Technical Field
The present invention relates to a method for searching for a gene polymorphism for identifying a gene whose expression level is different between alleles, and to a method for searching for a phenotype-associated gene polymorphism.
Background Art
Genes on the same genomic position but on the different alleles can be different in expression level. This phenomenon is a relatively new concept that has been reported recently (Knight JC. Allele-specific gene expression uncovered. Trends Genet. Mar; 20(3): 113-6. PMID: 15049300, 2004).
Genes expressed differentially between alleles are roughly classified into two types: imprinted gene and non-imprinted gene. In the phenomenon of the former, i.e. the imprinted gene, when one allele is inherited from one of the parents while the other allele from the other parent, one of the alleles is physiologically inactivated (e.g., methylated), whereby the expression of the gene is inhibited in cells or tissues. Also for the latter (i.e. non-imprinted genes), there are some cases where a difference in expression level is observed between alleles. With respect to such a difference, it is thought that a genomic polymorphism in a gene or a region adjacent thereto between alleles serves as a cis-acting element for regulating the expression of a gene in the vicinity of the polymorphism, thereby leading to a difference in gene expression level between alleles. Variations in expression of each allele resulting from different genomic DNA sequences are considered to be properties that aire inherited over generations, and such properties may influence differences in gene expression ■ levels among individuals and differences in body constitutions, pathological conditions and risks thereof, and responsiveness to drugs, among individuals.
2

Difference in gene expression levels between alleles can be most accurately assayed within a same cell, or under identical environmental conditions. In assaying a difference in gene expression levels between alleles, it is important from which allele a certain RNA is derived can be determined. To this end, the presence of a polymorphism that enables distinction of alleles (e.g., SNP) is required in an RNA sequence, which is a transcript of a gene, and the polymorphism in the RNA sequence is measured to determine a difference in gene expression level between alleles. There are several reports on determining a difference in gene expression level between alleles using such a polymorphism (SNP) on RNA (Cowles CR, Hirschhorn JN, Altshuler D, Lander ES, Detection of regulatory variation in mouse genes, Nat Genet. Nov; 32(3): 432-7, PMID, 12410233, 2002; Yan H, Yuan W, Velculescu VE, Vogelstein B, Kinzler KW, Related Allelic variation in human gene expression. Science. Aug 16; 297 (5584): 1143, PMID, 12183620, 2002; Bray NJ, Buckland PR, Owen MJ, O'Donovan MC, Cis-acting variation in the expression of a high proportion of genes in human brain, Hum Genet,, 2003 Jul; 113 (2): 149-53. Epub, May 01, PMID: 12728311, 2003).
However, the techniques employed in the reports are a combination of RT-PCR with a direct sequencing reaction or single-nucleotide extension, wherein cDNA is synthesized from mRNA and amplified, and then arbitrarily selected polymorphisms are individually subjected to typing. These techniques are not capable of simultaneously measuring many genes.
To date, extensive analysis of many genes using microarrays for SNP typing has been reported (Lo HS, Wang Z, Hu Y, Yang HH, Gere S, Buetow KH, Lee MP, Allelic variation in gene expression is common in the human genome. Genome Res. Aug; 13(8): 1855-62. PMID: 12902379, 2003). In this analysis, mRNA with poly(A) is converted into cDNA by the common RT method using a poly(T) primer, samples are prepared by the multiplex PCR technique using many specific primers in accordance.with the same protocol as a conventional genomic DNA typing technique, and samples are hybridized to the arrays to measure the expression levels of cDNA (mRNA) that differ between alleles based on the signal ratio. However, mature mRNA with poly (A) has only exon

sequences following splicing, and so such sequences are too short to comprise enough polymorphisms (SNPs) to be evaluated. Thus, because available polymorphisms (SNPs) are limited, it is difficult to find a gene whose expression level varies in every allele.
The correlation between genetic polymorphism and certain phenotype and gene expression (e.g., difference in disease or drug efficacy) has drawn attention. However, to study the correlation between certain genetic polymorphism and phenotype and gene expression, it is required to examine a huge number of SNPs for respective traits in the case of, for example, genomic SNPs (about 10 million according to the NCBI dbSNP (build 123) reported in October 2004), and thus it is indeed difficult to do so.
If genes whose phenotype and gene expression differs between alleles can be rapidly and effectively selected in order to study the correlation between the thus selected genetic polymorphism and the phenotype and gene expression, accordingly, the cause of a disease, effective therapeutic methods, or the like may be exannined by the completed procedures.
Disclosure of the Invention
Under the above circumstances, one object of the present invention is to provide a method for rapidly and effectively searching for a genetic polymorphism that enables identifying a gene whose expression level is different between alleles. Also, another object of the present invention is to provide a method of utilizing the genetic polymorphism searched for by the above method to search for a phenotype-associated genetic polymorphism.
We have conducted extensive studies in order to achieve the above objects. As a result, expecting that a gene whose expression level is different between alleles could be effectively discovered by utilizing a genetic polymorphism' on intranuclear RNA, we selected a. DNA polymerase, which, is capable of selectively amplifying intranuclear RNA, in the procedures for discovering a gene of interest. As a result, we succeeded in determining a genetic polymorphism (SNP) that enables identifying a gene whose

