GB2380999A

GB2380999A - Synthesis of oligonucleotide mixtures, and polynucleotide assembly

Info

Publication number: GB2380999A
Application number: GB0124945A
Authority: GB
Inventors: Sten Linnarsson
Original assignee: GLOBAL GENOMICS AB
Current assignee: GLOBAL GENOMICS AB
Priority date: 2001-10-17
Filing date: 2001-10-17
Publication date: 2003-04-23
Anticipated expiration: 2021-10-17
Also published as: GB2380999B; GB0124945D0

Abstract

Methods comprise synthesizing a population of oligonucleotides as an array on a solid support, such a population being useful in assembly of double-stranded polynucleotides. A method of assembly comprises optionally phosphorylating oligonucleotides in the array on the solid support, cleaving the population from the solid support to provide a mixture of oligonucleotides in suspension or solution; and assembling oligonucleotides in the mixture into the product double-stranded polynucleotide. Oligonucleotides form part of strands of the product double-stranded polynucleotide and may be assembled contiguously by ligation or may form a hybridised double-stranded polynucleotide containing nicks and/or gaps, and be assembled using ligation and polymerase chain reaction.

Description

SYNTHESIS OF OLIGONUCLEOTIDE MIXTURES, AND POLYNUCLEOTIDE ASSEMBLY The present invention relates to oligonucleotide synthesis and to polynucleotide assembly. In particular the invention relates to methods, apparatus, and systems for synthesising and using oligonucleotide mixtures. Furthermore, the invention relates to assembly of oligonucleotide mixtures into polynucleotides.

The present invention provides for synthesis of a plurality oligonucleotides in a population on an array, followed by release of the oligonucleotides as a mixture. Oligonucleotide mixtures may be assembled into double-stranded DNA molecules, and the invention further provides for assembling a doublestranded nucleic acid sequence from a mixed population of oligonucleotides. Such assembly is useful in a variety of fields, including cloning, subcloning, construction of genetically engineered organisms, in vitro selection and evolution, in situ mutagenesis, protein design, codon optimization, creation of virus vectors for gene therapy, creation of artificial gene regulatory circuits for control of gene expression, DNA computation and many other areas of interest to the skilled person.

The invention further provides for computer processor control of synthesis of mixed populations of oligonucleotides, and/or assembly of mixed populations of oligonucleotides into desired polynucleotides, and apparatus and systems for such synthesis and/or assembly.

While DNA sequencing technology has advanced very rapidly in recent years-to the point where the entire human genome has

been sequenced twice and the cost for a regular 600 base pair DNA sequence is not more than $5, or less than 1 cent per base - DNA synthesis remains expensive and difficult.

Presently, techniques for DNA synthesis techniques produce short oligonucleotides, usually less than 100 bp long, at a cost per base pair of $0.5. Thus, while the human genome could be sequenced for about $30 million, synthesizing oligonucleotides covering the two strands of the genome would cost $3 billion.

The current approach to synthesis of oligonucleotides utilises a solid-phase chemical reaction (see e. g. Gait, M. J. , Ed.

(1984) Oligonucleotid Synthesis: A Practical Approach. IRL Press, Oxford; also http ://www. interactiva. de/knowledge/nucleicchem/index. html). A support, usually controlled-pore glass or polystyrene, is activated to expose amine groups on its surface. See e. g. http ://www. interactiva. de/knowledge/nucleicchem/oligochem. html . A first nucleoside monomer is attached to the surface via a succinate ester link. The succinate ester is labile in a nucleophilic base and can thus easily be cleaved after synthesis, freeing the oligonucleotide from the support and leaving a 3'OH group.

During the process of elongation, it is necessary to protect the nucleobases of both the monomers and the elongating strand, such that they are unavailable for condensation. This is achieved by linking the nucleobase to an acyl functionality, for example a benzoyl (dAbz, dCbZ), acetyl (dCaC), isobutyryl (dG1bU) or dimethylformamidine (dGdm) functionality. The links to these protective groups are labile in a nucleophilic base, allowing the protective groups to be

removed after synthesis and thus allowing the oligonucleotides to be used in hybridization.

Since both the anchor attaching the oligonucleotides to the support and the links to the above mentioned acyl functionalities are labile in a nucleophilic base, the oligonucleotides can be released and deprotected by exposure to a single agent.

In addition, the 5'phosphate of the nucleoside monomer is replaced by a Dimethoxytrityl (DMT) blocking group to prevent uncontrolled oligomerization from taking place. http ://www. interactiva. de/knowledge/nucleicchem/oligochem. html is a good source for more details, including how to make protected bases from pure chemicals.

Synthesis begins when the DMT block is removed from the first, anchored nucleoside monomer, to leave a 5'hydroxyl group.

This is achieved by contacting the monomer with trichloroacetic acid in dichloromethane (detritylation). A protected phosphoramitide nucleoside monomer is then added together with tetrazole (coupling and activation), resulting in reaction between the phosphoramitide group of the monomer and the 5'hydroxyl of the growing chain to form a phosphite link.

Nucleosides on the growing chain which fail to react with the monomer are capped by with acetic anhydride and 1methylimidazole. This prevents the growing chain from undergoing subsequent reaction, and so prevents the formation of oligonucleotides with"missing"base pairs. The capped oligonucleotides will be shorter than the full length product and so can be readily separated during subsequent

purification. Thus, the capping step helps to ensure that the final product has the desired sequence, but also reduces the yield of the reaction. The more reaction steps that are required to generate an oligonucleotide, the lower the yield of the reaction will be, and this places an upper limit on the length of oligonucleotides which can be generated. Common coupling efficiencies of > 98.5% permit the synthesis of 60mers with 40% yield.

As a final step in each round of elongation, the phosphite link is oxidized (in, e. g. a mix of 12, H2O, pyridine and tetrahydrofuran) to a phosphotriester bond.

The cycle of activation-coupling-capping-oxidization is then repeated until the desired oligonucleotide has been formed.

After synthesis, the oligonucleotides are released from the

support by treatment with a nucleophilic base (e. g. NH4OH) and deprotected in the same base at higher temperature (e. g. 650C 1 hour), to make the nucleobases available for hybridisation.

Full-length oligonucleotides can then be purified by various methods (for example HPLC or PAGE). For gene expression purposes, PAGE is the preferred method as it gives biocompatible oligonucleotides with high purity, and fulllength species can be isolated simply by cutting the relevant band from the gel.

In a modification of this technique (Beattie et al 1988), multiple oligonucleotide sequences can be produced, by means of initiating the synthesis of each oligonucleotide sequence from a separate Teflon"wafer". A number of wafers are treated with the same monomer in any given round of elongation, but the wafers are sorted between each round, such

that each can be exposed to a different sequence of reactions.

Thus, a different oligonucleotide can be produced on each wafer, using the same number of rounds of elongation as are required to produce just one of these sequences.

If the short oligonucleotide fragments are subsequently to be assembled into a contiguous stretch of double-stranded DNA, then they must be phosphorylated. This is usually achieved enzymatically, because the cost of chemical phosphorylation is very high, as much as $20 per oligonucleotide. However, enzymatic phosphorylation is inefficient, commonly resulting in only-"60% phosphorylation. Unphosphorylated oligonucleotides which are integrated into a growing DNA chain can not be ligated to further oligonucleotides, thus blocking extension of the chain and reducing the yield of full length product. The problem grows with the number of oligonucleotides used and has been reported to limit ligation-based assembly to 14 oligonucleotides or less (Chalmers, 2001).

Arrays of oligonucleotides have been made attached to a chip for hybridisation experiments (Southern 5,700, 637 and 5,436, 327). According to this method, oligonucleotides are synthesized in situ using a plotter or ink-jet printer. The first base must be stably attached to the support medium throughout the synthesis and deprotection steps. In one approach, the oligonucleotide is anchored to the surface by a bond which is not cleaved by a nucleophilic base, for example a stable phosphoester (Maskos, 1992) or phosphoramidate bond (Matson, 1995). In another approach, the protecting groups may be altered such that they do not require a nucleophilic base for deprotection (Weilter et al 1996).

Light-directed photolithographic methods (US patent 5,445, 934) and ink-jet printing methods (Hughes, 2001 ; Recht, 1996) have been used for in situ oligonucleotide synthesis.

In the former method, the activation step has been rendered light-sensitive by replacing DMT with a light-sensitive group.

In each round light is shone through a mask, activating only the sites about to receive monomer. To join a number of nucleotides, four times that number of steps are required, as the monomers must be added sequentially. And for each new array design, four times the same number of masks must first be manufactured adding to the cost and inflexibility of the procedure.

The ink-jet method uses the standard DMT chemistry. The inkjet head can either deliver only activation chemicals, or it can deliver the actual monomers in parallel (Blanchard, 1998).

Using hydrophobic wells (i. e. hydrophilic patches surrounded by hydrophobic barriers laid down using standard photoresist technology), as suggested by Southern and others, a density of 50 spots per mm2 can be achieved (Hughes, 2001; Recht, 1996).

Because current techniques only allow for the synthesis of short oligonucleotides, methods for assembling these oligonucleotides into longer sequences are required.

A first group of methods are based on ligation. In the classic approach, (Khorana, 1979), short oligonucleotides (of around 10 to 12 base pairs) are individually synthesized.

These oligonucleotides represent contiguous sections of the first and second strands of the DNA sequence to be generated, wherein the sections of the first strand and the sections of the second strand are offset relative to each other, with an

overlap capable of annealing. Thus, an oligonucleotide of a first strand will generally have regions complementary to two oligonucleotides of the second strand. Once synthesized, the oligonucleotides are 5'phosphorylated using the enzyme polynucleotide kinase. Groups of 5 to 7 complementary oligonucleotides are then mixed and allowed to assemble into short double-stranded products, in which the nicks on each strand are offset relative to each other along the length of the double-stranded product. These nicks are ligated using a DNA ligase enzyme such as E. coli DNA Ligase or T4 DNA Ligase, or other available low-temperature enzyme, to produce a double-stranded DNA"duplex", which itself has single-stranded overhangs at each end.

