搜索

Virtual Screening of Transmembrane Serine Protease Inhibitors
跨膜丝氨酸蛋白酶抑制剂的虚拟筛选   

下载 PDF 引用 收藏 提问与回复 分享您的反馈 Cited by

本文章节

Abstract

The human family of type II transmembrane serine proteases includes 17 members. The defining features of these proteases are an N-terminal transmembrane domain and a C-terminal serine protease of the chymotrypsin (S1) fold, separated from each other by a variable stem region. Recently accumulated evidence suggests a critical role for these proteases in development of cancer and metastatic capacity. Both the cancer relevance and the accessibility of the extracellularly oriented catalytic domain for therapeutic and imaging agents have fueled drug discovery interest in the type II class of transmembrane serine proteases. Typically, the initial hit discovery processes aim to identify molecules with verifiable activity at the drug target and with sufficient drug-like characters. We present here protocols for structure-based virtual screening of candidate ligands for transmembrane serine protease hepsin. The methods describe use of the 3D structure of the catalytic site of hepsin for molecular docking with ZINC, which is a molecular database of > 30 million purchasable compounds. Small candidate subsets were experimentally tested with demonstrable hits, which provided meaningful cues of the ligand structures for further lead development.

Keywords: Molecular modeling(分子建模), Small-molecules(小分子), Molecular docking(分子对接), Cancer(癌症), Drug discovery(药物探索)

Background

Controlled proteolytic activity plays a fundamental role in cellular processes and signaling, as evidenced by the presence of proteases in all organisms, including viruses, prokaryotes and eukaryotes. Not surprisingly, aberrantly regulated protease activity is causal to wide variety of human pathologies such as cardiovascular and inflammatory diseases, osteoporosis, neurological disorders and cancer (Turk, 2006; Bachovchin and Cravatt, 2012). In particular, development of primary cancer and metastatic capacity has been linked to several different classes of proteases including matrix metalloproteinases (MMPs), cysteine proteases (cathepsins) and membrane-associated serine proteases (Lopez-Otin and Matrisian, 2007). Recent clinical, genetic and functional data, suggesting a critical role for membrane-associated serine proteases in solid cancers, including cancer of prostate, ovarian and breast, have prompted new interest in the development of small molecule serine protease inhibitors for the treatment of cancer. Hepsin is a type II transmembrane serine protease and an attractive target for serine protease drug development due to frequent hepsin overexpression in common solid cancers, such as prostate and breast cancer, confinement of its overexpression on the membranes of cancer cells and due to positioning of the catalytic domain to the extracellular, i.e., more reachable, side of the cells (Antalis et al., 2010).

We describe here protocols for structure-based virtual screening of serine protease inhibitors using the catalytic site of hepsin for docking with drug-like subset of ZINC database. On the basis of virtual screening results, altogether 24 candidate compounds were purchased for further biochemical validation. Cell-free ELISA-based fluorogenic enzymatic assay using recombinant hepsin and fluorogenic peptide substrate Boc-Gln-Arg-Arg-AMC (BACHEM) was used for experimental validation and 30% inhibition of peptidolytic activity was set as threshold. Cut-off value was based on a notion that with the initial high micromolar concentration of compounds more than 30% inhibition would be required to determine reasonable IC50 value (Goswami et al., 2015; Tervonen et al., 2016) (Figures 1A and 1B). With these criteria, 3 out of 24 tested compounds showed inhibition potency (Tervonen et al., 2016). In conclusion, even though structure-based virtual screening is often considered as a complementary drug screening approach, the protocols reported here allowed us to demonstrate the feasibility and provided meaningful structural scaffolds for further development of specific serine protease inhibitors. However, it is important to stress that virtual screening hits typically do not demonstrate high potency, which is also true to the present screen. High-affinity ligands typically become available only after skillful medicinal chemistry optimization of selected hit structures.


Figure 1. Schematic figure illustrating the workflow. A. Virtual screening (compound) library is prepared and the target protein structure analyzed and prepared. The virtual screening is carried out by docking (Glide SP/XP) and the putative binders are validated by in vitro biochemical assays. B. Examples of candidate ligands (a-d).

The virtual screening protocol presented here is a modified version from the original protocols described in our recent research paper (Tervonen et al., 2016). The modifications make the protocol fully compatible with the most updated versions of ZINC database and Schrödinger software. While the current virtual screening protocol is tailored for users of Schrödinger software and ZINC database, the protocol is versatile and also compatible with other molecular modeling packages. However, implementation of the present virtual screening protocol with other modeling packages would require a careful platform-specific validation of the docking settings. As a guidance, we provide a general overview of the validation procedures used in the present screen.