expression level is-different between aJleles,-thereby leading to the completion of the present invention.
Specifically, the present invention includes the following characteristics.
(1) A method for searching for a genetic polymorphism for identifying a gene
whose expression level is different between alleles, comprising the following steps of:
(a) synthesizing cDNA from total RNA or intranuclear RNA by reverse transcription using a random primer;
(b) selectively amplifying cDNA derived from long intranuclear RNA as a primary transcript using a random primer and a strand-displacing DNA polymerase that reacts at an isothermal temperature;

(c) detecting a genetic polymorphism present in the amplified cDNA;
(d) comparing the expression levels of cDNAs from the respective alleles on the genomic DNA whose genetic polymorphism is heterozygous, based on the detected polymorphism; and
(e) selecting the genetic polymorphism used for comparison, where the
expression levels of cDNAs from the respective alleles significantly differ from each
other.
In said method, examples of the DNA polymerase usable include (j)29 DNA polymerase.
Steps (c) and (d) preferably comprise labeling the amplified cDNAs, which is in turn subjected to hybridization with a genetic polymorphism-specific probe, to compare the expression levels of cDNAs from respective alleles based on the hybridization reaction.
In this method, single nucleotide polymorphism (SNP) can be used as the genetic polymorphism.
(2) A method for searching for a phenotype-associated genetic polymorphism
comprising using a genetic polymorphism searched for by the above method to evaluate
the correlation between the genetic polymorphism or gene expression level and the
5

phenotype.
In this method, examples of phenotypes include pathological conditions and severity of a disease, risks of developing a disease, responsiveness to drugs, responsiveness to foods, responsiveness to chemical substances, and responsiveness to environmental factors.
(3) A method for searching for a genetic polymorphism-associated phenotype comprising using the genetic polymorphism searched for by the above method to evaluate the correlation between the genetic polymorphism or gene expression level and the phenotype.
In this method, examples of phenotypes include pathological conditions and ■ severity of a disease, risks of developing a disease, responsiveness to drugs, responsiveness to foods, responsiveness to chemical substances, and responsiveness to environmental factors.
The present invention provides a method for rapidly and effectively searching for a genetic polymorphism that enables identifying a gene whose expression level is different between alleles. The genetic polymorphism searched for in this manner enables distinguishing of expression levels in different alleles. Accordingly, such a polymorphism can be used as an effective means for analyzing a phenotype associated with the gene of interest. Further, discoverying a correlation between the genetic polymorphism searched for in the aforementioned manner and a phenotype (e.g., risk of developing a disease, or drug responsiveness) may lead to examining the cause of a disease pj effective therapeutic methods.
Brief Dfescription of the Drawings
Fig. 1 is an electrophoresis showing cDNA obtained via amplification with the use of (j)29DNA polymerase.
Fig. 2 shows the positional relationship between the frequency distribution of
6

cDNA/genome signal ratios in the lymphocyte BL1395 and the genes on genome, in which the genetic polymorphism (the probe set) is present.
Fig. 3 shows the positional relationship between the frequency distribution of cDNA/genome signal ratios in the lymphocyte BL2122 and genes on the genome in which the genetic polymorphism (the probe set) is present.
Fig. 4 shows the comparison of cDNA/genome signal ratios determined from U133plus2 array and lOOK array (Xbal 50K) for expression analysis.
Fig. 5 shows the position of SNP sites on PPARy gene.
Fig. 6 shows a schematic diagram of identifying SNPs by direct sequencing.
Fig. 7 shows the correlation between the expression level of the PPARG gene of peripheral blood lymphocytes obtained from 30 Japanese individuals and the typing of genetic polymorphisms.
Fig. 8 shows the frequency distribution of the expression levels and genetic polymorphism typing of the PPARG gene in peripheral blood lymphocytes obtained from 30 Japanese individuals (Fig. 8A), and alleles of haplotype M and haplotype m (Fig. 8B).
Best modes for carrying out the Invention
Hereafter, the present invention is described in detail. This patent application claims priority from Japanese Patent Application No. 2004-366671 filed on December 17, 2004, and includes all or part of the contents as disclosed in the description and/or drawings thereof.
1. Searching for a genetic polymorphism that enables identifying a gene whose expression level is different between alleles
The present invention provides a method for studying a gene with expression levels that vary between alleles. The term "a gene whose expression level is different between alleles" -as used herein, refers. tO;.a gene in which the expression level in one allele differs from that in the other allele. Gene expressions in the respective alleles are distinguishable from each another using specific genetic polymorphisms an indicator;
7

