EP2087133A2

EP2087133A2 - Improved hybridisation of nucleic acids

Info

Publication number: EP2087133A2
Application number: EP07815599A
Authority: EP
Inventors: Peter Laurence Molloy; Maxine June Mccall; Horace R Drew
Original assignee: Commonwealth Scientific and Industrial Research Organization CSIRO
Current assignee: Commonwealth Scientific and Industrial Research Organization CSIRO
Priority date: 2006-11-22
Filing date: 2007-11-22
Publication date: 2009-08-12
Also published as: EP2087133A4; AU2007324273A1; WO2008061311A3; WO2008061311A2

Abstract

A method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising providing at least one oligonucleotide probe of about 7 to about 25 nucleotides, providing a sampl comprising target DNA, wherein the DNA is prepared so as to comprise fragments of up to about 100 bases, incubating the DNA with the at least one oligonucleotide probe under conditions suitable to enable hybridisation between probe and target DNA, removing unbound DNA, and detecting DNA hybridised to the at least one oligonucleotide probe, wherein either or both of the DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety. The probe oligonucleotide may be located on a solid support such as a microarray. The sample DNA may be analysed for CpG methylation through treating it with bisulfite.

Description

Improved Hybridisation of Nucleic Acids

Field of Invention

The present invention relates generally to methods for improving fidelity in nucleic acid hybridisation assays and PCR priming reactions. In particular, the invention provides methods for reducing mis-pairing in hybridisation between sample DNA and an oligonucleotide. The invention therefore finds application, inter alia, in micro-array DNA analysis, for example in DNA methylation analysis, and in the detection of point mutations and single nucleotide polymorphisms.

Background of the Invention

The reference in this specification to any prior publication (or information derived from it), or to any matter which is known, is not, and should not be taken as an acknowledgment or admission or any form of suggestion that that prior publication (or information derived from it) or known matter forms part of the common general knowledge in the field of endeavour to which this specification relates.

Nucleic acid hybridisation is a fundamental tool in molecular genetics. The inherent ability of single stranded DNA to form a double helix with complementary DNA or RNA is the basis for all nucleic acid hybridisation assays. The accurate determination of complementarity between two nucleic acid strands relies on correct Watson-Crick base pairing between the strands, ie the inherent base pairing between adenine and thymine (A-T) nucleotides and guanine and cytosine (G-C) nucleotides, and is an important condition for the implementation of any nucleic acid hybridisation assay. Despite preferential base pairing between A-T and G-C nucleotides, however, mis- pairing of nucleotides is an acknowledged occurrence during target DNA hybridisation to oligonucleotide arrays.

The requirement for correct Watson-Crick base pairing is recognised in particular in oligonucleotide sequencing or analysis of DNA methylation involving the treatment of sample DNA with bisulphite. Bisulphite treatment of sample DNA converts unmethylated cytosines to uracils, whereas methylated cytosines remain unchanged. Amplification by PCR of the treated DNA results in the copying of uracil bases with adenine in the complementary strand, and subsequent pairing with thymine in the copying of that strand. Thus, the occurrence of an unmethylated cytosine in the sample DNA sequence is represented by the presence of a thymine base in the amplified DNA sequence, whereas methylated cytosine in the sample DNA remains as cytosine in the amplified DNA. For an accurate distinction between representative methylated and unmethylated DNA, correct G-C and A-T base pairing by an interrogating DNA probe must take place. Incorrect base pairing will result in an erroneous measure of methylated and unmethylated DNA.

In the case of mammalian DNA, cytosine methylation is mostly found at CpG dinucleotides; however cytosines at other positions may be methylated, especially, for example, at CpNpG trinucleotides in plant DNA. Currently, accurate sequencing of methylation patterns is limited by the available technology and the need to amplify and sequence a large number of amplicons using conventional sequencing technology. A higher throughput technology with potential for multiplexing would significantly expand capacity in this fast growing experimental and clinical area. Thus, for dealing with large numbers of samples, improvements to DNA sequencing chip technology to allow accurate reading and quantification of cytosine methylation is preferred to other sequencing methods .

Such a chip preferentially uses pairs of probes for all sequences to be read. The pairs of probes are identical in sequence except for the base, C or T, which interrogates the methylation status of C at CpG sites in the original DNA sample. If the target binds to the probe containing C(pG), the C was methylated in the original sample; if the target binds to the probe containing T(pG), the C was unmethylated in the original DNA sample. The complements of these probe pairs, containing CpG and CpA at the interrogating sites, may also be used.

Thermodynamic parameters, for G-C and A-T base pairs and all combinations of mis-pairs, which take into account the effect of stacking with their nearest neighbours, show that the order of decreasing stability is: G-C > A-T > G-G > G-T >G-A > T-T > A-A > T-C >A-C >C-C (SantaLucia J. & Hicks, D., 2004). Since G-T is one of the most stable mis-pairs, it was expected in the development of a methylation sequencing chip that destabilising of G-T mis-pairs would be required. It was also expected that increasing the stability of A-T to be comparable to that of the strong G-C base-pair would be required so as to equalise the stabilities of the double helices formed when target binds correctly to the various short probes of different base sequences. However, most surprisingly, it was found that the control methylated targets did not readily form G-C pairs, as expected, but commonly bound preferentially to probes that formed G-T mis-pairs, while the control unmethylated targets formed quite stable double-helices with A-T pairs.

Clearly, mis-pairing of bases between target and probe must be tightly controlled for a methylation sequencing chip of this type where the target must be able to discriminate a single base between otherwise identical probes. G-T mis-pairing by the methylated target would falsely increase the estimate of unmethylated C in the sample. Accordingly, there is a need for the development of methods for reducing G-T mis- pairing to restore correct G-C pairing. Furthermore, such methods would clearly find application in hybridisation technologies other than those designed for measuring the methylation status of DNA.

Summary of the Invention

The present invention relates to methods for improving hybridisation fidelity, for example, on oligonucleotide arrays, and advantageous applications of these methods. Guanine to thymine (G-T) mis-pairing frequently occurs in target DNA hybridisation to probes in oligonucleotide arrays. The present inventors have surprisingly found that G- T mis-pairing may be controlled such that its incidence is reduced, resulting in increased fidelity of target DNA hybridisation to probe sequences.

According to a first aspect, the present invention provides a method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising:

(a) providing at least one oligonucleotide probe of about 7 to about 25 nucleotides;

(b) providing a sample comprising target DNA, wherein the DNA is prepared so as to comprise fragments of up to about 100 bases; (c) incubating the DNA with the at least one oligonucleotide probe under conditions suitable to enable hybridisation between probe and target DNA;

(d) removing unbound DNA; and

(e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein at least one of the DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety.

The hybridisation may take place in solution or with one or both of the DNA and the at least one oligonucleotide probe attached to a solid support. - A -

Optionally, the sample of step (b) may comprise bisulphite-treated DNA. The bisulphite-treated DNA may undergo a further step of amplification.

According to a second aspect, the present invention provides a method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising:

(b) providing a sample comprising amplified DNA of a target region, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases;

(c) incubating the amplified DNA with the at least one oligonucleotide probe under conditions suitable to enable hybridisation between probe and target DNA;

(d) removing unbound amplified DNA; and (e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein at least one of the amplified DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety.

The hybridisation may take place in solution or with one or both of the DNA and the at least one oligonucleotide probe attached to a solid support.

Optionally, the sample of step (b) may comprise bisulphite-treated DNA.

According to a third aspect, the present invention provides a method of increasing fidelity of hybridisation between a probe oligonucleotide and a target DNA, the method comprising: (a) providing at least one oligonucleotide probe of about 7 to about 25 nucleotides;

(b) providing a sample comprising amplified DNA of a target region, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases; (c) incubating the amplified DNA with the at least one oligonucleotide probe under conditions suitable to enable hybridisation between probe and target DNA; (d) removing unbound amplified DNA; and (e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein at least one of the amplified DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety. The hybridisation may take place in solution or with one or both of the DNA and the at least one oligonucleotide probe attached to a solid support.

Optionally, the sample of step (b) may comprise bisulphite-treated DNA.

According to a fourth aspect, the present invention provides a method of oligonucleotide array-based analysis of DNA, the method comprising: (a) providing a sample comprising target DNA;

(b) amplifying a DNA region comprising target DNA, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases;

(c) incubating amplified DNA with one or more oligonucleotide probes of about 7 to about 25 nucleotides immobilised and arrayed on a solid support under conditions to allow hybridisation between fragmented DNA and probes,

(d) washing the oligonucleotides to remove unbound amplified DNA;

(e) detecting DNA hybridised to the oligonucleotide probes; wherein at least one of the amplified DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety.

Optionally, prior to step (b) the method also comprises treating the DNA of the sample with bisulphite to convert unmethylated cytosine bases to uracil.

According to a fifth aspect, the present invention provides a method of oligonucleotide array-based analysis of DNA methylation, the method comprising:

(a) providing a sample comprising target DNA;

(b) treating the sample with bisulphite to conveit unmethylated cytosine bases to uracil;

(c) amplifying a DNA region comprising target DNA, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases;

(d) incubating amplified DNA with one or more oligonucleotide probes of about 7 to about 25 nucleotides immobilised and arrayed on a solid support under conditions to allow hybridisation between fragmented DNA and probes;

(e) washing the oligonucleotides to remove unbound amplified DNA;

(f) detecting DNA hybridised to the oligonucleotide probes; wherein at least one of the amplified DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety.

The oligonucleotide probe may entirely comprise standard DNA nucleotides or may comprise one or more modified nucleotides, such as nucleotides having modifications in the base and/or sugar and/or phosphate moieties. Likewise, target molecules may entirely comprise standard DNA nucleotides or may comprise one or more modified nucleotides, such as nucleotides having modifications in the base and/or sugar and/or phosphate moieties.

The modified nucleotides may be selected from 2'-O-methyl nucleotides, such as 2'-O-methyl adenosine to replace 2'-deoxy-adenosine, 2'-O-methyl-uridine or 2'-O- methyl-thymidine to replace 2'-deoxy-thymidine, and, 2-amino-adenosine to replace adenine.

The oligonucleotide probe may be of any desired length from about 7 to about 25 nucleotides, in one embodiment, the oligonucleotide probe is about 7 nucleotides in length. The oligonucleotide probe maybe about 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24 or 25 nucleotides long.

The at least one oligonucleotide probe may optionally comprise one or more universal bases or a mixture of normal or modified nucleotides at one or both ends to increase the melting temperature of the oligonucleotide-target double-helices. Universal bases include 5-nitro-indole.

In an embodiment of the invention as defined herein, the oligonucleotide probes may be immobilised on a solid support.

Sample may comprise genomic DNA, cDNA or synthetic DNA. Typically, the DNA of the sample is subjected to a fragmentation step to generate fragments of up to about 100 nucleotides. Fragmentation of DNA may be achieved by any means such that fragments of up to about 100 base pairs are generated comprising target sequences. Such methods include DNAaseI digestion of DNA of the sample or the incorporation of uracil in amplified DNA of the sample followed by treatment with uracil-DNA glycosylase to create abasic sites at every uracil base incorporated, and either alkali cleavage or endonuclease VIII cleavage of abasic sites. Alternatively, endonuclease V may be used to introduce nicks at the second or third phosphodiester bond 3' to the uracil site. DNA fragments having up to about 100 nucleotides may also be prepared by amplification techniques or restriction enzyme digestion.

Typically, the conditions suitable to enable hybridisation between probe and target DNA encourage the formation of short hybrids with probe melting temperature (Tm) independent of base sequence.

Detection of target DNA hybridised to the oligonucleotide probe may be achieved through the detection of an optional detectable label associated with the DNA of the sample, or by other means such as a labelling dye that only binds duplex DNA or through the measurement of one or more physical properties indicative of the presence of duplex DNA. Suitable detectable labels include biotin, digoxygenin, radioactive labels, and fluorescent labels such as Cy3, Cy5, Quasar570 (QSR570), Quasar670 (QSR670), and Qdots, and other labels that function through a similar labelling mechanism. Labelling dyes that only bind duplex DNA include chromomycin, SybrGreen, and YOYO-I. The one or more physical properties that can be measured to detect duplex DNA include surface plasmon resonance.

According to aspects and embodiments of the invention both the amplified DNA and the at least one oligonucleotide probe may comprise one or more modified nucleotides.

According to a sixth aspect the present invention provides a method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising: (a) providing at least one oligonucleotide probe of about 7 to about 25 nucleotides comprising one or more modified nucleotides;

(b) providing a sample comprising target DNA, wherein the DNA is prepared so as to comprise fragments of up to about 100 bases;

(c) incubating the DNA with the at least one oligonucleotide probe under conditions suitable to enable hybridisation between probe and target DNA;

(d) removing unbound DNA; and

(e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein the DNA of the sample is optionally labelled with a detectable moiety. The hybridisation may take place in solution or with one or both of the DNA and the at least one oligonucleotide probe attached to a solid support.

Optionally, the sample of step (b) may comprise bisulphite-treated DNA. The bisulphite-treated DNA may undergo a further step of amplification. According to a seventh aspect the present invention provides a method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising:

(a) providing at least one oligonucleotide probe of about 7 to about 25 nucleotides comprising one or more modified nucleotides; (b) providing a sample comprising amplified DNA of a target region, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases and wherein the amplified DNA comprises one or more modified nucleotides;

(c) incubating the amplified DNA with the at least one oligonucleotide probe, under conditions suitable to enable hybridisation between probe and target

DNA;

(d) removing unbound amplified DNA; and

Optionally, the sample of step (b) may comprise bisulphite-treated DNA.

According to an eighth aspect, the present invention provides a method of increasing fidelity of hybridisation between a probe oligonucleotide and a target DNA, the method comprising:

(a) providing at least one oligonucleotide probe of about 7 to about 25 nucleotides comprising one or more modified nucleotides;

(b) providing a sample comprising amplified DNA of a target region, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases and wherein the amplified DNA comprises one or more modified nucleotides; (c) incubating the amplified DNA with the at least one oligonucleotide probe, under conditions suitable to enable hybridisation between probe and target DNA;

(d) removing unbound amplified DNA; and (e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein the DNA of the sample is optionally labelled with a detectable moiety.

Optionally, the sample of step (b) may comprise bisulphite-treated DNA. According to a ninth aspect, the present invention provides a method of oligonucleotide array-based analysis of DNA, the method comprising:

(a) providing a sample comprising target DNA;

(b) amplifying a DNA region comprising target DNA, the amplified DNA prepared so as to comprise fragments of up to about 100 bases, and wherein the amplified DNA comprises one or more modified nucleotides;

(c) incubating amplified DNA with one or more oligonucleotide probes of about 7 to about 25 nucleotides immobilised and arrayed on a solid support under conditions to allow hybridisation between fragmented DNA and probes, wherein the one or more oligonucleotide probes comprise one or more modified nucleotides;

(d) washing the oligonucleotides arrayed on the solid support to remove unbound amplified DNA;

(e) detecting DNA hybridised to the oligonucleotide probes arrayed on the solid support; wherein the DNA of the sample is optionally labelled with a detectable moiety.

