劉羽晨,喬顯亮
大連理工大學(xué)環(huán)境學(xué)院 工業(yè)生態(tài)與環(huán)境工程教育部重點(diǎn)實(shí)驗(yàn)室,大連 116024
水生生物急性毒性QSAR模型研究進(jìn)展
劉羽晨,喬顯亮*
大連理工大學(xué)環(huán)境學(xué)院 工業(yè)生態(tài)與環(huán)境工程教育部重點(diǎn)實(shí)驗(yàn)室,大連 116024
化學(xué)品污染對人類健康和生態(tài)環(huán)境造成潛在風(fēng)險(xiǎn)。但是,危害性信息缺失是進(jìn)行化學(xué)品風(fēng)險(xiǎn)評價(jià)的主要挑戰(zhàn)。經(jīng)濟(jì)合作與發(fā)展組織(OECD)和美國環(huán)保署都提倡用非動(dòng)物實(shí)驗(yàn)替代方法來彌補(bǔ)數(shù)據(jù)缺失。定量結(jié)構(gòu)-活性關(guān)系(QSAR)被認(rèn)為是一種有應(yīng)用前景的替代技術(shù)。水生生物急性毒性是化學(xué)品風(fēng)險(xiǎn)評估和優(yōu)先污染物篩選中最常用的參數(shù)之一。但是,目前可獲得的實(shí)驗(yàn)毒性數(shù)據(jù)非常有限。本文總結(jié)了近年來發(fā)展的急性毒性預(yù)測模型,包括:(1)基于同類化合物建模;(2)基于數(shù)理統(tǒng)計(jì)建模;(3)基于化合物毒性作用模式建模。從模型預(yù)測能力、應(yīng)用域、機(jī)理解釋等角度對這3類模型進(jìn)行了比較。其中,基于作用模式構(gòu)建的模型一般具有較好的預(yù)測性能,并有助于機(jī)理解釋,將是今后水生生物急性毒性預(yù)測的發(fā)展方向。
定量結(jié)構(gòu)-活性關(guān)系(QSAR);水生生物急性毒性;作用模式
目前,美國化學(xué)文摘社(Chemical Abstracts Service, CAS)數(shù)據(jù)庫中的化學(xué)品約有9 108萬種[1],其中絕大多數(shù)是人為合成的有機(jī)物,這些物質(zhì)在人類生產(chǎn)、生活中產(chǎn)生了眾多有益影響,同時(shí)也對人類健康和環(huán)境造成了巨大威脅。2006年12月18日,歐盟通過立法發(fā)布了REACH法規(guī)[2](Registration, Evaluation, Authorization and Restriction of Chemicals),該法規(guī)要求進(jìn)入歐盟市場的所有化學(xué)品需要進(jìn)行預(yù)防性管理,并計(jì)劃對進(jìn)入歐盟市場的每年超過1 t的化學(xué)品進(jìn)行毒性評估。然而,這項(xiàng)任務(wù)面臨著巨大的挑戰(zhàn),一方面目前的化學(xué)品信息非常缺失[3-4],例如根據(jù)歐盟商品化的化學(xué)品清單,既有化學(xué)品和新化學(xué)品總數(shù)已達(dá)132 119種,但僅有2 198種化學(xué)品具有實(shí)驗(yàn)測試的急性毒性數(shù)據(jù)[5]。另一方面,傳統(tǒng)的測試方法耗費(fèi)大量的人力物力[6-7],從2001年至2005年,歐盟依據(jù)REACH法規(guī)在化學(xué)品毒理學(xué)測試中約花費(fèi)16億歐元[8]。因此,歐盟和美國都提倡采用一些預(yù)測技術(shù)如定量結(jié)構(gòu)-活性關(guān)系(QSAR)和交互比對(Read-Across)等方法來彌補(bǔ)化學(xué)品管理中的數(shù)據(jù)缺失[9]。
QSAR能夠通過計(jì)算化學(xué)品的特征參數(shù)來預(yù)測化學(xué)品理化性質(zhì)、毒理學(xué)效應(yīng)。其理論依據(jù)是分子結(jié)構(gòu)與有機(jī)物的理化性質(zhì)、環(huán)境中的遷移轉(zhuǎn)化行為和毒理學(xué)效應(yīng)是有內(nèi)在聯(lián)系的,而這種聯(lián)系是可以被認(rèn)識(shí)、表征并應(yīng)用的[10]。運(yùn)用QSAR可以對現(xiàn)有和尚未投入使用的化學(xué)品的相關(guān)性質(zhì)參數(shù)進(jìn)行預(yù)測和評價(jià),有助于化學(xué)品的管理以及對污染的預(yù)先防范[11]。經(jīng)濟(jì)合作與發(fā)展組織(OECD)于2007年提出了QSAR模型發(fā)展和使用準(zhǔn)則[12]:(1)有明確定義的環(huán)境(活性)指標(biāo);(2)有明確的算法;(3)能夠定義模型的應(yīng)用域;(4)有適當(dāng)?shù)臄M合度、穩(wěn)定性和預(yù)測能力;(5)最好能夠進(jìn)行機(jī)理解釋。符合該準(zhǔn)則的QSAR模型可以應(yīng)用于化學(xué)品風(fēng)險(xiǎn)評價(jià)、篩選和優(yōu)先控制等管理。該準(zhǔn)則的提出為QSAR模型的構(gòu)建和使用指明了方向。
