figure 21.8. Microarrays: (a) spotted; (b) GeneChip.

ability to compare data from one lab to another, and the absence of cross-hybridization caused by mixed clones or conserved sequences within large cDNAs that result in nonspecific cross-hybridization. In addition, variations in expressed gene sequence due to alternative exon usage, truncated message, and even mutations can be readily detected with custom arrays that are otherwise difficult to create from standard UNIGEN database information. Thus, the user is free to generate highly specific custom arrays to interrogate expressed in detail. Several commercial vendors, such as Agilent, have entered the field, which will result in standardization of arrays and results that would otherwise be difficult to achieve in "home brew" microarray facilities. This in turn will enhance their utility for clinical use, where repro-ducibility and reliability are of paramount importance.

c. Synthesized oligomeric DNA: Figure 21.8b illustrates the physical appearance of a typical commercial oligonu-cleotide array (Affymetrix GeneChip), where 25-mer segments of a gene are arrayed from 6' to 3' in "tiles," or areas containing that probe. Typically, 11 to 16 such 25-mers are generated across the coding sequence of the gene, paired with a one-nucleotide mismatch (below) (e.g., PM versus MM, or perfect match versus mismatch). The PM minus MM signal intensity for each pair is calculated, and the mean value for all tiles within a gene is calculated and reported as an expression value. A variety of correction algorithms are then applied to calculate a normalized value for each of the genes on the array, using values from internal standards on the array. In a typical analysis, the values for each gene in each sample are compared one to another in high-dimensional space, and the nearest neighbors, based on any of a number of criteria, are calculated and the patient samples "clustered" to identify those most similar to one another. When done well, very obvious clusters of "like" and "unlike" are readily appreciated, thereby establishing classes of tumor (Figure 21.9). Surprisingly, these clusters typically match known tumor groupings closely. Cases that do not frequently have subsequently been shown to be misdiagnosed. In other cases, the method identifies previously unrecognized subsets within tumor groups, many of which may have important clinical or therapeutic features of value for patient management in the future. In general, whole-transcriptome expression profiling has had an enormous impact on biomedical research and is poised to enter the clinical practice of medicine, particularly in oncology, where gene dysregulation is a regular feature of most cancers. Finding which genes has been the challenge. Microarray technology appears to accomplish that, rapidly and facilely, to great effect.

DNA Based

Analysis of DNA takes many forms. Methods to do so span virtually the entire history of molecular biology, more so than RNA or protein (at least as defined by modern sequencing and characterization methods). Progress in DNA analysis has been remarkable: from crude biochemical methods to coarse RFLP (restriction fragment polymorphism analysis) to modern methods that embody rapid sequence determination and identification of DNA from virtually any source, the ability to obtain precise, voluminous DNA sequence information is now virtually unlimited. The achievements of the human genome sequencing project seem to pale in the face of potential individual whole-genome sequencing within days at minimal cost: all 3.3. billion bases, times two (sense and antisense), times two (both haplotypes). In this context, it also worth mentioning that about one-third of the DNA extracted from a cell is mitochondrial in origin, and mitochondria undergo muta-tional damage over time, with resultant disease onset as well. Analysis of this DNA, too, will be of importance. It is likely that as highly automated, inexpensive methods of doing so become commonplace, broad DNA sequencing will likely become central to analysis of genetic anomalies in cancer and other diseases. At present, however, assays of more limited scope are the norm, as is illustrated here.

Southern Blot

The first widely used method for DNA identification without resort to laborious manual DNA sequencing methods was developed by E.M. Southern in 1975.33 This method is based

on the remarkable specificity of binding between nucleotides in the DNA helix. If the double helix is denatured ("melted"), the single-stranded DNA is then readily annealed with complementary DNA sequences. By denaturing DNA, elec-trophoresing it to separate different fragments by molecular weight, transferring the gel to a solid substrate such as nylon or nitrocellulose membranes, and incubating these filters with radioactive or fluorescently labeled complementary DNA probes (e.g., short sequences sufficient to establish absolute identity), it is possible to identify any sequence from the mix of genomic DNA. A side benefit is that loss or gain of chromosomal DNA can also be ascertained; comparison of a single-copy DNA probe with an amplified gene, for example, allows ready determination of copy number. An example from an MYCN-amplified childhood neuroblastoma is illustrated in Figure 21.10. Here, a variety of tumors with a single copy of the gene show comparably dense bands at the appropriate molecular weight. In contrast, the amplified tumor shows a marked increase in signal strength. Scintigraphy of radioactive DNA probes, fluorography of fluorescently labeled DNA, or densitometry of chromogen-labeled DNA allows precise quantitation of the signal, and therefore a ratio that defines copy number.