however, not all genetic polymorphisms enable differences in expression levels in different alleles to be distinguished. The present invention, accordingly, provides a method for rapidly and effectively searching for a genetic polymorphism that enables determiiiation of differences in expression levels between alleles.
(1) Synthesis of cDNA from intranuclear RNA and amplification thereof
In this method, cDNA is synthesized from intranuclear RNA. The term "intranuclear RNA" as used herein refers to a primary transcript that is not spliced after transcription from genomic DNA and remains in the nucleus before transferring to the cytoplasm. Thus, many intranuclear RNAs comprise both exons and introns of the genome and have a long strand. (For example, for the 21,804 reference sequences present on the genomic sequence of the Human Genome Build 34 (http://genome.ucsc.edu/) reported in April 2004, the average length is 85,284 bp, the median is 22,855 bp, and sequences longer than 5,000 bp account for about 84% of the all sequences.)
In order to measure differences in expression levels between alleles, genetic polymorphisms must be present in RNA, which is a gene transcript. Since intranuclear RNA, not spliced, might have a long strand, we expected that intranuclear RNA might contain many genetic polymorphisms that enable a gene whose expression level is different between alleles to be distinguished. For example, while the 21,804 reference sequences present on the genomic sequence of the Human Genome Build 34 (http.V/genome.ucscedu/) reported in April 2004 have an average mRNA length of 2,757 bp and a median mRNA length of 2,316 bp, the average length of intranuclear RNA comprising introns is 85,284 bp and the median thereof is 22,855 bp. Such a long
strand would enable evaluation of regions about 40 times the regions that could be
i evaloiate-d in mRNA, without considering the density of genetic polymorphisms-
accordingly.
To synthesize cDNA from intranuclear RNA, intranuclear RNA is selectively
extracted from a sample, from which cDNA is then synthesized. Alternatively, after

extraction of total RNA from a sample, cDNA is synthesized from the extracted RNA, and then only cDNA derived from longer intranuclear RNA is selectively amplified.
In one method, a nuclear fraction is first extracted from a sample. The sample is not particularly limited, provided that it is from a substance that is to be analyzed by this method for the genetic polymorphism for identifying a gene whose expression level is different between alleles. Examples of the sample that can be used include samples derived fi-om animals, plants, and microorganisms (e.g., fungi or bacteria), commercially available cell strains, and deposited cell strains. Samples are preferably from mammalians, more preferably humans. Also, the forms of samples are not particularly limited. When the samples are from humans for example, usable samples are in the form of body fluid, such as blood, saliva, lymph, airway mucus, bone marrow fluid, urine, and coeliac fluid, cells, or tissues.
Nuclear fractions can be extracted by a method known in the art. For example, cells are broken using a homogenizer, and nuclei can be separated via differential centrifugation or density gradient centrifugation (see, for example. Molecular cloning. Chapter 17.8, Preparation of nuclear extracts from tissue/cultured mammalian cells, CSHL Press, ISBN 0-87969-577-3).
Subsequently, cDNA is synthesized from intranuclear RNA prepared in the above-described manner by reverse transcription using a random primer. Use of random primer enables synthesis of cDNA from any sequence of RNA (i.e., intranuclear RNA) in samples.
The reversely-transcribed cDNA is then amplified using a random primer Because intranuclear RNA, not spliced, has a long strand, the RNA cannot be amplified with.a DNA polymerase as commonly used for amplification. Thus, a DNA polymerase that catalyzes a strand displacement under isothermal reaction conditions is used in this method. The DNA polymerase having such;properties includes, but not particularly limited to,. (}i29 DNA polymerase (Genomiphi™, Amersham Bioscience). The amplification reaction catalyzed by the (j)29 DNA polymerase is very stable, so cDNA synthesized by reverse transcription can be directly used for the amplification reaction

using (j)29 DNA polymerase without the need of any purification process. Additionally, since the yield is at the order of fig, extremely small amount's of samples, such as clinical specimens, can be amplified without loss of the sample caused by purification, and thus the use of (j)29 DNA polymerase is particularly preferable in the method of the present invention. Further, an alternative polymerase is Bst polymerase, which is commercially available from New England Biolabs (Lage JM, Leamon JH, Pejovic T, Hamann S, Lacey M, Dillon D, Segraves R, Vossbrinck B, Gonzalez A, Pinkel D, Albertson DG, Costa J, and Lizardi PM, Whole genome analysis of genetic alterations in small DNA samples using hyperbranched strand displacement amplification and array-CGH, Genome Res. 2003 Feb; 13(2): 294-307, PMID: 12566408).
The above-described DNA polymerases terminate the amplification reaction at the end of a DNA fragment during DNA amplification, so the amplification efficiency significantly lowers around the end. When a short DNA fragment is to be amplified, the amplification rate of the whole fragment lowers due to the short distance between both ends, resulting in selective amplification of a long DNA fragment, i.e., unspliced intranuclear RNA (Lage JM, Leamon JH, Pejovic T, Hamann S, Lacey M, Dillon D, Segraves R, Vossbrinck B, Gonzalez A, Pinkel D, Albertson DG, Costa J, and Lizardi PM, Whole genome analysis of genetic alterations in small DNA samples using hyperbranched strand displacement amplification and array-CGH., Genome Res., 2003 Feb; 13(2): 294-307, PMID: 12566408; General Amplification of Chromosomal DNA by phi29 DNA polymerase, Amersham Bioscience).
Accordingly, cDNA derived from intranuclear RNA can be selectively synthesized and amplified by preparing total RNA (including intranuclear RNA and mRNA) from samples, synthesizing cDNA from various RNA species by reverse' transcription using a random primer, and amplifying the cDNA using the aforementioned DNA polymerase (e.g.,.(j)29 DNA polymerase) that is capable of selectively amplifj'ing-a long-stranded cDNA (i.e.,. cDNA derived from long nuclear RNA). The method of' amplifying cDNA from intranuclear RNA can omit a procedure of selectively extracting intranuclear RNA and thus is preferable in the present invention. Total RNA can be
10

