Hot off the Press

In the Hot off the Press section of Molecular BioSystems members of the Editorial Board and their research groups highlight recent literature for the benefit of the community. This month the highlighted topics include X-ray structures of a bacterial drug-resistance transporter, a new method for the exploration of intra- and interchromosomal interactions and some items published recently in the RSC’s journals.

---

Global gene expression signatures for global chemical probe and drug discovery

High throughput methodologies have made a major impact on biology, chemistry and the activities at their interface, particularly chemical biology and drug discovery (Collins I. and Workman P., Nat. Chem. Biol., 2006, in press). The human genome sequence facilitates identification of new drug targets and has enabled the development of large scale genomic technologies such as genome-wide mRNA expression profiling (Clarke et al., Eur. J. Cancer, 2006, 40, 2560–2591). Three recent papers have further illustrated the power of gene expression profiling coupled to pattern-matching analysis to provide a systematic approach to elucidating the molecular connections between disease states and the effects of small molecule perturbations. In particular, they exemplify how gene expression signatures can be used to identify compounds with a desired mechanism of action—or to understand the mechanism of a compound with particular phenotypic properties. Linkage of compound expression profiles to those of disease states suggested possible new treatments. Of perhaps the greatest significance, the authors of the pilot project suggest its extension to provide a freely available community-wide resource, very much in the spirit of public genome projects. One of the suggested applications is that a publicly available compendium of global expression signatures for a very large compound library could be used to accelerate global drug discovery.

The three papers are based on the work at the Broad Institute of Massachusetts Institute of Technology and Harvard University. In the first, overarching article, Lamb et al. provide what they refer to as the ‘first generation connectivity map which links genome-wide mRNA profiles for various perturbed states’. This concept is not in itself new. Previous studies have shown that gene expression signatures induced by perturbagens such as drugs can be used as a diagnostic signature (Marton et al., Nat. Med., 1998, 4, 1293–1301; Hughes et al., Cell, 2000, 102, 109–126; Brownet al., Mol. Syst. Biol., 2006, 2, e-pub.). In 2000, Hughes et al. established a compendium of 300 gene expression profiles corresponding to mutations in the yeast S. cerevisiae and showed that these could be used to identify mechanisms of drug action. This complements the barcode-based chemical-genetic profiling approach in which drug mechanism is determined from patterns of sensitivity across 5000 yeast strains (Parsons et al., Cell, 2006, 126, 611–625). Pharmaceutical companies are using databases of gene expression responses in rat tissues following systemic compound administration to predict potential toxicities of new agents (Ganter et al., J. Biotechnol., 2005, 119, 219–244; Fielden et al., Toxicol. Pathol., 2005, 33, 675–683). In cancer research, drug-induced global mRNA signatures in malignant cells have been used to investigate the mechanism of action and discover potential biomarkers, as with inhibitors of the molecular chaperone HSP90 (Clarke et al., Oncogene, 2000, 19, 4125–4133) and histone deacetylase (HDAC; Glaser et al., Mol. Cancer Ther., 2003, 2, 151–163; also reviewed in Clarke et al., 2006).

Lamb et al. profiled 164 small molecule perturbagens, including US Food and Drug Administration-approved drugs and bioactive tool compounds. This is the first very large scale study to profile mammalian cell lines in this way. Given the enormous number of permutations of cell line, time and concentration, compromises had to be made. In most cases, compounds were used at a single concentration of 10 μM and an exposure time of 6 h in the well-characterized breast epithelial cell line MCF7. Additional conditions and cell lines were used in some cases. To facilitate the profiling of large numbers of samples, RNA was prepared using a semi-automated system and profiled using a newly developed higher throughput Affymetrix GeneChip microarray platform that allowed the profiling of ~22K genes in up to 96 samples in parallel. A total of 564 gene expression profiles were obtained, relating to 453 individual treatment-vehicle pairs. Gene signature comparisons were performed using rank-based pattern matching using the Kolmogorov-Smimov statistic. The data are freely available at http://www.broad.mit.edu/cmap. The approach was validated by comparison with external data for HDAC inhibitors, estrogens and phenothiazines. Following validation, several examples that illustrate the usefulness of the gene expression ‘Connectivity Map’ were provided. The published signature for the obese state was used to establish a link with peroxisome proliferator-activated receptor gamma (PPARg) agonists which are potent inducers of adipogenesis. Signatures associated with Alzheimer’s disease were connected to 4,5-dianilinopthalimide, an agent which can reverse the formation fibrils that have been causally related to neuronal cell death in the disease. A signature of dexamethasone resistance in acute lymphoblastic leukaemia (ALL) exhibited strong connectivity to rapamycin (sirolimus). Follow-up studies showed that rapamycin conferred dexamethasone sensitivity via an mTOR-dependent mechanism, involving decreased expression of the anti-apoptotic protein MCL1. The results suggest a combination of rapamycin and dexamethasone in ALL and exemplify a general strategy for overcoming resistance based on reversal of a resistance signature.