The maximum length of the duplexes which can be produced according to this first stage is controlled by a number of factors. Firstly, the length of the original oligonucleotides which can be used is limited by the potential for the formation of thermodynamically stable self-associations (secondary structures), such as hairpin structures, at low temperatures. The probability of forming secondary structures increases as the length of the sequence increases.

In addition, longer olignucleotides are more likely to have mutations, as a result of errors during the synthesis process.

The number of oligonucleotides which can be annealed in a single step is affected by the potential for mis-matching during hybridization. This problem is in fact exacerbated by the potential for secondary structures to form in singlestranded sequences, since this necessitates short overhangs, reducing the specificity of hybridization.

In order to generate a full-length sequence, these DNA duplexes are ligated to each other in a stepwise manner. Each duplex is added in turn to the growing sequence, the singlestranded sequence of the duplex is hybridized to the singlestranded overhang at the end of the growing sequence, and the nick is ligated, before the next duplex is added.

The multiple-step nature of this procedure results in the assembly of even a short double-stranded sequence being extremely time consuming, as well as extremely labour intensive. Moreover, purification of the intermediate duplex fragments after each round of annealing and ligation is inefficient, and results in a low yield of the desired fulllength product.

US 4,652, 639 describes a modified version of this method, wherein the initial oligonucleotides are longer, comprising between 18 and 24 nucleotides. Complementary pairs of oligonucleotides are placed in solution together and heated, and then cooled and allowed to hybridise, so as to form double-stranded segments with at least one 3-7 nucleotide terminal single-stranded overhang. Subsequently, these doublestranded segments are annealed and ligated to each other, via their complementary single-stranded overhangs. It is suggested that annealing of all the duplexes may occur in a single step. As above, the number of duplexes which may be assembled in a single step is limited by the degree of specificity which can be conferred by an overhang of 3-7 nucleotides.

An alternative modification of the method of Khorana (US patent 5, 942, 602) improves the ease of purification of intermediate products, by anchoring the elongating chain to a

solid support. A single-stranded"bridging"oligonucleotide is annealed to this initiating oligonucleotide, so as to form a double-stranded section with a terminal single-stranded overhang. Subsequently, an single-stranded"assembly" oligonucleotide having at one end a sequence complementary to the overhang is annealed and ligated. This process is repeated so as to create one complete strand of the desired length. A complete second strand is then produced by PCR.

A further method (US patent 6,110, 668) makes use of a first single-stranded template DNA molecule. Chemically synthesized oligonucleotides which correspond to contiguous sections of the second strand but which have one or more mutations in the sequence are then annealed to this template. The oligonucleotides are then ligated to each other using repeating cycles of melting, annealing and ligation, and the newly manufactured strand, which is a variant of the original sequence, is separated from the original strand by polymerase chain reaction amplification (PCR) using primers specific to the new strand. Because this reaction utilises a complete single-stranded template, the method is limited to the production of variants of cloned sequences.

Wholly PCR-based methods have also been used.

US patent 5, 503,995 describes a cyclical method of gene assembly whereby a single-stranded oligonucleotide is subjected to PCR to generate a double-stranded product, and is then subjected to enzymatic cleavage to remove several base pairs from the 5'terminus, leaving a short 3'single-stranded overhang. A second single-stranded oligonucleotide bearing at one end a region complementary to this short overhang is annealed, so as to produce a long 5'overhang. The cycle of

PCR, cleavage and annealing is repeated, wherein in each step the PCR is primer from the double-stranded region, and uses the 5'single-stranded overhang as a template. This method may proceed from a mixed population of oligonucleotides, provided that each can be enzymatically treated to provide a unique 3' overhang and also has a 3'region capable of selectively hybridizing to the appropriate overhang of the growing strand.

Because PCR can proceed using thermostable enzymes, higher temperatures can be used, and this reduces the formation of secondary structures in the single-stranded nucleic acids.

According to the method of Withers-Martinez et al (1999), oligonucleotides are synthesised which are each 40 nucleotides in length and which correspond to contiguous sections of the first and second strands of the desired sequence. Each oligonucleotide has a half complementary to a first oligonucleotide of the opposite strand, and a half complementary to a second oligonucleotide the opposite strand, such that the oligonucleotides will form double-stranded products with a 20 nucleotide overhang at each end. Multiple cycles of PCR are performed on the pooled oligonucleotides, wherein the reaction is primed from the double-stranded region of annealed, complementary oligonucleotides, and uses the 20bp overhang as a template for elongation. Thus, in each progressive round of PCR, the product lengthens by 20 nucleotides.

PCR based-methods have the disadvantage of introducing point mutations into the desired sequence during amplification. The number of point mutations in the product will increase with the length of the desired sequence. Withers-Martinez et al (1999) notes that point mutations in the final product must be

corrected by subsequent subcloning steps, and that this represents by far the most time-consuming stage of the method.

In a method which can be thought of as a synthesis of the above PCR-based method and a ligase-based method (Au et al 1998) short double-stranded duplexes are formed by the "ligation chain reaction". Here, an overlapping, complementary set of oligonucleotides is subject to a rapid cycle of heating and cooling, from 950C to 550 to 700 and back to95 C, in the presence of a heat-stable ligase. This first melts any double-stranded products then to allows complementary oligonucleotides to anneal and to be ligated by the heat stable ligase. Subsequently the duplexes produced in this way are joined using a PCR-based method.

Because of the costs of producing oligonucleotides, and the limitations associated with existing methods, novel DNA sequences are most often obtained by assembling new constructs from pieces of cloned DNA, using well known tools of molecular biology, such as restriction enzymes, ligases and in situ mutagenesis.

The present invention provides a method of producing mixtures of oligonucleotides in suspension or solution, and also provides a method of assembling mixed oligonucleotides into a contiguous double-stranded DNA sequence.

In one aspect, the present invention provides a method of producing a mixture of oligonucleotides in suspension or solution, the method comprising: synthesizing a population of oligonucleotides as an array on a solid support; cleaving the population from the solid support to provide the mixture in suspension or solution.

Prior to the cleaving, the oligonucleotides may be phosphorylated, this being advantageous where they are to be used in polynucleotide assembly as discussed.

Following provision of the mixture in solution, the oligonucleotides may be assembled into a polynucleotide. To allow this, the oligonucleotides in the population contain overlaps such that annealing in a ladder formation provides a contiguous double-stranded DNA molecule. It is preferred that the members of the oligonucleotide population are the same length, as discussed further below.

Assembly may be by means of further steps of denaturing by heating the oligonucleotides to a temperature sufficient to eliminate annealing between oligonucleotides in the mixture and secondary structure within the oligonucleotides, cooling the denatured oligonucleotides in the presence of a thermostable ligase.

Denaturation may be by heating to a temperature of melting all double-stranded structures, e. g. about 95 C.

Cooling to allow annealing of the oligonucleotides and ligation using the thermostable ligase is preferably a single cooling down to the lowest temperature required to complete all desired ligation reactions, e. g. to 55 C if 55 C is the melting point of the weakest desired hybridization, and is at a rate that permits all possible ligations at each particular successive temperature to go to completion, preferably at a relatively slow rate, e. g. about 1 C in 2 minutes.

In preferred embodiments of the invention, a single cooling is employed, bringing the mixture from the temperature at which all the oligonucleotides are denatured or melted down to the lowest temperature required to complete all desired ligation reactions in one reaction, with concomitant annealing and ligation by means of the thermostable ligase.

If secondary structures have formed despite performing a preferred embodiment with a single cooling, such structures can be melted by one or more further rounds of heating and cooling in the presence of ligase.

The required temperature sufficient for denaturation, removing all annealing between oligonucleotides and all secondary structure within oligonucleotides, and also the rate of cooling in the presence of thermostable ligase in order to allow proper annealing and ligation of the oligonucleotides into a contiguous double-stranded polynucleotide, depend on a number of factors, including length of the oligonucleotides, potential overhang between overlapping oligonucleotides, GC/AT content of different oligonucleotides and the population as a whole, and actual sequences of the oligonucleotides.

The variables controlling these process parameters can be considered thus: 1. Assume there is a population of oligonucleotides which are pairwise complementary in a staggered configuration, so that if one"upper"oligo hybridizes to two"lower"oligos (or vice versa), a nick is formed. Call these oligos U for upper and L1, L2 for the two lower oligos.

2. For each such nick, calculate the temperature at which it will disappear by melting, by obtaining or calculating the melting curves of each of the two hybridizations. The nick will disappear when either of the two"lower"oligos melt away (and vice versa for the opposite strand).

3. For each temperature, calculate the effective concentration of nicks (given the concentration of each oligo).

At a given temperature, there will be a ratio of hybridized to melted for the U/Ll pair and for the U/L2 pair (these ratios are given by the melting curve at the indicated temperature).

Call these ratios rl and r2.

Assume the oligos are present at the concentration C.

The effective nick concentration will then be rl*r2*C.

This result can easily be generalized to any set of individual concentrations, but in embodiments of the invention there will usually be mixes of oligos of approximately the same concentration.

4. Given the ligation rate of the enzyme (e. g. Kcat), calculate the time required, at each temperature, for (e. g.) 95% complete ligation of all nicks available at that temperature.

5. The annealing rate is not necessarily constant, but may be calculated for each temperature across the range of nick melting points.

This disregards the fact that some lower-temperature nicks will ligate at higher temperatures and thus already be ligated when the temperature reaches their melting point. Therefore, this gives an upper limit on the time required at each temperature.

In a preferred embodiment of the present invention, oligonucleotides in a mixture in suspension or solution are used in synthesis of a complete double-stranded polynucleotide of a known, desired sequence. The intended sequence determines the required sequences of the oligonucleotides in the population to be assembled, and these sequences then determine the conditions to be employed in the assembly, according to parameters as noted and the algorithms set out.

Accordingly, in preferred embodiments synthesis of the mixed oligonucleotides and/or assembly into a polynucleotide are controlled by a computer processor operating suitable apparatus. A user or operator provides as input data the sequence of at least one strand of the double-stranded DNA molecule desired to be synthesized (the other strand being complementary). Parameters such as length of oligonucleotide to be synthesized and then assembled may be selected by the user, or optimum oligonucleotide length and overlap may be preselected or determined by the computer processor employing the operating algorithms.

