Connecting the Dots Using Gene-Expression Profiles
http://www.100md.com
《新英格兰医药杂志》
Imagine you had a database of the gene-expression signatures of human diseases and their responses to therapeutic drugs. Its use could result in an unprecedented understanding of the interconnectivity of disease pathways and might also lead to the rational design of an optimal treatment or uncover new strategies for treating disease. Such a crystal ball does not yet exist, but a report by Lamb et al.1 offers a glimpse of this vision in a laboratory setting.
The traditional approach to developing new therapeutic drugs involves the painstaking identification of an individual drug target, such as a receptor or ion channel, and the subsequent development of antagonists or agonists to alter the phenotype of the target tissue. Although effective, this approach generally fails to detect unexpected off-target actions of drugs — such as the beneficial effect of an approved drug on another unrelated disease — which are often uncovered through serendipity. Furthermore, the tantalizing prospect that uncharacterized natural products could become effective therapeutics remains poorly realized.
Lamb et al. created a repository (dubbed the "connectivity map") of the gene-expression profiles of human cells, cultured in vitro and exposed to small bioactive molecules in a systematic manner. The first release of the connectivity map contains signatures of 164 distinct small molecules, including drugs approved by the Food and Drug Administration (FDA). It also includes signatures for several diseases, including Alzheimer's disease, the signature for which was obtained through a survey of the literature for reports of differentially expressed genes in the brains of patients with the disease.
The map represents a tool for the large-scale discovery of unexplored connections among small molecules, diseases, and the biologic pathways that join them. Easily accessible through the Internet, it consists of several elements: a large reference database of gene-expression signatures for diseases and small molecules obtained with the use of DNA microarray technology, software that classifies the signatures and compares them with one another, and an interface that allows users to "query" the database using their own signatures of interest to discover connections and form testable hypotheses.
What can the connectivity map do? Investigators can query it with the gene-expression profile of an uncharacterized compound to gain insight into the compound's possible mechanisms of action. Alternatively, they can seek information about whether a known or new bioactive agent has the potential to reverse a disease phenotype. Lamb et al. showed that querying the map with the signature of a drug with a known mechanism of action allows for the identification of additional compounds of similar action but different chemical structure. For example, they queried the map with a gene-expression signature (obtained previously by other investigators) of cultured breast cells exposed to estrogen (estradiol-17) and obtained other estrogen-receptor modulators — agonists and antagonists. This approach, when applied to phenothiazine (an antipsychotic drug), revealed the unintended off-target inhibition of prostaglandin synthesis. Consistent with this finding are the antiinflammatory action of phenothiazines and the established role of prostaglandins in driving inflammation.
The connectivity map has the potential to suggest new treatments for human diseases. The precept of this approach is that small molecules producing signatures that mimic a disease will identify pathways that are potential targets of the therapy of that disease. Conversely, bioactive compounds that induce a "reverse" signature of a disease (i.e., changes in gene expression in a direction opposite to that observed in the disease state) could represent new therapeutic agents. Accordingly, the authors identified a molecule that could reverse an Alzheimer's disease signature in cultured cells, and this same molecule was recently shown to reverse the in vitro formation of neurotoxic A42 fibrils, a hallmark of the pathogenesis of this disease.2
Lamb et al. also uncovered a potential approach to the treatment of acute lymphoblastic leukemia (ALL). In children with ALL, resistance to the standard therapeutic agent, dexamethasone, can develop. The authors used a gene signature of dexamethasone sensitivity — obtained from cultured bone marrow cells — to query the connectivity map, and they discovered that sirolimus induces a very similar gene-expression profile (Figure 1). This led them to hypothesize that sirolimus might sensitize resistant cells to dexamethasone. Supporting the hypothesis was their finding that treatment of a resistant lymphoid cell line with sirolimus rendered the cells sensitive to dexamethasone-dependent apoptosis. Hence, the combination of dexamethasone and sirolimus, an FDA-approved drug, may be an effective approach to treating ALL.
