Comprehensive gene expression analysis by transcript profiling

Plant Molecular Biology 48: 75–97, 2002. © 2002 Kluwer Academic Publishers. Printed in the Netherlands.

75

Comprehensive gene expression analysis by transcript profiling Jonathan Donson∗ , Yiwen Fang, Gregg Espiritu-Santo, Weimei Xing, Andres Salazar, Susie Miyamoto, Veronica Armendarez and Wayne Volkmuth Ceres, Inc., 3007 Malibu Canyon Road., Malibu, CA 90265, USA (∗ author for correspondence; e-mail [email protected])

Key words: cDNA-AFLP, gene expression, microarray, oligonucleotide-based array, SAGE, transcript profiling

Abstract After the completion of the genomic sequence of Arabidopsis thaliana, it is now a priority to identify all the genes, their patterns of expression and functions. Transcript profiling is playing a substantial role in annotating and determining gene functions, having advanced from one-gene-at-a-time methods to technologies that provide a holistic view of the genome. In this review, comprehensive transcript profiling methodologies are described, including two that are used extensively by the authors, cDNA-AFLP and cDNA microarraying. Both these technologies illustrate the requirement to integrate molecular biology, automation, LIMS and data analysis. With so much uncharted territory in the Arabidopsis genome, and the desire to tackle complex biological traits, such integrated systems will provide a rich source of data for the correlative, functional annotation of genes. Abbreviations: cDNA-AFLP, cDNA amplified fragment length polymorphism; EST, expressed sequence tag; LIMS, laboratory information management system; MPSS, massively parallel signature sequencing; PCR, polymerase chain reaction; RDA, representational difference analysis; SAGE, serial analysis of gene expression; SSH, suppression subtractive hybridization Introduction The publication of the ‘complete’ nuclear chromosomal sequence of Arabidopsis thaliana, following on from publication of the plant chloroplast and mitochondrial genome sequences (Unseld et al., 1997; Sato et al., 1999), has equipped biologists with the opportunity to describe a plant’s basic genetic determinants (Arabidopsis Genome Initiative, 2000). This monumental achievement begs the question, ‘How does this arrangement of nuclear chromosomal nucleotide sequences provide for the making and survival of a plant?’ Bioinformatic analysis allows the theoretical discrimination of open reading frames and genetic control elements; however, such studies by their nature carry inaccuracies (Pavy et al., 1999; Guigó et al., 2000; Cho and Walbot, 2001). Even where our knowledge of genomic structure allows us to infer the AFLP
is a registered trademark of Keygene N.V. The AFLP
technology is covered by patents owned by Keygene N.V.

presence of genes, we can say little about the functioning or control of these transcriptional units. While 69% of publicly annotated Arabidopsis genes have some sequence similarity to those with known functions in other organisms, only 9% of genes have been characterized experimentally (Arabidopsis Genome Initiative, 2000). Information on both the physical and functional annotation of the genome can be gained through transcript profiling (Hughes et al., 2001; Shoemaker et al., 2001). In recent years, transcript profiling has become synonymous with gene expression analysis, largely because of the technical difficulties and greater molecular complexity of proteomics and metabolomics (Smith, 2000; see other papers in this issue). Such correlations are acceptable in many cases (Celis et al., 2000), even though the term ‘gene expression’ is often used to refer more directly to the compendium of gene products that ultimately cause cellular responses, and these are more often than not proteins. In some

76 instances protein levels are not reflected by alterations in mRNA level, and their activities are often controlled by post-translational modifications (Gygi et al., 1999). This means one has to take care with both experimental design and interpretation if one wishes to extrapolate transcript levels to those of protein (proteomics) or protein activity (metabolomics). Even so, it is generally accepted and there is much experimental evidence to support the statement that while posttranscriptional events play a role in modulating gene expression the primary level of control is at transcription. In this framework, the accessibility of transcript profiling in recent years has allowed the establishment of various high-throughput methodologies of gene expression analysis. These methodologies differ in their convenience, expense, number of transcripts assayed and sensitivity (Table 1; Kuhn, 2001). However, as revealed by the genomic sequencing projects, full automation and data management are essential factors in all comprehensive transcript profiling.

Methods of transcript profiling Transcript profiling has been going on in one form or another for over 25 years (Bishop et al., 1974; Alwine et al., 1977; for review, see Goldberg, 2001). This period has spawned techniques such as northern transfer hybridization, S1 nuclease analysis and in situ hybridization. While these methods are characterized by good, well-defined sensitivities, they are timeconsuming, and therefore best suited for the in-depth analysis of a small number of genes. By comparison, current high-throughput transcript profiling technologies have relatively poorly defined sensitivities, so these early methods provide both a valuable means of confirming and extending results obtained with the more global approaches. Such ‘single-gene’ methods are part of the lexicon of most molecular biology labs (Ausubel et al., 2001; Sambrook and Russell, 2001), and have been extensively reviewed, and as such will not be considered further in this article. Instead, we will confine discussion to high-throughput approaches that allow for a global view of transcript levels, and where possible we will reference work with plants. These methods can be divided into two classes: (1) Direct Analysis, including procedures involving nucleotide sequencing and fragment sizing; and (2) Indirect Analysis, involving nucleic acid hybridization of mRNA or cDNA fragments. We have used two methods extensively, cDNA microarraying

and cDNA-amplified restriction fragment polymorphism (cDNA-AFLP), as representative technologies from each of these groups. All global methods of gene expression analysis demand good Laboratory Information Management Systems (LIMS), automation, and powerful data management and mining systems. Because these aspects have been introduced most widely in conjunction with cDNA microarraying, we will review them together with this technology.