Thus, further provided as an aspect of the present invention is a computer processor programmed to control a method of producing a mixture of oligonucleotides according to the present invention. Preferably, the computer processor controls operation of an ink-jet printing device for synthesis of oligonucleotides on an array as disclosed.

Still further, it is preferred that the computer processor controls assembly of oligonucleotides in the mixture into one or more polynucleotides as disclosed.

A system comprising an ink-jet printer controlled by a computer processor for synthesis of an oligonucleotide mixture, and a temperature-controlled reaction vessel into which an oligonucleotide mixture passes following synthesis and in which assembly of one or more polynucleotides takes place under control of the computer processor, is further provided by the present invention. Such a system may comprise a purifier, for instance comprising an electrophereser for a polyacrylamide gel from where a full-length band can be cut out. A detector and a collection device may be included so that an oligonucleotide mix can be run through a gel, and a full-length band detected and collected.

A further aspect provides a computer-readable device or computer program product having a software portion, such as diskette, CD-ROM etc. , carrying a programme for controlling a computer processing to operate a method and apparatus or system as disclosed herein.

In a further aspect, the invention provides a method for obtaining and optionally providing to a customer a desired polynucleotide, the method comprising accepting as a customer order input data comprising a polynucleotide sequence and/or identified vector or gene construct components, inputting the input data into a programmed computer processor, synthesizing a mixture of oligonucleotides as disclosed herein under control of the computer processor, preferably using an ink-jet

printer, assembling the polynucleotide from oligonucleotides within the mixture as disclosed herein, whereby the desired polynucleotide is obtained, and may be supplied to the customer.

Where known vector or gene construct components are specified as part of the input data within the customer order, sequences of these may be stored within the computer processor or retrievable by the computer processor (e. g. via the internet or otherwise from a database held on a server), and used in design of the oligonucleotides to be synthesized and then assembled. The art is replete with vectors, promoters and other regulatory sequences, antibiotic resistance markers and other components used in the construction of vectors and gene constructs for a multitude of uses. Sequences for many of these are already available in the art, or can be obtained by means of routine sequencing with the sequence data being stored for retrieval and use by the computer processor.

In one embodiment a customer or a user of a computer controlled system of the invention may design a polynucleotide sequence such as a complete vector by selection of components in silico, prior to instructing the computer processor to commence oligonucleotide synthesis. The system may comprise software allowing for in silico comparison of different constructs, e. g. at the behest of the user, to allow for optimisation prior to synthesis of a construct or vector of choice. Parameters such as codon usage, probability of homologous recombination, predicted secondary structure, and size of construct are among those which may be taken into account. The computer processor may be programmed with default parameters for any one or more variables to be taken

into account, or may allow for selection by the customer or user of chosen values for one or more of the variables.

Apparatus, including a computer processor controlled system or computer readable device carrying a control program, may be designed and employed in any aspect or embodiment of the present invention as disclosed herein.

Data storage within a system or apparatus of the invention, or accessible via a remote connection, may contain sequences of DNA components. A DNA Component may be any object capable of generating a sequence for incorporation into a polynucleotide to be synthesized, and may comprise one or more features for assembly (in silico prior to actual synthesis of oligonucleotides and assembly into a polynucleotide) with one or more other DNA Components. A Component can represent a peptide, a protein, a 5'UTR, a promoter, enhancer, other regulatory element, an intron, a splice acceptor, a loxP site, or any other component that may be included in a polynucleotide. For simplified retrieval, DNA Components may be classified or grouped (for example an intron with a splice acceptor and donor, or a 5'UTR, a protein and a 3'UTR to form an mRNA), e. g. in readily useable or preferred combinations.

This may facilitate rapid retrieval of components and design of polynucleotides for synthesis.

DNA Components can be considered rather like simple building blocks, like LEGO" pieces, which can be assembled into larger and larger constructs. A person operating the present invention may choose whatever components are desired for the polynucleotide, in any combination.

The programming may include categorisation of DNA Components that provide additional information or labelling that may, for example, be used to warn of structural inconsistencies or even prevent dangerous or otherwise undesirable combinations. For example, protein elements such as His-tags, GFP labels and so on may be labelled so they only fit into protein fusions, and the system may warn of the absence of a stop codon or some other inconsistency that may affect functionality.

More information is now provided on details of different aspects and embodiments of the present invention.

In a method of the invention of producing a mixture of oligonucleotides in suspension or solution, one embodiment of the method comprises: a) activating discrete patches of the surface of a solid support to provide an organised array of activated patches; b) contacting each activated patch with one or more protected nucleoside monomers, wherein a protected nucleoside monomer comprises a 5'blocking group and a protecting group linked to the nucleobase via a bond which is cleavable in a nucleophilic base, and wherein the protected nucleoside monomer added to each patch is selected from A, T, C or G, or a mixture thereof; c) anchoring a protected nucleoside monomer in contact with an activated patch to the activated patch surface via a bond which is cleavable in a nucleophilic base, to provide an array of patches of anchored nucleosides; d) activating some or all of the patches of anchored nucleosides by removing the 5'blocking group; e) contacting some or all of the patches with protected phosphoramitide nucleoside monomer, wherein the protected

phosphoramitide nucleoside monomer brought into contact to each patch is selected from A, T, C or G, and allowing the protected phosphoramitide nucleoside monomers to react with the activated anchored nucleoside to form phosphite bonds; f) capping any unreacted activated anchored nucleosides to prevent them elongating further ; g) oxidizing the phosphite bonds to phosphodiester bonds; h) repeating steps d) to g) a desired number of times, that number being defined as"n"times ; i) contacting the anchored oligonucleotides with said nucleophilic base, under conditions to provide a mixed population of deprotected oligonucleotides in suspension or solution, wherein said mixed population comprises a diversity of different nucleic acid sequences.

The present invention also provides for a method of producing a double-stranded polynucleotide consisting of a first strand and a second strand, the method comprising; a) providing in solution or suspension a mixture of oligonucleotides wherein a first subpopulation of the mixture consists of oligonucleotides ("strand one oligonucleotides") which when contiguously ligated together form said first strand of the double-stranded polynucleotide and a second subpopulation of the mixture consists of oligonucleotides ("strand two oligonucleotides") which when contiguously ligated together form said second strand of the doublestranded polynucleotide, wherein each strand one

oligonucleotide has a 5'portion complementary to a 3'portion of a first strand two oligonucleotide and/or has a 3'portion complementary to a 5'portion of a second strand two oligonucleotide, and each strand two oligonucleotide has a 3' portion complementary to a 5'portion of a first strand one

oligonucleotide and/or a 5'portion complementary to a 3' portion of a second strand one oligonucleotide; b) heating the mixture to melt the oligonucleotides within the mixture; c) adding a thermostable ligase to the mixture; and d) gradually cooling the mixture to produce said double-stranded polynucleotide.

Preferably the strand one and strand two oligonucleotides are the same length so the respective 5'and 3'portions correspond each to half of the relevant oligonucleotides.

For example, each of the oligonucleotides may be 40 nucleotides in length, so each of said 5'and 3'portions that are complementary to corresponding 3'and 5'portions of oligonucleotides of the other strand are 20 nucleotides. The oligonucleotides may be considered to"overlap"and provide a "ladder"or"zipper"that assembles into the full-length double-stranded polynucleotide.

In the subpopulations of oligonucleotides for synthesis of the two strands of the double-stranded polynucleotide, most of the oligonucleotides anneal to two oligonucleotides of the other strand. Thus, each of these strand one oligonucleotides has a

5'portion complementary to a 3'portion of a first strand two oligonucleotide and a 3'portion complementary to a 5'portion of a second strand two oligonucleotide, the first and second strand two oligonucleotides being adjacent to one another when assembled into the second strand of the double-stranded polynucleotide. However, except when making a circular double-stranded DNA, such as a plasmid vector, there are always two exceptions in the mixture, two oligonucleotides which are each at an end of a strand of the polynucleotide and

which each only hybridize or anneal to one of the oligonucleotides of the other strand. These may be a 5' terminal oligonucleotide and a 3'terminal oligonucleotide, either on the same strand or on opposite strands, or may be

two 5'terminal oligonucleotides, one on each strand, or two 3'terminal oligonucleotides, one on each strand.

There are several options for the ends of the double-stranded polynucleotide, depending on whether one or both of the ends is to be blunt-ended or sticky-ended, and whether if one or both of the ends is a sticky end the overhang of the sticky end (s) is on the first strand or the second strand, or one on each ; where the double-stranded polynucleotide has two sticky ends the overhands may both be on the same strand or on opposite strands. However, in each of these possibilities there are two terminal oligonucleotides in the mixture which only anneal to one other oligonucleotide, and these may be on the same strand (one 5'terminal, one 3'terminal) or on the different strands (both 5'terminal, both 3'terminal, or one 5'terminal and one 3'terminal). The mixed population of oligonucleotides may be provided by a method of synthesis according to the present invention.

As noted, one aspect of the invention provides a method of synthesizing a population of mixed oligonucleotides on a solid support, followed by release of this population from the solid support to provide a mixture of oligonucleotides in solution or suspension.

A solid support used in embodiments of the invention may for example be plain glass, which may be preferred for some embodiments, or may be for example controlled-pore glass or polystyrene, and may be activated by derivatization to provide

amine groups on the surface. Any suitable derivitization agentmay be used, for example aminopropyltrietoxysilane.

The first protected oligonucleosides are anchored to the activated regions of the solid support via link which can be cleaved, e. g. in a nucleophilic base. This link may be a succinate ester bond. In this case, the first protected nucleotide monomers may be 3'-O-succinate derivates of the protected nucleoside monomers. These will react with the amine groups exposed on the active regions of the surface to produce a succinate ester link.

In one embodiment of the invention, the same protected nucleic acid monomer (A, T, C or G) may be anchored to all of the activated patches on the surface, or a substantial region thereof. In this case, all the oligonucleotides produced on the surface or on that portion of the surface will have the same initial nucleotide. Alternatively, the protected nucleotide monomer added to each of the patches may be controlled, such that different patches are initiated with a different protected nucleoside monomer. This allows the oligonucleotides in a single region of the surface to commence with different nucleotides, and diversity to be provided across the whole of the sequence.