According to a tenth aspect, the present invention provides a method of oligonucleotide array-based analysis of DNAmethylation, the method comprising: (a) providing a sample comprising target DNA;

(b) treating the sample with bisulphite to convert unmethylated cytosine bases to uracil; (c) amplifying a DNA region comprising target DNA, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases and wherein the amplified DNA comprises one or more modified nucleotides;

(d) incubating amplified DNA with one or more oligonucleotide probes of about 7 to about 25 nucleotides immobilised and arrayed on a solid support under conditions to allow hybridisation between amplified DNA and probes, wherein the one or more oligonucleotide probes comprise one or more modified nucleotides;

(e) washing the oligonucleotides arrayed on the solid support to remove unbound amplified DNA; and

(f) detecting DNA hybridised to the oligonucleotide probes arrayed on the solid support; wherein the DNA of the sample is optionally labelled with a detectable moiety.

According to an eleventh aspect, the present invention provides a method, of oligonucleotide array-based analysis of DNA methylation, the method comprising:

(a) providing a sample comprising target DNA;

(b) treating the sample with bisulphite to convert unmethylated cytosine bases to uracil;

(e) washing the oligonucleotides arrayed on the solid support to remove unbound amplified DNA; and ^■

Brief Description of the Drawings The present invention will now be described, by way of non-limiting example only, with reference to the accompanying drawings.

Figure 1: DNA sequence of single stranded LTM19, LTU19, and LTM55 short DNA targets. The sequence of LTM 19 within LTM55 is underlined. Figure 2: DNA sequence of one strand of double-stranded 85 base pair DNA target, LTM85. The sequence of LTMl 9 within LTM85 is underlined.

Figure 3: Sequences of cloned DNA, LINE5, LINE8 and LINE9 used as target controls. The locations of the sequences of LTM19, within the clones LINE8 and LINE9, and LTUl 9, within the clone LINE5, are underlined. Figure 4: Fragmentation of 150 nucleotide DNA targets, (a) DNA target, internally labelled using d^Cy5CTP, was treated with DNAse 1 for 0, 1, 3 or 10 minutes, (b) DNA labelled internally with d^Cy5CTP and prepared with differing ratios of dTTP:dUTP [no dUTP (lane A), or with dTTP:dUTP ratios of 9:1 (lane B), 3:1 (lane C), or 1 : 1 (lane D)] were fragmented using UDG followed by alkali cleavage, (c) DNA labelled internally with d^Cy5CTP and prepared with differing ratios of 1 : 1 or 3 : 1 dTTP:dUTP and either dATP (A) or d(2-amino)ATP (2AA) as shown was fragmented with USER enzyme mix. Fragmented DNA samples were electrophoresed on 4% agarose gels and imaged with a phosphorimager.

Figure 5: Long, 150-nt, single-stranded DNA targets bind weakly to short, 9-nt, DNA probes on the microarray, with strength depending on the methylation status of the target DNA. Hybridisation was overnight at 23°C, in 3M TMACl, 1OmM Tris-Cl pH8, ImM EDTA, 5% sarkosyl, after which slides were washed in hybridisation solution containing no target DNA, and scanned on a Phosphorimager. Each row bears five probes, each probe spotted as adjacent duplicates. The methylated target should bind to all five probes in the top row and one probe in the third row, first column, as indicated by ovals in (a), and the unmethylated target should bind to all five probes in the second row, and one probe in the last row, first column, as indicated by the ovals in (b). Methylated target (LINE8) shows almost no binding, while unmethylated target (LINE5) binds. All probes are 9mers of normal DNA. Figure 6: Length of DNA target affects properties of binding to oligonucleotide probes on chip. The targets represent methylated DNA controls, (a) Long target, 150-nt LINE9, does not bind to any probe, (b) Target of intermediate length, 85-nt LTM85, binds well to probes, forming G-C base pairs as expected (circled spots in Figure), but also forming strong G-T mispairs (all spots not circled), (c) Short target, 19-nt LTM19, shows correct binding in general to form G-C base pairs (circled spots); there are also numerous G-T mis-pairs (all spots not circled), but fewer than for the 85-nt target (compare the number of spots not circled in (c) and (b)). All probes are 9-mers of normal DNA.

Figure 7: Mis-paired bases form between short DNA targets and DNA probes, with strength depending on base sequence. Hybridisation of target oligonucleotide LTM19 with array containing normal DNA probes was overnight at 23⁰C in 0.5M sodium phosphate, pH 7.2, 7% sarkosyl, with subsequent washing in 3x SSC for 5 minutes, and twice in 0.5x SSC for 5 minutes, (a) Circled spots indicate correct binding of target LTMl 9 with normal DNA probes; numbers indicate the specific probe sequence, eg. 9=LP9 ; the variation in intensities of uncircled spots underneath each of the circled spots indicate the level of sequence-dependent G-T mis-pairing of LTMl 9 to probes with related sequences, (b) Sequences of probes on array are shown. The probe pairs LP1/LP2 and LP3/LP4 are arrayed in the first column with the mispair probes beneath the matching probes. The remaining four columns contain sets of probes interrogating 2 CpG sites, with the perfect match probe at the top and double and single mismatch probes beneath. Interrogated sites are indicated by underlining in the probe sequences.

Figure 8:Effect of shortening probe length on the strength of binding to target LTM85. (a) The image shown was recorded after hybridising in 2.0M TMA-formate, 1OmM Tris pH8, ImM EDTA, 0.012% Tween-20 overnight at room temperature, washing in hybridisation solution (minus target) at room temperature, and scanning on a Phosphorlmager. Numbers on the left of the image indicate the numbers of nucleotides in the probes. Probes are spotted on the slides as adjacent pairs, (b) Sequences of the probes.

Figure 9: Effect of lengthening a probe with a central section of 9 defined bases by adding a random mix of bases at each end. Fragmented, methylated target LINE9 was hybridised to the probes overnight at room temperature in 1.9M TMA-formate, 1OmM Tris-Cl, ImM EDTA, 0.002% Tween-20; the slides were washed in hybridisation solution (minus DNA) and scanned. When the target binds to the probes LP 13Xn on the left, Watson-Crick base pairs are formed (a); when it binds to the probes LP 15Xn on the right, a G-T mis-pair is formed (b). (c). Base sequences of probes. X represents an equal mix of the normal bases A₃G₃C₅T.

Figure 10:Effect of 2'-O-methyl-nucleotides in 9-nt probes on the specificity and strength of target binding, (a) Sequences of the probe sets, (b) Probes in column "DNA" have normal DNA₃ "OU" have all T replaced by 2'-O-methyl-U, "OA" have all A replaced by 2'-O-methyl-A, "OAU" have all A and all T replaced by 2'-0-methyl-A and 2'-O-methyl-U, respectively, and "OaIl" have all nucleotides replaced by 2'-O-methyl- nucleotides. Generic probe names are on the left of the image. Target was LTM85. Hybridisation solution was 2.0M TMA formate, 0.056M MES, 5% DMSO₃ 2.5X Denhardt's solution, 5.8mM EDTA₃ 0.0115% Tween-20. Circles indicate locations of probes to which the methylated target should bind with no mismatches.

Figure 11: Effect of 2-amino-adenine and 2'-O-methyl-2-amino-adenine in 9-nt probes on the specificity and strength of target binding, (a) Probes in column "DNA" have normal DNA₃ "D" have all A replaced by 2-amino-A, and "M" have all A replaced by 2'-O-methyl-2-amino-A. Generic probe names are on the left of the image. Target was LTM85. Hybridisation solution was 2.7M TMACl₃ 0.056M MES, 5% DMSO₃ 2.5X Denhardt's solution, 5.8mM EDTA, 0.0115% Tween-20. Circles indicate locations of probes to which the methylated target should bind with no mismatches, (b) Sequences of the probe sets. Figure 12: Effect on the specificity and strength of target binding to probes containing nucleotides with 2'-O-methyl modifications in combination with universal bases at the termini. Probes in column "DNA" have 9-nt of normal DNA, "V2" have 9- nt of DNA with a universal base at each end, "V4" have two universal bases at each end, "OU" have all T replaced by 2'-O-methyl-U, "OA" have all A replaced by 2'-O-methyl- A, "OAU" have all A and all T replaced by 2'-O-methyl-A and 2'-O-methyl-U respectively, "0UV4" have all T replaced by 2'-O-methyl-U and two universal bases at each end, "OA V4" have all A replaced by 2'-O-methyl-A and two universal bases at each end, "0AUV4" have all A and all T replaced by 2'-O-methyl-A and 2'-O-methyl-U respectively and two universal bases at each end. The universal base was 5-nitro-indole. Target was LTM85. Hybridisation solution was 2.7M TMACl₃ 0.056M MES₃ 5%

DMSO₃ 2.5X Denhardt's solution, 5.8mM EDTA, 0.0115% Tween-20. Circles indicate locations of probes to which the methylated target should bind with no mismatches. Generic probe names are on the left of the image (a) Probes LP3 and LP4 and (b) probes LP13 and LP15.

Figure 13: Effect of 5-propynyl-C and 5-propynyl-U in 9-nt probes on the specificity and strength of target binding. Probes in column "DNA" have normal DNA₅ "P" have each interrogating C and T replaced by 5-propynyl-C and 5-propynyl-U, respectively, and "Pall" have all C and T replaced by 5-propynyl-C and 5-propynyl-U, respectively. Generic probe names are on the left of the image. Target was LTM85. Hybridisation solution was 2.7M TMACl, 0.056M MES, 5% DMSO, 2.5X Denhardt's solution, 5.8mM EDTA, 0.0115% Tween-20. Circles indicate locations of probes to which the methylated target should bind with no mismatches.

Figure 14(a): Names of probes, in the order in which they appear, as adjacent pairs, on the DNA chip. Probes within a row have the same base sequence, but contain different modifications as indicated by the column headings (all probes in the "DNAX2" column have normal nucleotides; all probes in the "DX2" column have every A as 2- amino-A; all probes in the "0AX2" column have every A as 2'-O-methyl-A; all probes in the "OATX2" column have every A as 2'-O-methyl-A and the interrogating base T, if present, as 2'-O-methyl-T. Probes highlighted in yellow form correct base pairs with the fully methylated target LINES', and probes highlighted in green form correct base pairs with the fully unmethylated target LINE5. All probes have 11 nucleotides, with 9 central nucleotides of defined base sequence, and one nucleotide at each end being an equal mix ofA, G, C and T.

Figure 14(b): Image of DNA chip after hybridisation with the fragmented methylated target LINE9. The generic names of probes in each row are on the left of the image, and the heading of each column indicates the type(s) of modified nucleotides in probes in that column. Circles indicate locations of probes which form correct base pairs with target LINE9, and which are highlighted in yellow in Figure 14(a). Spots not circled indicate target binding to probes through G-T mis-pairs.

Figure 14(c): Averaged intensities for adjacent spot pairs shown in Figure 14(b), for binding by methylated target LINE9. The intensity of each spot was corrected for local background, and then background-corrected intensities for identical probes were averaged, to give the tabulated value for a particular probe, with error in parenthesis.

The probe order is the same as in Figure 14(a). Figure 14(d): Relative intensities reveal extent of mis-pairing by methylated control target, LINE9, to probes with various modifications. The relative intensity is the intensity for target binding to the probe relative to the intensity for target binding to the probe's partner with correct base pairing. The relative intensities were calculated from data in Figure 14(c).

Figure 14(e): Image of DNA chip after hybridisation with the unmethylated target LINE5. The generic names of probes in each row are on the left of the image, and the heading of each column indicates the type of modified nucleotides in probes in that column. Circles indicate locations of probes which form correct base pairs with target LINE5, and which are highlighted in green in Figure 14(a). Spots not circled indicate target binding to probes through A-C mis-pairs.

Figure 14(f): Averaged intensities for adjacent spot pairs shown in Figure 14(e), for binding by unmethylated target LINE5. The intensity of each spot has been calculated as described in Figure 14(c). The probe order is the same as in Figure 14(a). Figure 14(g): Relative intensities reveal extent of mis-pairing by unmethylated control target, LINE5, to probes with various modifications. The relative intensity is the intensity for target binding to the probe relative to the intensity for target binding to the probe's partner with correct base pairing. The relative intensities were calculated from data in Figure 14(f). Figure 15: Improved specificity from incorporation of 2-amino adenine in target

DNA. (a) Human MLHl sequence after bisulphite treatment. Target sites for primers are underlined. Potential methylation sites are designated YG and those within the amplicon are shown in bold and labelled A to H. (b) Hybridisation of probes hBl, hB2, KEl, hE2, KFl, hF2, KGl and hG2 to KMLHl sites B, E, F and G. Sequences of probes are shown in the box to the right with the interrogated bases in bold and underlioned. hi the hybridisation panels probes in the the upper rows contain normal DNA, those in the second rows (OA) contained 2'0-methyl adenosine at the sites shown in bold in the probe sequences and those in the third row (OAU) contain 2'O-methyl adenosines and also 2'O-methyl uridine at the interrogated base position. Figure 16: Nucleotide sequence of the human TPEF gene following bisulphite treatment. Target sites for primers are underlined. CpG sites are labelled A to I and shown in bold with cytosines that may be C or T after conversion shown as Y. Figure 17: (a) Layout of probes listed in Example 13 on arrays shown in Figure 17(b) and (c). Results of hybridisation with methylated target (b) and unmethylated target (c) are shown.

Figure 18: Hybridisation of modified probes with TPEF target sites A, B and C. Probe sequences and array layout for sites A and B (a) and site C (c). Chemically modified bases are shown in bold and discriminating bases are underlined. Results of hybridisation with unmethylated and methylated target for sites A and B (b) and unmethylated and methylated target for site C (d) are shown.

