JP3897086B2 - Cancer-related gene expression observation array - Google Patents

Description

【００２７】
表１の計算ではまず、GENBANKから千種類の遺伝子配列を引用した。そのうち、無作為に２０、２００、４００、１０００種類の遺伝子配列セットを作成した。それら遺伝子配列セットからTmが７２℃以上８２℃以下となる３００塩基対のＤＮＡ配列を、各遺伝子について１個無作為に抽出した。それらの無作為抽出ＤＮＡ配列を遺伝子間で総当り的に互いに比較し、ブラストアルゴリズムに従ってホモロジーを計算した。なおブラストアルゴリズムは、Altschul, S. F.ら、Basic local alignment search tool, J. Mol. Biol. Vol.215, p.403-410, 1990に記載されていたものと同一とした。ブラストアルゴリズムで計算した統計的有意水準（ｐ値）が１０％以下の場合、比較された２種類のＤＮＡ配列のホモロジーは十分に高い（危険率１０％以下）とし、クロスハイブリダイゼーションが生じたとした。例えば表１で示した２００遺伝子の場合、２００個の遺伝子からそれぞれ無作為に抽出したＤＮＡ断片同士を比較したところ、７組の遺伝子、すなわち１４個の遺伝子同士でホモロジーが十分に高い（ｐ値＜１０％）という結果がえられた。この２００遺伝子中７組の遺伝子でクロスハイブリダイゼーションが生じたという結果を、クロスハイブリダイゼーションの生じた確率を７％（＝１４遺伝子／２００遺伝子）として表１に記載した。以下同様に、２０遺伝子ではクロスハイブリダイゼーションの確率はゼロ、４００遺伝子では２８％（＝１１２遺伝子／４００遺伝子）、１０００遺伝子では６８％（＝６８０遺伝子／１０００遺伝子）であった。類似機能を有する遺伝子同士やアイソフォーム同士でクロスハイブリダイゼーションが高頻度に見られた。表１示した計算の結果、遺伝子数が増加することでクロスハイブリダイゼーションが起こる確率が高くなることが裏付けられた。固定化する遺伝子の種類が増加することにより、クロスハイブリダイゼーションが避けられないことが分かる。遺伝子数が数千から数万に増加した場合、固定化ＤＮＡ断片の配列決定をよほど慎重に行わなければ、抗悪性腫瘍薬等の作用機序を高精度に解析することはできない。クロスハイブリダイゼーションを皆無とし、かつ融解温度を一定の範囲にそろえてハイブリダイゼーションを高ストリンジェントに行うためには、解析目的に応じて必要最小限の遺伝子数のみをアレイ上に固定化することが望ましい。例えば抗悪性腫瘍薬等の作用機序の解明を目的とする場合、ヒトがんの発生、浸潤、転移の要となる遺伝子を記述するＤＮＡ分子と相補結合するＤＮＡ断片および、薬物代謝の要となる遺伝子を記述するＤＮＡ分子と相補結合するＤＮＡ断片のみを固定化したアレイが良い。
【００２８】
ここで、図４を用いて固定化断片同士のｐ値とクロスハイブリダイセイションとの関係を説明する。固定化DNA断片同士の配列相同性が大きくなる（p値が小さくなる）につれ、クロスハイブリダイゼーションが顕著になることは予想がつく。このｐ値とクロスハイブリダイゼーションとの関係を知るために、以下のような実験を行った。表２に示した遺伝子のうち、IFI56に対応するＤＮＡ断片をガラス表面に固定化した。加えてＰＲＳＭに対応するＤＮＡ断片をガラス表面に固定化した。すなわち２種類の異なるＤＮＡ断片をガラス表面に並べて固定化した。一方、蛍光標識したIFI56遺伝子断片を、ガラス表面にふりかけて、ハイブリダイズさせた。IFI56遺伝子断片がIFI56に対応する固定化ＤＮＡ断片とのみハイブリダイズし、ＰＲＳＭに対応するＤＮＡ断片にはハイブリダイズしなければ、クロスハイブリダイゼーションは零である。すなわち、（ＰＲＳＭに対応するＤＮＡ断片とのハイブリダイズに由来する蛍光強度）と（IFI56に対応するＤＮＡ断片とのハイブリダイズに由来する蛍光強度）の比率は零となる。以後、この比率を信号強度比と呼ぶ。またIFI56遺伝子断片が、IFI56に対応するＤＮＡ断片とＰＲＳＭに対応するＤＮＡ断片の両者とハイブリダイズする場合、信号強度比は１になる。すなわちクロスハイブリダイゼーションが頻繁に起こるほど、信号強度比は１に近づく。図４に信号強度比とｐ値との関係をしめした。ｐ値が小さい（すなわち２種類の固定化ＤＮＡ断片同士の相同性が大きい）ほど、信号強度比が高い（クロスハイブリダイゼーションが多い）ことが分かる。但しｐ値が１０％を超えた時点で、信号強度比率は、０．０１程度まで小さくなった。この結果から、固定化ＤＮＡ断片同士のｐ値が１０％を超えるようにすれば、クロスハイブリダイゼーションを防ぐことが十分可能となることが分かる。
【００２９】
一方、ＤＮＡ分子同士のハイブリダイゼーションは、１本鎖ＤＮＡ同士で生じる現象である。１本鎖ＤＮＡは通常、溶液中ではその一部が折れ曲がって数個の相補的な塩基対を形成している。この構造はヘアピン構造と呼ばれる。このヘアピン構造のうち、異常に熱安定性が高いヘアピン構造が存在することが分かっている（平尾と三浦、１本鎖ＤＮＡの特殊構造 異常に熱安定性が高いミニヘアピン構造、蛋白質核酸酵素、vol.40, p.1583-1592,1995）。このミニヘアピン構造、もしくは特異な高次構造をとる短配列は、例えば、
GCGAAAGC（配列番号１）
GCGAAGC（配列番号２）
である。前者の（配列番号１）のTmは７６℃にも達する。この７６℃という温度は一般のハイブリダイゼーション温度（４０℃から６２℃）と比較して有意に高い。このため、前記配列番号１が固定化ＤＮＡ断片中に含まれた場合、ハイブリダイゼーション時、その配列部分ないしはより広範囲の領域で相補結合を形成しない確率が非常に高い。相補結合を形成しない状態とは、ある意味でミスマッチの状態と同一である。前述のように、２つのＤＮＡ間でミスマッチがあれば、両ＤＮＡのTmが低下し、ハイブリダイゼーション精度が低下することになる。このため例えば配列番号１や配列番号２のようなミニヘアピン構造はできる限り、固定化ＤＮＡ断片中に含まれない方がよい。このミニヘアピン構造を生じる可能性のある他の配列としては、短繰り返し配列、例えば、
GGGCGGCGGG（配列番号３）
TTTCATTATTGAAA（配列番号４）
等が考えられる。
【００３０】
ヒトＤＮＡにはヒト特有の、Alu配列と呼ばれる、反復配列を多く含む配列の存在が知られている。このAlu配列はヒトＤＮＡに数多く含まれている。またAlu配列は、メッセンジャーＲＮＡ内、特に５‘非翻訳領域または３’非翻訳領域にも存在する。仮に、測定対象のメッセンジャーＲＮＡサンプル中にＤＮＡ分子が混入していた場合、メッセンジャーＲＮＡを鋳型として蛍光標識をする工程中、混入ＤＮＡ内もAlu配列部分の存在により蛍光標識されてしまう可能性がある。そしてAlu配列部分を含む標識された混入ＤＮＡと、固定化ＤＮＡ断片とが相補結合することも考えられる。本来メッセンジャーＲＮＡのみを測定すべきが、ＤＮＡをも測定してしまう恐れがある。この混入ＤＮＡによる誤った解析を避けるためには、固定化ＤＮＡ断片中にAlu配列と相同性が有意に高い部分が含まれないことが望ましい。またAlu配列には反復構造が多く含まれるので、溶液中で、高次構造をとる可能性があり、前記のミニヘアピン構造と同様にミスマッチを引き起こすことが十分に考えられる。この意味でも固定化ＤＮＡ断片中にAlu配列と相同性が有意に高い部分が含まれないことが望ましい。例えば、Alu配列と固定化ＤＮＡ断片とのホモロジーがブラストアルゴリズムで算出した統計的有意水準（ｐ値）で１０％を超えるようにすれば、配列相同性が有意に低くなると考えられる。
【００３１】
ここで、 Alu配列と固定化ＤＮＡ断片との配列相同性とノイズ、Ｓ／Ｎ比との関係について図５を用いて説明する。 Ａｌｕ配列と固定化ＤＮＡ断片との配列相同性（ｐ値）が、測定結果に及ぼす影響を観察するために、以下のような実験を行った。まずガラス表面にIFI56に対応するＤＮＡ断片を固定化した。それに対して、表２に示した１０種類の遺伝子を蛍光標識し、ガラス表面にふりかけてハイブリダイズさせた。最も好ましくは、IFI56に対応する蛍光信号のみが観察されて、それ以外の信号は観察されないのが望ましい。今、Alu配列とのｐ値が２０％以上の配列（Alu配列とのホモロジーが十分低い）での蛍光強度を１に規格化する。Alu配列とのｐ値が、１０％、５％、１％、０．０１％以下の場合の、それぞれの蛍光強度（Alu配列とのｐ値が２０％の値を１に規格化した値）を図５に示す。すると、ｐ値が小さくなるにつれ、蛍光強度が増加することが分かった。この増加分は本来ハイブリダイズしないはずのＤＮＡ配列がハイブリダイズするために生じたものなので、ノイズである。ノイズ（すなわち、それぞれの蛍光強度から１を引いたもの）を図５に示す。ｐ値１０％でほぼ、ノイズは０．１であるが、ｐ値が１０％以下になると増加していく。また蛍光強度÷ノイズをＳＮ比率（S/N）と定義する。それぞれのＳＮ比率を計算したところ、ｐ値が１０％をきるにつれ、ＳＮ比が小さく（測定感度が悪く）なった。図５から、Alu配列とのｐ値が１０％以下の固定化ＤＮＡ断片では、繰り返し配列や、サンプル中に含まれるコンタミＤＮＡ由来のＡｌｕ配列の存在により、測定感度が有意に悪化することが分かった。Alu配列と固定化ＤＮＡ断片とのｐ値は１０％以上とすることで、測定感度の悪化を防ぐことができる。
【００３２】
前記、（１）クロスハイブリダイゼーション、（２）特異な高次構造をとる短配列や短繰り返し配列、（３）Aluとの類似配列等の存在が実際のハイブリダイゼーションに及ぼす影響を調べるために、１０種類の遺伝子を記述するメッセンジャーＲＮＡの相補ＤＮＡ（cＤＮＡ）配列を、ガラス表面に固定化した実験を行った。この固定化ＤＮＡ断片（正確にはcＤＮＡ断片だが、以後ＤＮＡ断片と呼ぶ）は、GENBANKの配列から無作為に３００、６００、９００、１２００、１５００、２１００塩基の領域を選択して合成したものである。測定サンプルは、ヒトすい臓ガン細胞（細胞名称CFPAC1）から抽出したメッセンジャーＲＮＡ群を鋳型とし、オリゴｄTプライマーを用いた逆転写反応により合成されたｃＤＮＡとした。なお逆転写して合成されたｃＤＮＡは、逆転写反応時に蛍光標識（蛍光試薬名Cy3-dCTP）された。実験で用いた１０種類の遺伝子の略号、GENBANKのアクセション番号、遺伝子名称を表２に示す。
【００３３】
【表２】

