WO2004053160A2 - Method to analyze polymeric nucleic acid sequence variations - Google Patents

Abstract

The present invention relates to methods to assess exon-intron boundaries, transcription start points as well as real experimental polyadenylation sites as utilized by a given cell, tissue or organism in a given time or situation (physiological or pathological). Self-nornmalisation (SN) provides a method to change the partial representation of transcripts in a self-normalising manner. Five prime exon rescue (FER) provides a method to analyse 5´ exon forks in sense-RNA/anti-RNA hybrids. String analysis of variant exon (SAVE) provides a method to link together complete RNA fragments resulting from RNase VI digestion. By using RNA linkers, SAVE also provides a method to generate strings of RNA fragments, to amplify them, to clone them, to sequence them and to purify them for further use as well.

Description

METHOD TO ANALYZE POLYMERIC NUCLEIC ACID SEQUENCE VARIATIONS

FIELD OF THE INVENTION

The invention relates to a method to analyse polymeric nucleic acid sequence variations, particularly but not exclusively exon-intron boundaries in an organism which utilises RNA splicing mechanisms. The invention further relates to methods for identifying nucleic acid sequences representing differences between two conditions being compared. The nucleic acid sequences so identified are useful as and in the development of screening tools for identifying molecules of therapeutic interest.

BACKGROUND OF THE INVENTION

Many cases of gene regulation through pre-mRNA splicing are classed as alternative splicing (AS). AS is an important mechanism for modulating gene function. It can change how a gene acts in different tissues and developmental states by generating distinct messenger RNA (mRNA) isoforms composed of different selections of exons. Thus, through the selection of a subset of exons, AS causes variations in the expressed protein. Alternative splicing has been implicated in many processes including sex determination, apoptosis, and acoustic tuning in the ear. Its functional implications can be either simple, generating a single alternative form, or complex, producing a remarkable diversity. In the Drosophila gene Dscam, combinatorial alternative splicing of 'cassettes' of exons produces thousands of distinct functional isoforms. This gene, homologous to the human gene for Down's syndrome cell adhesion molecule (DSCAM), appears to be involved in neuronal guidance, where such diversity could be useful as a molecular 'address' (for a recent review on the subject of alternative splicing, see Smith & Valcάrcel).

A variety of approaches have been used to study exon variation. They range from analysis of genome databases using prediction algorithms (Rrett et al), to microarray analysis using probes containing gene specific exon-intron boundaries (dubbed boundary probes) (Yeakley et al), or PCR based techniques that use boundary primers. The power of these methods however is notably limited by the self-imposed bias of the studies (only currently known or algorithm-predicted exon-intron boundaries are used in these studies). Additionally, these experimental approaches restrict their scope to expression levels of predetermined exon-intron boundaries. These boundaries are thought to be highly conserved although, in many cases, degenerate boundary sequences can be used by cells. More experimental approaches are thus needed to be able to analyze those cases. Moreover, since exon-intron boundaries can be alternatively used, more systematic approaches are needed to provide sequence definition of real experimental boundaries in physiological or pathological conditions (i.e. in normal versus cancer tissues).

Until recently, few techniques were available to look at differences in global gene expression. Mass approaches such as sequencing very large numbers of independent cDNA libraries were possible but are very expensive. A recently introduced technology, serial analysis of gene expression (SAGE), has in part overcome this problem, by generating 15 bp fragments through a specific restriction enzyme strategy. These fragments are ligated together and sequenced as a string. The frequencies of these fragments in the chimeric sequences narrowly represent an estimate of the frequencies of mRNAs in the population (Velculescu et α/). On a similar theme, Gene Calling evaluates restriction patterns of cDNA samples with multiple restriction enzymes such that each cDNA sample produces a characteristic profile. The identification and quantitation of the peaks for each cDNA molecule indicate the level of gene expression (Shimkets et al).

A new approach to analyze alternative exon expression has been recently described. The method, dubbed DATAS technology, is based on subtractive principles. It takes advantage of the ability of the enzyme ribonuclease H to release single stranded RNA (un-hybridized) fragments from mRNA/cDNA hybrids. When the hybrids correspond to two different mRNA sources, those fragments from RNase H digestion most likely represent alternatively expressed exons. In the DATAS technology, analysis of the differential RNA fragments is completed after synthesis of random primed complementary DNAs. These cDNAs are either sequenced or arrayed and used in a variety of studies that include differential display (Schweighoffer et al). The DATAS method however has several important drawbacks: 1) This method tends to miss entire populations of alternative exons. a) because it is based on the isolation of internal un-hybridized RNA loops, it does not recover 5' alternative exons (dubbed 5' forks), nor probably does 3' forks. This kind of exon represents 70 to 80% of the entire alternative exon pool. Indeed, in a recent analysis of the EST data base, alternative exons at 5' end (5' forks) occurred in 73 genes (18.5%), internal alternatives (loops) in 41 genes (10.4%), and 3' forks in 64 genes (16.2%) (Mironov et al). These figures bring the total estimates for 5' forks to up to 50% of all alternative splicing events, b) Because the technique uses non-normalized cDNA libraries as hybridization driver, exon pools obtained with the DATAS method, at best, misrepresent alternative exons from genes expressed at low levels. This is very relevant to our technique, since alternative splicing can lead to dramatic changes in protein function without an accompanying change in gene expression levels.

2) Typically, cDNA libraries, unless undergoing a major sequencing effort, do not provide reliable expression profile data, h order to draw gene expression information from cDNA libraries, they have to be modified to allow sequencing of a number of genes in a single sequencing event (for general cDNA libraries, this has been accomplished by the SAGE method).

3) The DATAS technology requires a sequencing reaction for each exonic fragment to be analyzed which, when studying a large number of splicing events, can represent a huge time and money consuming effort.

