High-throughput mammalian two-hybrid screening for protein-protein interactions using transfected cell arrays
© Fiebitz et al; licensee BioMed Central Ltd. 2008
Received: 05 September 2007
Accepted: 06 February 2008
Published: 06 February 2008
Most of the biological processes rely on the formation of protein complexes. Investigation of protein-protein interactions (PPI) is therefore essential for understanding of cellular functions. It is advantageous to perform mammalian PPI analysis in mammalian cells because the expressed proteins can then be subjected to essential post-translational modifications. Until now mammalian two-hybrid assays have been performed on individual gene scale. We here describe a new and cost-effective method for the high-throughput detection of protein-protein interactions in mammalian cells that combines the advantages of mammalian two-hybrid systems with those of DNA microarrays.
In this cell array protein-protein interaction assay (CAPPIA), mixtures of bait and prey expression plasmids together with an auto-fluorescent reporter are immobilized on glass slides in defined array formats. Adherent cells that grow on top of the micro-array will become fluorescent only if the expressed proteins interact and subsequently trans-activate the reporter. Using known interaction partners and by screening 160 different combinations of prey and bait proteins associated with the human androgen receptor we demonstrate that this assay allows the quantitative detection of specific protein interactions in different types of mammalian cells and under the influence of different compounds. Moreover, different strategies in respect to bait-prey combinations are presented.
We demonstrate that the CAPPIA assay allows the quantitative detection of specific protein interactions in different types of mammalian cells and under the influence of different compounds. The high number of preys that can be tested per slide together with the flexibility to interrogate any bait of interest and the small amounts of reagents that are required makes this assay currently one of the most economical high-throughput detection assays for protein-protein interactions in mammalian cells.
Most if not all biological processes require the cooperation of pairs of proteins or the formation of large functional complexes of proteins. Therefore the analysis of protein-protein interactions, either in vitro, using for example protein arrays, co-immunoprecipitation or affinity chromatography, or in vivo by two-hybrid assays is essential for the elucidation of biological processes and/or networks.
In classical yeast or mammalian two-hybrid based assays typically two proteins of interest are ectopically expressed as fusion proteins, one with the DNA Binding Domain (DBD) of for example GAL4 or LexA and the other with a transcriptional Activating Domain (AD), such that if both proteins show any interaction, the DBD and AD are functionally linked together at the promoter, reconstituting transcriptional activity [1–3]. This causes reporters that contain GAL4- or LexA binding sequences to be transcribed.
Since two-hybrid systems are in vivo assays they offer advantages over in vitro biophysical or biochemical methods. Indeed some protein-protein interactions are too weak and/or transient to be detected in vitro and some of these interactions require specific post-translational modifications of the proteins and/or specific co-factors in the cellular microenvironment. For the same reasons it is advantageous to determine protein interaction networks in mammalian cells, using mammalian two-hybrid assays . Until now high-throughput analyses of mammalian protein interactions were typically performed in yeast [5, 6] and putative interactions were then confirmed in mammalian two-hybrid assays on a gene-by-gene scale [7, 8].
We present here a novel assay for the parallel analysis of thousands of proteins for interacting partners in mammalian cells by combining cell arrays , with the more classical mammalian two-hybrid assay.
In this cell array protein-protein interaction assay (CAPPIA), nanoliter volumes of solutions containing a bait expression plasmid, a prey expression plasmid and a reporter plasmid complexed with transfection reagent are immobilized on glass slides in well-defined array formats. When these slides are overlayed with a monolayer of living cells, only those cells that grow on top of a particular spot of DNA will get transfected and will start to over-express specific chimeric bait and prey proteins. If these two proteins can interact with each other they will transactivate the autofluorescent reporter making that cluster of cells fluorescent while the surrounding cells remain non-fluorescent. Fluorescent cell clusters/features can then be analysed by simple fluorescence detection using common DNA array scanners or high-throughput microscopy, without the need for further manipulation of the slides such as immunofluorescence staining or enzyme-based detection.
Using known interacting proteins we demonstrate the specific and quantitative detection of protein-protein interactions on cell arrays in different mammalian cell lines. Furthermore, screening of 160 different combinations of prey and bait proteins including different domains of the human androgen receptor reveals that this assay is well suited for the detection of hormone-dependent protein interactions. The physiological significance of this interaction on cell arrays is further underscored by showing the dose response of this interaction to androgenic compounds as well as to antagonistic reagents.