【００３４】
図１に、実験により得られた、固定化ＤＮＡ断片の塩基長がハイブリダイゼーションに及ぼす影響を示す。図中の１０個の折れ線グラフのそれぞれは表２で示した１０遺伝子を用いた実験の結果である。折れ線グラフ中に遺伝子略号（表２に示す）を記した。折れ線グラフの横軸は固定化ＤＮＡ断片の塩基長を示す。折れ線グラフの縦軸は蛍光強度を示す。但し縦軸の蛍光強度は３００塩基長での蛍光強度を１として規格化した。蛍光強度は背景（バックグランド）の蛍光強度を差し引いた、正味の蛍光強度とした。ハイブリダイゼーション温度は６２℃とした。蛍光強度が高いほど、信号強度比（SN比）が高い、すなわち検出感度が高いことに相当する。なお折れ線グラフ中の値は、３回の実験の平均値、エラーバーはその標準偏差である。また各折れ線グラフの下方の横線は、各長さのＤＮＡ断片の相対位置関係を示している。５‘末端を左に、３’末端を右にして示した。上から順番に塩基長の長いものから短いものとなっている。横線がオーバラップしているところは、それぞれのＤＮＡ断片で共通の配列部分である。
【００３５】
図１の結果を見ると、IFI56, GSSの２遺伝子を対象とした場合、固定化ＤＮＡ断片長が長いほど、緩やかではあるが、蛍光強度すなわち検出感度が高くなる傾向が見られた。またPRSM, UBE, AKT, ADPRT、SMRTの５遺伝子では、測定誤差（エラーバー）を考慮すると、固定化ＤＮＡ断片長によらずほぼ一定の検出感度が得られた。前記のＤＮＡ断片長さが長いほど検出感度が増加する現象は、ＤＮＡ分子の長さが長いほどＤＮＡ分子同士の結合力の安定性が上昇することで説明できる。ＤＮＡ分子同士の結合力の安定性は、熱力学における自由エネルギー（ΔＧ）で見積もることができる。ΔＧを計算すると塩基長が長いほど小さい値が得られる。ΔＧが小さいほど熱力学的安定性が増加するのだ。なお、ＤＮＡ分子同士のΔＧの値は最近接塩基法により計算した（J. SantaLucia, Jr. A unified view of polymer, dumbbell, and oligonucleotide ＤＮＡ nearest-neighbor thermodynamics, Porc. Natl. Acad. Sci. USA, vol.95, p.1460-1465, 1998）。
【００３６】
しかし、CASP, PLG, GEMの３遺伝子では、統計的有意にＤＮＡ分子の長さと蛍光強度、すなわち検出感度が線形相関していない。CASPでいうと６００塩基、CASPでいうと６００塩基、GEMでいうと６００と１２００塩基で、明らかに蛍光強度が低い。この結果は（マイナス要因１）クロスハイブリダイゼーション、（マイナス要因２）特異な高次構造をとる短配列や短繰り返し配列、（マイナス要因３）Aluとの類似配列等の存在といったマイナス要因となる塩基配列部分が、該当部分に含まれているためではないかと推察し、解析を行った。その結果、CASPの６００塩基、GEMの６００塩基と１２００塩基の配列で、その他の配列と重複しない領域において、前記マイナス要因２が存在した。具体的にはCASPの６００塩基ではGCGAAAGC（配列番号１）もしくはその相補、反転、相補反転配列のいずれかが存在し、GEMの６００塩基と１２００塩基ではGCGAAGC（配列番号２）もしくは相補配列、反転配列、相補反転配列のいずれかが存在した。なおGCGAAAGCの相補配列はCGCTTTCG、反転配列はCGAAAGCG、相補反転配列はGCTTTCGCである。但し、PLGについては（１）から（３）のいずれにも該当しなかった。このPLGの結果から、まだ知られていないマイナス要因が存在する可能性が示唆された。図１の結果からも改めて、特異な高次構造をとる短配列や短繰り返し配列の存在が好ましくないことが確認された。また、固定化ＤＮＡ断片の熱力学における自由エネルギー（ΔＧ）を小さくすることは、少なくともマイナスにはならず、むしろプラスに作用することも確認された。例えば塩基長を３００塩基とした場合、ΔＧは、マイナス１４０からマイナス２５０程度が好ましい。但しΔＧの値は塩基長が長くなるほど小さくなり、塩基配列によっても異なる値をとるので、明確なしきい値を設けることは難しい。
【００３７】
図１の結果等を踏まえると、１枚のアレイ上に固定化するＤＮＡ断片は、ＤＮＡ断片のTmがハイブリダイゼーションの温度から３０℃を超えないこと、言いかえればある一定の範囲を超えないこと、そのTm条件を満たした上で自由エネルギー（ΔＧ）を出来る限り小さくすることが重要である。更に加えてクロスハイブリダイゼーションの原因配列の除外、特異な高次構造をとる短配列や短繰り返し配列（以後、有害短配列と呼ぶ）の除外、Aluとの類似配列の除外を行うことが好ましい。これらの条件を全て満足するためには、例えば図２に示した計算アルゴリズムで、固定化ＤＮＡ断片の配列決定を行えばよい。図２をアルゴリズムの順に従って説明する。はじめにFASTA形式等で記述されたＤＮＡもしくは、メッセンジャーＲＮＡの配列ファイルを読み込み工程（１）、続いて塩濃度やハイブリダイゼーション温度といった実験条件を入力する工程（２）、そして固定化ＤＮＡ断片の長さ範囲を入力する工程（３）を経て、固定化ＤＮＡ断片の候補となる複数のＤＮＡ配列を選択し、リストにする。続いて、そのリスト中のＤＮＡ配列候補の一つ一つについてTmを計算し、そのTmがハイブリダイゼーション温度（Th）と比較してある一定の範囲、例えばTh−αからTh−β（但しα、βの値は正負のどちらの値でもよい）を外れているＤＮＡ配列を候補リストから除外する工程（４）、特異な高次構造をとる短配列や短繰り返し配列のあるＤＮＡ配列を候補リストから除外する工程（５）、Alu配列とのホモロジーの高いＤＮＡ配列を候補リストから除外する工程（６）、他遺伝子配列とのホモロジーの高いＤＮＡ配列を候補リストから除外する工程（７）を行う。そして工程４から工程７の全ての条件を満足する配列のいずれかを固定化ＤＮＡ断片配列とする。複数の配列が工程４から工程７までを満足した際は、例えば自由エネルギー（ΔＧ）が最小の配列を固定化ＤＮＡ断片配列とすればよい。
【００３８】
工程５では例えば、文字列のパターンマッチングアルゴリズムを用いる。パターンマッチングの対象としては、有害短配列の相補配列、反転配列（GCGAAAGCであれば、CGAAAGCG）、相補反転配列について対象とすることが望ましい。計算速度を向上させるために文字列をいったん数字に変換しビットシフト計算等により比較してもよい。
【００３９】
工程６と工程７では、ブラスト(blast)アルゴリズム、ファスタ(FASTA)アルゴリズムなどの探索的(heuristic)アルゴリズムを用いてもよいし、スミスウォータマン(Smith-Waterman)法などのローカルアライメント(local alignment)アルゴリズムや、隠れマルコフモデル(Hidden Markov model)を用いたアルゴリズムなどを用いても良い。計算速度を向上させるために文字列をいったん数字に変換しビットシフト計算等を行ってもよい。
【００４０】
また工程７では、１枚のアレイ上に固定化する遺伝子配列同士のホモロジーを計算するのみならず、ＤＮＡ配列とＧＥＮＢＡＮＫ等のヒト遺伝子配列とのホモロジーを計算してもよい。測定対象試料には、固定化ＤＮＡ断片と本来ハイブリダイズする遺伝子以外にも、数多くの別遺伝子が含まれている。それら別遺伝子がハイブリダイズした場合、本来、個々の固定化ＤＮＡ断片がハイブリダイズすべき遺伝子のみの測定ができなくなる。そこで、ＤＮＡアレイ上に固定化するＤＮＡ断片候補の配列と、測定対象試料に含まれている可能性のある遺伝子群のＤＮＡ配列とを比較して、ホモロジーが有意に高いＤＮＡ配列は、固定化ＤＮＡ断片としては選択しないことが望ましい。ＧＥＮＢＡＮＫ等の遺伝子配列データベースには現在までに知られているヒトないしはげっ歯類、酵母、大腸菌などの配列が全て格納されている。固定化ＤＮＡ断片と、例えばＧＥＮＢＡＮＫ等のヒトなどの遺伝子配列とのＤＮＡ配列相同性が、ブラスト(Blast)アルゴリズムにより算出した統計的有意水準（ｐ値）で１０％を超えることとすれば、より高精度の測定が行える。
【００４１】
なおＤＮＡ配列のTmの計算には、前述の最近接塩基法（J. SantaLucia, Jr. A unified view of polymer, dumbbell, and oligonucleotide ＤＮＡ nearest-neighbor thermodynamics, Porc. Natl. Acad. Sci. USA, vol.95, p.1460-1465, 1998）と、経験式に基づく方法（Wetmure, J.G., ＤＮＡ probes: Applications of the principles of nucleic acid hybridization, Criti. Rev in Biochem. And Mol. Biol., vol.26, p.227-259, 1991）の２つがある。５０塩基対以下のＤＮＡ配列については前者の最近接塩基法がよく合致し、５０塩基対以上のＤＮＡ配列に対しては後者の経験式に基づく方法がよく合致するという報告もある。しかし前記J. SantaLucia, Jr.の論文によると、最近接塩基法も改良されており、配列長さによらず計算できることが示唆されている。そこで、ＤＮＡ配列のTm計算には、塩基長さに応じて最近接塩基法もしくは経験式に基づく方法のいずれか片方をもちいる、あるいは両者の方法で計算して得られた値の例えば平均値などを用いれば良い。
【００４２】
【発明の実施の形態】
本発明を例えば、抗悪性腫瘍剤等の薬剤の作用機序解明に用いる際の実施の形態について以下に記す。ガラス表面に固定化したヒトがんの発生、浸潤、転移の要となる遺伝子および、薬物代謝の要となる遺伝子のリストを表３から表１８に示す。
【００４３】
【表３】

