基于分子描述符和机器学习方法预测和虚拟筛选乳腺癌靶向蛋白HEC1抑制剂

HEC1(癌症高表达蛋白)是纺锤体检查点控制、着丝粒功能、细胞存活的关键的有丝分裂调节器,与原发性乳腺癌的不良预后有关.筛选具有高亲和力的HEC1新型抑制剂对探索乳腺癌的靶向治疗具有重要意义.本文从结构多样性的化合物库中筛选HEC1抑制剂.通过对分子描述符的特征筛选,采用支持向量机(SVM)和随机森林(RF)方法分别对HEC1抑制剂和非抑制剂建立了分类模型.经对比, RF模型显示了更好的预测精度.我们采用RF模型对HEC1抑制剂进行了虚拟筛选,从“in-house”实体库筛选得到2个潜在的HEC1抑制剂分子.随后对筛出的化合物进行了体外活性实验,发现对乳腺癌细胞株MDA-MB-468和MDA-MB-231均有一定程度的抗肿瘤活性.研究结果表明,机器学习方法对于设计和虚拟筛选HEC1抑制剂有良好的效果.     

基于四氢喹啉酸骈环戊烯骨架的蛋白酪氨酸磷酸酶1B抑制剂的设计、合成及构效关系

基于肿瘤、糖尿病等重大疾病的有限目标,选择蛋白酪氨酸磷酸酶(PTPs)家族这一生物体系中有代表性的PTP家族成员PTP1B,SHP-1,SHP-2,LAR,CDC25B及PRL-3进行研究.通过综合高通量筛选获得的"苗头"化合物结构信息,分析得到四氢喹啉骈环戊烯骨架.初步的构效关系研究表明,四氢喹啉酸骈环戊烯母核结构可...     

基于虚拟筛选和分子动力学模拟发现新型ULK1抑制剂

Unc-51样自噬激活激酶1（unc-51-like autophagy activating kinase 1,ULK1）作为自噬启动的重要调控因子,是肿瘤治疗的关键靶点之一。首先,以已知ULK1抑制剂为基础构建药效团模型,通过药效团模型筛选、分子对接以及分子力学广义波恩表面积（Molecular Mechanics/Generalized Born Surface Area,MM/GBSA）结合自由能计算等方法,对含有52万多个类药性小分子的数据库进行虚拟筛选,得到具有较高理论亲和力的化合物。随后,50ns的分子动力学模拟验证了蛋白质-配体复合物结合的稳定性,最后10ns的平均结合自由能的计算研究进一步验证了配体的结合能力。结果表明,6个化合物（F5258-0159、F3407-0428、F0529-1100、F0696-3531、F3222-5280、F6525-5596）具有骨架新颖、分子对接分数和结合自由能数值优异及与ULK1的结合状态稳定等特点,可以作为新型潜在的ULK1抑制剂用于肿瘤治疗的研究,也为新型ULK1抑制剂的设计和研发提供新的研究思路。     

天冬酰胺合成酶B抑制剂高效液相色谱筛选方法的建立与应用

建立了一种快速、高效测定天冬酰胺合成酶B(AS-B)酶活性的反相高效液相色谱法(RP-HPLC)。酶反应体系中的氨基酸经2,4-二硝基氟苯(DNFB)柱前衍生,通过RP-HPLC测定酶反应体系前后底物及产物的变化来分析酶的活性。采用的色谱柱为Agilent C18柱(250 mm×4.6 mm,5 μm),以50 mmol/L醋酸钠缓冲液(pH 6.2)-乙腈(体积比为15:85)为流动相,流速为1.0 mL/min,柱温为30 ℃,检测波长365 nm,于6 min内实现了各组分的基线分离。通过该方法测定反应动力学参数进行AS-B的抑制定量分析。将已知AS-B抑制剂L-谷氨酸-γ-甲酯作用于酶反应体系,测得的抑制剂的抑制常数与文献值相接近,证明该方法可用于AS-B抑制剂的筛选。     

基于高分辨质谱的甲基化心磷脂高通量分析方法的建立和应用

该文基于心磷脂标准品的甲基化反应,建立了高灵敏度高通量分析甲基化心磷脂的液相色谱 - 高分辨质谱方法。由于甲基化后的心磷脂具有特征性的碎片,因此建立了包含780个分子特征片段的模拟数据库用于脂质鉴定。衍生化后,CL（14∶0/14∶0/14∶0/14∶0）信号提高了10倍,定量下限（LOQ）达到10 ng/mL。将该方法应用于小鼠衰老过程中心磷脂的调控研究,通过Tracefinder本地二级数据库共鉴定了43个心磷脂分子,其中21个心磷脂在衰老的大脑中有差异变化。结果发现：含有FA（20∶4）、FA（22∶6）长链多不饱和脂肪酸的心磷脂在衰老过程中呈下降趋势。高分辨质谱平台为甲基化的心磷脂分析提供了一种高通量筛选方法,对于低丰度心磷脂的检测具有很大潜力。     