Materials and Reagents

  1. ViewPlate-96 black assay plates (PerkinElmer, catalog number: 6005225 )
  2. Recombinant human hepsin protein (R&D Systems, catalog number: 4776-SE-010 )
  3. DMSO (Sigma-Aldrich, catalog number: D8418 )
  4. Fluorogenic peptide substrate Boc-Gln-Arg-Arg-AMC (Bachem, catalog number: I-1655.0025 )
  5. Small-molecule compounds were purchased either from Asinex Corporation (Winston-Salem, NC, http://www.asinex.com) or MolPort (Riga, Latvia, https://www.molport.com/shop/index)
  6. WX-UK1 was a kind gift from Dr. Ramachandra (Aurigene Discovery Technologies Limited, Bangalore, India), however, it is also commercially available (for example, AURUM Pharmatech, catalog number: Z-3200 )
  7. Trizma base (Tris) (Sigma-Aldrich catalog number: T1503 )
  8. Calcium chloride (CaCl2) (Sigma-Aldrich, catalog number: 499609 )
  9. Brij-35 (Brij L23 30% solution in H2O) (Sigma-Aldrich, catalog number: B4184 )
  10. Sodium chloride (NaCl) (Sigma-Aldrich, catalog number: 31434-M )
  11. Activation buffer for recombinant human hepsin protein (see Recipes)
  12. Assay buffer for recombinant human hepsin protein (see Recipes)

Equipment

  1. ELISA plate reader FLUOstar Omega (BMG LABTECH, model: FLUOstar Omega )
  2. Computer hardware provided by CSC - IT Center for Science Ltd (Espoo, Finland)
    1. HP Proliant SL230s (Hewlett Packard Development Company, model: SL230s )
    2. Intel Xeon E5-2670 2.6GHz processors (100 cores with 16 GB memory per core) (Intel Corporation, model: E5-2670 )

Software

  1. Modeling software
    Schrödinger Suite (Schrödinger, LLC, New York, NY) with Protein Preparation Wizard (Sastry et al., 2013), LigPrep, MacroModel and Glide (Friesner et al., 2004 and 2006; Halgren et al., 2004) modules, and a protocol version 2016-2 (https://www.schrodinger.com/suites/small-molecule-drug-discovery-suite)
  2. Databases
    1. Screening and validation: ZINC database (downloadable via http://zinc15.docking.org/). ZINC database includes over 100 million purchasable compounds in 3D formats (Sterling and Irwin, 2015)
    2. Protein structure database (http://www.rcsb.org/pdb/home/home.do)

Procedure

  1. Preparation of the target protein structure for virtual screening
    The hepsin protein structure (PDB id 1O5E, from protein structure database, www.rcsb.org, see ‘Online tools’) can be automatically downloaded with Protein Preparation Wizard of Schrödinger suite using ‘biological unit’-option. With other types of softwares one should use the manual downloading and read in the structure by software preferred method. The pre-processing is carried out with default methods except keeping all the water molecules. H-bond refinement should be carried out with default pH value 7. All water molecules with at least 3 H-bonds with non-water atoms are saved and the whole structure is minimized with default settings with heavy atoms converged at RMSD 0.3Å. The Glide Grid file is created using the above mentioned structure and the center of the Grid was at Tyr94 (Figure 2). The center of the GRID is referring to the center of molecular field calculation during the GRID creation. Hydroxyl groups of Tyr94, Tyr146, Ser195 and Tyr228 are allowed to rotate for Grid generation. Addition of free rotatable bonds doubles the Glide calculation time and should be limited to keep the computational time at reasonable level. Also, it is strongly recommended that the Glide Grid file will be constructed so that OPLS3 force field is used, if possible. If another software is used, one should use the corresponding protein preparation method. The critical point is to make sure that the structural quality of the protein is checked, proper ionization status are fixed for polar sidechains and right rotamers for (pseudo)symetric residues are used.


    Figure 2. Definition of GRID region for docking. The center of the GRID is based on the location of residues TYR94, TYR146 and SER195 (residues marked with labels) and the GRIF region is shown with magenta box.

  2. Preparation of the virtual screening compound library
    The virtual screening docking library must be prepared using the methods and settings recommended by the authors of the virtual screening software. When choosing the virtual compound library, one should take into account the number of compounds intended for screening, possibilities for own synthesis and if the resulting hits are intended to be used as such or further modified in subsequent medicinal chemistry programs. We use the drug-like subset of ZINC library, downloaded as a SDF-file from the ZINC web page. SDF-file format is supported by all the major software brands. The SDF-file is prepared by Schrödinger LigPrep module with default settings (Figure 3). In short the preparation includes checking of 2D structure quality, fast 2D-3D transformation, analysis and modification of protonation state, modification of all reachable tautomers and further minimization of resulting molecules. In a typical case each molecule will be represented with 2-4 different ionization/protomer/tautomer states. For other types of software other than the Schrödinger software, we recommend ligand setup via LigPrep. In our own experience, the benefit of this method is that it yields high quality predictions for both ionization status (i.e., pKa) and also for tautomers. The prediction of tautomers option is not included in most of the other software packages.


    Figure 3. LigPrep window under Schrödinger Suite. The library for virtual screening is prepared according to the settings selected; the shown setting are the default ones in our protocol.

