胡駿馳 徐 盼 陳彥韜 劉婷婷 陸 冬 徐 圓 陽懷宇 羅小民 朱維良 鄭明月于坤千 羅 成 蔣華良,** 中國科學(xué)院上海藥物研究所 上?!?03 上??萍即髮W(xué)生命科學(xué)與技術(shù)學(xué)院 上海 00
科研信息化助力合理藥物設(shè)計(jì)新發(fā)展*
胡駿馳1徐盼1陳彥韜1劉婷婷2陸冬1徐圓1陽懷宇1羅小民1朱維良1鄭明月1于坤千1羅成1蔣華良1,2**
1中國科學(xué)院上海藥物研究所上海201203
2上??萍即髮W(xué)生命科學(xué)與技術(shù)學(xué)院上海201210
大數(shù)據(jù)時代,超級計(jì)算機(jī)促進(jìn)了合理藥物設(shè)計(jì)研發(fā),這種藥物研究模式的轉(zhuǎn)變推動了創(chuàng)新新藥發(fā)現(xiàn)又一個春天的到來??蒲行畔⒒A(chǔ)設(shè)施的升級,包括高性能計(jì)算機(jī)處理器技術(shù)的不斷更新,新的資源模式的出現(xiàn),使得超算技術(shù)與計(jì)算生物學(xué)、計(jì)算化學(xué)密切結(jié)合,進(jìn)一步為傳統(tǒng)的藥物發(fā)現(xiàn)和計(jì)算機(jī)輔助藥物設(shè)計(jì)增添了新功能,加速了分子動力學(xué)模擬和虛擬篩選等的研究進(jìn)程。文章從藥物設(shè)計(jì)學(xué)方法研究、藥代動力學(xué)模型設(shè)計(jì)和藥物設(shè)計(jì)方法的具體應(yīng)用三個方面對當(dāng)今合理藥物設(shè)計(jì)領(lǐng)域的新發(fā)展進(jìn)行了評述,詳盡闡釋了科研信息化在提高效率、降低成本、加快進(jìn)程、管控風(fēng)險及提升研發(fā)價值和創(chuàng)新能力方面的優(yōu)勢。
科研信息化,合理藥物設(shè)計(jì),計(jì)算機(jī)輔助藥物設(shè)計(jì)
大數(shù)據(jù)時代,促進(jìn)科研信息化已成為國家核心競爭力的發(fā)展戰(zhàn)略之一。為在科研中保持優(yōu)勢和創(chuàng)新力,歐盟、美國、英國相繼出臺多項(xiàng)政策促進(jìn)科研信息化發(fā)展。我國也于2015 年 8 月由國務(wù)院印發(fā)《促進(jìn)大數(shù)據(jù)發(fā)展行動綱要》,以期快速實(shí)現(xiàn)“發(fā)展科學(xué)大數(shù)據(jù)”的目標(biāo)??蒲行畔⒒幕A(chǔ)設(shè)施建設(shè)全面加速發(fā)展,使超級計(jì)算機(jī)成為全球科技發(fā)展競爭和經(jīng)濟(jì)增長的基本工具之一[1]。E 級(Exascale,百億億次計(jì)算能力)高性能計(jì)算機(jī)成為世界各國的追求目標(biāo)之一。近年來,我國高性能計(jì)算技術(shù)取得了舉世矚目的成就。而在科學(xué)研究的整個生命周期中,科研信息化可以促進(jìn)所有學(xué)科的協(xié)同、計(jì)算或數(shù)據(jù)密集型研究的創(chuàng)新,推動交叉學(xué)科的快速發(fā)展。其中,伴隨著合理藥物設(shè)計(jì)方法主導(dǎo)的藥物研發(fā)模式的轉(zhuǎn)變,新藥發(fā)現(xiàn)過程迎來了一個新的春天,從 2010 年后,美國食品和藥物管理局(FDA)批準(zhǔn)的新藥數(shù)量出現(xiàn)了顯著增長[2]。尤其 2015 年,F(xiàn)DA 批準(zhǔn)了 44 種新藥,達(dá)到最近 19 年的最高值[3]。
科研信息化基礎(chǔ)設(shè)施的升級,包括高性能計(jì)算機(jī)處理器技術(shù)的不斷更新,新的資源提供模式(網(wǎng)格和云計(jì)算)的出現(xiàn),使得超算技術(shù)與計(jì)算生物學(xué)、計(jì)算化學(xué)密切結(jié)合,同時也為傳統(tǒng)的藥物發(fā)現(xiàn)和計(jì)算機(jī)輔助藥物設(shè)計(jì)增添了新功能[4],加快了分子動力學(xué)模擬和虛擬篩選等的研究進(jìn)程。在生命科學(xué)研究中,云計(jì)算已成為未來必不可少的基礎(chǔ)設(shè)施的重要部分[5]。目前,云計(jì)算已被應(yīng)用到高通量 DNA 測序[6,7]、新一代測序技術(shù)(next-generation sequencing, NGS)[8]和蛋白質(zhì)組學(xué)[9]等的研究中。在生物信息學(xué)技術(shù)的基礎(chǔ)上,聯(lián)合運(yùn)用計(jì)算機(jī)輔助藥物設(shè)計(jì)、結(jié)構(gòu)生物學(xué)、化學(xué)信息學(xué)和網(wǎng)絡(luò)藥理學(xué)等學(xué)科的研究方法,有助于挖掘已知和發(fā)現(xiàn)未知的蛋白-蛋白相互作用。此外,計(jì)算技術(shù)的合理應(yīng)用也為在海量的蛋白組學(xué)信息中發(fā)現(xiàn)可靶性的蛋白-蛋白相互作用提供了新的可能[10]。
近年來,分子生物學(xué)和結(jié)構(gòu)生物學(xué)快速發(fā)展,闡明了大量生物靶標(biāo)大分子的三維結(jié)構(gòu)和功能;計(jì)算機(jī)科學(xué)的發(fā)展和高性能計(jì)算機(jī)的出現(xiàn),又極大地提高了數(shù)據(jù)計(jì)算、分析的速度和精度。在此基礎(chǔ)上,基于計(jì)算模擬的藥物分子設(shè)計(jì)已作為一種實(shí)用化的工具介入到了藥物研究的各個環(huán)節(jié),并已成為創(chuàng)新藥物研究的核心技術(shù)之一[11]。我國科研人員在藥物設(shè)計(jì)方法學(xué)研究方面也做了大量創(chuàng)新性工作,掌握了一些具有特色和優(yōu)勢的關(guān)鍵核心技術(shù):比如開發(fā)的藥物靶標(biāo)預(yù)測方法和程序 TarFisDock[12],具有自主知識產(chǎn)權(quán)的藥物設(shè)計(jì)軟件包 D3Pharm,北京大學(xué)來魯華教授課題組研發(fā)的全新藥物設(shè)計(jì)程序 LigBuilder[13,14],劉曉峰等人發(fā)展的基于“反向藥效團(tuán)匹配”策略的靶標(biāo)預(yù)測方法 PharmMapper(并提供了相應(yīng)的網(wǎng)頁服務(wù))[15]等。這些方法已在國際上得到廣泛應(yīng)用,占據(jù)了國際生物信息技術(shù)開發(fā)前沿領(lǐng)域的一席之地。
在過去的幾十年中,傳統(tǒng)的基于特定靶標(biāo)的藥物設(shè)計(jì)策略在先導(dǎo)化合物發(fā)現(xiàn)中起了主導(dǎo)作用,其優(yōu)點(diǎn)是用更低的成本來獲得更高的通量。雖然運(yùn)用上述策略能夠發(fā)現(xiàn)高活性高選擇性且結(jié)構(gòu)新穎的化合物,但成藥性不高,其原因主要是缺少對系統(tǒng)層面藥物作用機(jī)制的認(rèn)識及藥物的脫靶效應(yīng)。因此,藥物潛在的靶標(biāo)識別在新藥研發(fā)過程中具有十分重要的意義,有助于發(fā)現(xiàn)類藥化合物全新的治療作用以及一些意想不到的毒副作用,研究人員可據(jù)此及時調(diào)整研發(fā)策略。此外,隨著技術(shù)的進(jìn)步和發(fā)展,生物活性數(shù)據(jù)呈指數(shù)型增長,傳統(tǒng)的方法很難充分利用這些龐大的數(shù)據(jù)信息。利用相似性融合的思想,建立了一個計(jì)算速度快、耗費(fèi)資源少的基于小分子配體K最近鄰(k-Nearest Neighbor, KNN)融合指紋相似性的靶標(biāo)預(yù)測模型[16]。通過和近年來靶標(biāo)預(yù)測方面性能較為突出的相似性集成算法(Sim ilarity Ensemble Approach, SEA)[17]進(jìn)行對比,可以發(fā)現(xiàn) KNN 融合策略要優(yōu)于 SEA 算法。進(jìn)一步開發(fā)了一款界面友好的免費(fèi)在線靶標(biāo)預(yù)測工具 Tarpred[18],可以很好地幫助研究人員對小分子可能的治療靶標(biāo)和副作用靶標(biāo)進(jìn)行預(yù)測,以便調(diào)整新藥研發(fā)策略,及時規(guī)避風(fēng)險。眾多的工作與成果已經(jīng)表明,計(jì)算機(jī)輔助靶標(biāo)識別作為一種快速且低成本的研究方法,對于藥物多靶標(biāo)作用的預(yù)測、藥物脫靶效應(yīng)的發(fā)現(xiàn)和小分子調(diào)控網(wǎng)絡(luò)的構(gòu)建都具有十分重要的意義。下表列舉了一些國內(nèi)研究人員開發(fā)的在線靶標(biāo)預(yù)測工具(表 1)。
表1 國內(nèi)靶標(biāo)預(yù)測在線工具