【００４４】
【表４】

【００４５】
【表５】

【００４６】
【表６】

【００４７】
【表７】

【００４８】
【表８】

【００４９】
【表９】

【００５０】
【表１０】

【００５１】
【表１１】

【００５２】
【表１２】

【００５３】
【表１３】

【００５４】
【表１４】

【００５５】
【表１５】

【００５６】
【表１６】

【００５７】
【表１７】

【００５８】
【表１８】

【００５９】
続いて、図１のアルゴリズムを用いて計算した、表３から表１８に示した遺伝子の一部に対応する固定化ＤＮＡ断片配列リストを表１９から表３８に示す。
【００６０】
【表１９】

【００６１】
【表２０】

【００６２】
【表２１】

【００６３】
【表２２】

【００６４】
【表２３】

【００６５】
【表２４】

【００６６】
【表２５】

【００６７】
【表２６】

【００６８】
【表２７】

【００６９】
【表２８】

【００７０】
【表２９】

【００７１】
【表３０】

【００７２】
【表３１】

【００７３】
【表３２】

【００７４】
【表３３】

【００７５】
【表３４】

【００７６】
【表３５】

【００７７】
【表３６】

【００７８】
【表３７】

【００７９】
【表３８】

【００８０】
最後に、表３から表１８に示した遺伝子をガラス表面に固定化したＤＮＡアレイを用いて、抗悪性腫瘍剤（薬品名：５−ａｚａ−２‘−ｄｅｏｘｙ−ｃｙｔｉｄｉｎｅ）に対するヒト培養細胞（細胞名：ＨＴ２９）の遺伝子発現分布の測定結果を図３に示す。以下、順に従って説明する。
【００８１】
（１）ＤＮＡアレイ上の遺伝子リスト
表３に、本願請求項２記載の基準に従って選択したヒトがんの発生、浸潤、転移の要となる遺伝子および、本願請求項３記載の基準に従って選択した薬物代謝の要となる遺伝子のリストを示す。それぞれの遺伝子配列に対応するGENBANKのアクセション番号と、遺伝子名を記載した。
【００８２】
（２）遺伝子リストのうちの一部配列リスト
表４に、表３に示した遺伝子の一部に対応する固定化ＤＮＡ断片配列のリストを示す。それぞれのＤＮＡ断片配列、ＤＮＡ断片配列の塩基長(length)、ＤＮＡ断片配列の自由エネルギー（dＧ），経験式に基づく方法で計算したＤＮＡ断片配列の融解温度（Tm１）、最近接塩基法で計算したＤＮＡ断片配列の融解温度（Tm２）を併記した。ここでは融解温度(Tm)は、Tm１とTm２の平均値とした。表４のＤＮＡ断片配列のTｍ（＝0.5＊（Tm１＋Tm２））が７２℃から８２℃となるように設計した。
【００８３】
（３）実験結果の一例
図３に、ヒト結腸癌培養細胞（名称：ＨＴ２９細胞）に、５−ａｚａ−２‘−ｄｅｏｘｙ−ｃｙｔｉｄｉｎｅ（以下５−Ａｚａ−ＣｄＲ）処理した場合に、処理しない場合と比較して増加したメッセンジャーＲＮＡ量の増加率を、遺伝子ごとに示す。５−Ａｚａ−ＣｄＲ処理とは、５００ナノモルの５−Ａｚａ−ＣｄＲに細胞をそれぞれ２、１２、２４時間さらすことを意味する。それぞれの時間さらされたＨＴ２９細胞から、すぐにメッセンジャーＲＮＡを抽出した。同様に、薬剤処理をしないＨＴ２９細胞も同一の培養機中で培養し、それぞれの時間経過後、すぐにメッセンジャーＲＮＡを抽出した。オリゴｄTプライマーを用いた逆転写反応により、薬剤処理した細胞のメッセンジャーＲＮＡについては、Cｙ５−ｄCTPを用いて蛍光標識されたｃＤＮＡを合成した。また薬剤処理しない細胞のメッセンジャーＲＮＡについては、Cｙ３−ｄCTPを用いて蛍光標識されたｃＤＮＡを合成した。薬剤処理した細胞由来のｃＤＮＡ（Cｙ５標識）と薬剤処理しない細胞由来のｃＤＮＡ （Cｙ３標識）を混合して、同一のＤＮＡアレイとハイブリダイズさせた。ハイブリダイゼーション温度は６２℃、ハイブリダイゼーション時間は１２時間とした。ハイブリダイゼーション後、蛍光スキャナーによりCｙ５とCｙ３のそれぞれの蛍光強度を数値化した。
【００８４】
薬剤処理した細胞由来の蛍光強度（Cｙ５蛍光強度）と薬剤処理しない細胞由来蛍光強度（Cｙ３蛍光強度）の比、すなわちCｙ５蛍光強度／Cｙ３蛍光強度を値を計算し、図３の縦軸の値、メッセンジャーＲＮＡ量の増加率としてプロットした。図３の横軸はアレイ上の遺伝子の通し番号である。この実験の結果、Cｙ５蛍光強度／Cｙ３蛍光強度が３倍以上であった遺伝子は、例えば２時間後で、SIAT4A、PRSM1、ribosomal protein S5、plasminogen、ETR103、OCTS3、integrin 4、GEM、RGS14、HsCds18、IFN-induced protein 56, INFα-inducible protein 27、INF-induced 17kDa protein, INF-induced protein IFI-6-16, INF-inducible protein 10などであった。そして２、１２、２４時間を通じて、常にCｙ５蛍光強度／Cｙ３蛍光強度が３倍以上であったのはインタフェロン(IFN)αに関連する遺伝子群、IFN-induced protein 56, INFα-inducible protein 27、INF-induced 17kDa protein, INF-induced protein IFI-6-16, INF-inducible protein 10などであった。この結果は、Karpfらの報告（Karpf, A.R.ら、Inhibition of ＤＮＡ methyltransferase stimulates the expression of signal transducer and activator of transcription 1, 2, ad 3 genes in colon tumor cells, Proc. Natl. Acad. Sci. USA, vol.96, p.14007-14012, 1999）らの知見とも合致した。さらKarpfらの報告に対し、本発明では薬剤応答の時間依存の様子を再現性よく観察することができた。
【００８５】
本発明を実施することで、抗悪性腫瘍剤の一種である５−Ａｚａ−ＣｄＲに対するヒト細胞の薬剤応答の様子の時間依存性を再現性良くかつ高精度で観察した。５−Ａｚａ−ＣｄＲ以外の薬剤についても同様な結果を得ることができる。
【００８６】
【発明の効果】
抵悪性腫瘍薬等の作用機序を高精度で解明するためのＤＮＡアレイ、及びその製造方法を提供する。
【図面の簡単な説明】
【図１】固定化ＤＮＡ断片の塩基長がハイブリダイゼーションに及ぼす影響。
【図２】ＤＮＡ断片配列決定計算アルゴリズム。
【図３】ＨＴ２９細胞に、５−ａｚａ−２‘−ｄｅｏｘｙ−ｃｙｔｉｄｉｎｅ処理した場合に、処理しない場合と比較して増加したメッセンジャーＲＮＡ量の増加率。
【図４】固定化ＤＮＡ断片同士の配列相同性とクロスハイブリダイゼーションとの関係を示す図。
【図５】Ａｌｕ配列と固定化ＤＮＡ断片との配列相同性と測定結果との関係を示す図。
【符号の説明】
１.遺伝子配列ファイルの読み込み工程、２.塩濃度、ハイブリダイゼーション等の実験条件の入力工程、３.固定化ＤＮＡ断片の長さ範囲の入力工程、４. 固定化ＤＮＡ断片の融解温度を計算し、その融解温度がある一定の範囲を外れているＤＮＡ断片を候補リストから除外する工程、５. 特異な高次構造をとる短配列や短繰り返し配列のあるＤＮＡ ＤＮＡ断片を候補リストから除外する工程、６.Alu配列とのホモロジーの高いＤＮＡ断片を候補リストから除外する工程、７.他遺伝子配列とのホモロジーの高いＤＮＡ断片を候補リストから除外する工程。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a DNA array for elucidating the mechanism of action of an anti-neoplastic agent and the like, and a production method thereof.
[0002]
[Prior art]
Treatment of malignant tumors includes surgical therapy, radiosurgery therapy and chemotherapy. Chemotherapy for malignant tumors is intended to suppress selective growth and proliferation of tumor cells by utilizing the difference in sensitivity between normal cells and malignant tumor cells for antitumor drugs. However, antineoplastic drugs are not qualitatively different in their effects on tumor cells and normal cells, and are therefore toxic to normal cells. Therefore, a drug having a greater effect on a malignant tumor has a greater influence on normal cells and a greater possibility of causing serious side effects. However, the development of more powerful and effective drugs has always been made to counter the acquisition of tumor cell resistance (the appearance of tumor cells that are resistant to the drug by the continuous administration of antineoplastic drugs). It is desired. In addition, there are many cancers that are difficult to cure with drugs such as non-small cell lung cancer, pancreatic cancer, malignant melanoma, and kidney cancer, which is why the development of more powerful antineoplastic drugs is awaited.
[0003]
In order to develop a good antineoplastic drug, it is desired that the side effect is small as well as the strength against tumor. In order to satisfy these contradictory conditions, it is necessary to perform drug discovery after understanding the mechanism (mechanism of action) of the drug. For this purpose, it is desirable to understand the mechanism of action of drugs by digging down to the gene level.
[0004]
The genetic information described in the gene on the DNA molecule is transcribed into messenger RNA and translated into protein in the ribosome. This process from transcription to messenger RNA and translation to protein is called gene expression. In order to understand the mechanism of drug action at the genetic level, what genes and proteins are involved in the process from drug administration to metabolism and actual tumor toxicity, It is desirable to track over time. This is because gene and protein interactions and pathways (gene pathways) are revealed by this tracking. Once the genetic pathway is clarified, treatment strategies can be improved, such as reducing side effects by using other drugs that act on the genes on the pathway. It also provides guidelines for improving the drug itself.
[0005]
In order to clarify the genetic pathway, there are (1) a method for observing quantitative changes in messenger RNA and (2) a method for examining quantitative changes in protein and spatial distribution. The former includes DNA array (or DNA chip), Northern hybridization, and RT-PCR method. Examples of the latter include Yeast Two Hybrid and antibody staining. Among the above methods, as a technique for simultaneously viewing changes in several tens to several hundreds of genes that will be involved in the action of a drug, a method for viewing messenger RNA of (1), particularly a technique called a DNA array method, It is most suitable from the viewpoint of simplicity, throughput, and cost, and is currently being actively researched.
[0006]
DNA array method can measure thousands to tens of thousands of messenger RNAs at once (Duggan, DJ et al., Expression profiling using cDNA microarrays, Nature genetics supplement, vol.21, p10-14, 1999) . In general, a DNA array is obtained by immobilizing a complementary DNA (complementary DNA or cDNA) having a sequence complementary to messenger RNA as a probe on a glass slide (Lockhart, DJ et al., Expression monitoring by hybridization to high- density oligonucleotide arrays, Nature biotechnology, vol.14, p.1675-1680, 1996). The experimenter extracts messenger RNA from cells, organs, etc., and then synthesizes fluorescently labeled cDNA from messenger RNA using reverse transcriptase and a fluorescent labeling substance. Thereafter, the fluorescently labeled cDNA is sprinkled on the DNA array. Subsequently, the DNA array is held at a constant temperature for a fixed time, and the fluorescently labeled cDNA and the DNA array are complementarily bound (hybridized). Thereafter, the DNA array is washed, and the non-specifically bound fluorescently labeled cDNA is washed away (the non-specifically bound cDNA has a complementary binding force compared to the specifically bound cDNA). Finally, the fluorescence signal intensity corresponding to the individual probes immobilized on the DNA array is measured. The amount of messenger RNA is proportional to the fluorescence signal intensity hybridized with the corresponding probe. In this way, the amount of messenger RNA corresponding to various genes in organs and cells can be measured. For example, by taking out messenger RNA from cells or organs at every elapsed time after drug administration and measuring it by the DNA array method, it is possible to see the time change of the amount of messenger RNA for each gene. An increase in the amount of messenger RNA means that genetic information from the DNA molecule is actively transcribed, and corresponds to an increase in the function of the gene. By analyzing the time change of the function of this gene, information on the gene pathway can be obtained.
[0007]