Fluorescent In Situ Hybridization (FISH)

The remarkable specificity of DNA hybridization has given rise to many variations on DNA assays. One particularly useful method for clinical purposes is fluorescent in situ hybridization (FISH). In this method, fluorescently or affinity-tagged DNA probes some hundreds to several thousand bases in size are applied to tumor sections, cytospins, or imprints, for example, and hybridized under denaturing then reannealing conditions, conditions similar to those developed for Southern blots. The labeled DNA is now bound to target DNA and readily detected. This method can be used for a variety of purposes, such as detecting amplified genes, as in MYCN amplification in neuroblastoma (Figure 21.11) or, more commonly, to detect HER2/neu amplification in breast figure 21.11. Fluorescent in situ hybridization (FISH); MYCN; tissue section.

figure 21.11. Fluorescent in situ hybridization (FISH); MYCN; tissue section.

figure 21.12. DNA sequence.

cancer. Both are of great prognostic importance, imparting a graver prognosis than would otherwise be the case, and establishing the basis for more-aggressive therapy. FISH, then, is a valuable diagnostic and prognostic tool that begets more aggressive therapy when an amplified gene is detected.

DNA Sequencing

Direct detection of sequence variation in DNA has been greatly facilitated by the advent of high-throughput, automated, multichannel DNA sequencers. These machines were in fact responsible for the rapid progress made in the human genome sequencing project. The technology is now routinely used in research and diagnostic laboratories to determine the identity of, for example, PCR-amplified DNA, and to detect mutations or polymorphisms in genomic DNA. An example of heterozygous mutation of a gene is illustrated in Figure 21.12, where both G and C peaks are identified at the same location, indicating that there is a mutation in one allele of the gene, whereas the other allele remains wild type. Most often in cancer cells the wild-type allele is subsequently lost, resulting in unopposed function of the mutant and total loss of wild-type functionality, common in oncogenes such as p53.

Mutation Detection (WAVE, and Similar Methods)

Indirect means of DNA sequence detection have proliferated, especially before the availability of cheap, fast DNA sequencing. Several of the methods are based on aberrant migration in a gel due to conformational changes in the DNA double helix that results from base mismatches. They retain their importance as screening methods in particular, due to their low cost, high sensitivity, and high throughput, superior even to DNA sequencing. Their value stems from the fact that most mutation or polymorphisms screening assays presume a low incidence of positives. Rather than laboriously sequencing all samples, a prescreen that detects the rare anomalous case is used. When an anomaly is detected, it is simple to confirm the anomaly by direct DNA sequencing of the anom alous case. This is the principle behind SSCP (single-strand conformation polymorphism) and its many derivatives such as dHPLC (denaturing high performance liquid chromatography).34 A particularly useful method has been commercialized by Transgenomics, the so-called WAVE technology. In this method, homo- and hetero-DNA dimers are formed after denaturing and renaturing conditions and immobilization on a hydrophobic column matrix. Increasing concentrations of acetonitrile selectively elute triethylaminoethyl (TEAA)-bound DNA duplexes; heteroduplexes elute first, followed by homoduplexes. Mismatched dimers elute aberrantly, reflected in peak height and shape. Samples that show such anomalies are then subjected to confirmatory DNA sequencing, as described earlier. In this way, a large number of clinical samples can be screened, and only those showing anomalous elution patterns need be confirmed by DNA sequencing. A typical example of mutation detection by WAVE dHPLC is illustrated in Figure 21.13.


An interesting aspect of genomic DNA organization is the presence throughout the genome of millions of singlebase variations between individuals, called SNPs, for single-nucleotide polymorphisms. The HapMap project ( estimates there are 10 million such variants that occur with sufficient frequency as to warrant designation as polymorphisms. An example is illustrated in Figure 21.14, from the HapMap home page. These SNPs are not mutations; mutations occur rarely and are usually either inherited in the germ line from one parent

figure 21.14. Single nucleotide polymorphisms (SNPs): illustration from HapMap.

or arise de novo as a somatic mutation. In contrast, polymorphisms reflect human genetic variation between individuals. During sexual reproduction, chromosomal cross-over events result in the creation of mosaics of large blocks of DNA ("haplotype blocks") inherited from one or the other parent (Figure 21.15, also from the HapMap home page). This overall pattern is called a haplotype. Because these blocks of DNA are inherited as blocks, each individual inherits two haplotypes, one from each parent. These haplotypes have generally arisen as much as 150,000 years ago in the human germ line in Africa and have undergone further diversification with the spread of humanity over the globe. They show marked ethnic variation, resulting from expansion of isolated groups of humans, and variable age, due to additional polymorphisms that arose subsequent to dispersion of ethnic populations. The aggregate effect is a fingerprint of genetic diversity that can be used to identify the genetic background of any individual and, further, to associate that genetic makeup with disease propensity, severity, response to drugs, and many other parameters. Polymorphic variants of CYP450, for example, are powerful tools to predict drug metabolism in individuals, potentially guiding dosage for the individual as opposed to the "average." In oncology, idiosyncratic responses to drugs such as cyclophosphamide are linked to specific polymorphic variants of CYP450.35,36 This burgeoning field, termed "pharmacogenetics," is likely to increasingly dictate individualized therapy based on haplotype variants of critical genes such as CYP450 and many others that in aggregate dictate an individual's disease susceptibility and likely response to therapy.37-42

Whole-Genome Assays for SNPs

Given the apparent importance of polymorphic variation among individuals to important clinical issues such as disease susceptibility and response to therapy and prognosis, a practical method for genotyping individuals is needed for clinical use. The HapMap project referenced earlier is providing the data, including tag SNPs that will enable identification of haplotype blocks and therefore the individual's haplotype; only about 500,000 of these are needed to predict an individual's genotype, as opposed to direct detection of all 10 million

SNPs, or any large fraction thereof. A variety of methods for large-scale detection of SNPs are being developed. At present, two approaches merit discussion, as both are used for whole-genome SNP analysis.

a. Bead Arrays: The first, a multiplex bead assay developed by Michael et al.43 and commercialized by Illumina, Inc. (San Diego, CA), termed Beadarray, is capable of identifying up to approximately 1,500 SNPs per assay, by adsorbing a random bead mixture (~1,000 to ~1,500 types, specific for one SNP) onto the etched ends of approximately 50,000 optical fiber complexes. Each fiber binds a bead; with 1,500 bead types, each is represented with about 30-fold redundancy. A unique decoding scheme, based on unique "address" sequences incorporated into PCR-amplified genomic DNA sequences, is then used to identify each bead and assign a call value for each SNP.44,45 The result is about a 90+% successful call for SNPs from dbSNP. High-quality, validated SNPs results in more than 95% calls. SNPs can be drawn from the entire genome, or regionally at high density. This flexibility lends itself to disease association studies, where a locus is first identified, then probed in detail with regional SNP beads. However, there is a significant design and validation component for each SNP, and the cost per SNP is higher than SNP arrays (see following). Nonetheless, this technology has been widely used by the HapMap project, with great success.

b. SNP photolithography arrays: The second high-throughput method, available commercially from Affymetrix, is based on the same technology employed for the GeneChips discussed earlier. For SNP detection, the region of interest for a specific allele, termed "A" (e.g., the SNP and 14 surround ing nucleotides, designated -7 through +7) is represented by 25-mer oligonucleotide sequences, with the specific nucleotide of interest represented in position 13. Target sequences bind most stably to such sequences and yield the brightest signal. The brightest signal should thus occur for the 25-mer representing the SNP at position 0. To further control for nonspecific hybridization, a one-base mismatch sequence is tiled on the array immediately below the perfect match, yielding a 2 x 7 matrix of 14 tiles. This is then doubled, to include the same setup for the alternate allele ("B"), yielding a 4 x 7 matrix. Finally, this is replicated in the antisense direction, for a grand total of 56 tiles. To detect the target DNA on these probes, genomic DNA is fragmented by one or more restriction endonucleases, PCR primer adapters ligated to the ends, and preferentially PCR amplified by fragment size (e.g., 400-800 bp). These PCR products are then fragmented, denatured, end labeled, and hybridized to the SNP arrays, resulting in the hybridization pattern illustrated in Figure 21.16. With suitable software, the genotype for any given SNP can be called with greater than 95% confidence.46,47 At present, arrays capable of interrogating up to 126,000 SNPs are available, and the number is expected to rise to 500,000 and more in the near future. With reasonable selection of SNPs, this density will allow haplotyping with unprecedented precision and sensitivity across the entire genome. The disadvantage is that, unlike the bead array system already described, the user cannot specify specific SNPs for analysis.

10 Ways To Fight Off Cancer

10 Ways To Fight Off Cancer

Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.

Get My Free Ebook

Post a comment