Introduction:

In primer walking sequencing, a primer/template complex is extended with a polymerase and chain terminated to generate a nested set of fragments whence the sequence is read after electrophoresis and detection (radioactive or fluorescent). A second primer is then synthesised using the sequence information near to the end of the sequence obtained from the first primer. This second ('walking') primer is then used for sequencing the same template. Primer walking sequencing is more efficient in terms of generating less redundant sequence information than the alternative 'shotgun' approach.

Disadvantages of the Current Art:

The main disadvantage with primer walking is the resynthesis of the walking primer after each round of sequencing. Studier (J Kieleczawa et al, Science 258, p1787, (1992)) and others (T Azhikina ef al, PNAS, 90, p11460, (1993)) have proposed elegant schemes using modular primers constructed out of presynthesised sets to avoid de novo primer synthesis after each round of sequencing. None of these walking primer schemes, however, are suited to cycle sequencing.

Cycle sequencing requires primers that have annealing temperatures near to the optimal temperature for the polymerase enzyme used for the cycle sequencing. Primers between 18 and 24 residues long are generally used for cycle sequencing.

As the length of the primer increases, then the size of any presynthesised walking primer set needed in order to achieve a given probability of obtaining a perfect match within a given number of

contiguous overlapping frames also increases. For primers between 18 and 24 residues long, the size of the presynthesised walking primer set required makes primer walking an impractical proposition.

Advantages of the Current Invention:

The current invention allows the use of a manageably sized set of presynthesised walking primers for cycled sequencing.

The Current Invention: The invention provides a method for primer walking sequencing of a nucleic acid target, which method comprises performing a series of sequencing reactions, each involving hybridising a primer to the target and effecting chain extension/chain termination of the primer, wherein for each sequencing reaction there is used a primer selected from a presynthesised set of walking primers of which the annealing temperatures are raised and/or the annealing properties improved without increasing their sequence complexity.

The invention also provides a library of y oligonucleotides, where y is defined as from 2 to 20000, each oligonucleotide comprising n nucleotide residues N and x nucleotide analogue residues X wherein i) n is defined as 8 or 9, ii) x is defined as 3-5, iii) each nucleotide analogue residue X is defined as: either capable of base pairing with two or more of A C G and T, or forming stronger base interactions than A C G T, iv) the order of the nucleotide residues N and the nucleotide analogue residues X in the oligonucleotide is defined as: either random or the residue X at the 5'-end of the oligonucleotide chain.

A DNA sequence is herein considered as a series of contiguous overlapping identically sized frames of residues with each

frame translated from the next by one residue.

The following examples will define the terminology used herein:

Let each frame be n residues long:

e.g. within the sequence

GACTGTTACGACTTAGACCATAGAAGATCGATAGAC

TTACGACT is a frame of 8 residues

i.e. n=8

e.g. within the sequence

GACTGTTACGACTTAGACCATAGAAGATCGATAGAC

TACGACT is a frame of 7 residues

i.e. n=7

Let there be i such frames suitable for selecting the walking primer

e.g. within the sequence

e.g . GACTGTTACGACTTAGACCATAGAAGATCGATAGAC

ACGACTTA TACGACTT and TTACGACT are 3 frames of 8 residues i.e. i=3, n=8

e.g. within the sequence

GACTGTTACGACTTAGACCATAGAAGATCGATAGAC

GACTTAG CGACTTA ACGACTT and TACGACT are 4 frames of 7 residues

i.e. i=4, n=7

For each frame of n residues, the number of possible sequences is given by

₄ n

The values of 4 ⁿ for n between 5 and 9 are given below

n 4 ⁿ

5 1,024

6 4,096

7 16,384

8 65,536

9 262,144

Hence the sequence of a given 5 mer will occur about once every 1 kb on average in random sequence DNA, the sequence of a given 6 mer will occur about once every 4 kb on average, the sequence of a given 7 mer will occur about once every 16 kb on average, the sequence of a given 8 mer will occur about once every 65 kb on average and the sequence of a given 9 mer will occur about once every 262 kb on average. Thus one would expect a given 5 mer to form a perfect match at a unique location within a template of less than about 1 kb. For templates larger than about 1 kb, the 5 mer would be expected to form perfect matches at multiple locations within the template. Such formation of perfect matches at multiple locations within the template would make the 5 mer useless as a walking primer for sequencing.