In recent years, DNA pathways have been used to determine gene pathway information. In recent years, almost all yeast genes have been immobilized on a single array, and quantitative changes in messenger RNA when stimuli such as drug administration are applied are measured. An attempt has been made (DeRisi, JL et al., Exploring the metabolic and genetic control of gene expression on a genomic scale, Science, vol.278, p.680-686, 1997, Roberts, CJ et al., Signaling and circuitry of multiple MAPK pathways revealed by a matrix of global gene expression profiles, Science, vol.287, p.873-880, 2000). The relationship between drug administration and gene pathway has also been investigated by observing quantitative changes in messenger RNA after drug administration by destroying part of the yeast gene (Marton, MJ et al., Drug target validation and identification of secondary drug target effects using DNA microarrays, Nature Medicine, vol.4, p.1293-1301, or Stoughton et al. Methods for identifying pathways of drug action, US patent 5965352). However, the mechanism of action of drugs in yeast is not the same as the mechanism of action of drugs in humans. This is because the function and number of genes differ between yeast and human. In particular, disruption of yeast genes has been shown not only to change messenger RNA levels but also to changes in the number of chromosomes (Hughes, TR et al., Widespread aneupolidy revealed by DNA microarray expression profiling, Nature genetics, vol.25, p.333-337, 2000). This change in the number of chromosomes is a cause of serious diseases such as Down's syndrome in humans, and almost no phenomenon in normal cells excluding cancer cells. From this point of view, care must be taken when extrapolating knowledge obtained from yeast, particularly yeast that has disrupted a specific gene, to humans. It is desirable to accumulate knowledge on human cells rather than relying solely on information obtained from yeast.
[0008]
What is the difference in the amount of messenger RNA transcribed from individual genes (about 9700) when various anticancer drugs are administered to multiple cultured human cancer cells and when not administered? There is an example of studying whether it appears (Scherf, U. et al., A gene expression database fot the molecular pharmacology of cance RNAture genetics, vol. 24, p. 236-244, 2000). In this study, a cluster analysis was performed to determine which gene messenger RNA level was changed in each of various cultured human cancer cells for a certain drug. In addition, a cluster analysis of which gene messenger RNA amount was changed at the time of administration of various drugs to a certain human cancer cultured cell was performed at the same time.
[0009]
[Problems to be solved by the invention]
However, regarding the results of examining the difference in the amount of RNA described above, if the response mechanism to various drugs is originally common to each cell, the results of both cluster analyzes must be common. However, according to the above-mentioned paper by Scherf et al., The results of cluster analysis between cell types and messenger RNA and the results of cluster analysis between drug types and messenger RNA were very different. After all, the above-mentioned paper by Scherf et al. Does not provide new information on gene pathways. It should be noted that in the DNA array used by Scherf et al., Changes in the amount of messenger RNA corresponding to about 9700 genes have been measured with high accuracy separated for each gene, and it has not been examined. . As will be described later, it is considered that experiments separated with high accuracy for each gene can be performed only after a probe design technique for ensuring measurement accuracy is applied.
[0010]
As will be described later, since an existing array having several thousand to several tens of thousands of genes has a large number of genes to be immobilized, it is difficult to separate each gene with high accuracy. For this reason, an array having thousands to tens of thousands of genes is suitable for gene discovery, but is not necessarily suitable for analysis use for observing gene pathways.
[0011]
[Means for Solving the Problems]
In order to analyze the mechanism of action of antineoplastic drugs with high accuracy, it is inevitable that a DNA fragment that should bind only to one type of gene binds to another gene (cross-hybridization). It is obvious that it must not be. This becomes more difficult as will be described later as the number of genes immobilized on one array increases. Therefore, it is very difficult to eliminate cross-hybridization between each gene in a DNA array for searching which has thousands to tens of thousands of genes. In order to solve this problem, if the purpose of using a DNA array is elucidation of the mechanism of action of, for example, antineoplastic agents, only the genes related to the mechanism of action of the drug are minimized as much as possible ( as small as reasonably achievable). In addition, in a DNA array for yeast, there is a limit in reflecting the knowledge directly on humans. This is also a problem with DNA arrays targeting experimental animals such as mice and rats. In order to solve this problem, an array using human genes is preferable.
[0012]
In order to collect only the minimum necessary genes related to the mechanism of action of anti-neoplastic agents, the process after the drugs such as anti-neoplastic agents are taken into the body will be considered. After being taken into the body, the drug is metabolized by drug metabolizing enzymes with few exceptions. Some drugs lose their efficacy after being metabolized, some drugs exhibit their efficacy only after being metabolized, and others exhibit undesirable toxicity when metabolized. Therefore, in order to analyze the mechanism of action of antineoplastic drugs and the like at the gene level, it is necessary to analyze at least gene expression related to drug metabolism.
[0013]
In addition, drugs such as antineoplastic agents are originally expected to have medicinal effects such as killing malignant tumors, suppressing occurrence, suppressing invasion, and suppressing metastasis. A side effect is a case where an effect different from the expectation is brought about, and the side effect should be reduced as much as possible. Therefore, it is necessary to analyze the expression of genes related to cancer, particularly genes related to cancer occurrence, invasion and metastasis.
[0014]
That is, in order to analyze the efficacy of antineoplastic drugs with high accuracy, a DNA array in which (1) a drug metabolism-related gene and (2) a cancer-related gene are immobilized to a minimum amount is most suitable. . Next, the reasons for selecting drug metabolism-related genes and cancer-related genes to be immobilized on the DNA array will be shown.
[0015]
(A) Reasons for selecting drug metabolism-related genes
Drugs taken by the body are metabolized as much as possible. Various metabolic enzymes are involved in this metabolism. And the drug metabolic reaction involving the metabolic pathway in the living body usually does not occur randomly. Also, certain special drug metabolism pathways usually do not function alone. It has been found that the activity of one pathway affects the activity of the other pathway and mutually controls metabolism. One theory is that about a thousand genes are involved in drug metabolism, including minor ones. Drug metabolism reactions are generally classified into first phase reactions (functional group introduction reactions) and second phase reactions (conjugation reactions). In the first phase reaction, chemical reactions such as oxidation, reduction, hydrolysis, hydration, dethioacetylation, isomerization occur, and in the second phase reaction, glucuronic acid or glucose conjugation reaction, sulfate conjugation, methyl conjugation, acetyl conjugation, Each chemical reaction such as amino acid conjugation, glutathione conjugation, fatty acid conjugation, and condensation occurs.
[0016]
Covering all the various drug metabolism-related enzymes as described above results in an increase in the number of DNA fragments, which is contrary to the achievement of high-accuracy analysis without cross-hybridization as described above. Therefore, the major (major) contribution is selected from the first phase reaction and the second phase reaction. In humans, cytochrome 450 genes (CYP) 1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4, CYP3A7, CYP3A7, CYP3A7, CYP3A7, CYP3A7 Typical examples include dehydrogenase, dihydroxypyrimidine dehydrogenase, NADPH-cytochrome P450 reductase, DT-diaphorase, esterase, and epoxide hydrase, and more than 90% of the total drug metabolism is said to be due to the function of the gene ( Evans, WE and Relling, M., Pharmacogenomics: Translating functional genomics into rational therapeutics, Science, vol. 266, p. 487-491, 1999). Further, catechol O-methyl transferase, glutathione S-transferase, histamine methyl transferase, N-acetyl transferase, sulfotransferase, thiopurine methyl transferase, uridine 5′-triphosphate glucuronyl transferase are genes associated with the second phase reaction. It is also said to be involved in more than 80% of drug metabolism (Evans, WE and Relling, M., ibid). Thus, with regard to drug metabolism of at least 80% or more, an approximate drug metabolism mechanism can be analyzed by looking at the gene expression of the above gene group.