Direct analysis – nucleotide sequencing Large-scale expressed sequence tag (EST) sequencing Generating sequences from cDNA fragments serves two purposes, the discovery of new genes and the assessment of their expression levels in the representative tissue (Ewing et al., 1999; Mekhedov et al., 2000; White et al., 2000). The basis of the approach is that the level of an mRNA species in a specific tissue is reflected by the frequency of occurrence of its corresponding EST in a cDNA library. In this respect the technology is distinct from ratio-based methodologies, such as microarraying, in being immediately quantitative (Bohnert et al., 2001). EST technologies are attractive because they do not rely on established sequence data from the organism under study, and they also fit well with labs already equipped to carry out high-throughput DNA sequencing (Adams et al., 1991). However, even at a few dollars per sequence the process can be expensive if one desires to progress beyond cursory screening of abundant mRNAs to indepth analysis (Ohlrogge and Benning, 2000). In addition to the statistical problems in sampling small numbers from a large population (Audic and Claverie, 1997), there are also problems of bias in cloning and cDNA synthesis, though this last problem and others associated with data normalization are not specific to this technology. Auxiliary techniques are available that reduce the amount of sequencing. These include subtraction hybridization (Sargent, 1987) and related methods, representational difference analysis (RDA; Hubank and Schatz, 1994) and suppression subtractive hybridization (SSH; Diatchenko et al., 1996). Numerous variants on these technology themes are also available (Bonaldo et al., 1996; Sagerstrom et al., 1997).

high

high

Fragment sizing low

Indirect Hybridization high

Direct Nucleotide sequencing low high

[b] requires comprehensive reference database [c] short tags can show redundancy in database ∗ [b]∗ [d] need for extensive, unique automation or contract work

[4] less expensive than EST sequencing

[h] potential cross-hybridization problems [i] extensive clone and PCR fragment curation

∗ [b] and [d]∗ [8] highly amenable to automation [9] does not require clone and PCR fragment curation [g] requires accurate gene annotation [10] oligonucleotide design can avoid cross-hybridization [11] flexibility in defining extent of transcriptome to be analysed

cDNA microarrays ∗ [2], [8] and [11]∗ [12] ratio-based: cDNA samples are processed in parallel [13] compatible with labs having existing clone libraries

Oligo-Chips

∗ [6]∗

cDNA-AFLP

[7] specific primer sets overcome problems with DD

[6] low set-up costs

[5] longer tags than for SAGE allows more accurate annotation

[e] false-positive, sensitivity and reproductivity problems [f] extensive band isolation + sequencing ∗ [b] and/or [f]∗

∗ [a]∗

∗ [1] and [3]∗

∗ [1] and [4]∗

[a] statistically significant coverage can be expensive

Disadvantages

[1] numerable in nature [2] no requirement for existing sequence data [3] compatible with existing sequencing labs

Advantages

DD

MPSS

SAGE

EST sequencing

Cost of transcriptome screening Technology Partial Comprehensive 5000–10 000 genes >10 000 genes

Table 1. Major advantages and disadvantages of transcript profiling technologies. SAGE, serial analysis of gene expression; MPSS, massively parallel signature sequencing; DD, differential display; cDNA-AFLP, cDNA-amplified fragment length polymorphism analysis; Oligo-Chips, oligonucleotide-based arrays. Duplicate alphabet and numerical numbers (e.g. ∗ [a]∗ ) indicate advantages or disadvantages common to methods already cite din this table.

77

78 Serial analysis of gene expression (SAGE) By developing SAGE, Velculescu et al. (1995) ingeniously addressed the problem of reducing the costs of traditional EST sequencing, and further modifications have allowed this procedure to handle small amounts of tissue (1:10 000

NG

Global

RT-PCR NG

>2-fold and > 3× background Genes with >95% probable correlation with cosine test wave

>2-fold

RT-PCR

limit of detection = 1.74 (Incyte) >1.5-fold Correlation of high express and with occurence in cDNA libraries. >2-fold RT-PCR

>2.5-fold

Northern analysis on 40 genes

Northern analysis of 4 genes

>1.6-fold Average of triplicate spots

Northern analysis of 2 genes for 3 time points

Signal intensities >2000

NG

Northern analysis of 4 genes

>2-fold

NG on 2 replicate experiments

>3 fold in 3 out of 4 spots

>2-fold

NG

Northern analysis of 2 genes

Northern and sequence analysis of 7 fragments RT-PCR

Northern analysis of 4 genes 1 fragments verified in +3/+3 analysis Sequence of 200 fragments Fragments verified in +3/+2 analysis Northerns - 4 isolated cDNAs

Experimental confirmation

3-fold

>2.24–3.32

Equal lane loading Visual differences Constitutively expressed genes Equal lane loading >2-fold Constitutively expressed genes

Equal lane loading Visual differences Constitutively expressed genes

Global

NG

Thresholds

Equal lane loading Visual differences Constitutively expressed genes

Normalization

NG

NG

NG

NG

NG

NG

1 copy per cell

Sensitivity

Table 2. Comprehensive transcript profiling of plant species by array and cDNA-AFLP technologies. The table lists studies up to 1 April 2001. NG, information that is not given in the respective reference. In addition to the studies given in this table, the Arabidopsis Functional Consortium (AFGC) displays data from multiple cDNA microarray experiments (ca. 7500 genes) at http://afgc.stanford.edu/ (Wisman and Ohlrogge, 2000). This site allows the public to analyze data and set their own thresholds. The size of the database (ca. 200 arrays), variety of experiments and in particular its free access make it a substantial resource in plant research. Schaffer et al. (2001) have published some of the AFGC data.