除了藥物潛在作用靶標(biāo)預(yù)測,發(fā)展體系普適性強(qiáng)、計(jì)算效率高和計(jì)算精度高的配體-受體結(jié)合熱力學(xué)參數(shù)(如結(jié)合自由能 ΔGbinding、結(jié)合常數(shù) KA或解離常數(shù) KD等)和結(jié)合動力學(xué)參數(shù)(如結(jié)合速率常數(shù) kon和解離速率常數(shù) koff)的計(jì)算方法仍是藥物設(shè)計(jì)領(lǐng)域的重要研究課題之一。迄今為止,已發(fā)展了一系列配體-受體結(jié)合熱力學(xué)與結(jié)合動力學(xué)參數(shù)計(jì)算方法。其中,配體-受體結(jié)合熱力學(xué)參數(shù)的計(jì)算方法可以大致分為兩類:(1)計(jì)算速度快而計(jì)算精度有限的分子對接(Molecular Docking)和打分函數(shù)(Scoring Function)[21];(2)計(jì)算精度較高而計(jì)算速度較低的結(jié)合自由能計(jì)算方法[22,23],如線性相互作用能(Linear Interaction Energy, LIE)方法[24,25]、MM-PBSA/GBSA[26]、自由能微擾法(Free Energy Perturbation,F(xiàn)EP)[27]和熱力學(xué)積分法(Thermodynam ic Integration,TI)[28]等。配體-受體結(jié)合動力學(xué)參數(shù)的計(jì)算方法則主要包括基于分子動力學(xué)(Molecular Dynamics)的方法和基于分子對接的方法[29],兩種方法通過構(gòu)建配體-受體作用的馬爾科夫狀態(tài)模型(Markov State Model)[30]或配體-受體結(jié)合自由能全景圖(Binding Free Energy Landscape)[31],模擬配體-受體的結(jié)合或解離反應(yīng)路徑,計(jì)算和預(yù)測配體-受體結(jié)合動力學(xué)參數(shù)。其中,基于分子動力學(xué)的方法,對計(jì)算資源的要求高、計(jì)算時間長,對配體-受體結(jié)合能量的評價精確性較低,且對配體-受體結(jié)合構(gòu)象的采樣有限,因此,構(gòu)建的配體-受體結(jié)合自由能全景圖或馬爾科夫狀態(tài)模型的準(zhǔn)確性有限。目前,采用此類方法的研究仍未能準(zhǔn)確計(jì)算配體-受體結(jié)合動力學(xué)參數(shù)。中科院上海藥物所與大連理工大學(xué)合作,將蛋白質(zhì)折疊的能量全景圖理論和研究化學(xué)反應(yīng)機(jī)理的過渡態(tài)理論(Transition State Theory)應(yīng)用于藥物-靶標(biāo)相互作用研究。基于考慮蛋白質(zhì)柔性的分子對接方法以及改進(jìn)的 MM-GBSA 結(jié)合自由能計(jì)算方法,設(shè)計(jì)了藥物-靶標(biāo)結(jié)合模式采樣空間離散化策略,發(fā)展了藥物-靶標(biāo)結(jié)合自由能全景圖的構(gòu)建方法以及藥物-靶標(biāo)結(jié)合熱力學(xué)與結(jié)合動力學(xué)計(jì)算方法[29]。然而,目前這種基于分子對接構(gòu)象采樣的方法,僅適用于蛋白質(zhì)柔性尺度較小的體系,對蛋白質(zhì)變構(gòu)尺度較大的體系,如激酶、HIV 蛋白酶等體系仍不適用。
在藥物設(shè)計(jì)中,分子對接方法顯著加速了先導(dǎo)化合物的發(fā)現(xiàn)進(jìn)程,是目前創(chuàng)新藥物研發(fā)過程中應(yīng)用最為廣泛的虛擬篩選方法之一[32]。正確的配體-受體結(jié)合構(gòu)象預(yù)測和準(zhǔn)確的活性預(yù)測是分子對接的兩個重要目標(biāo),二者的實(shí)現(xiàn)取決于用于靶標(biāo)-配體相互作用評價的打分函數(shù)。然而,現(xiàn)有的計(jì)算方法在靶標(biāo)-配體相互作用評價方面仍存在著許多問題,包括:(1)無法較好地考慮受體柔性;(2)難以準(zhǔn)確描述配體結(jié)合過程的溶劑化效應(yīng);(3)化合物的活性結(jié)合構(gòu)象識別能力低;(4)靶標(biāo)-配體結(jié)合自由能預(yù)測精度差等[23]。針對上述問題,系統(tǒng)地研究和改進(jìn)了基于知識的靶標(biāo)-配體相互作用打分函數(shù)和計(jì)算模型,發(fā)展了一系列具有創(chuàng)新性的方法。在考慮受體柔性方面,提出了一種基于統(tǒng)計(jì)分析的自適應(yīng)構(gòu)象取樣方法,可用于構(gòu)建蛋白質(zhì)構(gòu)象變化的馬爾科夫狀態(tài)模型[33];在提高分子對接打分函數(shù)性能方面,發(fā)展了一種新型的二維氫鍵統(tǒng)計(jì)勢,顯著提高了蛋白-配體復(fù)合物結(jié)合構(gòu)象的識別能力[34];針對在分子對接打分函數(shù)中考慮溶劑效應(yīng)這一難點(diǎn)問題,發(fā)展了可以對蛋白質(zhì)結(jié)合腔的水合位點(diǎn)以及結(jié)晶水保守性進(jìn)行預(yù)測的統(tǒng)計(jì)勢方法 w PMF[35]。此外,沈倩誠等人[36]還發(fā)展了可以較好地對打分函數(shù)的“知識”基礎(chǔ)進(jìn)行擴(kuò)充的統(tǒng)計(jì)勢迭代優(yōu)化方法,通過與包括 Glide 在內(nèi)的 7 種常見的打分函數(shù)進(jìn)行比較,發(fā)現(xiàn)經(jīng)過優(yōu)化后的統(tǒng)計(jì)勢能圖物理意義更加清晰,得到的打分函數(shù)在結(jié)合親和性預(yù)測方面也有了明顯的改進(jìn)。
針對鹵鍵(有機(jī)化合物分子中的鹵素原子與另一分子中的 O、N、S 等帶有部分負(fù)電荷的重原子、芳香環(huán)結(jié)構(gòu)或 π 鍵體系之間的吸引作用)在藥物設(shè)計(jì)研究中的重要作用[37],計(jì)算模擬了鹵鍵作用的本質(zhì),以及與氫鍵及陽離子 -π 鍵作用的相互影響,發(fā)現(xiàn)它們之間既可以相互協(xié)同加強(qiáng)也可以互相拮抗減弱藥物與靶標(biāo)蛋白之間的結(jié)合[38]。在此基礎(chǔ)上,還探索了鹵鍵對化合物成藥性的影響及原因[39],發(fā)展了可以用于藥物設(shè)計(jì)的鹵鍵打分函數(shù)[40],可快速評價藥物與靶標(biāo)蛋白之間的鹵鍵作用與強(qiáng)度。據(jù)此進(jìn)一步開展了先導(dǎo)化合物結(jié)構(gòu)優(yōu)化及老藥重定位研究,提高了先導(dǎo)化合物的活性,發(fā)現(xiàn)了一些老藥的新用途信息[41,42]。這些新型靶標(biāo)-配體相互作用預(yù)測方法對于提高目前合理藥物設(shè)計(jì)在先導(dǎo)化合物發(fā)現(xiàn)方面的有效性和成功率具有積極意義。
藥代動力學(xué)性質(zhì)不良及毒性是造成藥物在開發(fā)后期失敗的重要原因之一。在藥物研發(fā)早期,開展吸收、分布、代謝、排泄和毒性(ADME/T)實(shí)驗(yàn)或計(jì)算評價,可以減少人力、物力的浪費(fèi),而后者優(yōu)勢尤為明顯。ADME/T 預(yù)測方法發(fā)展和應(yīng)用研究,已經(jīng)成為藥物發(fā)現(xiàn)和優(yōu)化研究的一個重要研究方向。
伴隨著超級計(jì)算機(jī)的發(fā)展,曾經(jīng)難以實(shí)現(xiàn)的眾多計(jì)算密集型工作得以有效開展。在藥物的理化性質(zhì)與類藥性研究方面,通過大規(guī)模實(shí)驗(yàn)值訓(xùn)練模型,已有多個成功的酸堿離解常數(shù) p K a 預(yù)測模型報道[43,44]。結(jié)合深度學(xué)習(xí)方法構(gòu)建的可自動選擇描述符的化合物水溶性 log S 預(yù)測模型[45]與使用相似化合物構(gòu)建的局部預(yù)測模型[46]的一致性模型為 log S 的預(yù)測提供了新的思路。通過采用多種描述符的連續(xù)函數(shù)[47]描述方法,類藥性描述符[47]能夠較為準(zhǔn)確地預(yù)測吸收等性質(zhì),但目前仍存在半衰期預(yù)測效果不好等問題[48]。
近年來,隨著計(jì)算能力和資源的提升,綜合考慮代謝[49]、P 糖蛋白(P-glycoprotein, P-gp)和細(xì)胞色素 P450(Cytochrome P450, CYP450)酶[50],以及滲透性和溶解性[51]對化合物吸收的影響成為可能,同時也提高了口服生物利用度等預(yù)測的準(zhǔn)確性。采用小分子與人血清白蛋白(Human Serum A lbum in, HSA)之間作用描述符及分子描述符,已有多個關(guān)于藥物 HSA 結(jié)合預(yù)測模型[52-54]的報道。在考慮了血漿蛋白結(jié)合及腦組織結(jié)合的影響,以及利用大數(shù)據(jù)和隨機(jī)森林(Random Forest, RF)、支持向量機(jī)(Support Vector Machine, SVM)建模等計(jì)算方法的基礎(chǔ)上,多個高準(zhǔn)確率的血腦屏障預(yù)測模型得以成功構(gòu)建[55,56]。中科院上海藥物所聯(lián)合運(yùn)用遺傳算法(Genetic Algorithm, GA)與 SVM 來構(gòu)建藥物血腦屏障滲透性定量預(yù)測模型,模型在分析發(fā)現(xiàn)羧基、極性表面積/氫鍵形成能力、親脂性和電荷分布方面有重要作用[57]。此外分子動力學(xué)模擬方法預(yù)測血腦屏障滲透性參數(shù)則為血腦屏障滲透的理論研究指出了新方向[58]。
