共查询到20条相似文献,搜索用时 15 毫秒
1.
Davis FP 《Molecular bioSystems》2011,7(2):545-557
Small molecules that modulate protein-protein interactions are of great interest for chemical biology and therapeutics. Here I present a structure-based approach to predict 'bi-functional' sites able to bind both small molecule ligands and proteins, in proteins of unknown structure. First, I develop a homology-based annotation method that transfers binding sites of known three-dimensional structure onto protein sequences, predicting residues in ligand and protein binding sites with estimated true positive rates of 98% and 88%, respectively, at 1% false positive rates. Applying this method to the human proteome predicts 8463 proteins with bi-functional residues and correctly recovers the targets of known interaction modulators. Proteins with significantly (p < 0.01) more bi-functional residues than expected were found to be enriched in regulatory and depleted in metabolism functions. Finally, I demonstrate the utility of the method by describing examples of predicted overlap and evidence of their biological and therapeutic relevance. The results suggest that combining the structures of known binding sites with established fold detection algorithms can predict regions of protein-protein interfaces that are amenable to small molecule modulation. Open-source software and the results for several complete proteomes are available at http://pibase.janelia.org/homolobind. 相似文献
2.
Neeliyath A. Ramakrishnan Marian J. Drescher Barbara J. Morley Philip M. Kelley Dennis G. Drescher 《The Journal of biological chemistry》2014,289(13):8750-8766
Mutations in otoferlin, a C2 domain-containing ferlin family protein, cause non-syndromic hearing loss in humans (DFNB9 deafness). Furthermore, transmitter secretion of cochlear inner hair cells is compromised in mice lacking otoferlin. In the present study, we show that the C2F domain of otoferlin directly binds calcium (KD = 267 μm) with diminished binding in a pachanga (D1767G) C2F mouse mutation. Calcium was found to differentially regulate binding of otoferlin C2 domains to target SNARE (t-SNARE) proteins and phospholipids. C2D–F domains interact with the syntaxin-1 t-SNARE motif with maximum binding within the range of 20–50 μm Ca2+. At 20 μm Ca2+, the dissociation rate was substantially lower, indicating increased binding (KD = ∼10−9) compared with 0 μm Ca2+ (KD = ∼10−8), suggesting a calcium-mediated stabilization of the C2 domain·t-SNARE complex. C2A and C2B interactions with t-SNAREs were insensitive to calcium. The C2F domain directly binds the t-SNARE SNAP-25 maximally at 100 μm and with reduction at 0 μm Ca2+, a pattern repeated for C2F domain interactions with phosphatidylinositol 4,5-bisphosphate. In contrast, C2F did not bind the vesicle SNARE protein synaptobrevin-1 (VAMP-1). Moreover, an antibody targeting otoferlin immunoprecipitated syntaxin-1 and SNAP-25 but not synaptobrevin-1. As opposed to an increase in binding with increased calcium, interactions between otoferlin C2F domain and intramolecular C2 domains occurred in the absence of calcium, consistent with intra-C2 domain interactions forming a “closed” tertiary structure at low calcium that “opens” as calcium increases. These results suggest a direct role for otoferlin in exocytosis and modulation of calcium-dependent membrane fusion. 相似文献
3.
Interactions between small molecules and proteins play critical roles in regulating and facilitating diverse biological functions, yet our ability to accurately re-engineer the specificity of these interactions using computational approaches has been limited. One main difficulty, in addition to inaccuracies in energy functions, is the exquisite sensitivity of protein–ligand interactions to subtle conformational changes, coupled with the computational problem of sampling the large conformational search space of degrees of freedom of ligands, amino acid side chains, and the protein backbone. Here, we describe two benchmarks for evaluating the accuracy of computational approaches for re-engineering protein-ligand interactions: (i) prediction of enzyme specificity altering mutations and (ii) prediction of sequence tolerance in ligand binding sites. After finding that current state-of-the-art “fixed backbone” design methods perform poorly on these tests, we develop a new “coupled moves” design method in the program Rosetta that couples changes to protein sequence with alterations in both protein side-chain and protein backbone conformations, and allows for changes in ligand rigid-body and torsion degrees of freedom. We show significantly increased accuracy in both predicting ligand specificity altering mutations and binding site sequences. These methodological improvements should be useful for many applications of protein – ligand design. The approach also provides insights into the role of subtle conformational adjustments that enable functional changes not only in engineering applications but also in natural protein evolution. 相似文献
4.