化學(xué)品對水生生物的毒性信息被作為化學(xué)品風(fēng)險(xiǎn)評估和優(yōu)先污染物篩選的關(guān)鍵指標(biāo)之一。魚類(如黑頭呆魚(Pimephales promelas))、水蚤類(如大型蚤(Daphnia magana))、纖毛蟲類(如梨形四膜蟲(Tetrahymena pyriformis))、藻類(如斜生柵列藻(Scenedesmus obliquus))等作為不同營養(yǎng)級(jí)的代表性水生生物常被用于急性毒性研究,一般采用半數(shù)致死濃度(LC50)和半數(shù)效應(yīng)濃度(EC50)作為毒性終點(diǎn)?;谀壳皵?shù)據(jù)庫和文獻(xiàn)中的急性毒性實(shí)驗(yàn)數(shù)據(jù),前人運(yùn)用不同類別的分子描述符和算法構(gòu)建了一系列QSAR模型[13-16]。目前,水生生物急性毒性QSAR模型主要可以分為3類:(1)基于同類化合物構(gòu)建的模型;(2)基于數(shù)理統(tǒng)計(jì)構(gòu)建的模型;(3)基于化合物毒性作用模式分類構(gòu)建的模型。下面將針對這3類模型進(jìn)行分別介紹。
將結(jié)構(gòu)相似的同類化合物作為訓(xùn)練集來建模在水生生物急性毒性QSAR模型中較為常見(見表1)。魚類的模型如鹵代苯類[17]、三唑類[18]、丙烯酸類[19]等。大型蚤的模型如多環(huán)芳烴類[20]、三唑類[21]、有機(jī)磷酸酯類[22]、苯甲酸類[23]等。纖毛蟲類的模型如芳香醛類[24]、環(huán)氧化合物類[25]等。藻類的模型如鹵代芳香族化合物[26]、腈類[27]、季胺類[28]等。這類模型主要包含一些毒性較大,受到廣泛關(guān)注的化合物。模型中的實(shí)驗(yàn)數(shù)據(jù)通常實(shí)驗(yàn)條件相同或相近,受實(shí)驗(yàn)條件差異而導(dǎo)致的數(shù)據(jù)不確定性的影響較小。此外,同類化合物一般具有相同的毒性作用機(jī)制,化合物的結(jié)構(gòu)特征明顯,易于提取特征結(jié)構(gòu)信息來建立模型。因此這類模型通常預(yù)測較準(zhǔn)確,簡單透明,描述符較少,有些只需疏水性描述符或電子描述符,如Wang等[29]運(yùn)用疏水性描述符建立全氟羧酸對發(fā)光菌的毒性QSAR模型,模型回歸系數(shù)較高。這類模型利于理解化合物結(jié)構(gòu)和毒性作用的關(guān)系,提高模型使用的可信度,其不足是覆蓋的化合物種類太少,應(yīng)用域都比較窄,預(yù)測應(yīng)用受到一定限制。另外,同類化合物也會(huì)包含不同的取代基團(tuán),往往會(huì)導(dǎo)致化合物的毒性產(chǎn)生差異。Song等[30]研究了6種萘醌類化合物對大型蚤的急性毒性,結(jié)果表明由于化合物的疏水性不同,使其毒性差異非常大,導(dǎo)致毒性類別也不盡相同。Zhang等[31]研究發(fā)現(xiàn)丙烯酸酯類化合物上的烷基取代基能使其毒性降低,使得丙烯酸酯類化合物并不都表現(xiàn)出過量毒性。因此,基于同類化合物建模不僅存在應(yīng)用域小的限制,而且由于同類化合物的結(jié)構(gòu)差異也給模型構(gòu)建和預(yù)測帶來一些挑戰(zhàn)。
表1 基于同類化合物構(gòu)建的QSAR模型Table 1 QSAR models for chemical classes
注:n,驗(yàn)證集化合物個(gè)數(shù);N,描述符個(gè)數(shù);R2,回歸系數(shù)。
Note: n, number of chemicals; N, number of descriptors; R2, the squared correlation coefficient.