另一方面,非 P-gp 類跨膜轉(zhuǎn)運(yùn)體的研究工作也取得了長足進(jìn)展,包括有機(jī)陽離子轉(zhuǎn)運(yùn)體 2(OCT2)抑制劑、多藥和毒物外排轉(zhuǎn)運(yùn)體 1(MATE1)抑制劑等多種藥物轉(zhuǎn)運(yùn)體抑制劑的發(fā)現(xiàn)[59,60]為預(yù)測模型的構(gòu)建奠定了基礎(chǔ)。國內(nèi)也有利用 GA-CG-SVM 方法成功構(gòu)建乳腺癌抗性蛋白(BCRP)底物預(yù)測模型的報道[61]。中科院上海藥物所采用組合藥效團(tuán)策略,成功構(gòu)建了 OCT2、MATE1 抑制劑組合藥效團(tuán)模型(圖1),模型識別準(zhǔn)確率大大提升,并發(fā)現(xiàn)了有重要作用的結(jié)構(gòu)特征和藥效團(tuán)模型[62,63]。
圖1 MATE 1 抑制劑預(yù)測模型構(gòu)建流程
CYP 代謝位點(diǎn)預(yù)測方面,通過采用柔性分子對接和構(gòu)象系綜考慮 CYP2D6 藥物代謝位點(diǎn)柔性,并根據(jù)候選位點(diǎn)與催化中心的距離、反應(yīng)位點(diǎn)的內(nèi)在反應(yīng)性等預(yù)測代謝位點(diǎn),取得較好結(jié)果[64,65]。SMARTCyp 軟件的出現(xiàn)使得根據(jù)分子 2D 結(jié)構(gòu)預(yù)測其 CYP3A 4 代謝位點(diǎn)成為可能[66],進(jìn)一步將 SMARTCyp 推廣到 CYP1A 2、2A6、2B6、2C19、2C8、2C9、2D6 和 2E1 代謝位點(diǎn)的預(yù)測,也取得了較好的預(yù)測結(jié)果[67-69];而采用描述符空間二次抽樣和系綜方法構(gòu)建CYP代謝位點(diǎn)預(yù)測模型,使模型成功實(shí)現(xiàn)了預(yù)測與訓(xùn)練集距離較遠(yuǎn)的分子[70]。針對II相代謝反應(yīng),中科院上海藥物所也有聯(lián)合使用遺傳算法和支持向量機(jī),采用表征活性的量化描述符、分子體積和脂溶性等描述符研究葡萄糖醛酸轉(zhuǎn)移酶(UGTs)的成功報道[71]。
致突變/致癌毒性研究方面,M cCarren 等[72]分析了對 Am es 預(yù)測模型有重要影響的描述符,并分析了不同數(shù)據(jù)集對模型的影響。唐赟等[73]采用 SVM、決策樹(Decision Tree, DT)、人工神經(jīng)網(wǎng)絡(luò)(A rtificial Neural Netw ork, ANN)、KNN 和樸素貝葉斯(Naive Bayesian,NB)方法,以及 CDK 分子指紋等為描述符構(gòu)建了預(yù)測模型。中科院上海藥物所則使用 Gaston 方法進(jìn)行致癌警示結(jié)構(gòu)和調(diào)節(jié)因子的挖掘,發(fā)現(xiàn)一些新警示結(jié)構(gòu)[74]。
采用 SVM、DT、RF和NB 方法、遞歸分割方法以及深度學(xué)習(xí)方法構(gòu)建急性毒性預(yù)測模型已取得了初步成功[75-77]。而 Drwal 等人[78]發(fā)展了一種綜合考慮化合物相似性與毒性碎片的急性毒性預(yù)測方法,并可根據(jù)毒效團(tuán)預(yù)測可能的靶標(biāo)。中科院上海藥物所采用即時學(xué)習(xí)方法構(gòu)建了大鼠急性毒性局部預(yù)測模型,并通過構(gòu)建一致性模型提升模型預(yù)測性能[76]。Davis 等人[79,80]構(gòu)建了 CTD(Comparative Toxicogenom ics Database)系統(tǒng)毒理學(xué)數(shù)據(jù)庫,為研究人員開展化合物作用機(jī)制、疾病機(jī)理提供幫助。
隨著計(jì)算技術(shù)的迅猛發(fā)展,高質(zhì)量ADME/T實(shí)驗(yàn)數(shù)據(jù)積累和公布,相關(guān)分子機(jī)制的闡明,以及新的先進(jìn)預(yù)測算法和抽樣方法等的應(yīng)用,大規(guī)模計(jì)算已經(jīng)并將繼續(xù)深遠(yuǎn)地影響包括藥代動力學(xué)在內(nèi)的眾多藥物開發(fā)過程。而由于技術(shù)的不斷進(jìn)步,ADME/T 預(yù)測模型的質(zhì)量將越來越高,并朝實(shí)用化邁進(jìn),可以樂觀地預(yù)見,在不遠(yuǎn)的將來,使用計(jì)算方法預(yù)判藥代動力學(xué)性質(zhì)及毒性反應(yīng),并為后期開發(fā)提供警示將成為藥物設(shè)計(jì)過程中的重要環(huán)節(jié)。
過去 30 年中,伴隨著計(jì)算機(jī)性能的快速提升和計(jì)算機(jī)輔助藥物設(shè)計(jì)方法研究的深入,藥物設(shè)計(jì)技術(shù)應(yīng)用領(lǐng)域已由原先的活性化合物發(fā)現(xiàn)與優(yōu)化向上下游拓展深化,逐步應(yīng)用到包括藥物作用靶標(biāo)發(fā)現(xiàn)、藥物開發(fā)階段的藥效學(xué)評價、代謝研究、安全性評價(ADME/T)以及制劑研究等各個新藥開發(fā)研究領(lǐng)域[11]。此外,計(jì)算技術(shù)的靶標(biāo)確證技術(shù)的發(fā)展、蛋白晶體解析技術(shù)的成熟以及眾多新方法、新技術(shù)的不斷涌現(xiàn)使得從藥物靶標(biāo)發(fā)現(xiàn)及確證到小分子以及多肽及蛋白相關(guān)的藥物設(shè)計(jì)取得了長足進(jìn)步,也產(chǎn)出了豐碩的成果。尤其是在發(fā)現(xiàn)新的生物活性化合物(H it)等方面有著出色的表現(xiàn)。從應(yīng)用領(lǐng)域看,藥物設(shè)計(jì)學(xué)科近年來的發(fā)展方向可大致分為如下3個方面(圖2):(1)靶標(biāo)發(fā)現(xiàn)、功能確證及機(jī)制研究;(2)小分子藥物設(shè)計(jì);(3)多肽藥物設(shè)計(jì)。
圖2 藥物設(shè)計(jì)學(xué)已成為與化學(xué)、生物學(xué)所齊名的藥物研發(fā)工具
靶標(biāo)是指存在于組織細(xì)胞內(nèi)與藥物相互作用,并賦予藥物效應(yīng)的特定分子。藥物作用新靶標(biāo)的發(fā)現(xiàn)以及現(xiàn)有靶標(biāo)的新功能研究,是小分子探針及原創(chuàng)(First-inclass)藥物的源頭?;瘜W(xué)相似性搜索、數(shù)據(jù)挖掘、機(jī)器學(xué)習(xí)、反向?qū)?,生物活性譜分析、基于模型的方法以及基于各類分子網(wǎng)絡(luò)方法的發(fā)展為尋找重大疾病具有成藥性的新型靶標(biāo)及開發(fā)新的靶向藥物提供了契機(jī)[81-84]。同時,這些方法的應(yīng)用也為解釋藥物的脫靶效應(yīng)及藥物的副作用預(yù)測提供了支撐與依據(jù)[85,86]。結(jié)合當(dāng)前國際藥物研究中藥物靶標(biāo)發(fā)現(xiàn)、功能研究和新分子實(shí)體發(fā)現(xiàn)一體化趨勢和我國藥物研發(fā)的需求,中科院上海藥物所在原有研究的基礎(chǔ)上,建立了集靶標(biāo)發(fā)現(xiàn)、功能確證和藥物設(shè)計(jì)一體化的技術(shù)平臺,成為我國藥物設(shè)計(jì)研究重要的中心之一,對我國藥物設(shè)計(jì)領(lǐng)域的發(fā)展作出了突出的貢獻(xiàn),在國內(nèi)外產(chǎn)生了較大的影響[87-89]。其中針對細(xì)菌感染和表觀遺傳等重要領(lǐng)域,發(fā)現(xiàn)了一系列新的候選靶標(biāo),并針對發(fā)現(xiàn)的潛在藥物靶標(biāo)進(jìn)行藥物發(fā)現(xiàn)研究,取得了較好的進(jìn)展。一系列新靶標(biāo)、新機(jī)制的發(fā)現(xiàn)深化了對于生命調(diào)控過程的理解,也拓展了藥物作用位點(diǎn)和機(jī)制,為新型高活低毒藥物的誕生提供了可能[90,91]。
