Arrays of arrays for high-throughput gene expression profiling.

Gene expression profiling using DNA arrays is rapidly becoming an essential tool for research and drug discovery and may soon play a central role in disease diagnosis. Although it is possible to make significant discoveries on the basis of a relatively small number of expression profiles, the full potential of this technology is best realized through more extensive collections of expression measurements. The generation of large numbers of expression profiles can be a time-consuming and labor-intensive process with current one-at-a-time technology. We have developed the ability to obtain expression profiles in a highly parallel yet straightforward format using glass wafers that contain 49 individual high-density oligonucleotide arrays. This arrays of arrays concept is generalizable and can be adapted readily to other types of arrays, including spotted cDNA microarrays. It is also scalable for use with hundreds and even thousands of smaller arrays on a single piece of glass. Using the arrays of arrays approach and parallel preparation of hybridization samples in 96-well plates, we were able to determine the patterns of gene expression in 27 ovarian carcinomas and 4 normal ovarian tissue samples, along with a number of control samples, in a single experiment. This new approach significantly increases the ease, efficiency, and throughput of microarray-based experiments and makes possible new applications of expression profiling that are currently impractical.

Analysis of gene expression profiles in normal and neoplastic ovarian tissue samples identifies candidate molecular markers of epithelial ovarian cancer.

Epithelial ovarian cancer is the leading cause of death from gynecologic cancer, in part because of the lack of effective early detection methods. Although alterations of several genes, such as c-erb-B2, c-myc, and p53, have been identified in a significant fraction of ovarian cancers, none of these mutations are diagnostic of malignancy or predictive of tumor behavior over time. Here, we used oligonucleotide microarrays with probe sets complementary to >6,000 human genes to identify genes whose expression correlated with epithelial ovarian cancer. We extended current microarray technology by simultaneously hybridizing ovarian RNA samples in a highly parallel manner to a single glass wafer containing 49 individual oligonucleotide arrays separated by gaskets within a custom-built chamber (termed "array-of-arrays"). Hierarchical clustering of the expression data revealed distinct groups of samples. Normal tissues were readily distinguished from tumor tissues, and tumors could be further subdivided into major groupings that correlated both to histological and clinical observations, as well as cell type-specific gene expression. A metric was devised to identify genes whose expression could be considered ideal for molecular determination of epithelial ovarian malignancies. The list of genes generated by this method was highly enriched for known markers of several epithelial malignancies, including ovarian cancer. This study demonstrates the rapidity with which large amounts of expression data can be generated. The results highlight important molecular features of human ovarian cancer and identify new genes as candidate molecular markers.

Optimizing the substrate specificity of a group I intron ribozyme.

Group I ribozymes can repair mutant RNAs via trans-splicing. Unfortunately, substrate specificity is quite low for the trans-splicing reaction catalyzed by the group I ribozyme from Tetrahymenathermophila. We have used a systematic approach based on biochemical knowledge of the function of the Tetrahymena ribozyme to optimize its ability to discriminate against nonspecific substrates in vitro. Ribozyme derivatives that combine a mutation which indirectly slows down the rate of the chemical cleavage step by weakening guanosine binding with additional mutations that weaken substrate binding have greatly enhanced specificity with short oligonucleotide substrates and an mRNA fragment derived from the p53 gene. Moreover, compared to the wild-type ribozyme, reaction of a more specific ribozyme with targeted substrates is much less sensitive to the presence of nonspecific RNA competitors. These results demonstrate how a detailed understanding of the biochemistry of a catalytic RNA can facilitate the design of customized ribozymes with improved properties for therapeutic applications.

Kinetic intermediates trapped by native interactions in RNA folding.

In the magnesium ion-dependent folding of the Tetrahymena ribozyme, a kinetic intermediate accumulates in which the P4-P6 domain is formed, but the P3-P7 domain is not. The kinetic barriers to P3-P7 formation were investigated with the use of in vitro selection to identify mutant RNA molecules in which the folding rate of the P3-P7 domain was increased. The critical mutations disrupt native tertiary interactions within the P4-P6 domain and increase the rate of P3-P7 formation by destabilizing a kinetically trapped intermediate. Hence, kinetic traps stabilized by native interactions, and not simply by mispaired nonnative structures, can present a substantial barrier to RNA folding.

Probing the interplay between the two steps of group I intron splicing: competition of exogenous guanosine with omega G.

One largely unexplored question about group I intron splicing is how the cleavage and ligation steps of the reaction are coordinated. We describe a simple in vitro trans-splicing model system in which both steps take place, including the exchange of ligands in the guanosine-binding site that must occur between the two steps. Using this model system, we show that the switch is accomplished by modulating the relative affinity of the binding site for the two ligands. While the terminal guanosine of the intron (omegaG) and exogenous guanosine compete for binding during the first step of splicing, no competition is apparent during the second step, when omegaG is bound tightly. These results help explain how the ribozyme orchestrates progression through the splicing reaction. In addition to providing a new tool to ask basic questions about RNA catalysis, the trans-splicing model system will also facilitate the development of therapeutically useful group I ribozymes that can repair mutant mRNAs.

Slow folding kinetics of RNase P RNA.