基于數(shù)理統(tǒng)計(jì)構(gòu)建的模型通常將多種類別化合物一起建模,通過不同的算法來得到擬合度較好的模型。這種建模方法一定程度上解決了基于同類化合物建模方法中應(yīng)用域較窄的問題。表2列出了近幾年發(fā)表的基于數(shù)理統(tǒng)計(jì)構(gòu)建的水生生物急性毒性模型。從表中可以看出,這些模型的數(shù)據(jù)集較大,但有些模型參數(shù)并不是很理想,如Kar和Ray[35]選取了297個(gè)化合物的大型蚤急性毒性數(shù)據(jù),使用了12個(gè)2D、3D描述符,回歸系數(shù)R2僅為0.738。Tao等[36]采用碎片常數(shù)法對217個(gè)化合物建模,雖然回歸系數(shù)高達(dá)0.97,但該方法選取了103個(gè)分子碎片和21個(gè)結(jié)構(gòu)修正因子,模型較復(fù)雜,不便于預(yù)測應(yīng)用。隨著一些機(jī)器學(xué)習(xí)類算法的引入,人工智能類模型被應(yīng)用于QSAR建模。Niculescu等[37]運(yùn)用概率神經(jīng)網(wǎng)絡(luò)算法建立大型蚤急性毒性的QSAR模型,模型的預(yù)測能力較好,回歸系數(shù)達(dá)0.85?;跈C(jī)器學(xué)習(xí)的方法構(gòu)建的模型雖然具有較好的擬合度,但模型算法和形式不夠透明,機(jī)理解釋性較差,會(huì)影響到結(jié)果的可信度和可接受程度,在一定程度上違背了OECD關(guān)于QSAR模型構(gòu)建和使用準(zhǔn)則的要求。
McFarland[41]研究認(rèn)為,化合物的毒性主要與其進(jìn)入生物膜的穿透能力和與作用位點(diǎn)的相互作用有關(guān)。不同化合物對不同受試生物的毒性大小和毒性機(jī)制會(huì)存在差異。前文介紹的兩類模型主要考慮到化合物本身的結(jié)構(gòu)與毒性的關(guān)系,并沒有考慮到有機(jī)物與生物受體之間作用關(guān)系對毒性的影響,顯然是不夠全面的。因此一些研究者嘗試基于毒性作用模式(mode of action, MOA)來構(gòu)建模型。應(yīng)用這一方法建模不僅依據(jù)化合物結(jié)構(gòu),還要考慮到化合物與生物受體之間的作用,一般需要先根據(jù)化合物結(jié)構(gòu)特征和毒性反應(yīng)機(jī)理對化合物進(jìn)行機(jī)制分類,再構(gòu)建模型。這類建模方法,一定程度上體現(xiàn)了OECD對QSAR構(gòu)建模型關(guān)于機(jī)理域的要求[12],即具有相同毒性機(jī)制的化合物可以組成訓(xùn)練集構(gòu)建模型,預(yù)測集化合物的毒性作用機(jī)理應(yīng)該與訓(xùn)練集化合物一致才能得到良好的預(yù)測結(jié)果。
3.1 毒性作用模式
認(rèn)識(shí)毒性作用模式是建立毒理學(xué)效應(yīng)QSAR模型的基礎(chǔ)和前提。目前一些研究者對毒性作用模式進(jìn)行了研究,提出了幾種分類方法(見表3)。McKim等[42]選取彩虹魚為受試生物,測試了多種污染物對其毒性影響。通過對魚的毒性反應(yīng)進(jìn)行統(tǒng)計(jì)分析,基于毒性作用模式將化合物分為6種,即非極性麻醉劑、極性麻醉劑、氧化磷酸化解偶聯(lián)劑、呼吸抑制劑、乙酰膽堿酶抑制劑和中樞神經(jīng)系統(tǒng)控制劑。Verhaar等[3,43-45]根據(jù)毒性作用模式將化合物分為4類,分別為惰性化學(xué)物質(zhì)、亞惰性化學(xué)物質(zhì)、活性化學(xué)物質(zhì)、特殊反應(yīng)化學(xué)物質(zhì),認(rèn)為4種作用模式與不同的結(jié)構(gòu)類別有關(guān),并提出了詳細(xì)的分類規(guī)則。Russom等[46]以魚類96 h- LC50值為毒性終點(diǎn),通過劑量-反應(yīng)關(guān)系、聯(lián)合毒性反應(yīng)研究、魚類急性毒性綜合癥(FATS)等,提出了8種毒性作用模式,總結(jié)了相應(yīng)的分類規(guī)則,并根據(jù)化合物所含的數(shù)據(jù)信息對各作用模式分類設(shè)置了置信度。Nendza和Wenzel[47]提出了9種機(jī)理,其中包括只與藻類有關(guān)的光合作用抑制和只與魚類有關(guān)的雌激素活性。Nendza和Muller[48]運(yùn)用體外成套測驗(yàn)(Battery test)方法將115種化合物按照這9種作用模式進(jìn)行分類,并從化合物描述符的角度嘗試對化合物分類,最終通過13個(gè)描述符用逐步判別分析的方法正確的將103個(gè)化合物分類,并認(rèn)為基于描述符的判別標(biāo)準(zhǔn)能夠用于作用模式下QSAR的預(yù)測和使用。
表2 基于數(shù)理統(tǒng)計(jì)構(gòu)建的QSAR模型Table 2 Statistically derived QSAR models
注:LC50,半數(shù)致死濃度;EC50,半數(shù)效應(yīng)濃度;IC50,半數(shù)抑制濃度;MLR,多元線性回歸;GA,遺傳算法;VSS,變量子集選擇;ANN,人工神經(jīng)網(wǎng)絡(luò);PLS,偏最小二乘;MLE,極大似然估計(jì);PNN,概率神經(jīng)網(wǎng)絡(luò);SVM,支持向量機(jī);DTB,決策樹;BPANN,反向傳播人工神經(jīng)網(wǎng)絡(luò)。
Note: LC50, Median lethal concentration; EC50, Median effective concentration; IC50, Median inhibitory concentration; MLR, Multiple linear regression; GA, Genetic algorithm; VSS, Variable subset selection; ANN, Artificial neural network; PLS, Partial least squares; MLE, Maximum likelihood estimation; PNN, Probabilistic neural network; SVM, Support vector machine; DTB, Decision treeboost; BPANN, Back-propagation artificial neural network.