Figure 1. The Concept of the Connectivity Map.
A gene signature (query signature) is obtained with the use of microarray analysis to identify differentially expressed genes (overexpressed shown in green, underexpressed in red) in a cell affected by disease or exposed to a drug, for example. This signature is compared with signatures (one such signature shown with colored bars; overexpressed genes shown in green, underexpressed shown in red, unchanged shown in black) stored in a reference database (Lamb et al.1 call this database the "connectivity map"). Each signature in the database is obtained by means of systematically exposing cultured human cells to a range of drugs; for each drug, the gene expression is compared in two cell cultures: one exposed to a drug and a control exposed to the vehicle only. Pattern-matching software searches the signatures in the database (shown as colored bars) to find those that are most similar (positive) and most dissimilar (negative) to the query signature (shown as colored circles) and generates a "connectivity score." Lamb et al. showed that this approach can be used to identify molecules that result in a gene-expression profile that mimics or reverses that induced by disease. They also showed that exposure to sirolimus results in a profile that mimics the dexamethasone-sensitive profile of bone marrow cells and that sirolimus restores sensitivity to dexamethasone when administered to resistant cells in vitro. It is therefore a candidate drug for restoring dexamethasone sensitivity in children with ALL. Adapted from Lamb et al.,1 with the permission of the publisher.
The connectivity map project may uncover unanticipated, meaningful relations between small molecules and human diseases, and since it is a public resource, it empowers the research community. Nonetheless, its usefulness remains difficult to gauge until we appreciate the extent to which genomic signatures obtained from in vitro experiments recapitulate the complexities of human disease.
Dr. Gullans reports holding equity in and being an employee of RxGen. No other potential conflict of interest relevant to this article was reported.
Source Information
From RxGen, New Haven, CT.
References
Lamb J, Crawford ED, Peck D, et al. The connectivity map: using gene-expression signatures to connect small molecules, genes, and disease. Science 2006;313:1929-1935.
Blanchard BJ, Chen A, Rozeboom LM, Stafford KA, Weigele P, Ingram VM. Efficient reversal of Alzheimer's disease fibril formation and elimination of neurotoxicity by a small molecule. Proc Natl Acad Sci U S A 2004;101:14326-14332.(Steven R. Gullans, Ph.D.)
The traditional approach to developing new therapeutic drugs involves the painstaking identification of an individual drug target, such as a receptor or ion channel, and the subsequent development of antagonists or agonists to alter the phenotype of the target tissue. Although effective, this approach generally fails to detect unexpected off-target actions of drugs — such as the beneficial effect of an approved drug on another unrelated disease — which are often uncovered through serendipity. Furthermore, the tantalizing prospect that uncharacterized natural products could become effective therapeutics remains poorly realized.
Lamb et al. created a repository (dubbed the "connectivity map") of the gene-expression profiles of human cells, cultured in vitro and exposed to small bioactive molecules in a systematic manner. The first release of the connectivity map contains signatures of 164 distinct small molecules, including drugs approved by the Food and Drug Administration (FDA). It also includes signatures for several diseases, including Alzheimer's disease, the signature for which was obtained through a survey of the literature for reports of differentially expressed genes in the brains of patients with the disease.
The map represents a tool for the large-scale discovery of unexplored connections among small molecules, diseases, and the biologic pathways that join them. Easily accessible through the Internet, it consists of several elements: a large reference database of gene-expression signatures for diseases and small molecules obtained with the use of DNA microarray technology, software that classifies the signatures and compares them with one another, and an interface that allows users to "query" the database using their own signatures of interest to discover connections and form testable hypotheses.
What can the connectivity map do? Investigators can query it with the gene-expression profile of an uncharacterized compound to gain insight into the compound's possible mechanisms of action. Alternatively, they can seek information about whether a known or new bioactive agent has the potential to reverse a disease phenotype. Lamb et al. showed that querying the map with the signature of a drug with a known mechanism of action allows for the identification of additional compounds of similar action but different chemical structure. For example, they queried the map with a gene-expression signature (obtained previously by other investigators) of cultured breast cells exposed to estrogen (estradiol-17) and obtained other estrogen-receptor modulators — agonists and antagonists. This approach, when applied to phenothiazine (an antipsychotic drug), revealed the unintended off-target inhibition of prostaglandin synthesis. Consistent with this finding are the antiinflammatory action of phenothiazines and the established role of prostaglandins in driving inflammation.