Finally we demonstrate that the possible combinatorial screens to detect protein interactions can be increased by combining slides that only carry a prey library with cell lines that carry a stable or transient bait construct. This flexibility, together with the high capacity of the cell arrays and the small amounts of reagents that are required, makes this assay currently one of the most economical high-throughput detection assays for protein-protein interactions in mammalian cells.
Development of a cell array based method for the detection of protein-protein interactions in various mammalian cells
The observation that specific protein-protein interactions can readily be analyzed in different cell lines is of particular interest for interactions that are dependent on cell-specific posttranslational modifications of expressed proteins or that depend on cell-specific co-factors. A cell type dependent difference in PPI-induced reporter expression is already evident for p53 and SV40T proteins that induced higher reporter levels when expressed in PC-3 as compared to HEK293T cells, even though both cell lines can be transfected with similar efficiencies, as reflected by the level of NF-κB-induced reporter expression in these cell lines.
Application of CAPPIA for screening of hormone-dependent protein interactions
Taken together these experiments clearly demonstrate that cell arrays provide a functional readout to identify and monitor PPI under different physiological conditions. Besides interrogating PPI in different cellular contexts, cell arrays can also be used to screen for ligand-dependent protein-protein interactions and to quantify the dose response of these to various compounds.
PR-stable bait and PR-trans bait cell arrays
In order to increase the possible combinatorial screens for protein interactions and hence further improve the high-throughput application of CAPPIA, we printed alternative slides on which the bait was omitted. Each spot on these so-called prey-reporter- (PR-) slides contains only the reporter and one of the prey constructs. To screen for interacting partners the bait is then introduced into the cells before adding them to the prey-reporter-arrays. This is either done using stably transfected cell lines (PR-stable-bait assay) or by transiently transfecting cells with the bait shortly before adding them to the arrays (PR-trans-bait assay). In both cases all the cells on these transfected slides express the bait. However, only cells that grow on top of a spot, with a prey that can interact with that bait, become fluorescent.
Although PRB slides offer a cost-effective and robust platform for simultaneous comparison of large number of interactions in different cell lines and under different culture conditions (time of culture, addition of agonists or antagonists), the PR slides are more suitable for the large-scale screening of novel bait-prey interactions. As shown in Additional file 3, the spot to spot distance on CAPPIA slides can easily be reduced to 1 mm. By printing each sample as one feature per slide and using a spot distance of 1 mm, up to 900 different preys can be printed on one slide. Replicate PR-slides can then be screened with any bait of interest, further increasing the high-throughput application of CAPPIA. In addition PR slides can be printed in large batches and can be stored for at least 6 months at 4°C and even longer at -80°C (data not shown), thus increasing the cost effectiveness of CAPPIA. The slides can be shipped on ice, allowing any research group with access to standard culture facilities to interrogate the prey library with their bait of interest.
We hereby present a new screening technology for the high-throughput interrogation of protein-protein interactions in mammalian cells. This cell-array-protein-protein-interaction-assay (CAPPIA) combines the advantages of mammalian two-hybrid systems with those of microarrays, a combination that is expected to save considerable time and expense. Indeed, CAPPIA slides are printed with the same robotic microarray devices used to print conventional DNA microarrays. Consequently, cell arrays require far less DNA, transfection reagents and cells as compared to assays performed in microwell plate format.
For comparison, the amount of transfection reagent required to transfect one sample in a 96 micro-well is sufficient to transfect more than 2000 samples in CAPPIA, a significant cost reduction since transfection reagents account for a major part of the expenses in these assays. Although cell arrays can theoretically carry up to 8000 features per slide [9–11, 13], the actual number of features per slide is dependent on the type of application. Since the efficiency of the simultaneous transfection of three plasmids is lower than that of a single plasmid and since the reporter will only be expressed after sufficient amounts of both prey and bait proteins are expressed in the same cell there is a need to increase the number of cells per spot in CAPPIA in order to guarantee the robust detection of interacting proteins. Consequently, in order to increase the sensitivity of the assay, the features on CAPPIA slides need to be bigger and hence the slide capacity will be lower. Still, CAPPIA slides can be printed with each containing up to 900 different preys as single spots when using a spot to spot distance of 1 mm. This is equivalent to 9 standard 96 micro-well plates and data can then be collected from a number of identical slides to obtain replicate data points per sample. In addition, by pooling 3 preys per feature even more preys can be interrogated per slide. The cost-effectiveness of the assay is further increased by using an auto-fluorescent based reporter eliminating the need for immuno-fluorescence staining or enzyme-based reporter detection. In contrast to microwell-based assays no extensive liquid handling infrastructure is required once the slides have been printed. In addition PR or PRB slides can be printed in large batches and stored frozen without losing their efficiency, thus increasing the flexibility and cost effectiveness of CAPPIA. Finally, PR slides can be screened with any bait of interest simply by using standard transfection and cell culture methods, available to many research groups.
