- Email: [email protected]
Hay in a needlestack?
News & Comment
TRENDS in Biotechnology Vol.19 No.11 November 2001
435
Journal Club
Proteomics in context The genome-sequencing project has brought us an encyclopaedia of the molecular parts of living systems. The next step is to understand the contextual meaning of the ‘words’ of the book of life, which essentially means to elucidate the rules of interaction between the proteins. In fact, recent efforts to map the network of interactions at a genome-scale using high-throughput two-hybrid systems and in vitro protein interaction studies have quickly provided a glimpse of the wiring diagram of regulatory networks. However, information about the cellular context of signaling events, such as the subcellular location of the interaction, still requires old-fashioned, ‘pre-genomic’ approaches centred around one particular protein at a time. Now, Remy and Michnick1 report an efficient experimental approach that allows the parallel detection and localization of biologically relevant protein–protein interactions within living mammalian cells. They use a method developed in their laboratory called protein-fragment complementation in which two rationally dissected fragments of dihydrofolate reductase (DHFR), a common selection marker, are each fused to a protein or peptide of interest. Because the DHFR fragments are unable to fold independently, cell survival in nucleotide-free media relies
on the ability of the fused proteins to interact, thereby positioning the two DHFR fragments in close proximity so that folding and reconstitution of enzyme activity can proceed. The insulin and growth factor receptor tyrosine kinase (RTK)-mediated translation initiation pathway and an overlapping pathway controlled by the serine–threonine kinase FK506-binding protein FRAP were chosen as test cases. The authors tested 148 combinations of 35 different proteins in the RTK–FRAP signal transduction pathways and identified 14 interaction partners, five of which had not been demonstrated previously. Following this initial survival screen, a fluorescence read-out assay allowed them to gather additional information regarding the nature and context of the identified interactions. Reassembled DHFR binds with high affinity to fluorescein-conjugated methotrexate in a 1:1 complex, which is retained in cells. The fluorescence signal provides a quantitative measure of the number of molecules of folded DHFR and, therefore, the number of interacting protein complexes. At the same time, visualization of the cells using fluorescence microscopy pinpoints the subcellular location of the interaction. To probe the position of a particular protein–protein interaction within the higher-order organization of the signaling network, the authors established
a pharmacological profile for each combination of protein partners. Protein interactions and localization were monitored by fluorescent DHFR following perturbation with an array of specific smallmolecule inhibitors or stimulators. By examining the response profiles the authors were able to assign a position within the signaling network to newly discovered interactions. For example, they identify a novel interaction between FRAP and PDK1 that co-localizes with the known interaction between FRAP and PKB, suggesting a point of crosstalk between the RTK and FRAP signaling pathways. This pharmacological profiling based on the protein-fragment complementation system can be applied to larger, more complex biochemical networks. It will enable us to combine the efficiency of broad-scale genomics screening with traditional, information-rich cell biology approaches to obtain insights into the collective function and cellular organization of gene products.
necessarily inherent to developing commercially viable drugs or lead molecules. Given the exciting prospect of identifying small molecules with novel biological activities, it is instructive to read the recent report of Liu et al.1, in which high-throughput screening was used to identify a small molecule disruptor of protein–protein interactions. A critical analysis of this and similar reports not only informs the design and performance of future screening experiments but also illustrates important differences between the responsibilities of industrial and academic screeners in interpreting and communicating their results.