靶標(biāo)發(fā)現(xiàn)的另一個重要發(fā)展趨勢是,基于生物信息學(xué)、系統(tǒng)生物學(xué)分析發(fā)現(xiàn)疾病相關(guān)的新的信號轉(zhuǎn)導(dǎo)通路和生物調(diào)控網(wǎng)絡(luò),分析潛在靶標(biāo)在生物網(wǎng)絡(luò)中所處的位置及其對網(wǎng)絡(luò)功能的影響,為深入理解疾病及其藥物治療奠定基礎(chǔ)。癌癥、糖尿病及心血管疾病等復(fù)雜疾病的自身特點(diǎn)使得傳統(tǒng)的針對單一靶標(biāo)藥物難以發(fā)揮效果達(dá)成治療目標(biāo)。近年,Pf izer 的研究人員與 Allen, B. K.及其合作者先后通過基于結(jié)構(gòu)的藥物設(shè)計(jì)方法以及結(jié)合機(jī)器學(xué)習(xí)的大規(guī)模虛擬篩選發(fā)現(xiàn)了作用于癌癥信號通路的雙靶標(biāo)抑制劑,確定了使用計(jì)算機(jī)方法進(jìn)行基于信號通路藥物研究的可行性[92,93]。而構(gòu)建疾病相關(guān)生物分子網(wǎng)絡(luò),分析網(wǎng)絡(luò)動力學(xué)性質(zhì),尋找網(wǎng)絡(luò)敏感位點(diǎn)則為藥物發(fā)現(xiàn)提供了新方向[94]。通過針對一些典型的生物功能模塊的深入研究,理解其網(wǎng)絡(luò)核心拓?fù)浣Y(jié)構(gòu)、參數(shù)限制、關(guān)鍵節(jié)點(diǎn)搜尋及調(diào)控機(jī)制,中科院上海藥物所成功構(gòu)建了包括炎-癌轉(zhuǎn)變網(wǎng)絡(luò)在內(nèi)的數(shù)個復(fù)雜生物網(wǎng)絡(luò)模型,從系統(tǒng)的角度理解蛋白調(diào)控網(wǎng)絡(luò)問題,為復(fù)雜疾病網(wǎng)絡(luò)的化學(xué)干預(yù)預(yù)測提供了新思路[95]。
傳統(tǒng)意義上靶標(biāo)確證及功能研究主要通過實(shí)驗(yàn)手段來進(jìn)行,耗時費(fèi)力,復(fù)雜體系不易操作。計(jì)算生物學(xué)方法作為一種生物學(xué)研究工具,特別是分子動力學(xué)模擬方法通過長時間的模擬生物大分子靶標(biāo)的構(gòu)象變化,可以從微觀層面研究其功能,有效地彌補(bǔ)了實(shí)驗(yàn)方法的不足。近年來,G 蛋白偶聯(lián)受體(Guanosine-binding Protein Coup led Receptor, GPCR)和離子通道蛋白作為一類重要的潛在藥物靶標(biāo),受到越來越多的重視。其動態(tài)構(gòu)象研究以及離子通道,特別是電壓門控離子(Kv)通道和酸敏感離子通道的門控機(jī)制研究取得了較大進(jìn)展。中科院上海藥物所應(yīng)用計(jì)算生物學(xué)方法率先在全長 B 類 GPCR 動態(tài)構(gòu)象研究領(lǐng)域取得突破[96],首次揭示全長 B 類 GPCR 的 ECD 和跨膜區(qū)具有兩種相對構(gòu)象,推動了 B 類 GPCR 的結(jié)構(gòu)功能關(guān)系研究,也為設(shè)計(jì)基于新機(jī)制的 B類 GPCR 的功能小分子提供了重要信息。美國科學(xué)家通過計(jì)算生物學(xué)方法研究了 Kv1.2/Kv2.1 通道從開放到靜息再到開放的狀態(tài)轉(zhuǎn)變,完整描述了 Kv通道的開放關(guān)閉過程[97]。針對另一類重要的藥物靶標(biāo)——酸敏感質(zhì)子通道(ASIC),中科院上海藥物所和中科院神經(jīng)科學(xué)所研究人員合作利用計(jì)算生物學(xué)方法率先研究了 ASIC1 的動態(tài)行為[98,99](圖 3),首次清楚地描述了ASIC1 動態(tài)行為與功能的關(guān)系,并據(jù)此提出了一個新的 ASIC1 門控模型:旋轉(zhuǎn)打開機(jī)制模型。
圖3 Kv 通道、ASIC 通道和 GPCR 的配體調(diào)控位點(diǎn)示意圖
小分子對跨膜蛋白動態(tài)構(gòu)象的調(diào)控是跨膜蛋白研究領(lǐng)域的另一個重要關(guān)注點(diǎn)。研究組成細(xì)胞膜的磷脂和膽固醇等脂類小分子與跨膜蛋白作用和調(diào)控跨膜蛋白的生理及病理功能的分子機(jī)制,是當(dāng)前離子通道結(jié)構(gòu)功能關(guān)系研究的重要研究方向。中科院上海藥物所通過大規(guī)模分子動力學(xué)模擬與實(shí)驗(yàn)研究結(jié)合,闡述了 PIP2 與 KCNQ2 通道結(jié)合的動態(tài)過程[100]。在此基礎(chǔ)上,通過更長時間(1μs)分子動力學(xué)模擬發(fā)現(xiàn)了 PIP2 調(diào)控 KCNQ2 通道的新機(jī)制[101]。此外,通過長時間 MD 模擬發(fā)現(xiàn)磷脂對同源蛋白的差異化調(diào)控不局限于離子通道。這些研究為在分子水平上探索磷脂分子調(diào)控膜蛋白提供了結(jié)構(gòu)基礎(chǔ)和新見解,同時為理解膽固醇分子調(diào)控 GPCR 的功能多樣性提供了結(jié)構(gòu)基礎(chǔ)[102,103]。
近年來,隨著新的候選靶標(biāo)不斷涌現(xiàn),在沒有現(xiàn)成活性分子參考的情況下,藥物設(shè)計(jì)技術(shù)在新靶標(biāo)第一代活性分子發(fā)現(xiàn)中起到了不可替代的作用。從第一代活性分子開始,結(jié)合藥物設(shè)計(jì)技術(shù)進(jìn)行結(jié)構(gòu)優(yōu)化,從而實(shí)現(xiàn)從無到有、從有到精的分子實(shí)體發(fā)現(xiàn)?;诮Y(jié)構(gòu)和基于配體的藥物設(shè)計(jì)技術(shù)在整個新藥研發(fā)階段中占有不可替代的獨(dú)特地位,為開發(fā)選擇性高、活性好的先導(dǎo)化合物提供了重要的技術(shù)支撐。利用基于結(jié)構(gòu)的藥物設(shè)計(jì)(Structure Based Drug Design, SBDD)技術(shù),根據(jù)同源模建或晶體結(jié)構(gòu),使用經(jīng)典分子對接方法進(jìn)行虛擬篩選現(xiàn)已成為富集活性先導(dǎo)化合物、加速藥物發(fā)現(xiàn)過程的有效手段[23]。眾多實(shí)驗(yàn)結(jié)果已經(jīng)證明,計(jì)算機(jī)輔助藥物設(shè)計(jì)在包括癌癥、代謝疾病、神經(jīng)系統(tǒng)疾病、心血管疾病、呼吸系統(tǒng)疾病等眾多疾病治療領(lǐng)域均有著杰出的應(yīng)用前景[104-110],屢有針對其關(guān)鍵蛋白的作用方式獨(dú)特兼具杰出活性與靶標(biāo)特異性的優(yōu)異活性化合物實(shí)體被發(fā)現(xiàn)[111-113]。近年,中科院上海藥物所通過與芝加哥大學(xué)、艾默里大學(xué)等單位合作,利用虛擬篩選策略,發(fā)現(xiàn)了高活性高選擇性小分子DC-AC50,可同時靶向兩種銅伴侶蛋白A tox1和CCS,選擇性調(diào)控銅離子轉(zhuǎn)運(yùn),從而特異性抑制腫瘤細(xì)胞增殖,且在多種動物實(shí)驗(yàn)中表現(xiàn)出良好的抗腫瘤活性[114]。目前,DC-AC50 相關(guān)成果已經(jīng)實(shí)現(xiàn)轉(zhuǎn)化。
與 GPCR 和激酶等靶標(biāo)不同,電壓門控通道是被電壓激活,沒有明確的常規(guī)內(nèi)源性配體結(jié)合口袋。確證激動劑的作用位點(diǎn)是電壓門控通道研究領(lǐng)域的難點(diǎn)之一,通過基于結(jié)構(gòu)的藥物設(shè)計(jì)發(fā)現(xiàn)電壓門控通道激動劑也進(jìn)而面臨很大挑戰(zhàn)。計(jì)算生物學(xué)研究直接推動了基于動態(tài)構(gòu)象的藥物設(shè)計(jì),使得這些藥物靶標(biāo)的理性藥物設(shè)計(jì)成為可能。KCNQ2 是癲癇相關(guān)的一類電壓門控鉀離子通道,其通道激動劑被證實(shí)可以緩解人類的癲癇癥狀。中科院上海藥物所通過綜合運(yùn)用動力學(xué)模擬、分子對接、定點(diǎn)突變和電生理測試等方法,發(fā)現(xiàn)了一個位于通道門控電荷通路(gating charge pathway)中的激動劑結(jié)合口袋[115]并針對該口袋開展基于結(jié)構(gòu)的藥物設(shè)計(jì)。經(jīng)電生理測試確認(rèn)了 9 個 KCNQ2 新激動劑,其中兩個在兩類動物模型中表現(xiàn)出優(yōu)異的抗癲癇活性。該研究為發(fā)現(xiàn)離子通道調(diào)制劑結(jié)合口袋提供了成功的案例,并且首次實(shí)現(xiàn)了“基于結(jié)構(gòu)的電壓門控鉀離子通道激動劑發(fā)現(xiàn)”,為離子通道藥物研究領(lǐng)域的一個重要進(jìn)展。
蛋白-蛋白相互作用(Protein-Pro tein Interaction,PPI)是許多生命過程得以順利完成的基礎(chǔ),同時也是一類重要的藥物作用靶標(biāo),其應(yīng)用前景廣泛、市場潛力巨大。相比于傳統(tǒng)的高通量篩選,使用基于片段的藥物設(shè)計(jì)與改進(jìn)的虛擬篩選方法在尋找 PPI 抑制劑方面更為有效[116]。靶向蛋白-蛋白相互作用藥物設(shè)計(jì)的要點(diǎn)是尋找作用界面上的關(guān)鍵氨基酸殘基(hot spots)[117]。作為一個新興的研究領(lǐng)域,近年來,蛋白-蛋白相互作用研究得到越來越多的關(guān)注[118,119]。目前,已被批準(zhǔn)用于臨床的基于多肽設(shè)計(jì)的藥物超過 50 種。國內(nèi)也已報道了使用計(jì)算模擬與實(shí)驗(yàn)驗(yàn)證相結(jié)合方式的包括電壓門控的鉀離子通道(Kv1.3)抑制性多肽設(shè)計(jì)等多個成功案例[120],中科院上海藥物所也有通過對表觀遺傳調(diào)控蛋白復(fù)合物 PRC2 的界面性質(zhì)研究發(fā)現(xiàn)第一個靶向 EED-EZH2 復(fù)合物界面的小分子抑制劑的成功案例[121]。
