMCMGMGMCM	GCCGA	50.9 ± 4.6	143.5 ± 14.9	6.39 ± 0.14	36.1	51.5 ± 7.9	145.4 ± 26.0	6.38 ± 0.35	36.1
						(44.0)	(119.9)	(6.85)	(42.2)
UMCLGMGLCM	GCCGA	53.1 ± 5.3	139.4 ± 15.6	9.92 ± 0.47	58.1	49.5 ± 5.1	128.5 ± 15.2	9.64 ± 0.36	57.8

aSolutions are 100 mM NaCl, 20 mM sodium cacodylate and 0.5 mM Na2EDTA, pH 7. Values in parentheses are thermodynamics predicted for RNA/RNA duplexes in 1 M NaCl (64).

bCalculated for 10−4 M oligomer concentration.

Table 3

Comparison of measured and predicted stabilities

Sequence	−ΔG37° (kcal/mol)		ΔG37° (measured) − ΔG37° (predicted) (kcal/mol)
	Measured	Additivity	Equation 1	Equations 1 + 2 + 3	Equations 1 + 4
5′ALCMULAMCLCMAL/3′UGAUGGU	11.68	0.54	−0.20	0.22	−0.05
5′AMCLUMALCMCLAM/3′UGAUGGU	11.45	−0.39	−0.03	0.40	0.13
5′AMCLUMAMCLCMAM/3′UGAUGGU	10.36	−0.27	−0.04	0.39	0.12
5′ALCMUMAMCLCMAM/3′UGAUGGU	9.06	0.33	0.12	−0.54	0.27
5′AMCMULAMCMCLAM/3′UGAUGGU	10.12	−0.04	−0.30	0.13	−0.14
5′ALCMUMALCMCMAL/3′UGAUGGU	9.87	0.37	0.01	0.43	0.16
5′GMCLUMALCMULGM/3′CGAUGAC	12.03	–	0.27	0.22	0.48
5′GMCLUMAMCMULGM/3′CGAUGAC	10.44	–	0.76	0.71	0.97
5′GMCLAMULGMGM/3′CGUACC	10.30	–	−0.42	−0.29	−0.56
5′GMGMCLAMULGM/3′CCGUAC	10.49	–	−0.61	−0.48	−0.75
5′CMGLGMCLAMUM/3′GCCGUA	10.22	–	−0.48	0.09	0.26
5′ULUMCLGMGMCM/3′AAGCCG	10.26	–	−0.45	−1.03	−0.66
5′GMULUMCLGMGL/3′CAAGCC	12.00	–	−1.15	−1.65	−1.68
5′CMGLGMCLAM/3′GCCGU	9.36	–	0.02	−0.03	0.39
5′UMCLGMGLCM/3′AGCCG	9.64	–	−0.07	−0.07	0.13

The data may also be fit to a nearest-neighbor model containing 30 of the LNA enhancement parameters associated with duplexes of RNA strands bound to 2′-O-methyl RNA/LNA chimeras. These parameters are listed in Supplementary Material. The number of occurrences for each nearest neighbor is limited, however, so the values are only roughly determined.

Predictions for RNA/RNA duplexes at 1 M NaCl can be used to approximate stabilities of 2′-O-methyl RNA/RNA duplexes at 0.1 M NaCl

The stabilities of RNA/RNA duplexes at 37°C and 1 M NaCl are predicted well by an Independent Nearest-Neighbor Hydrogen Bonding (INN-HB) model (64). In this model, the stability of an RNA/RNA duplex is approximated by: Here, is the free energy change for initiating a helix; each is the free energy increment of the jth type nearest neighbor (see Table 4) with n occurrences in the sequence; mterm-AU is the number of terminal AU pairs; is the free energy increment per terminal AU pair; ΔGsym is 0.43 kcal/mol at 37°C for self-complementary duplexes and 0 for non-self-complementary duplexes.