Liu et al. describe a high-throughput screen for small-molecule inhibitors of the interaction between human insulin-like growth factor (IGF-I) and insulin-like growth factor-binding protein 3 (IGFBP-3). This in vitro assay measures the amount of radiolabeled IGF-I in a protein isolate, and designates as ‘hits’ those compounds resulting in a >50% inhibition of retained radioactivity. After screening ~85 000 compounds from an in-house collection of small molecules, the authors identified the most potent inhibitor of the protein–protein interaction in a well containing L-3,4-dihydroxyphenylalanine. However, their attempts to validate their
1 Remy, I. and Michnick, S.W. (2001) Visualization of biochemical networks in living cells. Proc. Natl. Acad. Sci. U. S. A. 98, 7678–7683
Amy Brock [email protected] Sui Huang [email protected]
Hay in a needlestack? The explosions in combinatorial synthesis and high-throughput screening have brought the academic and industrial worlds closer together, both in methods and in mindset. New sources of chemical diversity and new technology platforms for biological screening now allow even modest academic laboratories to generate assay data at a rate comparable with some pharmaceutical or biotechnology projects. High-throughput biological assays, long a standard in the pharmaceutical industry, are now making significant progress in basic biological discovery. In the drive to publish results quickly, however, academic screeners risk pitfalls similar to those http://tibtech.trends.com
0167-7799/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved.
436
News & Comment
result using authentic samples of the compound failed, suggesting that either the original designation of the library compound was incorrect, or that the compound had undergone some transformation during storage in dimethylsulfoxide (DMSO). Liu et al. fractionated L-3,4-dihydroxyphenylalanine after a period of incubation in DMSO, in an attempt to mimic their storage conditions. The major active fraction was isolated and designated NBI31772, which can displace hIGF-I from all human IGFBPs, with EC50 values ranging from 1.2 to 16 nM. Further, NBI31772 suppresses IGFBP-3-mediated inhibition of hIGF-I-induced radioactive thymidine incorporation in 3T3 fibroblasts. Thus, the authors have identified a small molecule that displaces hIGF-I from its binding proteins, which they hope will lead to the release and increased bioavailability of hIGF-I in animals. Two aspects of the authors’ paper bear specific comment. First, NBI31772 was ultimately identified via fractionation of authentic L-3,4-dihydroxyphenylalanine,
TRENDS in Biotechnology Vol.19 No.11 November 2001
following an attempt to reproduce library storage conditions. Although this fraction demonstrated the desired biological activity, it remains unclear whether the active chemical entity in the library stock is in fact NBI31772. Indeed, no attempt was made by the authors to report the exploration of alternative possibilities, only that a biologically active entity had been discovered. Second, although the library sample was described as the ‘best’ inhibitor, no mention was made of the other ‘hits’ obtained in the primary screen, or where it fell among that distribution of hits. What fraction of the library scored positive by the assay criterion? Would the results be different if a more stringent ‘hit’ criterion had been applied? These omissions can be rationalized by considering an endpoint analysis of the authors’ aims. As researchers at a biotechnology company, Liu et al. are reasonably expected to focus exclusively on the goal of a promising lead. To translate such thinking into the academy, however, is less appropriate. Although the
identification of an isolated ‘hit’ in a small molecule screen is potentially interesting as a probe for further study of a particular biological outcome, the statistical results of small molecule screens, particularly when coupled to an examination of structure–activity relationships, might provide general information about the interaction of small molecules with biological systems. Of course, researchers in basic biology must focus their efforts and resources on interesting questions and promising results but the total information content of highthroughput techniques must be made available to the general community for future query.
those with no preference into near teetotalers. The work by Panayotis Thanos et al. (Journal of Neurochemistry, September issue) could have implications for the prevention and treatment of alcoholism in humans. DM
Copycat?