  3. Virtual screening and analysis of the results
    In our study (Tervonen et al., 2016) virtual screening was carried out as follows:
    1. SP-docking is performed using the above-defined Grid file and the best 10% of compounds are further redocked with XP settings.
      1. In the first stage VHTS-settings were used and then SP settings were utilized only for the best 50,000-100,000 compounds (ranked according to the Glide-scoring function) and further, for the best 10%, XP-settings.
      2. In the original study OPLS2005 force field was used. However, we now strongly recommend OPLS3 force field, since it gives much better results compared to the original method.
    2. Clustering compounds was initially performed by visually studying the resulting docking poses.
      1. The aim of the visual inspection was to make sure that only reasonable docking poses are accepted. Typically some 10-20% of docking poses are unrealistic due to missing H-bonding contacts or unrealistic ligand conformations and these can be effectively screened by visual inspection.
    3. Compounds were then clustered by using the Interaction Fingerprint script available within Schrödinger.
      1. This script analyzed each of the docking pose based on the interaction profile (interaction fingerprints) and the poses are clustered by using those profiles. In our study we used the optimal number of clusters as evaluated by the script with default linkage method. For a detailed description of the method and its uses, see Deng et al. (2004) and Singh et al. (2006).
    4. As a result 24 clusters were created.
    5. In each cluster the most representative compounds are located in the center of the cluster. Those compounds are selected for in vitro assays.

  4. Validation of screen results by in vitro cell-free peptide cleavage assay for hepsin enzymatic activity protocols
    1. Recombinant human hepsin (rhHepsin) is diluted to 100 µg/ml in activation buffer.
    2. rhHepsin is activated by incubating the solution at 37 °C for 24 h.
    3. rhHepsin is diluted into 0.1 nM solution with assay buffer just before use.
    4. Small-molecule compounds are dissolved in DMSO to give 10 mM stock solution.
    5. The assay is performed in assay buffer in a 96-well plate.
    6. 0.1 nM final concentration of rhHepsin and 10 µM final concentration of small-molecule compound (and equal volume of DMSO control) in 100 µl reaction volume are incubated for 30 min at room temperature. Assay buffer is used as blank control.
    7. The reaction is started by adding a final concentration of 30 µM peptide substrate BOC-Gln-Arg-Arg-AMC.
    8. The plate is analyzed with ELISA plate reader by 350 nm excitation and 450 nm emission capture at room temperature.
    9. The inhibition % is determined by using the following formula:


Data analysis

The described virtual screening protocol resulted in hundreds of potential hits. Using the protocols described above and in Tervonen et al. (2016), we performed an initial in vitro cell-free validation. More than 30% reduction in peptide cleavage relative to DMSO control with three technical replicates and two repeats was considered as a cut-off for hits. We ordered and tested 24 compounds identified by virtual screening as potential hits (Figure 1B) and few compounds made it close to the cut-off value (see Tervonen et al. [2016]). These compounds were used as skeletons for the next round of virtual screening, which is a reiterative process. In the study Tervonen et al. (2016) we chose to use WX-UK1, which was identified in parallel screen, because its binding pose against hepsin and superb inhibition efficiency in comparison to other tested molecules.

Notes

  1. Computer hardware
    In addition to listed hardware above, we note that any high-end PC computer with any of the major operating system (Windows, Mac or Linux) and with either Intel or AMD processors is suitable for the virtual screening job but the process will then require more time.
  2. Small-molecule compounds
    We note that while the virtual screening stage of the project was low-cost and non-time consuming, one bottleneck in the screening was the price of the compounds. The typical price for commercially available compounds ranges from 50 € to 200 € per mg. Thus, screening of hundreds of compounds in cell-free assays would be the most expensive stage of the project. Also, the fact that most of the compounds are sold with minimum 1 mg pack size (synthesis limit) and that the initial cell-free screening only requires fraction of that amount makes purchasing of vast amount of compounds unreasonable.
  3. Feasibility of virtual screening In our experience
    Virtual screening as a method for identification of molecular tool compounds performed well in finding a small set of skeleton compounds even though only relatively small number of compounds identified as hits by molecular modeling were eventually tested in vitro (cell free assays with recombinant hepsin). These results indicate that virtual screening is in principle a feasible method for identification of transmembrane serine protease inhibitors but it would be advisable to use virtual screening in combination with other approaches. For example, starting with skeleton compounds with previously established inhibitory action against the transmembrane serine protease of interest or against closely related serine protease. The virtual screening strategy described here would be greatly facilitated by a possibility to purchase just a small microgram quantity of the candidate molecule instead of a milligram. There would be a significant market for any contract research laboratory able to miniaturize compound synthesis for purpose of testing significant number of virtual screening hits with low-cost.