81

A. thaliana

A. thaliana

A. thaliana A. thaliana

Oryza sativa Panicle, callus, leaf and root

Schena et al. (1995)

Schenk et al. (2000)

Seki et al. (2001)

Wang et al. (2000b)

Yazaki et al. (2000)

Nitrate induction

jasmonate and ethylene treatments Drought and cold stress

3 -untranslated region - gene-specific primers

2 vector primers - 5 amino-modified Microarray - 1265 ESTs Full insert - vector primers

Microarray - ca. 1300 full-length cDNAs 2 vector primers Microarray - 5524 unique gene ESTs (Incyte) http://www.incyte.com

2 vector primers - 5 amino-modified

2 vector primers - 5 amino-modified Microarray - ca. 7800 unique ESTs (AFGC) 2 vector primers, http://afgc.stanford.edu/ Microarray - 45 ESTs

2 vector primers - 5 amino-modified Microarray - 150 defense-related ESTs 2 vector primers Microarray - 1443 ESTs

Microarray - 9861 cDNAs

Details of technology

Defense responses to Alternaria Microarray - 2375 defense-assoc. brassicicola, Salicylic acid, methyl and regulatory ESTs

Leaves v roots. HAT-transgenic v wt

flower buds, open Flowers Circadian and diurnal rhythms

A. thaliana

Leaf, roots,

A. thaliana

Ruan et al. (1998)

Wounding and insect feeding

Reymond et al. (2000) A. thaliana

Schaffer et al. (2001)

Systemic acquired resistance

A. thaliana

Petersen et al. (2000)

Study

Species

Ref

Table 2 continued.

1:20 000

Northern analysis of 3 genes

>2-fold and >2× background >2-fold

Alpha-tubulin control

Global

>2–3-fold

Limit of detection = 1.74 (Incyte) NG

Northern analysis of 80 genes Northern analysis of multiple genes

>2-fold >2-fold

NG

Northern analysis of 3 genes

NG

Northern analysis of 6 genes

Northern analysis of 3 genes

>2-fold

>2-fold and >3× background

Experimental confirmation

Thresholds

Equal amounts of mRNA dosed NG with equal amounts of human acetylcholine receptor mRNA Global >2.5-fold 2 standard devs > background

1:100 000 (Incyte) Global

1:300 000

NG

1:50 000

Global

Internal controls

≥1:100 000 1:50 000

16 control genes

Global

Normalization

NG

NG

Sensitivity

82

83

Figure 2. Establishment (A) and interrogation (B) of the cDNA-AFLP Reference Database for Arabidopsis thaliana. 1, 27 tissues were harvested to create 11 pooled samples; 2, mRNAs from these pools were analyzed by cDNA-AFLP analysis (+2/+3 differentiating nucleotides, 2 enzyme combinations); 3, 4, all cDNA-AFLP bands were eluted from the gels, PCR-amplified and sequenced; 5, the sequence information for each fragment was linked to its gel mobility in a database; 6, this database can be queried with AFLP
profiles to link band mobilities and intensities (expression data) with the individual band’s sequence (gene). AFLP
is a registered trademark of Keygene N.V. The AFLP
technology is covered by patents owned by Keygene N.V.

Of the other ‘selective fragment amplification’ methods, READS and TOGA generate only one fragment for each mRNA species. Arguments can be made in favor of the simplicity of such ‘one-fragment’ systems, but counter arguments can also be put forward for the value of redundancy in data sets, as illustrated by the multiple fragments per mRNA species obtained with cDNA-AFLP and GeneCalling. Stanssens and Zabeau have modified the cDNA-AFLP method so that only one restriction fragment is monitored for each cDNA (Breyne and Zabeau, 2001). Using this modified method, they profiled transcript levels during the cell cycle of tobacco, analyzing 18 000 cDNAAFLP tags, of which 10% exhibited a modulated banding pattern.

Indirect analysis The principles underlying the hybridization of complementary nucleotide sequences are embodied in the structure of duplexed nucleic acids, and have been exploited experimentally for decades (Gillespie and Spiegelman, 1965). In this time, nucleic acid hybridization has been used in a variety of guises in the quantification of plant RNA levels; in the 1970s, it was used to study sequence complexity (Goldberg et al., 1978). Also in that decade, Southern developed a method using a solid support in hybridization studies of DNA fragments separated by gel electrophoresis (Southern, 1975). This led to a major advance in the analysis of gene expression with the development of northern-transfer hybridization (Alwine et al., 1977). This technique has been immensely useful over the

84 years, though ironically the approach is less global and the method less genomic in nature than preceding solution-based systems. With the availability of nucleotide sequences and clones as physical reagents, hybridization-based approaches now allow for the simultaneous analysis of tens of thousands of genes; quite literally one can globally survey the transcription of a plant by hybridization. The interest in this form of transcript profiling has been spurred by the development of two parallel microarray-based technologies, one based on spotting cDNA fragments (cDNA microarrays; Schena et al., 1995), the other, the arrayed synthesis of oligonucleotides (GeneChips; Lockhart et al., 1996). These two methods have been extensively reviewed in recent years (Duggan et al., 1999; Lipshutz et al., 1999). Oligonucleotide-based arrays GeneChips, oligonucleotide-based arrays produced by Affymetrix, have been made to about 8200 different Arabidopsis ESTs (Zhu and Wang, 2000; Harmer et al., 2000) and 1500 maize ESTs (Baldwin et al., 1999); a full transcriptome Arabidopsis GeneChip is also under development (Zhu and Wang, 2000; Cho and Walbot, 2001). These arrays are produced by the synthesis of oligonucleotides directly onto a solid matrix using photolithographic masks to determine the correct sequence (Lockhart et al., 1996; Warrington et al., 2000). For the commercially available Arabidopsis GeneChips, 16 ‘probe pairs’ were synthesized per gene. One set of 16 consisted of ‘mismatch oligonucleotides’ that were identical to a ‘perfect match’ set except for the 13th nucleotide in each 25-mer. These mismatched oligonucleotides are used to assess cross-hybridization and local background signals. Affymetrix GeneChips are expensive to make, not least because of the need to manufacture the glass masks. However, digital micromirror arrays that form virtual masks may provide a cheaper and more accessible alternative (Singh-Gasson et al., 1999). There are other promising oligonucleotide-based technologies, including those that array 5 -terminally modified oligonucleotides (Kane et al., 2000; Okamoto et al., 2000), unmodified oligonucleotides (Ten Bosch et al., 2000), and phosphoamidites for the in situ synthesis of oligonucleotides (Shoemaker et al., 2001). In this latter example, exon arrays were constructed spanning 50 slides, containing 1090 408 60-mer probes representing 442 785 exons (Shoe-