extracted by a method known in the art. For example, the guanidine/cesium method or the acid guanidium-phenol-chloroform (AGPC) method can be employed.
(2) Genetic polymorphism and allele expression level
Expression levels of a gene (an amplified cDNA) from each allele are then compared, wherein the expression levels are compared between heterozygous alleles. The term "heterozygous alleles" refers to two alleles each having genetic polymorphisms different from each other in genomic DNA. When there are heterozygous alleles, accordingly, amplified cDNA from each allele can be distinguished. When the above expression levels are significantly different, the genetic polymorphism of interest can be selected as an indicator for identifying a gene whose expression level is different between alleles.
The term "genetic polymorphism" or "polymorphism" refers to a difference in gene that causes variations in traits or morphologies among individuals, and examples thereof include a single nucleotide polymorphism (SNP), a haplotype and the like. The terra "SNP" refers to a mutation of a single nucleotide in the nucleic acid of a certain gene or genes. Such SNP is known to occasionally cause variations in traits or morphologies among individuals. The term "haplotype" refers to a polymorphism represented by the type and the number of alleles in a continuous gene region or in several mutation sites in genes. The frequency of haplotype recombination is lower than that of usual recombination, and is likely to be genetically conserved. When the correlation between a polymorphism and a phenotype is to be studied, accordingly, it might be important to study the correlation with a given haplotype in addition to the correlation with respective mutations. Further, examples of genetic polymorphisms include insertion/deletion polymorphisms, polymorphisms resulting from a difference in numbers of repetition in a repeated sequence. ;i.and- restriction, fragment length polymorphisms. In: the method of the present invention, use. of a single: nucleotide polymorphism (SNP) is preferable because many methods of detecting SNP are available and differences between alleles can be easily distinguished based on difference in a
11

single nucleotide.
Genetic .polymorphisms that can be employed are not particularly limited, and
genetic polymorphism information can be easily obtained from public databases or the
like. For example, SNP and haplotype information for humans or mice can be obtained
from the NCBI database (http://www.ncbi.nlm.nih.gov/SNP/), and human SNP
information can be obtained from the JSNP database
(http://sap.ims.u-tokyo.ac.jp/indexJa.html). A person skilled in the art can easily obtain other genetic polymorphism information.
Detection of genetic polymorphisms and measurement of expression levels from alleles carrying genetic polymorphisms can be carried out in accordance with techniques known in the art.
For example, the detection of genetic polymorphisms and the measurement of gene expression levels from alleles can be carried out by hybridization with a probe specific for a single genetic polymorphism. A probe can be labeled with an adequate means, such as a fluorescent or radioactive substance, if needed. Any probe can be used, provided that such a probe comprises a site of genetic polymorphism and hybridizes specifically to the amplified cDNA. A specific probe design is known in the art. Any hybridization conditions can be employed, provided that genetic polymorphisms can be sufficiently distinguished. Under such conditions, a probe hybridizes in the case of a certain single genetic polymorphism but does not hybridize in the case of another genetic polymorphism. An example is stringent conditions, which are known in the art.
One end of a probe may be fixed onto a substrate and may be used as a DNA chip (microarray). In such a case, a DNA chip may.comprise probes corresponding to single genetic polymorphisms fixed thereon, or probes corresponding to both of the genetic polymorphisms: [fixed thereon. Detection of genetic polymorphi stns using such DNA chip is described in, for example, "DNA microarrays and latest PGR method," Masaaki Muramatsu and Hirbyuki Nawa (ed.), Shujunsha, 2000, chap. 10;
As a specific example of the technique for detecting genetic polymorphisms
12