Recipes

  1. Activation buffer for recombinant human hepsin (pH 8.0)
    0.1 M Tris
    10 mM CaCl2
    0.15 M NaCl
    0.05% Brij-35
  2. Assay buffer for recombinant human hepsin (pH 9.0)
    50 mM Tris

Acknowledgments

Protocol is adapted from Tervonen et al. (2016). This study was funded by the Academy of Finland, Sigrid Jusélius Foundation, the Finnish Cancer Society, the Research Funds of the Helsinki University Central Hospital, Jane and Aatos Erkko Foundation, and Helsinki Graduate Program in Biotechnology and Molecular Biology and Innovative Medicines Initiative Joint Undertaking under grant agreement No.115188.

References

  1. Antalis, T. M., Buzza, M. S., Hodge, K. M., Hooper, J. D. and Netzel-Arnett, S. (2010). The cutting edge: membrane-anchored serine protease activities in the pericellular microenvironment. Biochem J 428(3): 325-346.
  2. Bachovchin, D. A. and Cravatt, B. F. (2012). The pharmacological landscape and therapeutic potential of serine hydrolases. Nat Rev Drug Discov 11(1): 52-68.
  3. Deng, Z., Chuaqui, C. and Singh, J. (2004). Structural interaction fingerprint (SIFt): A novel method for analyzing three-dimensionla protein-ligand binding interactions. J Med Chem 47(2): 337-344.
  4. Friesner, R. A., Banks, J. L., Murphy, R. B., Halgren, T. A., Klicic, J. J., Mainz, D. T., Repasky, M. P., Knoll, E. H., Shelley, M., Perry, J. K., Shaw, D. E., Francis, P. and Shenkin, P. S. (2004). Glide: a new approach for rapid, accurate docking and scoring. 1. Method and assessment of docking accuracy. J Med Chem 47(7): 1739-1749.
  5. Friesner, R. A., Murphy, R. B., Repasky, M. P., Frye, L. L., Greenwood, J. R., Halgren, T. A., Sanschagrin, P. C. and Mainz, D. T. (2006). Extra precision glide: docking and scoring incorporating a model of hydrophobic enclosure for protein-ligand complexes. J Med Chem 49(21): 6177-6196.
  6. Goswami, R., Wohlfahrt, G., Törmäkangas, O., Moilanen, A., Lakshminarasimhan, A., Nagaraj, J., Arumugam, K. N., Mukherjee, S., Chacko, A. R., Krishnamurthy, N. R., Jaleel, M., Palakurthy, R. K., Samiulla, D. S. and Ramachandra, M. (2015). Structure-guided discovery of 2-aryl/pyridin-2-yl-1H-indole derivatives as potent and selective hepsin inhibitors. Bioorg Med Chem Lett 25(22): 5309-5314.
  7. Halgren, T. A., Murphy, R. B., Friesner, R. A., Beard, H. S., Frye, L. L., Pollard, W. T. and Banks, J. L. (2004). Glide: a new approach for rapid, accurate docking and scoring. 2. Enrichment factors in database screening. J Med Chem 47(7): 1750-1759.
  8. Lopez-Otin, C. and Matrisian, L. M. (2007). Emerging roles of proteases in tumour suppression. Nat Rev Cancer 7(10): 800-808.
  9. Sastry, G. M., Adzhigirey, M., Day, T., Annabhimoju, R. and Sherman, W. (2013). Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments. J Comput Aided Mol Des 27(3): 221-234.
  10. Singh, J.,Deng, Z., Narale, G. and Chuaqui, C. (2006). Structural interaction fingerprints: a new approach to organizing, mining, analyzing, and designing protein-small molecule complexes. Chem Biol Drug Des 67(1): 5-12.
  11. Sterling, T. and Irwin, J.J. (2015). ZINC 15--Ligand discovery for everyone. J Chem Inf Model 55(11): 2324-2337.
  12. Tervonen, T. A., Belitskin, D., Pant, S. M., Englund, J. I., Marques, E., Ala-Hongisto, H., Nevalaita, L., Sihto, H., Heikkila, P., Leidenius, M., Hewitson, K., Ramachandra, M., Moilanen, A., Joensuu, H., Kovanen, P. E., Poso, A. and Klefstrom, J. (2016). Deregulated hepsin protease activity confers oncogenicity by concomitantly augmenting HGF/MET signalling and disrupting epithelial cohesion. Oncogene 35(14): 1832-1846.
  13. Turk, B. (2006). Targeting proteases: successes, failures and future prospects. Nat Rev Drug Discov 5(9): 785-799.