[0017]
(b) Selection of cancer-related genes
Normally, the growth of cells is controlled, but cancer cells are cells that are mutated to grow autonomously without being controlled by the body. For this reason, cancer cells often have abnormalities in genes that regulate cell proliferation and cell cycle. For example, cell growth factors include epidermal growth factor (EGF) and its receptor (EGF receptor, EGFR), platelet-derived growth factor (PDGF) and its receptor (PDGFR), insulin-like Growth factor (IGF), its insulin-like growth factor (IGFR), its receptor (IGFR), fibroblast growth factor (FGF), its receptor (FGFR), vascular endothelial growth factor (VEGF, vascular endotherial growth factor) ) And its receptor (VEGFR), hepatocyte growth factor (HGF) and its receptor (HGFR), neurotrophic factor (NT, neurotropin), transforming growth factor-β (TGFβ, transforming growth factor-β ) An abnormality is seen in the family. In addition, there may be an abnormality in a receptor gene corresponding to each growth factor, for example, an EGF receptor (EGFR) with respect to EGF. As factors that induce cell proliferation and differentiation, physiologically active peptides secreted by lymphocytes and blood cells are collectively referred to as cytokines. Major cytokines include granulocyte colony stimulating factor (G-CSF), macrohage-colony stimulating factor (M-CSF), and granulocyte macrophage colony stimulating factor (GM-CSF, granulocyte). -macrophage colony stimulating factor), erythropoietin, thrombopoietin, stem cell factor, interleukin 1,2,3,4,5,6,7,8,9, Abnormalities are observed in 10, 11, 12, tumor necrosis factor (TNF), interferon, and the like. Cyclin, cyclin-dependent kinase (CDK), CDK inhibitor (CKI, CDK inhibitor) such as cyclin A, cyclin B, cyclin D, cyclin E, CDK1 CDK2, CDK4, CDK6, p16INK, p15, p21, p27, retinoblastoma (RB) gene, etc. are abnormal.
[0018]
Even if an abnormality occurs in the mechanism of cell death, the cell becomes cancerous. Most cell death that occurs in living bodies is cell death called apoptosis. Apoptotic pathways include (a) calcium pathway, (b) death signaling pathway, (c) ceramide pathway, (d) mitochondrial pathway, and (e) DNA damage pathway. (a) The calcium pathway is activated by activation of intracellular phosphazilinositol triphosphate receptors by the action of glucocorticoids. (b) The death signal pathway is triggered by binding of TNF alpha and Fas ligand to a cell surface receptor (TNF receptor, Fas). Thereafter, protein molecules associated with the intracellular domain of the receptor (TRADD, FADD, RAIDD etc. for TNF receptor 1, FADD, RIP, RAIDD etc. for Fas) are activated. Activation of these proteins activates caspase genes (CASP) 8, 1, 3 and the like. In addition, it is known that caspase gene-related genes such as TRAMP and TRAIL suppress apoptosis. (c) The ceramide pathway is activated by various stresses such as radiation, ultraviolet rays, heat and hydrogen peroxide. More specifically, apoptosis is induced by an increase in ceramide and activation of SAPK (stress-acivated protein kinase) / JNK (Jun terminal-N kinase). (d) The mitochondrial pathway is triggered by various stimuli such as radiation, ultraviolet light, heat, hydrogen peroxide, etc., and activates Bax2 (Bcl-2 associated X protein), Bcl-2, Bcl-xL, and associated caspase genes Activation occurs, and apoptosis is induced. (e) In the DNA damage pathway, p53, p21, p51, p73, MDM2 gene and the like are involved. Abnormalities of the apoptosis-related gene have been observed in cancer cells.
[0019]
A gene known to promote canceration of cells due to its abnormality is called an oncogene. The oncogenes are classified according to the signal transduction flow: (a) growth factor group (eg sis gene), (b) receptor tyrosine kinase group (eg erbB, fms, ret gene), (c) non-receptor type Tyrosine kinase group (eg fes gene), (d) GTP / GDP binding protein (eg ras gene), (e) serine / threonine kinase group (eg src, mos, raf gene), (f) nuclear protein group (eg , Myc, myb, fos, jun, erbA gene), (g) a signal transduction adapter molecule (for example, crk gene), and a fusion gene (for example, Bcr-Abl gene). In addition, genes on major signal transduction systems are often classified as oncogenes. Ras-MAP kinase pathways (eg Shc, Grb2, Sos, MEK, Rho, Rac genes), phospholipase C gamma-protein kinase C pathway (eg PLCγ, PKC genes), phosphadylinositol 3-phosphorus Acid kinase-Akt pathways (eg, PI3K, Akt, Bad genes), JAK-STAT pathways (JAK, STAT), GAP pathways (eg, GAP, p180, p62) and the like are known as oncogenes. As for nuclear transcription factors, Myc, Ear, Evi, Gli, Fos, Fra, jun, Maf, Ets, Myb, IRF1, IRF2, Rel, Nf-kappa B, AML, etc. are known to have carcinogenic activity. Yes.
[0020]
Loss of the function of a gene that suppresses canceration of a cell or a tumor suppressor gene advances the canceration step of the cell. The major tumor suppressor genes discovered so far are RB, p53, WT1, NF1, APC, VHL, NF2, CDKN / p16, BRCA1, BRCA2, TGFβRII, DPC4, PTCH, PTEN, MEN1, E-cadherin gene, nm23 Etc.
[0021]
When DNA is damaged, the cells sometimes go to apoptosis or may be repaired. When an error occurs in the DNA repair process, a mutation occurs in the gene. Mutations cause cell canceration with a high probability. Examples of DNA repair genes in humans include MSH2, MSH3, MSH6, MLH1, PMS1, PMS2, PMS3, methylguanine-DNA methyltransferase (MGMT), and the like.
[0022]
The matrix metalloproteinase (MMP) group is generally produced in excess in cancer cells as metastasis-related genes related to cancer metastasis. MMP degrades collagen, which is the main component of the intercellular matrix. In addition, abnormalities of TMP (tissue inhibitor of metalloprotease), an inhibitor of MMP, are also found in cancer cells. In addition, cadherin that allows cells to adhere to each other, catenin (α catenin, β catenin, γ catenin), hyaluronic acid receptor family (CD44), integrin family, immunoglobulin family, selectin family, sialomucin, which are the intracellular regions of cadherin It has been pointed out that the family and other abnormalities are associated with cancer.
[0023]
As mentioned above, oncogenes, tumor-related genes, tumor suppressor genes, growth factors, transcription factors, cytokines, apoptosis genes ), Cell cycle regulators, DNA damage repair genes, transcription-related genes, etc. Analyzes can be performed with the minimum necessary for the introduction.
[0024]
In order to perform hybridization between each DNA fragment immobilized on the DNA array and the sample-derived DNA fragment with high precision (or highly stringent), the hybridization temperature (Th, hybridization temperature) ) And the melting temperature (Tm, melting temperature) of the immobilized DNA fragment is important. This is because the following relationship is known between sequence homology between DNAs and Tm. The relationship is that the Tm of both DNAs decreases by 1.4 ° C. due to a 1% base sequence difference (mismatch) between the two DNAs. That is, when the immobilized DNA fragment is Tm equal Th, the sample-derived DNA fragment having 100% homology (homology) with the base sequence of the immobilized DNA fragment hybridizes. When Th and Tm differ by 10 ℃
100% − (10 ° C./1.4° C.) = 92.9%
Thus, the sample-derived DNA fragment having about 93% to 100% homology with the base sequence of the immobilized DNA fragment hybridizes. Similarly, when Th and Tm differ by 20 ° C., 85.7% to 100% homology, and when Th and Tm differ by 30 ° C., 78.6% to 100% homology will hybridize. When performing separation with high accuracy, it is desirable that only at least 80% to 100% homology hybridize. For this reason, it is necessary for the separation with high accuracy that the difference between the melting temperature of the immobilized DNA fragment and the hybridization temperature does not exceed 30 ° C.
[0025]
Another important condition for highly accurate hybridization between each DNA fragment immobilized on the DNA array and the sample-derived DNA fragment is to eliminate cross-hybridization. Cross-hybridization occurs because of the high homology between DNA sequences. Therefore, in order to prevent cross-hybridization, it is desired that the homology between the immobilized DNA fragment and a DNA fragment that does not originally hybridize with the immobilized DNA fragment among the DNA fragments derived from the sample is sufficiently low. This realization becomes more difficult as the number of DNA fragments to be immobilized (number of genes) increases. Table 1 shows the results of calculating how much cross-hybridization increases when the number of genes increases.
[0026]
[Table 1]