using DNA chip, a method involving the use of the GeneChip Human Mapping lOOK
Array (Affymetrix) is explained. The GeneChip® Human Mapping lOOK Array
comprises two arrays that can detect more than 100,000 SNPs in the genome. A sample
(e.g., genome or cDNA) is cleaved with a restriction enzyme (e.g., Xbal or Hindlll), an
adapter is attached, a single type of primer (each type for Xbal and Hindlll) specific for
the adapter is used to amplify the sample by PCR, and the amplified product is then
labeled. Two arrays are designed to be complementary to each SNP allele, SNP of the
sample is evaluated based on the signal after hybridization, and expression levels can be
compared between alleles based on signal intensity or signal ratio. Concerning details
of the DNA chip, reference may be made to the product information and the data sheet
posted on http;//www.affymetrix.co.jp/products/arrays/specific/100k;. and
http://www.affymetrix.co.jp/pdf/Mapping_100K.pdf.
Also, genetic polymorphisms can be detected via any techniques known in the art, in addition to the aforementioned techniques. Examples of the techniques that can be employed include a method involving the use of a primer specific for a genetic polymorphism, a method involving the use of a restriction fragment length polymorphism (RFLP), direct sequencing, denaturing gradient gel electrophoresis (DGGE), a method involving the utilization of chemical cleavage of mismatch (CCM), primer extension (PEX), the invader method, quantitative real-time PCR detection (the TaqMan method), and the like.
In the method of the present invention, use of a DNA chip (or a microarray) that enables detection of as many genetic polymorphisms as possible in a simple and rapid maimer is preferable.
As described above, when the gene expression level (or signal intensity) in each allele, is nieasured based oh the difference'in genetic polymorphisms' and the'gen'e expression levels.of between alleles significantly differ from each other, such genetic polymorphisms are selected. More specifically, the ratio of the allele exhibiting a high expression level to the allele exhibiting a low expression level is determined, and the
13

genetic polymorphism exhibiting a ratio of at least 1.3:1, and preferably at least 1.5:1, is selected. The ratio of 1:1. means that the gene expression levels are substantially the same in both alleles.
The genetic polymorphism exhibiting different expression levels, in comparison of the expression levels of each allele with the utilization of the genetic polymorphisms, can be selected as a genetic polymorphism for identifying a gene whose expression level is different between alleles. By merely detecting the thus-selected genetic polymorphism in tem, it is possible to determine whether or not a given specimen has an allele exhibiting a high expression level. Also, a gene whose expression level is different between alleles may correlate with a phenotype as described below, and so, utilizing the thus-selected genetic polymorphism, the correlation between the genetic polymorphism and the phenotype may be elucidated.
2. Method for searching phenotype-associated genetic polymorphism
Genetic polymorphisms are a genetic difference that causes variety in traits or morphologies among individuals. Thus, a genetic polymorphism may be correlated with a phenotype in some way or another. However, there are present numerous genetic polymorphisms, i.e., as many as at least about 10 milion SNPs on genome (the NCBI dbSNP build 123 reported in October 2004), and so it is difficult to select genetic polymorphisms that are associated with certain phenotypes from among such numerous SNPs. In contrast, a gene whose expression level is different between alleles may be correlated with a certain phenotype. Accordingly, the genetic polymorphisms that enable identifying a gene whose expression level is different between alleles searched for by the aforementioned method may more highly be correlated with phenotypes than other types of genetic polymorphisms.
The method for searching for a: phenotype-associated genetic polymorphism according to the present invention is characterized-in that the genetic polymorphism searched for by the aforementioned method is used to determine the correlation between the genetic polymorphism or gene expression level and the phenotype. Examples of the
14

phenotype include occurrence of a disease (e.g., pathological conditions and severity), risks of developing a disease, responsiveness to drugs, responsiveness to foods, responsiveness to chemical substances, and responsiveness to environmental factors (e;g.. ultraviolet rays or temperature).
Specifically, the method of the present invention can be implemented based on the association method, the affected sib-pair method, or the like, known in the art. In the association method, for example, both an analyte that exhibits a certain phenotype and an analyte that does not exhibit a certain phenotype are used to determine the correlation between a frequency of appearance of genetic polymorphisms searched for by the aforementioned method or gene expression level and a phenotype. Where the occurrence frequency of certain genetic polymorphism is significantly high in an analyte exhibiting a certain phenotype, it can be determined that differences in the genetic polymorphism affect quantitative regulation of the expression level of a phenotype-associated gene, or otherwise, changes in the genetic code of a genetic polymorphism, which lead to changes in amino acids and thus to changes in the nature of proteins, such as phenotype expression. In the affected sib-pair method, family members (such as brothers or sisters) having a same phenotype (e.g., a disease) are compared to identify a chromosome region in which a phenotype-associated gene is present. Such a technique is described in, for example, Sentan no genomu igaku wo shiru ("Learning the advanced genomic medicine"), Yusuke Nakamura, Yodosha, 2000, Chap. 1.
For example, a gene comprising a genetic polymorphism searched for by the above method exhibits a difference in the expression level of the gene, between alleles, in a same individual. Thus, the expression level is deduced to differ among individuals based on the allele type (i.e., the type of a genetic polymorphism). Measurement of genetic polymorphisms and. expression levels of marty individuals enables-verification.of the correlation between expression levels among'individuals and genetic polymorphism information.
The method of the present invention enables searching phenotype-associated
15