maker et al., 2001). This system could reliably detect transcripts at one copy per cell and the results correlated closely with comparable cDNA arrays (Hughes et al., 2001). Where they found discrepancies, these involved genes that were members of multi-gene families. Shoemaker et al. (2001) also produced tiling arrays consisting of overlapping oligonucleotides covering an entire genomic region. Such arrays are able to define gene structure because exons of the same transcript show identical expression patterns across all experimental treatments. Similar tiling arrays are proposed for the annotation of A. thaliana (Cho and Walbot, 2001). Although such arrays are high-throughput in nature, full-length cDNAs still have a role to play in annotation because of the reliability and precision of their sequences. Two forms of ink-jet printer (bubble-jet and piezoelectric) have been used for producing oligonucleotide arrays, and these systems promise reduced spot sizes. The delivery of phosphoamidite monomers (Blanchard et al., 1996; Shoemaker et al., 2001) has obvious advantages over the dispensing of synthesized oligonucleotides (Okamoto et al., 2000; Ansubel et al., 2001), as problems of washing and carry-over are greatly simplified. In a comparison of the sensitivity of detection of 50-mer amino-linker modified oligonucleotides over PCR products about 360 bp in length, no significant difference was found provided appropriate design criteria were followed (Kane et al., 2000). Ten Bosch et al. (2000) came to a similar conclusion, but also presented evidence for the differentiation of overlapping yeast genes. Using this technology, they have synthesized a full transcriptome oligonucleotide array for yeast. cDNA-microarrays The basic idea of spotted nucleic acid arrays for gene expression analysis is not new and has been used in some form for over twenty years (Kafatos et al., 1979). Recently, cDNA microarraying has caused a revolution in molecular biology, invited by the availability of genomic sequences. Its rapid incorporation into plant research labs is testament to the immediacy of gene expression studies and the demand for a more holistic approach throughout plant molecular biology. The encouragement of this global view of biology through the marriage of automation to the pertinent protocols has benefited from the altruistic behavior of its developers, who have actively dispensed their knowledge (http://cmgm.stanford.edu/pbrown/).

85

Figure 3. Schema for the generation and interrogation of cDNA microarrays. 1. PCR fragments are amplified from cDNA clones. 2. PCR fragments are purified, quality control-tested on agarose gels, buffer-adjusted and spotted on to coated glass microscope slides by means of a computer-controlled X-Y-Z robot. 3. Total or poly(A)+ RNA from both the test and reference sample is fluorescently labeled with either Cy3 or Cy5 nucleotides by reverse transcription. 4. The fluorescently labeled cDNAs are pooled and allowed to hybridize to the array. After hybridization, the array is washed to remove unhybridized molecules. The amount of fluorescent cDNA that hybridizes to each of the spots is then measured by a fluorescent scanner, through sequential exposures to excitation wavelengths specific to the two Cy dyes. Spectrally characteristic emissions from these dyes are captured with a confocal sensor. 5. Emission intensities are extracted and these are linked to the corresponding clone ID, and through extensive LIMS (Figure 4) to experimental details, array information and hybridization conditions. These data can be normalized (6) and merged to produce a pseudo-colored image (7), or analyzed as part of multiple experiments, by ratio-based data mining tools (8) (Figure 5).

Although DNA microarrays can be used in a variety of genomic studies, their major use to date has been in gene expression analysis, particularly in the case of plant research. Reviews of methodology are available (Eisen and Brown, 1999; Hegde et al., 2000, Ausubel et al., 2001), as are two recent books on the subject (Schena, 1999, 2000). The Internet is also a rich source of information on experimentation with microarrays, including reviews of equipment, protocols, and suppliers of pre-fabricated arrays (Ferea and Brown, 1999; http://www.gene-chips.com/; http://www.deathstarinc.com/science/biology/chips. html). The basic strategy for the manufacture and

interrogation of cDNA microarrays is described in Figure 3. As microarray experiments are conducted under conditions where there is a large excess of immobilized probe to labeled cDNA, the kinetics of hybridization are pseudo-first-order. Therefore, the intensity of the fluor’s emission is proportional to the level of the specific labeled cDNA in the hybridization solution. A particularly attractive aspect of the system is that it is ratio-based, with the two cDNA samples under analysis being hybridized in parallel. This is important because it removes the variability of array fabrication and individual hybridizations from the equation. This aspect makes radioactive nylon-