[0027]
In the calculation of Table 1, first, 1000 kinds of gene sequences were cited from GENBANK. Among them, 20, 200, 400, and 1000 kinds of gene sequence sets were randomly generated. A 300 base pair DNA sequence having a Tm of 72 ° C. or higher and 82 ° C. or lower was randomly extracted from each gene sequence set for each gene. Their randomly sampled DNA sequences were compared to each other across genes and homology was calculated according to the blast algorithm. The blast algorithm was the same as that described in Altschul, SF et al., Basic local alignment search tool, J. Mol. Biol. Vol. 215, p. 403-410, 1990. When the statistical significance level (p value) calculated by the blast algorithm was 10% or less, the homology of the two kinds of DNA sequences compared was sufficiently high (risk rate 10% or less), and cross-hybridization occurred. . For example, in the case of 200 genes shown in Table 1, when DNA fragments randomly extracted from 200 genes are compared with each other, 7 genes, that is, 14 genes, have sufficiently high homology (p value). The result was <10%). Table 1 shows the result of the occurrence of cross-hybridization in 7 sets of 200 genes, with the probability of occurrence of cross-hybridization being 7% (= 14 genes / 200 genes). Similarly, the probability of cross-hybridization was zero for 20 genes, 28% for 400 genes (= 112 genes / 400 genes), and 68% for 1000 genes (= 680 genes / 1000 genes). Cross-hybridization was frequently observed between genes having similar functions and between isoforms. As a result of the calculation shown in Table 1, it was confirmed that the probability of cross-hybridization increases as the number of genes increases. It can be seen that cross-hybridization is unavoidable due to an increase in the types of genes to be immobilized. When the number of genes increases from several thousand to several tens of thousands, the action mechanism of an antineoplastic drug or the like cannot be analyzed with high accuracy unless sequencing of the immobilized DNA fragment is performed with great care. In order to eliminate cross-hybridization and perform hybridization at high stringency with the melting temperature kept within a certain range, it is necessary to immobilize only the minimum number of genes on the array according to the purpose of analysis. desirable. For example, for the purpose of elucidating the mechanism of action of antineoplastic agents, etc., DNA fragments complementary to DNA molecules that describe genes responsible for human cancer development, invasion, and metastasis, and the key to drug metabolism An array in which only a DNA fragment that complementarily binds to a DNA molecule that describes the gene is immobilized.
[0028]
Here, the relationship between the p-value between the immobilized fragments and cross-hybridization will be described with reference to FIG. As the sequence homology between immobilized DNA fragments increases (p value decreases), it can be predicted that cross-hybridization will become more prominent. In order to know the relationship between the p value and cross hybridization, the following experiment was conducted. Among the genes shown in Table 2, a DNA fragment corresponding to IFI56 was immobilized on the glass surface. In addition, a DNA fragment corresponding to PRSM was immobilized on the glass surface. That is, two kinds of different DNA fragments were immobilized on the glass surface. On the other hand, the fluorescently labeled IFI56 gene fragment was sprinkled on the glass surface and hybridized. If the IFI56 gene fragment only hybridizes with the immobilized DNA fragment corresponding to IFI56 and does not hybridize to the DNA fragment corresponding to PRSM, cross-hybridization is zero. That is, the ratio of (fluorescence intensity derived from hybridization with a DNA fragment corresponding to PRSM) and (fluorescence intensity derived from hybridization with a DNA fragment corresponding to IFI56) is zero. Hereinafter, this ratio is referred to as a signal intensity ratio. When the IFI56 gene fragment is hybridized with both the DNA fragment corresponding to IFI56 and the DNA fragment corresponding to PRSM, the signal intensity ratio is 1. That is, the signal intensity ratio approaches 1 as the cross hybridization occurs more frequently. FIG. 4 shows the relationship between the signal intensity ratio and the p value. It can be seen that the smaller the p value (that is, the greater the homology between the two types of immobilized DNA fragments), the higher the signal intensity ratio (the more cross hybridization). However, when the p value exceeded 10%, the signal intensity ratio decreased to about 0.01. From this result, it can be seen that cross-hybridization can be sufficiently prevented if the p-value between immobilized DNA fragments exceeds 10%.
[0029]
On the other hand, hybridization between DNA molecules is a phenomenon that occurs between single-stranded DNAs. Single-stranded DNA is usually bent in solution to form several complementary base pairs. This structure is called a hairpin structure. Among these hairpin structures, it is known that there are hairpin structures with unusually high thermal stability (Hirao and Miura, special structures of single-stranded DNA, mini-hairpin structures with unusually high thermal stability, protein nucleic acid enzymes, vol.40, p.1583-1592,1995). This mini hairpin structure, or a short sequence with a specific higher order structure, for example,
GCGAAAGC (SEQ ID NO: 1)
GCGAAGC (SEQ ID NO: 2)
It is. The former (SEQ ID NO: 1) has a Tm of 76 ° C. This temperature of 76 ° C. is significantly higher than the general hybridization temperature (40 ° C. to 62 ° C.). For this reason, when SEQ ID NO: 1 is contained in the immobilized DNA fragment, there is a very high probability that a complementary bond will not be formed in the sequence portion or in a wider range during hybridization. In a sense, the state where no complementary bond is formed is the same as the mismatch state. As described above, if there is a mismatch between two DNAs, the Tm of both DNAs is lowered, and the hybridization accuracy is lowered. For this reason, for example, it is better not to include a mini hairpin structure such as SEQ ID NO: 1 or SEQ ID NO: 2 in the immobilized DNA fragment as much as possible. Other sequences that can give rise to this mini hairpin structure include short repeat sequences such as
GGGCGGCGGG (SEQ ID NO: 3)
TTTCATTATTGAAA (SEQ ID NO: 4)
Etc. are considered.
[0030]
It is known that human DNA has a sequence containing many repetitive sequences called Alu sequence, which is peculiar to humans. Many Alu sequences are contained in human DNA. Alu sequences are also present in messenger RNA, particularly in the 5 ′ untranslated region or the 3 ′ untranslated region. If DNA molecules are mixed in the messenger RNA sample to be measured, during the step of fluorescent labeling using the messenger RNA as a template, the contaminated DNA may be fluorescently labeled due to the presence of the Alu sequence portion. . And it is also conceivable that the labeled contaminating DNA containing the Alu sequence portion and the immobilized DNA fragment are complementarily bound. Originally, only messenger RNA should be measured, but DNA may also be measured. In order to avoid erroneous analysis due to this contaminated DNA, it is desirable that the immobilized DNA fragment does not contain a portion having a significantly high homology with the Alu sequence. In addition, since the Alu sequence contains a large number of repetitive structures, there is a possibility that a higher-order structure may be formed in the solution, and it is fully conceivable that a mismatch is caused in the same manner as the above-described mini hairpin structure. In this sense, it is desirable that the immobilized DNA fragment does not contain a portion having a significantly high homology with the Alu sequence. For example, if the homology between the Alu sequence and the immobilized DNA fragment exceeds 10% at the statistical significance level (p value) calculated by the blast algorithm, the sequence homology is considered to be significantly reduced.
[0031]
Here, the relationship between the sequence homology between the Alu sequence and the immobilized DNA fragment, noise, and S / N ratio will be described with reference to FIG. In order to observe the influence of the sequence homology (p value) between the Alu sequence and the immobilized DNA fragment on the measurement results, the following experiment was performed. First, a DNA fragment corresponding to IFI56 was immobilized on the glass surface. In contrast, 10 kinds of genes shown in Table 2 were fluorescently labeled and sprinkled on the glass surface for hybridization. Most preferably, only the fluorescence signal corresponding to IFI56 is observed and no other signals are observed. Now, the fluorescence intensity in a sequence having a p value of 20% or more with the Alu sequence (the homology with the Alu sequence is sufficiently low) is normalized to 1. Fluorescence intensities when the p value with the Alu sequence is 10%, 5%, 1%, or 0.01% or less (value obtained by normalizing the value with the p value of 20% with the Alu sequence to 1) Is shown in FIG. Then, it was found that the fluorescence intensity increased as the p value decreased. This increase is noise because it occurs because the DNA sequence that should not hybridize originally hybridizes. Noise (ie, each fluorescence intensity minus 1) is shown in FIG. The noise is 0.1 at a p value of 10%, but increases when the p value is 10% or less. Further, fluorescence intensity ÷ noise is defined as SN ratio (S / N). When the respective SN ratios were calculated, the SN ratio was decreased (measurement sensitivity was deteriorated) as the p value reached 10%. FIG. 5 shows that the measurement sensitivity of the immobilized DNA fragment having a p value of 10% or less with the Alu sequence is significantly deteriorated due to the presence of the repetitive sequence or the Alu sequence derived from the contaminating DNA contained in the sample. It was. By setting the p value of the Alu sequence and the immobilized DNA fragment to 10% or more, deterioration in measurement sensitivity can be prevented.
[0032]
In order to examine the influence of the presence of (1) cross-hybridization, (2) short sequences and short repeat sequences having a specific higher order structure, (3) similar sequences to Alu, etc. on actual hybridization, Experiments were conducted in which the complementary DNA (cDNA) sequence of messenger RNA describing 10 genes was immobilized on the glass surface. This immobilized DNA fragment (exactly a cDNA fragment, but hereinafter referred to as a DNA fragment) was synthesized by randomly selecting regions of 300, 600, 900, 1200, 1500, 2100 bases from the GENBANK sequence. is there. The measurement sample was cDNA synthesized by reverse transcription reaction using oligo dT primer, using a messenger RNA group extracted from human pancreatic cancer cells (cell name CFPAC1) as a template. The cDNA synthesized by reverse transcription was fluorescently labeled (fluorescent reagent name Cy3-dCTP) during the reverse transcription reaction. Table 2 shows the abbreviations of the 10 genes used in the experiments, GENBANK accession numbers, and gene names.
[0033]
[Table 2]

[0034]
FIG. 1 shows the influence of the base length of the immobilized DNA fragment obtained on the experiment on the hybridization. Each of the 10 line graphs in the figure is the result of an experiment using 10 genes shown in Table 2. Gene abbreviations (shown in Table 2) are indicated in the line graph. The horizontal axis of the line graph indicates the base length of the immobilized DNA fragment. The vertical axis of the line graph indicates the fluorescence intensity. However, the fluorescence intensity on the vertical axis was normalized assuming that the fluorescence intensity at 300 base length is 1. The fluorescence intensity was defined as the net fluorescence intensity obtained by subtracting the fluorescence intensity of the background (background). The hybridization temperature was 62 ° C. Higher fluorescence intensity corresponds to higher signal intensity ratio (SN ratio), that is, higher detection sensitivity. The values in the line graph are the average values of three experiments, and the error bars are the standard deviations. Moreover, the horizontal line below each line graph indicates the relative positional relationship of DNA fragments of each length. The 5 'end is shown on the left and the 3' end is shown on the right. In order from the top, the base length is long to short. The portions where the horizontal lines overlap are sequence portions common to the respective DNA fragments.
[0035]
From the results shown in FIG. 1, when the two IFI56 and GSS genes were targeted, the longer the immobilized DNA fragment length, the slower the fluorescence intensity, that is, the detection sensitivity tended to increase. In addition, for the five genes PRSM, UBE, AKT, ADPRT, and SMRT, when a measurement error (error bar) was taken into consideration, almost constant detection sensitivity was obtained regardless of the length of the immobilized DNA fragment. The phenomenon that the detection sensitivity increases as the DNA fragment length increases can be explained by the fact that the stability of the binding force between DNA molecules increases as the DNA molecule length increases. The stability of the binding force between DNA molecules can be estimated by free energy (ΔG) in thermodynamics. When ΔG is calculated, a smaller value is obtained as the base length is longer. The smaller the ΔG, the greater the thermodynamic stability. The value of ΔG between DNA molecules was calculated by the nearest neighbor method (J. SantaLucia, Jr. A unified view of polymer, dumbbell, and oligonucleotide DNA nearest-neighbor thermodynamics, Porc. Natl. Acad. Sci. USA, vol.95, p.1460-1465, 1998).
[0036]
However, in the three genes CASP, PLG, and GEM, the length of the DNA molecule and the fluorescence intensity, that is, the detection sensitivity are not statistically significantly correlated. The fluorescence intensity is clearly low at 600 bases in CASP, 600 bases in CASP, and 600 and 1200 bases in GEM. The results are negative bases such as (minus factor 1) cross-hybridization, (minus factor 2) short sequences and short repeats with specific higher order structures, (minus factor 3) the presence of similar sequences to Alu, etc. The sequence portion was inferred to be included in the corresponding portion, and analysis was performed. As a result, the negative factor 2 was present in the region of 600 bases of CASP, 600 bases of GEM and 1200 bases and not overlapping with other sequences. Specifically, at 600 bases of CASP, GCGAAAGC (SEQ ID NO: 1) or any of its complementary, inverted, and complementary inverted sequences exists, and at 600 bases and 1200 bases of GEM, GCGAAGC (SEQ ID NO: 2) or complementary sequences, inverted Either sequence or complementary inversion sequence was present. The complementary sequence of GCGAAAGC is CGCTTTCG, the inverted sequence is CGAAAGCG, and the complementary inverted sequence is GCTTTCGC. However, PLG did not fall under any of (1) to (3). The PLG results suggested that there may be negative factors that are not yet known. Again from the results of FIG. 1, it was confirmed that the presence of short sequences or short repeating sequences having a unique higher order structure was not preferable. It has also been confirmed that reducing the free energy (ΔG) in the thermodynamics of the immobilized DNA fragment does not at least become negative, but rather acts positively. For example, when the base length is 300 bases, ΔG is preferably about minus 140 to minus 250. However, since the value of ΔG becomes smaller as the base length becomes longer and takes a different value depending on the base sequence, it is difficult to provide a clear threshold value.
[0037]
Based on the results shown in FIG. 1 and the like, DNA fragments immobilized on one array must have a DNA fragment Tm not exceeding 30 ° C. from the hybridization temperature, in other words, not exceeding a certain range. It is important to reduce the free energy (ΔG) as much as possible while satisfying the Tm condition. In addition, it is preferable to exclude sequences causing cross-hybridization, exclude short sequences and short repeat sequences (hereinafter referred to as harmful short sequences) having a specific higher order structure, and exclude similar sequences to Alu. In order to satisfy all of these conditions, for example, the sequence of the immobilized DNA fragment may be determined by the calculation algorithm shown in FIG. FIG. 2 will be described in the order of the algorithm. First, read the DNA or messenger RNA sequence file described in FASTA format, etc. (1), then input the experimental conditions such as salt concentration and hybridization temperature (2), and the length of the immobilized DNA fragment Through the step (3) of inputting a range, a plurality of DNA sequences that are candidates for immobilized DNA fragments are selected and made into a list. Subsequently, Tm is calculated for each DNA sequence candidate in the list, and the Tm is compared with a hybridization temperature (Th) in a certain range, for example, Th-α to Th-β (where α , Β value may be either positive or negative) (4), removing candidate DNA sequences from the candidate list (4), candidate sequences for DNA sequences with short sequences or short repeats having a specific higher order structure (5), a step (6) for excluding DNA sequences having high homology with Alu sequences from the candidate list, and a step (7) for excluding DNA sequences having high homology with other gene sequences from the candidate list. . Any sequence that satisfies all the conditions in steps 4 to 7 is defined as an immobilized DNA fragment sequence. When a plurality of sequences satisfy Steps 4 to 7, for example, a sequence having the smallest free energy (ΔG) may be used as an immobilized DNA fragment sequence.
[0038]
In step 5, for example, a character string pattern matching algorithm is used. As a pattern matching target, it is desirable to target a complementary sequence of a harmful short sequence, an inverted sequence (CGAAAGCG in the case of GCGAAAGC), and a complementary inverted sequence. In order to improve the calculation speed, the character strings may be once converted into numbers and compared by bit shift calculation or the like.
[0039]
In step 6 and step 7, a heuristic algorithm such as a blast algorithm or a FASTA algorithm may be used, or a local alignment such as a Smith-Waterman method. An algorithm or an algorithm using a Hidden Markov model may be used. In order to improve the calculation speed, the character string may be once converted into a number and a bit shift calculation or the like may be performed.
[0040]
In step 7, not only the homology between gene sequences immobilized on one array but also the homology between a DNA sequence and a human gene sequence such as GENBANK may be calculated. The sample to be measured contains a large number of other genes in addition to the gene originally hybridized with the immobilized DNA fragment. When these different genes are hybridized, it is inherently impossible to measure only the genes to which individual immobilized DNA fragments should be hybridized. Therefore, comparing the DNA fragment candidate sequence to be immobilized on the DNA array with the DNA sequence of the gene group that may be contained in the sample to be measured, the DNA sequence with significantly high homology is immobilized. It is desirable not to select as a DNA fragment. Gene sequence databases such as GENBANK store all the sequences of humans, rodents, yeasts, Escherichia coli, etc. that have been known so far. If the DNA sequence homology between the immobilized DNA fragment and a gene sequence such as a human such as GENBANK exceeds 10% at the statistical significance level (p value) calculated by the Blast algorithm, High-precision measurement can be performed.
[0041]
For calculating the Tm of the DNA sequence, the nearest neighbor method (J. SantaLucia, Jr. A unified view of polymer, dumbbell, and oligonucleotide DNA nearest-neighbor thermodynamics, Porc. Natl. Acad. Sci. USA, vol. .95, p.1460-1465, 1998) and empirical methods (Wetmure, JG, DNA probes: Applications of the principles of nucleic acid hybridization, Criti. Rev in Biochem. And Mol. Biol., Vol. 26 , p.227-259, 1991). There are reports that the former closest base method is well matched for DNA sequences of 50 base pairs or less, and the latter method based on the empirical formula is well matched for DNA sequences of 50 base pairs or more. However, according to the paper by J. SantaLucia, Jr., the closest base method has also been improved, suggesting that the calculation can be performed regardless of the sequence length. Therefore, for the Tm calculation of the DNA sequence, either the nearest base method or the method based on the empirical formula is used according to the base length, or the average value of the values obtained by the both methods is calculated, for example. Etc. may be used.
[0042]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments when the present invention is used for elucidating the action mechanism of a drug such as an antineoplastic agent are described below. Tables 3 to 18 show a list of genes essential for occurrence, invasion and metastasis of human cancer immobilized on the glass surface, and genes essential for drug metabolism.
[0043]
[Table 3]