Compared to the more common high-throughput yeast two-hybrid assay, CAPPIA offers the advantage of testing mammalian PPI within a cellular context that more closely mimics the native protein environment. As a consequence the rate of false negative interactions, i.e. true interactions that can not be detected due to the inappropriate folding and/or post-translational processing of the proteins should be lower using CAPPIA, especially since phenotypically different cell types can be used. As with DNA microarrays, PR-slides interrogated with different baits allow a comparison between multiple identical assays. As such preys or baits that are found multiple times with unrelated proteins, so-called bait- or prey-unspecific false positives [22, 23] can be distinguished without need for further experimentation. High-throughput yeast two-hybrid assays also require extensive liquid handling infrastructures [5, 24] making the assays less accessible for many research groups.
Previously, Suzuki et al  described a PCR-mediated rapid sample preparation and a high-throughput assay system based on the mammalian two-hybrid method. Although the method they designed allows for the rapid preparation of high numbers of bait and prey samples, the actual two-hybrid assays are performed in microwells and require semiautomatic multiple dispensers as well as multiple reagents for downstream enzymatic detection of interacting proteins. CAPPIA, on the other hand allows the quantitative analysis of interactions comparable to assays performed in microwells, while offering the advantage of being economical and accessible to research groups equipped with basic cell culture infrastructures. Currently, large bait-prey libraries for use in CAPPIA are being generated using Gateway (Invitrogen) compatible destination pCMV-AD and pCMV-BD vectors and panels of pENTRY clones. Since CAPPIA is based on the GAL4-two-hybrid system that is already used by many research groups, existing baits and preys can easily be included in the assays. In particular, the trans-bait version of CAPPIA readily allows available bait constructs to be screened against large number of preys.
Taken together, the high capacity of the cell arrays, the flexibility to interrogate any bait of interest and the small amounts of reagents that are required make CAPPIA currently one of the most economical high-throughput detection assays for protein-protein interactions in mammalian cells. Microarray-based systems such as the one presented here should lead to the acceleration of the deciphering of functional protein complexes. In addition CAPPIA could also be an economical alternative for the identification of potential drug targets by combining screens for small molecule ligands that can disrupt or modulate interactions with selected prey-bait combinations expressed in array format inside 96 micro-wells .
WI-38, SV40-transformed fibroblasts; HeLa, epithelial cervical carcinoma; COS 7 African green monkey kidney cells; HEK293 embryonic kidney cells and its variant HEK293T were from ATTC. PC-3 human prostate cancer cells were from the German Collection of Microorganisms and Cell Cultures (Braunschweig); the HepG2, hepatocellular carcinoma was a kind gift from Dr. S. Sperling (MPI-MG, Berlin). Cells were cultured at 37°C, 5% CO2 in D-MEM (GIBCO) with 10% fetal bovine serum, penicillin/streptomycin and 1% L-glutamine.
HEK293-LBD cells were created by transfecting HEK293 with pBD-LBD and selecting for Neomycine-resistant transformants. HEK293 cells instead of HEK293T cells were used for these experiments since the latter is already Neomycine resistant.
Expression bait (pCMV-BD) and prey (pCMV-AD) vectors used to generate proteins of interest as GAL4 DBD fusions and fusions with the transactivation domain of NF-κB, respectively were from Mammalian Two-Hybrid Assay Kit from Stratagene. The regions coding for the different proteins or protein domains, summarized in Additional file 1, were introduced into pCMV-BD and pCMV-AD using standard methods. Positive and negative control plasmids from the kit were pBD-NF-κB, pBD-p53, pAD-SV40T and pAD-TRAF.