86 or nitrocellulose-based arrays less attractive (Desprez et al., 1998). Indeed, while absolute quantification of RNA levels can be inferred by incorporating appropriate dosing controls, generally it is the relative increase of a mRNA between different treatments or tissues that is of interest. For mRNA copy-number calculations, the highly optimized kinetics of olignucleotide arrays offers the better solution. Because spots can be arrayed at distances as low as 150 µm center-to-center, it is self-evident that representative PCR fragments from the complete transcriptome of A. thaliana can be deposited on a single slide. Full transcriptome arrays hold a huge attraction (DeRisi et al., 1997; Hihara et al., 2001), as one can assay traits without preconceived ideas. Even so, judiciously selected gene subsets can also be used effectively (Aharoni et al., 2000; Girke et al., 2000; Reymond et al., 2000; Schenk et al., 2000; Kawasaki et al., 2001). RDA (Welford et al., 1998; Nelson and Denny, 1999), SSH (Yang et al., 1999) and differential display-based methods (Display Systems, Vista, CA) have all been used to refine the complexity of probes on microarrays. Indeed, as microarraying becomes a standard laboratory technology emphasis may shift from full transcriptome coverage to systems that allow for more rapid screening of gene subsets. Then, full transcriptome arrays may be used largely as a means of refining candidate probes for smaller arrays. In this vein, Genometrix produces slides (VistaArrays) which contain 96 miniarrays, each of 256 probes, which can be used to process samples in parallel (Eggers, 2000). Automation and laboratory information management systems (LIMS) Although the first published report on cDNA microarrays in 1995 included 48 Arabidopsis ESTs (Schena et al., 1995), it is only in the last year that there has been an explosion of plant related microarray studies (Table 2). This suggests that the revolution going on in plant molecular biology labs has less to do with microarraying than with the introduction of automation. Up to 1 April 2001, there have been 18 published plant microarray papers (17 cDNA microarray, 1 GeneChip), of which 12 have involved analysis of A. thaliana. In addition, work has been reported on rice, maize, strawberry, petunia, ice plant, lima bean and the cyanobacterium Synechocystis. We have used cDNA microarraying to study the Arabidopsis transcriptome (Fang et al., 2000; Okamuro et al., 2000). As of the end of the millennium, these mi-

croarrays allowed profiling of greater than 40% of the transcriptome (Fang et al., 2000). To achieve this preferentially large coverage at the time, we made use of the A. thaliana full-length cDNA-sequencing program at Ceres, Inc. From our collection of cDNA clones, a set of about 10,000 was selected and gene-specific primers were designed close to the 3 end of each individual cDNA. To ensure the quality of the PCR product from these clones we instigated a highly automated system that tracks samples, including clones, PCR fragments and tissues, from the start of the process to final data analysis (Figure 4). While such a LIMS system might seem a luxury, it is in fact a substantial time-saver, as it reduces the number of quality control assays one has to perform with a more manual process. Automation, LIMS and effective data management are essential components of comprehensive transcript profiling (Ermolaeva et al., 1998; Bassett et al., 1999). We made several conscious decisions with regard to the generation of PCR fragments. On our robotics we use disposable tips for liquid dispensing to avoid cross-contamination of reagents and templates. To further ensure the specificity of the probes on the microarrays we also use gene specific primers. For anyone running high-throughput robotic procedures such as microarraying, assays to check for cross contamination are a constant concern, though with planning some of these issues can be addressed as part of data collected from every array hybridization. Contamination of microarrays can be insidious if through ill fortune low-expressing gene probes become contaminated with DNA complementary to high expressing genes. Sequencing or any other common assay method would not readily detect such contaminations. Consequently, preventative action is the best policy. It is interesting that apart from the work performed at Ceres, Inc., most current plant studies do not use gene-specific primers (Fang et al., 2000; Okamuro et al., 2000). Also, in contrast to the arrays we have produced, only Seki et al. (2001) have used full-length cDNAs. There are a large number of robots and spotting pins sold commercially for the printing of arrays (Meldrum, 2000; Mittal, 2001), as well as the option to build your own robot (http://cmgm.stanford.edu/ pbrown/), adapt existing robotics (Macas et al., 1998), or resort to a hand-held device (www.vpscientific.com). Commercial robots invariably come with good software for designing arrays and tracking clones, or it can be purchased separately (Zhou et al., 2000).

87

Figure 4. Laboratory Information Management System for cDNA microarray analysis. Arrows indicate the flow of materials and processes. Data flow in the facility is automated through the use of multiple robots including automated plate handling, bar codes and specialized software, except at instances where manual input is need. In these cases, stations have been set up with web interfaces to provided easy database input. Data reports are sent to and retrieved from specific groups of tables in the appropriate database by the individual processes and stations (indicated by common shading).

Microscope slides are the popular choice as an array support because they are non-porous and show low autofluorescence. There is now a large selection of commercially available coated microscope slides with amine or aldehyde surface chemistries, that have good hydrophobicity and that enhance the binding of DNA (Celis et al., 2000; Mittal, 2001). Even so, the original option, polylysine coated slides, is still widely used (Eisen and Brown, 1999; Fang et al., 2000; Kawasaki et al., 2001). In some instances amino-linked PCR fragments are preferred (Ruan et al., 1998; Aharoni et al., 2000; Maleck et al., 2000; Wang et al., 2000b), on the basis that this enhances the sensitivity of the system (Schena et al., 1996; Aharoni et al., 2000; Kawasaki et al., 2001). Sensitivities in the range 1:20 000 (Yazaki et al., 2000) to 1:300 000 (Seki et al.,

2001) have been reported in plant studies, though without any clear correlation with slide chemistry. Cross-hybridization During the analysis of experimental data particular attention has to be given to whether probes spotted on to the glass slide cross-hybridize to heterologous cDNAs. A discovery from the genomic sequencing project was that 65% of the genes in Arabidopsis are members of gene families, and the proportion of gene families with more than two members is considerably higher than in other sequenced eukaryotes (Arabidopsis Genome Initiative, 2000; Bevan et al., 2001). Indeed, the evolution of Arabidopsis is thought to have involved a number of large-scale duplications, followed by subsequent gene loss (Blanc et al., 2000; Vision et al., 2000). Even greater gene duplication is likely to be

