| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Editorial |
Department of Medicine, Johns Hopkins Bayview Medical Center, Baltimore, Maryland 20854
Address all correspondence and requests for reprints to: Dr. Paul M. Yen, Department of Medicine, Johns Hopkins Bayview Medical Center, 4940 Eastern Avenue, Baltimore, Maryland 21224. E-mail: pyen3{at}jhmi.edu.
In science, and in life, it is difficult to distinguish the trees from the forest. Often we are caught up in the minutiae that immediately confront us and we form only a limited picture of our nearby environs without any awareness of the broader horizons. In this connection, the recent application of microarrays to biological systems has revolutionized how we understand gene transcription by transforming our vision from the previous perspective of a few individual target genes to a panoramic view of the entire genome. For the first time, we can see the extent and grandeur of the forest, and it is thrilling! At the same time, particularly with the aid of powerful computer-based software, we can study individual members that are located anywhere in the genome and see how they can be related to each another.
Shortly after the successful demonstration of microarray prototypes (1, 2), oligonucleotide- and cDNA-based microarrays were used to investigate a wide range of biological phenomena ranging from cell cycle changes, differentiation, oncogenesis, and aging (3). The potential to study pharmacological and hormonal effects on gene expression profiles was readily appreciated, and studies of thyroid hormone (TH) action were among the first to take advantage of the new microarray technology (4, 5, 6). The earliest studies were performed in rodent liver and pituitary cells and yielded several unexpected findings. First, there were literally hundreds of target genes regulated at the transcriptional level by TH. Previously, only about 30 extant genes were known to be directly regulated by TH (7). Secondly, and surprisingly at the time, many of these genes were negatively regulated by TH. Only a handful of negatively regulated target genes had been identified previously, and most of them were expressed in the pituitary. Lastly, the abundance of newly identified target genes showed the breadth of TH effects on gene transcription and pointed to cellular pathways regulated by TH that hitherto had not been recognized.
Finding new target genes has turned out to be a major impetus for the use of microarrays in drug discovery and hormonal regulation. When such studies are conducted in cell culture or in vivo, several issues need to be considered. The regulation of mRNA from putative new target genes needs to be confirmed. Although repeat microarrays on the same samples are highly reproducible, it is important to perform multiple microarrays on samples from different experiments to address the issue of biological variability. Setting appropriate cut-off criteria for the fold increase or decrease of target genes also is important as it determines the sensitivity and specificity of the analyses. Additionally, when individual genes are investigated further, it may be important to use other techniques to measure mRNA expression such as Northern blotting or real-time RT-PCR to verify the hormonally induced changes, particularly if the changes are modest. As is the case for studying hormonal regulation by more conventional methods, dose-response and time-course patterns are important for determining the specificity of the mRNA response. Ideally, target gene mRNA should be regulated in a dose-dependent manner. Time-course studies also can help establish the status of target genes, particularly when expression is changed by only a modest amount. Temporal expression of individual target genes in response to hormone also can vary; thus, analyzing multiple time points increases the likelihood of detecting hormone-regulated genes. Finally, time-course studies may provide clues as to which genes are regulated directly by hormone [e.g. require TH receptors (TR) binding to promoters of target genes] vs. those that may be regulated indirectly by hormone (e.g. require the induction of other transcription factors by TH, which, in turn, regulate transcription of target genes). The former tend to occur early after hormone treatment, whereas the latter occur later. In the case of TH action, systematic promoter analyses of candidate target genes, with the identification of appropriate TH response elements by binding and functional studies, would be necessary to determine conclusively whether TRs directly participate in the transcription of a particular target gene.
In addition to identifying individual target genes within the genome, microarrays enable us to study the interrelationships among target genes. In particular, gene expression profiles can be clustered in terms of transcriptional pattern, function, or cellular pathway (5, 6, 8). These, in turn, have yielded information on patterns of positive and negative regulation in the absence or presence of TH as well as cross-talk and coregulation of target genes in cellular networks. Additionally, studies of TR
- and TRß-knockout mice have shed light on the potential contribution of specific TR isoforms on target gene regulation (6, 8). Interestingly, a comparison of hepatic gene regulation in severely hypothyroid mice and double-TR/TRß-knockout mice yielded different gene expression profiles (8). These findings help explain why the absence of TH gives a different phenotype than the absence of TR in mice. It is possible that basal transcriptional repression by unliganded TRs in the hypothyroid mice or perhaps even nongenomic effects of TH may have contributed to these differences (9, 10). From these examples, we see that the study of interrelationships among target genes can provide new insights and suggest unanticipated mechanisms for gene regulation that can be tested in more detail (perhaps in individual target genes).
Resistance to TH (RTH) is an autosomal dominant disorder in which patients have elevated serum TH levels with elevated or inappropriately normal serum TSH levels (11). The patients are resistant to the effects of high circulating TH in many tissues. Most RTH patients have a mutation in the TRß allele, whose aberrant product blocks the transcriptional activity of wild-type receptors (dominant negative activity) on target genes. In this issue, Moeller et al. (12) used cDNA microarrays to analyze TH regulation of gene expression in human fibroblasts from normal subjects and RTH patients. Their study, which represents one of the first investigations of human gene expression profiles in response to TH, sheds new light on target gene expression. The authors used 15K element cDNA microarray chips to identify 91 positively and five negatively regulated target genes. Among them were several novel target genes involved in development and glucose metabolism. On the basis of the number of target genes in a given tissue, these findings suggest that fibroblasts are less responsive to TH than the liver and pituitary gland but are more responsive than the cerebellum (4, 5, 6, 13).
The authors employed several of the foregoing criteria to characterize their putative target genes. They confirmed the expression of a number of target genes by RT-PCR and elegantly employed dose-response and time-course studies to verify the status of several target genes. They also used previously identified target genes as positive controls. Last, they compared TH-mediated expression of target genes in fibroblasts from normal subjects with those from RTH patients and thus were able to hypothesize which genes may require TR-mediated signaling (either directly or indirectly) and which genes may be regulated by non-TR-mediated signaling. In this connection, TH effects that occurred in both types of cells suggested the possibility that nongenomic effects (e.g. phosphorylation of transcription factors or activation of kinase signaling pathways) could play a role in the regulation of some of the target genes observed in these microarrays. It seems that a similar strategy could be employed for studying transcriptional effects of other hormones by using siRNAs of their cognate receptors.
In summary, this study by Moeller et al. (12) clearly demonstrates some of the strategies and controls required for the study of hormonal regulation by microarrays and should serve as a useful example to investigators considering similar studies of other hormones in relevant biological systems. Although microarrays yield large, and sometimes bewildering, amounts of data, careful experimental design and analyses of the data can yield important new information. As such, they can help us appreciate the beauty of the trees in the context of the forest.
Footnotes
Abbreviations: RTH, Resistance to TH; TH, thyroid hormone; TR, TH receptor.
Received December 14, 2004.
Accepted December 15, 2004.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