简介

II型跨膜丝氨酸蛋白酶的人类家族包括17个成员。这些蛋白酶的定义特征是糜蛋白酶(S1)折叠的N末端跨膜结构域和C末端丝氨酸蛋白酶,通过可变的茎区彼此分离。最近积累的证据表明这些蛋白酶在癌症和转移能力的发展中起关键作用。肿瘤相关性和用于治疗和成像剂的细胞外定向催化结构域的可及性促使药物发现对II型跨膜丝氨酸蛋白酶的兴趣。通常,初始命中发现过程旨在识别在药物靶标上具有可验证活性的分子,并具有足够的药物样特征。我们在这里提出了用于跨膜丝氨酸蛋白酶hepinin的候选配体的基于结构的虚拟筛选方案。该方法描述了使用三磷酸腺嘌呤的催化位点的3D结构与ZINC进行分子对接,ZINC是具有 3000万种可购买化合物的分子数据库。小的候选子集​​经过实验测试,具有可证实的命中,为进一步的铅开发提供了有用的配体结构线索。

受控的蛋白水解活性在细胞过程和信号传导中发挥重要作用,正如所有生物体中蛋白酶的存在所证明的,包括病毒,原核生物和真核生物。不足为奇的是,异常调节的蛋白酶活性是造成人类病态,如心血管疾病和炎症疾病,骨质疏松症,神经系统疾病和癌症的多种因素(Turk,2006; Bachovchin和Cravatt,2012)。特别是,原发性癌症和转移能力的发展已经与几种不同类型的蛋白酶相关,包括基质金属蛋白酶(MMPs),半胱氨酸蛋白酶(组织蛋白酶)和膜相关丝氨酸蛋白酶(Lopez-Otin和Matrisian,2007)。最近的临床,遗传和功能数据表明膜相关丝氨酸蛋白酶在固体癌症(包括前列腺癌,卵巢癌和乳腺癌)中的关键作用已经引起了人们对开发小分子丝氨酸蛋白酶抑制剂治疗癌症的新兴趣。 Hepsin是一种II型跨膜丝氨酸蛋白酶,是丝氨酸蛋白酶药物开发的有吸引力的靶标,因为普通固体癌症如前列腺癌和乳腺癌中的高频胰蛋白酶过表达,其对癌细胞膜的过表达的限制,以及由于催化结构域到胞外,即,更可达的细胞侧(Antalis等人,2010)。
 我们在这里描述了基于结构的虚拟筛选丝氨酸蛋白酶抑制剂的方案,使用hepsin的催化位点与ZINC数据库的药物样子集对接。在虚拟筛选结果的基础上,共购买了24种候选化合物进行进一步的生化鉴定。使用重组hepsin和荧光肽底物Boc-Gln-Arg-Arg-AMC(BACHEM)进行基于无细胞ELISA的荧光酶测定用于实验验证,并将30%的肽溶解活性抑制作为阈值。截止值是基于一个概念,即用最初的高微摩尔浓度的化合物,超过30%的抑制将需要确定合理的IC 50值(Goswami等人,2015; Tervonen,等等,2016)(图1A和1B)。根据这些标准,24种测试化合物中有3种显示出抑制效能(Tervonen等,2016)。总之,即使基于结构的虚拟筛查通常被认为是补充药物筛选方法,这里报告的方案允许我们证明可行性,并为进一步开发特异性丝氨酸蛋白酶抑制剂提供了有意义的结构支架。然而,重要的是要强调,虚拟筛选命中通常不会显示出高效力,这对于现在的屏幕也是如此。高亲和力配体通常仅在选定的命中结构的熟练药物化学优化后才可用。


图1.说明工作流程的原理图。A.制备虚拟筛选(化合物)文库,分析和制备靶蛋白结构。虚拟筛选通过对接(Glide SP / XP)进行,并且推定的结合物通过体外生化测定验证。 B.候选配体(a-d)的实例。

   这里提出的虚拟筛选协议是我们最近的研究论文(Tervonen等人,2016年)中描述的原始协议的修改版本。这些修改使协议与ZINC数据库和Schrödinger软件的最新版本完全兼容。虽然目前的虚拟筛选协议是针对Schrödinger软件和ZINC数据库的用户量身定制的,但该协议是通用的,并且与其他分子建模软件包兼容。然而,使用其他建模包实现当前虚拟筛选协议将需要对对接设置进行仔细的平台特定验证。作为指导,我们提供了本屏幕中使用的验证过程的一般概述。

关键字:分子建模, 小分子, 分子对接, 癌症, 药物探索

材料和试剂

  1. ViewPlate-96黑色测定板(PerkinElmer,目录号:6005225)
  2. 重组人类hepsin蛋白(R& D Systems,目录号:4776-SE-010)
  3. DMSO(Sigma-Aldrich,目录号:D8418)
  4. 荧光肽底物Boc-Gln-Arg-Arg-AMC(Bachem,目录号:I-1655.0025)
  5. 小分子化合物购自Asinex Corporation(Winston-Salem,NC, http://www .asinex.com )或MolPort(拉脱维亚里加, https: //www.molport.com/shop/index
  6. WX-UK1是Ramachandra博士(Aurigene Discovery Technologies Limited,Bangalore,India)的一种礼物,但是它也可以商购(例如,AURUM Pharmatech,目录号:Z-3200)
  7. Trizma碱(Tris)(Sigma-Aldrich目录号:T1503)
  8. 氯化钙(CaCl 2)(Sigma-Aldrich,目录号:499609)
  9. Brij-35(Brij L23 30%H 2 O溶液)(Sigma-Aldrich,目录号:B4184)
  10. 氯化钠(NaCl)(Sigma-Aldrich,目录号:31434-M)
  11. 重组人类hepsin蛋白的活化缓冲液(见配方)
  12. 重组人hepsin蛋白的测定缓冲液(参见食谱)