Michal Brylinski 《PLoS computational biology》2014,10(9)
Detecting similarities between ligand binding sites in the absence of global homology between target proteins has been recognized as one of the critical components of modern drug discovery. Local binding site alignments can be constructed using sequence order-independent techniques, however, to achieve a high accuracy, many current algorithms for binding site comparison require high-quality experimental protein structures, preferably in the bound conformational state. This, in turn, complicates proteome scale applications, where only various quality structure models are available for the majority of gene products. To improve the state-of-the-art, we developed eMatchSite, a new method for constructing sequence order-independent alignments of ligand binding sites in protein models. Large-scale benchmarking calculations using adenine-binding pockets in crystal structures demonstrate that eMatchSite generates accurate alignments for almost three times more protein pairs than SOIPPA. More importantly, eMatchSite offers a high tolerance to structural distortions in ligand binding regions in protein models. For example, the percentage of correctly aligned pairs of adenine-binding sites in weakly homologous protein models is only 4–9% lower than those aligned using crystal structures. This represents a significant improvement over other algorithms, e.g. the performance of eMatchSite in recognizing similar binding sites is 6% and 13% higher than that of SiteEngine using high- and moderate-quality protein models, respectively. Constructing biologically correct alignments using predicted ligand binding sites in protein models opens up the possibility to investigate drug-protein interaction networks for complete proteomes with prospective systems-level applications in polypharmacology and rational drug repositioning. eMatchSite is freely available to the academic community as a web-server and a stand-alone software distribution at http://www.brylinski.org/ematchsite.
This is a PLOS Computational Biology Software Article相似文献
5.
6.
7.
Interactions between proteins, drugs, water and B-DNA minor groove have been analyzed in crystal structures of 60 protein–DNA and 14 drug–DNA complexes. It was found that only purine N3, pyrimidine O2, guanine N2 and deoxyribose O4′ are involved in the interactions, and that contacts to N3 and O2 are most frequent and more polar than contacts to O4′. Many protein contacts are mediated by water, possibly to increase the DNA effective surface. Fewer water-mediated contacts are observed in drug complexes. The distributions of ligands around N3 are significantly more compact than around O2, and distributions of water molecules are the most compact. Distributions around O4′ are more diffuse than for the base atoms but most distributions still have just one binding site. Ligands bind to N3 and O2 atoms in analogous positions, and simultaneous binding to N3 and N2 in guanines is extremely rare. Contacts with two consecutive nucleotides are much more frequent than base–sugar contacts within one nucleotide. The probable reason for this is the large energy of deformation of hydrogen bonds for the one nucleotide motif. Contacts of Arg, the most frequent amino acid ligand, are stereochemically indistinguishable from the binding of the remaining amino acids except asparagine (Asn) and phenylalanine (Phe). Asn and Phe bind in distinct ways, mostly to a deformed DNA, as in the complexes of TATA-box binding proteins. DNA deformation concentrates on dinucleotide regions with a distinct deformation of the δ and backbone torsion angles for the Asn and δ, , ζ and χ for the Phe-contacted regions. 相似文献
8.