Figure 19: Micro-array probes containing modified nucleotides improve the discrimination between matched normal and tumour samples in the analysis of DNA methylation of the highly-repeated sequence, LINE, taken from a human patient with colorectal cancer. In each pair of images, LINE DNA extracted from normal tissue (left) and from tumour tissue (right) is bound to "DNA" probes (containing only normal DNA), "OU" probes (normal DNA with interrogating bases of normal C (upper) or 2'- O-methyl-U (lower)), "OA" probes (all A are 2'-O-methyl-A), "OAT" probes (all A are 2'~O~methyl-A, and interrogating bases are normal C (upper) or 2'-O-methyl-T (lower)). In each panel, the upper probes are specific for methylated target DNA, and lower probes are specific for unmethylated target DNA; each probe has been spotted on the micro-array as an adjacent pair. The number below each panel is the ratio of intensities (background-corrected) for target binding to the upper probes compared with the lower probes (intensities have been averaged over all probes of the same type spotted on the chip, including those appearing in the panel). The single number in bold italics below each pair of panels is the ratio of intensities of methylated:unmethylated (uppeπlower) probes for the normal sample compared with the tumour sample, and represents the discrimination achieved by each probe type in binding normal versus tumour DNA. (a) Matched probes with the generic sequences of LP 13 (specific for fully-methylated target DNA) and LP14 (specific for fully-unmethylated target DNA), (b) matched probes with the generic sequences of LP21 (specific for fully-methylated target DNA). and LP22 (specific for fully-unmethylated target DNA), and (c) matched probes with the generic sequences of LPBl (specific for fully-methylated target DNA) and LPB2 (specific for fully-unmethylated target DNA).

Figure 20: Methylation of TPEF gene in clinical samples, (a) Gel electrophoresis of restriction enzyme digested (HpyCH4TV and BstUΪ) DNA from normal (N) and tumour (T) tissue from five human patients following bisulphite treatment and PCR amplification, (b) Probe sequences and array layout for hybridisation. Chemically modified bases are shown in bold and discriminating bases are underlined, (c) Hybridisation of probes to patient DNA.

Definitions

Throughout this specification and the claims which follow, unless the context requires otherwise, the word "comprise", and variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps.

The articles "a" and "an" are used herein to refer to one or to more than one (i.e., to at least one) of the grammatical object of the article. Byway of example, "an element" means one element or more than one element.

As used in the context of the present invention, the term "probe" refers to an oligonucleotide. Optionally, the probe may be attached to a solid support.

As used in the context of the present invention, the term "solid support" refers to any surface on which probe oligonucleotides and/or target DNA may be immobilised. Such surfaces include nitrocellulose or vinyl membranes, glass, plastic or silicon slides, microbeads and other similar surfaces known in the art. The term "attached" encompasses both direct and indirect means of attachment of a nucleic acid molecule to the solid support. For example, indirect attachment may involve attaching, by any suitable means, the nucleic acid molecule to one or more linker molecules or compounds which in turn are bound to the surface of the solid support.

As used in the context of the present invention, the terms "DNA chip", "DNA microarray", "microarray" or "hybridisation array" refers to a solid support typically having oligonucleotide probes arrayed on its surface. Typically, the solid support utilised in the preparation of a chip or microarray is a nitrocellulose or nylon membrane, or a glass, plastic or silicon slide, or a bead.

As used in the context of the present invention, the term "fidelity" refers to accuracy in nucleotide base pairing. Accurate base pairing equates with Watson-Crick base pairing, ie preferential base pairing between standard 2'-deoxy adenosine and T- deoxy-thymidine nucleosides, and standard 2'-deoxy-cytidine and 2'-deoxy-guanosine nucleosides, and includes base pairing between standard and modified nucleosides and base pairing between modified nucleosides, where the modified nucleosides are capable of substituting for the appropriate standard nucleosides according to the Watson-Crick pairing. Thus, in accordance with the present invention, an increase in fidelity refers to an increase in Watson-Crick pairing over non- Watson-Crick pairing. As used in the context of the present invention, the term "oligonucleotide" refers to a nucleic acid comprising deoxyribonucleotides, and optionally one or more modified nucleotides, and/or universal bases. Modified nucleotides include ribonucleotides.

As used in the context of the present invention, the term "modified nucleotide" or "modified base" refers to a nucleotide that differs in structure from the standard or "unmodified" nucleotides 2 '-deoxy- adenosine, 2'-deoxy-thymidine, 2'-deoxy-cytidine and 2'-deoxy-guanosine, and that is capable of pairing with an unmodified nucleotide or a modified nucleotide. Modified nucleotides include ribonucleotides.

As used in the context of the present invention, the term "target" refers to a DNA molecule that is added in hybridisation buffer for binding to the probes. Optionally, the target is either internally labelled or end-labelled with a fluorophore or other detectable moiety. Sample may be known to contain target DNA, as applies when interrogating the methylation status of a target DNA, or, the sample may be 'searched' by an oligonucleotide array for target DNA sequence (as determined by probe sequence), such as in a gene expression array. As used in the context of the present invention, the term "sample" refers to a biological sample that comprises DNA that in turn typically comprises, or is suspected to comprise, target DNA. DNA of the sample maybe subjected to amplification and/or fragmentation.

As used in the context of the present invention, the term "methylated target" refers to DNA of the sample and the target DNA it comprises that originates from methylated genomic DNA, enzymatically methylated DNA fragments or enzymatically methylated synthetic DNA, and that have undergone treatment with bisulphite, a reaction that converts unmethylated cytosines in the DNA to uracils but does not convert methylated cytosines. Following PCR amplification of the bisulphite treated DNA, methylated cytosines in the starting DNA appear as cytosines in the amplified DNA. The term "methylated target" also refers to chemically synthesised DNA with cytosines at sites representing 5-methyl-cytosine. As used in the context of the present invention, the term "unmethylated target" refers to sample DNA and the target DNA it comprises originating from unmethylated genomic DNA, unmethylated DNA fragments or unmethylated synthetic DNA that has undergone treatment with bisulphite such that unmethylated cytosines in the DNA have been converted to uracils. Following PCR amplification of the bisulphite treated DNA, the uracils are replaced with thymines, and therefore unmethylated cytosines in the starting DNA are represented by thymines in. the amplified DNA. Unmethylated target also includes chemically synthesised DNA with thymine at sites representing unmethylated cytosine. As used in the context of the present invention, the term "target DNA region", or

"target region" in addition to referring to a single target region may also refer to multiple, independent target regions. Thus, multiple, independent target regions may be amplified simultaneously.

As used in the context of the present invention, the term "PCR" may refer to linear, non-exponential amplification of DNA in addition to exponential amplification of DNA, where the person skilled in the art would recognise that either form of amplification is appropriate for the purpose of the invention.

Detailed Description of the Invention

The present invention relates to methods for improving fidelity in hybridisation assays, and advantageous applications of these methods. As exemplified herein, methods in accordance with embodiments of the invention for improving hybridisation fidelity are shown to substantially improve oligonucleotide array-based sequencing of DNA methylation and allow the use of very short (7 to 11 base) oligonucleotides. It will be well appreciated by a person skilled in the art, however, that the methods demonstrated herewith for improving hybridisation fidelity are equally applicable to any nucleic acid hybridisation assay, a reduction in nucleotide mis-pairing in hybridisation being a universally recognisable and desirable advantage for most, if not all, DNA hybridisation-based assays. By way of example, methods of the invention can also be used in the detection of point mutations and single nucleotide polymorphisms (SNPs) and more generally in a variety of PCR-related applications.

Methods carried out in accordance with the invention find application in the diagnosis of, or predictor of susceptibility to, a disease or condition in a subject, which disease or condition is characterised by or associated with a variant genetic sequence. It will also be appreciated that methods of the invention may be employed in liquid or solid phase assay systems. Thus the methods are equally applicable to assays conducted in solution, such as in standard PCR and related reactions and to array-based assays in which one or more of the nucleic acid molecules to be hybridised is attached to a solid support.

As exemplified herein, using arrays composed of short oligonucleotide probes, the inventors have discovered a surprising relationship between the length of the target DNA hybridising to the probes and the fidelity of the hybridisation reaction. Reduction in the length of the target DNA is expected to increase the rate of hybridisation, but it has also been found that the extent of G-T mis-pairing decreases with decreasing length of the target DNA. G-T mis-pairing is also observed to be sequence dependent.

Although these measures can increase hybridisation fidelity by decreasing G-T mis-pairing, the present inventors have further discovered that the presence of modified nucleotides in the probe and/or target provides for increased fidelity in hybridisation for both short and longer target sequences.

Accordingly, a first aspect of the present invention provides a method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising:

(c) incubating the DNA with the at least one oligonucleotide probe under conditions suitable to enable hybridisation between probe and target DNA; (d) removing unbound DNA; and

(e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein at least one of the DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety. Typically, the conditions suitable to enable hybridisation between probe and target DNA encourage the formation of short.hybrids with probe melting temperature independent of base sequence. This maybe encouraged in the presence of TMA salts of singly-charged anions such as chloride, formate, and acetate. Preferably, hybridisation and wash solutions comprise TMA-formate at a concentration of between 0.8M and 2.0M, or TMA-chloride at a concentration of between 1.7M and 2.3M. More preferably, TMA-formate is present at 1.7M, or TMA-chloride is present at 2.0M.

Hybridisation and wash solutions may further comprise one or more additional reagents. Such reagents include Tween 20 and SDS.

Those skilled in the art will appreciate that in accordance with the present invention, modified nucleotides may be present in either the target DNA or the oligonucleotide probe, or both. Suitable modified nucleotides include, but are not limited to, 2'-O-methyl nucleotides, such as 2'-O-methyl adenosine, 2^?-O-methyl-uridine, 2'-O-methyl-thymidine, 2'-O-methyl-cytidine, 2'-O-methyl-guanosine, 2'-O-methyl-2- amino-adenosine; 2-amino-adenosine, 2-amino-purine, inosine; propynyl nucleotides such as 5-propynyl uridine and 5-propynyl cytidine; 2-thio-thymidine; universal bases such as 5-nitro-indole; locked nucleic acid (LNA), and peptide nucleic acid (PNA).

Detection of DNA methylation An exemplary application of the present invention relates to the analysis of DNA methylation status, often used in the diagnosis of, or determination of susceptibility to, disease.

Methylation of cytosine bases at CpG sites in genomic DNA can be detected by treating the DNA with bisulphite, a reaction that converts unmethylated cytosines to uracils but does not change methylated cytosines. After amplifying the bisulphite-treated DNA using PCR, the uracils are converted to thymines, and the methylated cytosines become normal cytosines (Frommer, et al. 1992; Clark, et al. 1994).

Thus the methylation status of genomic DNA can be determined by sequencing the DNA after bisulphite/PCR treatment. In order to utilise oligonucleotide microarrays for determining the methylation status of cytosines in an amplified PCR product, it is necessary to be able to clearly distinguish hybridisation of methylated and unmethylated target DNAs to oligonucleotide probes of the type nnniiCGnnn and nnnnTGnnn where n are variable nucleotides; these short probes may contain one or more CG and/or TG dinucleotide steps. Shorter probes have two advantages - firstly, the relative effect of single base mismatches on hybrid stability increases with decreasing length of the probe and secondly, the number of probes needed to include all possible sequence variations decreases substantially as probe length decreases. Conversely, hybrids with shorter probes are of lower stability and their stability is more strongly influenced by base composition. Considering the case of oligonucleotides containing 9 specified bases: if the central base is designated as the cytosine to be interrogated (i.e. all oligonucleotides are of the type nnnnCGnnn or nnnnTGnnn) all possible sequence variants, including multiple CpG sites, are contained in a set of 9374 oligonucleotides. Without the restriction of a central CpG interrogation position, approximately 50,000 oligonucleotide probes are needed to cover all combinations of CpG sites that could arise from bisulphite-treated DNA. Similar oligonucleotide sets could be prepared to interrogate other contexts of cytosine methylation, such as CpNpG sites in plant DNA. Alternatively, methylation sequencing chips with fewer probes may be used to interrogate defined sequences of interest, such as for diagnostic purposes.

The present invention will now be further described in the following examples of preferred but non-limiting embodiments thereof and the accompanying Figures.

Examples

Example 1 : Preparation of Target DNAs

Target DNAs for hybridisation to arrayed probes were either synthesised as oligonucleotides representing sequences that would be obtained after bisulphite conversion of DNA or were amplified by PCR from bisulphite treated DNA or clones of PCR amplicons. All oligonucleotides were purchased from Geneworks Pty Ltd (Adelaide, South Australia). DNA was bisulphite treated using standard methods (Clark et al., 1994) or a kit either from Qiagen (Epitect) or from Human Genetic Signatures (MethylEasy), and following the manufacturer's instructions. 1.1. Preparation of target DNA molecules with defined DNA sequences.

Methylated and unmethylated target DNA molecules with various numbers of nucleotides and of known sequence were prepared as controls to test the effect of target length on the binding of targets to the microarray.

1.1.1 Short DNA targets having 19 or 55 nucleotides. Short, single-stranded, DNA target molecules, containing 19 nucleotides, were synthesised with a fluorophore, QSR570 (similar characteristics to Cy3) or QSR670 (similar to Cy5), on the 3' end, and a single-stranded DNA target molecule of 55 nucleotides was synthesised with Cy5 on the 5' end. LTM19 and LTM55 represented the methylated form of the target of interest, and LTU 19 the unmethylated form (Figure 1). The underlined bases in LTM55 are the same bases as in the 19-mer methylated target, LTMl 9. 1.1.2 Double-stranded DNA target with 85 nucleotides

A double-helix of 85 base pairs was made by annealing two partially overlapping synthetic oligonucleotides (a 45-nucleotide molecule and a 55-nucleotide molecule 5'- end labelled with Cy5 (LTM55)) and filling with the Klenow fragment of DNA polymerase I. The Cy5 -labelled strand produced by this method, LTM85, is a longer version of LTM55 and includes the same bases as LTM19 (underlined, see Figure 2). The complementary strand is not labelled.

1.1.3 Long DNA targets (≥ 100 base pairs) •

Long DNA molecules which served as either fully-methylated or else fully- unmethylated controls were PCR amplified from human DNA after bisulphite treatment. The fully methylated targets were amplified using human genomic DNA that had been pre-treated with the methylase Sssl, so as to add methyl groups at all (or most) CpG sequences, and then was converted as C to U or else methyl-C to methyl-C using standard bisulphite methods. The unmethylated targets were PCR amplified using human sperm DNA from which all pre-existing methyl groups had been removed by whole-genome-amplification (with phi-29 DNA polymerase in the presence of dCTP), and then was converted using standard bisulphite methods. Standard PCR in a 25 μl volume contained IX Taq DNA polymerase buffer (GoTaq, Promega), with 2mM MgCl₂, 0.3μM forward primer F3, 0.3μM reverse primer R3, 200μM each of dGTP, dATP, dTTP and dCTP and 1 unit of Taq polymerase. The mixture was heated at 95°C for 1 minute, and then subjected to 30 cycles of 95°C for 10 seconds, 50°C for 20 seconds, 72°C for 60 seconds; finally the mixture was held at 72°C for 2 minutes. Primers F3 and R3 have the sequences 5' GGGRGAGGAGTTAAGATGGT and 5' CCCACTATCTAACACTCCCTAATA respectively. In each case, the PCR amplicon (from either Sssl or whole-genome-amplification) was cloned into the pGEM T-easy vector (Promega). Inserts of the desired sequence (typically all CpG or all TpG/CpA) were kept as target controls for further work using DNA chips.