The unreacted amine groups may be capped so as to prevent reaction with mononucleosides added at a later stage.

The protected nucleoside have two characterizing modifications. Firstly, the 5"phosphate is replaced by a blocking group to prevent uncontrolled elongation of the oligonucleotide strand. This blocking group is removed from the 5'terminus of the elongating strand at each cycle of

elongation, so as to activate the strand, allowing a further protected monomer to be added to the growing end. In one embodiment of the invention, this protecting group is DMT, which is removed by an activating agent, for example tetrazole, to leave a 5'hydroxyl group. A protected phosphoramitide-nucleoside monomer can then react with this hydroxyl group to form a phosphite linkage. This linkage is subsequently oxidized (for example, in a mixture of 12, H2O, pyridine and tetrahydrofuran) to produce a phosphotriester bond.

Secondly, the nucleobases of the protected monomers are preferably linked to a protecting group via a link which is cleavable in a nucleophilic base. Any suitable protecting group can be used, provided that they can be removed by contacting the protected nucleosides with a nucleophilic base. The protecting groups may be, for example, benzoyl (to protect A and C mononucleosides), acetyl (to protect C mononucleosides) isobutyryl (to protect G mononucleosides) or dimethylformamidine (to protect G mononucleosides).

Where the link anchoring the growing oligonucleotide to the solid surface and the link connecting the nucleobase to the protecting group are both cleavable in a nucleophilic base, the oligonucleotides will be released from the solid surface during the deprotection step. The combination of this chemistry and the array format provides an efficient method of producing as a solution or suspension a mixed population of oligonucleotides having a number of different sequences.

According to the various preferred embodiments of the present invention, the population of oligonucleotides comprises a number of different oligonucleotide species which may be about

or more than about 5,10, 15,20, 25,30, 35,40, 50,60, 70, 80,100, 120, 150, 180,200, 250,300, 400,500, 1,000 or 2,000 different oligonucleotides. For example, where the mixed population is to be used in synthesis of a doublestranded polynucleotide, the number will depend on the components to be included. A typical protein encoding sequence may be around 2 kb, requiring about 100 oligonucleotides of 40 nucleotides each. A typical promoter may be around 500 bp, corresponding to 25 40-mers. A typical construct for gene expression including a promoter, protein encoding sequence and polyA sequence may be around 5kb, corresponding to 250 40-mers. A typical gene targeting construct (e. g. for use in making a knockout mouse) may be around 8kb of genomic sequence, or about 400 40-mers. A typical library encoding a protein family may be around 20 different protein encoding sequences of around 2 kb each, requiring about 2,000 40-mers. Of course, the number of different oligonucleotides in the mixture, when to be used for polynucleotide synthesis, depends on the length of the oligonucleotides.

The maximum number of different sequences which may be produced is limited by the number of spots on an array; using the array technology which is currently available, up to approximately 20,000 different sequences may be produced.

In order to add nucleosides to the growing chain, the anchored nucleoside is first activated by removing the 5'terminal block, and then protected, e. g. by adding a phosphoramitide nucleoside monomer. This protected nucleoside monomer may be A, T, C or G.

In at least some rounds of elongation, different subsets of patches are contacted with different nucleoside monomers (A, T, C or G), so as to produce oligoncleotides with different sequences.

In addition, it may be desirable to perform some rounds of elongation in which a mixture of protected oligonucleotide monomers (i. e. , a mixture of nucleosides selected from A, T, C and G) is applied to one or more patches. This provides an oligonucleotide with a degenerate sequence at that nucleotide position. This may be particularly desirable if the oligonucleotides represent a library of variants on a given sequence or sequences, as described below.

Thus, the method may further comprise: activating some or all of the discrete patches of anchored nucleosides by removing the 5'first protecting group; adding to some or all of the patches a mixture of protected phosphoramitide nucleoside monomers selected from A, T, C or G, and allowing the protected phosphoramitide nucleoside monomers to react with the activated anchored nucleoside to form a phosphite bond; capping any unreacted activated anchored nucleosides to prevent them elongating further; oxidizing the phosphite bond to a phosphodiester bond; and repeating a number of times, wherein that number is defined as"y", subsequent to activating some or all of the discrete patches of anchored nucleosides by removing the 5' first protecting group, but prior to contacting the anchored oligonucleotides with a nucleophilic base.

In some embodiments of the invention, it may be desirable to produce oligonucleotides of different lengths. If so, some oligonucleotides should not be elongated during one or more rounds of the process. The patches which will not be elongated during that round may not be activated, and/or may not have monomer added to them.

Alternatively, it may be preferable to make all the oligonucleotides the same length. This is particularly true if the oligonucleotides are subsequently to be bulk purified. In this case, all of the anchored oligonucleotides should be elongated during each round of the process (i. e. , they should be activated and exposed to a protected monomer). As a result, all of the oligonucleotides will have a length of n+1 nucleotides, where no degenerate nucleotides are introduced, or n+l+y if one or more rounds of elongation are performed wherein a degenerate nucleotide position is introduced.

A mixed addition is possible at each step. At each cycle of additions, there is a choice between adding a single monomer or a mixture. Oligonucleotides may be generated of the same length, but some patches (e. g. where y > 0) may carry populations of oligonucleotides which differ in the positions where mixes were added.

Differential activation of patches, and/or differential reaction with protected phosphoramitide nucleoside monomers, may be achieved in a variety of ways.

In one method, differential activation takes place by means of a light-directed-photolithographic method (US patent 5,445, 934). According to this method, the 5'blocking group is light sensitive. Light is applied to the chip through a mask,

such that only a subset of patches of anchored oligonucleoside are exposed to the light and hence activated. The whole of the chip is then flushed with a particular phosphoramitide nucleoside monomer. The process in then repeated for each monomer sequentially, for each round of elongation.

More preferably, the differential activation and/or elongation takes place using an ink-jet-printing device. Ink-jet printing has the advantage of providing for efficient and fully automated production of an array of oligonucleotides, and also of having very low reagent consumption.

In ink-jet printing techniques, the activation of the anchored oligonucleotides is mediated by an activating agent.

An ink-jet printing device for use in embodiments of the invention comprises at least 5 heads, four of which each provide a protected phosphoramitide nucleoside monomer selected from A, C, G and T, and one providing activating agent. The 5'blocking group may be DMT, and the activating agent may be tetrazole. The ink-jet printing device may also comprise a further head delivering a mixture of A, T, C and/or G protected phosphoramitide nucleosides, to produce a degenerate nucleotide position, as described above.

Preferred, though is that mixes are added by ejecting nucleosides from multiple heads onto the same patch at the same time, so an extra head is unnecessary.

An ink-jet printer is generally computer controlled, the computer being programmed to operate synthesis in accordance with the present invention. The computer may receive input data corresponding to desired sequences of oligonucleotides to be synthesized. As discussed, the oligonucleotides may

together provide sub-populations that can be assembled into one or more double-stranded polynucleotides in accordance with further aspects of the invention.

When using the ink-jet printing device, a regular microscope glass slide provided with a pattern of hydrophobic wells may be used as a support. These hydrophobic wells may be prepared according to the general protocol of Blanchard et al. (1996).

An illustrative example according to this protocol follows: a. Photoresist is sprayed evenly on the surface. b. A film carrying four 1 x 1 cm patterns of 100 x 100 urn circles separated by 40 um is placed on top of the photoresist. The film can be made as follows: i. The pattern is designed in a computer; ii. The pattern is transferred to a photographic negative using a digital film printer (e. g.

Polaroid ProPalette, capable of producing-4 um pixels); iii. The film is developed and dried. c. Each 1 x 1 cm pattern is called a"unit"and contains about 5000 black circular patches. d. The photoresist is exposed to UV light through the film. e. The resist is developed in NaOH (12 g in 1 1 H20), exposing the glass surface between the patches. f. (tridecafluorotetrahydrooctyl)-triethoxysilane is vapor-deposited on the exposed surface, producing a highly hydrophobic coating between the patches. g. The remaining resist is removed by washing in solvent (e. g. ethanol or acetone).

Subsequently, the patches may be activated by derivatization with aminopropyltrietoxysilane (or any other common support derivatization), exposing an amine group, as discussed. Then, 3'-0-succinate derivatives of the protected nucleosides may be reacted to the exposed amine groups.

If it is desired that the first nucleoside should be varied between neighbouring patches, so as to give complete control over the entire sequence of the oligonucleotide, then it may be necessary to include in the ink-jet printing device further heads, capable of dispensing 3'-0-succinate derivates of each of the protected nucleotide monomers A, T, C and G.

Alternatively, the slide may be pre-treated with a 3'-0succinate derivate of a nucleoside monomer, so as to give the same first anchored monomer for each of the patches on the slide or on a substantial portion thereof.

Using an ink-jet printing method as described above, it is straightforward to synthesize for example 4 x 5000 patches, each 0.01 mm2 (100x100 um) in a matter of hours on a regular glass slide. Each patch contains some 2.5 fmol of product. A mix of 20 000 40-mers is capable of representing the two strands of 400 kb of DNA, which might for example be a library of 400 members of a family of 1 kb genes, etc. If all patches are not required for the downstream process, a basic pattern can be repeated across the support; this would allow some increase in yield (e. g. from 1 to 400 fmol if only 1 kb is synthesized).

After the oligonucleotides have been synthesized, they may be phosphorylated. Preferably, this phosphorylation is chemical, since chemical phosphorylation provides a higher efficiency

than enzymatic phosphorylation. This may be achieved using standard phosporamitide chemistry (e. g., PhosphalinkTM Applied Biosystems), flushed across the array. Chemical phosphorylation is particularly desirable if the oligonucleotides synthesized by a method as disclosed herein are to be used subsequently in a method of gene assembly, since incorporation of unphosphorylated oligonucleotides into a product blocks the subsequent extension of that product, reducing the yield of the full length nucleotide sequence.