88 found with other plant species. However, duplication of sequence, does not necessarily mean duplication of function (Cho and Walbot, 2001), so for comprehensive transcript profiling one might still wish to assay all genes. Two plant-based studies have attempted an assessment of this problem (Girke et al., 2000; McGonigle et al., 2000). In a study by Girke et al. (2000) a FAD2 gene from Arabidopsis was arrayed in three different forms of identical length and GC content, but with nucleotide similarities to the native gene of 100%, 90% and 80%, respectively. The fragment of 80% similarity showed no detectable hybridization. In the same paper, four ferredoxin sequences and three acyl-ACP-desaturase sequences from other species with more variable clusters of similarity to the Arabidopsis sequences showed cross hybridization thresholds between 60–70%. On this basis the authors estimated that cross-hybridization occurred with their system if related genes have greater than 70– 80% sequence identity. Similarly, in a study of the maize glutathione S-transferase gene family, McGonigle et al. (2000) found that gene expression data behaved independently for genes below ca. 80% similarity. Studies on yeast estimate that cross hybridization becomes significant at or above 75% sequence similarity (Spellman et al., 1998). Given that crosshybridization will occur, solutions include a more judicial choice of probes, representing 3 -UTR regions of mRNAs (Yazaki et al., 2000), or the use of oligonucleotide-based microarrays that can discriminate between nucleotide sequences of less than 93% similarity (Lipshutz et al., 1999). However, even if oligonucleotide arrays eventually become the predominate microarray platform, one area where cDNA microarrays, with their longer probes, may have an advantage is in experiments with heterologous species; for example, Arabidopsis microarrays might be used in parallel experiments with Brassica species (Girke et al., 2000). Sample preparation The choice of samples and the quality of labeled cDNAs are major factors in determining the sensitivity of microarraying. When it comes to sampling of tissues, studies on plant microarrays and GeneChips have illustrated how genes show diurnal and circadian responses (Harmer et al., 2000; Schaffer et al., 2001). These phenomena are not new observations (Kreps and Kay, 1997), though the data do provide an estimate of the number of similarly affected genes

in Arabidopsis. However, these studies, and an example of sample-to-sample variation in the paper of Kawasaki et al. (2001), illustrate at a molecular level the importance of good experimental and sampling design, if one aspires to assay a single variable. Inevitably, selection of tissues must be as sophisticated as the methods used for transcript profiling, as ultimately one wishes to profile transcripts in one or a small cluster of cells. To date, published plant microarray experiments have involved the analysis of heterogeneous samples, either whole plants or tissues, each of which consists of multiple cell types, spatially distinct from one another. Differences in mRNA levels in a small number of cells may be swamped by the dilution effect of RNA from millions of different cells. There is no universal answer to this problem, and possible solutions include cell culturing, cell sorting, ablation experiments and the judicious use of mutants (Sheen et al., 1995; McCabe et al., 1997; Liu et al., 1999; Reymond et al., 2000; Bohnert et al., 2001). Laser capture microdissection (LCM) has been used in animal systems for the procurement of microscopic and pure subpopulations of cells from tissues (Emmert-Buck et al., 1996), but its use in plant tissues may be complicated by the cell walls. Ultimately, all modern profiling methods will be at their most powerful in understanding mechanisms of development and differentiation when combined with classical genetic and cytological approaches. Conclusions from expression profiling are influenced by the ability to make comparisons across a large number of diverse and precisely executed experiments. In designing an experiment one must decide on the most appropriate control samples to use in such comparisons. For pairwise analysis this is generally a simple matter, but for experiments with multiple samples universal controls might be desirable to allow comparison across different data sets (Eisen and Brown, 1999). Care must also be taken in the choice of controls spiked into the hybridization solution to monitor microarray performance and to facilitate normalization. The experimental design process extends to the seemingly obvious repetition of experiments in order to assess variability, though this is not always followed (Lee et al., 2000). Having sampled your tissue, one needs to decide what RNA fraction should be analyzed. Differential transcript levels between similar samples, perhaps treated and untreated, can be determined by measuring mRNA levels in total cell RNA. This allows a correlation to be drawn between a change in level of a

89 particular transcript and the treatment. However, since mRNA levels do not always correlate with protein levels the extrapolation to measuring translation is less easily made (Gygi et al., 1999). The correlation of differential transcript levels with the translationally active versus the translationally inactive pools of mRNA is particularly difficult if the samples are disparate in nature. A solution to this is through the labeling of polysome-associated RNA (translationally active mRNA) as against mRNAs associated with ribonucleoprotein particles or monosomes (translationally inactive mRNA) (Zong et al., 1999). Plant polysome fractions can be readily separated by sucrose gradient centrifugation (Jackson and Larkins, 1976). This process can be taken a step further in that purification of membrane-bound polysomes will afford a selection of mRNAs of proteins destined for membranes and secretion (Diehn et al., 2000). Similar strategies might also be envisaged for the study of other subcellular localization. cDNA microarray analysis traditionally uses a relatively large amount of mRNA (1–2 µg). This is usually labeled by the incorporation of fluorescently labeled nucleotides into first-strand cDNA or by the cross linking of N-hydroxysuccinimide-activated fluorescent dyes to aminoallyl groups incorporated into the cDNA (http://cmgm.stanford.edu/pbrown/). A consequence of the need to work with more refined tissues is that these levels of mRNA may not be available. In conjunction with LCM, an amplification method based on cyclical rounds of T7 polymerase in vitro transcription has been developed (Luo et al., 1999; Salunga et al., 1999; Wang et al., 2000a). This method makes use of the linear amplification of T7 polymerase, and has been shown to produce results that correlate well with original mRNA levels, in contrast to the exponential nature of PCR-based methods (Lockhart and Winzeler, 2000). Luo et al. (1999) report the amplification of mRNA from 500–1000 brain cells for microarray analysis. Alternative strategies include increasing the effective concentration of a subset of the RNA population under study, by reducing its complexity. This can be done by generating probes from differential display or cDNA-AFLP fragments (Trenkle et al., 1998). Other than the amplification of the labeled sample, there are also procedures that allow for kinetically more efficient hybridizations and the amplification of fluorescent signals (Stears et al., 2000).