目前,基于網(wǎng)絡(luò)的藥物設(shè)計(jì)還處在起步階段,已構(gòu)建的分子網(wǎng)絡(luò)模型數(shù)量、規(guī)模有限,對靶點(diǎn)/蛋白間相互影響的認(rèn)識也較為膚淺。但基于生物體自身調(diào)控網(wǎng)絡(luò)的復(fù)雜性,從系統(tǒng)層面對疾病與藥物的理解至關(guān)重要,基于調(diào)控網(wǎng)絡(luò)的藥物設(shè)計(jì)方法也是藥物設(shè)計(jì)領(lǐng)域的一大方向。隨著對于疾病網(wǎng)絡(luò)層面的理解的加深,分子網(wǎng)絡(luò)模型正在不斷發(fā)展與完善,可以預(yù)見,在不遠(yuǎn)的將來,基于對疾病的系統(tǒng)理解,癌癥、神經(jīng)系統(tǒng)疾病等復(fù)雜疾病的多靶標(biāo)藥物聯(lián)用治療將成為可能。而另一方面,由于蛋白-蛋白相互作用界面的關(guān)鍵殘基難以預(yù)測和模擬,現(xiàn)階段作用于蛋白-蛋白相互作用的小分子抑制劑數(shù)量還相對較少,如何有效地模擬和預(yù)測蛋白-蛋白作用過程仍是藥物設(shè)計(jì)學(xué)領(lǐng)域的一大難題。經(jīng)典的小分子虛擬篩選方式似乎并不適用于蛋白-蛋白作用界面抑制劑的篩選[122],而如何捕獲蛋白-蛋白相互作用反應(yīng)過程的動態(tài)構(gòu)型、使用柔性蛋白構(gòu)象進(jìn)行分子庫虛篩則是蛋白-蛋白相互作用小分子設(shè)計(jì)的先決條件,也是藥物設(shè)計(jì)學(xué)發(fā)展的重要方向。
綜上所述,由于藥物研發(fā)流程的復(fù)雜性,近年來大量自動化設(shè)備的使用,研發(fā)產(chǎn)生的海量數(shù)據(jù)也有著不同類型,這標(biāo)志著醫(yī)藥研發(fā)大數(shù)據(jù)時代的來臨?;谘邪l(fā)大數(shù)據(jù)時代的建模與模擬,需要建立企業(yè)級的研發(fā)信息高速公路。如今,制藥公司和學(xué)術(shù)機(jī)構(gòu)的合作也更為緊密,甚至形成了所謂的“開源藥物”研發(fā)模式。另外,研發(fā)領(lǐng)域出現(xiàn)了眾多移動應(yīng)用,并與云端形成合作平臺,如全球利用云技術(shù)進(jìn)行罕見病的合作研究,研發(fā)模式與互聯(lián)網(wǎng)時代同進(jìn)化。其中最為關(guān)鍵的就是整合不同來源和結(jié)構(gòu)的數(shù)據(jù),解析不同來源數(shù)據(jù)之間的關(guān)聯(lián),對原始數(shù)據(jù)進(jìn)行不同維度的數(shù)據(jù)解析。中科院上海藥物所已經(jīng)成功構(gòu)建了包涵情報信息系統(tǒng)、GRP 實(shí)驗(yàn)管理系統(tǒng)和 GLP 新藥申報系統(tǒng)等多項(xiàng)內(nèi)容的“新藥研發(fā)集成化平臺”,初步形成完整的新藥研發(fā)標(biāo)準(zhǔn)化流程管理系統(tǒng)。
信息技術(shù)改變了藥物研發(fā)模式。藥物研發(fā)信息化可以幫助我們提高效率、降低成本,加快研發(fā)進(jìn)程,進(jìn)行風(fēng)險控制,并提升研發(fā)價值和創(chuàng)新能力,使研發(fā)業(yè)務(wù)真正向智能化邁進(jìn),讓藥物研發(fā)各參與者擺脫盲人摸象的狀態(tài)。
1 Ge H, Wang Y, Li C, et al. M o lecular dynam ics-based virtual screening: Accelerating the drug discovery process by highperformance com puting. J. Chem. In f. M odel, 2013, 53(10):2757-2764.
2 M ullard A. 2014 FDA drug approvals. Nat. Rev. Drug Discov.,2015, 14 (2): 77-81.
3 Booth B L. This time may be different. Nat. Biotechnol., 2016,34 (1): 25-30.
4 Pitera J W. Current developments in and importance of highperformance computing in drug discovery. Curr. Opin. Drug Disc., 2009, 12 (3): 388-396.
5 Stein L D. The case for cloud computing in genome informatics. Genome Biol., 2010, 11 (5): 79-82.
6 Jourdren L, Bernard M, Dillies M-A, et al. Eoulsan: A cloud com puting-based framew ork facilitating high throughput sequencing analyses. Bioinformatics, 2012, 28 (11): 1542-1543.
7 Moghadam B T, Alvarsson J, Holm M, et al. Scaling predictive modeling in drug development w ith cloud computing. J. Chem. Inf. Model, 2015, 55 (1): 19-25.
8 Kwon T, Yoo W G, Lee W-J, et al. Next-generation sequencing data analysis on cloud computing. Genes Genom., 2015, 37 (6):489-501.
9 Mohammed Y, Mostovenko E, Henneman A A, et al. Cloud parallel p rocessing o f tandem m ass spectrom etry based proteom ics data. J. Proteome Res., 2012, 11(10): 5101-5108.
10 M ilroy L-G, Grossmann T N, Hennig S, et al. M odulators of protein-protein interactions. Chem. Rev., 2014, 114(9): 4695-4748.
11 Zheng M Y, Liu X, Xu Y, et al. Computational methods for drugdesign and discovery: Focus on china. Trends Pharmacol. Sci.,2013, 34(10): 549-559.
12 Li H L, Gao Z T, Kang L, et al. A web server for identifying drug targets w ith docking approach. Nucleic Acids Res., 2006, 34(SI):219-224.
13 Wang R X, Gao Y, Lai L H. Ligbuilder: A multi-purpose program for structure-based drug design. J. Mol. Model, 2000, 6(7-8):498-516.
14 Yuan Y X, Pei J F, Lai L H, et al.Ligbuilder 2: A practical de novo drug design approach. J. Chem. Inf. M odel, 2011, 51(5):1083-1091.
15 Liu X F, Ouyang S S, Yu B, et al.Pharmm apper server:A w eb server for potential drug target identification using pharmacophore mapping approach. Nucleic Acids Res., 2010,38(2): 609-614.
16 Liu X, Xu Y, Li S S, et al. In silico target fishing: Addressing a“big data” problem by ligand-based sim ilarity rankings w ith data fusion. J. Chem informatics, 2014, 6(1): 33-33.