Here we describe a new technology to analyze en masse pools of differentially (or alternatively) expressed exons. This technology comprises several methods: Self- normalization of cDNA libraries (SN), Five prime Exon Rescue (FER) and String Analysis of (complete) Nariant Exons (SANE). The approach is based on the following steps:

- The self-normalization of 5 ' enriched cDΝA libraries,

- The hybridization of sense and antisense RΝA pools from different conditions,

- The isolation of full length RΝA exonic fragments of a complete spectrum of different lengths followed by the

- Ligation in chain of such fragments, cloning and sequencing of the resulting libraries.

Our approach systematically solves the major problems of current technologies. What follows is a detailed description of the technique, a list of advantages with respect to current methods employed in the study of alternative splicing as well as an account of examples of possible applications. A related application of SANE is described in our co-pending international patent application PCT/GB2003/003118, the entire contents of which are incorporated herein by reference.

SUMMARY OF THE INVENTION

The present invention comprises a method to investigate the exon-intron boundaries with a genome-wide range in any organism that utilizes RNA splicing mechanisms. This invention is ideally fitted to the comparative study of two or more sources of RNA that harbour differences in splicing, transcription initiation, poly- adenylation or RNA editing. It represents a clear advantage when compared to currently used methods to study alternative transcripts of the same gene. Since it is based on the differential hybridisation of RNA duplexes, SANE as described herein avoids dealing with DΝA. Moreover, FER permits the isolation of entire exon populations (notably 5 ' alternative exons) that are missed when using current technologies.

Because it is based on the string ligation of RΝA fragments, SANE sequences are also indicative of relative abundance of exon expression. Moreover, SANE has additional economic advantages since it provides information of multiple exons in a single sequencing event.

Practically, it starts with the isolation of total RΝA from the sources to be compared. These sources can be blood, tissue culture, tissues or whole organisms. As shown below, a typical comparison can be established between a normal tissue and its diseased counterpart (i.e. normal lung versus lung cancer).

Pools of messenger RΝA are typically composed of species that are represented in different abundances. This is a very important feature to take into account when doing gene expression profiling. These types of studies evaluate the different abundance of specific RΝA species and directly correlates the number of copies of a specific mRΝA with its functional significance in a given condition (i.e. cancer)

A differential feature of alternative splicing, however, is the fact that it can be functionally relevant (i.e. dramatically change the function of a specific protein) even if an mRΝA is present in a low or very low copy number within an mRΝA mixture. This type of species are usually missed or, at best, underrepresented, unless we use a normalization approach. For instance, cDΝA libraries can be normalised by hybridising the source of test RΝA against a pool of cDΝA from different tissues or conditions. The goal of normalisation being in that case to equalize the proportional representation of the different RNA species (i.e. the more abundant species will proportionally decrease its representation in the entire pool) by subtracting the most abundant mRNAs (Swaroopet et al, 1991). In the SN method we apply a similar principle to a different goal: to self-normalize RNA pools. Self-normalization is thus an important part of our invention. It allows us to, for example, increase the representation of less abundant species within a given RNA pool. By doing that internally in both the tRNA and dRNA pools (see page 8), we can promote (in a tRNA-dRNA mixture) the encounter and subsequent (hybrid) formation of duplexes of low copy number species which only differ in their pattern of alternative splicing.

Typical substrates of our technology are RNase VI digestion derived fragments after tRNA-dRNA hybridisation. A fundamental step in our technology is the conversion of native mRNA into amplified RNA (aRNA). This step introduces fundamental advantages in our technology. Essentially, it allows the rescue of the entire population of 5' alternative exons; this is a pool that is missed from the analytical methods used by current technologies. Working with tRNA-dRNA hybrids represents an additional advantage because, after digestion with RNase VI, alternative exons from both the tRNA and dRNA pools can be isolated in a single step. Finally, this step allow us to avoid dealing with having to eliminate DNA from DNA/RNA hybrids. At this point, un-hybridised tRNA species are retained after passing the RNase VI digests through a polyU retaining column (see below). Since RNase VI digests contain reactive 5' (phosphate) and 3' (hydroxyl) groups, they are perfect substrates for T4 RNA ligase. Thus a central step in SAVE is the chain ligation of pools of RNA fragments from the RNase VI digestion of RNA-RNA hybrids. Alternatively, these fragments can be ligated in the presence of a hinge RNA linker that is complementary to the consensus sequence for a low frequency endonuclease (e.g. Notl). The latter version of intercalated chain can be later used for the isolation, purification and sequencing of individual exonic fragments (see below). Either type of chain is later used in a second ligation reaction with 5' and 3' RNA linkers. The latter step is essential in order to PCR amplify the pool of strings. Previous to sequencing, the final step in SAVE is the cloning of exon strings into a suitable cloning vector. Individual clones are thus sequenced and the sequences are searched against different genetic databases. Typical searches are shown in figure 10 and figure 11, whereby different segments of a single string of exons display a perfect match to different genes (typical pattern). As mentioned above, this invention is also aiming at isolating and purifying individual exons from clonal strings. More specifically, strings intercalated with restriction hinges can be digested with the corresponding specific restriction enzyme and the different individual exons submitted to agarose electrophoresis and gel purification. Individual fragments can be typically spotted on solid surfaces (i.e. micro- arrays for differential display studies).

BRIEF DESCRIPTION OF THE DRAWINGS

Fi l a) Predicted formation of a single stranded 5' RNA fork when a messenger RNA that contains an alternatively spliced/transcribed 5' exon hybridises with a cDNA that misses such an exon. b) It also shows the predicted outcome of adding RNase VI resulting in the digestion of the RNA duplexes hybrids harbouring alternate 5' exons and the formation of RNA fragments chemically active at both ends. (Note: the padlock symbol represents the 3' cordycepin modification that shields the 3' ends of driver transcripts from further RNA ligation).