Hybridization conditions Microarray hybridizations are performed for a period of 4–16 h in either formamide or SSC-based solutions (Eisen and Brown, 1999; Hegde et al., 2000). Volumes are kept to a minimum (ca. 0.033 µl per mm2 of coverslip) under a floating coverslip. To allow the processing of twelve microarrays at one time, Genomic Solutions, Michigan, (www.genomicsolutions.com), sells an automated hybridization and washing station. However, one area that sets current cDNA-microarrays apart from membrane-based arrays is that they are not generally reusable (Desprez et al., 1998; Baldwin et al., 1999), though a newer slide derivatization purports to make this possible (Beier and Hoheisel, 1999). Nanogen and Illumina are two companies that have experimented with alternative technologies that reduce hybridization times and allow for array reuse. Nanogen has produced a chip where the probes are independently, electronically activated (Sosnowski et al., 1997; Edman et al., 1997). Current models have only 99 probe sites, making them more suitable for small-scale expression profile studies. Even so, the electronics allow the labeled DNA interrogating the chip to be effectively concentrated, thus speeding hybridization times dramatically. Controlled reversal of the electric field can also achieve a defined washing stringency. With the independently controlled probes, this raises the prospect of having highly defined hybridization and washing conditions, for what is traditionally a passive process controlled by salt concentrations and temperature. The electronic process also facilitates the effective removal of all labeled cDNA, so that the chips can be re-used. Such a system could allow multiple hybridization experiments to be performed extremely quickly. The system is currently being beta tested as a gene expression system. By contrast, Illumina produces self-assembled bead arrays on bundles of individual, selectively etched optical fibers (Ferguson et al., 1996; Walt, 2000). Oligonucleotides can be either synthesized directly on to these beads or pre-synthesized molecules can be attached. The sequences on the beads are registered by fluorescent tagging of the beads or by decoding secondary DNA tags by hybridization. This system is reported to assay very small volumes and to be highly sensitive, allowing for potentially faster hybridizations. In addition, multiple hybridization cycles can be performed with the same array (Steemers et al., 2000). Other systems in development for improving hybridization kinetics and signal detection

90 have been recently reviewed (Steel et al., 2000; Blohm and Guiseppi-Elie, 2001). The parallel hybridization of two cDNA samples, a major factor in the efficiency of microarraying, is afforded by the dual detection of fluorescent moieties, typically Cy3 and Cy5. The fluorescent signals from cDNA bound to probes on a microarray are monitored by scanning systems that are mostly confocal in nature (http://cmgm.stanford.edu/pbrown/) (Schermer, 1999; Basarsky et al., 2000; Mittal, 2001). Some commercial models now allow the use of multiple absorption and emission wavelengths, facilitating third-channel normalization of hybridizations and more than two samples to be hybridized to each microarray. Data from these machines is typically captured as two 16bit TIFF images, one for each fluor. Commercial and public programs are available for extracting data from these image files, and these also include features for spot finding and the flagging of physical artifacts, such as dust (Bassett et al., 1999; Mittal, 2001). Data analysis Transcript profiling produces extremely large data sets, and even relatively simple studies can run into millions of data points (e.g. Arabidopsis Functional Consortium (AFGC; http://afgc.stanford.edu/) (Wisman and Ohlrogge, 2000). It is self-evident that such data sets cannot be organized on simple spreadsheets, but instead requires effective database resources for management and analysis (Bassett et al., 1999; Zhou et al., 2000). Molecular biology and computation are inexorably entwined in the science of transcript profiling (Figures 4 and 5). Normalization of extracted data is a requirement for all gene expression profiling. This is because, irrespective of the method, part of the processing of biological material and data extraction is inevitably done separately for each sample. A number of normalization approaches can be employed, including the use of ‘housekeeping genes’, that are hypothesized to show a constant expression between samples, and global approaches, where the total level of gene expression for a large portion of the total genes under study is considered to stay constant (Table 2). Where samples are similar, global normalization works well. For more divergent samples, normalization with a designated subset of genes may have an advantage. Because of the nature of cDNA-microarraying the most useful data output is the ratio of transcript levels between two samples. The first task on being faced

with these data is to decide what is ‘significant’. For plant microarray studies, ratios of 1.5 (McGonigle et al., 2000) to 3.32 (Aharoni et al., 2000) have been chosen, with provision for signals to be above a certain background threshold (Table 2). In some cases these values have been based on experiments where variability was assessed between duplicate RNA preparations (Wang et al., 2000b; Kawasaki et al., 2001), or in more advanced studies, between two duplicate experiments (Schenk et al., 2000; Kawasaki et al., 2001). Hughes et al. (2000) clearly illustrated the importance of such control experiments by establishing a gene-specific error model, conducting 63 control experiments compared to 300 compendium experiments in yeast. In this way, the effect of intrinsically fluctuating mRNA species can be considered. This is particularly important for GeneChip experiments where each array is interrogated with only a single labeled cDNA population (Harmer et al., 2000; Zhu and Wang, 2000). Alternative strategies have involved assessments from multiple repetitions of the same experiment and statistical analysis of variance (Aharoni et al., 2000; Seki et al., 2001). Correlative analysis An analysis of ratios provides a valuable means of looking at pair-wise comparisons (Arimura et al., 2000; Wang et al., 2000b; Seki et al., 2001). However, particularly for associating gene function, greater insights can be obtained when one draws correlations between the coordinated responses of genes (Figure 5; Fang et al., 2000) (Okamuro et al., 2000; Maleck et al., 2000; Schaffer et al., 2001; Kawasaki et al., 2001). Hughes et al. (2000) present a compelling case for functionally annotating yeast by creating a compendium of transcript profiles from 300 diverse mutations and chemical treatments. The principle is a simple one – ‘guilt by association’, genes whose expression profiles cluster together over a range of experimental treatments are more likely to be functionally linked than those that do not (Chu et al., 1998). As such, the functions of uncharacterized genes might be defined by their clustering with genes of known function over a diverse range of experiments (Chu et al., 1998) (Figure 5; Fang et al., 2000). Of the many clustering methods, the pairwise averagelinkage cluster analysis of Eisen et al. (1998), with its visually pleasing graphical representation, has been the most popular in plant microarray studies (Figure 5; Fang et al., 2000) (Maleck et al., 2000; Schaffer