In another application, gene expression profiles were used to identify a new class of HSP90 pathway inhibitors. The original search was for agents that would revert the gene expression signature of an androgen-activated state to that of the non-proliferative androgen-deprived state in androgen-dependent prostate cancer cells. Among the hits were a large number of triterpenoid celastrol and gedunin analogues. The Connectivity Map was used to reveal a very strong similarity in the expression signatures of celastrol, gedunin and the established HSP90 inhibitors, namely geldanamycin analogues and radicicol. The molecular chaperone HSP90 regulates the conformational stability and activation of a range of ‘client’ proteins, including the androgen receptor and many kinases involved in cancer (Whitesell L. and Lindquist S. L., Nat. Rev. Cancer, 2005, 5, 761–772). Celastrol and gedunin were shown to have the same effects as 17-AAG in cells, all inducing the depletion of androgen receptor, BCR-ABL, mutant-FLT3 and EGF-receptor. They also prevented the essential ATP binding to HSP90 and celastrol reduced the interaction of HSP90 with the co-chaperone p23—another feature of HSP90 inhibitors. Celastrol and gedunin did not, however, bind to the N-terminal ATP pocket of HSP90, in contrast to most known inhibitors (McDonald et al., Curr. Top. Med. Chem., 2006, 6, 1193–1203), suggesting a novel mechanism of action. The effects were not totally surprising since celastrol (a constituent of the Chinese medicinal plant known as the Thunder of God Vine) is known to induce a heat shock response (as do ATP site HSP90 inhibitors), through activation of heat shock factor 1, as well as acting as a proteasome inhibitor, leading to accumulation of ubiquinated proteins (Westerheide et al., J. Biol. Chem., 2006, 281, 9616–9622; Yang et al., Cancer Res., 2006, 66, 4758–4765). Further studies are required to elucidate precisely how celastrol and gedunin inhibit HSP90 function: this may involve direct effects outwith the N-terminal ATP site, as with novabiocin and cisplatin, or at other sites, potentially affecting the binding of co-chaperones. The findings may be clinically significant, since the geldanamycin analogue 17-AAG shows promise in the clinic (Yang et al, 2006) and celastrol inhibits the growth of prostate cancer xenografts (Banerji et al., Clin. Oncol., 2005, 23, 4152–4161).

Together with previous profiling studies discussed above, the three featured papers clearly exemplify the value of gene expression data bases in mechanistic studies and potentially in drug discovery. Lamb et al. propose an expansion, initially involving the profiling of all US Food and Drug Administration (FDA)-approved drugs, together with inhibitory RNAs against ten diverse cell lines. The choice of cell line together with drug concentration and time of exposure will need to be carefully considered. The ultimate goal would be to provide a comprehensive compendium of gene expression covering all cellular states and very large chemical libraries. This could act as a global resource for mechanistic exploration and the discovery of chemical probes.

Other not unrelated public domain projects are already well underway. The US National Cancer Institute’s Initiative for Chemical Genetics is a publicly accessible research facility for high throughput screening (HTS) against diverse compound libraries (Tolliday et al., Cancer Res., 2006, 66, 8935–8942). Emphasis is on phenotypic or chemical genetic screens. Data are publicly available in Chembank (http://chembank.broad.harvard.edu/). The US National Institutes of Health Molecular Libraries Initiative, part of the NIH Roadmap for Medical Research, uses HTS to identify small molecule chemical probes acting on biological targets (Austin et al., Science, 2006, 306, 1138–1139). The focus is on targets that are viewed by the pharmaceutical industry as not easily druggable. Data are publicly available via Pubchem (http://pubchem.ncbi.nlm.nih.gov/). Patenting the chemical probes is discouraged. There seems little doubt that public initiatives of this type will yield some useful chemical probes. The availability of readily searchable databases containing the biological effects of large diverse compound libraries would be of great potential value for research, in the same way that the human genome sequence has proved invaluable. The inclusion of gene expression signatures would add considerable value. Characterization of chemical probes across thousand of targets and processes will enhance their utility as tools that are complementary to genetic reagents like RNA interference. Unnecessary duplication of research would be reduced. The vision of finding a chemical probe for every protein (Schreiber S. L., Bioorg. Med. Chem., 1998, 6, 1127–1152) will become more tangible.

What is not yet clear is how much these initiatives will directly aid drug discovery. The chemical probes will certainly help target validation and the discovery of biomarkers. But chemical probes are not the same as drugs with respect to many properties and the value of the probes for use in animal models will generally be very limited (Lipinski C. and Hopkins A., Nature, 2004, 432, 855–861). The intellectual property issue is also a significant one. Nevertheless, the value to biological and medical research of a range of accessible, well-annotated chemical probes does in our view support the development of ‘open source’ chemical biology initiatives.