Following synthesis of full-length anchored oligonucleotides, and optionally phosphorylation of the anchored oligonucleotides, the oligonucleotide mix can be released by contacting the anchored oligonucleotides with a nucleophilic base, such as NH40H. The oligonucleotides may then be deprotected as a mixture in the same solution at a higher temperature; i. e. , the links between the nucleobase and its protecting group may be cleaved by exposure to the nucleophilic base, so as to make the nucleobases available for subsequent hybridisation.

After oligonucleotides have been released from the surface, they may be bulk purified according to any suitable method, for example HPLC and/or PAGE. Performing these purification methods on the bulk product significantly reduced costs, while providing a product pure enough for cloning or other techniques of molecular biology.

Subsequently, these mixtures of oligonucleotides may be used for any desired purpose. This may include use of the mixed oligonucleotides as primers in PCR reactions. A mixed pool of oligonucleotides may be useful if it is desirable to amplify simultaneously more than one double-stranded nucleotide;

alternatively, the mixed pool of oligonucleotides may represent degenerate primers.

The oligonucleotides may also be used in a method of in vitro selection, as described in more detail below.

In addition, the mixed oligonucleotides may be used in a method of gene assembly. This may be any one of the suitable methods available to the skilled person. It may be a method of the present invention as described below.

Also provided by the present invention is an apparatus for use in a method of oligonucleotide synthesis as described above, the apparatus comprising an ink-jet printer under computer processor control, the computer programmed to control the inkjet printer to synthesize a mixture of oligonucleotides in accordance with a method disclosed herein, and the ink-jet printer comprising reservoirs containing the reagents required for the synthesis, wherein a reservoir is provided containing each of the following reagents: monomers A, T, C and G, and activator; optionally also one or more synthesis chemicals such as capping fluid, oxidization fluid and detritylation fluid, deprotection fluid (e. g. NH4OH) ; wherein each reservoir is operatively connected to a nozzle of the ink-jet printer for delivery of the reagent to an array for synthesis of a mixture of oligonucleotides under control of the computer processor. A collection tube for the mix may also be included.

As discussed, according to another aspect of the invention, there is provided a method of assembling one or more doublestranded polynucleotides from a mixed population of singlestranded oligonucleotides. The method comprises providing in

solution or suspension a mixture of oligonucleotides, which are designed to correspond to contiguous sections of the upper and lower strand of one or more double-stranded nucleic acids.

For each of the nucleic acids to be assembled, the mixture contains at least two oligonucleotides corresponding to contiguous sections of the upper strand and at least 2 oligonucleotides corresponding to contiguous sections of the lower strand, although generally the number of oligonucleotides is greater as already discussed.

I The oligonucleotides assemble following heating of the mixture to a sufficiently high temperature to denature all secondary structure, as discussed. Thermostable ligase is added and the mixture is then cooled gradually preferably once, but optional more than once e. g. two or three times, to allow for slow annealing and ligation.

During this process of slow annealing, high melting-point oligonucleotide pairs tend to anneal and ligate first (preventing them from misannealing at lower, and for them, less stringent temperatures). As the temperature is progressively lowered, oligonucleotide pairs with progressively lower melting points anneal and ligate. Thus, the method provides for step-by-step gene assembly, which reduces the potential for mismatches, but at the same time provides for a method which is not labour intensive, since the successive steps are mediated by a progressive reduction in temperature.

According to a further advantage of the invention, the high temperature of the reaction reduces the potential for the formation of self-hybridised secondary structures in the

oligonucleotides, particularly in the case of long oligonucleotides.

It is preferred that the oligonucleotides used as the starting point for gene assembly are at least 16 nucleotides in length, and they may be about 20 or 25 nucleotides. More preferably, they are at least 30 nucleotides in length, and may be about or at least about 40,50 or 60 nucleotides in length. The oligonucleotides may have a length of about 100 nucleotides. A length of 30-50 nucleotides may be preferred, e. g. about 40.

A double-stranded nucleic acid obtained by a method of the invention may be, or may be at least, about 50 base pairs in length, possibly about or at least about 100,200, 300 or about or at least about 500 or 1000 base pairs in length, or more.

The feasibility of assembling double-stranded nucleic acid sequences of 10kb can be demonstrated by theoretical considerations. Thermostable ligases are capable of achieving an error rate as low as 1/1000.500 ligation reactions would be needed to assemble a 10kb double-stranded sequence from 40bp single-stranded oligonucleotides. Thus, the success rate of the reaction, and the yield of full length product, may be as high as 0. 999500, or 60. 6%.

It will be understood that the method can easily be used to assemble 10 genes of lkb in a single step or to assemble 1 gene of 10kb, with the same degree of accuracy.

A significant advantage of this method is that, when used in combination with the method of synthesizing a population of mixed oligonucleotides in solution or suspension, the process

of gene synthesis may be largely, if not entirely, automated from start to finish.

It may be preferred that the oligonucleotides used in a method of gene assembly as described herein are of approximately the same length. Preferably, the range of lengths of the oligonucleotides is not greater than 10 nucleotides, more preferably not greater than 5 nucleotides, still more preferably not greater than 2 nucleotides. It may be that the oligonucleotides are all of exactly the same length. This tends to standardize the melting point of the oligonucleotide pairs, reducing the range of temperatures though which the mixture must be cooled in order to anneal all the oligonucleotides to their complementary sequences. Similarly, sequences with a high content of AT nucleotides can have a significantly lower melting point than other oligonucleotide pairs; it may be desirable to standardize the differences in melting point caused by this effect by adding tetramethyl ammonium chloride (TMAC) to stabilize AT bonds.

The double-stranded nucleic acid may be a produced as an insert in a vector. This vector may be, for example, a plasmid or a viral genome. In one embodiment of the invention, the vector is assembled from a series of oligonucleotides, together with and in the same way as the insert sequence. In another embodiment, a vector which comprises two terminal single-stranded overhangs at the desired insertion site is added to the oligonucleotide mixture before heating. These overhangs are complementary to singlestranded overhangs of the assembled double-stranded nucleic acid, allowing the double-stranded nucleic acid to be annealed to and ligated into the vector.

In order to be compatible with hybridisation at high temperatures, it is preferred that the terminal singlestranded overhangs of the vector are 8 nucleotides or more in length. Still more preferably, they may be at least 15 nucleotides in length.

Vectors with these long overhangs may be produced by a variety of methods. In one method, vectors may be used which carry suitably placed recognition sites for a nicking enzyme, such as N. BstNBI (New England Biolabs), which cleaves one strand of DNA. If recognition sites are placed on opposite strands a staggered cut is obtained with an offset determined by the spacing of the recognition sites. This however requires that the vectors be specially designed for the purpose, and almost all commercial plasmid vectors, for example, carry an N. BstNBI site in their backbone which would have to be removed. Such sites are common enough in natural sequences to make the approach impractical in larger vectors such as BACs or viruses/phages.

As an alternative, a linearised vector with long, singlestranded overhangs may be constructed from two separate pools of vector, wherein the first pool of vector is subjected to restriction enzyme digestion with restriction enzymes A and B, and a second pool of vector is subject to restriction enzyme digestion with restriction enzymes B and C. To produce vector with two single-stranded overhangs of a number of nucleotides defined as"x", the restriction sites of enzymes A, B and C should be x nucleotides apart. If the restriction sites for enzymes A, B and C are unevenly spaced, the two overhangs of the linearised vector will be of different lengths.

In order to assemble a linearised vector with long, singlestranded overhangs, the two pools of restriction digested vectors are mixed. They are then heated to melt, and cooled to anneal. Where a strand of a vector cut with restriction enzymes A and B reanneals to a strand of a vector cut with restriction enzymes B and C, a product bearing long, terminal overhangs will be formed.

This method is especially suited to make use of the multiple cloning sites in existing plasmids, BACs, cosmids or viruses.

It may be preferred that restriction enzymes A, B and C are Type II restriction enzymes, which cut double-stranded DNA so as to leave a blunt end. If the restriction enzymes cut so as to leave a"sticky end", with a short overhang, the terminal regions of the long single-stranded overhangs from the two different vectors will be complementary, and thus the vector will have a tendency to circularise at low temperatures. At the high temperatures used in the method of gene synthesis, however, any short regions of affinity at the ends of the long overhangs will be melted apart and the vector will be fully linearised.

Vectors carrying the desired overhangs may be affinity purified using e. g. a solid support (sepharose, magnetic beads, plastic beads, etc) with an attached complementary sequence, in much the same way as polyA+ mRNA is isolated using an oligo-d sequence.

In a third approach, vectors with long, single-stranded overhangs may be produced by cutting a vector with a Type II restriction enzyme, which leaves a short single-stranded overhang, or"sticky end". Then, a single-stranded linker

sequence of the desired length, having a terminal sequence complementary to the short overhang, may be annealed and ligated to the vector. This may take place at regular ligation temperature (i. e. 4-37 C), after which the activated vector may be used at elevated temperatures in method of polynucleotide assembly according to the invention. This approach may be particularly useful when only a single cloning site is available.

In other embodiments the double-stranded nucleic acids produced in a method of the invention may be cloned into a vector as a subsequent step.

A vector employed in an embodiment of the invention may be, for example, a viral genome, or a plasmid. To create a modified viral vector carrying a desired construct, the construct may be made in a plasmid first or may be made directly in the viral genome. In the latter case the naked DNA may be transfected into a suitable host and live viral particles recovered. At least for small viruses, direct synthesis may be preferable.

Once vectors containing the double-stranded nucleic acid sequence have been obtained, they may be used to transform cells. These cells may be bacterial or eukaryotic.

Transformation of cells may be useful for a variety of applications, including protein expression, gene therapy and vaccine production, or the production of transgenic or knockout non-human animals.

The ability to assemble a double-stranded nucleic acid from oligonucleotides can be useful in a variety of contexts. The

oligonucleotides may have been synthesized de novo according to a method of the invention.

Cloning. If the sequence of a desired gene (or in general, any nucleic acid sequence) is known, it can easily be assembled from short oligonucleotides according to a method of the invention. For genes which have already been cloned, and their sequence published, it may be easier and quicker to assemble the gene than to obtain it from a supplier.

Furthermore, the method may be used to provide a clone of a gene predicted from genome sequencing (or EST sequencing etc), which may never previously have been physically isolated and cloned. This can be especially important for genes which are difficult to clone (i. e. rare transcripts or GC-rich genes).