Fig 2 a) Predicted formation of a single stranded 3' RNA fork when a messenger RNA that contains an alternatively spliced 3' exon hybridises with a cDNA that misses such an exon. b) Predicted outcome of adding RNase VI resulting in the digestion of the RNA duplexes hybrids harbouring alternate 3' exons and the formation of RNA fragments chemically active at both ends.

Fig 3 a) Predicted formation of a single stranded internal RNA loop when a polyA RNA that contains an alternatively spliced internal exon hybridises with a polyU RNA that misses such an exon. b) Predicted outcome of the addition of RNase VI resulting in the digestion of the RNA duplexes hybrids harbouring alternate internal exons and the formation of RNA fragments chemically active at both ends.

Fig 4

A system used to capture non-hybridized in vitro transcribed polyU RNA. First, after RNase VI digestion, the purified RNA soup (single stranded RNA) is hybridised against a biotinylated poly-dA oligonucleoti.de A (a). The polyU/polydA hybrids are run through an streptavidin column (b). The ready to ligate exon soup is found in the flow-through while the polyU/polydA hybrids, including the previously non- hybridized polyU RNA (e), are retained in the column (d).

Fig 5

Hinging of RNA strings by intercalating Not I double stranded nucleotide hinges. Not I RNA hinges are ligated to RNase digests in order to generate hinged strings.

Fig 6

Shielding of both hinged or un-hinged RNA strings. Mono-modified linkers (either 3 'OH or 5'P modified RNA linkers) are ligated to the 5' and 3' ends respectively of RNA strings.

Fig 7

A technique to convert shielded RNA strings into their complementary DNA. The technique, dubbed RT-SS or RT-PCR uses DNA primers homologous or complementary to the 3 'OH or 5'P modified RNA linkers, respectively.

Fig 8.

Restriction digestion of the complementary DNA of shielded RNA strings with the appropriate restriction enzyme (for example, EcoRI). These digests are cloned in an equally EcoRI digested suitable vector.

Fig 9. Restriction digestion of the cloned complementary DNA of shielded RNA strings with the appropriate hinge restriction enzyme (i.e. Not I).

Fig 10 shows a typical Blast search of sequences from a single gene fragment.

Fig 11 shows a Blast search using an entry sequence from an exon string clone. In the example depicted, the string is composed of three different gene fragments.

DETAILED DESCRIPTION OF THE INVENTION

In a first embodiment, the present invention is applied to the study of alternative splicing.

A simple approach to obtaining RNA fragments corresponding to alternatively spliced exons consists in the hybridisation of two pools of RNA of a different tissue origin (i.e. normal versus cancer tissue, one tissue versus a different one, etc.) followed by RNase VI digestion.

Unlike transcription, alternative splicing provides a means to introduce gene functional diversity without changes in the amount of m NA. Usual ways of making cDNA libraries however tend to misrepresent mRNA species that are expressed at low levels. In theory, these mRNAs are subjected to AS to the same extent as abundant mRNAs. Therefore, it is important to increase the partial representation of those transcripts and, as consequence, of AS events within rare transcripts. To this end, we designed a strategy to "equalize" the relative abundance of transcripts within an mRNA sample. The strategy, dubbed Self-Normalization or SN, is based on common practices in current protocols for synthesis of cDNA libraries (i.e. representational difference analysis or RDA, Kuvbachieva et al, 2002), the main difference being that, in our invention, we perform a self-hybridisation or homo-hybridisation of RNA/DNA moieties instead of trans-hybridisation (i.e. hetero-hybridisation).

In order to generate adequate RNA RNA duplexes, in another embodiment, we synthesize two types of self-normalised cDNA libraries: Sense and anti-sense cDNA libraries. In addition, both sense and anti-sense cDNA libraries have the singularity of being enriched in mRNA CAP sequences (or 5' end sequences). This step is critical in order to obtain a representation of 5' end alternative exons. Thus, in a different embodiment, a strategy to be able to recover a complete population of alternative exons is provided. The strategy, dubbed Five prime Exon Rescue or FER, is based in the generation of sense and anti-sense mRNA pools of two RNA extracts to be compared, followed by the 5' modification of tester RNA pool, the differential hybridisation of sense with anti-sense mRNAs and the separation of both 5' and 3' RNA forks from un-hybridized RNA through the use of a variety of capture and purification columns.

Differential hybridisation between two situations to be compared consists in mixing tRNA from one tissue with dRNA from the other. Hybridisation of complementary species leaves un-hybridized single stranded segments of RNA that correspond to differentially expressed/spliced exons. These asymmetric hybrids can be isolated and treated with an enzyme that eliminates all RNA segments that form duplexes (in other words, RNA segments that are symmetric/complementary in both pools). This reaction yields free segments of single stranded RNA (ssRNA) that correspond to differential exons.

Since products of RNase VI digestion contain reactive 5'P and 3 'OH ends, we can ligate pools of RNase VI digested hybrids with the T4 RNA ligase enzyme. T4 RNA ligase welds single stranded RNA fragments through a phosphodiester bond.