 

Lamb J, Crawford ED, Peck D, Modell JW, Blat IC, Wrobel MJ, Lerner J, Brunet JP, Subramanian A, Ross KN, Reich M, Hieronymus H, Wei G, Armstrong SA, Haggarty SJ, Clemons PA, Wei R, Carr SA, Lander ES, Golub TR, Science, 2006, 313, 1929–1935.

Hieronymus H, Lamb J, Ross KN, Peng XP, Clement C, Rodina A, Nieto M, Du J, Stegmaier K, Raj SM, Maloney KN, Clardy J, Hahn WC, Chiosis G, Golub TR, Cancer Cell, 2006, 10, 321–330.

Wei G, Twomey D, Lamb J, Schlis K, Agarwal J, Stam RW, Opferman JT, Sallan SE, den Boer ML, Pieters R, Golub TR, Armstrong SA, Cancer Cell, 2006, 10, 331–342.

Reviewed by: Paul Workman and Paul A. Clarke, The Institute of Cancer Research, UK

The authors are funded by Cancer Research UK Programme Grant Number [CUK] C309/A2187. Paul Workman is a Cancer Research UK Life Fellow.

Structural asymmetry of a bacterial drug-resistance transporter suggests a peristaltic pump mechanism

Resistance to certain drugs in Escherichia coli is mediated by a transporter that spans the inner and outer membranes, comprising the proteins AcrA, AcrB and TolC. Part of TolC forms a β barrel across the outer membrane, while the rest of this structure is a tube that spans the periplasmic space. Using the proton motive force, AcrB in the inner membrane feeds the undesirable drug molecules from the cytoplasm and periplasm into TolC, driving them from the cell. AcrA serves to connect these two major components of the efflux system, which is remarkable in its ability to mediate the transport of a wide variety of structurally unrelated small charged and neutral molecules, including various antibiotics. X-Ray structures of trimeric AcrB have now been determined by using crystals that allow the individual structures of each monomer to be distinguished, which for technical reasons was not the case in an earlier study. Three different monomer conformations are revealed, which are interpreted to represent the consecutive states of a transport cycle and suggest a transport mechanism that differs from that which is generally accepted. In the new mechanism, the substrate accesses the extramembraneous, periplasmic domain of AcrB through the transmembrane domain of one of the three subunits, rather than the central transmembrane pore, which is thought to contain lipid. The extramembraneous domain then undergoes a conformational change that traps the substrate in an internal hydrophobic pocket. In a second conformational change, the substrate is released into the TolC tube. Again, transport through the central pore of the periplasmic domain of AcrB, as proposed earlier, is ruled out, in this case because the opening at the periplasmic end is too narrow for known substrates. The sequential structural changes are driven by “downhill” proton transport (periplasm to cytoplasm) through AcrB at a site distinct from those involved in substrate movement. This is an elegant mechanism that bears some resemblance to that established for ATP generation by F₁F_O-ATPase. However, crucial supporting data are lacking. In the absence of structures with bound substrate and the lack of direct evidence for the central transmembrane pore being filled with lipid, the new mechanism remains speculative. Additional biochemical and biophysical studies are required to substantiate the proposed peristaltic mechanism of multi-drug transport.

 

Seeger MA, Schiefner A, Eicher T, Verrey F, Diederichs K, Pos KM., Science, 2006, 313(5791), 1295–8.

Reviewed by: Lakmal Jayasinghe and Amy Mason, University of Oxford

4C approach: a new tool for genome-wide exploration of intra- and interchromosomal interactions

In eukaryotes, huge lengths of chromosomes as well as the need for expression of different genes in different tissues and/or at different stages of cell differentiation were driving force for development of sophisticated regulatory systems that govern the genome function. There is a growing body of data supporting the idea that physical association between non-homologous chromosomes may impact their gene expression. Most of the studies on chromosome interactions use chromosomal conformation capture technique (3C) based on cross-linking of physically touching segments of genome via contacts between their DNA-bound proteins. The interacting partners are digested with certain endonucleases, ligated and quantified by PCR (Fig. 1). However, prior knowledge about interacting complexes is required for the 3C method to be applied. Now, geneticists in Sweden describe a related circular chromosome conformation capture (4C) approach, which requires the knowledge of sequence of only one site, whose interaction with any other sites is under study(Fig. 2). Using 4C, researchers were able to uncover the networks of intra- and interchromosomal interactions specific to the maternal imprinting control region H19 in course of embryonic stem cells maturation. Strikingly, experimental data show that the H19 region can touch several different chromosomes at once. This work demonstrates a new layer of complexity in spatial conformations of chromosomal machinery and one more time emphasizes the importance of epigenetic events for proper functioning of eukaryotic genome.