Figure 3 shows the base sequences of cloned targets LINE5, LINE8 and LINE9. Underlined bases in LINE5 are the same as those in the unmethylated short target, LTU19. Underlined bases in LINE8 and LINE9 are the same as those in the methylated short target, LTMl 9.

1.1.3.1 Labelling of long target DNAs.

Cy5- or Cy3-containing long-target controls were each amplified from a plasmid DNA template of desired sequence (either Sssl or whole-genome-amplification, as described above, diluted 1/50) in 25 μl under standard PCR conditions except that 200 μM dCTP was substituted with a mix of lOμM dCTP and lOμM d^Cy5CTP (in a molar ratio of 1:1). Each completed PCR was checked by electrophoresing lμL on a 2.5% agarose mini-gel. Alternatively, the PCR mixture contained lOμM dCTP and lOμM d^biotinCTP (in a molar ratio of 1 : 1), for subsequent detection of the target (after the hybridisation step) using streptavidin conjugates.

In some cases, PCR amplifications were performed in duplicate or quadruplicate (to make more material), and the identical reactions were pooled at this stage. As an entirely optional step, the completed PCR could be purified (to remove Taq polymerase, buffer, unreacted primers and dNTPs) using a silica-based spin column (Rapid PCR Purification System, Marligen Biosciences). However, this extra step was not essential, and may generally be omitted to save time and effort. Each completed (and unpurified) PCR reaction contains sufficient labelled target DNA for two to four hybridisation experiments.

Example 2: Fragmentation of single-stranded and double-stranded DNAs for targets having normal or modified nucleotides.

2.1 Fragmentation of internally labelled, double-stranded targets using DNAaseI to produce short, double-stranded internally labelled targets

Double-stranded, internally labelled, DNA products of 150 bp were prepared by PCR (1.1.3.1) and products were purified over Marligen columns to remove unreacted dNTPs and excess primers. The purified products in lOOμl were adjusted to 0.5X NEB buffer 2 (New England Biolabs) containing 5mM MgCl₂. About 0.5 units of DNAaseI (Affymetrix) were added, and the DNA was digested for 1, 3 or 10 minutes. The reaction was stopped in excess EDTA. 2μl from each reaction was analysed on a 4% agarose/Nu- sieve gel to determine the sizes of the fragmented DNA (Figure 4a).

2.2. Fragmentation of internally-labeled, single-stranded DNA using UDG and NaOH Bisulphite-treated, methylated, genomic DNA was amplified by PCR with a biotin-labeled primer (biotin-F3) and an unlabelled primer (R3) in the presence of normal dATP and dGTP, a 1:1 mix of Cy5-labelled dCTP and normal dCTP, and dTTP and dUTP in the following ratios: (A) 200μM T with OμM U, (B) 180μM T with 20μM U, (C) 150μM T with 50μM U, and (D) lOOμM T with lOOμM U. Excess primers and unreacted dNTPs were removed using Marligen columns. The purified DNA of 150bp was mixed with magnetic beads coated with streptavidin (M-280) to bind the strand bearing the biotin; the other strand was released in 10OmM NaOH at 20°C. The released strand was neutralised with an equal amount of KOAc/EDTA, diluted to low salt, and then treated with Uracil-DNA glycosylase (UDG) for 1 hour at 37°C to create an abasic site for every U base incorporated during the PCR. The mixture was then adjusted to 10OmM NaOH and heated at 95 °C for 10 minutes to break the sugar-phosphate chain at each abasic site. The solution was then neutralised to pH 6 using an equal amount of KOAc, pH 5. The sizes of the fragmented targets produced for various T:U ratios are shown in Figure 4(b). The uncleaved target, shown in lane A is 150-nt. Increasing the amount of dUTP, relative to dTTP, produced shorter targets.

2.3. Fragmentation of DNA targets with USER enzyme

To prepare samples of fragmented DNA targets, DNA was PCR-amp lifted as described in 1.1.3, using T:U ratios of 3:1 (150μM dTTP and 50μM dUTP) and 1 :1 (1 OOμM dTTP and 1 OOμM dUTP). In parallel reactions dATP was substituted with d(2- amino-adenine)TP. Pooled, duplicated reactions were purified over Marligen columns. Mixtures were then heated at 95°C for 1 minute, cooled briefly to room temperature, and 1 unit of USER enzyme was added (NewEngland Biolabs). After incubating at 37⁰C for 1 hour, mixtures were again heated at 95°C for 1 minute, cooled briefly to room temperature, and an additional 1 unit of USER enzyme was added. Mixtures were again incubated for at least 1 hour (up to 3 hours) at 37⁰C. The extent of DNA fragmentation was checked by electrophoresing 2μL of the mixture on a 4% agarose mini-gel (Figure 4c).

Treatment with USER enzyme (a mixture of UDG (Uracil DNA glycosylase) and Endonuclease VIII, a DNA glycosylase-lyase), results in cleavage of the DNA wherever U is incorporated, as the UDG excises the uracil bases and the Endonuclease VIII breaks the phosphodiester chain at the abasic sites. Thus, target fragment lengths may be controlled as desired by adjusting the T:U ratio (eg the target DNA may be fragmented to shorter lengths by lowering the T:U ratio from 3:1 to 1:1).

As an alternative to USER enzyme, endonuclease V may be used to fragment uracil-containing DNA molecules. Endonuclease V nicks DNA containing uracil at the second or third phosphodiester bond 3' to the uracil site (Miyazaki, K. (2002).

Example 4. Probe Design.

Pairs of probes were designed for binding with Watson-Crick base pairing to sites in the target DNA which contained CpG! Where the C in CpG sites is methylated, the target will retain the sequence CpG after bisulphite treatment and PCR. Where the C in CpG sites is unmethylated, the target will have the sequence TpG after bisulphite treatment and PCR (with CpA in the complementary strand). Thus, the pairs of probes have identical base sequences except for the interrogating bases G or A (for interrogating CpG or TpG, respectively, in the target) or, alternatively, they have the interrogating bases C or T (for interrogating CpG or CpA in the complementary strand of the target). Most probe pairs were tested with C and T as the interrogating bases.

Several probe pairs are required to determine the methylation status of closely- spaced CpG sites. For example, the sequence 5 'NNCGNCGNN in a target DNA requires the four probes, 5 'NNCGNCGNN, 5 'NNTGNTGNN, 5 'NNTGNCGNN and 5 'NNCGNTGNN, to determine the status of cytosine methylation in these closely spaced sites.

Example 5. Preparation of DNA chips.

5.1 Preparation of oligonucleotide probes

Oligonucleotides were synthesised by Gene Works Pty Ltd (Adelaide, South Australia). All probes contained an amine linker on one end, mostly the 3' end. The concentration of each probe was calculated by measuring the absorbance at

260nm, and using the following extinction coefficients for each nucleotide (e₂₆o 1 mol^"1 cm^"1, dA 15400, dG 11500, dC 7400, dT 8700, d-(2-amino)-A 8500, d-(5-propynyl)-U 3500, d-(5-propynyl)-C 5100, 2'-O-methyl-A 15400, 2'-O-methyl-G 11500, 2'-O- methyl-C 7200, 2'-O-methyl-T 8700, 2'-O-methyl-U 9900, 2'-O-methyl-2-amino-A 8500, d-(5-nitro-indole) 16000). The purity of each oligonucleotide was analysed by recording the absorbance spectrum between 220 and 300nm. AU oligonucleotides with 3' amine linkers were also labelled on their 5' ends with P³²-phosphate, electrophoresed on 20% polyacrylamide gels containing 7M urea, and scanned with a Phosphorlmager (Molecular Dynamics); a probe was subsequently gel-purified if this analysis showed that shorter fragments were present in significant amounts (>20%) together with the desired full-length molecule.

5.2 Printing oligonucleotide probes on slides,

Probe solutions were prepared for printing on Codelink slides typically as 25 μM solutions in 5OmM sodium phosphate buffer, pH 8.5. Codelink slides (Amersham Biosciences) consist of standard glass microscope slides coated with a hydrophilic polymer containing N-hydroxysuccinimide ester reactive groups. Probes were arrayed on the slides using a Floating Pin Replicator (VP478A), and a Scientific Glass Slide Indexing System (VP470), from V&P Scientific. The oligonucleotide probes were bound through their amine linkers to activated Codelink slides following the manufacturer's instructions. Areas on the slide not printed with oligonucleotide probes were subsequently blocked with ethanolamine. Slides arrayed with the oligonucleotide probes were stored in a desiccator until ready for use.

Example 6. Hybridisation.

Solutions of target DNA were denatured by heating at 90°C for 1 minute, and then either quickly chilled on ice or cooled briefly at room temperature. An aliquot (up to 25 μl) of the denatured target DNA was added to hybridisation solution to give a final volume of 50 to 100 μl, and mixed well. Various hybridisation solutions were found to be suitable (data not shown) and some of these are described in particular examples, but typically, hybridisation solution consisted of 2M TMACl (tetra-methyl-ammonium- chloride, Sigma) or 1.7M TMA-Formate (tetra-methyl-ammonium-formate, Sigma), 5OmM Tris-HCl pH 7.6 (Sigma), 5mM EDTA (Sigma), 0.01% Tween-20 (Pierce), 0.05% SDS (Sigma), and 10OnM short synthetic DNA, or 25μL target DNA solution prepared as described above in Example 2. TMA concentrations were chosen for a given TMA salt to give approximately equal Tms for target binding to probes with a range of base sequences. When a TMA salt is used in the hybridisation buffer, any TMA salt with a singly charged anion is suitable, in particular, chloride, formate or acetate. 50-100μL of hybridisation buffer containing target DNA was placed on a DNA chip, and covered with a glass cover slip. The slide was placed in a sealed box, humidified by paper soaked in water, and left to sit at room temperature for 15-30 minutes (or longer, 1 hr to overnight, if convenient). The slide was then washed briefly at room temperature three times with 150μL hybridisation buffer containing no target.

Where the target DNA was labelled with Cy5 or Cy3, slides were scanned directly on a Phosphorlmager (Typhoon, Molecular Dynamics) using a red 633nm laser and a 670nm filter for Cy5-labelled DNA or a green 532 laser and a 555nm filter for Cy3- labelled DNA to determine the extent of hybridisation of the target to the probes attached to the slide. If necessary, the slides were further washed either with hybridisation buffer (containing no target) at room temperature, or at a higher temperature (typically 37°C), and scanned again.

If prepared with d^biotinCTP, the target was labelled with R-phyco after the hybridisation and washing steps. 50 μl of Ix hybridisation solution containing 0.1 μg Streptavidin R-phycoerythrin conjugate (SAPE, Molecular Probes (Invitrogen)) was squirted onto the chip, covered with a clean glass slip, and left to sit for 15 minutes at 2O⁰C. The cover slip was removed, and unbound material was washed off three times with 150 ul of Ix hybridisation solution. Fluorescence on the slide was read on a Phosphorimager using a green 532 nm laser and TAMRA 580 nm filter. The same slide arrayed with oligonucleotide probes may be re-used by rinsing briefly under distilled water at 20°C, and then soaking in distilled water at 65°C for 10 minutes (twice optionally), and air drying.

Example 7. G-T mis-pairing is a serious problem when normal DNA targets hybridise to normal DNA probes.

7.1 Long targets of methylated DNA bind only weakly to short probes on the DNA chip. hi contrast to hybridisation with short oligonucleotide targets (shown in Example 7.2), initial experiments with long targets of defined length (150-nt), which were PCR products from bisulphite-treated genomic DNA, showed very weak binding to the short (9-mer) DNA probes arrayed on the Codelink slides in a variety of hybridisation solutions (data not shown). Target DNA corresponding to LINE5 and LINE8 was prepared as in Example 2.2 (A) using a biotin-labelled forward primer and a Cy5-end- labelled reverse primer but was not subjected to fragmentation. Under hybridisation conditions shown in Figure 5 unmethylated control target bound effectively with correct specificity to oligonucleotide probes (Figure 5b). Methylated control target DNA bound poorly, if at all (Figure 5 a).

7.2 The length of a normal DNA target influences its extent and specificity of binding to oligonucleotide probes on a chip.

Target DNAs of differing lengths were tested to determine whether target length influenced binding to the microarrays. The targets represent methylated DNA. Hybridisation of targets to probes was overnight at 23°C in a solution containing 2.7M TMACl, 0.056M MES, 5% DMSO, 2.5X Denhardt's solution, 5.8mM EDTA, 0.0115% Tween-20. The slides were washed in hybridisation solution without target, and scanned. All probes were 9-mers of normal DNA.

A 150-nt long methylated DNA target, LINE9, showed no binding to the 9-nt probes on the chip (Figure 6(a)). LINE9 was prepared from methylated genomic DNA as described in Example 7.1; it is a different clone from LINE8 which was used in Example 7.1, but it has the same sequence as LINE8 in the region which binds to the probes printed on the DNA chip. Thus, poor binding of long 150-nt methylated DNA targets to the DNA probes was confirmed.

An 85-nt methylated DNA target (prepared as in 1.1.2), LTM85, bound well to the oligonucleotide probes but with substantial G-T mis-pairing (Figure 6(b)). A 55-nt methylated DNA target, LTM55, showed binding similar to the 85-nt target (data not shown). A short methylated DNA target of 19-nt, LTM19, bound well to the probes (Figure 6 (c)). Compared with the 85-nt target, the 19-nt target formed fewer and weaker G-T mispairs, but G-T pairs were still present (compare intensities of spots not circled in Figures 6(b) and (c)). These experiments showed that, for both targets and probes composed of normal

DNA, shorter targets needed to be made in order to observe probe binding, and also means for reducing or eliminating mis-pairing, particularly G-T mis-pairing, needed to be found.