1 Liu, X-J et al. (2001) Identification of a nonpeptide ligand that releases bioactive insulinlike growth factor-I from its binding protein complex. J. Biol. Chem. 276, 32419–32422
Paul A. Clemons [email protected] Bridget K. Wagner [email protected]
In Brief
Latest CF gene therapy bmn.com (6 September) gives a commentary on two reports in the July issue of Human Gene Therapy detailing new clinical trials of gene therapy treatments for cystic fibrosis (CF). CF has long been a good candidate for gene therapy, but results have yet to match initial expectations. The new research by workers at Genzyme Corporation (Framingham, MA, USA) and the University of Cincinnati Medical Center (Cincinnati, OH, USA) describes clinical effects of adenoviral vector or plasmid DNA delivery vehicles of the CF transmembrane conductance regulator (CFTR) gene. Both groups report limited (30–60%) incorporation of the CFTR gene and also noted inflammatory side effects arising from the delivery vehicle. MJD
Gene therapy reduces drinking in ‘alcoholic’ rats Scientists at the US Department of Energy’s Brookhaven National Laboratory (Upton, NY, USA) have shown that increasing the level of a brain protein important for transmitting pleasure signals can turn rats that prefer alcohol into light drinkers, and http://tibtech.trends.com
Helicobacter halted The stomach-ulcer-causing bacterium Helicobacter pylori might soon be literally stopped in its tracks following results published by a team of UK scientists (Microbiology, September issue). David Kelly (University of Sheffield, Sheffield, UK) has identified proteins responsible for sending chemical signals from the bacterial cell surface to the flagellum, which propels the bacteria away from harmful chemicals in its environment. Building on the sequencing of the H. pylori genome, which was published in 1994, the Sheffield group has been studying mutants that have defective chemotaxis genes, and have shown that they can interfere with the sensing and movement properties of H. pylori. Kelly believes that this could be one way to prevent infection and lead to the development of new anti-H. pylori drugs. MJD
For only US$ 1500, a Californian company is offering celebrities the chance to establish dubious copyright over their DNA. ‘A lot of people are going to want to clone people they admire,’ says Andre Crump, president of the DNA Copyright Institute (DCI) (San Francisco, CA, USA). For high-profile individuals worried they might fall victim to being cloned from a few living cells left behind on a glass or exchanged in a handshake, DCI is offering to record their DNA fingerprint, check that it is unique and store it. Ten people have already taken advantage of DCI’s services and for an extra fee, the company will also try to register the pattern with the US Copyright Office (New Scientist, 25 August). DM
Proteomics set for billion dollar growth A report by Front Line Strategic Management Consultants, Inc (Westfield, MA, USA) has forecast that the proteomics market will grow from its present value of US$ 560m to almost US$ 2.8b by 2005, (BioTech International, September issue). This equates to 40% year-on-year growth, and will be driven by the pharmaceutical
0167-7799/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved.
TRENDS in Biotechnology Vol.19 No.11 November 2001
435
Journal Club
Proteomics in context The genome-sequencing project has brought us an encyclopaedia of the molecular parts of living systems. The next step is to understand the contextual meaning of the ‘words’ of the book of life, which essentially means to elucidate the rules of interaction between the proteins. In fact, recent efforts to map the network of interactions at a genome-scale using high-throughput two-hybrid systems and in vitro protein interaction studies have quickly provided a glimpse of the wiring diagram of regulatory networks. However, information about the cellular context of signaling events, such as the subcellular location of the interaction, still requires old-fashioned, ‘pre-genomic’ approaches centred around one particular protein at a time. Now, Remy and Michnick1 report an efficient experimental approach that allows the parallel detection and localization of biologically relevant protein–protein interactions within living mammalian cells. They use a method developed in their laboratory called protein-fragment complementation in which two rationally dissected fragments of dihydrofolate reductase (DHFR), a common selection marker, are each fused to a protein or peptide of interest. Because the DHFR fragments are unable to fold independently, cell survival in nucleotide-free media relies
on the ability of the fused proteins to interact, thereby positioning the two DHFR fragments in close proximity so that folding and reconstitution of enzyme activity can proceed. The insulin and growth factor receptor tyrosine kinase (RTK)-mediated translation initiation pathway and an overlapping pathway controlled by the serine–threonine kinase FK506-binding protein FRAP were chosen as test cases. The authors tested 148 combinations of 35 different proteins in the RTK–FRAP signal transduction pathways and identified 14 interaction partners, five of which had not been demonstrated previously. Following this initial survival screen, a fluorescence read-out assay allowed them to gather additional information regarding the nature and context of the identified interactions. Reassembled DHFR binds with high affinity to fluorescein-conjugated methotrexate in a 1:1 complex, which is retained in cells. The fluorescence signal provides a quantitative measure of the number of molecules of folded DHFR and, therefore, the number of interacting protein complexes. At the same time, visualization of the cells using fluorescence microscopy pinpoints the subcellular location of the interaction. To probe the position of a particular protein–protein interaction within the higher-order organization of the signaling network, the authors established
a pharmacological profile for each combination of protein partners. Protein interactions and localization were monitored by fluorescent DHFR following perturbation with an array of specific smallmolecule inhibitors or stimulators. By examining the response profiles the authors were able to assign a position within the signaling network to newly discovered interactions. For example, they identify a novel interaction between FRAP and PDK1 that co-localizes with the known interaction between FRAP and PKB, suggesting a point of crosstalk between the RTK and FRAP signaling pathways. This pharmacological profiling based on the protein-fragment complementation system can be applied to larger, more complex biochemical networks. It will enable us to combine the efficiency of broad-scale genomics screening with traditional, information-rich cell biology approaches to obtain insights into the collective function and cellular organization of gene products.