盡管不同研究者提出的毒性作用模式的類別有差異,但整體上可分為麻醉劑類和其他反應(yīng)模式。麻醉劑作用模式被認(rèn)為是基本的或是最弱的效應(yīng),又稱基線毒性,其他反應(yīng)模式毒性大于麻醉劑類毒性,稱為過量毒性。可以采用Kow預(yù)測的有機(jī)物基線毒性值與毒性實(shí)驗(yàn)值的毒性比(toxic ratio)來區(qū)分基線毒性和過量毒性,一般認(rèn)為毒性比>10可以作為過量毒性。von der Ohe等[49]以大型蚤急性毒性數(shù)據(jù)作為毒性終點(diǎn),以化合物結(jié)構(gòu)警示(structural alerts)作為判定規(guī)則提出了3種方法來區(qū)分過量毒性和麻醉劑。由于3種方法對麻醉劑化合物和過量毒性化合物的預(yù)測能力上各有優(yōu)勢,所以建議將3種方法聯(lián)合使用來進(jìn)行分類。Nendza等[50]綜合了化合物結(jié)構(gòu)和物理化學(xué)性質(zhì)提出了一種區(qū)分基線毒性和過量毒性的判定規(guī)則,能夠較準(zhǔn)確的識(shí)別基線毒性化合物。
3.2 基于毒性作用模式的QSAR模型
目前,基于毒性作用模式分類構(gòu)建了一些QSAR模型[44,51-54]。其中Verhaar[43]提出的分類方法使用較多[44,51-53]。這些模型中,非極性麻醉劑和極性麻醉劑類模型回歸效果較好(見表4),原因在于這類化合物與機(jī)體中特殊的受體沒有相互作用,毒性的大小主要取決于化合物的疏水性,在建模時(shí)僅用疏水性參數(shù)(如辛醇水分配系數(shù),Kow)便可得到擬合度較好的模型[55-57]。極性麻醉劑能夠提供氫鍵供體,產(chǎn)生的毒性比非極性麻醉劑毒性略大,因此,增加描述產(chǎn)生氫鍵能力的描述符(如最低空軌道能,ELUMO)后,模型的擬合度得到提升[58]。對于活性化合物和特殊反應(yīng)類化合物2類模型擬合效果并不理想,原因在于毒性作用復(fù)雜,包括的反應(yīng)機(jī)制很多,如活性化合物還包含邁克爾加成反應(yīng),親核取代反應(yīng)等反應(yīng)類型。另外,還有一些化合物無法根據(jù)Verhaar方法進(jìn)行分類,仍需進(jìn)一步的研究來使得化合物的分類規(guī)則更準(zhǔn)確和系統(tǒng)。Bearden和Schultz[54]將化合物分為非極性麻醉劑、極性麻醉劑、弱酸呼吸解偶聯(lián)劑、軟親電試劑、親電試劑,對黑頭呆魚和梨形四膜蟲進(jìn)行了模型構(gòu)建(表4)。雖然分類方法不盡相同,但模型的整體體現(xiàn)與基于Verhaar分類所構(gòu)建的模型很相似。對于非極性麻醉劑、極性麻醉劑采用log Kow就可以得到良好的模型,其他3類化合物需要引入其他描述符才能得到較優(yōu)的模型。
表3 毒性作用模式Table 3 Mode of action
表4 基于毒性作用模式的QSAR模型Table 4 QSAR models for mode of action
注:IGC50為半數(shù)生長抑制濃度。
Note: IGC50represents median inhibitory growth concentration.
目前,有些毒性作用模式分類方法,因其分類規(guī)則較復(fù)雜,在QSAR建模中的應(yīng)用還比較少。將毒性作用機(jī)制與其化合物結(jié)構(gòu)建立聯(lián)系會(huì)一定程度上提高其可操作性。但是,結(jié)構(gòu)相似的化合物毒性作用模式也會(huì)存在差別,如擁有相同的子結(jié)構(gòu)(substructure)的化合物仍會(huì)表現(xiàn)出不同的毒性機(jī)制。有研究者基于統(tǒng)計(jì)學(xué)方法從描述符角度對化合物毒性作用模式進(jìn)行表征和分類[48,59-60]。如Aptula等[59]基于結(jié)構(gòu)將221個(gè)酚類化合物分為四類,運(yùn)用逐步線性判別分析方法來表征不同描述符對各類別化合物的影響,以實(shí)現(xiàn)基于描述符來劃分具有不同毒性機(jī)制的化合物,劃分的準(zhǔn)確率可以達(dá)到86%~89%。盡管基于描述符進(jìn)行化合物分類具有較好的操作性,但仍有一些化合物不能準(zhǔn)確的區(qū)分。如Ren[60]運(yùn)用判別分析方法對化合物進(jìn)行分類,發(fā)現(xiàn)對極性麻醉劑和親電試劑區(qū)分比較困難。
有研究者嘗試運(yùn)用定量活性-活性關(guān)系(QAAR)對不同物種進(jìn)行分析建模。該方法需要根據(jù)兩種生物之間的實(shí)驗(yàn)數(shù)據(jù)建立相關(guān)性模型,以實(shí)現(xiàn)種間毒性數(shù)據(jù)的預(yù)測。研究不同物種的種間毒性關(guān)系,可以用一種生物的毒性來估計(jì)其他物種的毒性,尋找替代的測試生物以減少動(dòng)物實(shí)驗(yàn),同時(shí)還有助于認(rèn)識(shí)化合物的毒性作用機(jī)制。Zhang等[61]研究表明,海洋細(xì)菌和淡水細(xì)菌的種間關(guān)系較好,不同魚類之間也表現(xiàn)出較好的種間關(guān)系。另外,結(jié)構(gòu)相似的同類化合物對不同物種間也表現(xiàn)出較好的相關(guān)性,如有機(jī)磷酸酯類化合物對大型蚤和鯉魚[22],苯三唑類化合物對大型蚤和虹鱒魚[62],醛類對黑頭呆魚和梨形四膜蟲[63]都表現(xiàn)出較好的相關(guān)性。但對于多種類化合物,不同化合物對不同物種的毒性大小和作用機(jī)制不盡相同。Tremolada等[64]較系統(tǒng)的研究了不同生物種間的毒性關(guān)系,研究表明不同種類魚之間的毒性相關(guān)性較好;若將有特殊反應(yīng)的化合物(有機(jī)磷農(nóng)藥和氨基甲酸酯)除去后,在大型蚤與魚類之間也得到了較好的種間毒性關(guān)系;而綠藻與大型蚤和魚類之間在毒性效應(yīng)上表現(xiàn)出更多的差異。Lessigiarska等[65]對藻類、大型蚤和魚類種間關(guān)系做了研究,發(fā)現(xiàn)在斑馬魚和羊角月牙藻、大型蚤和彩虹魚之間模型的回歸系數(shù)分別為0.674和0.665,其余6個(gè)種間關(guān)系模型并沒有表現(xiàn)出較好的回歸系數(shù)。Zhang等[31]研究表明,在大型蚤和梨形四膜蟲之間,醇類、酮類、酯類等化合物的毒性機(jī)制種間較為一致,苯胺及其衍生物的種間毒性差異顯著。這說明雖然不同生物之間存在一些相似的毒性機(jī)制,但是這種種間的一致性并不具有通用性,應(yīng)用這種方法預(yù)測特別需要關(guān)注種間可能存在的差異性。
綜上,前人基于毒性作用模式建模取得一些進(jìn)展,其中水生生物急性毒性麻醉類化合物模型性能較好,而對具有反應(yīng)活性化合物的QSAR建模仍存在一定的困難。原因在于這類化合物毒性作用機(jī)制相互交叉,作用模式分類復(fù)雜。為了降低毒性作用模式分類的復(fù)雜性,研究者們采取了一些方法來進(jìn)行毒性作用模式分類和化合物毒性的預(yù)測,但都有各自的優(yōu)勢和局限。有關(guān)毒性分類方法有待進(jìn)一步的研究。例如,可以將不同毒性分類方法整合,保留其共性和各自的優(yōu)勢,并將毒性機(jī)制通過相關(guān)的分子結(jié)構(gòu)參數(shù)來進(jìn)行表征,提高毒性分類方法的可操作性和統(tǒng)一性,以便于在模型構(gòu)建中進(jìn)行應(yīng)用。
盡管目前已有較多水生生物急性毒性的QSAR模型,但很多模型存在自身的局限性,不太符合OECD提出的QSAR模型建立和使用準(zhǔn)則?;谕惢衔飿?gòu)建的模型,一般擬合能力較好,但存在應(yīng)用域較窄的限制?;跀?shù)理統(tǒng)計(jì)構(gòu)建的涵蓋多類化合物的模型,應(yīng)用域有所拓展,但會(huì)存在模型性能較差或算法不透明等問題?;谧饔媚J綄衔镞M(jìn)行分類建模,一方面具有較為明確的模型應(yīng)用域和較好的預(yù)測能力,同時(shí)易于進(jìn)行機(jī)理解釋,將是今后構(gòu)建水生生物急性毒性QSAR模型的主要方向。從目前所建的模型來看,麻醉毒性機(jī)制較為明確,模型擬合效果較好。過量毒性的毒性反應(yīng)機(jī)制非常復(fù)雜,分類方法不太統(tǒng)一,給過量毒性化合物的模型構(gòu)建帶來一定的挑戰(zhàn)。后續(xù)的研究需要在毒性作用機(jī)制研究的基礎(chǔ)上,將多種毒性作用分類方法進(jìn)行整合,以提高分類規(guī)則的可操作性和統(tǒng)一性,為過量毒性的化學(xué)物QSAR模型的構(gòu)建奠定基礎(chǔ)。
[1] 美國化學(xué)文摘社[OL]. [2014-10-30]. http://www.cas.org/