基于分子描述符和机器学习方法预测和虚拟筛选MMP-13对MMP-1的选择性抑制剂

基质金属蛋白酶-13(MMP-13)为预防和治疗骨关节炎(OA)提供了充满希望的靶标.通过抑制剂来阻断MMP-13的活性将会对治疗OA疾病产生潜在的作用.然而,宽谱抑制剂同样抑制MMP家族的其它成员,特别是MMP-1,这将会导致肌与骨的综合症.因此,设计和发现潜在的MMP-13相对于MMP-1的高效选择性抑制剂,在对治疗OA新型药物的研发中具有相当重要的现实意义.本研究通过两种机器学习方法(ML):支持向量机(SVM)和随机森林(RF)来建立分类模型,用于预测不同结构的MMP-13对MMP-1的选择性抑制剂.所建这些模型的预测效果都已经达到了令人满意的精度.在这两种ML模型中,RF对于MMP-13选择性抑制剂和非抑制剂的精度分别达到97.58%和100%.同时,与MMP-13对MMP-1的选择性抑制最相关的分子描述符也基于不同的特征选择方法被两种模型挑选出来.最后,用预测效果最好的RF模型虚拟筛选了ZINC数据库的"fragment-like"子集,从而得到了一系列潜在的候选药物.研究表明,机器学习方法,特别是RF方法,对于发现潜在的MMP-13选择性抑制剂十分有效.同时还得到了一些与MMP-13的选择性抑制相关的分子描述符.     

荧光光度法二氢叶酸还原酶抑制剂筛选模型的建立和应用

本文应用二氢叶酸(DHFA)和辅酶Ⅱ(NADPH)的荧光性质,建立了二氢叶酸还原酶(DHFR)活力测定的反应体系,并对酸度等反应条件进行了优化。在该体系中,用双倒数作图法分别测定了二氢叶酸还原酶对二氢叶酸和辅酶Ⅱ的表观米氏常数。将所建立的体系应用于氨甲喋呤,甲氧苄氨嘧啶等已知抑制剂的抑制率的测定,结果满意。     

基于分子描述符和机器学习方法预测和虚拟筛选MMP-13对MMP-1的选择性抑制剂

基质金属蛋白酶-13 (MMP-13)为预防和治疗骨关节炎(OA)提供了充满希望的靶标. 通过抑制剂来阻断MMP-13的活性将会对治疗OA疾病产生潜在的作用. 然而,宽谱抑制剂同样抑制MMP家族的其它成员,特别是MMP-1,这将会导致肌与骨的综合症. 因此,设计和发现潜在的MMP-13 相对于MMP-1 的高效选择性抑制剂,在对治疗OA新型药物的研发中具有相当重要的现实意义. 本研究通过两种机器学习方法(ML)：支持向量机(SVM)和随机森林(RF)来建立分类模型,用于预测不同结构的MMP-13 对MMP-1 的选择性抑制剂. 所建这些模型的预测效果都已经达到了令人满意的精度. 在这两种ML模型中,RF对于MMP-13选择性抑制剂和非抑制剂的精度分别达到97.58%和100%. 同时,与MMP-13对MMP-1的选择性抑制最相关的分子描述符也基于不同的特征选择方法被两种模型挑选出来. 最后,用预测效果最好的RF模型虚拟筛选了ZINC数据库的“fragment-like”子集,从而得到了一系列潜在的候选药物. 研究表明,机器学习方法,特别是RF方法,对于发现潜在的MMP-13选择性抑制剂十分有效. 同时还得到了一些与MMP-13的选择性抑制相关的分子描述符.     

单分子力谱研究活细胞上多药耐药相关蛋白1的表达

采用原子力显微镜的单分子力谱(SMFM)技术研究了多药耐药相关蛋白1(MRP1)与其抗体间的相互作用, 并考察了人舌癌细胞系TCA8113经高剂量平阳霉素(BLM)反复间歇诱导前后细胞表面MRP1的表达差异. 实验结果表明, MRP1与其抗体之间存在特异性相互作用力, 当针尖运动速率为2.5 μm/s时, 作用力大小约为(182±35) pN; 而且药物诱导后MRP1在人舌癌细胞上的表达明显增强. 本工作为了解活细胞水平上MRP1的表达提供了新方法, 有助于肿瘤细胞多药耐药性(MDR)的研究.     