Table 4

Preliminary nearest-neighbor free energy parameters (kcal/mol at 37°C) for 2′-O-methyl RNA/RNA duplexes in 0.1 M NaCl compared with parameters for RNA/RNA duplexes in 1 M NaCl

2′-O-MeRNA/RNA parameters	ΔG37° (kcal/mol)	RNA/RNA parametersa	ΔG37° (kcal/mol)a
5′AMAM3′	(−0.79 ± 0.44)b	5′AA3′	−0.93 ± 0.03
3′U U 5′		3′UU5′
5′UMUM3′	−0.99 ± 0.20
3′A A 5′
5′AMUM3′	−0.73 ± 0.26	5′AU3′	−1.10 ± 0.08
3′U A 5′		3′UA5′
5′UMAM3′	−1.28 ± 0.29	5′UA3′	−1.33 ± 0.09
3′A U 5′		3′AU5′
5′CMUM3′	−2.18 ± 0.26	5′CU3′	−2.08 ± 0.06
3′G A 5′		3′GA5′
5′AMGM3′	(−1.53 ± 0.41)b
3′U C 5′
5′CMAM3′	−1.96 ± 0.26	5′CA3′	−2.11 ± 0.07
3′G U 5′		3′GU5′
5′UMGM3′	−1.58 ± 0.36
3′A C 5′
5′GMUM3′	−2.65 ± 0.35	5′GU3′	−2.24 ± 0.06
3′C A 5′		3′CA5′
5′AMCM3′	−1.60 ± 0.23
3′U G 5′
5′GMAM3′	(−2.62 ± 0.52)b	5′GA3′	−2.35 ± 0.06
3′C U 5′		3′CU5′
5′UMCM3′	−1.97 ± 0.24
3′A G 5′
5′CMGM3′	−2.11 ± 0.39	5′CG3′	−2.36 ± 0.09
3′G C 5′		3′GC5′
5′GMGM3′	−2.88 ± 0.27	5′GG3′	−3.26 ± 0.07
3′C C 5′		3′CC5′
5′CMCM3′	−2.85 ± 0.19
3′G G 5′
5′GMCM3′	−3.49 ± 0.36	5′GC3′	−3.42 ± 0.08
3′C G 5′		3′CG5′
Initiation	3.59 ± 0.95		4.09 ± 0.22
Per terminal AU	0.29 ± 0.15		0.45 ± 0.04
Symmetry correction (‘self-complementary’)	0		0.43
Symmetry correction (non-self-complementary)	0		0

aXia et al. (64).

bThere are only one or two occurrences of these nearest-neighbor sequences in the database.

Similar sequence dependent parameters may also be applicable to 2′-O-methyl RNA/RNA heteroduplexes because they are expected to have A-form conformations similar to those of RNA/RNA homoduplexes (68). This was tested by comparing the predicted stabilities of RNA/RNA duplexes in 1 M NaCl at 37°C with those measured for 2′-O-methyl RNA/RNA duplexes in 0.1 M NaCl at 37°C. The predicted thermodynamics are listed in parentheses in Tables 1 and 2. On average at 37°C, the RNA/RNA duplexes in 1 M NaCl are 0.12 ± 0.01 kcal/mol of phosphate pairs more stable than the 2′-O-methyl RNA/RNA duplexes in 0.1 M NaCl. Presumably, much of this difference is due to a sequence independent effect of salt concentration, which would primarily affect the ΔS° for duplex formation (22,69). Thus, a reasonable approximation for the first term on the right hand side of Equation 1 is: Note that ΔGsym from the RNA/RNA calculation is subtracted because a 2′-O-methyl RNA/RNA duplex cannot be self-complementary because the backbones differ. For the duplexes studied here, the number of phosphate pairs is one less than the number of base pairs.

The effects of LNA substitutions are likely not very dependent on salt concentration. Thus, it is probable that in 1 M NaCl or in the presence of Mg2+ (70) that (2′-O-MeRNA/RNA) can be approximated by (RNA/RNA, 1 M NaCl).

Table 3 compares measured values for duplexes with more than one LNA to predictions from combining Equation 1–3. The measured values average −10.5 kcal/mol and the root-mean-square difference between measured and predicted values is 0.6 kcal/mol with the largest difference being 1.7 kcal/mol. Again, the sequence with the largest difference contains a 5′GMUL/3′CA nearest neighbor so the prediction would be improved if Equation 1 was corrected for the apparent instability of this motif.