A CMV-driven EGFP construct was generated by PCR amplification of EGFP from pIRES2-EGFP (Clontech) and TA cloning into pcDNA4/HisMax TOPO (Invitrogen).
The autofluorescent reporter Gal4-pZsGreen was created by cloning the GAL4 upstream activating sequences from pGAL/lacZ (Invitrogen) into the multicloning sites of pZsGreen1-1 (Clontech). The pCIS-CK plasmid was from Stratagene.
Samples were prepared as described using the lipid-DNA-method [9, 10]. Briefly, DNA mixtures containing reporter and prey (for PR slides) or reporter, prey and bait constructs (for PRB slides) in EC buffer supplemented with 0.2 M sucrose were treated for 5 minutes at RT with 1.5 μl Enhancer (Effectene, QIAGEN). Subsequently, 7 μl Effectene (QIAGEN) was added and 10 minutes later an equal amount of 0.1% gelatine (SIGMA) solution in water was added. Samples were used immediately or stored for several days at 4°C. Where required pCIS-CK was used to obtain a final DNA concentration of 50 ng/μl.
Preparation of arrayed slides
Arrays were printed using a sciFlexArrayer non-contact piezodispensing system (Scienion AG) with 70 μm nozzle. Features were printed with a spot-to-spot distance of 1.5 mm using about 6 nl sample per feature. Slides used were home made poly-L-lysine (SIGMA) coated plain microscope glass slides.
For small number of samples, manual spotting is a useful alternative to the use of automated array robots. About 10 nl sample was manually deposited per spot using long tips (PreCision safe seal tips, 10 μl, Biozym) by gently touching the filled tip on top of the slide surface. Slides were positioned on a frame allowing the arrayed dispensing of spots at 1.5 mm intervals.
Printed slides were dried for a minimum of 1 hour before transfection or were stored for longer times desiccated at 4°C or -20°C.
Transfection of cell arrays
For HEK293T, HEK293, HEK293-LBD, PC-3, WI-38 and HepG2, 1 × 107 cells were seeded in a 60 cm2 cell culture plate one day before transfection. For HeLa and Cos7, 5 × 106 cells were pre-cultured in a 145 cm2 plate. On the day of transfection cells were seeded at 3.5 × 106 (HEK293T, HEK293, HEK293-LBD, PC-3, Cos7, HepG2 and WI-38) or 1 × 106 (HeLa) cells per slide in 8 ml complete media in quadriperm (Vivascience) dishes. For hormone-dependent interactions the medium was supplemented with R1881 (Methyltrienolone, Perkin Elmer) and Medroxyprogesterone acetate (MPA, Schering AG) or hydroxyflutamide (OH-Flu, Schering AG) at the indicated concentrations. After 48 or 72 hours the slides were washed with PBS and fixed with PBS/4M sucrose/3.7% formaldehyde solution for 30 min. After washing in PBS, slides were mounted with Fluoromount-G (Southern Biotech) and stored in the dark at 4°C until analysis.
Signal detection and analysis
Fluorescent signals were detected using a DNA microarray reader (BIOCCD Image Reader, PE Applied Biosystems) with excitation 470/30 and emission 510/20 filters for EGFP and ZsGreen. The entire slide was uniformly scanned with same focus and exposure time settings. Pictures were handled by AxioVision LE (ZEISS) and image densitometry analysis was done with AlphaEase FC Stand Alone Software, version 4.0.0 (Alpha Inotech). Using the "Spot Denso" tool integrated in the software a "Spot denso overlay" was generated, stored and loaded for each new scanned slide that was printed with the same feature size and pitch parameters. Using this overlay the relative grey value of each feature of the slide is automatically analysed and exported as an excel file for further processing. Background fluorescence of cells is subtracted before data is exported. Data was collected from identical slides to achieve a number of replicate data points per sample as indicated. Exposure times were chosen so that the weaker, experimental signals are within linear range. No selection markers were used to monitor transfection of individual plasmids. Signals were normalized for transfection efficiency relative to the level of EGFP expression on the same slide and the mean value from replicate spots was calculated for each sample.
DNA binding domain
cell array protein-protein interaction assay
prey, reporter, bait
- PR prey:
androgen receptor ligand binding domain
androgen receptor N-terminal domain
We thank Claus Hultschig, Regine Schwartz and Tillmann Wegwerth of the Max Planck Institute for Molecular Genetics for their help with array preparations. The study was supported by the European Commission (integrated project MolTools) and the German Federal Ministry of Education and Research (grant no 0313068, 01GS0416, 01GR0414, 31P3674).