[2] European Parliament and Council Regulation (EC) No 1907/2006 Commission of the European Communities 2006. Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC and repealing Council Regulation (EEC) No 793/93 and Commission Regulation (EC) No 1488/94 as well as Council Directive 76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/EC. Off. J. Eur. Union, L396/1 of 30.12.2006 [R/OL]. [2014-10-30]. http://publications.europa.eu
[3] Verhaar H J M, Solbe J, Speksnijder J, et al. Classifying environmental pollutants: Part 3. External validation of the classification system [J]. Chemosphere, 2000, 40(8): 875-883
[4] Allanou R, Hansen B G, Van der Bilt Y. PublicAvailability of Data on EU High Production Volume Chemicals[M]. European Commission, 1999
[5] Strempel S, Scheringer M, Ng C A, et al. Screening for PBT Chemicals among the "Existing" and "New" Chemicals of the EU [J]. Environmental Science & Technology, 2012, 46(11): 5680-5687
[6] Taylor K, Stengel W, Casalegno C, et al. Experiences of the REACH testing proposals system to reduce animal testing [J]. ALTEX, 2014, 31(2): 107-128
[7] Rovida C, Hartung T. Re-evaluation of animal numbers and costs for in vivo tests to accomplish REACH legislation requirements for chemicals-A report by the transatlantic think tank for toxicology (t(4)) [J]. ALTEX, 2009, 26(3): 187-208
[8] Hartung T, Rovida C. Chemical regulators have overreached [J]. Nature, 2009, 460(7259): 1080-1081
[9] Daginnus K. Characterisation of the REACH pre-registered substances list by chemical structure and physicochemical properties [M]. European Commission, Joint Research Centre, 2010
[10] 陳景文, 李雪花, 于海瀛, 等. 面向毒害有機(jī)物生態(tài)風(fēng)險(xiǎn)評價(jià)的(Q)SAR技術(shù): 進(jìn)展與展望[J]. 中國科學(xué): B輯, 2008, 38(6): 461-474
Chen J W, Li X H, Yu H Y, et al. (Q)SAR technologies oriented to the ecological risk assessment of hazardous organic chemicals: Advances and perspective [J]. Science in China: B, 2008, 38(6): 461-474 (in Chinese)
[11] Tunkel J, Mayo K, Austin C, et al. Practical considerations on the use of predictive models for regulatory purposes [J]. Environmental Science & Technology, 2005, 39(7): 2188-2199
[12] OECD (Organisation for Economic Co-Operation and Development). Guideline document on the validation of (quantitative) structure-activity relationships [(Q)SAR] models. ENV/JM/MONO(2007)2. [R]. Paris: OECD, 2007
[13] Lyakurwa F S, Yang X H, Li X H, et al. Development of in silico models for predicting LSER molecular parameters and for acute toxicity prediction to fathead minnow (Pimephales promelas) [J]. Chemosphere, 2014, 108: 17-25
[14] Moosus M, Maran U. Quantitative structure-activity relationship analysis of acute toxicity of diverse chemicals to Daphnia magna with whole molecule descriptors [J]. SAR and QSAR in Environmental Research, 2011, 22(7-8): 757-774
[15] Roberts D W, Roberts J F, Hodges G, et al. Aquatic toxicity of cationic surfactants to Daphnia magna [J]. SAR and QSAR in Environmental Research, 2013, 24(5): 417-427
[16] Asadollahi-Baboli M. Aquatic toxicity assessment of esters towards the Daphnia magna through PCA-ANFIS [J]. Bulletin of Environmental Contamination and Toxicology, 2013, 91(4): 450-454
[17] 馮子雅, 馮長君. 鹵代苯對呆頭魚, 發(fā)光菌急性毒性的構(gòu)效關(guān)系研究[J]. 廣州化工, 2013, 41(6): 16-18
Feng Z Y, Feng C J. QSAR study on the acute toxicity of halogeno-benzene compounds to photobacterium phosphoreum and black pimephales by molecular shape indices [J]. Guangzhou Chemical Industry, 2013, 41(6): 16-18 (in Chinese)
[18] Ding F, Guo J, Song W H, et al. Comparative quantitative structure-activity relationship (QSAR) study on acute toxicity of triazole fungicides to zebrafish [J]. Chemistry and Ecology, 2011, 27(4): 359-368
[19] Furuhama A, Aoki Y, Shiraishi H. Development of ecotoxicity QSAR models based on partial charge descriptors for acrylate and related compounds [J]. SAR and QSAR in Environmental Research, 2012, 23(7-8): 731-749