7.3 Formation of G-T mis-pairs by short DNA targets is sequence dependent. In general, our experiments showed that shorter DNA targets have less extensive

G-T mis-pairing with the oligonucleotide probes than longer targets. However, the strength of the G-T mis-pairing by the shorter targets, is sequence dependent (Figure 7). Synthetic target, LTMl 9, (19-nucleotide target representing methylated DNA) was hybridised to normal DNA probes (9-mers) arranged as pairs interrogating single CpG sites or sets of four probes interrogating 2 CpG sites. Circled spots in Figure 7(a) show correct binding of the target to probes LPl, LP3, LP5, LP9, LP13 and LP17. Uncircled spots elsewhere reveal G-T mis-pairing between target and probe. Probes LPl and LP3 each contain the same single interrogating site stepped by 2 bases; LTMl 9 binds strongly to the correct probes (circled) forming a G-C base pair, and weakly (with a G-T mis-pair) to the paired probes LP2 and LP4 (immediately underneath the circled probes). Sequences LP5, LP9, LP13 and LP17 each have two interrogated sites and hence require a total of four probes each to determine their methylation states. For each set of probes LP5, LP9, LP13 and LP17, LTM19 will form two G-C base pairs at the interrogated sites with probes in the top line (circled) in Figure 7(a), two G-T mis-pairs with probes in the second line (under 5, 9, 13, and 17), one G-C base pair and one G-T mis-pair with probes in the third line, and one G-T mis-pair and one G-C base pair with probes in the fourth line (where the locations of the G-C and G-T are swapped compared with the third line). The relative intensities of the uncircled spots for the probe sets LP5, LP9, LP 13, and LP 17 show that G-T mis-pairing with methylated target LTM 19 is strongest for sequences LP5 and LP9, less strong for sequence LPl 3, and weak for sequence LP 17, indicating that sequences flanking the interrogated sites influence the strength of the G-T mis-pair. The interrogated sites in the LP 13 and LP 17 sets of probes are the same, except that their position in the probe is displaced by two bases. Likewise the probe pairs LP1/LP2, LP3/LP4 and LP5/LP7 all interrogate the same CpG site. Comparison of the levels of mis-pairing when there is a single base mismatch demonstrates the importance of position of the mismatch in the hybridising sequence. In summary, G-T mis-pairing occurs in a sequence and position-dependent manner for short targets binding to short probes, with both target and probe consisting of normal DNA, and the extent of G-T mis-pairing increases as the target lengthens.

Example 8. Investigating the effect of various design parameters for normal DNA probes on the strength and specificity of target binding. A number of parameters were investigated to determine the optimal design of the oligonucleotide probes in order to achieve strong and highly specific binding by the target DNA, with Tms approximately equalised over probes of widely varying base sequence. These parameters included varying the length of the defined base sequence in the probes, extending the length of the probe by keeping the number of defined bases constant and adding various numbers of N bases at either end of the probe (where N is either the "universal" base, 5-nitro-indole, or N is an equal mix of the normal bases A, G₅C, T), extending the probe out from the surface of the coated chip by inserting a spacer chain of 18 carbon atoms between the surface of the chip and the defined sequence, comparing connection of the probe to the chip through either the 5' end or the 3' end, and comparing the effect of switching the target sequence to the probe and the probe sequence to the target.

8.1 Effect of shortening the probe The number of probes required for sequencing the methylation status of all cytosine bases in all available sequence contexts may be minimised by reducing the probe length. In order to determine how short the probes could be, while still obtaining reasonable target-binding strength, probes with a common base sequence and containing 9, 8, 7, 6, or 5 nucleotides, were hybridised with the methylated target, LTM85, in 2M TMA-formate, 1OmM Tris pH8, ImM EDTA, 0.012% Tween-20, overnight at room temperature. As shown in Figure 8(a), the amount of target binding to probes decreased with decreasing probe length, with no target observed to bind to the probe containing just 5 nucleotides, LP3(5), under these experimental conditions. While it was rather surprising that target binding to the probe with as few as 6 nucleotides was observed, the experiment showed that probes with at least 7 nucleotides were required to obtain signals in a workable range. Sequences of probes are shown in Figure 8(b).

8.2. Effect of lengthening a probe at each end by surrounding the central 9 nucleotides of defined base sequence with 1, 2, 3 or 4 nucleotides of random sequence.

While the number of defined bases in each probe must be kept small to minimise the number of probes required for sequencing the methylation status of-cytosine bases in all contexts, the Tm of each short probe may be increased by effectively lengthening the probe with "undefined" bases. This may be done in a number of ways. Li one method, we have made a series of probes in which the base sequence of the central 9 nucleotides are defined, and have added at each end either 1, 2, 3 or 4 nucleotides taken from an equal mix of the four normal bases, A, G, C or T; this effectively makes probes with 11, 13, 15 and 17 nucleotides, respectively. Probes LP 13 and LP 15 were chosen as the base probes for the series because of the significant level of mis-pairing of the 9-mer probes. Methylated target LINE9, prepared and fragmented as described in Example 2.2, was hybridised to the probes, the slides washed in hybridisation solution at room temperature, and scanned. Figure 9(a) shows correct binding of the fragmented, methylated target to the series of probes, LP 13Xn, of increasing length, where a G-C base-pair is formed at the interrogated site.Increasing the probe length from 11 (X2) to 17 (X8) nucleotides does not appear to substantially affect the amount of target bound. Figure 9(b) shows binding of the target to the series of probes LP 15Xn, of increasing length, in which a single, central, G-T mispair is formed. Optimal specificity of binding is observed for the LP13X2/LP15X2 probe pair with hybridisation and washing at room temperature. Such 1 lmer probes comprising central defined 9 bases flanked by one random nucleotide at each end were found to work well under a variety of conditions for various probe sequences (data not shown).

8.3 Other parameters investigated

Extending the length of probe by keeping the number of defined bases constant and adding various numbers of the "universal" base 5-nitro-indole produced an effect on binding strength but little effect on specificity. A single 5-nitro-indole introduced at both ends of the probe caused the Tm to increase, and the addition of a second universal base at both ends caused the Tm to further increase slightly, but little effect was observed on specificity of binding (data not shown). Inserting a spacer chain of 18 carbon atonis between the surface of the chip and the defined sequence had little effect on target binding to probes (data not shown).

Target was found to bind with equal preference to probe connected to the chip through either the 5' end or the 3' end (data not shown), and switching the target sequence to the probe and probe sequence to the target had no effect on target binding (data not shown).

Example 9: Investigating the effect of modified nucleotides in probes on increasing the strength and specificity of target binding.

9.1 Certain modified nucleotides in probes confer increased specificity in base pairing when binding to an 85-nt methylated target, compared with normal nucleotides in probes.

Sets of probes were synthesised with the same base sequence, but with various modifications to specific nucleotides, in order to determine which modifications best reduced mis-pairing, particularly the troublesome G-T mis-pair, when probes were bound by longer targets. An 85-nt methylated target labelled with 5'-Cy5, LTM85, was hybridised as described in Figure 10 and scanned on a Phosphorlmager.

In the images shown for Example 9 (Figures 10 to 13), probes in the same line have the same base sequence but have modifications introduced as indicated by the label at the top of each column. For example, the probe in line LPl and column DNA consists of normal DNA and has the base sequence 5' TTTTAGCGT, while the probe in line LPl and column OU is probe LPlOU which has the same sequence as LPl but with all T replaced by 2'-O-methyl-U, 5' UUUUAGCGU. Positions of probes which should bind correctly to the methylated target LTM85 to form G-C pair(s) at the interrogated site(s) are circled. Spots appearing in other places indicate G-T mis-pairing on target binding. Each probe has been spotted as an adjacent pair.

9.1.1 Probes with 2 '-O-methyl nucleotides

Three probe sets with different sequences, LP1/2, LP3/4, and LP13/14/15/16, were selected to test the effect on fidelity of binding, and Tm, of nucleotides bearing - 0-CH₃ groups on the C2' atom of the sugar moiety. Sets of these probes were made in which all T were replaced by 2'-O-methyl-U ("OU" probes), all A were replaced by 2'- O-methyl-A ("OA" probes), all A and all T were replaced by 2'-O-methyl-A and 2'-O- methyl-U, respectively ("OAU" probes), and, for LP3/4 only, all nucleotides were replaced by their 2'-O-methyl equivalents ("OaIl" probes). The effect of target LTM85 binding to these probes with modified nucleotides was compared with the equivalent normal DNA probes. All probes had 9 nucleotides.

All combinations of 2'-O-methyl-nucleotides in the three probe sets tested resulted in reduced G-T mis-pairing with the methylated target, compared with probes containing normal DNA, as shown by the reduction in intensities of spots not circled in columns "OU", "OA", "OAU" and "OaIl", compared with "DNA" in the same row, in Figure 10(b).

The Tms for correct binding of target to probes (circled spots in Figure 10(b)) were affected for modifications of the types "OU" and "OAU" to varying degrees depending on the sequence. The Tm for probe LPlOU (with five 2'-0-methyl-U, including four consecutive Us) was greatly reduced relative to LPl with normal DNA, while the Tm for probe LP30U (with three 2.'-O-methyl-U, including two consecutive Us) was reduced to a slightly lesser extent compared with LP3. Since the Tm of LP 13 OU (with one 2'-O-methyl-U) was similar to that of LP 13, it appears that three or more 2'-O-methyl-U substitutions in 9-nt probes will reduce strength of binding with the target (under the experimental conditions used). Interestingly, the reduction in Tm by 2'- O-methyl-U was compensated by other 2'-O-methyl modifications, as the spot intensities for probes with 2'-O-methyl-A combined with 2'-0-methyl-U were greater than their respective partners with solely 2'-O-methyl-U modifications (compare LPlOAU with LPlOU, LP30AU with LP3OU), and, when probes included 2'-O- methyl-G/C, so that all nucleotides were modified, the Tm was raised further (compare LP3Oall with LP30AU and LP30U in Figure 10(b)). For all three sets of sequences, probes "OA", with all A replaced by 2'-O- methyl-A, produced the desired effect of improving the specificity of binding while not affecting Tm. Importantly, the replacement of A by 2'-O-methyl-A in the 9-nt probes conferred correct binding with a target of 85-nt, while probes with normal DNA mis- paired with the same target. Thus, longer targets may be used on DNA chips containing probes with 2'-0-methyl-A, than on DNA chips containing probes with normal DNA.

9.1.2 Probes with 2-amino-adenine, and 2 '-O-methyl-2-amino-adenine

Three probe sets with different sequences, LP 1/2, LP3/4, and LP 17/18/19/20, were selected to test the effect on fidelity of binding, and Tm, of nucleotides bearing T- deoxy-2-amino-A in place of A ("D" probes), and two probe sets, LP1/2 and LP3/4, were tested with 2'-O-methyl-2-amino-A in place of A ("M" probes). An amine group on the C2 atom of adenine allows the modified base to form three hydrogen bonds with thymine, which, in principle, should result in the Tm of (2-amino-A)-T base pairs being approximately equivalent to those of G-C base pairs. The effect of target LTM85 binding to these probes with modified nucleotides was compared with the equivalent normal DNA probes. All probes had 9 nucleotides.

Probe LPlD, LP3D and LP17D have, respectively, one, two and three 2-amino- A nucleotides. The relative increase in Tm for the "D" probes compared with their equivalent unmodified "DNA" probes was greater for LP 17 and LPl than for LP3 (compare spot intensities for LP17D with LP 17, LPlD with LPl, and LP3D with LP3, in Figure 1 l(a)). Thus, an increase in Tm is dependent either on the sequence flanking the 2-amino-A and/or on the location of the 2-amino-A in the probe (in LP17D there are 2- amino-A at each end of the probe) more than on the number of 2-amino-As in the probe. AIl three probe sets containing 2-amino-A also showed reduced G-T mis-pairing with the 85-nt target, compared with all-DNA probes (in Figure 1 l(a), compare spots not circled in column "D" with spots not circled in column "DNA" on the same row). Thus, probes with A replaced by 2-amino-A show both increased binding strength and increased discrimination with normal DNA targets, compared with probes of normal DNA.

2' deoxy-2-amino-A and 2'-O-methyl-A, when used separately in 9-nt probes, provided good specificity for binding with longer targets, compared with probes of normal DNA which show some non-specific binding. However, when these two modifications to the adenine nucleotide were used in conjunction as 2'-O-methyl-2- amino-A, the Tm appeared to be reduced (compare LPlM with LPlD, and LP3M with LP3D in Figure 11 (a)).

9.1.3 Probes with 2 '-O-methyl-nucleotides, in combination with universal bases at either end A universal base is a moiety that can bind equally well to all four bases of DNA.

5-nitro-indole was selected to test if a universal base placed at each end of a probe could effectively increase the Tm of a probe, while preserving the number of the defined bases. In Example 8.3 5-nitro-indole was shown to be effective in increasing the Tm of probes with normal DNA, but without affecting specificity of target binding. Here, 5-nitro- indole is used in combination with 2 '-O-methyl-nucleotides to test if both Tm and specificity for probes may increase.

Sequences of probes:

"DNA" probes with normal DNA nucleotides LP3 5' TTAGCGTGA-NH₂ 3'

LP4 5 ' TTAGTGTGA-NH₂ 3 '

LP 13 5 ' TGAGCGACG-NH₂ 3 '

LP 15 5 ' TGAGTGACG-NH₂ 3 ' "OU" probes in which all T are replaced by 2'-O-methyl-U, as indicated by U LP13OU 5' UGAGCGACG-NH₂ 3' LP 15OU 5' UGAGUGACG-NH₂ 3' ^•

"OA" probes in which all A are replaced by 2'-O-methyl-A, as indicated by A LP13OA 5' TGAGCGACG-NH₂ 3' LP15OA 5' TGAGTGACG-NH₂ 3' "OAU" probes in which all A and all U are replaced by 2'-O-methyl-A and 2'-O- methyl-U, as indicated by A and U, respectively. LP30AU 5 ' UUAGCGUGA-NH7. 3 ' LP40AU 5 ' UUAGUGUGA-NH? 3 ' LP130AU 5' UGAGCGACG-NH₂ 3' LP 150AU 5 ' UGAGUGACG-NH₂ 3 '

"V2" probes with normal DNA nucleotides, and one universal base at both the 5' and 3' ends, indicated by V. LP3V2 5' VTTAGCGTGAV-NH₂ 3'

LP4V2 5 ' VTTAGTGTGAV-NH₂ 3 '

"V4" probes with normal DNA nucleotides, and two universal bases at both the 5' and 3' ends, indicated by VV. LP3V4 5' VVTTAGCGTGAVV-NH₂ 3'

LP4V4 5 ' VVTTAGTGTGAVV-NH₂ 3 '

"OUV4" probes with two universal bases at both the 5' and 3' ends, indicated by VV, and in which all T are replaced by 2'-O-methyl-U, as indicated by U LP13OUV4 5' VVUGAGCGACGVV-NH_? 3' LP 15OUV4 5 ' WUGAGUGACGVV-NH₇ 3 '

"0AV4" probes with two universal bases at both the 5' and 3' ends, indicated by VV, and in which all A are replaced by 2'-O-methyl-A, as indicated by A LP13OAV4 5' VVTGAGCGACGVV-NH₂ 3' LP 15OAV4 5 ' VVTGAGTGACGVV-NH₂ 3 '

"OAUV4" probes with two universal bases at both the 5' and 3' ends, indicated by VV, and in which all A and all U are replaced by 2'-O-methyl-A and 2'-O-methyl-U, as indicated by A and U, respectively.