Aoife M. Doyle David A. Ross Kaballa Maganja Kathy Baisley Clemens Masesa Aura Andreasen Mary L. Plummer Angela I. N. Obasi Helen A. Weiss Saidi Kapiga Deborah Watson-Jones John Changalucha Richard J. Hayes for the MEMA kwa Vijana Trial Study Group 《PLoS medicine》2010,7(6)
Background
The ability of specific behaviour-change interventions to reduce HIV infection in young people remains questionable. Since January 1999, an adolescent sexual and reproductive health (SRH) intervention has been implemented in ten randomly chosen intervention communities in rural Tanzania, within a community randomised trial (see below; ). The intervention consisted of teacher-led, peer-assisted in-school education, youth-friendly health services, community activities, and youth condom promotion and distribution. Process evaluation in 1999–2002 showed high intervention quality and coverage. A 2001/2 intervention impact evaluation showed no impact on the primary outcomes of HIV seroincidence and herpes simplex virus type 2 (HSV-2) seroprevalence but found substantial improvements in SRH knowledge, reported attitudes, and some reported sexual behaviours. It was postulated that the impact on “upstream” knowledge, attitude, and reported behaviour outcomes seen at the 3-year follow-up would, in the longer term, lead to a reduction in HIV and HSV-2 infection rates and other biological outcomes. A further impact evaluation survey in 2007/8 (∼9 years post-intervention) tested this hypothesis. NCT00248469Methods and Findings
This is a cross-sectional survey (June 2007 through July 2008) of 13,814 young people aged 15–30 y who had attended trial schools during the first phase of the MEMA kwa Vijana intervention trial (1999–2002). Prevalences of the primary outcomes HIV and HSV-2 were 1.8% and 25.9% in males and 4.0% and 41.4% in females, respectively. The intervention did not significantly reduce risk of HIV (males adjusted prevalence ratio [aPR] 0.91, 95%CI 0.50–1.65; females aPR 1.07, 95%CI 0.68–1.67) or HSV-2 (males aPR 0.94, 95%CI 0.77–1.15; females aPR 0.96, 95%CI 0.87–1.06). The intervention was associated with a reduction in the proportion of males reporting more than four sexual partners in their lifetime (aPR 0.87, 95%CI 0.78–0.97) and an increase in reported condom use at last sex with a non-regular partner among females (aPR 1.34, 95%CI 1.07–1.69). There was a clear and consistent beneficial impact on knowledge, but no significant impact on reported attitudes to sexual risk, reported pregnancies, or other reported sexual behaviours. The study population was likely to have been, on average, at lower risk of HIV and other sexually transmitted infections compared to other rural populations, as only youth who had reached year five of primary school were eligible.Conclusions
SRH knowledge can be improved and retained long-term, but this intervention had only a limited effect on reported behaviour and no significant effect on HIV/STI prevalence. Youth interventions integrated within intensive, community-wide risk reduction programmes may be more successful and should be evaluated.Trial Registration
ClinicalTrials.gov Please see later in the article for the Editors'' Summary NCT00248469相似文献9.
Yibing Shan Venkatesh P. Mysore Abba E. Leffler Eric T. Kim Shiori Sagawa David E. Shaw 《PLoS computational biology》2022,18(3)
Protein-protein interactions (PPIs) are ubiquitous biomolecular processes that are central to virtually all aspects of cellular function. Identifying small molecules that modulate specific disease-related PPIs is a strategy with enormous promise for drug discovery. The design of drugs to disrupt PPIs is challenging, however, because many potential drug-binding sites at PPI interfaces are “cryptic”: When unoccupied by a ligand, cryptic sites are often flat and featureless, and thus not readily recognizable in crystal structures, with the geometric and chemical characteristics of typical small-molecule binding sites only emerging upon ligand binding. The rational design of small molecules to inhibit specific PPIs would benefit from a better understanding of how such molecules bind at PPI interfaces. To this end, we have conducted unbiased, all-atom MD simulations of the binding of four small-molecule inhibitors (SP4206 and three SP4206 analogs) to interleukin 2 (IL2)—which performs its function by forming a PPI with its receptor—without incorporating any prior structural information about the ligands’ binding. In multiple binding events, a small molecule settled into a stable binding pose at the PPI interface of IL2, resulting in a protein–small-molecule binding site and pose virtually identical to that observed in an existing crystal structure of the IL2-SP4206 complex. Binding of the small molecule stabilized the IL2 binding groove, which when the small molecule was not bound emerged only transiently and incompletely. Moreover, free energy perturbation (FEP) calculations successfully distinguished between the native and non-native IL2–small-molecule binding poses found in the simulations, suggesting that binding simulations in combination with FEP may provide an effective tool for identifying cryptic binding sites and determining the binding poses of small molecules designed to disrupt PPI interfaces by binding to such sites. 相似文献
10.
11.
Tsukasa Fukunaga Haruka Ozaki Goro Terai Kiyoshi Asai Wataru Iwasaki Hisanori Kiryu 《Genome biology》2014,15(1):R16
RNA-binding proteins (RBPs) bind to their target RNA molecules by recognizing specific RNA sequences and structural contexts. The development of CLIP-seq and related protocols has made it possible to exhaustively identify RNA fragments that bind to RBPs. However, no efficient bioinformatics method exists to reveal the structural specificities of RBP–RNA interactions using these data. We present CapR, an efficient algorithm that calculates the probability that each RNA base position is located within each secondary structural context. Using CapR, we demonstrate that several RBPs bind to their target RNA molecules under specific structural contexts. CapR is available at https://sites.google.com/site/fukunagatsu/software/capr. 相似文献
12.
Anna Philips Kaja Milanowska Grzegorz ?ach Janusz M. Bujnicki 《RNA (New York, N.Y.)》2013,19(12):1605-1616
RNA molecules have recently become attractive as potential drug targets due to the increased awareness of their importance in key biological processes. The increase of the number of experimentally determined RNA 3D structures enabled structure-based searches for small molecules that can specifically bind to defined sites in RNA molecules, thereby blocking or otherwise modulating their function. However, as of yet, computational methods for structure-based docking of small molecule ligands to RNA molecules are not as well established as analogous methods for protein-ligand docking. This motivated us to create LigandRNA, a scoring function for the prediction of RNA–small molecule interactions. Our method employs a grid-based algorithm and a knowledge-based potential derived from ligand-binding sites in the experimentally solved RNA–ligand complexes. As an input, LigandRNA takes an RNA receptor file and a file with ligand poses. As an output, it returns a ranking of the poses according to their score. The predictive power of LigandRNA favorably compares to five other publicly available methods. We found that the combination of LigandRNA and Dock6 into a “meta-predictor” leads to further improvement in the identification of near-native ligand poses. The LigandRNA program is available free of charge as a web server at http://ligandrna.genesilico.pl. 相似文献
13.
Intrinsically disordered proteins (IDPs) were found to be widely associated with human diseases and may serve as potential drug design targets. However, drug design targeting IDPs is still in the very early stages. Progress in drug design is usually achieved using experimental screening; however, the structural disorder of IDPs makes it difficult to characterize their interaction with ligands using experiments alone. To better understand the structure of IDPs and their interactions with small molecule ligands, we performed extensive simulations on the c-Myc370–409 peptide and its binding to a reported small molecule inhibitor, ligand 10074-A4. We found that the conformational space of the apo c-Myc370–409 peptide was rather dispersed and that the conformations of the peptide were stabilized mainly by charge interactions and hydrogen bonds. Under the binding of the ligand, c-Myc370–409 remained disordered. The ligand was found to bind to c-Myc370–409 at different sites along the chain and behaved like a ‘ligand cloud’. In contrast to ligand binding to more rigid target proteins that usually results in a dominant bound structure, ligand binding to IDPs may better be described as ligand clouds around protein clouds. Nevertheless, the binding of the ligand and a non-ligand to the c-Myc370–409 target could be clearly distinguished. The present study provides insights that will help improve rational drug design that targets IDPs. 相似文献
14.
Matti Uusitupa Markku Peltonen Jaana Lindstr?m Sirkka Aunola Pirjo Ilanne-Parikka Sirkka Kein?nen-Kiukaanniemi Timo T. Valle Johan G. Eriksson Jaakko Tuomilehto for the Finnish Diabetes Prevention Study Group 《PloS one》2009,4(5)
Background
The Finnish Diabetes Prevention Study (DPS) was a randomized controlled trial, which showed that it is possible to prevent type 2 diabetes by lifestyle changes. The aim of the present study was to examine whether the lifestyle intervention had an effect on the ten-year mortality and cardiovascular morbidity in the DPS participants originally randomized either into an intervention or control group. Furthermore, we compared these results with a population-based cohort comprising individuals of varying glucose tolerance states.Methods and Findings
Middle-aged, overweight people with IGT (n = 522) were randomized into intensive intervention (including physical activity, weight reduction and dietary counseling), or control “mini-intervention” group. Median length of the intervention period was 4 years and the mean follow-up was 10.6 years. The population-based reference study cohort included 1881 individuals (1570 with normal glucose tolerance, 183 with IGT, 59 with screen-detected type 2 diabetes, 69 with previously known type 2 diabetes) with the mean follow-up of 13.8 years. Mortality and cardiovascular morbidity data were collected from the national Hospital Discharge Register and Causes of Death Register. Among the DPS participants who consented for register linkage (n = 505), total mortality (2.2 vs. 3.8 per 1000 person years, hazard ratio HR = 0.57, 95% CI 0.21–1.58) and cardiovascular morbidity (22.9 vs. 22.0 per 1000 person years, HR = 1.04, 95% CI 0.72–1.51) did not differ significantly between the intervention and control groups. Compared with the population-based cohort with impaired glucose tolerance, adjusted HRs were 0.21 (95% CI 0.09–0.52) and 0.39 (95% CI 0.20–0.79) for total mortality, and 0.89 (95% CI 0.62–1.27) and 0.87 (0.60–1.27) for cardiovascular morbidity in the intervention and control groups of the DPS, respectively. The risk of death in DPS combined cohort was markedly lower than in FINRISK IGT cohort (adjusted HR 0.30, 95% CI 0.17–0.54), but there was no significant difference in the risk of CVD (adjusted HR 0.88, 95% CI 0.64–1.21).Conclusions
Lifestyle intervention among persons with IGT did not decrease cardiovascular morbidity during the first 10 years of follow-up. However, the statistical power may not be sufficient to detect small differences between the intervention and control groups. Low total mortality among participants of the DPS compared with individuals with IGT in the general population could be ascribed to a lower cardiovascular risk profile at baseline and regular follow-up.Trial Registration
ClinicalTrials.gov NCT00518167相似文献15.