One would expect a given 6 mer to form a perfect match at a unique location within a template of less than about 4 kb. For templates larger than about 4 kb, the 6 mer would be expected to form perfect matches at multiple locations within the template. Such formation of perfect matches at multiple locations within the template would make the 6 mer useless as a walking primer for sequencing.

One would expect a given 7 mer to form a perfect match at a unique location within a template of less than about 16 kb. For templates larger than about 16 kb, the 7 mer would be expected to form perfect matches at multiple locations within the template. Such formation of perfect matches at multiple locations within the template would make the 7 mer useless as a walking primer for sequencing. One would expect a given 8 mer to form a perfect match at a unique location within a template of less than about 65 kb. For templates larger than about 65 kb, the 8 mer would be expected to form perfect matches at multiple locations within the template. Such formation of perfect matches at multiple locations within the template would make the 8 mer useless as a walking primer for sequencing.

One would expect a given 9 mer to form a perfect match at a unique location within a template of less than about 262 kb. For templates larger than about 262 kb, the 9 mer would be expected to form perfect matches at multiple locations within the template. Such formation of perfect matches at multiple locations within the template would make the 9 mer useless as a walking primer for sequencing.

Sequencing templates and template sizes are variable and will be obvious to those skilled in the art. The following guidelines are given for example only. Plasmid templates may be in the range 2kb to 15 kb

M13 templates may be in the range 8 kb to 15 kb

Lambda templates may be in the range 45 kb to 55.kb

Cosmid templates may be in the range 45 kb to 55 kb

Bacterial artificial chromosome templates may be in the range 50 kb to 150 kb

Yeast artificial chromosome templates may be in the range 100 kb to 1 ,000 kb

Thus primers of 7 residues or longer will be required in order to generate a perfect match at a unique site within plasmid and M13 templates.

Primers of 8 residues or longer will be required in order to generate a perfect match at a unique site within lambda and cosmid templates.

Primers of 9 residues or longer will be required in order to generate a perfect match at a unique site within bacterial artificial chromosome templates.

Primers of 10 residues or longer will be required in order to generate a perfect match at a unique site within yeast artificial chromosome templates. For primer walking sequencing using a presynthesised set of

walking primers, it is possible to calculate the probability of finding a perfect match between a primer within the presynthesised set and one of the sequences within i contiguous overlapping frames of n residues near to the end of the sequence read from the previous primer. Let S be the number of primers n residues long comprising the set of walking primers

For a single frame n residues long, the probability that there will be a perfect match between the frame sequence and a primer n residues long in the set of walking primers is given by

S/4 ⁿ

For a single frame n residues long, the probability that there will not be a perfect match between the frame sequence and a primer n residues long in the set of walking primers is given by

1-(S/4 ⁿ )

For i contiguous overlapping frames n residues long, the probability that there will not be a perfect match between any of the frame sequences and a primer n residues long in the set of walking primers is given by

P = (1-(S/4 ⁿ ))'

Thus

logP = ilog(1-(S/4 ⁿ ))

i = logP/log(1-(S/4 ⁿ ))

S = 4 ⁿ (1-(P) ^1/i )

S/4 ⁿ = (1-(P) ^{1 i} )

The following table values give the fraction of the walking primer set needed to give the percentage probability (P) shown of a perfect match within the number of contiguous overlapping frames (i) shown.