genetic polymorphisms. A genetic polymorphism searched for in this manner is useful for diagnosing the development of a disease or the risk of developing a disease or for evaluating responsiveness to drugs in advance.
In the present invention, a genetic polymorphism searched for by the above method is used to determine the correlation between the genetic polymorphism or gene expression level and a phenotype. Thus, it becomes possible to search for a phenotype associated with a genetic polymorphism.
Based on the "inherent properties" of a protein encoded by a certain gene (e.g., involvement of PPAKy in lipid metabolism), for example, differences in genetic polymorphism of the gene are deduced to produce different phenotypes. (For example, PPARy is presumed to be associated with a phenotype involved in the lipid metabolism, such as diabetes.) Accordingly, actual validation of both the genetic polymorphism and the phenotype (e.g., reactivity of an antidiabetic agent, ACTOS^'^, in various individuals as compared with the results of genetic polymorphism typing of PPARy) enables verification of the actual involvement thereof.
By using the aforementioned method, we actually discovered that the expression level of human peroxisome proliferator-activated receptor y (PPARy or PPARG) genes differs between alleles, and also discovered genetic polymorphisms that would enable determination of differences in genetic expression level between alleles.
Examples
Hereafter, the method of the present invention will be described in more detail with reference to the following examples; however, it should be understood that the technical .scope of the present invention is not limited, thereto. [Example 1]
In this'example; cDNA was synthesized from/intranuclear RNA and amplified.
Total RNAs (1. ^g each) of lymphoid cell lines, BLl 395 (ATCC CRL-5957) and BL2122 (ATCC GRL-5967), established by EB virus, were treated with DNAase and subjected to reverse transcription using reverse transcriptase (Superscript III RT enzyme,
16

Invitrogen) in accordance with the protocol included therein, to prepare single-stranded cDNA. A portion (I jil) of the resulting 20 |il of reaction solution was added, without . purification, to a reaction solution containing a random primer and the phi29 enzyme as described in the protocol of Genomiphi^*^ (sold by Amersham Bioscience), the reaction was allowed to proceed at 30°C for 16 hours, and cDNA was obtained (yield: 2.34 }ig and 2.27 \xg, respectively).
The results of electrophoresis of the thus amplified cDNA are shown in Fig. 1. In Fig. 1, lane 1 and lane 2 show cDNA prepared using phi29 DNA polymerase as described above.
As shown in Fig. 1, a smear from 10 Kb or more to about 3 Kb, centering around about 8Kb, was obtained (lanes 1 and 2 of Fig. 1). The fact that the median cDNA length was 2,316 bp after synthesis from common mRNA suggests that only long cDNA was selectively amplified. Specifically, it was demonstrated that the use of the phi29 enzyme enabled a selective amplification of cDNA from long-stranded intrainuclear RNA.
[Example 2]
In this example, cDNA obtained from intranuclear RNA in Example 1 was used to perform experiments on SNP typing and gene expression level. (1) Verification of cDNA amplification from intranuclear RNA using phi29 enzyme
At the outset, 250 ng of the cDNA amplified in Example 1 was subjected to a reaction in accordance with the protocol of the 11 OK array (Affimetrix). Specifically, cDNA amplified with phi29 as described in Example 1 and genomic DNA similarly amplified with phi29.were amplified in accordance with the protocol of common lO.OK, the signal intensity ratio (cDNA signal intensity/genomic DNA signal intensity) was determined, and the frequency distribution thereof was then examined. By dfeterminirig the signal intensity ratio, some sequences are found to be easily amplified with the aid of phi29 and other sequences are found to be less likely to be amplifiisd therewith, depending on differences in secondary structures, which result from differences in
17

sequences (bias of amplification). By dividing the cDNA signal value by the signal-value of the genome amplified with.phi29, however, such a bias of amplification, can be eliminated.
As a result, as shown in Fig. 2 and Fig. 3, a shape similar to a normal distribution curve, which appeared to be noises, was observed in regions with low signal ratios, and there were many signals (probe sets) based on the genetic polymorphisms in regions containing no genes (the light gray zone of a bar chart in Fig. 2 and in Fig. 3). In contrast, portions with potent signal ratios were observed in such a manner that the portions are off the aforementioned normal distribution curve, on the right side of Fig. 2 and Fig. 3 (portions indicated with arrows in Fig. 2 and in Fig. 3), and there were many signals (probe sets) based on genetic polymorphisms that were present in regions with genes (substantially regions from the intron). Based on the shape of the frequency distribution and the positional relationship between the probe sets and the gene on genome, it was found that the gene (cDNA)-derived signal was separated from the noises via the assay. By assaying the cDNA signal/genome signal ratio, it was found that portions exhibiting high signal ratios may be considered as representing signals resulting from the expressed gene (primarily cDNA derived from intranuclear RNA).
Also, whether or not the gene expression level assayed with the use of common microarrays is correlated with the cDNA signal/genomic DNA signal ratio analyzed in the above-described manner, was determined using common arrays for expression analysis (Affymetrix U133plus2.0 array; http://www.affymetrix.co.jp/pdf/HG_DS.pdf).
Specifically, total RNA of BL2122 was prepared and analyzed using the Affymetrix U133plus2.0 array in accordance with a common protocol. The entirety was averaged so as. to bring the average signal value of about 54,000 probe sets to 100, two groups, a group of genes exhibiting a high signal ratio and a high expression level (a score of 100 or greater) and .a group of genes; exhibiting a lowsignal ratio and- a low expression level (a score of 10. or-smaller), weresubjected to determination of a cDNA signal/genome signal ratio from the aforementioned information using probes of SNPs present on the genomic region carrying introns fixed on the lOOK array (Xbal 5 OK array
18