Andrew R. Zolopa Janet Andersen Lauren Komarow Ian Sanne Alejandro Sanchez Evelyn Hogg Carol Suckow William Powderly for the ACTG A study team 《PloS one》2009,4(5)
Background
Optimal timing of ART initiation for individuals presenting with AIDS-related OIs has not been defined.Methods and Findings
A5164 was a randomized strategy trial of “early ART” - given within 14 days of starting acute OI treatment versus “deferred ART” - given after acute OI treatment is completed. Randomization was stratified by presenting OI and entry CD4 count. The primary week 48 endpoint was 3-level ordered categorical variable: 1. Death/AIDS progression; 2. No progression with incomplete viral suppression (ie HIV viral load (VL) ≥50 copies/ml); 3. No progression with optimal viral suppression (ie HIV VL <50 copies/ml). Secondary endpoints included: AIDS progression/death; plasma HIV RNA and CD4 responses and safety parameters including IRIS.282 subjects were evaluable; 141 per arm. Entry OIs included Pneumocytis jirovecii pneumonia 63%, cryptococcal meningitis 12%, and bacterial infections 12%. The early and deferred arms started ART a median of 12 and 45 days after start of OI treatment, respectively.The difference in the primary endpoint did not reach statistical significance: AIDS progression/death was seen in 20 (14%) vs. 34 (24%); whereas no progression but with incomplete viral suppression was seen in 54 (38%) vs. 44 (31%); and no progression with optimal viral suppression in 67 (48%) vs 63 (45%) in the early vs. deferred arm, respectively (p = 0.22). However, the early ART arm had fewer AIDS progression/deaths (OR = 0.51; 95% CI = 0.27–0.94) and a longer time to AIDS progression/death (stratified HR = 0.53; 95% CI = 0.30–0.92). The early ART had shorter time to achieving a CD4 count above 50 cells/mL (p<0.001) and no increase in adverse events.Conclusions
Early ART resulted in less AIDS progression/death with no increase in adverse events or loss of virologic response compared to deferred ART. These results support the early initiation of ART in patients presenting with acute AIDS-related OIs, absent major contraindications.Trial Registration
ClinicalTrials.gov NCT00055120相似文献16.