The reaction above constitutes a key step in our technology. It allows us to create random strings of RNA segments that are a representation of relative abundance of differential exons. In order to rescue and analyse those fragments, we have developed several strategies. Perhaps the more important one is accomplished with the utilization of molecular hinges that are intercalated between RNase VI digests. Typically, an example of molecular hinge can be represented by an RNA linker that is complementary to the recognition signature of a low frequency cutter DNA restriction enzyme (in the example described below, Not I). Typically, the linker is modified on both ends (5'P and 3 OH), therefore becoming a substrate for T4 RNA ligase. RNase VI digests can therefore be ligated to both ends of those molecular hinges. Strings of hinged RNA fragments can be converted into their complementary DNA (cDNA) and eventually disjointed into the original individual fragments by using the appropriate restriction enzyme (Not 1 in the example below). We can find many uses to those DNA fragments that are complementary to alternate RNA exons. In an example below we use them to spot solid surfaces such as 96 or 384 microtiter well plates or microarrays to be used in expression profiling and other similar studies. Strings of RNA (hinged or not) are submitted to final molecular modifications in order to shield them from further ligations and, in addition, to enable the cloning of their complementary DNAs. In this embodiment, RNA linkers that are modified only on one end (they are either 3 'OH or 5'P modified) are used in conjunction with T4 RNA ligase and pools of RNA strings to create shielded RNA strings. These linkers are complementary to the recognition signature of a low frequency cutter DNA restriction enzyme (in the example described below, EcoRI), thereby enabling the convenient cloning of shielded strings' cDNAs into vectors that encode the same restriction enzyme site.

Plasmids with a multicloning site including a unique EcoRI restriction enzyme site, are digested as is the cDNA of shielded RNA strings. Digested vector and insert are subsequently ligated and used to transform bacteria. Clones expressing cDNAs of individual strings are isolated, plasmid purified and used in two types of embodiments: a) Individual clones can be submitted to a DNA sequencing reaction. Sequences obtained with this approach are compared with available public gene databases in order to identify, characterize, and map each individual exon included in each sequenced string. b) Individual clones can be subjected to a restriction digestion using the specific hinge restriction endonuclease (Not 1 in the example below). This digestion directs the release of individual alternate exonic fragments that are typically gel purified. c) Gel purified alternate exons can be used as probes in multiple sub- embodiments. For example, they can be individually spotted, at the adequate concentrations, on solid surfaces such as in 384 and 96 well plate formats as well as microarrays and ultimately used in gene profiling studies.

In another embodiment, probes, oligos or otherwise sequence information obtained using the SANE method can be applied in the development of in vitro diagnostics DΝA based tests. These tests can be based either on hybridisation techniques (plates, microarrays or other surfaces, solution), or on DΝA polymerisation techniques (generally speaking, PCR or electrophoretic fragment length analysis as in STR analysis).

In another embodiment, exonic sequences obtained with the SANE technology can be analysed for their codon code. Alternate exons often code for individual protein domains. A protein domain is a segment of a protein that typically can be separated, synthesized and purified in a functional form. Thus, protein domains encoded in some of the SANE exons can be, for instance, synthesized (either in vitro as in the in vitro synthesis of peptides or in vivo as in the in vivo expression in a micro-organism or a cell of a peptide), purified and used as tools in functional studies, as therapeutic targets, as drugs, or as immunogens to raise specific antibodies against the particular domain encoded by the alternate exon.

EXAMPLES

For example purposes, the methods of this invention can be applied to the comparative study of variant exons in samples of human lung cancer. This study requires a minimum of two samples (ideally biopsies, ideally normal and cancer lung tissues from the same patient) that are processed as follows:

1. Sample processing and RΝA extraction.

Tissue samples are dounce lysed using a Douncer (1 min) in the presence of 1 ml of RΝA Wiz lysis buffer. Total RΝA is purified following standard phenol/chloroform/isopropanol purification methods as is well known in the art.

2. RT-Second Strand and RT-PCR reactions

For the synthesis of the first strand complementary DΝA, 0.5-3 ug total RΝA is incubated with lul of 0.5ug/ul oligo dT(15)-T7 primer (for antisense cDΝA libraries) or poly-dT primer (for sense cDΝA libraries) at 70C for 3-4min, snap cool on ice and mixed with 12ul of a 1^st Strand Master mix (4ul 5X First strand buffer; lul lug/ul CAPSWITCH primer -for antisense cDΝA libraries- or CAPT7 -for sense cDΝA libraries-; 2ul 0.1M DTT; lul RNaseIN (Promega Cat# Ν2111); 2ul lOmM dNTP (Pharmacia Cat# 27-2035-02); 2ul Superscript II (Gibco BRL Cat# 18064-071). The reaction is carried out at 42C for 90min in a thermal cycler.

For the synthesis of the second strand, a total of 128ul of a 2^nd Strand Mater Mix (106 ul DEPC H2O; 15ul Advantage PCR buffer; 3ul 10 mM dNTP mix ; lul RNase H (2U/ul Gibco BRL Cat# 18021-071); 3ul Advantage Polymerase (Clontech Cat# 8417-1)) is added to each tube. The samples are then incubated at 37C for 5min to digest mRNA, and either 94C for 2min to denature, 65C for lmin for specific priming and 75C for 30min for extension, or Smart PCR with some modifications (essentially, 94C for lmin, followed by (94 C 20", 68C 6') 22 cycles. The reaction is stopped with 7.5ul 1M NaOH solution containing 2mM EDTA and incubated at 65C for lOmin to inactivate enzyme.

For second strand cleanup, lul O.lug/ul or lug/ul Linear Acrylamide (Ambion Cat# 9520; comes at 5ug/ul) and 150ul Phenol: Chloroform: Isoamyl alcohol 25:24:1 (Boehringer Mannhem Cat #101001) is added to the PCR tube, mixed by pipetting and spun at 14,000rpm for 5min at room temperature. The aqueous phase is then transferred to an RNase/DNase-ftee tube, added 70ul of 7.5M ammonium acetate (Sigma Cat# A2706) and gently mixed. After adding 1ml 95% room temperature ethanol, the sample is centrifuged at 14,000rpm for 20min at room temperature. The PCR reaction is evaluated in a 1% agarose gel.