LP3OAUV4 5' WUUAGCGUGAVV-NH? 3' LP4OAUV4 5' WUUAGUGUGAVV-NH₇. 3' LP13OAUV4 5' WUGAGCGACGW-NH₇ 3' LP15OAUV4 5' WUGAGUGACGVV-NH₇ 3' When one ("V2") or two ("V4") universal bases were added to each end of a normal DNA probe to effectively lengthen the probe, the binding strength between target LTM85 and the probes increased when correct base pairs formed (in Figure 12(a)) compare the circled spots for LP3"DNA" with LP3V2 and LP3V4), but also when G-T mis-pairs formed (in Figure 12(a) compare intensities of uncircled spots for LP4"DNA", LP4V2 and LP4V4).

The addition of two universal bases to each end of probes bearing certain 2'-O- methyl-nucleotides also increased the binding strength between target LTM85 and probes when correct base pairs formed (in Figure 12, compare LP3OAU with LP3OAUV4 in (a), and LP13OU with LP13OUV4, LP13OA with LP13OAV4, and LPl 3OAU with LP13OAUV4 in (b)). While G-T mis-pairing was enhanced for probes bearing 2'-0-methyl-U modifications in combination with universal bases at each end (in Figure 12)(b) compare LP15OU with LP15OUV4), such mis-pairing was not seen for probes bearing universal bases together with either 2'-O-methyl-A or 2'-0-methyl-A in combination with 2'-O-methyl-U (in Figure 12 compare LP4OAU with LP4OAUV4 in (a), and LP15OA with LP15OAV4, and LP15OAU with LP15OAUV4 in (b)).

Thus, the use of universal bases at the ends of probes, combined with internally placed 2'-O-methyl-A (either alone (OAV4), or in combination with 2'-O-methyl-U (OAUV4)), provides good fidelity, and strength, for target binding. 9.1.4 Probes with 5-propynyl-U and 5-propynyl-C

When G mis-pairs with T, the T base slides forward into the major groove of the double helix in order to be optimally positioned for formation of hydrogen bonds between the bases. Replacing the methyl group at the C5 atom of the T base with a more hydrophobic group should make the forward motion of the T base into the water-filled major groove less energetically favourable, in principle, and thus reduce G-T mis- pairing. Two probe sets with different sequences, LP3/4, and LP 17/18/19/20, were selected to test the effect on fidelity of binding, and Tm, of nucleotides bearing propynyl groups at the C5 atoms of the pyrimidine bases, C and T. A set of probes ("Pall") was made in which all C and T were replaced by 5-propynyl-C and 5-propynyl-U, respectively, and a second set ("P") contained 5-propynyl-C and 5-propynyl-U only at the interrogation sites. The effect of target LTM85 binding to these probes with modified nucleotides was compared with the equivalent normal DNA probes. All probes had 9 nucleotides.

Sequences of probes:

"DNA" probes with normal DNA nucleotides

LP3 5' TTAGCGTGA-NH2 3'

LP4 5' TTAGTGTGA-NH2 3'

LP 17 5 ' AGCGACGTA-NH2 3 ' LP18 5ΑGTGATGTA-NH2 3'

LP 19 5' AGTGACGT A-NH2 3'

LP20 5' AGCGATGTA-NH2 3'

"P" probes in which the interrogating C and T are replaced by 5-propynyl-C and 5- propynyl-U, as indicated by C and U, respectively LP3P 5' TTAGCGTGA-NH2 3' LP4P 5' TTAGUGTGA-NH2 3' LP 17P 5 ' AGCGACGT A-NH2 3 '

LP 18P 5 ' AGUGAUGTA-NH2 3 '

LP 19P 5 ' AGUGACGT A-NH2 3 '

LP20P 5 ' AGCGAUGTA-NH2 3 '

"Pall" probes in which all C and T are replaced by 5-propynyl-C and 5-propynyl-U, as indicated by C and U, respectively

LP3Pall 5 ' UUAGCGUGA-NH2 3 '

LP4Pall 5 ' UUAGUGUGA-NH2 3 ' LP 17PaIl 5' AGCGACGUA-NH2 3'

LP 18PaIl 5 ' AGUGAUGUA-NH2 3 '

LP 19PaIl 5 ' AGUGACGUA-NH2 3 '

LP20Pall 5 ' AGCGAUGUA-NH2 3 '

The introduction of the hydrophobic propynyl group on the interrogating C base in probes LP3 did not diminish the strength of binding by the methylated target LTM85, as shown in Figure 13 (compare the circled spots for LP3P with LP3). With the propynyl group on the interrogating T base in probe LP4 G-T mis-pairing was-reduced (in Figure 13, compare uncircled spots for LP4P with LP4). For the LP17 set of probes the inclusion of propynyl-C and propynyl-U at the interrogating position did not improve discrimination compared with normal DNA probes. When all C and T bases in each probe were replaced by 5-propynyl-C and 5-propynyl-U, the correct binding of LTM85 with probes LP3 and LP 17 was diminished compared with the unmodified DNA probes (in Figure 13 compare LP3Pall with LP3, and LP 17PaIl with LP 17), while G-T mis- pairing was slightly reduced for some probes (compare LP 19PaIl with LP 19) but increased for others (compare LP 18PaIl with LP 18).

Thus, 5-propynyl groups on the interrogating bases of probes can assist the fidelity of binding by the target in a sequence dependent manner, but inclusion in all pyrimidine bases of the probe is disadvantageous.

Example 10. Quantification of intensities on the DNA chips The human eye and brain can very effectively integrate the information contained in images of microarrays, as shown in Example 9. As an alternative method of presenting the data, the intensities of spots in the images have been quantified and presented in tables for numerical scrutiny.

Fragmented methylated target LINE9 and unmethylated target LINE5 (prepared as in Example 2.3 with USER enzyme) were hybridised for 2 hours at room temperature to probes on the DNA chip in 1.7M TMA-formate, 5OmM Tris-HCl, pH 7.5, 5mM EDTA, 0.02% Tween-20, 0.04% SDS, and washed twice at room temperature in the same solution (without targets). The slides were scanned on a Phosphorlmager.

The amount of target bound to each probe was determined as follows. The raw fluorescence at each probe spot was measured on the Phosphorlmager by recording the intensity within a circle drawn tightly around the spot. The background was measured at four positions adjacent to the probe spot. The background counts were averaged, and subtracted from the raw intensity. The background-corrected intensities for identical probes were averaged, and these are reported in Figures 14(c) and (f). Standard deviations are given in parentheses. The arrangement in which the probes were spotted on the DNA chip is indicated in Figure 14(a). Probes LP3 and LP4 interrogate the methylation status of a single CpG site, where the methylated target should bind to LP3 and the unmethylated target should bind to LP4. LP13, LP14, LP15 and LP16 interrogate the methylation status of two closely-spaced CpG sites, where the fully methylated target should bind to LP 13, the fully unmethylated target should bind to LP 14, and the target in which one CpG is methylated and the other not should bind sequence-specifically to LP 15 or, for the alternative combination, to LP 16. Likewise, probes LP21, LP22, LP23 and LP24 interrogate two closely-spaced CpG sites, with LP21 being specific for the fully methylated target, and LP22 for the fully unmethylated target. LP61 and LP62 interrogate a single CpG site, where the methylated target should bind to LP61 and the unmethylated target to LP62. To aid understanding, the probes which bind fully methylated targets are colour-coded yellow, and those which bind fully unmethylated targets are colour-coded green, in Figure 14(a).

Probe sequences, in which X is an equal mix of A, G, C and T; D is 2-amino-A; A is T- O-methyl-A; T is 2'-O-methyl-T:

LP3X2 5 ' XTTAGCGTGAX-NH₂ 3 '

LP4X2 5 ' XTTAGTGTGAX-NH₂ 3 '

LP3DX2 5 ' XTTDGCGTGDX-NH2 3 ' LP4DX2 5' XTTDGTGTGDX-NH2 3'

LP3OAX2 5 ' XTTAGCGTGAX-NH₂ 3 '

LP4O AX2 5 ' XTTAGTGTGAX-NH₂ 3 '

LP4OAT1X2 5 ' XTTAGTGTGAX-NH2 3 ' LP13X2 5' XTGAGCGACGX-NH₂ S'

LP14X2 5' XTGAGTGATGX-NH₂ 3'

LP 15X2 5 ' XTGAGTGACGX-NH₂ 3 ' LP 16X2 5' XTGAGCGATGX-NH₂ 3'

LP13DX2 5' XTGDGCGDCGX-NH2 3'

LP14DX2 5' XTGDGTGDTGX-NH2 3'

LP15DX2 5' XTGDGTGDCGX-NH2 3'

LP16DX2 5' XTGDGCGDTGX-NH2 3'

LP13OAX2 5' XTGAGCGACGX-NH₂ 3'

LP14OAX2 5' XTGAGTGATGX-NH₂ 3'

LP15OAX2 5' XTGAGTGACGX-NH₂ 3'

LP16OAX2 5' XTGAGCGATGX-NH₂ 3'

LP14OAT2X2 5' XTGAGTGATGX-NH2 3'

LP15OAT1X2 5' XTGAGTGACGX-NH2 3'

LP16OAT1X2 5' XTGAGCGATGX-NH2 3'

LP21X2 5' XGCGACGTAGX-NH₂ 3'

LP22X2 5' XGTGATGTAGX-NH₂ 3'

LP23X2 5' XGCGATGTAGX-NH₂ 3'

LP24X2 5' XGTGACGTAGX-NH₂ 3'

LP21DX2 5' XGCGDCGTDGX-NH2 3'

LP22DX2 5' XGTGDTGTDGX-NH2 3'

LP23DX2 5' XGCGDTGTDGX-NH2 3'

LP24DX2 5' XGTGDCGTDGX-NH2 3'

LP21OAX2 5' XGCGACGTAGX-NH₂ 3'

LP22OAX2 5' XGTGATGTAGX-NH₂ 3'

LP23OAX2 5' XGCGATGTAGX-NH₂ 3'

LP24OAX2 5' XGTGACGTAGX-NH₂ 3'

LP22OAT2X2 5' XGTGATGT AGX-NH2 3'

LP23OAT1X2 5' XGCGATGTAGX-NH2 3'

LP24OAT1X2 5' XGTGACGTAGX-NH2 3'

LP61X2 5' XGTATCGGGTX-NH₂ 3'

LP62X2 5' XGTATTGGGTX-NH₂ 3'

LP61DX2 5' XGTDTCGGGTX-NH2 3'

LP62DX2 5' XGTDTTGGGTX-NH2 3'

LP61OAX2 5' XGTATCGGGTX-NH₂ 3'

LP62OAX2 5' XGTATTGGGTX-NH₂ 3'

LP62OAT1X2 5' XGTATTGGGTX-NH2 3'

The amount of methylated, control target, LINE9, which is bound to the probes on the DNA chip is shown visually in Figure 14(b), and is quantified in Figure 14(c). The amount of unmethylated, control target, LINE5, which is bound to the probes on the DNA chips is shown visually in Figure 14(e) and is quantified in Figure 14(f). Figures 14(d) and (g) reveal the intensity of target binding to a particular probe, relative to the intensity of target binding to the correct probe (which would give Watson-Crick base pairs), using intensity data from Figures 14(c) and (f) respectively. 10.1 Methylated target LINE9 For example, the methylated target, LINE9, should bind to all probes with the LP3 generic name, forming a central G-C base pair. As shown by the circled spots in the top line of Figure 14(b), all LP3 probes (LP3X2, LP3DX2, LP3OAX2, and LP3OAX2 repeated) are indeed bound by the methylated target. The averaged intensities for each of these circled pairs of spots, corrected for background, are given in the top row of Figure 14(c). Since the methylated target forms correct base pairs with the LP3 probes, the intensities of target binding for these probes are compared with themselves to generate the data in Figure 14(d), and so the relative intensities for probes LP3 in the top line of Figure 14(d) are all 1.0. Probe LP4 has the same sequence as LP3 except for a central base change from

C (in LP3) to T (in LP4); thus any binding by the fully methylated control target, LINE9, to any probe LP4 indicates the extent of G-T mis-pairing for this LP3/LP4 probe pair. As shown by the pairs of spots in the second row of Figure 14(b), fragmented LINE9 binds strongly to LP4 when it contains normal DNA nucleotides (first column), indicating a strong G-T mis-pair is formed. The strength of binding is reduced for probes LP4DX2 (where all A are 2-amino-A, second column)) and LP4OAX2 (where all A are 2'-O- methyl-A, third column), and G-T mis-pairing is almost eliminated when the probe has all A as 2'-O-methyl-A combined with the interrogating T as 2'-O-methyl-T (probe LP4OAT1X2, fourth column, second row, of Figure 14(b)). The averaged intensities of the spot pairs for the LP4 probes imaged in the second row of Figure 14(b) are given in the second row of Figure 14(c). The intensity for target LINE9 incorrectly binding the normal DNA probe LP4X2 is 372 (first column), for incorrectly binding LP4DX2 is 101, for LP4OAX2 is 50, and for LP4OAT1X2 is 8. The intensity of each incorrectly bound LP4 probe, relative to the intensity of the correctly bound LP3 probe with the same modifications, is given in the second row of Figure 14(d). Thus, LINE9 incorrectly binds LP4X2 by a factor of 372/354 = 1.05, compared to correct binding to LP3X2; in other words, the G-T mis- pairing by LINE9 is 5% greater than correct binding to normal DNA probes of 11 nucleotides. LINE9 incorrectly binds LP4DX2 by a factor of 101/335 = 0.3, compared to correct binding to LP3DX2; thus, under the experimental conditions presented here, G-T mis-pairing is at 30% the level of correct pairing, when A are present as 2-amino-A in probes LP3DX2/LP4DX2. Similarly, G-T mis-pairing is 50/287 = 0.17 for LP4OAX2 compared with LP3OAX2 (where A are present as 2'-O-methyl-A). For the LP3/LP4 sequence, G-T mis-pairing is reduced to 3% when all A are 2'-0-methyl-A and the interrogated T is 2'-O~methyl-T, as shown by very weak incorrect binding of LINE9 to probe LP4OAT1X2 compared with correct binding to LP3OAX2 (8/287 = 0.03).

The third row in each of Figures 14(b), (c) and (d) shows correct binding of methylated target LINE9 to probes with the generic name and sequence of LP13.

Intensities in the fourth (LP 14), fifth (LP 15) and sixth (LP 16) rows of Figure 14(c) are compared to those in the third row (LP 13) to calculate the relative intensities of the G-T mis-pairs given in Figure 14(d). For the LP 13/LP 14/LP 15/LP 16 probe set, which interrogates two closely-spaced CpG sites, probes of the type 0AX2 (all A are 2'-O- methyl- A) and O ATX2 (all A are 2 ' -O-methyl- A with interrogated T as 2 ' -O-methyl-T) confer good mis-pair discrimination compared with probes of normal DNA, or containing 2-amino-A.