Tunca Doan Ece Akhan Güzelcan Marcus Baumann Altay Koyas Heval Atas Ian R. Baxendale Maria Martin Rengul Cetin-Atalay 《PLoS computational biology》2021,17(11)
Predictive approaches such as virtual screening have been used in drug discovery with the objective of reducing developmental time and costs. Current machine learning and network-based approaches have issues related to generalization, usability, or model interpretability, especially due to the complexity of target proteins’ structure/function, and bias in system training datasets. Here, we propose a new method “DRUIDom” (DRUg Interacting Domain prediction) to identify bio-interactions between drug candidate compounds and targets by utilizing the domain modularity of proteins, to overcome problems associated with current approaches. DRUIDom is composed of two methodological steps. First, ligands/compounds are statistically mapped to structural domains of their target proteins, with the aim of identifying their interactions. As such, other proteins containing the same mapped domain or domain pair become new candidate targets for the corresponding compounds. Next, a million-scale dataset of small molecule compounds, including those mapped to domains in the previous step, are clustered based on their molecular similarities, and their domain associations are propagated to other compounds within the same clusters. Experimentally verified bioactivity data points, obtained from public databases, are meticulously filtered to construct datasets of active/interacting and inactive/non-interacting drug/compound–target pairs (~2.9M data points), and used as training data for calculating parameters of compound–domain mappings, which led to 27,032 high-confidence associations between 250 domains and 8,165 compounds, and a finalized output of ~5 million new compound–protein interactions. DRUIDom is experimentally validated by syntheses and bioactivity analyses of compounds predicted to target LIM-kinase proteins, which play critical roles in the regulation of cell motility, cell cycle progression, and differentiation through actin filament dynamics. We showed that LIMK-inhibitor-2 and its derivatives significantly block the cancer cell migration through inhibition of LIMK phosphorylation and the downstream protein cofilin. One of the derivative compounds (LIMKi-2d) was identified as a promising candidate due to its action on resistant Mahlavu liver cancer cells. The results demonstrated that DRUIDom can be exploited to identify drug candidate compounds for intended targets and to predict new target proteins based on the defined compound–domain relationships. Datasets, results, and the source code of DRUIDom are fully-available at: https://github.com/cansyl/DRUIDom. 相似文献
17.
18.
Background
Understanding the residue covariations between multiple positions in protein families is very crucial and can be helpful for designing protein engineering experiments. These simultaneous changes or residue coevolution allow protein to maintain its overall structural-functional integrity while enabling it to acquire specific functional modifications. Despite the significant efforts in the field there is still controversy in terms of the preferable locations of coevolved residues on different regions of protein molecules, the strength of coevolutionary signal and role of coevolution in functional diversification.Methodology
In this paper we study the scale and nature of residue coevolution in maintaining the overall functionality and structural integrity of proteins. We employed a large scale study to investigate the structural and functional aspects of coevolved residues. We found that the networks representing the coevolutionary residue connections within our dataset are in general of ‘small-world’ type as they have clustering coefficient values higher than random networks and also show smaller mean shortest path lengths similar and/or lower than random and regular networks. We also found that altogether 11% of functionally important sites are coevolved with any other sites. Active sites are found more frequently to coevolve with any other sites (15%) compared to protein (11%) and ligand (9%) binding sites. Metal binding and active sites are also found to be more frequently coevolved with other metal binding and active sites, respectively. Analysis of the coupling between coevolutionary processes and the spatial distribution of coevolved sites reveals that a high fraction of coevolved sites are located close to each other. Moreover, ∼80% of charge compensatory substitutions within coevolved sites are found at very close spatial proximity (< = 5Å), pointing to the possible preservation of salt bridges in evolution.Conclusion
Our findings show that a noticeable fraction of functionally important sites undergo coevolution and also point towards compensatory substitutions as a probable coevolutionary mechanism within spatially proximal coevolved functional sites. 相似文献19.
Hydrogen constitutes nearly half of all atoms in proteins and their positions are essential for analyzing hydrogen-bonding interactions and refining atomic-level structures. However, most protein structures determined by experiments or computer prediction lack hydrogen coordinates. We present a new algorithm, HAAD, to predict the positions of hydrogen atoms based on the positions of heavy atoms. The algorithm is built on the basic rules of orbital hybridization followed by the optimization of steric repulsion and electrostatic interactions. We tested the algorithm using three independent data sets: ultra-high-resolution X-ray structures, structures determined by neutron diffraction, and NOE proton-proton distances. Compared with the widely used programs CHARMM and REDUCE, HAAD has a significantly higher accuracy, with the average RMSD of the predicted hydrogen atoms to the X-ray and neutron diffraction structures decreased by 26% and 11%, respectively. Furthermore, hydrogen atoms placed by HAAD have more matches with the NOE restraints and fewer clashes with heavy atoms. The average CPU cost by HAAD is 18 and 8 times lower than that of CHARMM and REDUCE, respectively. The significant advantage of HAAD in both the accuracy and the speed of the hydrogen additions should make HAAD a useful tool for the detailed study of protein structure and function. Both an executable and the source code of HAAD are freely available at http://zhang.bioinformatics.ku.edu/HAAD. 相似文献
20.