Fig. 1 Section a illustrates the 3C approach, which the 4C approach improves upon. Reprinted (abstracted/excerpted) with permission from Dekker et al., Science, 2002, 295, 1306–1311. Copyright 2002 AAAS.

Fig. 2 Schematic representation of the 4C assay. The red and blue arrows and rectangles indicate the nested primers within the plasmid. Reprinted by permission from Macmillan Publishers Ltd: Nature Genetics, Zhihu Zhao, Gholamreza Tavoosidana, Mikael Sjölinder, Anita Göndör, Piero Mariano, Sha Wang, Chandrasekhar Kanduri, Magda Lezcano, Kuljeet Singh Sandhu, Umashankar Singh, Vinod Pant, Vijay Tiwari, Sreenivasulu Kurukuti and Rolf Ohlsson, DOI: 10.1038/ng1891, copyright 2006

 

Zhihu Zhao, Gholamreza Tavoosidana, Mikael Sjölinder, Anita Göndör, Piero Mariano, Sha Wang, Chandrasekhar Kanduri, Magda Lezcano, Kuljeet Singh Sandhu, Umashankar Singh, Vinod Pant, Vijay Tiwari, Sreenivasulu Kurukuti and Rolf Ohlsson, Nature Genetics. Published online: 8 October 2006, DOI:10.1038/ng1891

Reviewed by: Bohdan Ostash, Harvard Medical School

Hot off the RSC Press

The future of proteomics

Microfluidic techniques will lead to high-throughput protein profiling within a decade, suggest scientists.

Aaron Wheeler and Sergio Freire from the University of Toronto, Canada, have looked at the future of proteomics, the study of all the proteins coded by the genome, in the light of existing microfluidics technologies.

Currently, most protein profiles are generated through a series of steps which can take many days to perform. Wheeler and Freire looked towards their specialism, microfluidics, for a solution and describe microfluidics and protein profiling as a ‘natural fit.’

Proteomics has become an increasingly popular research topic, said Wheeler, and its ‘increasing importance has not been accompanied by a corresponding improvement in analytical tools.’ Whilst genomics, the study of the entire genome, has a wide range of high-throughput techniques associated with it, proteomics still relies heavily on low-throughput techniques.

The reasons for this are manifold, explained Wheeler. Samples can contain a vast number of proteins, some of which are structurally similar but functionally different. Also, there is often a large difference between the concentrations of the least and most abundant proteins and the varying properties of the proteins mean that experimental conditions cannot be standardised. Wheeler and Freire suggest that microfluidics can reduce reagent consumption and decrease analysis time by integrating multiple processes.

Whilst there are still hurdles to overcome, the scientists conclude that microfluidic tools for high-throughput proteomics will be a reality in the near future. Larry Kricka, of the University of Pennsylvania, US, agrees that 10 years is a realistic timeframe, adding, ‘microfluidics and the associated microminiaturisation technologies represent one of the most promising directions in analysis.’

 

SLS Freire and AR Wheeler, Lab Chip, 2006, DOI: 10.1039/b609871a

Reviewed by: Laura Howes, Royal Society of Chemistry, Cambridge, UK.

Optical tweezers for genomics

Scientists in the US have illuminated the state-of-the-art in optical tweezer research, using laser beams to trap and move small biological objects.

Piero Bianco and Yuji Kimura from the University at Buffalo, highlight how researchers can now use optical tweezers to study interactions between DNA and proteins.

Optical tweezers, or optical traps, are conceptually like eyebrow tweezers. An infrared laser is focused through an optical microscope on a precise spot (Fig. 3), trapping objects as small as several nanometres across. Moving the laser beam allows these objects to be picked up and moved.

Numerous DNA binding proteins have been studied using optical tweezers, including members of the polymerase family. Polymerases are responsible for replicating DNA and transcribing DNA into RNA. Scientists have used optical tweezers to show how a polymerase pauses and back tracks during transcription. This is consistent with a proofreading process, said Bianco.

Ulf Landegren from Uppsala University, Sweden, said that the research is proving ‘fundamental for understanding how the genome is regulated.’

Optical tweezers have significantly contributed to the understanding of the dynamic behaviour of individual protein molecules at single molecule levels, said Bianco. He predicts that the method will eventually help scientists to discover how breaks in double stranded DNA are repaired.

Fig. 3 Schematics showing the principle of optical tweezers based on ray optics. The ability to trap and manipulate small objects, such as polystyrene beads, results from light possessing momentum which is in the direction of propagation of the beam. Reproduced from Analyst, 2006, 131, 868 by permission of The Royal Society of Chemistry.

 

Y Kimura and PR Bianco, Analyst, 2006, 131, 868

Reviewed by: Nina Athey-Pollard, Royal Society of Chemistry, Cambridge, UK.

---