used herein), thereby to observe a frequency distribution.
As a.result; as shown in Fig. 4, most genes with low expression level (scores of 10 or smaller with the use of U133plus2.0) were found to be present on the|portion, which is considered to be noises as above and exhibits a shape similar to a normal distribution. In contrast, most genes with high expression level (scores of'100 or greater with the use of U133plus2.0) were found to be in the right portion, which is considered to capture signals from cDNA. Accordingly, the cDNA/genome signal ratios determined in this example was correlated with the actual gene expression levels, and long intranuclear RNA comprising introns before being spliced was likely to be more selectively amplified with phi29.
(2) Detection of difference in expression levels between alleles
The cDNA/genome signal ratios for two types of alleles (A and B) in BL1395 and BL2122 cell lines (from feinales) were measured as described in (1) above to determine the RNA (cDNA) levels from allele A and from allele B, and to study differences in expression levels between allele A and allele B, the ratio of the two (i.e., a ratio of the cDNA/genome ratio of allele A to the cDNA/genome ratio of allele B) was determined. Where the ratio is 1:1, the expression levels of the alleles can be equivalent. As a result, statistically significant differences in the expression levels between alleles were observed (i.e., 4.24 and 5.06) in the X chromosome, wherein one allele is well known to be inactivated due to physiological imprinting, as compared with other autosomal chromosomes (1.69 and 2,01), as shown in Table 1.
Table 1

Autosomal chromosome X chromosome t-test p-value
BL1395 1.69. 4.24 . 1.12x10"^
BL2122. 2.01 5.06 1.70x10"*
Thus, it was demonstrated that the measurement of expression levels of a- gene
19

between alleles based on SNPs with the use of lOOK array (50K Xbal array) in cDNA, which was amplified with phi29 polymerase, would enable the determination of an expression level for each allele.
[Example 3]
In this example, differences in expression levels of the PPARG gene between alleles were studied.
Among the genes that had been confirmed to be expressed differentially between alleles in Example 2, PPARG (peroxisome proliferator-activated receptor y) gene could be selected.
The 50K Xbal array included in the lOOK array is designed such that probe sets are comprised at 7 SNP sites in total in the genomic PPARG gene region. When the BL1395 lymphoid cell line was analyzed in Example 2, three SNPs, i.e. rsl051O410, rs10510411 and rs 10510412 (NCBI dbSNP database IDs), which are closely located within 5'-side 300-bp-region of the PPARG gene, were found to be polymorphisms via genomic typing (i.e., two alleles were distinguishable from each other). Fig. 5 shows the position of these 3 SNPs in the PPARG gene. In Fig. 5, open stars indicate an SNP (i.e., informative SNP) that enables determination of differences in expression levels of the PPARG gene between alleles.
The expression ratio between two alleles (i.e., a ratio of the cDNA/genome ratio of allele A to the cDNA/genome ratio of allele B) was 4 times or higher for any of SNPs, as shown in Table 2.
Table 2

SNP id Expression ratio between alleles
rs 10510410 4.55
rs 10510411 4.85
rsl0510412 6.75
20

The existing ratio of alleles carrying these 3 SNPs in the region was determined by direct sequencing. The summarized results are shown in Fig. 6. In the case of rsl0510410 of the BL1395 sample, for example, A/C heterozygotes were found in genomic DNA (c) as shown ia Fig.6, and there was no change in the genomic DNA amplified with phi29. In the case of cDNAs amplified with phi29 (a and b), the signals from allele A were lowered, and thus a substantial waveform consisting of allele C alone was observed. Specifically, rs 10510410 is an A/C heterozygote on the genome; however, the expression level of the gene that is actually expressed from allele C was found to be high. This result was consistent with the result obtained using 50K Xbal array. Expression of the PPARG gene was not observed in BL2122 lymphoid cell line.
Thirty Japanese individuals were subjected to typing via direct sequencing to study the correlation between these 3 SNPs and the expression of the PPARG gene of peripheral blood lymphocytes. Specifically, the correlation between the 3 SNP types and the expression levels of the PPARG gene was analyzed. The expression analysis of the PPARG gene was carried out using the CodeLink, which is the array for expression analysis (Amersham Bioscience), in accordance with a common protocol thereof, and the signals of all probes for each array were averaged so as to bring the median to 1.
Consequently, as shown in Fig. 7 and Fig. 8A (tables and frequency distribution), specimens were classified in accordance with the existence frequency in 30 Japanese individuals as follows: C,- A, and A homozygotes of alleles (m) with a low abundance frequency in rsl0510410, rsl05104]l, and rsl0510412; A, G, and G homozygotes of alleles (M) with a high existence frequency in rsl0510410, rsl 0510411, and rsl0510412; and heterozygotes thereof In the case of homozygotes of alleles with a low abundance frequency (mm homozygotes, shaded in Fig. 7 and in Fig. 8A), the expression level was higher than in the cases of other types of homozygotes (i.e., mM and MM homozygotes) (the average .of mm: 1.58; the average of others; 0.80). Among the 3 top speciniens exhibiting the; highest expression levels, 2 specimens were of the ram type:
Thus, the presence of such SNPs (haplotypes) was found to be correlated with the expression level of the PPARG gene, and the SNP typing was suggested to be
21