- Fields S, Song O: A novel genetic system to detect protein-protein interactions. Nature. 1989, 340 (6230): 245-246. 10.1038/340245a0.PubMedView ArticleGoogle Scholar
- Uetz P: Two-hybrid arrays. Curr Opin Chem Biol. 2002, 6 (1): 57-62. 10.1016/S1367-5931(01)00288-5.PubMedView ArticleGoogle Scholar
- Drees BL: Progress and variations in two-hybrid and three-hybrid technologies. Curr Opin Chem Biol. 1999, 3 (1): 64-70. 10.1016/S1367-5931(99)80012-X.PubMedView ArticleGoogle Scholar
- Dang CV, Barrett J, Villa-Garcia M, Resar LM, Kato GJ, Fearon ER: Intracellular leucine zipper interactions suggest c-Myc hetero-oligomerization. Mol Cell Biol. 1991, 11 (2): 954-962.PubMedPubMed CentralView ArticleGoogle Scholar
- Stelzl U, Worm U, Lalowski M, Haenig C, Brembeck FH, Goehler H, Stroedicke M, Zenkner M, Schoenherr A, Koeppen S, Timm J, Mintzlaff S, Abraham C, Bock N, Kietzmann S, Goedde A, Toksoz E, Droege A, Krobitsch S, Korn B, Birchmeier W, Lehrach H, Wanker EE: A human protein-protein interaction network: a resource for annotating the proteome. Cell. 2005, 122 (6): 957-968. 10.1016/j.cell.2005.08.029.PubMedView ArticleGoogle Scholar
- Rual JF, Venkatesan K, Hao T, Hirozane-Kishikawa T, Dricot A, Li N, Berriz GF, Gibbons FD, Dreze M, Ayivi-Guedehoussou N, Klitgord N, Simon C, Boxem M, Milstein S, Rosenberg J, Goldberg DS, Zhang LV, Wong SL, Franklin G, Li S, Albala JS, Lim J, Fraughton C, Llamosas E, Cevik S, Bex C, Lamesch P, Sikorski RS, Vandenhaute J, Zoghbi HY, Smolyar A, Bosak S, Sequerra R, Doucette-Stamm L, Cusick ME, Hill DE, Roth FP, Vidal M: Towards a proteome-scale map of the human protein-protein interaction network. Nature. 2005, 437 (7062): 1173-1178. 10.1038/nature04209.PubMedView ArticleGoogle Scholar
- Luo Y, Batalao A, Zhou H, Zhu L: Mammalian two-hybrid system: a complementary approach to the yeast two-hybrid system. Biotechniques. 1997, 22 (2): 350-352.PubMedGoogle Scholar
- Leonhardt SA, Altmann M, Edwards DP: Agonist and antagonists induce homodimerization and mixed ligand heterodimerization of human progesterone receptors in vivo by a mammalian two-hybrid assay. Mol Endocrinol. 1998, 12 (12): 1914-1930. 10.1210/me.12.12.1914.PubMedView ArticleGoogle Scholar
- Ziauddin J, Sabatini DM: Microarrays of cells expressing defined cDNAs. Nature. 2001, 411 (6833): 107-110. 10.1038/35075114.PubMedView ArticleGoogle Scholar
- Hu YH, Vanhecke D, Lehrach H, Janitz M: High-throughput subcellular protein localization using cell arrays. Biochem Soc Trans. 2005, 33 (Pt 6): 1407-1408.PubMedView ArticleGoogle Scholar
- Vanhecke D, Janitz M: High-throughput gene silencing using cell arrays. Oncogene. 2004, 23 (51): 8353-8358. 10.1038/sj.onc.1208027.PubMedView ArticleGoogle Scholar
- Mannherz O, Mertens D, Hahn M, Lichter P: Functional screening for proapoptotic genes by reverse transfection cell array technology. Genomics. 2006, 87 (5): 665-672. 10.1016/j.ygeno.2005.12.009.PubMedView ArticleGoogle Scholar
- Hu YH, Warnatz HJ, Vanhecke D, Wagner F, Fiebitz A, Thamm S, Kahlem P, Lehrach H, Yaspo ML, Janitz M: Cell array-based intracellular localization screening reveals novel functional features of human chromosome 21 proteins. BMC Genomics. 2006, 7: 155-10.1186/1471-2164-7-155.PubMedPubMed CentralView ArticleGoogle Scholar