[20] Al-Fahemi J H. The use of quantum-chemical descriptors for predicting the photoinduced toxicity of PAHs [J]. Journal of Molecular Modeling, 2012, 18(9): 4121-4129
[21] Cassani S, Kovarich S, Papa E, et al. Evaluation of CADASTER QSAR models for the aquatic toxicity of (Benzo)triazoles and prioritisation by consensus prediction [J]. Alternatives to Laboratory Animals: ATLA, 2013, 41(1): 49-64
[22] Zvinavashe E, Du T, Griff T, et al.Quantitative structure-activity relationship modeling of the toxicity of organothiophosphate pesticides to Daphnia magna and Cyprinus carpio [J]. Chemosphere, 2009, 75(11): 1531-1538
[23] Kamaya Y, Fukaya Y, Suzuki K. Acute toxicity of benzoic acids to the crustacean Daphnia magna [J]. Chemosphere, 2005, 59(2): 255-261
[24] Roy K, Das R N. QSTR with extended topochemical atom (ETA) indices. 14. QSAR modeling of toxicity of aromatic aldehydes to Tetrahymena pyriformis [J]. Journal of Hazardous Materials, 2010, 183(1): 913-922
[25] Schramm F, Muller A, Hammer H, et al. Epoxide and thiirane toxicity in vitro with the ciliates Tetrahymena pyriformis: Structural alerts indicating excess toxicity [J]. Environmental Science & Technology, 2011, 45(13): 5812-5819
[26] Zeng M, Lin Z F, Yin D Q, et al. A Kow-based QSAR model for predicting toxicity of halogenated benzenes to all algae regardless of species [J]. Bulletin of Environmental Contamination and Toxicology, 2011, 86(6): 565-570
[27] Huang C P, Wang Y J, Chen C Y. Toxicity and quantitative structure-activity relationships of nitriles based on Pseudokirchneriella subcapitata [J]. Ecotoxicology and Environmental Safety, 2007, 67(3): 439-446
[28] Jing G H, Zhou Z M, Zhuo J. Quantitative structure-activity relationship (QSAR) study of toxicity of quaternary ammonium compounds on Chlorella pyrenoidosa and Scenedesmus quadricauda [J]. Chemosphere, 2012, 86(1): 76-82
[29] Wang T, Zhou X H, Wang D L, et al. Using molecular docking between organic chemicals and lipid membrane to revise the well known octanol-water partition coefficient of the mixture [J]. Environmental Toxicology and Pharmacology, 2012, 34(1): 59-66
[30] Song W H, Guo J, Ding F, et al. Study on acute toxicity and structure-activity relationship of Daphnia magna exposed to naphthoquinones [J]. Environmental Toxicology and Pharmacology, 2011, 32(1): 102-106
[31] Zhang X J, Qin W C, He J, et al. Discrimination of excess toxicity from narcotic effect: Comparison of toxicity of class-based organic chemicals to Daphnia magna and Tetrahymena pyriformis [J]. Chemosphere, 2013, 93(2): 397-407
[32] Pavan M, Netzeva T I, Worth A P. Validation of a QSAR model for acute toxicity [J]. SAR and QSAR in Environmental Research, 2006, 17(2): 147-171
[33] In Y, Lee S K, Kim P J, et al. Prediction of acute toxicity to fathead minnow by local model based QSAR and global QSAR approaches [J]. Bulletin of the Korean Chemical Society, 2012, 33(2): 613-619
[34] Mazzatorta P, Smiesko M, Lo Piparo E, et al.QSAR model for predicting pesticide aquatic toxicity [J]. Journal of Chemical Information and Modeling, 2005, 45(6): 1767-1774
[35] Kar S, Roy K. QSAR modeling of toxicity of diverse organic chemicals to Daphnia magna using 2D and 3D descriptors [J]. Journal of Hazardous Materials, 2010, 177(1-3): 344-351
[36] Tao S, Xi X H, Xu F L, et al. A fragment constant QSAR model for evaluating the EC50values of organic chemicals to Daphnia magna [J]. Environmental Pollution, 2002, 116(1): 57-64
[37] Niculescu S P, Lewis M A, Tigner J. Probabilistic neural networks modeling of the 48 h- LC50acute toxicity endpoint to Daphnia magna [J]. SAR and QSAR in Environmental Research, 2008, 19(7-8): 735-750
[38] Panaye A, Fan B T, Doucet J P, et al. Quantitative structure-toxicity relationships (QSTRs): A comparative study of various non linear methods. General regression neural network, radial basis function neural network and support vector machine in predicting toxicity of nitro- and cyano- aromatics to Tetrahymena pyriformis [J]. SAR and QSAR in Environmental Research, 2006, 17(1): 75-91