The results for 2′-O-methyl RNA/RNA duplexes provide preliminary nearest-neighbor free energy increments for predicting stabilities of such duplexes

The comparison of predicted RNA/RNA stabilities with those measured for 2′-O-methyl RNA/RNA duplexes suggests that the INN-HB model will also be applicable to 2′-O-methyl RNA/RNA duplexes (71). The results in Tables 1 and 2 were combined with those for several other duplexes (E. Kierzek, R. Kierzek, and D.H. Turner, unpublished data) to give preliminary INN-HB parameters for 2′-O-methyl RNA/RNA duplexes in 0.1 M NaCl (see Table 4). Three nearest neighbors are only represented once or twice in the database, and these parameters are in parentheses. The parameters for 2′-O-methyl RNA/RNA and RNA/RNA duplexes are similar, especially if the RNA/RNA Watson–Crick nearest-neighbor parameters are each made less favorable by 0.12 kcal/mol, which largely accounts for the difference in salt concentration as suggested above. Evidently, the first term on the right hand side of Equation 1 can also be approximated by: Table 3 compares predictions from combining Equations 1 and 4 with measured values for duplexes with more than one LNA. The root-mean-square difference between measured and predicted values is 0.6 kcal/mol with the largest difference being the 1.7 kcal/mol associated with the duplex containing a 5′GMUL/3′CA nearest neighbor. Undoubtedly, this model can be expanded and refined by more measurements, but it appears sufficient to aid sequence design for many applications.

Complete LNA substitution is no more stabilizing than substitution at every other nucleotide starting at the second nucleotide from the 5′ end

The effect of complete LNA substitution for a 2′-O-methyl RNA backbone was studied for the sequences 5′ALCLULALCLCLAL/3′r(UGAUGGU) and 5′GLCLULALCLULGL/3′r(CGAUGAC). As summarized in Table 2, the stabilities of these duplexes at 37°C are within experimental error of those measured for 5′AMCLUMALCMCLAM/3′r(UGAUGGU) and 5′GMCLUMALCMULGM/3′r(CGAUGAC), respectively. Evidently, the most effective use of LNA nucleotides is to space them every other nucleotide with the first LNA placed at the second nucleotide from the 5′ end.

Internal mismatches make duplex formation less favorable

Table 5 contains thermodynamic parameters measured for the formation of duplexes containing single mismatches and the difference in stabilities relative to completely Watson–Crick complementary duplexes (Tables 1 and 2). All internal mismatches make duplex formation less favorable by at least 2 kcal/mol at 37°C corresponding to at least a 25-fold less favorable equilibrium constant for duplex formation. In general, terminal mismatches destabilize much less than internal mismatches. In fact, when the 3′ terminal UL of 5′AMCMUMAMCMCMUL makes a GU pair, the duplex is stabilized by 0.14 kcal/mol at 37°C relative to a terminal AU pair.

Table 5

Thermodynamic parameters of duplex formation between RNA and 2′-O-methyl oligoribonucleotides with and without LNA substitutions: effects of mismatchesa