The seventh row in each of Figures 14(b), (c) and (d) shows correct binding of methylated target LINE9 to probes with the generic name and sequence of LP21. The set of probes LP21/LP22/LP23/LP24 interrogates two closely-spaced CpG sites, and rows eight, nine and ten in Figure 14(d) indicate the level of mis-pairing by LINE9 to probes LP22, LP23 and LP24, relative to correct binding to probe LP22. For the sequence of this probe set, all probes with modified nucleotides (DX2, 0AX2, and OATX2) show good mis-pair discrimination relative to probes of normal DNA. The last two rows of Figure 14(b) and (c) show very strong and correct binding to generic probes LP61. There is also a single, very strong G-T mis-pair formed with probes LP62X2 (normal DNA, 935/2125 = 0.44 relative to LP61X2) and LP62DX2 (2- amino-A, 784/1199 = 0.65 relative to LP61DX2). The G-T mis-pair is reduced to 15% (LP62OAX2/LP61OAX2 = 192/1262 = 0.15) when all A in probes are 2 '-O-methyl- A, and is negligible (1%) when all A in probes are 2'-O-methyl-A and interrogating T are 2'-O-methyl-T (LP62OAT1X2/LP61OAX2 = 14/1262 = 0.01).

Thus, modifications of either all A as 2 '-O-methyl- A, or else all A as 2'-O- methyl-A combined with interrogated T as 2 '-O-methyl-T, were successful in conferring excellent fidelity and strength of binding by a methylated target to probe sequences represented in the sets LP3/LP4, LP 13/LP 14/LP 15/LP 16, LP21/LP22/LP23/LP24, and LP61/LP62.

10.2 Unmethylated target, LINE5 The unmethylated target, LINE5, should bind to probes LP4, forming a central A-T base pair. As shown by the circled spots in the second line of Figure 14(e), all LP4 probes (LP4X2, LP4DX2, LP4OAX2 and LP4OAT1X2) are bound by the unmethylated target, and there is minimal mis-pairing to analogous probes (LP3) in the row immediately above; in this case any mis-pairs of LINE5 with LP3 probes would be of the A-C type. The averaged intensities for each LP3 and LP4 probes are given in the first and second rows, respectively, of Figure 14(f), and the first row of Figure 14(g) shows the intensities of the LP3 probes relative to those of their matched LP4 probes. The extent of A-C mis-pairing by the unmethylated target to LP3 probes, relative to correct binding to the LP4 probes, is less than 5% for all probes.

The set of probes LP 13/LP 14/LP 15/LP 16 interrogates two closely spaced CpG sites, and correct binding by the unmethylated target is to the LP 14 probes (in the fourth rows of Figures 14(e) and (f)). Thus, incorrect binding of the unmethylated target LINE5 to probes LP 13, LP 15 and LP 16 is given in rows three (immediately above LP 14), five and six (the two rows below LP 14) in Figures 14(e), (f) and (g). The quantified data in Figure 14(g) support the visual results in Figure 14(e) indicating that probes of the type 0AX2 and OATX2 confer fidelity of binding (for this set all probes, including normal DNA, show less than 3% mis-pairing) by unmethylated targets to probes bearing the sequence of this set. Similarly, the intensities of binding by unmethylated target LINE5 to probes

LP22 (row eight) with A-C mis-pairing to probes LP21, LP23, and LP24 (rows seven, nine and ten), and correct binding to probes LP62 (bottom row) with mis-pairing to LP61 (second bottom row) can be seen in Figures 14(e) and (f), with relative bindings in Figure 14(g). For both sets of probe sequences, all probes confer good discrimination, but the Tm for correct binding by LINE5 to LP22OAT2X2 is drastically reduced when the interrogating T are 2'-O-methyl-T (compare intensity for LP22OAT2X2 with LP22OAX2 in Figures 14(e) or (f)). The low intensity observed for target binding correctly to LP22OAT2X2, combined with minimal target mispairing to probes LP21OAX2, LP23OAT1X2, and LP24OAT1X2, make the relative intensities for these probes less reliable than for other probes in Figure 14(g).

Thus, probes containing either normal DNA, or all A as 2-amino-A, or all A as 2'-O-methyl-A, confer good discrimination and strength in binding to the unmethylated target LINE5. Probes with all A as 2'-O-metliyl-A (but not combined with interrogating T as 2'-O-methyl-T) are preferred for probes with sequences similar to LP22.

Example 11. Solution melting temperatures of duplexes containing modified nucleotides 11-mer or 13-mer DNA target oligonucleotides corresponding to 11 or 13-mer probe DNAs were annealed (at 2 μM each) and the melting temperatures of duplexes determined by following SYBR Green fluorescence with increasing temperature in an ABI7700 real time PCR machine. Melting temperatures were measured in two buffers: "NaCl" buffer contained 0.1 M NaCl, 40 mM Tris, 10 mM EDTA, pH 8, 0.02% Tween 20, 0.04% SDS and 0.5 μM SYBR Green; "TMACl" contained 2.0 M tetramethylammonium chloride, 40 mM Tris, 10 mM EDTA, pH 8, 0.02% Tween 20, 0.04% SDS and 32 μM SYBR Green (a higher concentration of SYBR Green being required under the high salt conditions).

The synthetic targets, corresponding to sites C or D and E of the LINE-I sequence, were:

11-C-G XTCACGCTAAX, 11-C-A XTCACACTAAX, 11-ED-GG XCGTCGCTCAX, 13-C-G XXTCACGCTAAXX,

13-ED-GG XXCGTCGCTCAXX, 13-ED-GA XXCGTCACTCAXX where X is an equal mixture of all four bases A, G, C and T Each chemically-synthesized target (whether an 11-mer or a 13-mer) was studied in combination with six different synthetic probes of equal length. Three of those target- probe mixtures lead to normal G/C matched pairs, while the other three lead to G/T mismatches. The probes contained mixtures of all four bases (X) at the terminal one (11- mers) or two (13-mers) positions or equivalent probes containing the universal base 5- nitro-indole at these positions (indicated by V prefix in Table 1 below). Probes contained normal bases (N columns in Table 1), two 2 '-O-methyl- adenosines (OA or VA columns in Table 1), or two 2'-O-methyl-adenosines and one 2'-O-methyl-thymidine or 2'-O- methyl-uridine at their central step (OAT or VAU columns in Table 1). The VAU probes contained two 2'-O-methyl-adenosines and three or four 2'-O-methyl-uridines for 13-C- G partners, or one/two 2'-O-methyl-uridines for 13-ED-GG and 13-ED-GA partners.

In each cell is shown the melting temperature for the perfectly matched duplex (G/C or A/T) and for duplexes containing mismatches (G/T or A/C) in each buffer, NaCl or TMAC. Melting temperature differences between paired and mismatched sequences are shown (Δ).

Table 1. Double-helical DNA melting temperatures in degrees C as measured in solution for LINE-I targets and 11 or 13-mer probes using SYBR Green I

Target Salt N OA OAT

G/C G/T Δ G/C G/T Δ G/C G/T Δ

11 -C-G NaCI 40.3 26.9 13.4 37.7 23.6 14.1 37.7 15.0 23.0

TMACI 48.5 35.6 12.9 43.5 30.4 13.1 43.5 21.0 22.5

2 G/C 2 G/T Δ 2 G/C 2 G/T Δ 2 G/C 2 G/T Δ

11 -ED- NaCI 52.1 24.2 27.9 42.2 11.6 30.6 42.2 11.0 31.2

GG TMACI 53.0 28.5 24.5 45.5 17.5 28.0 45.5 14.0 31.5

Target Salt N OA OAT

A/T A/C Δ A/T A/C Δ A/T A/C Δ

11 -C-A NaCI 35.0 15.3 19.7 32.6 13.2 19 .4 33.0 13.2 19 .8

TMACI 44.4 28.1 16.3 40.6 21.7 18 .9 40.7 21.7 19 .0

Target Salt V VA VAU

G/C G/T Δ G/C G/T Δ G/C G/T Δ

13-C-G NaCI 46.4 31.4 15. 0 - - - 33.9 14.5 19.4

TMACI 52.7 39.0 13. 7 - - - 41.2 20.1 21.1

13-ED- NaCI 53.6 43.0 10 .6 49.7 33.2 16.5 44.4 22.0 22.0

GG TMACI 56.9 45.9 11 .0 51.4 37.4 14.0 47.6 30.0 17.6

Most modifications produced slight effects on the melting temperatures of fully- matched G/C or A/T double helices, as well as for mismatched G/T or A/C. Each 2'-O- methyl-adenosine lowered the Tm by about 3^°C, each 2'-O-methyl-thymidine had little effect, each 2'-O-methyl-uridine lowered the Tm by about 2 C, while each "V" or 5- nitro-indole added typically 2 C.

The inclusion of 2'0-methyl-adenosine along with 2'0-methyl- thymidine or 2'0-methyl-uridine typically increases the difference in melting temperature of the fully matched and mismatched duplexes. For example, for site C the inclusion of three 2'-O- methyl nucleotides increases the mismatch discrimination from about 13⁰C for normal DNA to about 22⁰C for the substituted DNA. A more modest difference is seen for the duplexes containing two mismatches at sites D and E with base mixtures at the termini, but the differences are significantly greater when the termini comprise the "universal" base 5-nitro-indole. The single base mismatch discrimination from hybridisation of 3- ED-GA with probes covering sites D and E is also significantly improved by inclusion of 2'-O-methyl adenosines alone or further in combination with 2'O-methyl uridine, hi some cases we have observed that the relative effect on mismatch discrimination is greater in the NaCl than the TMAC buffer. Mismatch discrimination of C-A base pairs is generally good and is not significantly affected by 2'-O-methyl substitution.

Example 12. Incorporation of modified bases in target DNA. DNA polymerases are able to incorporate a range of modified bases, including propynyl derivatives and 2-amino adenine, into DNA. Since improved specificity from incorporation of 2-amino adenine in probes was observed, the inventors investigated whether improved hybridisation specificity could be obtained through its incorporation during PCR into the target DNA. Target DNA was prepared from fully methylated DNA after bisulphite treatment by two rounds of PCR using primers targeted to the hMLHl gene (Figure 15(a)). In the bisulphite-treated sequence potential methylation sites are designated YG and those within the final amplicon are shown in bold and denoted A through to H. In the second round of amplification either 100% 2-amino dATP or 100% normal dATP was included in the PCR, along with a 1 :3 mix of dUTP:dTTP and 1 : 1 Cy5 dCTP:dCTP as in Example 2. After fragmentation with the USER enzyme mix target DNAs were hybridised with a microarray containing probes designed to CpG sites B, E, F and G of the hMLHl gene. Figure 15(b) shows hybridisation of probes to sites B, E, F and G (see Figure 15(a). Probe sequences are as follows: hBl 5' NATAGCGATTN hB2 NATAGTGATTN

hEl 5' NTAAGCGTATN hE2 NTAAGTGTATN

hFl 5' NGTAGCGGGTN hF2 NGTAGTGGGTN

hGl 5' NTAGTCGTTTN hG2 NTAGTTGTTTN

Probes in the second and third rows of each set of probes contain 2'0-methyl adenosine at the locations shown in bold and those in the third row additionally contain 2'0-methyl uridine in place of thymidine at the discriminating nucleotide position (underlined).

As shown in Figure 15(b), the top row of each set of probes contained normal DNA , with the left hand pair of spots corresponding to correct hybridisation with the methylated sequence and the right pair of spots G-T mispairing. For all four CpG sites significant improvement in specificity and in some cases also sensitivity, is seen when the target DNA contains 2-amino adenine. Reduced mispairing is seen for sites B and E for both normal and 2-amino adenine substituted target DNA. The additional inclusion of 2'O-methyl uridine at the interrogation site reduced mispairing somewhat for sites F and G.

Example 13. Hybridisation of modified probes with representing CpG sites in the TPEF gene.

The human TPEF gene (also known as TMEFF2) is frequently methylated in colorectal cancer. Nested PCR primers were designed to amplify a region of the TPEF promoter shown in Figure 16 (target sites for primers underlined). The CpG sites A to I within the amplicon are indicated and the cytosines that could be C or T after conversion labelled Y. TPEF sequences were amplified from fully methylated and unmethylated DNA by two rounds of PCR. Second round synthesis included dUTP and either Cy5- dUTP or biotin-dUTP and fragmented target DNA was prepared by USER cleavage as described in Example 2.

Targets were applied to microarrays at 2O⁰C followed by brief washing at 37⁰C as in Example 6. Results obtained using Cy5 fluorescence were very similar to those using biotin and strepatvidin R-phycoerythrin (Example 6).

TPEF site H was studied using four kinds of chemically-modified nucleotide as follows (modified nucleotides are shown in bold and discriminating nucleotides are underlined):

- (No modification) THl 5' NAGTTCGTTGN

TH2 5'NAGTTTGTTGN

OA (2'-0-methyl-adenosine)

THlOA 5' NAGTTCGTTGN

OAU (2'-0-methyl-adenosine plus 2'-O-methyl-uridine)

TH20AU 5' NAGTTUGTTGN

OC (2'-O-methyl-cytidine) THlOCl 5' NAGTTCGTTGN

One OT (2'-O-methyl-thymidine)

TH2OT1 5' NAGTTTGTTGN

Two OT (2'-O-methyl-thymidine)

TH2OTlot3 5' NAGTTTGTTGN TH2OTlot4 5' NAGTTTGTTGN TH2OTlot7 5' NAGTTTGTTGN Three OT (2'-O-methyl-thymidine)

TH2OTlot3ot4 5' NAGTTTGTTGN TH2OTlot4ot7 5' NAGTTTGTTGN TH2OTlot3ot7 5' NAGTTTGTTGN

The name of any probe tells both its PCR target site and also its kind of modification. For example "TH" means "TPEF site H, "1 or 2" means "cytosine or thymidine at CpG position", while "THlOA" means "TPEF site H-I, 2'-0-methyl- adenosine". Chemically-modified bases are shown in bold: "A" for 2'-O-methyl- adenosine, "C" for 2'-0-methyl-cytidine, "T" for 2'-O-methyl-thymidine, or "U" for T- O-methyl-uridine. Layouts of the probes on the arrays are shown in Figure 17(a). Correct base-pairing locations of the probes iii the hybridisation dot blots shown in Figure 17(b) and (c) have been outlined with thin rectangles, while incorrect base- pairing locations remain unlabelled.

Figure 17(b) shows results for hybridisation of the methylated TPEF target and quantification by densitometry. It can be seen that one 2'-O-methyl T produces a slight reduction of G-T mispairing from 0.70-0.76 for normal DNA to 0.36-0.38, while two T- O-methyl T produce a more substantial reduction to 0.07-0.22, and three a further reduction down to 0.01-0.13. Results for the unmethylated TPEF target are shown in Figure 17(c). A-C mispairing is less of a problem than for G-T. Most of the "OT" probes that reduced G-T mispairing in (a) still bind well with correct A-T pairing in (b), at relative levels of 0.68-1.05 for one or 0.44-0.92 for two 2'-O-methyl Ts. One probe containing three 2'-O-methyl Ts binds with a normal affinity of 0.47-0.62, while two others bind with reduced affinities of 0.12-0.28.