[39] Singh K P, Gupta S. In silico prediction of toxicity of non-congeneric industrial chemicals using ensemble learning based modeling approaches [J]. Toxicology and Applied Pharmacology, 2014, 275(3): 198-212
[40] Jin X Q, Jin M H, Sheng L X. Three dimensional quantitative structure-toxicity relationship modeling and prediction of acute toxicity for organic contaminants to algae [J]. Computers in Biology and Medicine, 2014, 51: 205-213
[41] McFarland J W. Parabolic relation between drug potency and hydrophobicity [J]. Journal of Medicinal Chemistry, 1970, 13(6): 1192-1196
[42] Mckim J M, Bradbury S P, Niemi G J. Fish acute toxicity syndromes and their use in the QSAR approach to hazard assessment [J]. Environmental Health Perspectives, 1987, 71: 171-186
[43] Verhaar H J M, Vanleeuwen C J, Hermens J L M. Classifying Environmental-Pollutants .1. Structure-Activity-Relationships for Prediction of Aquatic Toxicity [J]. Chemosphere, 1992, 25(4): 471-491
[44] Verhaar H J M, Ramos E U, Hermens J L M. Classifying environmental pollutants .2. Separation of class 1 (baseline toxicity) and class 2 ('polar narcosis') type compounds based on chemical descriptors [J]. Journal of Chemometrics, 1996, 10(2): 149-162
[45] Enoch S J, Hewitt M, Cronin M T D, et al. Classification of chemicals according to mechanism of aquatic toxicity: An evaluation of the implementation of the Verhaar scheme in Toxtree [J]. Chemosphere, 2008, 73(3): 243-248
[46] Russom C L, Bradbury S P, Broderius S J, et al. Predicting modes of toxic action from chemical structure: Acute toxicity in the fathead minnow (Pimephales promelas) [J]. Environmental Toxicology and Chemistry, 1997, 16(5): 948-967
[47] Nendza M, Wenzel A. Discriminating toxicant classes by mode of action-1. (Eco)toxicity profiles [J]. Environmental Science and Pollution Research, 2006, 13(3): 192-203
[48] Nendza M, Muller M. Discriminating toxicant classes by mode of action: 2. Physico-chemical descriptors [J]. QSAR-Quantitative Structure Activity Relationships, 2000, 19(6): 581-598
[49] von der Ohe P C, Kuhne R, Ebert R U, et al. Structural alerts-A new classification model to discriminate excess toxicity from narcotic effect levels of organic compounds in the acute daphnid assay [J]. Chemical Research in Toxicology, 2005, 18(3): 536-555
[50] Nendza M, Muller M, Wenzel A. Discriminating toxicant classes by mode of action: 4. Baseline and excess toxicity [J]. SAR and QSAR in Environmental Research, 2014, 25(5): 393-405
[51] Lyakurwa F S, Yang X H, Li X H, et al. Development and validation of theoretical linear solvation energy relationship models for toxicity prediction to fathead minnow (Pimephales promelas) [J]. Chemosphere, 2014, 96: 188-194
[52] Papa E, Villa F, Gramatica P. Statistically validated QSARs, based on theoretical descriptors, for modeling aquatic toxicity of organic chemicals in Pimephales promelas (fathead minnow) [J]. Journal of Chemical Information and Modeling, 2005, 45(5): 1256-1266
[53] Katritzky A R, Tatham D B, Maran U. Theoretical descriptors for the correlation of aquatic toxicity of environmental pollutants by quantitative structure-toxicity relationships [J]. Journal of Chemical Information and Computer Sciences, 2001, 41(5): 1162-1176
[54] Bearden A P, Schultz T W. Structure-activity relationships for Pimephales and Tetrahymena: A mechanism of action approach [J]. Environmental Toxicology and Chemistry, 1997, 16(6): 1311-1317
[55] Worgan A D, Dearden J C, Edwards R, et al. Evaluation of a novel short-term algal toxicity assay by the development of QSARs and inter-species relationships for narcotic chemicals [J]. QSAR & Combinatorial Science, 2003, 22(2): 204-209
[56] Roberts D W, Costello J F. Mechanisms of action for general and polar narcosis: A difference in dimension [J]. QSAR & Combinatorial Science, 2003, 22(2): 226-233
[57] Pavan M, Worth A, Netzeva T. Preliminary Analysis of an Aquatic Toxicity Dataset and Assessment of QSAR Models for Narcosis [R]. European Commission, 2005