Mismatch	Oligonucleotides (5′ to 3′)	RNA (5′ to 3′)	Average of curve fits				Tm−1 versus log (CT/4) plots
			−ΔH° (kcal/mol)	−ΔS° (eu)	−ΔG37° (kcal/mol)	Tmb (°C)	−ΔH° (kcal/mol)	−ΔS° (eu)	−ΔG37° (kcal/mol)	Tmb (°C)	ΔΔG37°c (kcal/mol)	ΔTmb (°C)
AL-C	AMCMUMALCMCMAM	UGGCAGU	48.7 ± 4.5	140.9 ± 14.8	5.00 ± 0.17	27.5	44.7 ± 6.3	127.4 ± 20.9	5.15 ± 0.30	27.7	3.20	−21.0
AL-C	ALCMUMALCMCMAL	UGGCAGU	51.4 ± 8.6	144.8 ± 27.7	6.44 ± 0.18	36.4	47.8 ± 8.7	133.2 ± 28.1	6.47 ± 0.43	36.6	3.40	−21.0
AL-G	AMCMUMALGMCMAM	UGCGAGU	46.6 ± 20.5	127.5 ± 65.2	7.07 ± 0.32	40.6	40.2 ± 6.3	107.2 ± 20.6	6.93 ± 0.31	40.1	2.25	−12.4
AM-G	AMCMUMAMGMCMUM	UGCGAGU	48.9 ± 4.9	144.6 ± 16.4	4.08 ± 0.27	22.2	45.3 ± 2.7	132.2 ± 9.2	4.32 ± 0.16	22.6	4.86	−29.9
AM-G	ALCMULAMCLCMAL	UGGGAGU	49.4 ± 3.8	135.4 ± 11.9	7.39 ± 0.16	42.5	40.8 ± 1.9	108.0 ± 6.2	7.29 ± 0.03	42.9	4.39	−23.5
AM-G	AMCMUMAMCMCMAM	UGGGAGU	42.4 ± 4.3	122.9 ± 14.4	4.26 ± 0.25	21.2	49.3 ± 5.4	146.6 ± 18.5	3.81 ± 0.37	20.8	3.32	−20.5
CL-A	AMCMUMCLCMCMAM	UGGAAGU	39.3 ± 6.2	110.0 ± 20.5	5.17 ± 0.22	26.6	42.5 ± 5.9	120.8 ± 19.7	5.01 ± 0.36	26.3	6.67	−36.8
CL-A	AMCMUMAMCMCMCL	AGGUAGU	52.2 ± 6.3	147.2 ± 19.9	6.50 ± 0.36	36.8	44.5 ± 16.3	122.1 ± 52.6	6.55 ± 1.71	37.8	2.69	−11.0
CL-U	AMCMUMCLGMCMAM	UGCUAGU	59.1 ± 8.9	177.2 ± 30.0	4.18 ± 0.37	25.1	62.2 ± 3.4	187.7 ± 11.5	4.00 ± 0.19	24.9	7.13	−35.7
CM-U	AMCMUMCMGMCMAM	UGCUAGU	62.4 ± 5.7	188.7 ± 19.4	3.90 ± 0.29	24.4	69.2 ± 2.9	211.7 ± 9.8	3.55 ± 0.16	24.2	7.58	−36.4
CL-U	AMCMUMCLUMCMAM	UGCUAGU	62.3 ± 10.6	187.0 ± 35.3	4.29 ± 0.46	26.2	62.9 ± 5.9	188.9 ± 19.7	4.32 ± 0.31	26.5	4.82	−25.1
GL-A	AMCMUMGLCMCMAM	UGGAAGU	35.6 ± 7.6	98.7 ± 26.5	5.02 ± 0.63	24.4	37.5 ± 2.3	105.6 ± 7.9	4.73 ± 0.16	22.8	5.62	−37.1
GL-A	GLCMUMAMCMCMAM	UGGUAGA	46.1 ± 5.1	127.3 ± 15.8	6.59 ± 0.16	37.4	43.8 ± 2.9	120.0 ± 9.5	6.59 ± 0.09	37.4	2.52	−15.6
GL-U	AMCMUMGLCMCMAM	UGGUAGU	52.5 ± 4.3	145.8 ± 13.5	7.32 ± 0.12	41.7	49.7 ± 1.7	136.8 ± 5.4	7.27 ± 0.02	41.7	3.23	−18.4
GM-U	AMCMUMGMCMCMAM	UGGUAGU	53.6 ± 8.2	152.4 ± 26.6	6.36 ± 0.09	36.0	53.5 ± 1.4	152.3 ± 4.5	6.28 ± 0.03	35.6	4.22	−24.5
GL-U	AMCMUMGLUMCMAM	UGAUAGU	51.3 ± 3.6	146.4 ± 11.8	5.85 ± 0.10	32.9	47.7 ± 1.9	134.7 ± 6.3	5.93 ± 0.06	33.1	3.04	−19.4
GL-U	GLCMUMAMCMCMAM	UGGUAGU	56.1 ± 6.4	156.8 ± 20.5	7.53 ± 0.26	42.6	48.0 ± 2.3	130.5 ± 7.3	7.48 ± 0.04	43.2	1.63	−9.8
UL-C	AMCMUMULAMCMAM	UGUCAGU	47.1 ± 10.4	141.0 ± 36.1	3.33 ± 0.77	17.2	53.0 ± 13.6	161.9 ± 46.4	2.84 ± 1.13	16.8	4.04	−22.5
UL-C	AMCMUMULUMCMAM	UGACAGU	43.2 ± 11.6	125.9 ± 39.9	4.17 ± 0.88	20.9	43.2 ± 10.8	126.4 ± 37.0	4.01 ± 0.97	19.9	2.84	−19.6
UL-G	AMCMUMULGMCMAM	UGCGAGU	60.8 ± 11.0	174.2 ± 35.5	6.78 ± 0.19	38.3	55.9 ± 4.7	158.3 ± 15.5	6.76 ± 0.13	38.3	2.20	−11.9
UM-G	AMCMUMUMGMCMAM	UGCGAGU	65.4 ± 7.9	195.8 ± 26.3	4.70 ± 0.21	28.5	61.6 ± 7.2	182.9 ± 24.1	4.84 ± 0.35	28.7	4.12	−21.5
UL-G	AMCMUMAMCMCMUL	GGGUAGU	57.6 ± 11.2	161.3 ± 35.9	7.52 ± 0.29	42.4	58.4 ± 9.0	164.6 ± 29.0	7.38 ± 0.38	41.6	−0.14	0.6
UL-U	AMCMUMULGMCMAM	UGCUAGU	59.1 ± 3.8	175.7 ± 12.4	4.60 ± 0.14	27.2	71.5 ± 4.8	217.3 ± 16.1	4.07 ± 0.22	26.7	4.89	−23.5