necessarily inherent to developing commercially viable drugs or lead molecules. Given the exciting prospect of identifying small molecules with novel biological activities, it is instructive to read the recent report of Liu et al.1, in which high-throughput screening was used to identify a small molecule disruptor of protein–protein interactions. A critical analysis of this and similar reports not only informs the design and performance of future screening experiments but also illustrates important differences between the responsibilities of industrial and academic screeners in interpreting and communicating their results.
Liu et al. describe a high-throughput screen for small-molecule inhibitors of the interaction between human insulin-like growth factor (IGF-I) and insulin-like growth factor-binding protein 3 (IGFBP-3). This in vitro assay measures the amount of radiolabeled IGF-I in a protein isolate, and designates as ‘hits’ those compounds resulting in a >50% inhibition of retained radioactivity. After screening ~85 000 compounds from an in-house collection of small molecules, the authors identified the most potent inhibitor of the protein–protein interaction in a well containing L-3,4-dihydroxyphenylalanine. However, their attempts to validate their
1 Remy, I. and Michnick, S.W. (2001) Visualization of biochemical networks in living cells. Proc. Natl. Acad. Sci. U. S. A. 98, 7678–7683
Amy Brock [email protected] Sui Huang [email protected]
Hay in a needlestack? The explosions in combinatorial synthesis and high-throughput screening have brought the academic and industrial worlds closer together, both in methods and in mindset. New sources of chemical diversity and new technology platforms for biological screening now allow even modest academic laboratories to generate assay data at a rate comparable with some pharmaceutical or biotechnology projects. High-throughput biological assays, long a standard in the pharmaceutical industry, are now making significant progress in basic biological discovery. In the drive to publish results quickly, however, academic screeners risk pitfalls similar to those http://tibtech.trends.com
0167-7799/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved.
436
News & Comment
result using authentic samples of the compound failed, suggesting that either the original designation of the library compound was incorrect, or that the compound had undergone some transformation during storage in dimethylsulfoxide (DMSO). Liu et al. fractionated L-3,4-dihydroxyphenylalanine after a period of incubation in DMSO, in an attempt to mimic their storage conditions. The major active fraction was isolated and designated NBI31772, which can displace hIGF-I from all human IGFBPs, with EC50 values ranging from 1.2 to 16 nM. Further, NBI31772 suppresses IGFBP-3-mediated inhibition of hIGF-I-induced radioactive thymidine incorporation in 3T3 fibroblasts. Thus, the authors have identified a small molecule that displaces hIGF-I from its binding proteins, which they hope will lead to the release and increased bioavailability of hIGF-I in animals. Two aspects of the authors’ paper bear specific comment. First, NBI31772 was ultimately identified via fractionation of authentic L-3,4-dihydroxyphenylalanine,
TRENDS in Biotechnology Vol.19 No.11 November 2001
following an attempt to reproduce library storage conditions. Although this fraction demonstrated the desired biological activity, it remains unclear whether the active chemical entity in the library stock is in fact NBI31772. Indeed, no attempt was made by the authors to report the exploration of alternative possibilities, only that a biologically active entity had been discovered. Second, although the library sample was described as the ‘best’ inhibitor, no mention was made of the other ‘hits’ obtained in the primary screen, or where it fell among that distribution of hits. What fraction of the library scored positive by the assay criterion? Would the results be different if a more stringent ‘hit’ criterion had been applied? These omissions can be rationalized by considering an endpoint analysis of the authors’ aims. As researchers at a biotechnology company, Liu et al. are reasonably expected to focus exclusively on the goal of a promising lead. To translate such thinking into the academy, however, is less appropriate. Although the
identification of an isolated ‘hit’ in a small molecule screen is potentially interesting as a probe for further study of a particular biological outcome, the statistical results of small molecule screens, particularly when coupled to an examination of structure–activity relationships, might provide general information about the interaction of small molecules with biological systems. Of course, researchers in basic biology must focus their efforts and resources on interesting questions and promising results but the total information content of highthroughput techniques must be made available to the general community for future query.