Understanding the folding mechanisms of large, highly structured RNAs is important for understanding how these molecules carry out their function. Although models for the three-dimensional architecture of several large RNAs have been constructed, the process by which these structures are formed is only now beginning to be explored. The kinetic folding pathway of the Tetrahymena ribozyme involves multiple intermediates and both Mg2+-dependent and Mg2+-independent steps. To determine whether this general mechanism is representative of folding of other large RNAs, a study of RNase P RNA folding was undertaken. We show, using a kinetic oligonucleotide hybridization assay, that there is at least one slow step on the folding pathway of RNase P RNA, resulting in conformational changes in the P7 helix region on the minute timescale. Although this folding event requires the presence of Mg2+, the slow step itself does not involve Mg2+ binding. The P7 and P2 helix regions exhibit distinctly different folding behavior and ion dependence, implying that RNase P folding is likely to be a complex process. Furthermore, there are distinct similarities in the folding of RNase P RNA from both Bacillus subtilis and Escherichia coli, indicating that the folding pathway may also be conserved along with the final structure. The slow folding kinetics, Mg2+-independence of the rate, and existence of intermediates are basic features of the folding mechanism of the Tetrahymena group I intron that are also found in RNase P RNA, suggesting these may be general features of the folding of large RNAs.

The kinetic folding pathway of the Tetrahymena ribozyme reveals possible similarities between RNA and protein folding.

We have probed the nature of the individual kinetic steps in the folding of the Tetrahymena ribozyme by studying the folding kinetics of mutant ribozymes. After rapid formation of the first structural subdomain, a slow step precedes stable formation of the second subdomain. The two central helices of the second subdomain form in an interdependent manner, and this structural subunit therefore also constitutes a kinetic folding unit. The slow folding step includes formation of tertiary interactions in a triple-helical scaffold that orients the two subdomains of the RNA. The rapid and early formation of short range secondary structure, the hierarchical formation of kinetic folding units corresponding to structural subdomains, and the formation of tertiary interactions between subdomains late during the folding process appear to be common features of the folding mechanism for both RNA and proteins.

The P9.1-P9.2 peripheral extension helps guide folding of the Tetrahymena ribozyme.

We have previously proposed a hierarchical model for the folding mechanism of the Tetrahymena ribozyme that may illustrate general features of the folding pathways of large RNAs. While the role of elements in the conserved catalytic core of this ribozyme during the folding process is beginning to emerge, the participation of non-conserved peripheral extensions in the kinetic folding mechanism has not yet been addressed. We now show that the 3'-terminal P9.1-P9.2 extension of the Tetrahymena ribozyme plays an important role during the folding process and appears to guide formation of the catalytic core.

Kinetic intermediates in RNA folding.

The folding pathways of large, highly structured RNA molecules are largely unexplored. Insight into both the kinetics of folding and the presence of intermediates was provided in a study of the Mg(2+)-induced folding of the Tetrahymena ribozyme by hybridization of complementary oligodeoxynucleotide probes. This RNA folds via a complex mechanism involving both Mg(2+)-dependent and Mg(2+)-independent steps. A hierarchical model for the folding pathway is proposed in which formation of one helical domain (P4-P6) precedes that of a second helical domain (P3-P7). The overall rate-limiting step is formation of P3-P7, and takes place with an observed rate constant of 0.72 +/- 0.14 minute-1. The folding mechanism of large RNAs appears similar to that of many multidomain proteins in that formation of independently stable substructures precedes their association into the final conformation.

Unique fatty acid composition of normal cartilage: discovery of high levels of n-9 eicosatrienoic acid and low levels of n-6 polyunsaturated fatty acids.

We report here the finding that normal, young cartilages, in distinction from all other tissues examined, have unusually high levels of n-9 eicosatrienoic (20:3 cis-delta 5,8,11) acid and low levels of n-6 polyunsaturated fatty acids (n-6 PUFA). This pattern is identical to that found in tissues of animals subjected to prolonged depletion of nutritionally essential n-6 polyunsaturated fatty acids (EFA). This apparent deficiency is consistently observed in cartilage of all species so far studied (young chicken, fetal calf, newborn pig, rabbit, and human), even though levels of n-6 PUFA in blood and all other tissues is normal. The n-9 20:3 acid is particularly abundant in phosphatidylethanolamine, phosphatidylinositol, and the free fatty acid fractions from the young cartilage. Several factors appear to contribute to the reduction in n-6 PUFA and the appearance of high levels of the n-9 20:3 acid in cartilage: 1) limited access to nutritional sources of EFA due to the impermeability and avascularity of cartilage, 2) rapid metabolism of n-6 PUFA to prostanoids by chondrocytes, and 3) a unique fatty acid metabolism by cartilage. Evidence is presented that each of these factors contributes. Previously, EFA deficiency has been shown to greatly suppress the inflammatory response of leukocytes and rejection of tissues transplanted into allogeneic recipients. Because eicosanoids, which are derived from EFA, have been implicated in the inflammatory responses associated with arthritic disease, reduction of n-6 PUFA and accumulation of the n-9 20:3 acid in cartilage may be important for maintaining normal cartilage structure.