aSolutions are 100 mM NaCl, 20 mM sodium cacodylate and 0.5 mM Na2EDTA, pH 7.

bCalculated for 10−4 M oligomer concentration. The ΔTm is the difference in Tm between the mismatched and fully Watzon-Crick complementary duplexes.

cThe is the diffence in between the mismatched and fully between the mismatched and fully Watson-Crick complementary duplexes.

For four cases, the effect of a mismatch with an LNA nucleotide was compared with that for the equivalent 2′-O-methyl nucleotide. In each case, the mismatch penalty for the LNA was less than that for 2′-O-methyl RNA. However, for an AM-G mismatch flanked by LNAs in the context 5′ALCMULCLCMAL/3′r(UGAGGGU), the LNAs enhanced the mismatch penalty by ∼1 kcal/mol relative to a completely 2′-O-methyl RNA strand. Thus, oligonucleotides containing LNA may discriminate best against mismatches flanked by LNAs.

DISCUSSION

Oligonucleotide hybridization to RNA has many applications, ranging from quantifying gene expression (18–20,56) to designing therapeutics (4,8,21,46,52–54). LNA nucleotides have characteristics useful for these purposes. For example, LNA usually stabilizes duplexes (4,44,48,51) and is more resistant than RNA and DNA to nuclease digestion (4,6,51). The results presented here provide insights that are useful for designing 2′-O-methyl RNA/LNA chimeric oligonucleotides for various purposes. Some trends may be general for RNA A-form helixes and thus may also be relevant to other chimeras with nucleotides that favor A-form conformations.

The results suggest several principles for the design of 2′-O-methyl RNA/LNA chimeras for hybridization to RNA

The database in Tables 1 and 2 is too small to generate a complete model for the design of 2′-O-methyl RNA/LNA chimeras. Nevertheless, several trends are apparent: (i) LNA substitution at the 5′ end has little effect on duplex stability. (ii) Except for a 3′ terminal UL, interior and 3′ terminal LNA substitutions have similar effects. (iii) Most of the stability enhancement is achieved when LNA nucleotides are separated by one 2′-O-methyl nucleotide. (iv) The effects of LNA substitutions are approximately additive when the LNA nucleotides are separated by at least one 2′-O-methyl nucleotide. Thus, in most cases, LNA nucleotides are used most effectively for duplex stabilization when separated by at least one 2′-O-methyl nucleotide and not placed on the 5′ end of the chimera. The additivity observed suggests that a nearest-neighbor model (64,72–75) will be able to predict well the stabilities of duplexes formed between RNA and 2′-O-methyl RNA/LNA chimeras. While the database is not sufficient to determine all the parameters for such a model, the simpler model of Equation 1 provides reasonable approximations for stabilities at 37°C, though a correction should probably be applied for duplexes containing a 5′GMUL/3′CA nearest neighbor. It also appears that approximations for the stabilities of 2′-O-methyl RNA/RNA duplexes can be provided by nearest-neighbor models for 2′-O-methyl RNA/RNA duplexes (Equation 4) or even for RNA/RNA duplexes (Equations 2 and 3).