[58] Lee Y G, Hwang S H, Kim S D. Predicting the toxicity of substituted phenols to aquatic species and its changes in the stream and effluent waters [J]. Archives of Environmental Contamination and Toxicology, 2006, 50(2): 213-219
[59] Aptula A O, Netzeva T I, Valkova I V, et al. Multivariate discrimination between modes of toxic action of phenols [J]. Quantitative Structure-Activity Relationships, 2002, 21(1): 12-22
[60] Ren S. Determining the mechanisms of toxic action of phenols to Tetrahymena pyriformis [J]. Environmental Toxicology, 2002, 17(2): 119-127
[61] Zhang X J, Qin H W, Su L M, et al. Interspecies correlations of toxicity to eight aquatic organisms: Theoretical considerations [J]. Science of the Total Environment, 2010, 408(20): 4549-4555
[62] Cassani S, Kovarich S, Papa E, et al. Daphnia and fish toxicity of (benzo) triazoles: Validated QSAR models, and interspecies quantitative activity-activity modelling [J]. Journal of Hazardous Materials, 2013, 258: 50-60
[63] Dimitrov S, Koleva Y, Schultz T W, et al. Interspecies quantitative structure-activity relationship model for aldehydes: Aquatic toxicity [J]. Environmental Toxicology and Chemistry, 2004, 23(2): 463-470
[64] Tremolada P, Finizio A, Villa S, et al. Quantitative inter-specific chemical activity relationships of pesticides in the aquatic environment [J]. Aquatic Toxicology, 2004, 67(1): 87-103
[65] Lessigiarska I, Worth A, Sokull-Klüttgen B, et al. QSAR investigation of a large data set for fish, algae and Daphnia toxicity [J]. SAR and QSAR in Environmental Research, 2004, 15(5-6): 413-431
◆
Progress in Quantitative Structure-Activity Relationship Models for Acute Aquatic Toxicity
Liu Yuchen, Qiao Xianliang*
Key Laboratory of Industrial Ecology and Environmental Engineering of Ministry of Education, Department of Environmental Science and Technology, Dalian University of Technology, Dalian 116024,China
22 November 2014 accepted 5 January 2015
Chemical contaminations lead potential risks to both human health and ecological environment. However, the lack of available data on the hazardous properties of chemicals is the major challenge for the risk assessment of chemicals. Non-animal alternative methods are encouraged to fill in data gaps by OECD and US EPA. Quantitative structure-activity relationship (QSAR) approach is regarded as one promising alternative technique. Information on acute toxicity to aquatic organisms are commonly used in the risk assessment and screening of priority substances. But, the available experimental toxicity data are very limited currently. In this paper, three types of prediction models of acute toxicity are summarized, including (1) models for particular chemical classes; (2) statistically derived models that are developed without an a priority mechanistic hypothesis; (3) models for a given mode or mechanism of action (MOA). The predictive ability, applicability domain and mechanism interpretation of the three type models are compared. QSAR models based on MOA, which generally demonstrate rather good predictive ability and facilitate the interpretation of mechanism meanwhile, will become the main trend for predicting acute toxicity to aquatic organisms.
QSAR; acute aquatic toxicity; mode of action
國家科技部863課題(2012AA06A301);國家自然科學(xué)基金面上項(xiàng)目(21277018);中央高校基本科研業(yè)務(wù)費(fèi)專項(xiàng)(DUT14ZD213)
劉羽晨(1991-),女,碩士,研究方向?yàn)槲廴旧鷳B(tài)化學(xué),E-mail: 286871921@qq.com;
*通訊作者(Corresponding author), E-mail: xlqiao@dlut.edu.cn
10.7524/AJE.1673-5897.20141122001
2014-11-22 錄用日期:2015-01-05
1673-5897(2015)2-26-10
X171.5
A
喬顯亮(1974-),男,博士,副教授,研究方向?yàn)槲廴旧鷳B(tài)化學(xué)。
劉羽晨, 喬顯亮. 水生生物急性毒性QSAR模型研究進(jìn)展[J]. 生態(tài)毒理學(xué)報(bào), 2015, 10(2): 26-35
Liu Y C, Qiao X L. Progress in quantitative structure-activity relationship models for acute aquatic toxicity [J]. Asian Journal of Ecotoxicology, 2015, 10(2): 26-35 (in Chinese)