Subcloning. For various applications in molecular biology, it is necessary to construct sequences containing not just a coding sequence, but also elements such as promoters, poly-A sites, enhancers and other modifiers. At present, such constructs are created by stepwise application of digestion (with restriction enzymes) and ligation, sometimes also requiring in situ mutagenesis. Using the present invention, such constructs can be designed in silico and directly assembled in one step.

In situ mutagenesis. Often, it is necessary to modify a gene at a single site (e. g. when changing the corresponding amino acid, introducing restriction enzyme sites or repairing or introducing mutations). The present invention allows the modified sequence to be produced by assembling the gene de novo from oligonucleotides of the desired sequence. Even if the sequence of the gene is not known, it may often be easier

and less expensive to sequence it and then reassemble the modified sequence, than to perform in situ mutagenesis.

Codon optimization. When expressing eukaryotic genes in prokaryotes (and vice versa), one usually obtains suboptimal yields because of codon usage differences. By changing the codons of the protein to match the codon usage of the target organism, yields can be much improved. This can easily be achieved with the present invention.

Gene regulatory circuits. Since the present invention permits an extremely rapid design/assembly cycle with complete control over the placement of genes, promoters etc. , more complex constructs than are now common can be created. For gene expression, especially in gene therapy, one may create (in for example a virus, see above) a whole regulatory circuit consisting of a number of regulatory genes controlling one or more target genes. A bistable switch (Gardner, 2000) and an oscillator (Elowitz, 2000) have been constructed, and other kinds of circuits can easily be imagined, e. g. logic gates, clocks, amplifiers, filters etc. Such genetic circuits can be stored and exchanged as data structures where a protein of interest is simply plugged-in before the construct is synthesized.

DNA Computers. Complex gene regulatory circuits can be built to perform calculations or logic functions. Inserted into bacterial hosts (for example) they would form powerful computational devices (Gibbons, 1997; Lee, 1999; Rozen, 1996 ; Beaver, 1995; Kari, 2000).

Library assembly. The above method represents a powerful tool for constructing entire gene families or libraries in a single

step. During synthesis, various versions of an oligonucleotides can be generated, by contacting the patches with different nucleoside monomers during particular rounds of elongation. In addition, it may be desirable to introduce one or more degenerate bases, by adding a mixture of protected phosporamitide nucleoside monomers to some or all of the patches in a given round of elongation.

In the assembly step, any one of the various alternative oligonucleotides may anneal and ligate to the elongating double-stranded nucleotide. Thus, a large variety of full length double-stranded products can be produced.

As an example, if a 2 kb gene is made from 100 40-mers, and it is desired to produce a family of genes, such that in the family as a whole 8 out of every 20 bp is represented by each possible residue, then each 40-mer must be synthesized in 64 different versions. The total number of oligonucleotides is thus 6400. However, the resulting assembled library would contain up to 8100 different variants of the gene with mutations 2.5 bp apart on average. Thus, a very large degree of diversity in the library may be obtained, while retaining control over where the diversity is allowed to arise.

In practice, it may be desirable to begin from a multiple alignment of a gene family of interest. For each amino acid position, there will be a characteristic range of variation, which will be greater in positions which have little or no impact on the final structure of the protein. For each position, a set of amino acid variants can be selected which model the characteristic natural variants. For example, in positions which are highly polymorphic (indicating that the amino acid in that position has little impact on the protein),

it may be better to model the variation simply by choosing a neutral amino acid, whereas positions with a strong bias towards a few amino acids would be modelled as exactly as possible (see e. g. Tomandl et al J Comput Aided Mol Des 11 (1) 29-38 1997). The model amino acid variability can then be translated into model triplet code variability.

Then, upper and lower strand oligonucleotides may be designed and synthesized, such that the population as a whole contains different nucleotides at each variable position. These oligonucleotides can then be assembled and cloned as described above. During assembly, any one of the variant sequences can be annealed at each stage, to produce a gene library with an enormous variety of sequences all belonging to the same general profile.

Gene Assembly for in vitro selection. Libraries assembled as described above may be used for in vitro selection.

There are currently two main approaches to in vitro selection.

One is based on fully random libraries of RNA or DNA, usually 20-400 nucleotides long (SELEX [Tuerk, 1990]). Such libraries have been generally found to contain ligands for any given epitope at more than 1 part in 1015. Multiple rounds of selection from such a library can be used to refine a selectable biological property. The nucleic acid sequence responsible for conferring this property can then be cloned, sequenced and purified.

The other main method is gene shuffling, where one starts with a family of related sequences. The starting library is usually small for practical reasons, containing on the order of five to twenty sequences. In each round of selection the library is

randomly cut with a DNAse, heated and reannealed, then PCR amplified to form unbroken molecules. After multiple rounds of selection and shuffling, chimeras can be found which improve on the pre-existing gene family members in some selectable way. However the search space in this case is small, as only a few recombination events are usually observed between the relatively small number of starting genes, and also as closely spaced polymorphisms will rarely be segregrated.

Many variants of these methods exist, but they all retain the fundamental dichotomy between the fully random starting material of SELEX and the small number of initial variants in gene shuffling.

To improve on the existing methods, either the oligonucleotide mixes produced by the present invention (as DNA or as RNA, or in general as any synthetic DNA/RNA-like oligomer) themselves, or assembled gene libraries as described above can be used for in vitro selection. In general, any existing display and selection method can be used on such oligonucleotide mixes or assembled gene library.

This approach is in some respects an ideal compromise between SELEX and gene shuffling. Very large libraries of fully designed and controlled oligonucleotides can be used as the starting point for a more focused version of SELEX. Very large libraries of all possible variants of a gene family (produced as above) may be used in the same way for direct selection, or in combination with the gene shuffling approach.

An additional advantage is that no gene family members need be physically isolated and cloned. This permits, for example, in vitro selection and gene shuffling to be extended to members

of gene families which are only predicted from genomic open reading frames.

It may be desirable to use repeated rounds of library synthesis, selection and sequencing to perform in vitro evolution with the aim of improving a desired biological property. Thus, after synthesis of an initial library (as above), selection is applied to find, say, the best 500 sequences. These are sequenced and aligned and a new library is constructed which represents variants of the 500 selected sequences. The resulting library is again large, but more refined and concentrated around the fitter sequences. If sequence space is imagined as a landscape with hills representing fitter sequences, then each cycle will produce a library converging more and more on the hilltops (Kaufman, 1993).

Further aspects and embodiments will be apparent to the skilled person in the light of the present disclosure, including with reference to the following examples.

All documents mentioned in this specification are incorporated herein by reference.

EXAMPLE 1 A set of oligonucleotides was designed wherein the oligonucleotides corresponded to contiguous segments of the whole of the upper and lower strand of the human insulin gene.

The oligonucleotides which corresponded to the end of the 5' ends of each strand of the insulin sequence also possessed "cloning arms" designed to allow the full-length gene to be inserted into a vector. These oligonucleotides were

individually synthesized by MWG Biotech, Germany, and were chemically 5'-phosphorylated.

The oligonucleotide sequences are shown below, in a 5'to 3' orientation. Lower case represents the"cloning arms" cohesive single-strand overhangs of a high-temperature rescue vector: insulin upper strand: zipl ggggtgcaggaattcgatATGGCCCTGTGGATGCGCCTCC zip2 TGCCCCTGCTGGCGCTGCTGGCCCTCTGGGGACCTGACCC zip3 AGCCGCAGCCTTTGTGAACCAACACCTGTGCGGCTCACAC zip4 CTGGTGGAAGCTCTCTACCTAGTGTGCGGGGAACGAGGCT zip5 TCTTCTACACACCCAAGACCCGCCGGGAGGCAGAGGACCT zip6 GCAGGTGGGGCAGGTGGAGCTGGGCGGGGGCCCTGGTGCA zip7 GGCAGCCTGCAGCCCTTGGCCCTGGAGGGGTCCCTGCAGA zip8 AGCGTGGCATTGTGGAACAATGCTGTACCAGCATCTGCTC zip9 CCTCTACCAGCTGGAGAACTACTGCAACTAGCGATCGACT lower strand: ziplO GCAGCGCCAGCAGGGGCAGGAGGCGCATCCACAGGGCCAT zipll TTCACAAAGGCTGCGGCTGGGTCAGGTCCCCAGAGGGCCA zipl2 GTAGAGAGCTTCCACCAGGTGTGAGCCGCACAGGTGTTGG zipl3 TCTTGGGTGTGTAGAAGAAGCCTCGTTCCCCGCACACTAG

zipl4 TCCACCTGCCCCACCTGCAGGTCCTCTGCCTCCCGGCGGG zip15 CAAGGGCTGCAGGCTGCCTGCACCAGGGCCCCCGCCCAGC zipl6 GTTCCACAATGCCACGCTTCTGCAGGGACCCCTCCAGGGC zip17 TTCTCCAGCTGGTAGAGGGAGCAGATGCTGGTACAGCATT zipl8 gacggtatcgataagcttgatAGTCGATCGCTAGTTGCAGTAG

The inventors produced a vector with single-stranded overhangs complementary to the"cloning arms"of the above oligonucleotides. pBluescript vector was cut with EcoRV/HincII and EcoRV/SmaI in separate tubes, as described in materials and methods. The products were mixed, melted and then reannealed. Where a strand of a vector cut with EcoRV/HincII reannealed with a strand of vector cut with EcoRV/SmaI, a new vector was formed having two 18/21 base pair overhangs.

The vector and oligonucleotides were mixed. The mixture contained 5fmol of each oligonucleotide, which corresponds to the amount of product that would be expected from five single spots on an ink-jet microarray. The mixture was then heated to 950C for five minutes, and then cooled to 65 C. DNA ligase was added, and the solution was cooled over the course of approximately 1 hour to 55 C, to anneal the product.

Parallel control experiments were carried out, in which the mixture lacked oligonucleotides, or lacked ligase.

All three mixtures were used to transform DHa5 cells, at three different volumes (0. 5pl, 2pl and 5ul). The cells were cultured as described in materials and methods.