17 Keiser M J, Roth B L, Armbruster B N, et al. Relating protein pharmacology by ligand chem istry. Nat. Biotechnol., 2007,25(2): 197-206.
18 Liu X, Gao Y, Peng J L, et al. Tarpred: A web app lication for predicting therapeutic and side effect targets of chem ical compounds. Bioinformatics, 2015, 31(12): 2049-2051.
19 Luo H, Chen J, Shi L M, et al. Drar-cpi: A server for identifying drug repositioning potential and adverse drug reactions via the chem ical-protein interactome. Nucleic Acids Res., 2011, 39(2):492-498.
20 Wang X, Chen H P, Yang F, et al. Idrug: A web-accessible and interactive drug discovery and design p latform. J. Chem informatics, 2014, 6(28): 205-230.
21 Huang S Y, Grinter S Z, Zou X. Scoring functions and their evaluation methods for protein-ligand docking: Recent advances and future directions. Phys. Chem. Chem. Phys., 2010, 12(40):12899-12908.
22 Singh N, Warshel A. Absolute binding free energy calculations:On the accuracy of computational scoring of protein-ligand interactions. Proteins, 2010, 78(7): 1705-1723.
23 Kitchen D B, Decornez H, Furr J R, et al. Docking and scoring in virtual screening for drug discovery: Methods and applications. Nat. Rev. Drug Discov., 2004, 3(11): 935-949.
24 Aqvist J, M ow bray S L. Sugar recognition by a glucose/ galactose receptor-evaluation of binding energetics from molecular-dynam ics simulations. J. Biol. Chem., 1995, 270(17):9978-9981.
25 Aqvist J, M edina C, Samuelsson J E. New method for predicting binding-affinity in computer-aided drug design. Protein Eng.,1994, 7(3): 385-391.
26 Aqvist J. Long-range electrostatic effects on peptide folding. FEBS Lett., 1999, 457(3): 414-418.
27 Wang W, Donini O, Reyes C M, et al. Biomolecular simulations:Recent developments in force fields, simu lations of enzyme catalysis, protein-ligand, protein-protein, and protein-nucleic acid noncovalent interactions. Annu. Rev. Biophys. Biom ol. Struct., 2001, 30(30): 211-243.
28 Adcock S A, M ccammon J A. M olecular dynam ics: Survey of methods for simulating the activity of proteins. Chem. Rev.,2006, 106(5): 1589-1615.
29 Bai F, Xu Y, Chen J, et al. Free energy landscape for the binding process of huperzine a to acetylcholinesterase. Proc. Natl. Acad. Sci. USA, 2013, 110(11): 4273-4278.
30Buch I, Giorgino T, De Fabritiis G. Complete reconstruction of an enzyme-inhibitor binding process by molecular dynam ics simulations. Proc. Natl. Acad. Sci. USA, 2011, 108(25): 10184-10189.
31 Okazaki K-I, Takada S. Dynam ic energy landscape view of coupled binding and protein conformational change: Induced-fit versus population-shift mechanisms. Proc. Natl. Acad. Sci. USA,2008, 105(32): 11182-11187.
32 Shoichet B K. Virtual screening of chem ical libraries. Nature,2004, 432(7019): 862-865.
33 Li S S, Xiong B, Xu Y, et al. M echanism of the all-alpha to allbeta conformational transition of rfah-ctd: Molecular dynam ics simulation and markov state model. J. Chem. Theory Com put.,2014, 10(6): 2255-2264.
34 Zheng M Y, Xiong B, Luo C, et al. Know ledge-based scoring functions in drug design: 3. A two-dimensional know ledge-based hydrogen-bonding potential for the prediction of protein-ligand interactions. J. Chem. Inf. Model, 2011, 51(11): 2994-3004.
35 Zheng M Y, Li Y L, Xiong B, et al. Water pm f for predicting the properties of water molecules in protein binding site. J. Comput. Chem., 2013, 34(7): 583-592.
36 Shen Q C, X iong B, Zheng M Y, et al. Know ledge-based scoring functions in drug design: 2. Can the know ledge base be enriched?. J. Chem. Inf. Model, 2011, 51(2): 386-397.
37 Lu Y X, Shi T, Wang Y, et al. Halogen bonding-a novel interaction for rational drug design?. J. Med. Chem., 2009,52(9): 2854-2862.
38 Lu Y X, Wang Y, Xu Z J, et al. C-x center dot center dot center dot h contacts in biomolecular systems: How they contribute to protein-ligand binding affinity. J. Phys. Chem. B, 2009, 113(37):12615-12621.
39 Xu Z J, Yang Z, Liu Y T, et al. Halogen bond: Its role beyond drug-target binding affinity for drug discovery and development. J. Chem. Inf. Model, 2014, 54(1): 69-78.
40 Liu Y T, Xu Z J, Yang Z, et al. A know ledge-based halogen bonding sco ring function fo r predicting pro tein-ligand interactions. J. Mol. Model, 2013, 19(11): 5015-5030.
41 Xu Z J, Liu Z, Chen T, et al. U tilization of halogen bond in lead optim ization: A case study of rational design of potent phosphodiesterase type 5 (pde5) inhibitors. J. M ed. Chem., 2011,54(15): 5607-5611.
42 Ren J, He Y, Chen W Y, et al. Thermodynam ic and structural characterization of halogen bonding in pro tein-ligand interactions: A case study of pde5 and its inhibitors. J. Med. Chem., 2014, 57(8): 3588-3593.
43 Settimo L, Bellman K, Knegtel R M A. Com parison of the accuracy of experimental and predicted pka values of basic and acidic compounds. Pharm. Res., 2014, 31(4): 1082-1095.
44 Fraczkiew icz R, Lobell M, Goeller A H, et al. Best of both worlds: Combining pharma data and state of the art modeling technology to im prove in silico pk(a) prediction. J. Chem. Inf. Model, 2015, 55(2): 389-397.
45 Lusci A, Po llastri G, Baldi P. Deep architectures and deep learning in chemoinformatics: The prediction of aqueous solubility for drug-like molecules. J. Chem. Inf. M odel, 2013,53(7): 1563-1575.
46 Raevsky O A, Po lianczyk D E, Grigorev V Y, et al. In silico prediction of aqueous solubility: A comparative study of local and global predictive models. M ol. Inf., 2015, 34(6-7): 417-430.
47 Bickerton G R, Paolini G V, Besnard J, et al. Quantifying the chemical beauty of drugs. Nat. Chem., 2012, 4(2): 90-98.
48 Ritchie T J, Macdonald S J F. How drug-like are ‘ugly’ drugs:Do drug-likeness metrics predict adme behaviour in humans?. Drug Discov. Today, 2014, 19(4): 489-495.
49 Tian S, Li Y Y, Wang J M, et al. A dm e evaluation in d rug discovery. 9. Prediction of oral bioavailability in humans based on molecular properties and structural fingerprints. M ol. Pharm.,2011, 8(3): 841-851.
50 Xu X, Zhang W X, Huang C, et al. A novel chemometric method for the prediction of human oral bioavailability. Int. J. Mol. Sci.,2012, 13(6): 6964-6982.
51 Newby D, Freitas A A, Ghafourian T. Decision trees to characterise the roles of permeability and solubility on the prediction of oral absorption. Eur. J. Med. Chem., 2015, 90: 751-765.
52 Li H Y, Chen Z X, Xu X J, et al. Predicting human plasmaprotein binding of drugs using plasma protein interaction qsar analysis (ppi-qsar). Biopharm. Drug Dispos., 2011, 32(6): 333-342.
53 Chen L J, Chen X. Results of molecular docking as descriptors to predict human serum album in binding affinity. J. Mol. Graph.,2012, 33: 35-43.
54 Lexa K W, Dolghih E, Jacobson M P. A structure-based model for predicting serum album in binding. PLoS One, 2014, 9(4):e93323.
55 M artins I F, Teixeira A L, Pinheiro L, et al. A bayesian approach to in silico blood-brain barrier penetration modeling. J. Chem. Inf. M odel, 2012, 52(6): 1686-1697.
56 Lanevskij K, Dapkunas J, Juska L, et al. Qsar analysis of bloodbrain distribution: The influence of p lasma and brain tissue binding. J. Pharm. Sci., 2011, 100(6): 2147-2160.
57 Zhang D Q, Xiao J F, Zhou N N, et al. A genetic algorithm based support vector machine model for blood-brain barrier penetration prediction. Biomed Res. Int., 2015, 2015: 292683.
58 Carpenter T S, Kirshner D A, Lau E Y, et al. A method to predict blood-brain barrier permeability of drug-like compounds using molecular dynam ics simulations. Biophys. J., 2014, 107(3): 630-641.
59 Kido Y, Matsson P, Giacomini K M. Profiling of a prescription drug library for potential renal drug-drug interactions mediated by the organic cation transporter 2. J. Med. Chem., 2011, 54(13):4548-4558.
60 Wittw er M B, Zur A A, Khuri N, et al. Discovery of potent,selective multidrug and toxin extrusion transporter 1 (mate1,slc47a1) inhibitors through prescription drug profiling and computational modeling. J. Med. Chem., 2013, 56(3): 781-795.
61 Zhong L, M a C Y, Zhang H, et al. A p rediction model o f substrates and non-substrates of breast cancer resistance protein(bcrp) developed by ga-cg-svm method. Com put. Biol. M ed.,2011, 41(11): 1006-1013.
