Special Topic Human Genomics and Microarrays: Implications for the Plastic Surgeon

Jana Cole, M.D., and Frank Isik, M.D.

Seattle, Wash.

The Human Genome Project was launched in 1989 in an effort to sequence the entire span of human DNA. Although coding sequences are important in identifying mutations, the static order of DNA does not explain how a cell or organism may respond to normal and abnormal biological processes. By examining the mRNA content of a cell, researchers can determine which genes are being activated in response to a stimulus. Traditional methods in molecular biology generally work on a one gene: one experiment basis, which means that the throughput is very limited and the whole picture of gene function is hard to obtain. To study each of the 60,000 to 80,000 genes in the human genome under each biological circumstance is not practical. Recently, microarrays (also known as gene or DNA chips) have emerged; these allow for the simultaneous determination of expression for thousands of genes and analysis of genome-wide mRNA expression. The purpose of this article is twofold: first, to provide the clinical plastic surgeon with a working knowledge and understanding of the fields of genomics, microarrays, and bioinformatics and second, to present a case to illustrate how these technologies can be applied in the study of wound healing. (Plast. Reconstr. Surg. 110: 849, 2002.)

A phenomenal scientific achievement occurred 40 years ago: the chemical structure of DNA was cracked by Watson and Crick.1 In the year 2000, another scientific milestone was achieved. On June 26, researchers told the world that they had identified the order of all 3 billion chemical units that make up the human genome.2 The first draft sequence of the entire human chromosome set was identified. Although the precise number of human chromosomes was still under debate when Watson and Crick1 made their discovery, we now know that there are 46 human chromosomes, which between them house 3 billion base pairs of DNA and encode about 60,000 to 80,000 proteins.

The effort to sequence the entire span of human DNA, the Human Genome Project, was launched in 1989 as a consortium between the National Institutes of Health and the Department of Energy. The Human Genome Project served to develop technologies for genomic analysis, to examine the ethical, legal, and social implications of human genetics research, and to train scientists to use these tools and resources to pursue biological studies that will improve human health. At first glance, the sequencing of the human genome and the related technology that has emerged do not seem to affect the practicing plastic surgeon. Many laboratory discoveries impact very little on our day-to-day practice. The discovery of a promising endogenous molecule or application of a novel technique initially seems to be a panacea for many clinical problems, but then does not change our practice as initially hoped and hyped. The purpose of this article is twofold: first, to provide the clinical plastic surgeon with a working knowledge and understanding of the fields of genomics, microarrays, and bioinformatics and second, to present a case to illustrate how these technologies can be applied in the study of wound healing.
THE WORKINGS
OF A

CELL

Genes and their products (i.e., RNA and proteins) are thought to function in a complicated and orchestrated way to create the mystery of life. The human body begins as a fusion product of two cells, the egg and the sperm. The successive division of the fusion product results in the formation of the adult organism,

From the Division of Plastic Surgery, University of Washington School of Medicine. Received for publication June 25, 2001; revised November 19, 2001. DOI: 10.1097/01.PRS.0000019918.86678.91

849

850

PLASTIC AND RECONSTRUCTIVE SURGERY,

September 1, 2002

which contains approximately 1014 cells. Although each adult cell preserves the entire DNA content of the original single cell, the progeny cells have become specialized in function (i.e., differentiated). The cell differentiation pathway is dictated by the restricted spatial and temporal expression of specific genes that are coded for by the DNA. Which genes or clusters of genes induce differentiation to a particular cell type, such as a chondrocyte or adipocyte, are key questions now being asked by many investigators and companies. In addition, the response of differentiated cells to an external stimulus such as tissue injury depends on the expression of different clusters of genes at different times.35 As research improves and matures our comprehension of cellular differentiation and cellular response mechanisms, we will gain insight into the causes of many congenital anomalies and cancer and possibly even develop therapeutic strategies to treat them. As a brief review, we will go over the necessary definition of terms and cell function. The nucleus of a human cell contains 23 pairs of chromosomes, with each chromosome being

made up of DNA molecules. DNA molecules consist of two long chains held together by complementary base pairs. Each chain is a long, unbranched polymer composed of only four types of subunits or bases. These are the deoxyribonucleotides containing adenine (A), guanine (G), cytosine (C), and thymidine (T). There is complementary base pairing between A and T and between C and G. This means that each strand of the DNA molecule is a mirror image of the other and the molecule exists as a double helix. The order in which these subunits are linked together represents a blueprint for the cell, providing all the necessary information to construct a cell or even generate a clone of an individual (Fig. 1). The information contained in the chain of the DNA molecule is read in triplicate (e.g., AGC TAG ATG. . .), with each possible triplicate combination (codon) coding for one of the 20 amino acids, the building blocks of proteins. Contained within the long chain of the DNA molecule are short segments that code for proteins. A protein is made by a series of steps called transcription and translation. The blueprint for making the protein resides on the

FIG. 1. Illustration of the transfer of information from DNA to protein. This proceeds by means of mRNA. During transcription, one strand on the DNA serves as a template for the new mRNA. This transfer of information is accomplished by complementary base pairing between the bases A and U and C and G. The mRNA can cross the nuclear membrane into the cytoplasm. The mRNA then binds to ribosomes (location where the protein is made), which translate the information in the mRNA into protein. This step is called translation. The mRNA is then usually degraded after the protein is produced.

Vol. 110, No. 3 /

HUMAN GENOMICS AND MICROARRAYS

851

chromosome in the nucleus, but the protein machinery resides in the cytoplasm, and they are separated by the nuclear membrane. The transfer of information from DNA to protein proceeds by means of an RNA intermediate called messenger RNA (mRNA). Both DNA and RNA are linear polymers of nucleotides, but they differ in three ways. The sugar phosphate backbone in RNA contains ribose instead of deoxyribose, RNA contains the base uracil (U) instead of thymidine (T), and RNA exists as a single strand instead of a double helix. During transcription, one strand on the DNA serves as a template for the new mRNA. The mRNA is again transcribed by complementary base pairing (A 3 U and C3 G) and provides a mirror image of the DNA template. The mRNA can cross the nuclear membrane into the cytoplasm. The mRNA then binds to ribosomes (location where the protein is made), which translate the information in the mRNA into protein. This step is called translation. The mRNA is usually degraded after the protein is produced. Although critical for the function of a cell, messenger RNA does not have a function outside the nucleus other than to transport the code necessary to build the protein. Protein is the ultimate molecule that provides structure and function for the cell. The vast majority of the DNA in a cell, even if the cell is actively proliferating and migrating, is inactive. In other words, only a small fraction of the 60,000 to 80,000 genes are being transcribed into mRNA. Because mRNA is often rapidly degraded shortly after the pro-

tein is made, the mRNA content of a cell at any given time represents what that cell is doing or responding to at that moment. This is an important point that is exploited by the microarray technology.
GENOMICS
AND

MICROARRAYS

The term genome refers to all the genetic material in all the chromosomes of a particular organism. For us to understand the molecular basis of health and disease, we need to know more than the coding sequence of the genome. Although coding sequences are important in identifying mutations linked to certain cancers, the static order of the DNA does not tell us how a cell or organism may respond to normal and abnormal biological processes. We need to know which gene products are made in biological processes that affect human health and disease, from embryonic development to cancer development. The best method to investigate this currently is by examining the mRNA content of a cell. Traditional methods in molecular biology generally work on a one gene: one experiment basis, which means that the throughput is very limited and the whole picture of gene function is hard to obtain. To study each of the 60,000 to 80,000 genes individually under each biological circumstance is not practical. Recently, newer high-throughput techniques have emerged, such as differential display,6 serial analysis of gene expression (SAGE),7 and microarrays (also known as gene or DNA chips).8 Of these high-throughput techniques, microarrays are much more efficient; they allow for the simultaneous deter-

FIG. 2. This diagram demonstrates the attraction of certain nucleotides to each other (A3 T and G3 C) and how, over a long stretch of DNA, the sequence alignment becomes critical for hybridization (reforming of the double helix) to occur. In microarray experiments, known sequences are immobilized on a solid surface. The labeled sample bathes the microarray. If there is sufficient match between two complementary sequences, then a stable duplex will form. The rate of double helix formation during hybridization is limited by the rate at which the two complementary nucleic acids happen to collide, which depends on their concentration in the solution. Hybridization rates can therefore be used to determine the concentration of any desired RNA or DNA sequence in a mixture.

852

PLASTIC AND RECONSTRUCTIVE SURGERY,

September 1, 2002

mination of expression for thousands of genes, and they analyze genome-wide patterns of mRNA expression. We will focus our discussion on this technology.
Microarrays

Complementary base-pairing (i.e., A3 T and G3 C for DNA; A3 U and G3 C for RNA) or hybridization is the underlying principle of DNA microarrays (Fig. 2). The natural pairing of certain nucleotides, for example, adenine with thymidine and guanine with cytosine, forms the tight duplex of DNA.9 An array means an orderly arrangement. In microarrays, the premise is to have known and unique complementary DNA (cDNA) sequences immobilized on a support surface while the radiolabeled sample (mRNA or DNA) passes over each spot, as shown in Figure 2. If there is a matching mRNA for that immobilized cDNA,

then the labeled mRNA will hybridize (form a double helix) to that spot. Depending on the labeling method, the spot will then either fluoresce or be radioactive, which is easily detected by scanning techniques. Because there is a known amount of cDNA spotted, the amount of fluorescence or radioactivity can be quantified. There are two competing formats for the microarrays. In one format, cDNA (each representing a gene 500 to 5000 bases long) is immobilized to a solid surface such as glass by high-speed robotics. This method, traditionally called cDNA microarray, was developed principally at Stanford University.8,10 14 In the competing format, an array of oligonucleotides (20 to 25 oligos) is synthesized either in situ (onchip) or by conventional synthesis followed by on-chip immobilization.15 This method, historically called DNA chips, was developed at Af-

FIG. 3. (Above, left) Example of a nylon microarray (from Research Genetics) that contains 4000 known human cDNA sequences. Each spot represents a unique gene sequence. (Above, right and below, left) cDNA microarray membranes after hybridization with two different radiolabeled mRNA sources and scanned by phosphorimager. Darker spots indicate increased gene expression. (Below, right) Image generated by the Pathways software program (Research Genetics) comparing the two mRNA samples. Genes that predominate in sample one are shown in red. Genes that predominate in sample two are shown in green. Those genes with equal expression in both samples are shown in yellow.

Vol. 110, No. 3 /

HUMAN GENOMICS AND MICROARRAYS

853

fymetrix, Inc., which sells its products under the GeneChip trademark. In either case, the array is exposed to labeled DNA or mRNA samples, hybridized, scanned, and analyzed to determine the identity and abundance of each gene in the sample. There are two major applications for either microarray technology: (1) identification of DNA sequence (gene or gene mutation) and (2) determination of mRNA expression level or abundance. In the first application, special sets of microarrays have been constructed that are specific in identifying certain gene mutations. For instance, the BRCA1 and BRCA2 gene mutations have been linked to the development of ovarian and breast cancer. Patients who carry the mutation can be identified rapidly using specially constructed cDNA microarrays that have the immobilized BRCA1 and BRCA2 gene mutations and the nonmutated BRCA1 and BRCA2 genes.16,17 If patients are not carriers, then their DNA will not hybridize to the BRCA1 and BRCA2 gene mutation spots, only to the nonmutated spots. This method requires obtaining a blood sample or buccal swab to obtain the genomic DNA. Because every cell in the body contains the same DNA, identification of the inheritable mutation in any cell type identifies the patient as a carrier of the mutation. This allows rapid screening of the patients DNA without having to sequence their DNA. In addition, several other microarrays have been constructed for the identification of mutations associated with many other types of cancers.18 25 The other application of microarrays involves monitoring the mRNA expression of thousands of genes simultaneously. Unlike the previous application of microarrays that assayed the patients genomic DNA content and sequence, expression arrays assay the patients, or cells, mRNA content. This allows for massively parallel gene expression and gene discovery studies, which will be discussed later in this article. A single experiment can provide information on thousands of genes simultaneously (i.e., which genes are increased or decreased in specific biological processes); this is a dramatic increase in throughput over the one gene: one experiment method. Although currently restricted to several thousands of genes, this technology in the near future promises to monitor the entire human genome on a single membrane or glass slide so that researchers can determine the expression of all genes during

any biological process. Microarrays represent the biological equivalent of the integrated chip.
Bioinformatics

The generation of such vast volumes of data requires specialized methods to catalogue, group, analyze, and interpret the biological data. The field of bioinformatics is defined as the application of computers, databases, and computational models to the management of this enormous amount of biological information. In general, most of the analysis tools in use today use computational methods to group (cluster) genes or experiments with similar profiles of changes in expression levels.26 The assumption is that by distinguishing genes that behave similarly, it is possible to gain insight into shared regulatory aspects or shared functions by the cluster of genes. Hierarchical clustering is the most commonly used tool in gene expression analysis. Pairwise matrices can be used to identify genes sharing a similar expression pattern across multiple experiments. In one such method, all values are paired, and modified Pearson correlations are calculated for each possible pairwise combination and used in distance matrices. This allows hierarchical clustering of groups of genes that behave most similarly. Cluster, a hierarchical clustering-based algorithm is one of the most common tools used to analyze microarray data.26 Self-organizing maps are ideally suited for exploratory data analysis.27 They are considered superior to hierarchical clustering when analyzing messy data that contains outliers and irrelevant variables. The basic concept is that you impose a partial structure on the data and then adjust the structure according to the data. The input data are the raw expression values obtained from the microarray experiments. The output is a series of maps representing similar patterns of gene expression. The infancy of high-throughput gene analysis is reflected in the analytical tools used to decipher the voluminous information. With either method, the analysis is based on the simple premise that if it is variable, perhaps it is important, and if a group of genes are similarly variable, perhaps they all share a similar function. Although simple observations about the relative expression of genes in different sample groups will not lead to conclusions about physiologic or pathologic processes, they can be

854

PLASTIC AND RECONSTRUCTIVE SURGERY,

September 1, 2002

used to generate hypothesis-driven mechanistic experiments that define the function of specific genes in specific biological processes. It is these studies that will ultimately validate the importance of expression profiling by microarrays. Currently, microarrays are largely used as a screening tool, as exemplified in the example below.
CLINICAL CORRELATION

In our laboratory, we have applied cDNA microarray technology to the study of wound healing. We performed a series of experiments using microarrays to determine the gene expression profile of normal human skin, 3 acutely wounded skin,4 and hypertrophic scarring.5 In the sequence to follow, we will provide a step-by-step illustration of how the technique can be used, and we will highlight some of the analyses that can identify which genes may be relevant to normal wound healing.
Step 1

are shown in red, those that predominate in sample two are shown in green. Those genes that have an equal gene expression in both samples are shown in yellow. The data from each comparison can be viewed as a histogram to determine those genes that have increased or decreased in expression. The histogram shown in Figure 4, above, is representative of two normal skin samples. Note that more than 99 percent of the genes are expressed within a three-fold difference in expression, showing a gaussian distribution of gene expression in normal skin from person to person. The histogram in Figure 4, below, illustrates a normal skin sample compared with the same persons injured skin at 30 minutes. Approximately 2 percent of genes (124 of 4000) in the wounded skin are increased greater than two-fold, and less than 1 percent are increased more than three-fold (22 of 4000). There was little downregulation of gene expression at 30 minutes.
Step 4

With Institutional Review Board approval, we obtained skin samples from a healthy woman who was undergoing elective breast reconstruction 10 years after a mastectomy. Using an 8 mm punch, we obtained normal skin samples and injured skin samples at 30 minutes, 1 hour, 2 hours, 4 hours, and 1 month after wounding from the latissimus dorsi donor site. The 1-month biopsy was obtained from a small dehiscence of the donor site and represents an open epithelializing wound.
Step 2

The mRNA from intact skin was extracted, reverse-transcribed into 33 P-radiolabeled cDNA, and hybridized onto high-density cDNA microarray membranes of 4000 genes (Fig. 3, above, left). This membrane was then scanned on a phosphorimager to produce the raw images seen in Figure 3 (above, right, and below, left).
Step 3

The two images were then analyzed by the Pathways software program (Research Genetics, Inc.) to determine the intensity of each cDNA on each membrane; intensity correlates with abundance. The radioactive intensities between the two samples can then be compared (Fig. 3, below, right), and a gene expression profile of relative intensities can be produced. Those genes that predominate in sample one

To analyze the data further, we used cluster analysis to group genes based on expression patterns over the different time points (Fig. 5). We then tabulated those genes that were most upregulated during acute injury (30 minutes to 4 hours; Table I) or chronic injury (1 month; Table II). Those genes most upregulated immediately after injury are involved in transcription, signaling, and inflammation. This is expected, because the initial cellular response after coagulation is inflammation.28 Genes expressed in the chronic wound show a different expression profile. Structural genes such as intracellular keratins in the epithelium and extracellular collagen types are most upregulated, consistent with the process of epithelialization. We are thus able to create an expression profile of those genes upregulated and downregulated during the wound healing process. A myriad of further analysis is possible with this data set. We could temporally follow the expression changes in a subset of genes such as growth factors, collagens, or inflammatory mediators.35 In summary, we were able to identify 210 of 4000 genes examined (5 percent) that changed in expression in response to injury on the basis of the few time points we examined. If we consider that there are 60,000 to 80,000 genes in the human genome, we could extrapolate that wound healing may potentially involve up to 4000 genes. Obviously, this is a

Vol. 110, No. 3 /

HUMAN GENOMICS AND MICROARRAYS

855

FIG. 4. (Above) Histogram shows a comparison of the gene expression profile from two normal skin samples. More than 99 percent of the genes fall within a three-fold change in expression, demonstrating the limited variability of gene expression in normal skin from one person to the next. (Below) Histogram shows a comparison of the gene expression profile of normal skin compared with the same patients skin 30 minutes after injury. Note the increased gene expression (red shift to the right) in the acutely wounded skin compared with the quiescent state.

guess and demonstrates the need for largescale studies over longer periods of time to define accurately how a wound heals at the mRNA level. These studies are underway by both investigators and pharmaceutical companies. Although arduous, these are powerful studies that are bound to raise new questions. By knowing which genes are expressed during normal wound healing, we could, in the future, identify and categorize nonhealing wounds or determine the effect of radiation and steroids on the normal wound healing expression response. In the future, a simple biopsy of a nonhealing wound may be sent to the lab for analysis to help direct treatment just as we now send off a urine sample for culture to guide our antibiotic choices.
CLINICAL IMPACT

At first glance, the sequencing of the human genome may not seem to be of value to the

clinical plastic surgeon. After all, we will not be performing cDNA expression profiling in the office in the foreseeable future. However, the information from the Human Genome Project has already impacted plastic surgeons. Patients who are BRCA1 and BRCA2 mutation carriers are being counseled for prophylactic bilateral mastectomies29,30; some then opt for immediate breast reconstruction.31 The most promising results of the Human Genome Project and its offspring technologies, such as microarrays, will be in the future as we better understand biological processes at the most basic level. For example, gene expression analyses will aid in understanding the development process: for example, which sets of genes are coordinately regulated to achieve normal development of the craniofacial skeleton and limbs. This has obvious implications in determining the cause and perhaps the potential treatment of certain developmental anomalies. Some syndromes that are difficult to diagnose

856

PLASTIC AND RECONSTRUCTIVE SURGERY,

September 1, 2002

FIG. 5. Graphic representation of cluster analysis demonstrated for one patient. Because of size restraints, only 800 of the 4000 genes are represented in black on the left. This column represents baseline gene expression in healthy unwounded skin (Nl). Genes that are upregulated at different time points are shown in green, and genes that are downregulated are shown in red. In this subset of 800 genes, at 30 minutes, the majority of the represented genes are upregulated (green). In comparison, at the 60 minute time point, there is a general downregulation of these same genes (red). A subset of these genes is listed in the text box at right. Note that the cluster analysis groups genes with similar expression patterns; this is not based on the most variable genes. The most variable genes are listed in Tables I and II.

early in the neonatal period are likely to be aided by mRNA expression analysis or DNA mutational analysis provided by microarrays, thus enabling rapid, accurate, and early diagTABLE I Genes Expressed 30 Minutes after Injury

nosis. Obviously, the human genome research effort also raises important yet unanswered ethical, legal, and social questions.
TABLE II Genes Expressed 1 Month after Injury

Gene

Increase (fold)

Gene Function

Increase (%)

Function

Suppressor of cytokine signaling 3,4 Inositol phosphate signaling Regulator of G protein signaling Transcription factor 3 CCAAT box-binding transcript 1 Elongation factor 1 Macrophage-stimulating 1 TNF receptor-1 TNF Metallothionein 1L
TNF, tumor necrosis factor.

20 12 6 5 7 10 9 6 5 4

Signaling Signaling Signaling Transcription Transcription Transcription Inflammation Inflammation Inflammation Inflammation
tor.

Keratin Keratin 5 Collage type I (2) Collagen type III (1) Actin (2) Regulator of G-protein signaling 1 MHC class II transactivator MHC class I1 Matrix metalloproteinase 21 Latent TGF- binding protein 1

37 18 33 27 27 6 10 7 5 5

Structural Structural Structural Structural Structural Signaling Inflammation Inflammation Inflammation Growth factor

MHC, major histocompatability complex; TGF, transforming growth fac-

Vol. 110, No. 3 /

HUMAN GENOMICS AND MICROARRAYS

857

TABLE III List of Useful Web Sites

Introductory Websites National Center for Biotechnology Information: http://www3.ncbi.nlm.nih.gov/ Weizmann Institute Bioinformatics unit: http://bioinformatics.weizmann.ac.il// National Center for Genome resources: http://www.ncgr.org/ SWISS PROT: http://www.expasy.ch/sprot/ Genes and diseases: http://www.ncbi.nlm.nih.gov/disease/ Genome databases Genome channel: http://compbio.ornl.gov/channel/ Entrez genome: http://www3.ncbi.nlm.nih.gov/entrez/query.fcgi?dbGenome Whitehead institute/MIT center for genome research: http://www-genome.wi.mit.edu/ Nucleotide and sequence databases Genbank: http://www.ncbi.nlm.nih.gov/Genbank/index.html DbEST: http://www.ncbi.nlm.nih.gov/dbEST/ Unigene: http://www.ncbi.nlm.nih.gov/UniGene/index.html Nucleotide and protein sequence analysis BLAST: http://www.ncbi.nlm.nih.gov/BLAST PROSITE: http://www.expasy.ch/prosite/ Expression data and analysis The Brown Laboratory: http://cmgm.stanford.edu/pbrown/ Stanford Genome Center: http://genome-www.stanford.edu/ The Microarray Project at NHGRI: http://www.nhgri.nih.gov/DIR/LCG/15K/HTML Whitehead/MIT Center for Genome Research: http://waldo.wi.mit.edu/MPR/ Proteomics Danish Center for Human Genome Research: http://biobase.dk/cgi-bin/celis EXPASY: http://www.expasy.ch/

The use of cDNA microarrays is likely to foster a new level of understanding and categorization of cancers, which may impact our treatment of melanoma. Future investigation is likely to prove that expression profiling is a better prognosticator than Breslow depth32 and that expression profiling of excised melanomas may even prove pivotal for performing or not performing sentinel lymph node biopsy.11,33 As better biological markers of invasive or aggressive melanoma are identified, plastic surgeons may have to change not only the current treatment algorithms but also reconstruction algorithms. For example, if genetic markers of locally aggressive melanoma can be identified by microarrays, than those patients may be better served by skin grafting the defect to detect recurrences earlier. The tool is already being used a great deal by pharmaceutical companies to understand the global impact a drug has on cells and tissues. Microarrays are being used to determine the

response of bacteria to certain antibiotics.34 Microarrays are also proving worthwhile to monitor toxicity of various chemicals and drugs to cells.35,36 The expression profile of a patient may be a very useful way to study the response to certain drugs.37 With time, pharmaceutical companies will design drugs that enhance or inhibit a specific molecule or groups of molecules that were initially identified by microarrays. The Human Genome Project has provided enormous information about the basic set of inherited instructions for the development and functioning of a human being and is profoundly changing our understanding of cell and tissue function. The identification of which genes are involved in specific biological processes will lead to a better understanding of disease. In the near future, genomic technologies will improve our diagnostic abilities, provide new opportunities for screening and prevention, and impact the treatment we provide to our patients. Once their strengths and limitations become better defined, microarrays and the other products of the Human Genome Project are likely to find a vital role in the practice of medicine and plastic surgery. Table III provides a list of useful Web sites for further information. Jana Cole, M.D. Division of Plastic Surgery University of Washington Medical Center Box 356410 1959 N.E. Pacific Street Seattle, Wash. 98195 janacole@u.washington.edu
REFERENCES 1. Watson, J. D., and Crick, F. H. Molecular structure of nucleic acids: A structure for deoxyribose nucleic acid. Nature 248: 765, 1974. 2. Venter, J. C., Adams, M. D., Myers, E. W., et al. The sequence of the human genome. Science 291: 1304, 2001. 3. Cole, J., Tsou, R., Wallace, K., Gibran, N., and Isik, F. Comparison of normal human skin gene expression using cDNA microarrays. Wound Repair Regen. 9: 77, 2001. 4. Cole, J., Tsou, R., Wallace, K., Gibran, N. S., and Isik, F. F. The early gene expression profile of human skin to injury using high-density cDNA microarrays. Wound Repair Regen. 9: 360, 2001. 5. Tsou, R., Cole, J. K., Nathens, A. B., et al. Analysis of hypertrophic and normal scar gene expression with cDNA microarrays. J. Burn Care Rehabil. 21: 541, 2000. 6. Liang, P., and Pardee, A. B. Differential display of eukaryotic messenger RNA by means of the polymerase chain reaction. Science 257: 967, 1992.

858
7. Velculescu, V. E., Zhang, L., Vogelstein, B., and Kinzler, K. W. Serial analysis of gene expression. Science 270: 484, 1995. 8. Schena, M., Shalon, D., Davis, R. W., and Brown, P. O. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270: 467, 1995. 9. Arnott, S. Crystallography of DNA: Difference synthesis supports Watson-Crick base pairing. Science 167: 1694, 1970. 10. Shalon, D., Smith, S. J., and Brown, P. O. A DNA microarray system for analyzing complex DNA samples using two-color fluorescent probe hybridization. Genome Res. 6: 639, 1996. 11. DeRisi, J., Penland, L., Brown, P. O., et al. Use of a cDNA microarray to analyse gene expression patterns in human cancer. Nat. Genet. 14: 457, 1996. 12. Schena, M., Shalon, D., Heller, R., et al. Parallel human genome analysis: microarray-based expression monitoring of 1000 genes. Proc. Natl. Acad. Sci. U.S.A. 93: 10614, 1996. 13. Spellman, P. T., Sherlock, G., Zhang, M. Q., et al. Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol. Biol. Cell 9: 3273, 1998. 14. Iyer, V. R., Eisen, M. B., Ross, D. T., et al. The transcriptional program in the response of human fibroblasts to serum. Science 283: 83, 1999. 15. Vasiliskov, A. V., Timofeev, E. N., Surzhikov, S. A., et al. Fabrication of microarray of gel-immobilized compounds on a chip by copolymerization. Biotechniques 27: 592, 1999. 16. Hedenfalk, I., Duggan, D., Chen, Y., et al. Gene-expression profiles in hereditary breast cancer. N. Engl. J. Med. 344: 539, 2001. 17. Favis, R., Day, J. P., Gerry, N. P., et al. Universal DNA array detection of small insertions and deletions in BRCA1 and BRCA2. Nat. Biotechnol. 18: 561, 2000. 18. Kononen, J., Bubendorf, L., Kallioniemi, A., et al. Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat. Med. 4: 844, 1998. 19. Gray, J. W., and Collins, C. Genome changes and gene expression in human solid tumors. Carcinogenesis 21: 443, 2000. 20. Lau, W. Y., Lai, P. B., Leung, M. F., et al. Differential gene expression of hepatocellular carcinoma using cDNA microarray analysis. Oncol. Res. 12: 59, 2000. 21. Welsh, J. B., Zarrinkar, P. P., Sapinoso, L. M., et al. Analysis of gene expression profiles in normal and neoplastic ovarian tissue samples identifies candidate molecular markers of epithelial ovarian cancer. Proc. Natl. Acad. Sci. U.S.A. 98: 1176, 2001. 22. Bruder, C. E., Hirvela, C., Tapia-Paez, I., et al. High

PLASTIC AND RECONSTRUCTIVE SURGERY,

September 1, 2002

23.

24. 25.

26.

27.

28. 29.

30.

31.

32.

33.

34.

35.

36.

37.

resolution deletion analysis of constitutional DNA from neurofibromatosis type 2 (NF2) patients using microarray-CGH. Hum. Mol. Genet. 10: 271, 2001. Lu, J., Liu, Z., Xiong, M., et al. Gene expression profile changes in initiation and progression of squamous cell carcinoma of esophagus. Int. J. Cancer 91: 288, 2001. Cooper, C. S. Applications of microarray technology in breast cancer research. Breast Cancer Res. 3: 158, 2001. Hui, A. B., Lo, K. W., Yin, X. L., Poon, W. S., and Ng, H. K. Detection of multiple gene amplifications in glioblastoma multiforme using array-based comparative genomic hybridization. Lab. Invest. 81: 717, 2001. Eisen, M. B., Spellman, P. T., Brown, P. O., and Botstein, D. Cluster analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. U.S.A. 95: 14863, 1998. Toronen, P., Kolehmainen, M., Wong, G., and Castren, E. Analysis of gene expression data using self-organizing maps. FEBS Lett. 451: 142, 1999. Peacock, E. E. J. Wound Repair, 3rd Ed. Philadelphia: Saunders, 1984. Welsch, P. L., and King, M. C. BRCA1 and BRCA2 and the genetics of breast and ovarian cancer. Hum. Mol. Genet. 10: 705, 2001. Friedman, L. S., Ostermeyer, E. A., Szabo, C. I., et al. Confirmation of BRCA1 by analysis of germline mutations linked to breast and ovarian cancer in ten families. Nat. Genet. 8: 399, 1994. Solomon, J. S., Brunicardi, C. F., and Friedman, J. D. Evaluation and treatment of BRCA-positive patients. Plast. Reconstr. Surg. 105: 714, 2000. Breslow, A. Thickness, cross-sectional areas and depth of invasion in the prognosis of cutaneous melanoma. Ann. Surg. 172: 902, 1970. Loftus, S. K., Chen, Y., Gooden, G., et al. Informatic selection of a neural crest-melanocyte cDNA set for microarray analysis. Proc. Natl. Acad. Sci. U.S.A. 96: 9277, 1999. Wilson, M., DeRisi, J., Kristensen, H. H., et al. Exploring drug-induced alterations in gene expression in Mycobacterium tuberculosis by microarray hybridization. Proc. Natl. Acad. Sci. U.S.A. 96: 12833, 1999. Nuwaysir, E. F., Bittner, M., Trent, J., Barrett, J. C., and Afshari, C. A. Microarrays and toxicology: The advent of toxicogenomics. Mol. Carcinog. 24: 153, 1999. Gerhold, D., Lu, M., Xu, J., et al. Monitoring expression of genes involved in drug metabolism and toxicology using DNA microarrays. Physiol. Genomics 5: 161, 2001. Afshari, C. A., Nuwaysir, E. F., and Barrett, J. C. Application of complementary DNA microarray technology to carcinogen identification, toxicology, and drug safety evaluation. Cancer Res. 59: 4759, 1999.