Analysis was extended to the neighbouring sites A and B and to site C using the biotin-labelled target (Figure 18). Probes sequences and their layouts on the arrays of Figure 18(b) are shown in Figure 18(a). Probes in the second row of the arrays fully match the unmethylated target DNA; probes in the top row match the methylated target and will have two mismatches with the unmethylated target. Chemically modified bases are shown in bold, and the discriminating bases are underlined. The bottom two rows of probes contain one mismatch with either the methylated or unmethylated probe. As shown in Figure 18(b) for sites A and B, for the unmethylated target there is minimal hybridisation with the methylated probes with two mismatches (top row) and mismatch hybridisation to probes with one mismatch is substantially reduced in the 2'O-methyl modified probes compared with normal DNA. The incorporation of 2'O-methyl nucleotides also reduces hybridisation of the fully matched target in this sequence context. In the right hand panel it can be seen that hybridisation of the methylated target with normal DNA probes gives a high level of mismatching with probes containing either single or double mismatches. Mismatch hybridisation is substantially reduced by the inclusion of 2'O-methyl nucleotides with a fifty-fold reduction for the probe containing three 2'O-methyl Ts; the relative specificity compared to binding to the correct unmethylated probes is also substantially improved (6-fold). For site C (Figure 18 (c) and (d)) similar effects are seen.

The data demonstrate that significant improvements in the specificity of hybridisation can be obtained through inclusion of 2'O-methyl adenosine, uridine and thymidine nucleotides in oligonucleotide probes in a variety of different sequence contexts.

Example 14. Testing the microarray using DNA from a human patient with colorectal cancer The microarray with oligonucleotide probes of the type "X-9-X" was used to determine if differences in methylation could be seen in DNA samples taken from an individual with colorectal cancer. The methylation status of DNA from the tumour, and from adjacent normal tissue, was compared for a multiply-repeated LINE sequence (long interspersed nuclear element). Demethylation of Ll (LINE) promoter sequences has been reported in tumour tissue, and also to a degree in normal adjacent tissue, from patients with colorectal cancer (Suter, et al., 2004).

The DNA was treated with bisulphite using a kit from Human Genetic Signatures. The reaction was for 4 hours at 75⁰C with heat pulses for 1 minute at 95°C every hour. 20 ng of bisulphite-converted sample was PCR-amplified incorporating a biotin label and with a 1 :3 ratio of dUTP:dTTP and then fragmented with USER enzyme as in Example 2.3. Fragmented DNA was denatured by heating and hybridisation to probes was done in 5OmM Tris HCl, pH 7.5, 5mM EDTA, pH 8, 2.0M TMA-Cl, 0.01% Tween 20, 0.05% SDS at room temperature for one hour. After 3 washes at room temperature, bound target was detected using streptavidin-R-phyco (Example 6)

Spot intensities were quantified by measuring the intensity within a circle drawn tightly around each spot. Background was measured at four positions adjacent to each probe spot. The background counts were averaged, and subtracted from the spot intensity. Background corrected intensities for identical probes were averaged. The averaged intensity for target DNA binding to fully-methylated probes was divided by the averaged intensity for target DNA binding to the matched fully-unmethylated probes to obtain the ratio of methylated:unmethylated binding for each probe pair; this was done for DNA extracted from both the normal tissue and the tumour tissue samples. Then, the ratio of methylated:unmethylated probe binding by DNA extracted from the normal tissue was divided by the ratio of methylated:unmethylated probe binding by DNA extracted from the tumour tissue, for each probe pair, to determine the relative differences in probe-pair binding by bisulphite-treated DNA extracted from normal compared with tumour tissue.

Figure 19 shows the binding of the bisulphite-treated, PCR-amplified, and fragmented LINE target to LINE-specific probes on the microarray, where the probes in the top line of each image are specific for fully-methylated target DNA, and probes in the bottom line are specific for fuUy-unmethylated DNA; specifically, the probes have the generic base sequence of (a) LP 13 (top) and LP 14 (bottom), (b) LP21 (top) and LP22 (bottom), and (c) LPBl (top) and LPB2 (bottom). Each pair of images shows the extent of target DNA binding to the matched probes; the ratio of methylated:unmethylated binding for each probe pair is given underneath each image, along with the ratio for normal:tumour samples in bold-italics. Ratios of methylated:unmethylated >1.0 indicate that more target DNA binds the methyl-specific probe, and ratios of normal:tumour > 1.0 indicate that DNA from the normal sample is more methylated than DNA from the tumour sample. Data for target DNA binding to probes LP13/LP14 (a), LP21/LP22 (b), and

LPB1/LPB2 (c), all indicate that LINE DNA extracted from the normal sample is more methylated than LINE DNA extracted from the tumour sample, for the selected sequence region in this patient. Our microarray data, showing an increase in demethylation of C at CpG sites in LINE in tumour tissue compared with matched normal tissue, are consistent with previous reports.

In each of Figures 19(a), (b) and (c), pairs of images are shown for target binding to probes with the same base sequence but containing modified nucleotides. The increased values of the normal:tumour ratios for probe pairs containing modified nucleotides, compared with normal:tumour ratios for normal-DNA probe pairs of the same sequence, indicate that the probes with these modified nucleotides better discriminate the differences in DNA methylation between tumour and matched normal samples than probes of normal DNA nucleotides.

Example 15. Detecting methylation of the TPEF gene in colorectal cancer DNA.

Cy5 and biotin labelled PCR products' from bisulphite treated DNAs were prepared from control methylated and unmethylated DNAs as well as from matched tumour and normal DNA from patients with colorectal cancer. Methylation at specific sites within these DNAs were analysed using the restriction enzymes HpyCH4IV and BstUI that cut at ACGT and CGCG respectively — these correspond to sites I and A/B respectively in the TPEF sequence (Figure 16). Cutting will occur at these sites only if the CpG site was methylated in the original DNA and the C is retained after bisulphite treatment. Digests of Cy5-labelled amplicons in Figure 20(a) show the detection of methylated DNA in tumour samples 2, 3 and 4 in comparison to the unmethylated DNA from adjacent normal tissue.

Biotin-labelled Target DNA prepared from four patients was hybridised to an array of probes for detection of methylation at CpG sites A and B. The sequences and layout of these probes, that contained various modified nucleotides are shown in Figure 20(b). The positions of the interrogated sites are underlined and positions of modifications shown in bold. The left hand panel of Figure 20(c) shows hybridisation with control unmethylated DNA and with the DNA from non-diseased tissue of four patients (patients 2 to 5 from Figure 20(a)). Probes that match fully with unmethylated target DNA are boxed. It can be seen that the pattern of hybridisation is essentially the same for unmethylated DNA and DNAs and DNA from normal tissue. Mispairing to probes of normal DNA containing single mismatches is evident in the lower two rows of the left columns. This is essentially eliminated for probes containing 2'O-methyl uridine or thymidine.

The right hand panel of Figure 20(c) shows hybridisation with a fully methylated control target or with tumour DNA from four patients. Probes on the top row and right hand side of the array should match with fully methylated DNA. As well as normal DNA bases, these contain a range of modified nucleotides as indicated. Hybridisation is similar in all cases and there is minimal mispairing of these probes with the unmethylated target DNA (left side). The boxed probes on the lower left contain single mismatches to fully methylated DNA and show considerable mispairing with unmethylated target. Incorporation of 2'O-methyl nucleotides, particularly three 2'O- methyl thymidines substantially reduces this mispairing. The same is true for mispairing with the doubly-mismatching probes for unmethylated DNA (second row, left side of arrays). Inspection of the tumour DNAs shows significant hybridisation with probes detecting methylated DNA for the first three patients. No methylation is evident for the fourth patient. The relative intensity of the spots is consistent with the DNAs from patient tumours containing a mix of methylated and unmethylated sequences. References:

Clark, SJ., et al. (1994) Nucleic Acids Res., 22, 2990-2997. Ebert, M.P.A., et al. (2005) Neoplasia 7, 771-778. Frommer, M., et al. (1992). Proc. Natl. Acad. ScL USA, 89, 1827-1831. Germani, A., et al. (2004) . J. Agric. Food Chem. 52, 4535-4540. Melchoir, W.B. and von Hippel, P.H. (1973) Proc. Natl. Acad. Sci. USA 70, 298-302. Miyazaki, K. (2002) Nucleic Acids Research 30, el39.

SantaLucia J. and Hicks, D. (2004) Annu. Rev. Biophys. Biomol. Struct. 33, 415-440. Suter, CM., et al. (2004) Int. J. Colorectal Dis. 19, 95-101.

Claims

Claims:

1. A method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising: (a) providing at least one oligonucleotide probe of about 7 to about 25 nucleotides;

(d) removing unbound DNA; and

(e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein either or both of the DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety.

2. The method according to claim 1, wherein the at least one oligonucleotide probe is about 7 to about 12 nucleotides in length.

3. The method according to claim 2, wherein the at least one oligonucleotide probe is about 7 nucleotides in length.

4. The method according to any one of claims 1 to 3 wherein the at least one oligonucleotide probe comprises one or more universal bases or a mixture of normal or modified nucleotides at one or both ends to increase the melting temperature of oligonucleotide-target double-helices .

5. The method according to any one of claims 1 to 4 wherein each of the at least one oligonucleotide probe and the DNA of the sample comprise modified nucleotides.

6. The method according to any one of claims 1 to 5, wherein the modified nucleotides are selected from the group consisting of 2'-O-methyl nucleotides, 2-amino- adenine, 2-ammo-purine, inosine, propynyl nucleotides; 2-thio-thyrnidme; universal bases, locked nucleic acid (LNA), and peptide nucleic acid (PNA).

7. The method according to claim 6, wherein the 2'-O-methyl nucleotides are selected from 2'-O-methyl adenosine, 2'~O-methyl-uridine, 2'-O-methyl-thymidme, 2'-O-methyl- cytidine, 2'-O-methyl-guanosine, 2'-O-methyl-2-amino-adenosine.

8. The method according to claim 6, wherein the propynyl nucleotides are selected from 5-propynyl uracil and 5-proρynyl cytosine.

9. The method according to claim 6, wherein the universal base is 5-nitro-indole.

10. The method according to any one of claims 1 to 9, wherein the DNA of the sample is subjected to a fragmentation step to generate fragments of up to about 100 bases.

11. The method according to any one of claims 1 to 10, wherein the at least one oligonucleotide probe is immobilised on a solid support.

12. The method according to any one of claims 1 to 10, wherein the hybridisation is carried out in solution.

13. The method according to any one of claims 1 to 12, wherein the DNA of the sample is treated with bisulphite to convert unmethylated cytosines to uracils and subsequently amplified prior to incubation with the at least one oligonucleotide probe.

14. The method according to claim 13, wherein the amplified DNA requires fragmentation to produce fragments of up to about 100 bases.

15. A method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising: (a) providing at least one oligonucleotide probe of about 7 to about 25 nucleotides;

(b) providing a sample comprising amplified DNA of a target region, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases; (c) incubating the amplified DNA with the at least one oligonucleotide probe under conditions suitable to enable hybridisation between probe and target DNA; (d) removing unbound amplified DNA; and (e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein either or both of the amplified DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety.

16. A method of increasing fidelity of hybridisation between a probe oligonucleotide and a target DNA, the method comprising:

(b) providing a sample comprising amplified DNA of a target region, wherein the amplified DNA is prepared so as to comprise fragments of up to about

100 bases;

(c) incubating the amplified DNA with the at least one oligonucleotide probe under conditions suitable to enable hybridisation between probe and target DNA; (d) removing unbound amplified DNA; and

(e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein either or both of the amplified DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety.

17. A method of oligonucleotide array-based analysis of DNA, the method comprising:

(a) providing a sample comprising target DNA;

(b) amplifying a DNA region comprising target DNA, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases; (c) incubating amplified DNA with one or more oligonucleotide probes of about

7 to about 25 nucleotides immobilised and arrayed on a solid support under conditions to allow hybridisation_/between fragmented DNA and probes,

(d) washing the oligonucleotides to remove unbound amplified DNA;

(e) detecting DNA hybridised to the oligonucleotide probes; wherein either or both of the amplified DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety. _.

18. A method of oligonucleotide array-based analysis of DNA methylation, the method comprising:

(a) providing a sample comprising target DNA;

(e) washing the oligonucleotides to remove unbound amplified DNA;

(f) detecting DNA hybridised to the oligonucleotide probes; wherein either or both of the amplified DNA and the at least one oligonucleotide probe comprises one or more modified nucleotides, and wherein the DNA of the sample is optionally labelled with a detectable moiety.

19. A method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising:

(a) providing at least one oligonucleotide probe of about 7 to about 25 nucleotides comprising one or more modified nucleotides; (b) providing a sample comprising target DNA, wherein the DNA is prepared so as to comprise fragments of up to about 100 bases;

(d) removing unbound DNA; and (e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein the DNA of the sample is optionally labelled with a detectable moiety.

20. A method of reducing nucleotide mis-pairing in hybridisation between a probe oligonucleotide and a target DNA, the method comprising:

(a) providing at least one oligonucleotide probe of about 7 to about 25 nucleotides comprising one or more modified nucleotides; (b) providing a sample comprising amplified DNA of a target region, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases and wherein the amplified DNA comprises one or more modified nucleotides; (c) incubating the amplified DNA with the at least one oligonucleotide probe, under conditions suitable to enable hybridisation between probe and target DNA;

(d) removing unbound amplified DNA; and

(e) detecting DNA hybridised to the at least one oligonucleotide probe; wherein the DNA of the sample is optionally labelled with a detectable moiety.

21. A method of increasing fidelity of hybridisation between a probe oligonucleotide and a target DNA, the method comprising:

DNA;

(d) removing unbound amplified DNA; and

22. A method of oligonucleotide array-based analysis of DNA, the method comprising:

(a) providing a sample comprising target DNA;

(b)amplifying a DNA region comprising target DNA, the amplified DNA prepared so as to comprise fragments of up to about 100 bases, and wherein the amplified DNA comprises one or more modified nucleotides;

(d)washing the oligonucleotides arrayed on the solid support to remove unbound amplified DNA;

23. A method of oligonucleotide array-based analysis of DNA methylation, the method comprising:

(a) providing a sample comprising target DNA;

(c) amplifying a DNA region comprising target DNA, wherein the amplified DNA is prepared so as to comprise fragments of up to about 100 bases and wherein the amplified DNA comprises one or more modified nucleotides;

(e) washing the oligonucleotides arrayed on the solid support to remove unbound amplified DNA;