第十章+分子對(duì)接

1、1第十章第十章 分子對(duì)接分子對(duì)接2What is docking ? Docking : finding the binding orientation of two molecules with known structures Goals of Docking Fitting a small molecular (drug molecule) into a protein Docking two proteins together & calculate Binding Free Energy3To perform a docking Structures of protein/small

2、 moleculeMolecule Databases MethodProtein/DNA PDB xray/NMRSmall Molecules CSD xray/neutron diffraction Software Program to do docking: Dock, AutoDock, Gold, FlexX, etc. Powerful Computers CPU, diskspaceCambridge Structural Database 4Classification According to the molecules involved: Protein-Ligand

3、docking Protein-Protein docking Docking calculations are used for many purposes Search for drug design lead compounds Study conformational possibilities of compounds known to interact Find binding determinants 5Protein-Protein DockingBoth molecules are relatively rigidLarge search spaceUses steric c

4、onstraints and energetics of potential binding conformation to reduce search spaceRequires good representation of docking surface6Protein-Protein Docking7Protein-Ligand Docking Identify interacting sites between a protein with known (rigid) structure and a flexible ligand (usually small molecule) EX

5、TREMELY large search space Many possible binding sites for potential interactions8Protein-Ligand DockingGbinding = Gcomplex Gprotein GligandTaskComparesConstant termsStructurePose1 vs. Pose2Gprotein, GligandEnrichmentLigand1 vs. Ligand2GproteinSpecificityDifferent ligands in different proteinsNone9T

6、he Docking Process Several algorithms can be used to do the docking Molecular Dynamics Monte Carlo Simulated Annealing Genetic Algorithms Fragment-based Methods Distance Geometry Methods10From Structure to LigandEnergy minimize structures of target and ligand Molecular dynamics simulation time avera

7、ged structuresCreate pseudo-intercalation binding site11 Dock ligand into pseudo-intercalation siteManual, automatic, and flexible ligand docking Energy minimize to determine G complex , G protein Determine Gligand = interaction energy of ligand with surroundings when explicitly solvatedDocking12 Vi

8、sual inspection of compound 1 docking Compounds 3 and 4 Similarly docked 13Docking results Attempts at docking compound 2 were unsuccessfulDistorted the quadraplex structure due to bulky side chains14Models of Docking Rigid Docking Assume the configurations of protein and ligand cannot change rigid

9、Most commonly used model Partial Flexibility Protein and ligand assumed to be flexible only at the binding site Can be modeled by adding a few DOF in the protein binding site to the combined protein/ligand C-space Full Flexibility All DOF of the ligand and the protein are taken into accountWe chose

10、a model and then simulate how a ligand behaves around a receptor and see if it bindBUT FIRST we need some kind of a guiding/scoring function.15Scoring Function We would like to have a function which: - given a configuration of protein and the ligand- returns a number representing goodness or energy

11、of the configuration. Desired properties: (Ideally) Lowest value when the ligand is naturally docked Higher value everywhere else Should be able to distinguish between correctly and incorrectly docked structures. Should be fast! to compute.16Scoring Function: Examples Quantum mechanical models Takes

12、 5 days per configuration on a super duper computer Van der Waals + Electrostatic Potential energy (very popular) Hydrogen bonding Surface Area Combination of above17Solvation Effect18 式中式中A和和B分別表示受體分子和底物分子，分別表示受體分子和底物分子， 為氣態(tài)下分子對(duì)接過程的自由能變?yōu)闅鈶B(tài)下分子對(duì)接過程的自由能變化，約為對(duì)接過程的焓變化，約為對(duì)接過程的焓變; 分別為受體分子，底物分子以及復(fù)分別為受體分子，底

13、物分子以及復(fù)合物分子的溶劑化自由能；合物分子的溶劑化自由能； 則表示對(duì)接過程中的熵變。則表示對(duì)接過程中的熵變。 實(shí)際上不能采用嚴(yán)格的自由能計(jì)算方法來準(zhǔn)確評(píng)價(jià)靶酶分子和每個(gè)底物分子之間的結(jié)合自由能。 只能采用較為簡單的自由能評(píng)價(jià)方法，即：只考慮非鍵相互作用能和用基于分子表面的溶劑化來考慮分子對(duì)接過程中溶劑化能的變化。,ABABsolvsolvsolvGGGABABbindgassolvsolvsolvABABgasPBSAPBSAPBSAGGGGGHT SGGG gasGT S19Types of Scoring Functions Force field based: non-bonded i

14、nteraction terms as the score, sometimes in combination with solvation terms Empirical: multivariate regression methods to fit coefficients of physically motivated structural functions by using a training set of ligand-receptor complexes with measured binding affinity Knowledge-based: statistical at

15、om pair potentials derived from structural databases as the score Consensus scoring functions approach Other: scores and/or filters based on chemical properties, pharmacophore, contact, shape complementary20Force Field Based Scoring Functions Advantages FF terms are well studied and have some physic

16、al basis Transferable, and fast when used on a pre-computed grid Disadvantages Only parts of the relevant energies, i.e., potential energies & sometimes enhanced by solvation or entropy terms Electrostatics often overestimated, leading to systematic problems in ranking complexes Solvation and Entrop

17、y Terms: Solvation terms decomposed into nonpolar and electrostatic contributions (e.g., DOCK):ligirecjijjibijijaijijDrqqrBrAE11332e.g. AMBER FF in DOCKnpsolvelecsolvnonbondbindEEEE,21Empirical Scoring Functions Goals: Reproduce the experimental values of binding energies and with its global minimum

18、 directed to the X-ray crystal structure Advantages: Fast & direct estimation of binding affinity DisadvantagesOnly a few complexes with both accurate structures & binding energies known Discrepancy in the binding affinities measured from different labsHeavy dependence on the placement of hydrogen a

19、tomsHeavy dependence of transferability on the training setNo effective penalty term for bad structures22Knowledge-based Potentials of Mean Force Scoring Functions (PMF) Assumptions An observed crystallographic complex represents the optimum placement of the ligand atoms relative to the receptor ato

20、ms The Boltzmann hypothesis converts the frequencies of finding atom A of the ligand at a distance r from atom B of the receptor into an effective interaction energy between A and B as a function of r Advantages Similar to empirical, but more general (much more distance data than binding energy data

21、) Disadvantages The Boltzmann hypothesis originates from the statistics of a spatially uniform liquid, while receptor-ligand complex is a two-component non-uniform medium PMF are typically pair-wise, while the probability to find atoms A and B at a distance r is non-pairwise and depends also on surr

22、ounding atoms23Consensus Scoring and Others Too many scoring functions, none prevails in terms of predictivity Combined approach: one scoring function to sample configuration space, the other(s) to optimize and/or score: 2 docking methods & 13 scoring functions to significantly reduces false positiv

23、e rate (Charifson et al. 1999) Postprocessing of docking results with a filter function followed by re-scoring (Stahl & Bohm 1998) ADAM, FlexX, Hammerhead SYBYL Cscore (Tripos) : FlexX, PMF, DOCK energy, GOLD score C2 (Accelrys) : LigScore2, PLP, PMF, Ludi, Jain FRED (OpenEye) : ChemScore, PB-SA, Ch

24、emGauss, PLP, ScreenScore DOCK: AMBER FF, PMF, contact scores, ChemScore24DOCK (Kuntz, UCSF)Receptor Structure X-ray crystal NMR homologyBinding SiteMolecular Surface of Binding SiteSpheres describing the shape of binding site andfavorable locations of potential ligand atomsMatching heavy atoms of l

25、igands to centers ofspheres to generate thousandsof binding orientationsScoring Orientations1. Energy scoring (vdw and electrostatic)2. Contact scoring (shape complementarity)3. Chemical scoring4. Solvation termsVirtual Screening for MTS/HTS and Library Design: ligands in the order of their best sco

26、resBinding Mode Analysis for Lead Optimization: binding orientations and scores for each ligandsLigands 3D structure atomic charges potentials labelingFilters25Other Docking SoftwaresDOCK: (Kuntz et al. 1982)DOCK 4.0 (Ewing & Kuntz 1997)AutoDOCK (Goodsell & Olson 1990)AutoDOCK 3.0 (Morris et al. 199

27、8) GOLD (Jones et al. 1997)FlexX: (Rarey et al. 1996) GLIDE: (Friesner et al. 2004)ADAM (Mizutani et al. 1994)CDOCKER (Wu et al. 2003)CombiDOCK (Sun et al. 1998)DIVALI (Clark & Ajay 1995)DockVision (Hart & Read 1992)FLOG (Miller et al. 1994) GEMDOCK (Yang & Chen 2004)Hammerhead (Welch et al. 1996)LI

28、BDOCK (Diller & Merz 2001)MCDOCK (Liu & Wang 1999)PRO_LEADS (Baxter et al. 1998)Hex (Ritchie & Kemp 2000)SDOCKER (Wu et al. 2004)QXP (McMartin & Bohacek 1997)Validate (Head et al. 1996).26Homology Modelingprotein of known 3D structure modeled 3D structure of target protein Build the lock, then find

29、the keyIf you know the 3D structure of the target receptorDONTBasic Modeling Strategies27 If you know the 3D structure of the target receptor Ligand-Based Molecular DesignInfer the lock by inspecting the keysInfer Binding PocketDONTBasic Modeling Strategies28If you know the 3D structure of the targe

30、t receptorReceptor-Based Design Build or Find the key that fits the lockBasic Modeling StrategiesDO29Virtual Screeningcorporate databaseknown activesstructures found30Lead Finding: de novo DesignLinkingProtein Define active site Generate ligand molecules Atom connection Fragment connection Estimatio

31、n31Protein-based Design of Combinatorial Libraries 一一. DOCK是目前應(yīng)用最為廣泛的分子對(duì)接程序之一。是目前應(yīng)用最為廣泛的分子對(duì)接程序之一。 特點(diǎn)：特點(diǎn)： 能自動(dòng)模擬配體分子在受體活性位點(diǎn)的作用情況，并把理能自動(dòng)模擬配體分子在受體活性位點(diǎn)的作用情況，并把理論預(yù)測最佳的方式記錄下來。論預(yù)測最佳的方式記錄下來。 能夠?qū)ε潴w的三維結(jié)構(gòu)數(shù)據(jù)庫進(jìn)行自動(dòng)搜索，被廣泛應(yīng)用能夠?qū)ε潴w的三維結(jié)構(gòu)數(shù)據(jù)庫進(jìn)行自動(dòng)搜索，被廣泛應(yīng)用于基于受體結(jié)構(gòu)的數(shù)據(jù)庫搜索的藥物設(shè)計(jì)中。于基于受體結(jié)構(gòu)的數(shù)據(jù)庫搜索的藥物設(shè)計(jì)中。基本步驟：基本步驟： 1. 配體和受體相互作用位點(diǎn)的確定

32、；配體和受體相互作用位點(diǎn)的確定； 2. 評(píng)分系統(tǒng)的生成；評(píng)分系統(tǒng)的生成； 3. DOCK計(jì)算；計(jì)算； 4. DOCK結(jié)果的處理與分析。結(jié)果的處理與分析。321. 活性位點(diǎn)的確定活性位點(diǎn)的確定 活性位點(diǎn)的確定，通過軟件包中活性位點(diǎn)的確定，通過軟件包中 sphgen 程序完成。程序完成。 Sphgen 程序，它生成受體表面所有的凹陷的負(fù)像，并對(duì)這些負(fù)像程序，它生成受體表面所有的凹陷的負(fù)像，并對(duì)這些負(fù)像進(jìn)行聚類分析。下圖顯示了一個(gè)和兩個(gè)表面點(diǎn)相切的負(fù)像以及在進(jìn)行聚類分析。下圖顯示了一個(gè)和兩個(gè)表面點(diǎn)相切的負(fù)像以及在活性口袋中互相疊合的多個(gè)負(fù)像。活性口袋中互相疊合的多個(gè)負(fù)像。圖圖1a中，黑色的小球代表表

33、面點(diǎn)。在中，黑色的小球代表表面點(diǎn)。在DOCK程序中，表面點(diǎn)采用了程序中，表面點(diǎn)采用了Richards提出的模型。在這些表面點(diǎn)的基礎(chǔ)上，采用提出的模型。在這些表面點(diǎn)的基礎(chǔ)上，采用sphgen程序程序生成了負(fù)像，它實(shí)際上由一些與分子表面點(diǎn)相切的圓球疊加而成。生成了負(fù)像，它實(shí)際上由一些與分子表面點(diǎn)相切的圓球疊加而成。332. 匹配原則匹配原則 在生成的負(fù)像的基礎(chǔ)上，就可以進(jìn)行配體分子和在生成的負(fù)像的基礎(chǔ)上，就可以進(jìn)行配體分子和活性口袋之間的匹配。在這里，配體也采用一組活性口袋之間的匹配。在這里，配體也采用一組球集來表示，和負(fù)像不同的是，配體所用的球集球集來表示，和負(fù)像不同的是，配體所用的球集表示配體

34、所占的空間區(qū)域。表示配體所占的空間區(qū)域。 如果一個(gè)配體分子能和活性口袋形成比較好的匹如果一個(gè)配體分子能和活性口袋形成比較好的匹配，那么配體的球集一定能和活性口袋中的負(fù)像配，那么配體的球集一定能和活性口袋中的負(fù)像形成好的疊合。配體分子和負(fù)像之間的匹配原則形成好的疊合。配體分子和負(fù)像之間的匹配原則是基于配體和受體之間球集的內(nèi)坐標(biāo)的比較。是基于配體和受體之間球集的內(nèi)坐標(biāo)的比較。34例例1 如果配體用如果配體用m個(gè)球體表示，受體的活性口袋中負(fù)像個(gè)數(shù)為個(gè)球體表示，受體的活性口袋中負(fù)像個(gè)數(shù)為n，那么他們之間的可能匹配的數(shù)目為，那么他們之間的可能匹配的數(shù)目為n!/(n-m)!。為了簡化。為了簡化計(jì)算，可以采

35、用下面的操作進(jìn)行匹配。計(jì)算，可以采用下面的操作進(jìn)行匹配。1,先把配體球集中的一個(gè)球先把配體球集中的一個(gè)球i和活性位點(diǎn)的一個(gè)負(fù)像和活性位點(diǎn)的一個(gè)負(fù)像k進(jìn)行匹配進(jìn)行匹配2,接著計(jì)算配體球接著計(jì)算配體球i到其它配體球到其它配體球j之間的距離之間的距離dij，以及負(fù)像，以及負(fù)像k到到其它負(fù)像其它負(fù)像l之間的距離之間的距離dkl.如果滿足下面的條件，則配體球如果滿足下面的條件，則配體球j和負(fù)像和負(fù)像l可以有效的匹配。式中可以有效的匹配。式中取值一般在取值一般在1.0-2.0 之間之間 |d |dijij-d-dklkl| | 3,第二匹配點(diǎn)確定以后，可以確定第三個(gè)匹配點(diǎn)。第三個(gè)匹配第二匹配點(diǎn)確定以后，可

36、以確定第三個(gè)匹配點(diǎn)。第三個(gè)匹配點(diǎn)必須滿足兩組如上式所示的約束。同樣，第四個(gè)點(diǎn)必須點(diǎn)必須滿足兩組如上式所示的約束。同樣，第四個(gè)點(diǎn)必須滿足三個(gè)上式約束，每個(gè)有效匹配的配對(duì)點(diǎn)不能少于滿足三個(gè)上式約束，每個(gè)有效匹配的配對(duì)點(diǎn)不能少于4個(gè)。個(gè)。4,按照上述操作得到受體和配體的匹配點(diǎn)集以后，還要進(jìn)行進(jìn)按照上述操作得到受體和配體的匹配點(diǎn)集以后，還要進(jìn)行進(jìn)一步的調(diào)整。一步的調(diào)整。3. 得分函數(shù)得分函數(shù)按照匹配原則得到了配體和受體之間的匹配情況之后，就要通按照匹配原則得到了配體和受體之間的匹配情況之后，就要通過合理的得分函數(shù)來選擇最優(yōu)的結(jié)果。過合理的得分函數(shù)來選擇最優(yōu)的結(jié)果。DOCK提供了多種提供了多種得分函數(shù)來

37、評(píng)價(jià)配體和受體之間的結(jié)合情況，包括原子接得分函數(shù)來評(píng)價(jià)配體和受體之間的結(jié)合情況，包括原子接觸得分以及能量得分。觸得分以及能量得分。 原子接觸得分原子接觸得分 是是 DOCK提供簡單的評(píng)價(jià)表面匹配的評(píng)價(jià)函數(shù)。提供簡單的評(píng)價(jià)表面匹配的評(píng)價(jià)函數(shù)。 這個(gè)得分函數(shù)為配體和受體之間接觸重原子數(shù)的簡單加和。這個(gè)得分函數(shù)為配體和受體之間接觸重原子數(shù)的簡單加和。 所謂接觸原子指在一定距離之間的原子（一般定義為所謂接觸原子指在一定距離之間的原子（一般定義為4.5A)如果原字間距離太近，則這兩個(gè)原子被定義為原子碰撞。如果原字間距離太近，則這兩個(gè)原子被定義為原子碰撞。(1)碰撞的原子顯然不利于配體和受體的匹配，按規(guī)定

38、作為罰碰撞的原子顯然不利于配體和受體的匹配，按規(guī)定作為罰分從總得分中予以扣除。分從總得分中予以扣除。36 (2) 能量得分能量得分 DOCK把配體和受體之間的非鍵相互作用能作為能量匹配把配體和受體之間的非鍵相互作用能作為能量匹配的評(píng)價(jià)函數(shù)。具體公式如下：的評(píng)價(jià)函數(shù)。具體公式如下： 式中，式中，E表示配體和受體之間的相互作用能；表示配體和受體之間的相互作用能；rij 為原子為原子i 和和原子原子j 之間的距離；之間的距離；Aij 和和 Bij 為范得華排斥和吸引參數(shù)；為范得華排斥和吸引參數(shù)；a 和和b 表示范得華吸引和排斥方次；表示范得華吸引和排斥方次；qi和和qj為原子和原子上的為原子和原子上