effective for determining individual PPARG activity, diagnosing and screening for diseases that may be associated with PPARG, and determining responsiveness to a drug, that targets PPARG.
In the specimens from 30 individuals, the combination of major allele M with minor allele m was completely consistent in 3 SNP sites (rs 10510410, rsl0510411, and rs 10510412). (Fig. 7). Thus, the 3 SNPs were in complete linkage disequilibrium and formed haplotypes. As shown in Fig. 8B, two haplotypes M and m are present, and haplotype m exhibited a higher expression level of PPARG (Fig. 7). By determining SNPs in the haplotypes or in the vicinity thereof that is in linkage disequilibrium, the aforementioned object was considered to be attained.
A genetic polymorphism that enables determination of a gene whose expression level is different between alleles and that is searched for by the method of the present invention, is suggested to be correlated with gene expression level and to influence phenotype.
All publications, patents, and patent applications cited herein are incorporated herein by reference in their entirety.
Industrial Applicability
The present invention provides a method for rapidly and effectively searching for a genetic polymorphism that enables determination of a gene whose expression level is different between alleles. The thus-searched for genetic polymorphism enables distinguishing expression levels between alleles. Accordingly, such a genetic polymorphism can be an effective means for analyzing a phenotype associated- with a gene of interest. Further, if the correlation between the thus searched genetic polymorphism and a phenotype (e.g., a risk- of developing a disease or drug responsiveness) is found, then its finding will be utilizable for studying a cause ofa disease or effective therapeutic methods.
Sequence Listing Free Text
SEQ ID NOs.: 1 to . 3: partial sequences of human peroxisome-proliferator-activated receptors y ("n" represents g or t in SEQ ID NO: 3).
22

WE CLAIM :
1. A method for searching for a genetic polymorphism for identifying a gene
whose expression level is different between alleles, comprising the following steps of:
(a) synthesizing cDNA from total RNA or-intranuclear RNA by reverse transcription using a random primer;
(b) selectively amplifying cDNA derived from long intranuclear RNA as a primary transcript using a random primer and a strand-displacing DNA polymerase that reacts at an isothermal temperature;

(c) detecting a genetic polymorphism present in the amplified cDNA;
(d) comparing the expression levels of cDNAs from the respective alleles on the genomic DNA whose genetic polymorphism is heterozygous, based on the detected polymorphism; and
(e) selecting the genetic polymorphism used for comparison, where the
expression levels of cDNAs from the respective alleles significantly differ from each
other.
2. The method according to claim 1, wherein the DNA polymerase is i|i29 DNA polymerase.
3. The method according to claim 1 or 2, wherein the steps (c) and (d) comprise labeling the amplified cDNAs, which is in turn subjected to hybridization with a genetic polymorphism-specific probe, to compare the expression levels of cDNAs from respective alleles based on the hybridization reaction.
4. The method according to any one of claims 1 to 3, wherein the genetic polymorphism is a single nucleotide polymorphism (SNP).
5. A method for searching for a phenotype-associated genetic polymorphism
23

comprising using a genetic polymorphism searched for by the method according to any one of claims 1 to 4 to evaluate the correlation between the genetic polymorphism or gene expression level and the phenotype.
6. The method according to claim 5, wherein the phenotype is selected from the group
consisting of pathological conditions and severity of a disease, risks of developing a
disease, responsiveness to drugs, responsiveness to foods, responsiveness to chemical
substances, and responsiveness to environmental factors.
7. A method for searching for a phenotype associated with a genetic polymorphism
comprising using a genetic polymorphism searched for by the method according to
any one of claims 1 to 4 to evaluate the correlation between the genetic polymorphism
or gene expression level and the phenotype.
8. The method according to claim 7, wherein the phenotype is selected from the group
consisting of pathological conditions and severity of a disease, risks of developing a
disease, responsiveness to drugs, responsiveness to foods, responsiveness to chemical
substances, and responsiveness to envirormiental factors.