3. Self-Normalization

Pools of mRNAs in a given cell are mixtures of mRNA species that are present at different levels. Although changes in cell phenotype can be the result of differences in the amount of mRNA expression levels, alternative splicing without accompanying changes in gene expression levels can also be responsible for dramatic differences in protein function and cell phenotype (Gee SL et al, 2000). It is thus of central importance to our technology to be able to obtain a representation of those cases (genes with low mRNA expression levels that are subjected to alternative splicing) that are likely to account for a considerable amount of AS species. As a general goal, we would like to change the relative abundance of the high and low frequency mRNA subpopulations in a given pool. This has been traditionally a concern in making cDNA libraries with a democratic representation of transcripts independently of their relative abundance (Swaroop et al, 1991). Today's protocols for cDNA libraries synthesis often include a normalization step that consists in the depletion of the more commonly abundant transcripts by hybridisation of a (tracer) test mRNA pool with a driver cDNA mixture of a multi-cell type origin (typically in a 1/100 ratio) (Li et al, 1994, Kuvbachieva et al, 2002). The association of complementary nucleic acid molecules is extremely sequence-specific, in such a way that hybridisation of the most represented species will be favoured in comparison with those less represented. After hybridisation, driver-tracer hybrids are removed. This constitutes the proper normalization step. The small fraction of mRNA that has not been hybridised is separated from the bulk of DNA-RNA hybrids, by a commercially available oligodT column (Qiagen). Usually, the process must be performed reiteratively in order to remove the most abundant sequences common to both the driver and the tracer. For our own purposes we call this type of population redistribution heterodepletion.

We assumed that the procedure above could be applied to the depletion of abundant transcripts within the same population of mRNAs (we dubbed it homodepletion or self-normalization). In preliminary experiments with controlled amounts of known transcripts, we are able to change (either decrease the concentration of high abundance or increase the concentration of low abundance mRNA species) the relative abundance in average one log per normalization step. This is in accordance with previous studies on heterodepletion.

The normalization technology is carried out, in more detail, as follows:

Messenger RNA is purified from total RNA and analysed by formaldehyde gel electrophoresis. Then, first strand cDNA as well as double strand cDNA, are synthesized as described previously. First strand DNA or PCR amplified first and second strand DNA (proper PCR amplified cDNA library) are treated with RNaseH, as recommended by the manufacturer (USB), and purified by Phenol/CHC13 extraction The recovered aqueous fraction is precipitated by adding 0.1 volume of 3M-ammonium acetate and 2.5 volumes of ethanol.

First strand or PCR amplified double strand DNA are hybridised with the RNA (tester), in such a way that the molar concentration rate is around 1 RNA (tracer): 100 DNA (tester). This hybridisation is carried out in a dry oven at 37°C, in a buffer containing 80% formamide, 250mM NaCl, 25mM Hepes (pH 7.5) and 5mM EDTA. Although this is the preferred hybridisation method other methods well known in the art can be used to generate DNA/RNA duplexes. After the hybridisation step, the sample is purified by EtOH precipitation, as above. Single and double-stranded polynucleotides can be separated from one another using an oligo-dT column (oligotex resin, Qiagen Corp.), as suggested by the manufacturer. The oligotex resin will retain solely those mRNA molecules that were not forming hybrids with driver DNA (i.e. the less abundant mRNA species). The eluted mRNA from the oligotex column is reverse transcribed (first and double strand DNA), and then in vitro transcribed to RNA (see below: In vitro transcription). This RNA is analysed and will constitute the first normalized RNA library. Normalisation efficiency is analysed by Real Time PCR (ReTiPCR), that checks the gene expression profiling of reference genes using specific primers that amplify high (actin) and low (fibrinogen) abundance transcripts (Sympson et al, 1997). This technology has been developed to detect variant gene expression based on the polymerase chain reaction. ReTiPCR evaluates product accumulation, using SYBR Green dye detection in a light cycler apparatus (Roche), during the time of reaction. This is an accurate approach to gene quantification.

Whether we perform one or several rounds of self-normalization, eluted RNA from the last normalization round will be used to generate a self-normalized library. In order to obtain a ready for in vitro transcription library both T7-CAP-PCR and dT-T7 primers can be used in conjunction with poly-dT and CAPSWITCH primers respectively in a classical RT-PCR reaction (see above: generation of SMART cDNA libraries). This RT-PCR reaction will yield either sense (T7-CAP-PCR and poly-dT) or anti-sense (dT-T7 and CAPSWITCH) double stranded cDNA libraries. Such libraries are phenol/chloroform purified and the DNA dried until used in the in vitro transcription reaction.

4. In vitro transcription and modification of cDNA libraries

Sense and anti-sense cDNA libraries, as above, are the template in classical in vitro transcription experiments as follows:

The following reagents (MegaScript, A bion Corp.) will be added to the dried pellets of T7 sense and antisense cDNA libraries: 8 μl of NTP mix, 2μl 5X reaction buffer, 2μl enzyme mix, and 8μl H2O. The reaction mix will be incubated at 37°C for 6 to 16 hours.

The reaction product (either sense poly A RNA when using a sense cDNA library or antisense polyU RNA when using an anti-sense cDNA library) is visualized in a formaldehyde agarose gel electrophoresis as is well known in the art. Alternatively, the analysis can be done using a 2100 Bioanalyzer (Agilent Technologies).

In some experiments and for libraries to be used as a tester, the in vitro generated polyA (sense) RNA pool is 5' modified. First, 5' phosphate is removed using Calf Intestinal Phosphatase (CIP) or other enzymes that remove ligable 5' phosphates. This reaction is followed by the 5' addition of a single ligatable phosphate. The latter reaction is carried out in the presence of polynucleotide kinase (PNK) as is well known in the art.