39、的部分電荷；部分電荷；D為介電函數(shù)。為介電函數(shù)。 為了考察溶劑效應(yīng)對(duì)分子對(duì)接的影響，為了考察溶劑效應(yīng)對(duì)分子對(duì)接的影響，Kuntz的科研小組的科研小組在上式中引入基于普適波恩模型的去溶劑化能量項(xiàng)。計(jì)算在上式中引入基于普適波恩模型的去溶劑化能量項(xiàng)。計(jì)算結(jié)果表明引入溶劑效應(yīng)可以改善模型精度，但這部分工作結(jié)果表明引入溶劑效應(yīng)可以改善模型精度，但這部分工作在最新的在最新的DOCK軟件包中還沒有體現(xiàn)。軟件包中還沒有體現(xiàn)。37ijjibijijaijijrecjligiDrqqrBrAE332114 格點(diǎn)對(duì)接格點(diǎn)對(duì)接 在能量匹配得分計(jì)算中，按照一般的做法，需要計(jì)算配體在能量匹配得分計(jì)算中，按照一般的做法，需

40、要計(jì)算配體和受體之間所有在截?cái)嘀抵畠?nèi)的原子對(duì)之間的相互作用能，和受體之間所有在截?cái)嘀抵畠?nèi)的原子對(duì)之間的相互作用能，這個(gè)過程顯然是非常耗時(shí)的。為了加快計(jì)算的速度，這個(gè)過程顯然是非常耗時(shí)的。為了加快計(jì)算的速度，DOCK采取了格點(diǎn)（采取了格點(diǎn)（GRID）對(duì)接的技術(shù)（）對(duì)接的技術(shù)（Meng 1992）。）。 格點(diǎn)對(duì)接的思路；就是把勢能函數(shù)中與受體相關(guān)的函數(shù)項(xiàng)格點(diǎn)對(duì)接的思路；就是把勢能函數(shù)中與受體相關(guān)的函數(shù)項(xiàng)在空間某些區(qū)域取值預(yù)先計(jì)算得到，這些與受體有關(guān)的特在空間某些區(qū)域取值預(yù)先計(jì)算得到，這些與受體有關(guān)的特征量在分子對(duì)接中不再變化，只要考慮不同的配體就可以征量在分子對(duì)接中不再變化，只要考慮不同的配體就可

41、以了。了。 格點(diǎn)的范圍：一般要包含整個(gè)活性口袋區(qū)域，通常采用立格點(diǎn)的范圍：一般要包含整個(gè)活性口袋區(qū)域，通常采用立方體來定義。在定義的立方體空間區(qū)域等分放置相應(yīng)的格方體來定義。在定義的立方體空間區(qū)域等分放置相應(yīng)的格點(diǎn)。然后計(jì)算格點(diǎn)上的特征量的數(shù)值。點(diǎn)。然后計(jì)算格點(diǎn)上的特征量的數(shù)值。 在在DOCK中，每個(gè)格點(diǎn)具有三個(gè)特征值：計(jì)算表面接觸匹中，每個(gè)格點(diǎn)具有三個(gè)特征值：計(jì)算表面接觸匹配的格點(diǎn)值；計(jì)算靜電得分的格點(diǎn)值；計(jì)算范德華得分的配的格點(diǎn)值；計(jì)算靜電得分的格點(diǎn)值；計(jì)算范德華得分的格點(diǎn)值。格點(diǎn)值。38(1). 表面接觸匹配的格點(diǎn)值表面接觸匹配的格點(diǎn)值 這個(gè)值用來描述格點(diǎn)附近重原子的數(shù)目。這個(gè)值用來描述格點(diǎn)附近重原子的數(shù)目。 首先程序定義了原子和格點(diǎn)接觸的范圍，然后：首先程序定義了原子和格點(diǎn)接觸的范圍，然后： 當(dāng)重原子和格點(diǎn)的距離在這個(gè)范圍之內(nèi)的時(shí)候，格點(diǎn)上的當(dāng)重原子和格點(diǎn)的距離在這個(gè)范圍之內(nèi)的時(shí)候，格點(diǎn)上的數(shù)值加數(shù)值加1。 當(dāng)格點(diǎn)距離原子太近，則格點(diǎn)上的數(shù)值減當(dāng)格點(diǎn)距離原子太近，則格點(diǎn)上的