[0044]
[Table 4]

[0045]
[Table 5]

[0046]
[Table 6]

[0047]
[Table 7]

[0048]
[Table 8]

[0049]
[Table 9]

[0050]
[Table 10]

[0051]
[Table 11]

[0052]
[Table 12]

[0053]
[Table 13]

[0054]
[Table 14]

[0055]
[Table 15]

[0056]
[Table 16]

[0057]
[Table 17]

[0058]
[Table 18]

[0059]
Next, Table 19 to Table 38 show immobilized DNA fragment sequence lists corresponding to a part of the genes shown in Tables 3 to 18 calculated using the algorithm of FIG.
[0060]
[Table 19]

[0061]
[Table 20]

[0062]
[Table 21]

[0063]
[Table 22]

[0064]
[Table 23]

[0065]
[Table 24]

[0066]
[Table 25]

[0067]
[Table 26]

[0068]
[Table 27]

[0069]
[Table 28]

[0070]
[Table 29]

[0071]
[Table 30]

[0072]
[Table 31]

[0073]
[Table 32]

[0074]
[Table 33]

[0075]
[Table 34]

[0076]
[Table 35]

[0077]
[Table 36]

[0078]
[Table 37]

[0079]
[Table 38]

[0080]
Finally, using a DNA array in which the genes shown in Table 3 to Table 18 are immobilized on the glass surface, cultured human cells (cells) against an antineoplastic agent (drug name: 5-aza-2′-deoxy-cytidine) The measurement result of the gene expression distribution of name: HT29) is shown in FIG. In the following, description will be given in order.
[0081]
(1) Gene list on DNA array
Table 3 shows a list of genes that are essential for the occurrence, invasion, and metastasis of human cancer selected according to the criteria described in claim 2, and genes that are important for drug metabolism selected according to the criteria described in claim 3. Show. The GENBANK accession number and gene name corresponding to each gene sequence are described.
[0082]
(2) Partial sequence list in gene list
Table 4 shows a list of immobilized DNA fragment sequences corresponding to a part of the genes shown in Table 3. Each DNA fragment sequence, DNA fragment sequence base length (length), DNA fragment sequence free energy (dG), melting temperature of DNA fragment sequence calculated by empirical method (Tm1), calculated by nearest base method The melting temperature (Tm2) of the obtained DNA fragment sequence is also shown. Here, the melting temperature (Tm) was an average value of Tm1 and Tm2. The DNA fragment sequences in Table 4 were designed so that the Tm (= 0.5 * (Tm1 + Tm2)) was 72 ° C. to 82 ° C.
[0083]
(3) An example of experimental results
FIG. 3 shows an increased messenger when human colon cancer cultured cells (name: HT29 cells) were treated with 5-aza-2′-deoxy-cytidine (hereinafter, 5-Aza-CdR) compared with the case where they were not treated. The increase rate of RNA amount is shown for each gene. 5-Aza-CdR treatment means exposing cells to 500 nanomolar 5-Aza-CdR for 2, 12, 24 hours, respectively. Messenger RNA was immediately extracted from each time exposed HT29 cells. Similarly, HT29 cells without drug treatment were also cultured in the same incubator, and messenger RNA was extracted immediately after each time. With respect to the messenger RNA of the drug-treated cell by reverse transcription reaction using an oligo dT primer, a fluorescently labeled cDNA was synthesized using Cy5-dCTP. As for messenger RNA of cells not treated with a drug, fluorescently labeled cDNA was synthesized using Cy3-dCTP. Drug-treated cell-derived cDNA (Cy5-labeled) and non-drug-treated cell-derived cDNA (Cy3-labeled) were mixed and hybridized with the same DNA array. The hybridization temperature was 62 ° C. and the hybridization time was 12 hours. After hybridization, each fluorescence intensity of Cy5 and Cy3 was digitized by a fluorescence scanner.
[0084]
The ratio of fluorescence intensity derived from drug-treated cells (Cy5 fluorescence intensity) and cell-derived fluorescence intensity not treated with drugs (Cy3 fluorescence intensity), ie, Cy5 fluorescence intensity / Cy3 fluorescence intensity was calculated, and the value on the vertical axis in FIG. , And plotted as the rate of increase in messenger RNA amount. The horizontal axis in FIG. 3 is the serial number of the gene on the array. As a result of this experiment, a gene having a Cy5 fluorescence intensity / Cy3 fluorescence intensity of 3 times or more, for example, after 2 hours, SIAT4A, PRSM1, ribosomal protein S5, plasminogen, ETR103, OCTS3, integer4, GEM, RGS14, HsCds18 IFN-induced protein 56, INFα-inducible protein 27, INF-induced 17 kDa protein, INF-induced protein IFI-6-16, INF-inducible protein 10 and the like. Throughout 2, 12, and 24 hours, the Cy5 fluorescence intensity / Cy3 fluorescence intensity was always more than 3 times that of genes related to interferon (IFN) α, IFN-induced protein 56, INFα-inducible protein 27, INF-induced 17 kDa protein, INF-induced protein IFI-6-16, INF-inducible protein 10 and so on. This result was reported by Karpf et al. (Karpf, AR et al., Inhibition of DNA methyltransferase stimulates the expression of signal transducer and activator of transcription 1, 2, ad 3 genes in colon tumor cells, Proc. Natl. Acad. Sci. USA, vol.96, p.14007-14012, 1999). Furthermore, in response to the report of Karpf et al., The present invention was able to observe the time dependence of the drug response with good reproducibility.
[0085]
By carrying out the present invention, the time dependence of the state of drug response of human cells to 5-Aza-CdR, which is a kind of antineoplastic agent, was observed with high reproducibility and high accuracy. Similar results can be obtained for drugs other than 5-Aza-CdR.
[0086]
【The invention's effect】
Provided are a DNA array for elucidating the mechanism of action of a malignant tumor drug or the like with high accuracy and a method for producing the same.
[Brief description of the drawings]
FIG. 1 shows the influence of the base length of an immobilized DNA fragment on hybridization.
FIG. 2 shows a DNA fragment sequencing calculation algorithm.
FIG. 3 shows the rate of increase in the amount of messenger RNA increased when HT29 cells were treated with 5-aza-2′-deoxy-cytidine compared to when they were not treated.
FIG. 4 is a diagram showing the relationship between sequence homology between immobilized DNA fragments and cross-hybridization.
FIG. 5 is a diagram showing the relationship between the sequence homology between the Alu sequence and the immobilized DNA fragment and the measurement results.
[Explanation of symbols]
1. Loading process of gene sequence file 2. Input process of experimental conditions such as salt concentration and hybridization 3. Input process of length range of immobilized DNA fragment 4. Calculate melting temperature of immobilized DNA fragment 5. a step of excluding DNA fragments whose melting temperature is outside a certain range from the candidate list, 5. a step of excluding DNA DNA fragments having a short sequence having a specific higher order structure or a short repeat sequence from the candidate list 6. A step of excluding DNA fragments having high homology with Alu sequences from the candidate list, 7. A step of excluding DNA fragments having high homology with other gene sequences from the candidate list.

Claims (3)

DNA fragments that complementarily bind to DNA molecules or cDNA molecules that describe genes associated with human cancer development, invasion, or metastasis, and DNA that complementarily bind to DNA molecules or cDNA molecules that describe drug metabolism- related genes Both fragments are immobilized on a solid surface,
The sequence homology between the DNA fragments exceeds 10% as a statistical significance level (p value) calculated by a homology search (homology search) algorithm, and any of the DNA fragments is a sequence with a human Alu sequence. The homology statistical significance level (p value) exceeds 10%, the difference between the melting temperature of the DNA fragment and the hybridization temperature does not exceed 30 ° C., and the DNA structure has a unique higher order structure A DNA array obtained by immobilizing the DNA fragment group that does not contain short DNA sequences and short repeat sequences. Genes associated with the development, invasion, and metastasis of human cancer according to claim 1 are cancer genes (oncogenes, tumor-related genes), tumor suppressor genes, growth factors, transcription factors 2. A transcription factor, a cytokine, an apoptosis gene, a cell cycle regulator, a DNA repair gene, and a transcription-related gene. DNA array. The drug metabolism- related gene according to claim 1 is a drug-metabolizing enzyme-related gene, in particular cytochrome 450 gene (CYP) 1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP3A4, CYP3E4, CYP3E4 , Alcohol dehydrogenase, aldehyde dehydrogenase, dihydroxypyrimidine dehydrogenase, NADPH-cytochrome P450 reductase, DT-diaphorase, esterase, epoxide hydrase and other first phase drug metabolism related genes, catechol O-methyltransferase, glutathione S-transferase , Histamine methyltransferase, N-acetyltransferase, sulfotransferase 2. The DNA array according to claim 1, wherein the DNA array is one or more of phase II drug metabolism-related genes such as lysine, thiopurine methyltransferase, and uridine 5′-triphosphate glucuronyltransferase.

Images