In another set of experiments, the antisense dRNA pool can be modified in order to shield its transcripts from unwanted RNA ligation reactions. More specifically, a modified nucleotide triphosphate can be covalently added to the 3' end of transcripts. This NTP (i.e. 2'hydroxy-adenine or cordycepin) is chemically modified in such a way that it does not admit further 3' polymerization with natural NTPs. Thus, cordycepin is covalently linked to the 3' end of transcripts through the formation of a phosphodiester bond. This reaction is catalyzed by several types of enzymes, including PolyA polymerase or Terminal Deoxynucleotidyl Transferase (TDT). In a preferred embodiment, the antisense dRNA pool is also submitted to a 5' dephosphorylation as above. This latter step is important in order to render the dRNA pool non reactive in future RNA ligation reactions. An additional experiment can be optionally performed that introduces biotinylated UTP nucleotides at the 3' end of dRNAs in the presence of TDT. Ideally the latter reaction is followed by the addition of cordycepin. This biotinylated dRNA can be used later to isolate tRNA/dRNA hybrids in a streptavidin biotin-capturing system.

5. RNA-RNA Hybridization and ribonuclease VI digestion

Purified samples from the previous step can be used in RNA/RNA Hybridisation reactions. Hybridisation is carried out at RoT values of 5-500, depending on the experiment, in a buffer containing 80% formamide (from a deionised stock), 250 m-M NaCl, 25 mM HEPES (pH 7.5), and 5 mM EDTA. Hybridisation is carried out either at 37°C in a dry oven with continuous rotation; even volumes as small as 5 μL did not require mineral-oil overlays. Many other RNA/RNA hybridisation protocols, well known in the art, can be non-exclusively tested at this point, to reach optimal hybridisation results. After hybridisation, the sample is precipitated by adding 2.5 volumes of absolute ethanol and incubated it for 30 min on ice, centrifuged for lO min at 15,000 rpm and washed once with 70% ethanol; the pellet containing the RNA/cDNA hybrids is carefully resuspended in 10 μL of water on ice. In order to eliminate the matched RNA, RNA/RNA hybrids are usually digested with RNase VI as recommended by the manufacturer (Ambion). The sample is then precipitated in the presence of 1 μl O.lug/ul or lug/ul Linear Acrylamide and 2.5 M ammonium acetate after addition of 2.5 volumes of 95% room temperature ethanol. The precipitate (RNA Exon Soup or RES) is vacuum dried and used in RNA ligation experiments. 6. Capture of non-hybridised RNA

Non-hybridised driver RNA is eliminated from the RES by passing the mixture through a poly-dA column. This column can be designed in several ways. In one version, an oligonucleotide poly-dA (26mer) is covalently attached to agarose beads and these beads used to retain polyUridine containing transcripts. In a preferred version, an oligonucleotide poly-dA (26mer) is 5' labelled with biotin. In this version, the oligonucleotide is first hybridised with the RNase VI digest followed by the separation (retention) of the polyU/poly-dA hybrids from the exon soup in a streptavidin column.

In another version of this example, double stranded RNA can be separated from single stranded RNA by passing the RNA/RNA hybridisation mixture through a hydroxyapatite column.

7. RNA ligation

The reaction master mix contains the following:

o lμl lOx ligase buffer, o 1 μl RNase inhibitor o 0.5μl T4 RNA ligase (6U/μl) o lμl RNA substrate

Total volume is adjusted to lOμl/reaction with H₂O and add 6μl of master-mix to each reaction tube containing the dried RES, and incubated overnight at 16°C. In some experiments, the sample is divided in two halves. One is directly used as a template in RT-Second strand or RT-PCR reactions as below. The other half is purified in an Rneasy column previous to use in those reactions.

8. String Hinging

Hinge ligation. The oligoNotLinkRNA (5'AAUCAGAAGGCGGCCGCAAGA- OH), is in vitro phosphorylated in the presence of T4 polynucleotide kinase (incubating in essentially T4 PNE buffer for 1 hr @37°C. T4 PNK buffer, 50mM Tris-Cl ph7.5, lOmM MgC12, 1 mM DTT, ImM ATP, and 10 units of T4 PNK), purified using a Prepare Bio-6 Chromatograph column (Bio-Rad Cat# 732-6222) as follows: Wash column one time with 700ul DEPC H2O and spin at 700xg for 2min at room temperature. Remove flow-through. Spin again at 700xg for 2min to dry column completely. Load 60μl sample (lOμl ligation reaction plus 50μl DEPC water) to the centre of the Bio-6 column and spin at 700xg for 4min. Transfer samples to new PCR tubes. Completely dry samples by speedvac.

For the RNA ligation experiment, pellets containing the RES are resuspended in an RNA ligation mix containing lμl lOx ligase buffer, lμl RNase inhibitor, 0.5μl T4 RNA ligase (6U/μl) lμl NotLinkRNA phosphorylated linker and 6 μl DEPC H2O.

9. String Shielding and Library Amplification

Modified RNA primers (EcoLinkRNA (P-AAUACCGAGAAUUCCCUUGCG- 3'), VRNA (5'-ACUGACAUGGACUGAAGGAGUAG-OH)) are used in ligation reactions in order both to shield RNA strings and to allow cloning of their complementary DNAs. Thus EcoLinkRNA and NRNA primers are used in an RNA ligation mix containing the purified hinged strings. This reaction generates shielded strings with a ligated 5' NRNA linker and a 3' EcoLinkRNA linker. After completion (16°C overnight incubation), the reaction is precipitated in the presence of 1 μl O.lug/ul or lug/ul Linear Acrylamide and 2.5 M ammonium acetate after addition of 2.5 volumes of 95% room temperature ethanol and the dried pellet used RT-PCR reactions. The latter reaction is carried out as above except for the primers (EcoLinkPCR instead of the dT-T7 primer and VPCR instead of the CAPSWITCH primer).