S/4 ⁿ =

P i 80 % 90 % 99 % 99.9 %

5 0.275 0.369 0.602 0.749

10 0.149 0.206 0.369 0.499

20 0.077 0.109 0.206 0.292

50 0.032 0.045 0.088 0J29

100 0.016 0.023 0.045 0.067

200 0.008 0.011 0.023 0.034

500 0.003 0.005 0.009 0.014

The number of 5 mers needed in order to give the percentage probability (P) shown of a perfect match within the number of contiguous overlapping frames (i) shown is given in the following table:

P i 80% 90% 99% 99.9%

5 282 378 616 767

10 153 211 378 511

20 79 112 211 299

50 33 46 90 132

100 16 24 46 69

200 8 11 24 35

500 3 5 9 14

The number of 6 mers needed in order to give the percentage probability (P) shown of a perfect match within the number of contiguous overlapping frames (i) shown is given in the following table:

P i 80% 90% 99% 99.9%

5 1126 1511 2466 3068

10 610 844 1511 2044

20 315 446 844 1196

50 131 184 360 528

100 66 94 184 274

200 33 45 94 139

500 12 20 37 57

The number of 7 mers needed in order to give the percentage probability (P) shown of a perfect match within the number of contiguous overlapping frames (i) shown is given in the following table:

P i 80% 90% 99% 99.9%

5 4506 6046 9863 12272

10 2441 3375 6046 8176

20 1262 1786 3375 4784

50 524 737 1442 2114

100 262 377 737 1098

200 131 180 377 557

500 49 82 147 229

The number of 8 mers needed in order to give the percentage probability (P) shown of a perfect match within the number of contiguous overlapping frames (i) shown is given in the following table:

P i 80% 90% 99% 99.9%

5 19022 24183 39453 49086

10 9765 13500 24183 32702

20 5046 7143 13500 19137

50 2097 2949 5767 8454

100 1049 1507 2949 4361

200 524 721 1507 2228

500 197 328 590 918

The number of 9 mers needed in order to give the percentage probability (P) shown of a perfect match within the number of contiguous overlapping frames (i) shown is given in the following table:

P i 80% 90% 99% 99.9%

5 72090 96731 157811 196346

10 39059 54002 96731 130810

20 20185 28574 54002 76546

50 8389 11796 23069 33817

100 4194 6029 11796 17564

200 2097 2884 6029 8913

500 786 1311 2359 3670

The following example clearly shows that only a fraction of the complete set of presynthesised walking primers of length n residues needs to be used in order to have a very high probability of finding a perfect match between one of the presynthesised walking primers and one of i contiguous overlapping frames of n residues where i is smaller than the average length of sequence read per cycle (about 400-500 bases).

For this example, the number (i) of contiguous overlapping frames of 8 mers (n=8) needed in order to give the percentage probability (P) shown of a perfect match for a set of presynthesised walking primers of size S is given in the following table:

i for ... s 90 % probability of 99 % probability of 99.9 % probability fmding a perfect finding a perfect of finding a perfect match match match

100 1494 2988 4482

500 298 596 894

1000 148 296 444

1500 99 198 297

2000 74 148 222

3000 49 98 147

4000 36 72 108

5000 29 58 87

Thus, for example, a presynthesised set of 1,500 8 mers would give a 90 % chance of finding a perfect match between one of the 8 mers and one of 99 contiguous overlapping frames 8 residues long.

The same presynthesised set of 1,500 8 mers would give a 99 % chance of finding a perfect match between one of the 8 mers and one of 198 contiguous overlapping frames 8 residues long, The same presynthesised set of 1 ,500 8 mers would give a

99.9 % chance of finding a perfect match between one of the 8 mers and one of 297 contiguous overlapping frames 8 residues long.

In the above example, such a presynthesised set of walking primers would allow a very high probability of selecting an 8 mer presynthesised walking primer after each round of sequencing. A set of 1 ,500 8 mers is a manageable amount for synthesis and accessing for each round of priming. Such 8 mer primers would also be expected to generate a perfect match at a unique site within all templates of a size up

to and including lambda and cosmid templates. Such simple 8 mer primers would not, however, be suitable for cycle sequencing. As stated above, cycle sequencing requires primers that have annealing temperatures near to the optimal temperature for the polymerase enzyme used for the cycle sequencing. Primers between 18 and 24 residues long are generally used for cycle sequencing.