When cells are transformed with pBluescript vector, blue colonies demonstrate the presence of uncut vector. Only blue colonies were observed on the control plates.

The experimental plate containing cells transformed with 0. 5pl had 5 white colonies. These white colonies were picked, and lysed, and plasmid DNA was isolated using standard procedures.

PCR analysis was then performed, and the products were separated by agarose gel electrophoresis. It was found that all of the white colonies contained a 400bp fragment, indicative of the presence of a full length human insulin insert.

Thus, this experiment demonstrates that it is possible to synthesize a 400bp gene, and ligate it into a vector, in a single step. It further demonstrated that this is possible using the very low concentrations of oligonucleotides that are expected to be present when the oligonucleotides are produced on an array, according to a method of the invention.

MATERIALS AND METHODS 1. Preparation of vector a. Cut pBluescript II KS+ with EcoRV/HincII and EcoRV/SmaI i. 5 pg in 100 pl- > 50 ng/ul- > 25 fmol/pl b. Phenol/chloroform, ethanol precipitation, resuspend in 50 pi c. Mix the two products d. Heat to 95 C for 10 minutes e. Slowly anneal to room temperature 2. Preparation of oligonucleotides a. Adjust to a single concentration (100 pmol/ul) b. Mix 5 pi of each in 100 pi total volume- > 25 fmol/pl each 3. Ligation a. Mix templates and buffer in three tubes i. 10 pi oligo mix (control: no oligonucleotides)

ii. 10 pi vector iii. 5 pi buffer iv. 24 pl H2O b. Heat to 95 C for 5 minutes c. Cool to 650C d. Add 1 pl NEB Taq DNA ligase (control: no ligase) e. Anneal slowly to 55 (e. g. during one hour) f. Phenol/chloroform, ethanol precipitation, resuspend in 10 l (i. e. 25 fmol/pl) 4. Transformation of E Coli a. Dilute 1: 25 (i. e. 1 fmol/pl) b. Transform Invitrogen Subcloning Efficiency DH50 cells ( 105 CFU/pg non-supercoiled) i. 0.5 pi (0.5 fmol- > 1 ng- > 100 colonies) ii. 2 pl (2 fmol- > 4 ng- > 400 colonies) iii. 5 l (5 fmol- > 10 ng- > 1000 colonies) c. Plate on IPTG/X-gal agar plates d. Select white colonies (blue ones are from uncut vector) 5. Rapid assessment of insert size a. Pick white colonies into PCR mix, replica to a new plate b. Amplify c. Separate on agarose d. Look for-400 bp fragment

REFERENCES 1. Chalmers & Curnow (2001) Biotechniques 30,249-52.

2. Gardner et al. (2000) Nature 403,339-42.

3. Elowitz & Leibler (2000) Nature 403,335-8.

4. Gibbons et al. (1997) Curr Opin Biotechnol 8,103-6.

5. Lee et al. (1999) Mol Cells 9,464-9.

6. Rozen et al. (1996) Curr Biol 6,254-7.

7. Beaver (1995) J Comput Biol 2,1-7.

8. Kari & Landweber (2000) Methods Mol Biol 132,413-30.

9. Hughes et al. (2001) Nat Biotechnol 19,342-7.

10. Recht et al. (1996) J Mol Biol 262,421-36.

11. Blanchard (1998) Genet Eng 20, 111-23.

12. Maskos & Southern (1992) Nucleic Acids Res 20,1679-84.

13. Maskos & Southern (1992) Nucleic Acids Res 20,1675-8.

14. Matson et al. (1995) Anal Biochem 224,110-6.

15. Weiler & Hoheisel (1996) Anal Biochem 243,218-27.

16. Tuerk & Gold (1990) Science 249,505-10.

17. Kauffman (1993) The origins of order (Oxford University Press, Oxford).

Claims

CLAIMS : 1. A method of producing a mixture of oligonucleotides in suspension or solution, the method comprising: synthesizing a population of oligonucleotides as an array on a solid support; cleaving the population from the solid support to provide the mixture in suspension or solution.
2. A method according to claim 1 wherein the oligonucleotides are the same length.
3. A method according to claim 1 or claim 2, comprising: a) activating discrete patches of the surface of a solid support to provide an organised array of activated patches ; b) contacting each activated patch with one or more protected nucleoside monomers, wherein a protected nucleoside monomer comprises a 5'blocking group and a protecting group linked to the nucleobase via a bond which is cleavable in a nucleophilic base, and wherein the protected nucleoside monomer added to each patch is selected from A, T, C or G, or a mixture thereof; c) anchoring a protected nucleoside monomer in contact with an activated patch to the activated patch surface via a bond which is cleavable in a nucleophilic base, to provide an array of patches of anchored nucleosides; d) activating some or all of the patches of anchored nucleosides by removing the 5'blocking group; e) contacting some or all of the patches with protected phosphoramitide nucleoside monomer, wherein the protected phosphoramitide nucleoside monomer brought into contact to each patch is selected from A, T, C or G, and allowing the

<Desc/Clms Page number 51>

protected phosphoramitide nucleoside monomers to react with the activated anchored nucleoside to form phosphite bonds ; f) capping any unreacted activated anchored nucleosides to prevent them elongating further; g) oxidizing the phosphite bonds to phosphodiester bonds; h) repeating steps d) to g) a desired number of times, that number being defined as"n"times ; i) contacting the anchored oligonucleotides with said nucleophilic base, under conditions to provide a mixed population of deprotected oligonucleotides in suspension or solution, wherein said population comprises a diversity of different nucleic acid sequences.
4. A method according to claim 3 wherein the protected nucleoside monomer is anchored to the activated patches via a succinate ester bond.
5. A method according to claim 3 or claim 4 wherein the protecting group is a benzoyl, acetyl, isobutryl or dimethlyformamidine group.
6. A method according to any one of claims 3 to 5 wherein anchored oligonucleotides consist of a number of nucleotides, which number of nucleotides corresponds to n+1.
7. A method according to any one of the preceding claims, wherein the population of oligonucleotides comprises at least 50 different nucleic acid sequences.
8. A method according to claim 7 wherein the population of oligonucleotides comprises at least 500 different nucleic acid sequences.

<Desc/Clms Page number 52>
9. A method according to any one of claims 3 to 8, further comprising: j) activating some or all of the discrete patches of anchored nucleosides by removing the 5'first protecting group; k) adding to some or all of the patches a mixture of protected phosphoramitide nucleoside monomers selected from A, T, C or G, and allowing the protected phosphoramitide nucleoside monomers to react with the activated anchored nucleoside to form a phosphite bond; 1) capping any unreacted activated anchored nucleosides to prevent them elongating further; m) oxidizing the phosphite bond to a phosphodiester bond; and repeating a number of times defined as"y", subsequent to activating some or all of the discrete patches of anchored nucleosides by removing the 5'first protecting group, but prior to contacting the anchored oligonucleotides with a nucleophilic base.
10. A method according to claim 9 wherein anchored oligonucleotides consist of a number of nucleotides, that number corresponding to n+l+y.
11. A method according to any one of the preceding claims comprising phosphorylating the oligonucleotides prior to the cleaving.
12. A method according to claim 11 comprising chemically phosphorylating the oligonucleotides, for example by contacting the anchored oligonucleotides with a phosphoramitide.

<Desc/Clms Page number 53>
13. A method according to claim 11 or claim 12 further comprising assembling oligonucleotides in the mixture into a double-stranded polynucleotide.
14. A method according to claim 13 comprising assembling the oligonucleotides by: denaturing by heating the oligonucleotides to a temperature sufficient to eliminate annealing between oligonucleotides in the mixture and secondary structure within the oligonucleotides, cooling the denatured oligonucleotides in the presence of a thermostable ligase to provide a double-stranded polynucleotide.
15. A method according to claim 14 wherein heating is to a temperature of at least about 95 C.
16. A method according to claim 14 or claim 15, wherein cooling is down to 55 C over a period of about two hours.
17. A method of producing a double-stranded polynucleotide consisting of a first strand and a second strand, the method comprising; a) providing in solution or suspension a mixture of oligonucleotides wherein a first subpopulation of the mixture consists of oligonucleotides ("strand one oligonucleotides") which when contiguously ligated together form said first strand of the double-stranded polynucleotide and a second subpopulation of the mixture consists of oligonucleotides ("strand two oligonucleotides") which when contiguously ligated together form said second strand of the doublestranded polynucleotide, wherein each strand one

<Desc/Clms Page number 54>

oligonucleotide has a 5'portion complementary to a 3'portion of a first strand two oligonucleotide and/or has a 3'portion complementary to a 5'portion of a second strand two oligonucleotide, and each strand two oligonucleotide has a 3' portion complementary to a 5'portion of a first strand one oligonucleotide and/or a 5'portion complementary to a 3' portion of a second strand one oligonucleotide; b) heating the mixture to a melt the oligonucleotides within the mixture; c) adding a thermostable ligase to the mixture; and d) cooling the mixture to produce said double-stranded polynucleotide.
18. A method according to claim 17 wherein the strand one and strand two oligonucleotides are at least 16 nucleotides in length.
19. A method according to claim 18 wherein the strand one and strand two oligonucleotides are 40 nucleotides in length.
20. A method according to any one of claims 13 to 19 wherein the oligonucleotides have a range of lengths, which range does not exceed 10 nucleotides.
21. A method according to any one of claims 13 to 20, wherein the double-stranded polynucleotide is at least about 300bp in length.
22. A method according to any one of claims 13 to 21 wherein the double-stranded polynucleotide is a vector.

<Desc/Clms Page number 55>
23. A method according to any one of claims 13 to 21 comprising ligating the double-stranded polynucleotide to one or more vectors.
24. A method according to claim 23 wherein the ligating follows annealing of single-stranded overhangs of the doublestranded polynucleotide and vector or vectors.
25. A method according to claim 24 wherein the single-stranded overhangs comprise at least 8 nucleotides.
26. A method according to claim 20 wherein the one or more vectors is or are provided by subjecting a first sample of a vector to restriction by first and second restriction enzymes, and subjecting a second sample of a vector to restriction by second and third restriction enzymes, wherein said first, second and third restriction enzymes are different, and wherein the restriction sites of said first, second and third restriction enzymes are at least 8 nucleotides apart, mixing the samples, heating the mixture of samples to melt the strands and then reannealing to provide vectors with singlestranded over-hangs at least 8 nucleotides in length.
27. A method according to any one of claims 22 to 26 which further comprises transforming the vector or vectors into cells.
28. A computer processor programmed to control a method of producing a mixture of oligonucleotides according to any one of claims 1 to 12.
29. A polynucleotide obtained in accordance with the method of any one of claims 1 to 23.