- Tsai MJ, O'Malley BW: Molecular mechanisms of action of steroid/thyroid receptor superfamily members. Annu Rev Biochem. 1994, 63: 451-486. 10.1146/annurev.bi.63.070194.002315.PubMedView ArticleGoogle Scholar
- Mangelsdorf DJ, Thummel C, Beato M, Herrlich P, Schutz G, Umesono K, Blumberg B, Kastner P, Mark M, Chambon P, Evans RM: The nuclear receptor superfamily: the second decade. Cell. 1995, 83 (6): 835-839. 10.1016/0092-8674(95)90199-X.PubMedView ArticleGoogle Scholar
- Evans RM: The steroid and thyroid hormone receptor superfamily. Science. 1988, 240 (4854): 889-895. 10.1126/science.3283939.PubMedView ArticleGoogle Scholar
- Beato M, Herrlich P, Schutz G: Steroid hormone receptors: many actors in search of a plot. Cell. 1995, 83 (6): 851-857. 10.1016/0092-8674(95)90201-5.PubMedView ArticleGoogle Scholar
- Doesburg P, Kuil CW, Berrevoets CA, Steketee K, Faber PW, Mulder E, Brinkmann AO, Trapman J: Functional in vivo interaction between the amino-terminal, transactivation domain and the ligand binding domain of the androgen receptor. Biochemistry. 1997, 36 (5): 1052-1064. 10.1021/bi961775g.PubMedView ArticleGoogle Scholar
- Ahrens-Fath I, Politz O, Geserick C, Haendler B: Androgen receptor function is modulated by the tissue-specific AR45 variant. Febs J. 2005, 272 (1): 74-84. 10.1111/j.1432-1033.2004.04395.x.PubMedView ArticleGoogle Scholar
- Langley E, Zhou ZX, Wilson EM: Evidence for an anti-parallel orientation of the ligand-activated human androgen receptor dimer. J Biol Chem. 1995, 270 (50): 29983-29990. 10.1074/jbc.270.50.29983.PubMedView ArticleGoogle Scholar
- Kemppainen JA, Langley E, Wong CI, Bobseine K, Kelce WR, Wilson EM: Distinguishing androgen receptor agonists and antagonists: distinct mechanisms of activation by medroxyprogesterone acetate and dihydrotestosterone. Mol Endocrinol. 1999, 13 (3): 440-454. 10.1210/me.13.3.440.PubMedView ArticleGoogle Scholar
- Serebriiskii IG, Golemis EA: Two-hybrid system and false positives. Approaches to detection and elimination. Methods Mol Biol. 2001, 177: 123-134.PubMedGoogle Scholar
- Ito T, Chiba T, Ozawa R, Yoshida M, Hattori M, Sakaki Y: A comprehensive two-hybrid analysis to explore the yeast protein interactome. Proc Natl Acad Sci U S A. 2001, 98 (8): 4569-4574. 10.1073/pnas.061034498.PubMedPubMed CentralView ArticleGoogle Scholar
- Shioda T, Andriole S, Yahata T, Isselbacher KJ: A green fluorescent protein-reporter mammalian two-hybrid system with extrachromosomal maintenance of a prey expression plasmid: application to interaction screening. Proc Natl Acad Sci U S A. 2000, 97 (10): 5220-5224. 10.1073/pnas.97.10.5220.PubMedPubMed CentralView ArticleGoogle Scholar
- Suzuki H, Fukunishi Y, Kagawa I, Saito R, Oda H, Endo T, Kondo S, Bono H, Okazaki Y, Hayashizaki Y: Protein-protein interaction panel using mouse full-length cDNAs. Genome Res. 2001, 11 (10): 1758-1765. 10.1101/gr.180101.PubMedPubMed CentralView ArticleGoogle Scholar
- Bailey SN, Wu RZ, Sabatini DM: Applications of transfected cell microarrays in high-throughput drug discovery. Drug Discov Today. 2002, 7 (18 Suppl): S113-8. 10.1016/S1359-6446(02)02386-3.PubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.