those with no preference into near teetotalers. The work by Panayotis Thanos et al. (Journal of Neurochemistry, September issue) could have implications for the prevention and treatment of alcoholism in humans. DM
Copycat?
1 Liu, X-J et al. (2001) Identification of a nonpeptide ligand that releases bioactive insulinlike growth factor-I from its binding protein complex. J. Biol. Chem. 276, 32419–32422
Paul A. Clemons [email protected] Bridget K. Wagner [email protected]
In Brief
Latest CF gene therapy bmn.com (6 September) gives a commentary on two reports in the July issue of Human Gene Therapy detailing new clinical trials of gene therapy treatments for cystic fibrosis (CF). CF has long been a good candidate for gene therapy, but results have yet to match initial expectations. The new research by workers at Genzyme Corporation (Framingham, MA, USA) and the University of Cincinnati Medical Center (Cincinnati, OH, USA) describes clinical effects of adenoviral vector or plasmid DNA delivery vehicles of the CF transmembrane conductance regulator (CFTR) gene. Both groups report limited (30–60%) incorporation of the CFTR gene and also noted inflammatory side effects arising from the delivery vehicle. MJD
Gene therapy reduces drinking in ‘alcoholic’ rats Scientists at the US Department of Energy’s Brookhaven National Laboratory (Upton, NY, USA) have shown that increasing the level of a brain protein important for transmitting pleasure signals can turn rats that prefer alcohol into light drinkers, and http://tibtech.trends.com
Helicobacter halted The stomach-ulcer-causing bacterium Helicobacter pylori might soon be literally stopped in its tracks following results published by a team of UK scientists (Microbiology, September issue). David Kelly (University of Sheffield, Sheffield, UK) has identified proteins responsible for sending chemical signals from the bacterial cell surface to the flagellum, which propels the bacteria away from harmful chemicals in its environment. Building on the sequencing of the H. pylori genome, which was published in 1994, the Sheffield group has been studying mutants that have defective chemotaxis genes, and have shown that they can interfere with the sensing and movement properties of H. pylori. Kelly believes that this could be one way to prevent infection and lead to the development of new anti-H. pylori drugs. MJD
For only US$ 1500, a Californian company is offering celebrities the chance to establish dubious copyright over their DNA. ‘A lot of people are going to want to clone people they admire,’ says Andre Crump, president of the DNA Copyright Institute (DCI) (San Francisco, CA, USA). For high-profile individuals worried they might fall victim to being cloned from a few living cells left behind on a glass or exchanged in a handshake, DCI is offering to record their DNA fingerprint, check that it is unique and store it. Ten people have already taken advantage of DCI’s services and for an extra fee, the company will also try to register the pattern with the US Copyright Office (New Scientist, 25 August). DM
Proteomics set for billion dollar growth A report by Front Line Strategic Management Consultants, Inc (Westfield, MA, USA) has forecast that the proteomics market will grow from its present value of US$ 560m to almost US$ 2.8b by 2005, (BioTech International, September issue). This equates to 40% year-on-year growth, and will be driven by the pharmaceutical
0167-7799/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved.