![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
,3
Departments of
*
Pediatrics and
Molecular and Cellular Biology, Baylor College of Medicine, Children’s Nutrition Research Center, Houston, TX 77030 and
**
Torrey Mesa Research Institute, Syngenta, San Diego CA 92121
3To whom correspondence should be addressed. E-mail: khirschi{at}bcm.tmc.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONDNA microarray technologiesUsing the toolsREFERENCES |
|---|
KEY WORDS: • microarray • genomics • genotype • differential gene expression
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONDNA microarray technologiesUsing the toolsREFERENCES |
|---|
|
DNA microarray technologies |
|---|
|
TOPABSTRACTINTRODUCTIONDNA microarray technologiesUsing the toolsREFERENCES |
|---|
Since the advent of gene cloning, scientists have gained much insight into the control of cellular behavior by studying the regulation of individual genes. However, understanding how one’s favorite gene is coordinately regulated among thousands of genes within the cellular microenvironment has only become possible within the last several years. Microarray technology provides a forum for such studies. Microarray analysis has also enabled researchers to move beyond understanding how a specific stimulus regulates a specific gene or set of genes in a pathway and toward understanding how such stimuli modulate many regulatory pathways in concert.
Successful utilization of such powerful technology begins with the
creation of an ordered set of nucleotide sequences, an array of DNA.
DNA microarrays can be produced in various forms, some examples of
which are shown in Figure 1
. DNA arrays can be synthesized using photochemical techniques or with
ink jet technology (Marton et al. 1998
). These arrays
will eventually accommodate the entire population of genes carried
within the human genome, but, currently, the most instructive
experiments have been performed with the complete set of yeast genes.
|
). Researchers have observed differential signal depending
on the fluorescent tag used, so it is often necessary to repeat the
hybridizations after switching the fluorescent tags on the control and
experimental RNA samples.
In the case of the GeneChip (Affymetrix, Inc., Santa Clara,
CA), a single RNA sample is converted into a fluorescent probe
and hybridized to a single chip, and the signal is also read using a
fluorimager device (Lockhart and Winzeler 2000
). Several
hybridization controls are used to generate normalized expression
levels that allow comparison of RNA levels among different GeneChips.
Both approaches require highly reproducible methods that must be
validated empirically by hybridizing the same probe to two different
chips and comparing the observed expression levels. This control
experiment gives the investigator an indication of the variability of
the system and is expressed as the lowest fold difference that is
considered reliable; a twofold difference is usually the lower limit.
The cDNA microarray approach lends itself to pair-wise comparisons,
as mentioned, control versus experimental (red vs. green in Fig. 2
). However, as long as an appropriate control sample is available,
multiple experimental samples can be analyzed (i.e., mock-treated
cells at T = 0 and at 1 h vs. cells treated with compounds X
or Y at T = 0 and at 1 h). Typically, replicate chips are
hybridized with the same probes and the values averaged. Because of the
computer controlled on-chip synthesis method for the
oligonucleotide array, as opposed to the spotting of polymerase chain
reaction products for the cDNA array chips, the results from a GeneChip
experiment typically have lower error rates (Lipshutz et al. 1999
). In addition, the creation of cDNA arrays often depends
on expressed sequencing tag-sequencing results to define singleton
clones (i.e., nonredundant spots on the slide). Because of the inherent
variability in expressed sequencing tag sequence data, results from
cDNA microarrays can require resequencing of clones to determine
identity and uniqueness of genes of interest, which can be laborious
when large numbers of clones are involved (Lipshutz et al. 1999
).
|
). The images of fluorescence levels (Fig. 2)
are processed and
normalization performed to control for variability in probe quality.
The results from both arrays can be expressed as change in expression
levels relative to control for each gene or probe set. It is then
possible to sort the data and construct clusters of genes that display
coregulation. In the example shown at the bottom of Figure 2
, red
indicates genes whose expression has increased, black is unchanged and
green is reduced relative to control. The red rows are clusters of
genes that are responding in a similar manner, in one case, to only one
treatment, and in another case, to two different treatments. This type
of clustering allows the investigator to determine which genes are
regulated in the same manner, suggesting relatedness. The identities of
the genes within a cluster can then lead to insights into the cell’s
response to the experimental condition. The real power of
large-scale gene expression experiments is not the creation of
lists of regulated genes, but it is a view of changes in biochemistry
on a grand scale. The yeast GeneChip, with nearly all of the yeast open
reading frames represented, makes it possible to observe effects on an
entire genome under a given condition (Wodicka et al. 1997
). This type of perspective requires a substantial amount
of time to analyze fully all of the changes; an example is given below.
In some cases, hundreds or even thousands of genes are affected in an
experiment. As more and more microarray experiments are performed and
the results compiled, we may begin to see commonalities and differences
in how cells integrate signals and modify responses.
|
Use of DNA microarrays, both oligonucleotide- and cDNA-based, presents a researcher with novel issues. These issues can be summarized in the following statement: when a researcher obtains a list of regulated genes as output from a chip experiment, that list may contain from a few tens upward to a few thousand genes. Thus, it is not always practical to confirm all of the putative changes in gene expression. Therefore, it is important to anticipate sources of error beforehand. A common practice to minimize sample-to-sample variability is to perform replicate experiments and then to pool RNA samples for each treatment from each replicate sample. These pooled RNA samples are then used to produce the labeled probes. Because of variability in the production of cDNA arrays, at least four replicate arrays typically are used per hybridization and the values for each spot are averaged. The oligonucleotide arrays tend to have fewer errors and the need for replicate hybridizations is also less; in some cases, no replicate hybridizations are performed. Finally, it is very important that control samples are matched as precisely as possible to experimental samples so that any changes in expression can be specifically attributed to the treatment and not to mechanical error. For example, a proposed experiment may require the use of two different incubators that have slightly different carbon dioxide levels. If the experimental sample is in one incubator and the control sample is in the other, an investigator could be mislead into thinking that a large cluster of genes is regulated by the treatment, but, in fact, they are responding to different levels of carbon dioxide.
Although it is tempting to get caught up in ensuring the accuracy of
microarray expression data, it should be noted that the true power of
genome-wide expression monitoring is not the listing of relative
RNA levels, but rather the predictive value that such information has
on the physiological responses being studied. Thus, the gold being
mined from the data is which pathways are showing transcriptional
control under the experimental conditions and that typically depends on
the response of several genes, minimizing the impact of slight
inaccuracies of the estimation of expression from any one gene. Once
the researcher has defined the transcriptional responses at play in a
given experiment, those conclusions are the basis for the next set of
experiments that should be performed and it is the outcome of those
next experiments that confirms the projections of the expression data
(Harmer et al. 2000
).
|
Using the tools |
|---|
|
TOPABSTRACTINTRODUCTIONDNA microarray technologiesUsing the toolsREFERENCES |
|---|
Although this concept may be heretical (particularly to individuals in
medical fields), microarray analysis is one illustration that important
insights can be gained from the systematic production of simple kinds
of biological information without a central hypothesis (Brent 2000
). The ongoing microarray research is often observational
rather than experimental and, as we mentioned above and demonstrate in
an example below, the findings are sometimes better described
as inferences that require additional testing rather than as a simple
conclusion that biologist are trained to seek. Although the
accumulation of microarray data sets alone cannot lead to important
insights, it can serve as a foundation for modeling regulatory
mechanisms and networks, as well as mapping disease progression.
Common applications.
Regardless of whether the experiments begin with a specific hypothesis or a fishing rod, the desired endpoints of microarray-based endeavors can be roughly divided into two categories: determination of genotype and analysis of differential gene expression.
Microarray-based genotype analysis involves the isolation of DNA from
an experimental sample, and comparison of its genetic code with that of
a defined control sample. To detect single base pair changes in a gene
of interest (i.e., polymorphisms), the probe must consist of
oligonucleotides representing permutations of this gene. Some sequence
alterations will be associated with aberrant gene function; some will
represent benign changes. The hybridization pattern of the sample DNA
with the probe will reveal the presence of, or propensity to develop, a
specific disease. This approach has been successfully applied to the
prediction/detection of various pathologies for which causal gene
mutations have been defined, most notably, cancer (Wang et al. 1998
, Wen et al. 2000
, Lehman et al. 2000
).
A more common application of microarray technology in basic and clinical research is the detection of differences in gene expression between and among samples (or populations of samples). As described above, these analyses require messenger RNA isolation from experimental and control samples, which are then reverse-transcribed, and levels of resulting cDNA compared. Similarities and differences in gene expression can be equally instructive depending on the research question.
This approach has been successfully used for gene discovery (i.e., new
genes in a characterized gene family or known regulatory pathway) and
is similarly used by pharmaceutical companies for drug discovery
(Scherf et al. 2000
). Pharmacologists and toxicologists
alike have taken advantage of this approach to understand individual
and population variances in physiological response and sensitivity to
particular compounds (Nuwaysir et al. 1999
,
Afshari et al. 1999
).
In the past, drug discovery was a long and arduous task. Initially, a biochemical pathway was implicated in a pathophysiological process. Armed with potential insights about the enzymes that are required for a functional pathway, biochemists and medicinal chemists would collaborate to identify and optimize the therapeutic behavior of candidate molecules that bind to specific target enzymes. Clinical trials then identified the candidate molecules with therapeutic efficacy in human subjects. This "needle in a haystack" approach costs $50 to $500 million for each new drug brought to market. DNA microarray methods promise to streamline many of these steps.
Indeed, microarray analysis is currently being incorporated into many
steps in drug development, including target identification and
validation (demonstrating that affecting the enzyme or biochemical
process has therapeutic utility), optimization of efficacy and
reduction of toxicity (Roses 2000
). A promise of this
emerging technology is that medical professionals will be able to
select the most effective drugs, or those with the fewest side effects,
for individual patients. DNA microarray analysis also will facilitate
identification of prospective clinical trail participants who might be
expected to respond optimally or adversely to specific therapeutics.
Applications in nutrition research.
The technology necessary to perform elaborate DNA microarray analysis and to analyze the vast data sets currently is perceived as prohibitively expensive by most academic scientists. At present, the majority of DNA microarray data are being generated in corporations; these data have been slow in finding their way into the public domain. However, this technology is becoming more broadly accessible, as indicated by the large increase in the number of published studies involving data obtained from array analyses. Ultimately, the field of nutrition will be fundamentally changed by molecular genetics, the availability of massive amounts of DNA sequence information and the development of technologies to exploit its use, such as microarray analysis.
Consider the power of using a series of mouse mutants consuming a diet
of well-defined plant mutants (Fink 1998
). Using
comparative microarray technologies with these precise genetic
backgrounds, one can define the specific nutrients that promote
vitality and longevity and their molecular targets. For example, one
might find that a particular set of genes is always activated in a
specific mouse mutant that dies of coronary artery disease when fed a
wild-type plant. However, this same animal exhibits no induction of
these genes and survives without complications when fed a mutant plant
unable to produce a particular lipid. The ability to perform such exact
feeding studies would transform heterogeneous nutrition feeding studies
into an exact science.
Although these types of experiments in humans may seem like science
fiction, they are already being achieved in simple models systems, with
well-defined genomes, such as the budding yeast
Saccharomyces cerevisiae. A recent example is a study by
Lyons et al. (2000
) that used DNA microarray analysis to
identify the yeast genes involved in zinc homeostasis.
Previously, the same investigators had identified a transcription
factor Zap1p that senses cellular zinc status and increases expression
of target genes in response to zinc deficiency (Zhao et al. 1998
). They also found the zinc-responsive DNA element
(ZRE) within the promoter region of these target genes (Zhao and Eide, 1997
). In their recent study, the authors performed three
individual experiments that combined the elegant yeast genetic system
with state-of-the-art DNA microarray analysis (Fig. 4
). First, using DNA microarrays, they determined that 15% of the genes
in yeast are significantly altered by changing the zinc content of the
media; 458 yeast genes are more highly expressed in zinc-deficient
media. The second experiment used genetic analysis in combination with
DNA microarrays. They compared the ability of wild-type yeast
strains and Zap1-deficient yeast strains to alter gene expression in
response to zinc deficiency. This experiment revealed that 214 genes
are more highly expressed in the wild-type strain, which suggests
that these genes require functional Zap1p for expression. Of these 214
genes, 111 of these overlap with the 476 genes induced during zinc
deficiency. Next, the investigators used a computer program to scan the
promoter region of these 111 genes for any putative ZRE elements. This
approach identified 46 genes that overlap with the 111 genes identified
in the previous two experiments.
|
A decade ago, a relatively small set of genes, thought to be important
in zinc homeostasis, would have been included in an array. These types
of experiments would not use the full potential of the array. A major
advantage of using arrays, especially those that contain probes for the
entire genome (such as the yeast chips used in this study), is that it
is not necessary to guess what the important genes or mechanisms are in
advance. Instead of testing a hypothesis, a broader, more thorough and
less biased view of the cellular response is obtained (Lockhart and Winzeler 2000
).
The impressive list of Zap1 target genes generated in this study illustrates the elegance of combining microarray expression profiling with computer analysis. Only a handful of genes involved in zinc homeostasis have been identified by classical genetic studies in yeast. This novel combined approach enabled the discovery of new genes that could not be identified by mutant analysis alone.
It is important to bear in mind that this study has no immediate functional significance. It simply suggests that genes, regulated by Zap1p in a zinc-responsive manner, may be involved in zinc homeostasis. More than 50% of the genes identified in this study have no known function. However, by using the tools available in yeast, it will be a relatively straightforward procedure to analyze the mutant phenotype of these genes individually and in combinations.
Another potential use of microarray technology in the field of nutrition science is analogous to that of pharmacological or toxicological studies in which nutritive components of foods are viewed as drugs. By studying the cellular and molecular reaction to specific nutrients, in specific doses, between population groups and among individuals within those groups, one can begin to define the optimal cellular doses to produce the desired molecular responses in a given subpopulation.
For the field of nutrition to adequately harness the powerful tools of
genomic biology, it is imperative that collaborations be formed with
plant biologists, mathematicians, engineers and computer scientists.
With the proper design and implementation of microarray experiments,
these expression profiles will provide the basis for understanding how
nutrients affect gene expression to orchestrate the function of
organisms (Young 2000
). By capturing genetic pathways in
computational form, the field of nutrition can augment the
understanding of the metabolic pathways that are the benchmark of
nutritional sciences. These genetic and metabolic pathways then can be
manipulated through diet, environmental stimuli and genetic targeting.
|
FOOTNOTES |
|---|
2 Supported by United States Department of
Agriculture Grant CRIS-6250-21520-041 and National Institutes of Health
Grants R01-GM57427 and CHRC 5-P30 (to K.D.H.); and United States
Department of Agriculture Grant CRIS-6250-51000-033, National
Institutes of Health Grant R01-HL61408, American Heart Association
National Award SDG-9930054N, and Baylor College of Medicine Gillson
Longenbaugh Foundation Award (to K.K.H.).
4 Abbreviations used: DNA, deoxyribonucleic acid;
RNA, ribonucleic acid; cDNA, complimentary deoxyribonucleic acid; ZRE,
zinc-responsive DNA element.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONDNA microarray technologiesUsing the toolsREFERENCES |
|---|
1.
Afshari C. A., Nuwaysir E. F., Barrett J. C. Application of complementary DNA microarray technology to carcinogen identification, toxicology, and drug safety evaluation. Cancer Res 1999;59:4759-4760
2. Brent R. Genomic biology. Cell 2000;100:169-183[Medline]
3.
Fink G. R. Anatomy of a revolution. Genetics 1998;149:473-477
4.
Harmer S. L., Hogenesch J. B., Straume M., Chang H.-S., Han B., Zhu T., Wang X., Kreps J. A., Kay S. A. Orchestrated transcription of key pathways in Arabidopsis by the circadian clock. Science 2000;290:2110-2113
5.
Lehman T. A., Haffty B. G., Carbone C. J., Bishop L. R., Gumbs A. A., Krishnan S., Shields P. G., Modali R., Turner B. C. Elevated frequency and functional activity of a specific germ-line p53 intron mutation in familial breast cancer. Cancer Res 2000;60:1062-1069
6. Lipshutz R. J., Fodor S. P., Gingeras T. R., Lockhart D. J. High density synthetic oligonucleotide arrays. Nat. Genet. 1999;21:20-24[Medline]
7. Lockhart D. J., Winzeler E. A. Genomics, gene expression and DNA arrays. Nature 2000;403:827-836
8.
Lyons T. J., Gasch A. P., Gaither L. A., Botstein D., Brown P. O., Eide D. J. Genomic-wide characterization of the Zap1p zinc-responsive regulon in yeast. PNAS USA 2000;97:7957-7962
9. Marton M. J., DeRisi J. L., Bennett H. A., Iyer V. R., Meyer M. R., Robets C. J., Stoughton R., Burchard J., Slade D., Dai H., Bassett D. E., Jr, Hartwell L. H., Brown P. O., Friend S. H. Drug target validation and identification of secondary drug target effects using DNA microarrays. Nat. Med. 1998;4:1293-1301[Medline]
10. Nuwaysir E. F., Bittner M., Trent J., Barrett J. C., Afshari C. A. Microarrays and toxicolgy: the advent of toxicogenomics. Mol. Carcinog. 1999;24:153-159[Medline]
11. Roses A. D. Pharmacogenetics and the practice of medicine. Nature 2000;405:857-865[Medline]
12. Scherf U., Ross D. T., Waltham M., Smith L. H., Lee J. K., Tanabe L., Kohn K. W., Reinhold W. C., Myers T. G., Andrews D. T., Scudiero D. A., Eisen M. B., Sausville E. A., Pommier Y., Botstein D., Brown P. O., Weinstein J. N. A gene expression database for the molecular pharmacology of cancer. Nat. Genet. 2000;24:236-244[Medline]
13. Wang D. G., Fan J. B., Lander E. S. Large-scale identification, mapping, and genotyping of single-nucleotide polymorphisms in the human genome. Science 1998;15:1077-1082
14. Wen W. H., Bernstein L., Lescallett J., Beazer-Barclay Y., Sullivan-Halley J., White M., Press M. F. Comparison of TP53 mutations identified by oligonucleotide microarray and conventional DNA sequence analysis. Cancer Res 2000;15:2716-2722
15. Wodicka L., Dong H., Mittman M., Ho M.-H., Lockhart D. J. Genome-wide expression monitoring in Saccharomyces cerevisiae. Nat. Biotechnol. 1997;15:1359-1367[Medline]
16. Young R. A. Biomedical discovery with DNA arrays. Cell 2000;102:9-15[Medline]
17.
Zhao H., Butler E., Rodgers J., Spizzo T., Duesterhoeft S., Eide D. J. Regulation of zinc homeostasis in yeast by binding of the ZAP1 transcriptional activator to zinc-responsive promoter elements. J. Biol. Chem. 1998;273:28713-28720
18. Zhao H., Eide D. J. Zap1p, a metalloregulatory protein involved in zinc-responsive transcriptional regulation in Saccharomyces cerevisiae. Mol. Cell. Biol. 1997;17:5044-5052[Abstract]
This article has been cited by other articles:
|
|
 |
|
  J. Kaput and R. L. Rodriguez Nutritional genomics: the next frontier in the postgenomic era Physiol Genomics, January 15, 2004; 16(2): 166 - 177. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. A. Martinez, M. S. Corbalan, A. Sanchez-Villegas, L. Forga, A. Marti, and M. A. Martinez-Gonzalez Obesity Risk Is Associated with Carbohydrate Intake in Women Carrying the Gln27Glu {beta}2-Adrenoceptor Polymorphism J. Nutr., August 1, 2003; 133(8): 2549 - 2554. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Kato and T. Kimura Evaluation of the Effects of the Dietary Intake of Proteins and Amino Acids by DNA Microarray Technology J. Nutr., June 1, 2003; 133(6): 2073S - 2077. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Nur, A. A. C. M. Peijnenburg, H. P. J. M. Noteborn, H. Baykus, and R. Reifen DNA Microarray Technology Reveals Similar Gene Expression Patterns in Rats with Vitamin A Deficiency and Chemically Induced Colitis J. Nutr., August 1, 2002; 132(8): 2131 - 2136. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. R. Young 2001 W.O. Atwater Memorial Lecture and the 2001 ASNS President's Lecture: Human Nutrient Requirements: The Challenge of the Post-Genome Era J. Nutr., April 1, 2002; 132(4): 621 - 629. [Full Text] [PDF]
|
|
|
|
|
|
J. D. Hirschman and J. A. Kresge Synthesis of a Symposium: Innovative Non- or Minimally-Invasive Technologies for Monitoring Health and Nutritional Status in Mothers and Young Children J. Nutr., May 1, 2001; 131(5): 1643S - 1645. [Full Text]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