基于计算机辅助药物设计的VEGFR2和PD-L1双靶点抑制剂的筛选

肝细胞癌(HCC)是世界上最具侵袭性的实体恶性肿瘤之一,尽管近几十年来在肝切除术、射频消融和肝移植等方面取得了显著进展,但HCC患者的愈后仍不乐观,主要原因是HCC的高复发率和转移率以及该疾病对化疗和靶向治疗的耐受性。VEGFR2和PD-L1抑制剂联合应用能对小鼠体内的HCC肿瘤进行显著抑制并延长其寿命,因此开发VEGFR2与PD-L1双靶点抑制剂对HCC的治疗具有十分重要的意义。本文基于计算机辅助药物设计(CADD),经过以分子对接为基础的虚拟筛选、MM-GBSA和ADMET评价以及QSAR模型预测,得到了一系列VEGFR2和PD-L1的双靶点活性化合物,有望为后续抑制剂的开发提供思路和方向。     

A whole-cell assay for the high throughput screening of calmodulin antagonists

Cell-based screening systems for pharmaceuticals are desired over molecular biosensing systems because of the information they provide on toxicity and bioavailability. However, the majority of sensing systems developed are molecular biosensing type screening systems and cannot be easily adapted to cell-based screening. In this study, we demonstrate that protein-based molecular sensing systems that employ a fluorescent protein as a signal transducer are amenable to cell-based sensing by expressing the protein molecular sensing system in the cell and employing these cells for screening of desired molecules. To achieve this, we expressed a molecular sensing system based on the fusion protein of calmodulin (CaM) and enhanced green fluorescent protein (EGFP) in bacterial cells, and utilized these cells for the screening of CaM antagonists. In the presence of Ca²⁺, CaM undergoes a conformational change exposing a hydrophobic pocket that interacts with CaM-binding proteins, peptides, and drugs. This conformational change induced in CaM leads to a change in the microenvironment of EGFP, resulting in a change in its fluorescence intensity. The observed change in fluorescence intensity of EGFP can be correlated to the concentration of the analyte present in the sample. Dose-response curves for various tricyclic antidepressants were generated using cells containing CaM-EGFP fusion protein. Additionally, we demonstrate the versatility of our system for studying protein-protein interactions by using cells to study the binding of a peptide to CaM. The study showed that the CaM-EGFP fusion protein within the intact cells responds similarly to that of the isolated fusion protein, hence eliminating the need for any isolation and purification steps. We have demonstrated that this system can be used for the rapid screening of various CaM antagonists that are potential antipsychotic drugs.     

Homogeneous single-label tyrosine kinase activity assay for high throughput screening

Protein post-translational modifications (PTMs) are regulatory mechanisms carried out by different enzymes in a cell. Kinase catalyzed phosphorylation is one of the most important PTM affecting the protein activity and function. We have developed a single-label quenching resonance energy transfer (QRET) assay to monitor tyrosine phosphorylation in a homogeneous high throughput compatible format. Epidermal growth factor receptor (EGFR) induced phosphorylation was monitored using Eu³⁺-chelate labeled peptide and label-free phosphotyrosine specific antibody in presence of a soluble quencher molecule. In the QRET kinase assay, antibody binding to phosphorylated Eu³⁺-peptide protects the Eu³⁺-chelate from luminescence quenching, monitoring high time-resolved luminescence (TRL) signals. In the presence of specific kinase inhibitor, antibody recognition and Eu³⁺-chelate protection is prevented, allowing an efficient luminescence quenching. The assay functionality was demonstrated with a panel of EGFR inhibitors (AG-1478, compound 56, erlotinib, PD174265, and staurosporine). The monitored IC₅₀ values ranged from 0.08 to 155.3 nM and were comparable to those found in the literature. EGFR activity and inhibition assays were performed using low nanomolar enzyme and antibody concentration in a 384-well plate format, demonstrating its compatibility for high throughput screening (HTS).     

10-23 DNA enzymes for targeting MDR1 mRNA

The properties of the 31-mer 10-23 DNA enzyme and its analog with the terminal 3"-3" internucleotide linkage, which are complementary to an mRNA region of the multi-drug resistance gene MDR1, were investigated. DNA enzymes can selectively cleave RNA with high efficiency in a catalytic mode as exemplified by a synthetic 18-mer fragment of MDR1 mRNA.     