10. Library Cloning and Screening

As a way of example, RT-PCR generated complementary DNAs from RNA shielded strings, as well as a suitable plasmid vector are digested with the appropriate restriction enzyme (i.e. EcoRI). The digested insert (library of chained cDNAs) and vector (pGEM-T, Promega Corp.; pCR4-TOPO, Invitrogen Corp.) are gel purified, as well described in the art, and ligated together in the presence of T4 DNA ligase. The products of such ligation are ultimately used to transform competent E. coli cells, that are grown in ampicillin selection plates. Individual clones from such transformation are grown and lysed to obtain purified plasmid DNA. This DNA is later sequenced or used in other reactions such as digestion with hinge restriction endonuclease and subsequent agarose gel purification of the yielded DNA fragments (individual exonic fragments). Those fragments can be, for example, spotted on solid surfaces to use as probes in, for instance, gene expression studies.

11. Sequence Validation

Different methods well known in the art can be available to use to this end. These include bioinformatics tools to blast string sequences against gene banks, to annotate and map, and to predict exon/intron boundaries, transcription start points, polyadenylation sites or Open Reading Frames.

Materials

Polymeric Nucleic Acids used as Primers

SEQ ID No 1: NotLinkRNA (5'-AAUCAGAAGGCGGCCGCAAGA-OH),

SEQ ID No 2: NotFlankRNA (P-AACGUGCUCGACUAGAUGAGCGGCCGCAAGUGAC

GUCUGCACGUCA-OH),

SEQ ID No3: EcoLinkRNA (P-AAUACCGAGAAUUCCCUUGCG-3'),

SEQ ID No 4: NRNA (5'-ACUGACAUGGACUGAAGGAGUAG-OH),

SEQ ID No 5: EcoLinkPCR (5'-CGCAAGGGAATTCTCGGTATT-3'),

SEQ ID No 6: NotLinkPCR (5'-AATCAGAAGGCGGCCGCAAGA-3'),

SEQ ID No 7: dT-T7 (5'AAACGACGGCCAGTGAATTGTAATACGACTCACTATA

GGCGCT(15)3'),

SEQ ID No 8: CAPSWITCH (5'-TGCTGCGGAAGACGACAGAAGGG-3'),

SEQ ID No 9: TS (5'-AAGCAGTGGTAACAACGCAGAGTACGCGGG-3').

SEQ ID No 10: beta-actin:GTCGACAACGGCTCCGGCATG (forward)

SEQ ID No 11: GGAAGGCTGGAAGAGTGCCTC (reverse);

SEQ ID No 12: Fibrinogen:GGCCTCATCTGCCATTTTATAGCTC.(forward)

SEQ ID No 13 : GAAGCAACTTGAACAGGTCATTGC (reverse)

SEQ ID No 14: dA(26)biotin (AAAAAAAAAAAAAAAAAAAAAAAAAA)

SEQ ID No 15: CAP-T7-II (GCATTAGCGGCCGCGAAATTAATACGACTCACTAT

AGGGAGATGCTGCGGAAGACGACAGAAG)

SEQ ID No: 16: VPCR (5'-ACTGACATGGACTGAAGGAGTAG-OH), REFERENCES

Patent Documents

EP1062364 Dec, 2000 Schweighoffer et al

EP 1198595 Apr., 2002 Tocque et al

US5695937 Dec, 1997 Kinzler et al

US6251590 Jun., 2001 Schweighoffer et al

US6372432 Apr., 2002 Tocque et al

US6383743 May, 2002 Kinzler et al

Other references

Brett,D., Hanke,J., Lehmam,G., Haase,S., Delbruck,S., Krueger,S., Reich,J. and Bork,P. (2000) EST comparison indicates 38% of human mRNAs contain possible alternative splice forms. FEBSLett., 474, 83-86

E ing ,B. and Green,P. (2000) Analysis of expressed sequence tags indicates

35,000 human genes. Nat. Genet, 25, 232-234

Gee SL, Aoyagi K, Lersch R, Hou N, Wu M, Conboy JG. (2000)Alternative splicing of protein 4.1R exon 16: ordered excision of flanking introns ensures proper splice site choice. Blood. 95:692-9

Hawkins, J.D. (1988) A survey on intron and exon lengths. Nucleic Acids Res.,

16, 9893-9905

Ji, H., Zhou,Q., Wen,F., Xia,H., Lu,X. and Li,Y. (2001) AsMamDB: an alternative splice database of mammals. Nucleic Acids Res., 29, 260-263

Kuvbachieva AA, Goffmet AM. (2002)A modification of Representational

Difference Analysis, with application to the cloning of a candidate in the Reelin signalling pathway. BMCMol Biol 3:6

Li, W.-B., Gruber, C.E., Lin, J.-J., Lim, R., DAlessio, J.M., and lessee, J.A.

1994. The isolation of differentially expressed genes in fibroblast growth factor stimulated BC3H1 cells by subtractive hybridization. BioTechniques 16: 722-

729 Liang, F., Holt,L, Pertea,G., Karamycheva,S., Salzberg,S.L. and

Quackenbush . (2000) Gene Index analysis of the human genome estimates approximately 120,000 genes. Nat. Genet, 25, 239-240

Mironov,A.A., Fickett .W. and Gelfand,M.S. (1999) Frequent alternative splicing of human genes. Genome Res., 9, 1288-1293

Mironov AA, Novichkov PS, Gelfand MS. Pro-Frame: similarity-based gene recognition in eukaryotic DNA sequences with errors. Bioinformatics 2001

Jan;17(l):13-5

Needleman, S.B. and Wunsch,C.D. (1970) A general method applicable to the search for similarities in the amino acid sequence of two proteins. J Mol Biol,

48, 443-453

Relogio A, Schwager C, Richter A, Ansorge W, Nalcarcel J. Optimization of oligonucleotide-based DΝA microarrays. Nucleic Acids Res. 2002; 30:e51.