设备

  1. ELISA板阅读器FLUOstar Omega(BMG LABTECH,型号:FLUOstar Omega)
  2. 电脑硬件由CSC提供 - IT中心科技有限公司(芬兰埃斯波)
    1. HP Proliant SL230s(Hewlett Packard Development Company,型号:SL230s)
    2. 英特尔至强E5-2670 2.6GHz前瞻(100核,每个内核16 GB内存)(英特尔公司,型号:E5-2670)

软件

  1. 建模软件
    使用蛋白质制备向导(Sastry等人,2013)的SchrödingerSuite(Schrödinger,LLC,New York,NY)),LigPrep,MacroModel和Glide(Friesner等人, 2004和2006; Halgren等人,2004)模块和协议版本2016-2( https://www.schrodinger.com/suites/small-molecule-drug-discovery-suite
  2. 数据库
    1. 筛选和验证:ZINC数据库(可通过 http://zinc15.docking.org/)。 ZINC数据库包含超过1亿种3D格式的购买化合物(Sterling and Irwin,2015)
    2. 蛋白质结构数据库( http://www.rcsb.org/pdb/home/home.do

程序

  1. 用于虚拟筛选的靶蛋白结构的制备
    hepsin蛋白质结构(PDB id 10E,来自蛋白质结构数据库, www.rcsb.org ,请参阅"在线工具")可以使用"生物单位"选项自动下载薛定with with with with with with。。。。。。。。。。使用其他软件,应使用手动下载,并以软件首选方式读取结构。除了保留所有的水分子之外,使用默认方法进行预处理。应使用默认pH值7进行H键精制。所有具有至少3个非水原子H键的水分子都被保存,整个结构在默认设置下被最小化,重原子会聚在RMSD0.3Å。 Glide Grid文件使用上述结构创建,Grid的中心位于Tyr94(图2)。 GRID的中心是指GRID创建期间分子场计算的中心。 Tyr94,Tyr146,Ser195和Tyr228的羟基被允许旋转以进行网格生成。增加自由可旋转键可以将Glide计算时间增加一倍,并应将其计算时间保持在合理的水平。此外,强烈建议,如果可能,将构造Glide Grid文件,以便使用OPLS3强制字段。如果使用另一种软件,应使用相应的蛋白质制备方法。关键点是确保检查蛋白质的结构质量,使用极性侧链固定正确的电离状态,并使用(psuedo)对称残基的右旋转异构体。


    图2.用于对接的GRID区域的定义。 GRID的中心基于残留TYR94,TYR146和SER195(残留标记的残基)的位置,GRIF区域以品红色框显示。

  2. 虚拟筛选化合物库的准备工作
    虚拟筛选对象库必须使用虚拟筛选软件的作者推荐的方法和设置进行准备。当选择虚拟化合物库时,应考虑到用于筛选的化合物的量,自身合成的可能性,并且如果所得到的命中旨在如此使用或在随后的药物化学程序中进一步修饰。我们使用ZINC库的药物样子集,从ZINC网页下载为SDF文件。 SDF文件格式由所有主要的软件品牌支持。 SDF文件由SchrödingerLigPrep模块准备,具有默认设置(图3)。简而言之,准备工作包括检查2D结构质量,快速2D-3D转化,质子化状态的分析和修改,所有可达互变异构体的修饰以及进一步最小化所得分子。在典型情况下,每个分子将用2-4种不同的电离/原子/互变异构体状态表示。对于除Schrödinger软件以外的软件,我们建议通过LigPrep进行配置设置。根据我们自己的经验,这种方法的好处是它可以产生电离状态(即,pKa)以及互变异构体的高质量预测。大多数其他软件包中不包括互变异构体选项的预测。


    图3.SchrödingerSuite下的LigPrep窗口。虚拟筛选库根据所选的设置进行准备;所示的设置是我们协议中的默认设置。

  3. 虚拟筛选和分析结果
    在我们的研究(Tervonen等人,2016)中,虚拟筛选进行如下:
    1. 使用上面定义的网格文件执行SP对接,最好的10%的化合物将用XP设置进行更新。
      1. 在第一阶段使用VHTS设置,然后SP设置仅用于最佳的50,000-100,000种化合物(根据Glide评分功能进行排名),而且,对于最佳的10%XP设置。
      2. 在原始研究中使用OPLS2005力场。然而,我们现在强烈建议OPLS3力场,因为它比原始方法提供了更好的结果。
    2. 聚类化合物最初通过视觉研究所得到的对接姿势进行。
      1. 目视检查的目的是确保只接受合理的对接姿势。通常约10-20%的对接姿势是不现实的,因为缺少H键接触或不切实际的配体构象,这些可以通过目测进行有效的筛选。
    3. 然后使用Schrödinger中提供的Interaction Fingerpint脚本对化合物进行聚类。
      1. 该脚本基于交互简档(交互指纹)分析每个对接姿势,并且通过使用这些配置文件聚集姿势。在我们的研究中,我们使用默认链接方法通过脚本评估的最佳聚类数。有关方法及其用途的详细描述,请参见Deng等人。 (2004)和Singh等人。 (2006)
    4. 结果造成了24个群集。
    5. 在每个簇中,最具代表性的化合物位于簇的中心。选择这些化合物用于体外测定。

  4. 通过体外实验验证筛选结果无细胞肽裂解测定用于hepsin酶活性方案
    1. 重组人hepsin(rhHepsin)在活化缓冲液中稀释至100μg/ml
    2. 通过将溶液在37℃下孵育24小时来激活rhHepsin
    3. 在使用前,用测定缓冲液将rhHepsin稀释成0.1nM溶液。
    4. 将小分子化合物溶于DMSO中,得到10mM储备溶液
    5. 测定在96孔板中的测定缓冲液中进行
    6. 在100μl反应体积中将0.1nM终浓度的rhHepsin和10μM终浓度的小分子化合物(和等体积的DMSO对照)在室温下孵育30分钟。分析缓冲区用作空白控制。
    7. 通过加入终浓度为30μM肽底物BOC-Gln-Arg-Arg-AMC开始反应。
    8. 通过350nm激发和450nm发射捕获在室温下用ELISA板读数器分析板。
    9. 抑制%通过使用以下公式确定:


数据分析

描述的虚拟筛选协议导致了数百个潜在的命中。使用上述协议和Tervonen等人。 (2016),我们进行了一个初始的体外无细胞验证。通过三次技术重复和两次重复,相对于DMSO对照,肽切割的减少30%被认为是命中的截止值。我们订购并测试了通过虚拟筛选鉴定的24种化合物作为潜在命中(图1B),并且少量化合物使其接近截止值(参见Tervonen等人。这些化合物被用作下一轮虚拟筛选的骨架,这是一个重复的过程。在Tervonen等人的研究中。 (2016),我们选择使用在平行屏幕上鉴定的WX-UK1,因为与其他测试分子相比,它与hepsin的结合姿势和极好的抑制效率。

笔记

  1. 电脑硬件
    除了上面列出的硬件之外,我们还注意到,任何具有任何主要操作系统(Windows,Mac或Linux)以及Intel或AMD处理器的高端PC计算机都适用于虚拟筛选工作,但该过程将需要更多时间。
  2. 小分子化合物
    我们注意到,虽然项目的虚拟筛选阶段是低成本和非耗时的,但筛选的一个瓶颈是化合物的价格。市售化合物的典型价格从50欧元到200欧元/毫克。因此,在无细胞测定中筛选数百种化合物将是项目中最昂贵的阶段。此外,大多数化合物以最小1mg包装尺寸(合成限度)出售,并且初始无细胞筛选仅需要该部分的量使得购买大量化合物不合理。
  3. 虚拟筛选的可行性在我们的经验中
    虚拟筛选作为用于鉴定分子工具化合物的方法,在发现一小组骨架化合物方面表现良好,即使通过分子建模鉴定为命中的相对较少数量的化合物最终在体外测试(无细胞用重组hepsin进行测定)。这些结果表明,虚拟筛选原则上是鉴定跨膜丝氨酸蛋白酶抑制剂的可行方法,但建议使用虚拟筛选与其他方法结合使用。例如,从具有先前对目的的跨膜丝氨酸蛋白酶具有抑制作用或与密切相关的丝氨酸蛋白酶具有抑制作用的骨架化合物开始。这里描述的虚拟筛选策略将极大地促进了购买仅一个小微克数量的候选分子而不是毫克的可能性。任何合同研究实验室都将有一个重要的市场,能够将复合合成小型化,以便以低成本测试大量虚拟筛选命中。

食谱

  1. 重组人类hepsin(pH 8.0)的活化缓冲液
    0.1 M Tris
    10mM CaCl 2
    0.15 M NaCl
    0.05%Brij-35
  2. 重组人hepsin(pH 9.0)的测定缓冲液 50 mM Tris

致谢

协议是从Tervonen等人改编而来的。 (2016)。这项研究由芬兰科学院,SigridJusélius基金会,芬兰癌症协会,赫尔辛基大学中央医院研究基金,Jane和Aatos Erkko基金会以及赫尔辛基生物技术与分子生物学和创新药物倡议联合承诺研究生计划授予协议No.115188。