29. A computer processor programmed to control a method of producing a mixture of oligonucleotides and assembling a

<Desc/Clms Page number 56>

double-stranded polynucleotide according to any one of claims 13 to 16.

30. A computer-readable device carrying a program for a computer processor according to claim 28 or claim 29.

31. A method for obtaining and optionally providing to a customer a desired polynucleotide, the method comprising providing, optionally by accepting as a customer order, input data comprising a polynucleotide sequence and/or identified vector or gene construct components, inputting the input data into a programmed computer processor, synthesizing a mixture of oligonucleotides in accordance with the method of any one of claims 1 to 12 under control of the computer processor, preferably using an ink-jet printer, assembling the polynucleotide from oligonucleotides within the mixture in accordance with the method of any one of claims 13 to 26, whereby the desired polynucleotide is obtained, and may be supplied to a customer.

32. A method according to claim 31 wherein one or more nucleotide sequences forming components of the desired polynucleotide sequence is retrieved by the computer processor and used in design of the oligonucleotides to be synthesized and then assembled.

33. A system comprising an ink-jet printer under control of a computer processor programmed for control of performance of a method according to any one of claims 1 to 12, reservoirs of the ink-jet printer each containing reagents required for performing the method and each operatively connected to nozzles of the ink-jet printer, and a reaction chamber of adjustable temperature for performance of a method according

<Desc/Clms Page number 57>

to any one of claims 13 to 26 also under control of the computer processor.

34. A mixture of oligonucleotides obtained in accordance with the method of any one of claims 1 to 12.

35. Use of a mixture of oligonucleotides according to claim 34 in assembly of a double-stranded polynucleotide.

36. A polynucleotide obtained in accordance with the method of any one of claims 13 to 26.

<Desc/Clms Page number 58>

CLAIMS :

1. A method of producing a double-stranded polynucleotide consisting of a first strand and a second strand, the method comprising: (i) synthesizing a population of oligonucleotides as an array on a solid support ; (ii) phosphorylating the oligonucleotides in the array on the solid support; (iii) cleaving the population from the solid support to provide a mixture of oligonucleotides in suspension or solution; wherein a first subpopulation of the mixture consists of oligonucleotides ("strand one oligonucleotides") which when contiguously ligated together form said first strand of the double-stranded polynucleotide and a second subpopulation of the mixture consists of oligonucleotides ("strand two oligonucleotides") which when contiguously ligated together form said second strand of the double-stranded polynucleotide, wherein each strand one oligonucleotide has a 5'portion complementary to a 3'portion of a first strand two oligonucleotide and/or has a 3'portion complementary to a 5' portion of a second strand two oligonucleotide, and each strand two oligonucleotide has a 3'portion complementary to a 5'portion of a first strand one oligonucleotide and/or a 5' portion complementary to a 3'portion of a second strand one oligonucleotide ; (iv) heating the mixture to a melt the oligonucleotides within the population; (v) adding a thermostable ligase to the mixture ; and (vi) cooling the mixture to produce said double-stranded polynucleotide.

<Desc/Clms Page number 59>

2. A method according to claim 1 wherein heating is to a temperature of at least 952C.

3. A method according to claim 1 or claim 2, wherein cooling is down to 552C over a period of about two hours.

4. A method according to any one of claims 1 to 3 comprising chemically phosphorylating the oligonucleotides prior to cleaving, for example by contacting the oligonucleotides in the array on the solid support with a phosphoramitide.

5. A method according to claim any one of claims 1 to 4 wherein the strand one and strand two oligonucleotides are at least 16 nucleotides in length.

6. A method according to claim 5 wherein the strand one and strand two oligonucleotides are 40 nucleotides in length.

7. A method according to any one of claims 1 to 6 wherein the oligonucleotides are the same length.

8. A method according to any one of claims 1 to 6 wherein the oligonucleotides have a range of lengths, which range does not exceed 10 nucleotides.

9. A method according to any one of claims 1 to 8, wherein the double-stranded polynucleotide is at least 300bp in length.

10. A method according to any one of claims 1 to 9 wherein the double-stranded polynucleotide is a vector.

<Desc/Clms Page number 60>

11. A method according to anyone 0 : claims 1 to 9 comprising ligating the double-stranded polynucleotide to one or more vectors.

12. A method according to claim 11 wherein the ligating follows annealing of single-stranded overhangs of the double- stranded polynucleotide and vector or vectors.

13. A method according to claim 12 wherein the single-stranded overhangs comprise at least 8 nucleotides.

14. A method according to claim 11 wherein the one or more vectors is or are provided by subjecting a first sample of a vector to restriction by first and second restriction enzymes, and subjecting a second sample of a vector to restriction by second and third restriction enzymes, wherein said first, second and third restriction enzymes are different, and wherein the restriction sites of said first, second and third restriction enzymes are at least 8 nucleotides apart, mixing the samples, heating the mixture of samples to melt the strands and then reannealing to provide vectors with singlestranded over-hangs at least 8 nucleotides in length.

15. A method according to any one of claims 10 to 14 which further comprises transforming the vector or vectors into cells.

16. A method according to any one of claims 1 to 15 wherein the synthesizing of a population of oligonucleotides as an array comprises: a) activating discrete patches of the surface of a solid support to provide an organised array of activated patches;

<Desc/Clms Page number 61>

b) contacting each activated patch with one or more protected nucleoside monomers, wherein a protected nucleoside monomer comprises a 5'blocking group and a protecting group linked to the nucleobase via a bond which is cleavable in a nucleophilic base, and wherein the protected nucleoside monomer added to each patch is selected from A, T, C or G, or a mixture thereof; c) anchoring a protected nucleoside monomer in contact with an activated patch to the activated patch surface via a bond which is cleavable in a nucleophilic base, to provide an array of patches of anchored nucleosides ; d) activating some or all of the patches of anchored nucleosides by removing the 5'blocking group; e) contacting some or all of the patches with protected phosphoramitide nucleoside monomer, wherein the protected phosphoramitide nucleoside monomer brought into contact to each patch is selected from A, T, C or G, and allowing the protected phosphoramitide nucleoside monomers to react with the activated anchored nucleoside to form phosphite bonds; f) capping any unreacted activated anchored nucleosides to prevent them elongating further; g) oxidizing the phosphite bonds to phosphodiester bonds ; h) repeating steps d) to g) a desired number of times, that number being defined as"n"times ; i) contacting the anchored oligonucleotides with said nucleophilic base, under conditions to provide a mixed population of deprotected oligonucleotides in suspension or solution, wherein said population comprises a diversity of different nucleic acid sequences for assembly into the doublestranded polynucleotide.

<Desc/Clms Page number 62>

17. A method according to claim 16 wherein the protected nucleoside monomer is anchored to the activated patches via a succinate ester bond.

18. A method according to claim 16 or claim 17 wherein the protecting group is a benzoyl, acetyl, isobutryl or dimethlyformamidine group.

19. A method according to any one of claims 16 to 18 wherein anchored oligonucleotides consist of a number of nucleotides, which number of nucleotides corresponds to n+1.

20. A method according to any one of the preceding claims, wherein the population of oligonucleotides comprises at least 50 different nucleic acid sequences.

21. A method according to claim 20 wherein the population of oligonucleotides comprises at least 500 different nucleic acid sequences.

22. A method according to any one of claims 16 ro 21, further comprising : j) activating some or all of the discrete patches of anchored nucleosides by removing the 5'first protecting group; k) adding to some or all of the patches a mixture of protected phosphoramitide nucleoside monomers selected from A, T, C or G, and allowing the protected phosphoramitide nucleoside monomers to react with the activated anchored nucleoside to form a phosphite bond; 1) capping any unreacted activated anchored nucleosides to prevent them elongating further;

<Desc/Clms Page number 63>

m) oxidizing the phosphite bond to a phosphodiester bond; and repeating a number of times defined as"y", subsequent-co ac-. ivazing some or all of the. discrete patches of anchored nucleosides by removing the 5'first protecting group, but prior to contacting the anchored oligonucleotides with a nucleophilic base.

23. A method according to claim 22 wherein anchored oligonucleotides consist of a number of nucleotides, that number corresponding to n+l+y.

24. A computer processor programmed to control apparatus performing a method of producing a double-stranded polynucleotide comprising the steps in a method according to any one of claims 1 to 23.

25. A computer-readable device carrying a program for a computer processor according to claim 24.

26. A method for obtaining and optionally providing to a customer a desired polynucleotide, the method comprising providing, optionally by accepting as a customer order, input data comprising a polynucleotide sequence and/or identified vector or gene construct components, inputting the input data into a programmed computer processor, producing a doublestranded polynucleotide by synthesizing a mixture of oligonucleotides under control of the computer processor, preferably using an ink-jet printer, and assembling the double-stranded polynucleotide from oligonucleotides within the mixture in accordance with the method of any one of claims 1 to 23, whereby the desired polynucleozide is obtained, and may be supplied to a customer.

<Desc/Clms Page number 64>

27. A method according to claim 26 wherein one or more nucleotide sequences forming components of the desired polynucleotide sequence is retrieved by the computer processor and used in design of the oligonucleotides to be synthesized and then assembled.

28. A system comprising an ink-jet printer under control of a computer processor programmed for control of performance of a method of producing a double-stranded polynucleotide according to any one of claims 1 to 23 by synthesizing a mixture of oligonucleotides under control of the computer processor using the ink-jet printer, reservoirs of he ink-jet printer each containing reagents required for performing the method and each operatively connected to nozzles of the ink-jet printer, and assembling the double-stranded polynucleotide from oligonucleotides within the mixture in a reaction chamber of adjustable temperature also under control of the computer processor.