The magnitude and sequence dependence of the stabilization due to LNAs are surprising. Ribose and therefore probably 2′-O-methyl ribose sugars in single strands are typically found in roughly equal fractions in C2′-endo and C3′-endo conformations. If the methylene bridge of an LNA only locks the sugar into the C3′-endo conformation, then the expected stabilization due to preorganization would be: ΔΔG° = −RT ln 2, which is −0.4 kcal/mol at 37°C (310.15 K). The stabilization observed for a 5′ terminal LNA is roughly −0.4 kcal/mol, but the average stabilizations for internal LNAs and 3′ terminal AL, CL and GL are more favorable at −1.3 and −1.2 kcal/mol, respectively. Moreover, if stabilization was only due to preorganization of an LNA sugar, then the effect would not saturate when alternate sugars are LNA. Evidently, the LNA substitution also affects the 5′ neighboring base pair in a way that enhances the stabilization beyond that expected from preorganization of a single sugar. Interestingly, NMR structures of DNA/LNA chimeras bound to RNA show that only the DNA sugar 3′ of the LNA is driven to a C3′-endo conformation for the sequence d(5′CTGATLATGC)/3′GACUAUACG, but all non-terminal DNA sugars are C3′-endo when all three Ts are LNAs (76).

Comparison with single LNA substitutions in DNA/DNA duplexes

McTigue et al. (48) measured the thermodynamic effects of single internal LNA substitutions in 100 DNA/DNA duplexes. The free energy increments at 37°C for LNA substitutions ranged from +0.83 to −1.90 kcal/mol with an average of −0.55 kcal/mol. This compares with a range from +0.18 to −2.17 kcal/mol and an average of −1.32 kcal/mol for the single internal LNA substitutions in Table 1. The comparision suggests that single LNA substitutions are on average more stabilizing to 2′-O-methyl RNA/RNA duplexes than to DNA/DNA duplexes. This may reflect the expectation that LNA substitutions do not have a large effect on the conformations of 2′-O-methyl RNA/RNA duplexes, but alter the conformations of DNA/DNA duplexes.

LNA substitutions should be useful for probing RNA with short 2′-O-methyl RNA oligonucleotides

RNA structure can be probed with short oligonucleotides on microarrays (3). To optimize such methods, it is necessary to have tight binding that is sequence independent and that discriminates against mismatches. It appears that LNA nucleotides can be used to achieve this. For example, free energy increments for 2′-O-methyl RNA/RNA nearest neighbors range from −0.7 to −3.5 kcal/mol, corresponding to 5′AMUM/3′UA and 5′GMCM/3′CG, respectively (Table 4). The average increment of −1.3 kcal/mol of internal and 3′ terminal LNA can help compensate for such less favorable stability of AU relative to GC pairs. The stability enhancement from LNA can also allow the use of shorter oligonucleotides.

The potential disadvantage to LNA substitutions in 2′-O-methyl RNA oligonucleotides is that discrimination against mismatches containing an LNA may be less than with a complete 2′-O-methyl RNA backbone. This was clearly true for three of the four cases where such direct comparisons were made. Nevertheless, internal mismatches with LNA nucleotides are considerably destabilizing, averaging a penalty of 4.1 kcal/mol at 37°C (Table 5), which translates to almost a 1000-fold weaker binding due to a single mismatch. When LNAs flanked an AM-G mismatch, the mismatch penalty at 37°C was 4.4 kcal/mol compared with 3.3 kcal/mol in the absence of LNAs. Such an effect may reflect enhanced rigidity due to LNA, which thereby prevents a mismatch from adopting a favorable conformation. Thus, it may be advantageous to use LNAs to flank nucleotides likely to give small mismatch penalties.

SUPPLEMENTARY DATA

Supplementary Data is available at NAR Online.