参考文献

  1. Antalis,TM,Buzza,MS,Hodge,KM,Hooper,JD和Netzel-Arnett,S。(2010)。切割边缘:膜锚定的丝氨酸蛋白酶活性在细胞周围微环境中。生物化学J 428(3):325-346 。
  2. Bachovchin,DA和Cravatt,BF(2012)。  药理学风景和丝氨酸水解酶的治疗潜力。 Nat Rev Drug Discov 11(1):52-68。
  3. Deng,Z.,Chuaqui,C. and Singh,J。(2004)。  结构相互作用指纹(SIFt):一种分析三维1a蛋白 - 配体结合相互作用的新方法。/em> 47(2):337-344。
  4. Friesner,RA,Banks,JL,Murphy,RB,Halgren,TA,Klicic,JJ,Mainz,DT,Repasky,MP,Knoll,EH,Shelley,M.,Perry,JK,Shaw,DE,Francis, Shenkin,PS(2004)。  Glide:一种新方法快速,准确的对接和打分。 1.对接精度的方法和评估。 J Med Chem 47(7):1739-1749。
  5. Friesner,RA,Murphy,RB,Repasky,MP,Frye,LL,Greenwood,JR,Halgren,TA,Sanschagrin,PC and Mainz,DT(2006)。  额外的精度滑移:对接和评分结合了蛋白质 - 配体复合物的疏水性外壳模型。 Med Chem 49(21):6177-6196。
  6. Goswami,R.,Wohlfahrt,G.,Törmäkangas,O.,Moilanen,A.,Lakshminarasimhan,A.,Nagaraj,J.,Arumugam,KN,Mukherjee,S.,Chacko,AR,Krishnamurthy,NR,Jaleel,M 。,Palakurthy,RK,Samiulla,DS and Ramachandra,M。(2015)。  2-芳基/吡啶-2-基-1H-吲哚衍生物的结构引导发现作为有效和选择性的hepsin抑制剂。 > Bioorg Med Chem Lett 25(22):5309-5314。
  7. Halgren,TA,Murphy,RB,Friesner,RA,Beard,HS,Frye,LL,Pollard,WT and Banks,JL(2004)。  Glide:一种快速,准确对接和打分的新方法。 2.数据库筛选中的丰富因素。 J Med Chem。47(7):1750-1759。
  8. Lopez-Otin,C.和Matrisian,LM(2007)。蛋白酶在肿瘤抑制中的新作用。 Nat Rev Cancer 7(10):800-808。
  9. Sastry,GM,Adzhigirey,M.,Day,T.,Annabhimoju,R。和Sherman,W。(2013)。  结构相互作用指纹:组织,开采,分析和设计蛋白质 - 小分子复合物的新方法化学生物药物Des 67/1):5-12。
  10. Sterling,T.和Irwin,J.J。 (2015)。< a class ="ke-insertfile"href ="https://www.ncbi.nlm.nih.gov/pubmed/?term=ZINC+15+%E2%80%93+Ligand+发现+ for +所有人"target ="_ blank"> ZINC 15 - 所有人的配体发现。 J Chem Inf Model 55(11):2324-2337。
  11. Tervonen,TA,Belitskin,D.,Pant,SM,Englund,JI,Marques,E.,Ala-Hongisto,H.,Nevalaita,L.,Sihto,H.,Heikkila,P.,Leidenius,M.,Hewitson ,K.,Ramachandra,M.,Moilanen,A.,Joensuu,H.,Kovanen,PE,Poso,A。和Klefstrom,J.(2016)。< a class ="ke-insertfile"href = http://www.ncbi.nlm.nih.gov/pubmed/26165838"target ="_ blank">排除的hepsin蛋白酶活性通过伴随地增加HGF/MET信号传导和破坏上皮内聚力赋予致癌性。致癌基因 35(14):1832-1846。
  12. Turk,B.(2006)。  靶向蛋白酶:成功,失败和未来前景。 Nat Rev Drug Discov 5(9):785-799。
  • English
  • 中文翻译
免责声明 × 为了向广大用户提供经翻译的内容,www.bio-protocol.org 采用人工翻译与计算机翻译结合的技术翻译了本文章。基于计算机的翻译质量再高,也不及 100% 的人工翻译的质量。为此,我们始终建议用户参考原始英文版本。 Bio-protocol., LLC对翻译版本的准确性不承担任何责任。
Copyright: © 2017 The Authors; exclusive licensee Bio-protocol LLC.
引用:Poso, A., Tervonen, T. and Klefström, J. (2017). Virtual Screening of Transmembrane Serine Protease Inhibitors. Bio-protocol 7(8): e2246. DOI: 10.21769/BioProtoc.2246.
提问与回复

(提问前,请先登录)bio-protocol作为媒介平台,会将您的问题转发给作者,并将作者的回复发送至您的邮箱(在bio-protocol注册时所用的邮箱)。为了作者与用户间沟通流畅(作者能准确理解您所遇到的问题并给与正确的建议),我们鼓励用户用图片或者视频的形式来说明遇到的问题。由于本平台用Youtube储存、播放视频,作者需要google 账户来上传视频。

当遇到任务问题时,强烈推荐您提交相关数据(如截屏或视频)。由于Bio-protocol使用Youtube存储、播放视频,如需上传视频,您可能需要一个谷歌账号。