62 Xu Y, Liu X, Li S S, et al. Combinatorial pharmacophore modeling of organic cation transporter 2 (oct2) inhibitors:Insights into multiple inhibitory mechanisms. Mol. Pharm.,2013, 10(12): 4611-4619.
63 Xu Y, Liu X, Wang Y L, et al. Combinatorial pharmacophore modeling of multidrug and toxin extrusion transporter 1 inhibitors: A theoretical perspective for understanding multiple inhibitory mechanisms. Sci. Rep., 2015, (5): 13684.
64 Li J, Schneebeli S T, By lund J, et al. Idsite: An accurate approach to predict p450-mediated drug metabolism. J. Chem. Theory Comput., 2011, 7(11): 3829-3845.
65 M oors S L C, Vos A M, Cumm ings M D, et al. Structure-based site of metabolism prediction for cytochrome p450 2d6. J. Med. Chem., 2011, 54(17): 6098-6105.
66 Rydberg P, G loriam D E, Zaretzki J, et al. Smartcyp: A 2d m ethod for prediction o f cy tochrome p450-m ediated d rug metabolism. ACS Med. Chem. Lett., 2010, 1(3): 96-100.
67 Rydberg P, Olsen L. Ligand-based site of metabolism prediction for cytochrome p450 2d6. ACS Med. Chem. Lett., 2012, 3(1):69-73.
68 Liu R F, Liu J, Tawa G, et al. 2d smartcyp reactivity-based site of metabolism prediction for major drug-metabolizing cytochrome p450 enzymes. J. Chem. Inf. Model, 2012, 52(6): 1698-1712.
69 Zaretzki J, Rydberg P, Bergeron C, et al. Rs-predictor models augmented w ith smartcyp reactivities: Robust metabolic regioselectivity predictions for nine cyp isozymes. J. Chem. Inf. Model, 2012, 52(6): 1637-1659.
70 Tyzack J D, Mussa H Y, Williamson M J, et al. Cytochrome p450 site of metabolism prediction from 2d topological fingerprints using gpu accelerated probabilistic classifiers. J. Chem informatics, 2014, 6(2), 29.
71 Peng J L, Lu J, Shen Q C, et al. In silico site of metabolism prediction for human ugt-catalyzed reactions. Bioinformatics,2014, 30(3): 398-405.
72 M ccarren P, Springer C, Whitehead L. An investigation into pharmaceutically relevant mutagenicity data and the influence on ames predictive potential. J. Chem informatics, 2011,3(22):12264-12269.
73 Xu C Y, Cheng F X, Chen L, et al. In silico prediction of chem ical ames mutagenicity. J .Chem. Inf. M odel, 2012, 52(11):2840-2847.
74 Wang Y, Lu J, Wang F, et al. Estimation of carcinogenicity using molecular fragments tree. J. Chem. Inf. Model, 2012, 52(8):1994-2003.
75 Li X, Chen L, Cheng F X, et al. In silico prediction of chem ical acute oral toxicity using multiclassification methods. J. Chem. Inf. Model, 2014, 54(4): 1061-1069.
76 Lu J, Peng J L, Wang J N, et al. Estimation of acute oral toxicity in rat using local lazy learning. J. Chem informatics, 2014, 6(1):334-336.
77 Wang S C, Li Y Y, Wang J M, et al. Admet evaluation in drug discovery. 12. Development of binary classification models for prediction of herg potassium channel blockage. Mol. Pharm.,2012, 9(4): 996-1010.
78 Drwal M N, Banerjee P, Dunkel M, et al. Protox: A web server for the in silico prediction of rodent oral toxicity. Nucleic Acids Res., 2014, 42(W 1): 53-58.
79 Davis A P, Murphy C G, Johnson R, et al. The comparative toxicogenom ics database: Update 2013. Nucleic A cids Res,2013, 41(D1): 104-114.
80 Kongsbak K, Hadrup N, Audouze K, et al. Applicability o f com putational system s biology in toxicology. Basic Clin. Pharmacol. Toxicol., 2014, 115(1): 45-49.
81 Csermely P, Korcsmaros T, Kiss H J M, et al. Structure and dynam ics of molecu lar netw orks: A novel paradigm o f drug discovery a comprehensive review. Pharmacol. Ther., 2013,138(3): 333-408.
82 Harrold J M, Ramanathan M, Mager D E. Netw ork-based approaches in drug discovery and early development. Clin. Pharmacol. Ther., 2013, 94(6): 651-658.
83 Wang L R, Ma C, Wipf P, et al. Targethunter: An in silico target identification tool for predicting therapeutic potential of small organic molecules based on chemogenom ic database. AAPS J.,2013, 15(2): 395-406.
84 Yizhak K, Gabay O, Cohen H, et al. Model-based identification o f d rug targets that revert disrup ted m etabo lism and its application to ageing. Nat. Commun., 2013, 4: 1-11.
85 Lounkine E, Keiser M J, W hitebread S, et al. Large-scale prediction and testing of drug activity on side-effect targets. Nature, 2012, 486(7403): 361-367.
86 Schenone M, Dancik V, Wagner B K, et al. Target identification and m echanism o f action in chem ical bio logy and d rug discovery. Nat. Chem. Biol., 2013, 9(4): 232-240.
87 Chen S J, Wang Y L, Zhou W, et al. Identifying novel selective non-nucleoside DNA methy ltransferase 1 inhibitors through docking-based virtual screening. J. Med. Chem., 2014, 57(21):9028-9041.
88 Zhang J, Liu H C, Zhu K K, et al. Antiinfective therapy w ith a small molecule inhibitor of staphylococcus aureus sortase. Proc. Natl. Acad. Sci. USA, 2014, 111(37): 13517-13522.
89 Huang Y, Yan J L, Li Q, et al. Meclofenam ic acid selectively inhibits fto demethylation of m(6)a over alkbh5. Nucleic Acids Res., 2015, 43(1): 373-384.
90 Chen Z, Wang X, Zhu W P, et al. Acenaphtho 1,2-b pyrrolebased selective fibroblast grow th factor receptors 1 (fgfr1)inhibitors: Design, synthesis, and biological activity. J. Med. Chem., 2011, 54(11): 3732-3745.
91 Sun W Y, Xie Z Q, Liu Y F, et al. Jx06 selectively inhibits pyruvate dehydrogenase kinase pdk1 by a covalent cysteine modification. Cancer Res., 2015, 75(22): 4923-4936.
92 Cui J J, Tran-Dube M, Shen H, et al. Structure based drug design of crizotinib (pf-02341066), a potent and selective dual inhibitorof mesenchymal-epithelial transition factor (c-met) kinase and anaplastic lymphoma kinase (alk). J. Med. Chem., 2011, 54(18):6342-6363.
93 A llen B K, M ehta S, Ember S W J, et al. Large-scale com putational screening identifies first in c lass m ultitarget inhibitor of egfr kinase and brd4. Sci. Rep., 2015, 5: 16924.
94 Yang K, Bai H J, Qi O Y, et al. Finding multiple target optimal intervention in disease-related molecular netw ork. Mol. Syst. Biol., 2008, 4(1): 228.
95 Lu J Y, Zeng H L, Liang Z J, et al. Network modelling reveals the mechanism underlying colitis-associated colon cancer and identifies novel combinatorial anti-cancer targets. Sci. Rep.,2015, 5: 2045-2322.
96 Yang L L, Yang D H, De Graaf C, et al. Conformational states of the full-length glucagon receptor. Nat. Commun., 2015, 6(7859):2708-2713.
97 Jensen M O, Jogini V, Borhani D W, et al. Mechanism of voltage gating in potassium channels. Science, 2012, 336(6078): 229-233.
98 Yang H Y, Yu Y, Li W G, et al. Inherent dynamics of the acidsensing ion channel 1 correlates w ith the gating mechanism. Plos Biol., 2009, 7(7): e1000151.
99 Yang H Y, Yu Y, Li W G, et al. Conformational sampling on acid-sensing ion channel 1 (asic1): Implication for a symmetric conformation. Cell Res., 2009, 19(8): 1035-1037.
100 Zhang Q S, Zhou P Z, Chen Z X, et al. Dynam ic pip2 interactions w ith voltage sensor elements contribute to kcnq2 channel gating. Proc. Natl. Acad. Sci. USA, 2013, 110(50):20093-20098.
101 Chen L P, Zhang Q S, Qiu Y G, et al. M igration of pip2 lipids on voltage-gated potassium channel surface influences channel deactivation. Sci. Rep., 2015, 5: 15079.
102 Cang X H, Du Y, Mao Y Y, et al. Mapping the functional binding sites o f cholesterol in beta(2)-adrenergic recep tor by longtime molecular dynam ics simulations. J. Phys. Chem. B, 2013,117(4): 1085-1094.
103 Cang X H, Yang L L, Yang J, et al. Cholesterol-beta(1)ar interaction versus cholesterol-beta(2)ar interaction. Proteins,2014, 82(5): 760-770.
104 Bouyssou T, Hoenke C, Rudolf K, et al. Discovery of olodaterol,a novel inhaled beta(2)-adrenoceptor agonist w ith a 24 h bronchodilatory efficacy. Bioorg. Med. Chem. Lett., 2010, 20(4):1410-1414.
105 Curreli F, Zhang H T, Zhang X H, et al. Virtual screening based identification of novel small-molecule inhibitors targeted to the hiv-1 capsid. Bioorg. Med. Chem., 2011, 19(1): 77-90.
106 Chen H F, Chan B K, Drummond J, et al. Discovery of selective lrrk2 inhibitors guided by computational analysis and molecular modeling. J. M ed. Chem., 2012, 55(11): 5536-5545.