The current invention relies upon raising the annealing temperature (or, more generally, improving the annealing properties) of a presynthesised set of walking primers (of the type described above) WITHOUT increasing the size of the walking primer set required in the following ways:

A. Adding residues to the primers that are capable of base pairing with each of the four DNA bases (A, C, G and T). Examples are:

i) inosine:

(which pairs with all four bases with the preference C>A>G~T).

ii) 5-nitro indole:

(which can pair with all four bases, but mainly stabilises by favourable stacking interactions).

iii) 5-nitro-pyrrole:

(which pairs with all four bases, but mainly stabilises by favourable stacking interactions).

iv) K (2-amino-6-methoxyaminopurine):

(which pairs with pyrimidines).

v) P (6H,8H-3,4-dihydropyrimido[4,5-c][1 ,2]oxazin-7-one):

R

(which pairs with purines).

vi) Others that may be used which are obvious to those skilled in the art. B. Using bases in the primers that form stronger base pairing interactions than the normal A, C, G and T bases. Examples are:

i) 2-aminoadenine

R

(which can be used in place of A),

ii) 5-methylcytosine

R

(which can be used in place of C).

iii) Others that may be used which are obvious to those skilled in the art.

The bases listed above may be collectively termed degenerate bases. The number of such degenerate bases, in the walking primers of the pre-synthesised set with which the invention is concerned, is preferably from 1 to 20. These degenerate bases may be interspersed along the length of each oligonucleotide walking primer, or may be concentrated, at the 5'-end, or at the 3'-end, or in the middle. Good results

have been obtained using 3, 4 or 5 degenerate bases concentrated at the 5'-end of the oligonucleotide.

Example 1

The following oligonucleotides were synthesised:

5' GTCACGAC 3' (AM1)

5' XXX GTCACGAC 3' (AM2) where X denotes a 5-nitroindole residue.

25 pmol of each of these primers were added to CsCI purified M13mp8 template and cycle sequenced on a Vistra DNA Systems DNA Labstation 625 (Amersham lntemational pic.) using the Labstation Thermo Sequenase™ fluorescent dye-terminator cycle sequencing kit (RPN2435, Amersham lntemational pic), the Labstation FMP™ fluorescent dye- terminator precipitation kit (RPN2433, Amersham International pic.) and the 1-32 dye-terminator cycle sequencing, v2.0 method. The method was edited for this example by reducing the cycle sequencing annealing temperature to 40°C followed by 2 sec/°C ramping to the 4 minute extension at 60 degrees.

The samples were then electrophoresed and analysed on an Applied Biosystems 373A fluorescent sequencer.

The results clearly showed a marked improvement in both the signal strength and sequence quality by virtue of the added 5-nitroindole residues at the 5 ' end of primer AM2. Signal strength was 8 times higher with AM2 than with AM1. Readable sequence was 460-500 bases with AM2 compared to 340-380 with AM1.

Example 2

Oligonucleotides were synthesised of general formula 5' XΛ X _m GTCACGAC 3' as follows

X m

5-nitroindole 1, 2, 3, 4, 5

5-nitroindazole 3, 4, 5

3-niropyrrole 3 benzimidazole 1 - 6

The oligonucleotides containing 5-nitroindole residues all performed better than AM1 , in the test described in Example 1 , with those where m was 3 or 4 being marginally the best . The performance of the oligonucleotides containing

5-nitroindazoie residues were comparable to AM2.

The performance of the oligonucleotide containing 3-nitropyrrole residues was better than AM1 (the signal was twice as strong). On the basis of these results, it can be predicted that a library of presynthesised oligonucleotides, containing 8 or 9 ordinary bases and 3, 4 or 5 degenerate bases particularly those listed above, will give improved results when used in a primer walking sequencing method as described herein.