Schweighoffer F, Ait-Ikhlef A, Resink AL, Brinkman B, Melle-Milovanovic D,

Laurent-Puig P, Kearsey j, Bracco L. Qualitative gene profiling: a novel tool in genomics and in pharmacogenomics that deciphers messenger RΝA isoforms diversity. Pharmacogenomics. 2000; 1:187-97

Sharp ,P.A. (1994) Split genes and RΝA splicing. Cell, 11, 805-815

Shimkets RA, Lowe DG, Tai JT, Sehl P, Jin H, Yang R, Predki PF, Rothberg

BE, Murtha MT, Roth ME, Shenoy SG, Wmdemuth A, Simpson JW, Simons

JF, Daley MP, Gold SA, McKenna MP, Hillaii K, Went GT, Rothberg JM. Gene expression analysis by transcript profiling coupled to a gene database query. Nat

Biotechnol 1999; 17: 798-803

Swaroop, A., Xu, J., Agarwal, Ν., and Weissman, S.M. 1991. A simple and efficient cDΝA library subtraction procedure: Isolation of human retina-specific cDΝA clones. Nucleic Acids Res. 19: 1954

Sympson PJ. (1997 Induction of fibrinogen biosynthesis and secretion from cultured pulmonary epithelial cells. Blood, 89: 873-882.

Smith CW, Nalcarcel J. Alternative pre-mRΝA splicing: the logic of combinatorial control. Trends Biochem Sci. 2000; 25: 381-8.

Taveau M, Stockholm D, Spencer M, Richard I. Quantification of splice variants using molecular beacon or scorpion primers. Anal Biochem. 2002; 305:

227-35 Velculescu NE, Zhang L, Nogelstein B, Kinzler KW. Serial analysis of gene expression. Science. 1995; 270: 484-7.

Yeakley JM, Fan JB, Doucet D, Luo L, Wickham E, Ye Z, Chee MS, Fu XD.

Profiling alternative splicing on fiber-optic arrays. Nat Biotechnol. 2002; 20:

353-8.

Claims

1. A method to experimentally analyse boundaries within polymeric deoxy ribonucleic acid (DNA) or ribonucleic acid (RNA) molecules, comprising: a) Hybridisation of first and second pluralities of different RNA molecules, derived from first and second samples, to provide a plurality of RNA / RNA hybrids; b) Obtaining from said RNA / RNA hybrids a plurality of RNA fragments with free 5'P and 3 'OH groups, wherein said RNA fragments are differentially processed exons; c) Ligating at least two of said RNA fragments to form a string; d) Reverse transcription of the product of reaction (c); and e) PCR amplification of the product of reaction (d).

2. The method of claim 1, wherein the ligation is carried out in the presence of a hinge RNA linker, said hinge RNA linker comprising a nucleotide sequence which is complementary to the recognition site of a DNA restriction enzyme.

3. The method of claim 1 or 2, wherein the ligation is carried out in the presence of a PCR RNA linker, said PCR RNA linker comprising a nucleotide sequence for hybridising to a PCR primer.

4. The method of any preceding claim, further comprising cloning of the PCR products into a recombinant vector.

5. The method of any preceding claim, further comprising sequencing of the PCR products.

6. The method of any preceding claim, further comprising digestion of the PCR products with an appropriate restriction enzyme to release a fragment comprising a differentially processed exon.

7. The method of any preceding claim, further comprising purifying the PCR products.

8. The method of any preceding claim, wherein the RNA fragments of step b) are obtained by digestion of the RNA / RNA hybrids with a ribonuclease.

9. The method of claim 8, wherein the ribonuclease is RNase VI.

10. The method of any preceding claim, comprising producing a peptide having an amino acid sequence corresponding to the nucleotide sequence of at least one of the RNA fragments.

11. The use of a product of the method of any preceding claim as a probe.

12. A method to change the relative representation of polymeric deoxy ribonucleic acid (DNA) or ribonucleic acid (RNA) molecules, within a given DNA or RNA sample, comprising a) RT-PCR of an mRNA sample to generate a cDNA library; b) hybridization of said mRNA sample with said cDNA library, to provide a plurality of cDNA / RNA hybrids and a plurality of single stranded RNA molecules; and c) selectively extracting single stranded RNA from the sample.

13. The method of claim 12, further comprising: d) reverse transcribing said single stranded RNA to provide a cDNA sample; and e) transcribing said cDNA sample to provide an RNA sample.

14. The method of any of claims 12 or 13, further comprising repeating the steps of the method at least once more.

15. The method of any of claims 1 to 11, wherein at least one of said first and second pluralities of RNA molecules is obtained using the method of claims 12 to 14.

16. A method to experimentally analyse 5' exon forks comprising a) Generation of RNA duplexes between sense RNA and antisense RNA from two different samples; b) Generation of a population of RNA fragments with free 5'P and 3 'OH groups from said RNA duplexes; and c) Purifying RNA exonic fragments with free 5'P 3 'OH groups from said population of RNA fragments.

17. The method of claim 16, wherein said RNA fragments are generated with RNase VI.

18. The method of claim 16 or 17, wherein said RNA exonic fragments are purified by means of a poly-U capture poly-A column.