107 Lane J R, Chubukov P, Liu W, et al. Structure-based ligand discovery targeting orthosteric and allosteric pockets o f dopam ine receptors. Mol. Pharmacol., 2013, 84(6): 794-807.
108 Lecka J, Ben-David G, Simhaev L, et al. Nonhydrolyzable atp analogues as selective inhibitors of human npp1: A combined com putational/experimental study. J. Med. Chem., 2013, 56(21):8308-8320.
109 B row n L M, Rogers K E, A roonsakool N, et al. A llosteric inhibition of epac com putational modeling and experimental validation to identify allosteric sites and inhibitors. J. Biol. Chem., 2014, 289(42): 29148-29157.
110 Link J O, Taylor J G, Xu L H, et al. Discovery of ledipasvir (gs-5885): A potent, once-daily oral ns5a inhibitor for the treatment of hepatitis c virus infection. J. Med. Chem., 2014, 57(5): 2033-2046.
111 Wu Y R, He C, Gao Y, et al. Dynam ic modeling of human 5-lipoxygenase-inhibitor interactions helps to discover novel inhibitors. J. Med. Chem., 2012, 55(6): 2597-2605.
112 Li X, Zuo Y Y, Tang G H, et al. Discovery of a series of2,5-diam inopyrim idine covalent irreversible inhibitors of bruton’s tyrosine kinase w ith in vivo antitum or activity. J. Med. Chem., 2014, 57(12): 5112-5128.
113 Meng F W, Cheng S F, Ding H, et al. Discovery and optimization of novel, selective histone methyltransferase set7 inhibitors by pharmacophore- and docking-based virtual screening. J. M ed. Chem., 2015, 58(20): 8166-8181.
114 Wang J, Luo C, Shan C L, et al. Inhibition o f hum an copper trafficking by a small molecule significantly attenuates cancer cell proliferation. Nat. Chem., 2015, 7(12): 968-979.
115 Li P, Chen Z X, Xu H Y, et al. The gating charge pathway of an epilepsy-associated potassium channel accommodates chemical ligands. Cell Res., 2013, 23(9): 1106-1118.
116 Sheng C Q, Dong G Q, M iao Z Y, et al. State-of-the-art strategies for targeting protein-protein interactions by small-molecule inhibitors. Chem. Soc. Rev., 2015, 44(22): 8238-8259.
117 Wanner J, Fry D C, Peng Z W, et al. Druggability assessment of protein-protein interfaces. Future M ed. Chem., 2011, 3(16):2021-2038.
118 Christ F, Voet A, Marchand A, et al. Rational design of smallmolecule inhibitors of the ledgf/p75-integrase interaction and hiv replication. Nat. Chem. Biol., 2010, 6(6): 442-448.
119 Li H M, Xiao H, Lin L, et al. Drug design targeting proteinprotein interactions (ppis) using multiple ligand simultaneous docking (m lsd) and drug repositioning: Discovery of raloxifene and bazedoxifene as novel inhibitors of il-6/gp130 interface. J. M ed. Chem., 2014, 57(3): 632-641.
120 Chen Z Y, Hu Y T, Hong J, et al. Tox in acid ic residue evolutionary function-guided design of de novo peptide drugs for the immunotherapeutic target, the kv1.3 channel. Sci. Rep.,2015, 5: 9881.
121 Kong X Q, Chen L M, Jiao L Y, et al. Astem izole arrests the proliferation of cancer cells by disrupting the ezh2-eed interaction of polycomb repressive complex 2. J. Med. Chem.,2014, 57(22): 9512-9521.
122 Pei J F, Yin N, Ma X M, et al. Systems biology brings new dimensions for structure-based drug design. J. Am. Chem. Soc.,2014, 136(33): 11556-11565.
胡駿馳中科院上海藥物所博士研究生。主要從事包括表觀遺傳代謝酶的機(jī)理研究、藥物發(fā)現(xiàn)和優(yōu)化等研究。先后發(fā)現(xiàn)了包括DC_05、DCE_42等在內(nèi)的多個結(jié)構(gòu)新穎、活性優(yōu)良的表觀遺傳酶小分子抑制劑。E-mail: hujunchi@simm.ac.cn
Hu JunchiPh.D. candidate of Shanghai Institute of M ateria M edica (SIMM), Chinese Academ y of Sciences. He is mainly engages in research on epigenetic enzymes, as well as drug discovery and optim ization. He discovered several small molecule inhibitors targeting epigenetic regulation with novel structure, such as DC_05 and DCE_42. E-mail: hujunchi@simm.ac.cn
蔣華良男,中科院上海藥物所所長,研究員。長期致力于藥學(xué)基礎(chǔ)研究和新藥研發(fā)。發(fā)展了多種靶標(biāo)發(fā)現(xiàn)和藥物設(shè)計(jì)方法,受到國際同行認(rèn)可、被廣泛應(yīng)用,用戶數(shù)達(dá) 3 000 余個;其中,應(yīng)用藥物-受體相互作用動力學(xué)原理,發(fā)展了精確預(yù)測藥效的新方法,解決了藥效難以預(yù)測這一長期困擾藥物設(shè)計(jì)領(lǐng)域的難題,為藥物研發(fā)提供了新的工具;把理論計(jì)算與實(shí)驗(yàn)研究結(jié)合起來,闡明了多種疾病相關(guān)靶標(biāo)蛋白質(zhì)動態(tài)行為與功能的關(guān)系;針對多種重要靶標(biāo)發(fā)現(xiàn)了數(shù)十個新結(jié)構(gòu)類型的先導(dǎo)化合物,正在進(jìn)行臨床前研究或臨床研究,其中兩個候選藥物已經(jīng)轉(zhuǎn)讓給企業(yè)進(jìn)行后續(xù)藥物開發(fā)。以通訊作者身份在 Neuron、PLoS Biol.、PNAS 等國際重要雜志上發(fā)表一系列原創(chuàng)性成果,他引6 000 余次,并應(yīng)邀在 Nature Chem. Biol., TiPS, Drug Discov Today, Curr. Med. Chem. 等期刊上撰寫綜述 10 余篇。作為第一完成人獲得國家自然科學(xué)獎二等獎、何梁何利獎等獎項(xiàng);先后被聘為“973”等重大項(xiàng)目的首席科學(xué)家和專家組成員,被聘為美國藥學(xué)重要雜志 J. Med. Chem. 副主編和 J. Biol. Chem. 等 5 個國際雜志編委。E-mail: hljiang@simm.ac.cn
Jiang HualiangM ale, professor and the director of Shanghai Institute of M ateria M edica (SIMM), Chinese Academ y of Sciences. In 1997,he received the prize from “the National Science Foundation for Distinguished Young Scholars”, and in 2007, he received the Second Class Prize of National Natural Science Award of China. Since 2013, he started to serve as the director of State Key Laboratory of Drug Research. He has also served as an associate editor of J Med Chem since 2009. Dr. Jiang's research interests are in computational biology and drug design. He has been engaged in the establishment of the innovative drug research platform of “integrating target discovery and drug design”. He developed a series of new methods for drug target discovery and drug design, which has been w idely used and drawn a great deal of attention from the research community. He has discovered drug lead compounds w ith novel structure for several important drug targets, among which some com pounds are undergoing preclinical or clinical studies. Based on the methodology that closely combines the theoretical calculation and experimental validation, he has directed several research programs investigating the relationship between conformation changes of drug targets and their pharmacological functions, and revealed new functions and mechanisms of the drug targets. Currently, he has published more than 200 research articles w ith the corresponding authorship in various journals including: PLoS. Biol., Neuron, Proc. Natl. Acad. Sci. U.S.A., J. Am. Chem. Soc., J. Biol. Chem., J. Med. Chem., and so on, and he published over 10 co-authorship papers o n Nature, Science, Nature Chem. Biol.,and Proc. Natl. Acad. Sci. U.S.A.. He was also invited to w rite more than ten review s for Nature Chem. Biol., Trends Pharmacol. Sci., Drug Discov. Today, Curr. Med. Chem., Proc. Natl. Acad. Sci. U.S.A. and so on. E-mail: hljiang@simm.ac.cn
e-Science Platform Construction Accelerates the Development of Rational Drug Design
Hu Junchi1Xu Pan1Chen Yantao1Liu Tingting2Lu Dong2Xu Yuan1Yang Huaiyu1Luo Xiaomin1Zhu Weiliang1Zheng Mingyue1Yu Kunqian1Luo Cheng1Jiang Hualiang1,2
(1Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai 201203, China;2School of Life Science and Technology, ShanghaiTech University, Shanghai 201210, China)
As the big data era approaching, supercomputer-guided rational drug discovery and design provides a new opportunity for modern drug discovery. The infrastructure upgrades of e-science platform, including the high-performance computers' microprocessors development,and the appearance of new resource pattern, connect supercomputer w ith computational biology and computational chem istry, add new features to the traditional drug discovery and computer-aided drug design, as well as speed up the research process of the molecular dynam ics simulation and the virtual screening. To illustrate the advantage of e-science platform in efficiency enhancement, cost reduction, process acceleration,risk management, as well as value improvement of research and innovative ability, this review covers a comprehensive range of the current development of rational drug design from three respects: the methodology research of drug design, the pharmacokinetics model design, and thedrug design application.
e-S cience, rational drug design, computer-aided drug design
10.16418/j.issn.1000-3045.2016.06.005
**通訊作者
*修改稿收到日期:2016 年 3月9日