The use of proton transfer reaction mass spectrometry for high throughput screening of terpene synthases

In this work, we introduce the application of proton transfer reaction mass spectrometry (PTR-MS) for the selection of improved terpene synthase mutants. In comparison with gas chromatography mass spectrometry (GC-MS)-based methods, PTR-MS could offer advantages by reduction of sample preparation steps and analysis time. The method we propose here allows for minimal sample preparation and analysis time and provides a promising platform for the high throughput screening (HTS) of large enzyme mutant libraries. To investigate the feasibility of a PTR-MS-based screening method, we employed a small library of Callitropsis nootkatensis valencene synthase (CnVS) mutants. Bacterial cultures expressing enzyme mutants were subjected to different growth formats, and headspace terpenes concentrations measured by PTR-Qi-ToF-MS were compared with GC-MS, to rank the activity of the enzyme mutants. For all cultivation formats, including 96 deep well plates, PTR-Qi-ToF-MS resulted in the same ranking of the enzyme variants, compared with the canonical format using 100 mL flasks and GC-MS analysis. This study provides a first basis for the application of rapid PTR-Qi-ToF-MS detection, in combination with multi-well formats, in HTS screening methods for the selection of highly productive terpene synthases.     

A time-resolved fluorescence probe for dipeptidyl peptidase 4 and its application in inhibitor screening

The prevalence of type 2 diabetes is increasing dramatically throughout the world. Recently, dipeptidyl peptidase 4 (DPP4) was identified as a potential antidiabetes target. Many DPP4 inhibitors, such as sitagliptin and vildagliptin, have been developed and marketed, but superior therapeutic agents are still required. Therefore, we have developed new methodology for screening of DPP4 inhibitors. Absorption-based measurements with para-nitroaniline or fluorescence-based measurements with the coumarin derivative 7-amino-4-methylcoumarin are often used for the screening of protease inhibitors, including DPP4 inhibitors, but these strategies are not sufficiently sensitive because of interfering background absorption and fluorescence, thus giving rise to many false-positive and false-negative results. Therefore, we have designed and synthesised a novel DPP4 probe (Gly-Pro-BCD-Tb; Gly=glycine, Pro=proline, andBCD defines the backbone of the probe comprising an aniline derivative as on/off switch, a 7-amino-4-methyl-2(1H)-quinolinone (cs-124) as antenna moiety, and a diethylenetriamine-N,N,N',N',N'-pentaacetic acid (DTPA) as chelator moiety, Tb=terbium) for time-resolved fluorescence (TRF) measurements. TRF measurements with Gly-Pro-BCD-Tb showed high sensitivity and reliability in the inhibitory assay relative to Gly-Pro-MCA (MCA=4-methylcoumarin-7-amide), a conventional fluorescence probe for DPP4. Further, we employed our probe for high-throughput DPP4 inhibitor screening with 3841 randomly selected compounds and found that epibestatin, an epimer of bestatin (a well-known anticancer drug and general aminopeptidase inhibitor), showed dose-dependent DPP4 inhibitory activity. Interestingly, bestatin did not exhibit DPP4 inhibitory activity. We believe that this screening system will be useful for the discovery of DPP4 inhibitors with novel structural scaffolds.     

Suppression of MDR1 gene expression by chemically modified siRNAs

The ability of chemically modified siRNAs targeted to MDR1 mRNA to inhibit P-glycoprotein expression and to restore sensitivity of cancer cells to antibiotic vinblastine was investigated. The effects of chemical modifications on RNA stability in cell culture medium and inhibition of MDR1 gene expression were tested. We found that siRNAs containing 2′-O-methyl ribonucleotides within either sense or/and antisense strands display high stability in serum but exhibit a significant reduction in the biological activity. The protection of 3′-ends of siRNA by introduction of 3′-3′-inverted phosphodiester bonds and two 2′-O-methyl ribonucleotides in protruding 3′-ends considerably increase their biological activity, which allows a 30-fold decrease in the cytostatic agent concentration required for cancer cell death. The data obtained demonstrate that the chemically modified siRNAs can be considered as potential therapeutics, which enhances the efficiency of cancer chemotherapy. Published in Russian in Izvestiya Akademii Nauk. Seriya Khimicheskaya, No. 7, pp. 1227–1235, July, 2006.     

High throughput screening for rapid development of membranes and membrane processes

The objective of this study was to investigate the feasibility of high throughput (HT) screening techniques for pressure-driven membrane processes. For this purpose, a HT-filtration module, allowing to perform 16 pressure-driven separations simultaneously, was designed. The potential of the developed equipment and of the HT-screening concept in general was validated by demonstrating both the reproducibility of experimental flux and selectivity data, and the scalability of these data between the HT-module and a conventional dead-end filtration set-up. Data were obtained with two solvent resistant nanofiltration (SRNF) membranes: a laboratory-prepared polyimide (PI) and a commercial MPF-50 membrane. The reproducibility of the data was highly encouraging, proving that this HT-approach can be a useful tool to rapidly screen a large array of operational parameters in membrane processes and of synthesis parameters in the development of new membranes.