The connectivity map has the potential to suggest new treatments for human diseases. The precept of this approach is that small molecules producing signatures that mimic a disease will identify pathways that are potential targets of the therapy of that disease. Conversely, bioactive compounds that induce a "reverse" signature of a disease (i.e., changes in gene expression in a direction opposite to that observed in the disease state) could represent new therapeutic agents. Accordingly, the authors identified a molecule that could reverse an Alzheimer's disease signature in cultured cells, and this same molecule was recently shown to reverse the in vitro formation of neurotoxic A42 fibrils, a hallmark of the pathogenesis of this disease.2
Lamb et al. also uncovered a potential approach to the treatment of acute lymphoblastic leukemia (ALL). In children with ALL, resistance to the standard therapeutic agent, dexamethasone, can develop. The authors used a gene signature of dexamethasone sensitivity — obtained from cultured bone marrow cells — to query the connectivity map, and they discovered that sirolimus induces a very similar gene-expression profile (Figure 1). This led them to hypothesize that sirolimus might sensitize resistant cells to dexamethasone. Supporting the hypothesis was their finding that treatment of a resistant lymphoid cell line with sirolimus rendered the cells sensitive to dexamethasone-dependent apoptosis. Hence, the combination of dexamethasone and sirolimus, an FDA-approved drug, may be an effective approach to treating ALL.
Figure 1. The Concept of the Connectivity Map.
A gene signature (query signature) is obtained with the use of microarray analysis to identify differentially expressed genes (overexpressed shown in green, underexpressed in red) in a cell affected by disease or exposed to a drug, for example. This signature is compared with signatures (one such signature shown with colored bars; overexpressed genes shown in green, underexpressed shown in red, unchanged shown in black) stored in a reference database (Lamb et al.1 call this database the "connectivity map"). Each signature in the database is obtained by means of systematically exposing cultured human cells to a range of drugs; for each drug, the gene expression is compared in two cell cultures: one exposed to a drug and a control exposed to the vehicle only. Pattern-matching software searches the signatures in the database (shown as colored bars) to find those that are most similar (positive) and most dissimilar (negative) to the query signature (shown as colored circles) and generates a "connectivity score." Lamb et al. showed that this approach can be used to identify molecules that result in a gene-expression profile that mimics or reverses that induced by disease. They also showed that exposure to sirolimus results in a profile that mimics the dexamethasone-sensitive profile of bone marrow cells and that sirolimus restores sensitivity to dexamethasone when administered to resistant cells in vitro. It is therefore a candidate drug for restoring dexamethasone sensitivity in children with ALL. Adapted from Lamb et al.,1 with the permission of the publisher.
The connectivity map project may uncover unanticipated, meaningful relations between small molecules and human diseases, and since it is a public resource, it empowers the research community. Nonetheless, its usefulness remains difficult to gauge until we appreciate the extent to which genomic signatures obtained from in vitro experiments recapitulate the complexities of human disease.
Dr. Gullans reports holding equity in and being an employee of RxGen. No other potential conflict of interest relevant to this article was reported.
Source Information
From RxGen, New Haven, CT.
References
Lamb J, Crawford ED, Peck D, et al. The connectivity map: using gene-expression signatures to connect small molecules, genes, and disease. Science 2006;313:1929-1935.
Blanchard BJ, Chen A, Rozeboom LM, Stafford KA, Weigele P, Ingram VM. Efficient reversal of Alzheimer's disease fibril formation and elimination of neurotoxicity by a small molecule. Proc Natl Acad Sci U S A 2004;101:14326-14332.(Steven R. Gullans, Ph.D.)