Figure 5. Computational approaches to the deconvolution of transcript profiling data. A series of in-house and commercial programs illustrate the use of visualization as a means of data assimilation. 1. A scatter plot will identify genes whose expression differs between two samples. 2, 3. A Venn diagram allows one to identify common genes between two or more data sets (3). 4, 5. Multiple data sets can be compared by cluster analysis. Each horizontal represents one experiment, while each vertical line represents a single gene; the most similar expression patterns will be placed next to each other (5). 6. We have created a database linking correlations in expression data to sequence and function. 7. Conclusions obtained from the data analysis can be tested by methods such as quantitative RT-PCR.

91

92 et al., 2001; Kawasaki et al., 2001). Even so, its replacement of actual data with averaged data at each iteration, the use of similarity measures as opposed to Euclidean distance, and the creation of a hierarchical format where one does not naturally exist, are factors to consider when evaluating the results (Tavazoie et al., 1999). To alleviate some of these problems, data can be first partitioned into preliminary groupings using methods such as self-organizing maps (SOMs; Tamayo et al., 1999) and k-means clustering (Tavazoie et al., 1999). However, these methods require a decision on how many clusters you wish to define, although there are other methods where this is not necessary (Heyer et al., 1999). Needless to say, all these methods have prejudices, and the similarity metrics in themselves are not an indication of the significance of clusters. We have set up an expression database linking correlations in expression data to sequence and function (Volkmuth, unpublished data; Figure 5; Fang et al., 2000). Recently, a number of commercial software packages have become available that lead the user through expression data analysis with visually pleasing normalization, correlation and clustering features (Figure 5) (Richmond and Somerville, 2000; Goodman, 2001; Mittal, 2001). Downstream analysis Nobody handling large gene expression data sets would question the importance of computationally based organization and analysis (Figures 4 and 5). Ultimately, however, because of a number of technical issues, such as sample tracking, cross-hybridization and contamination, one should independently confirm important results obtained by expression profiling. The favorite method for plant microarray studies has been northern analysis, though as an automated system real-time RT-PCR is an attractive alternative (Table 2, Figure 5; Sambrook and Russell, 2001). Plant expression studies have shown good qualitative agreement between northern and microarray data (Ruan et al., 1998; Maleck et al., 2000; Wang et al., 2000b). The value of clustering genes within and between experiments can be enhanced immensely by linkage of the expression database to other information stockpiles (Ermolaeva et al., 1998; Marcotte et al., 1999; Burks, 1999). Such databases include the National Center for Biotechnology Information (NCBI; Wheeler et al., 2001) which provides a source of gene annotations, protein domain databases (Pfam; Bateman et al., 2000) and the Kyoto Encyclopedia

of Genes and Genomes (KEGG; Kanehisa and Goto, 2000), a database of metabolic pathways. In addition, transcript profiles are informative for genome annotation, as they facilitate characterization of genes and help identify promoter elements (Spellman et al., 1998; Harmer et al., 2000). At Ceres, Inc., we are able to link expression data to our full-length cDNA collection, providing correlations with superior annotation. Part of this collection of full-length cDNA sequences has been provided to TIGR in order to help with their genomic annotation of Arabidopsis. In another expression-based genome study, Cohen et al. (2000) showed that genes near each other on yeast chromosomes show correlated expression, suggesting a positional influence of chromosome structure on gene expression. As one can see, multifarious genomic characteristics can be integrated with transcriptional profiling. Likewise, this can be extended to include information from proteomics (e.g. protein profiles and protein interaction analysis) and metabolomics, indeed all genomic platforms, as part of a well-organized data management system. Phenotypic data are perhaps best integrated with expression data by cross-reference to plant insertional mutants, noting also when and where mRNAs are expressed (Azpiroz-Leehan and Feldmann, 1997; Parinov and Sundaresan, 2000; Hughes et al., 2000; Petersen et al., 2000). The net result of this data organization should be a multi-dimensional representation of a plant, both bioinformatically (Vidal, 2001) and visually (Streicher et al., 2000). This is surely the aspiration of plant genomics, a truly holistic view of biology where disciplines such as genetics, anatomy and biochemistry merge in one vision and model of a plant. Such a plan will also provide for more insightful reductionist approaches. In an effort to reach that goal, transcript profiling has now joined nucleotide sequencing as a truly global means of plant analysis.

Acknowledgements We wish to thank our colleagues at Cenes, Inc., and Keygene N.V. The authors are grateful to Dick Flanell for comments on the text and to Amy Hunt for help in preparing the manuscript.

References Adams, M.D., Kelley, J.M., Gocayne, J.D., Dubnick, M., Polymeropoulos, M.H., Xiao, H., Merril, C.R., Wu, A., Olde, B.,

93 Moreno, R.F., Kerlavage, A.R., McCombie, W.R. and Venter, J.C. 1991. Complementary DNA sequencing: expressed sequence tags and human genome project. Science 252: 1651– 1656. Aharoni, A., Keizer, L.C.P., Bouwmeester, H.J., Sun, Z., AlvarezHuerta, M., Verhoeven, H.A., Blaas, J., van Houwelingen, A.M.M.L., de Vos, R.C.H., van der Voet, H., Jansen, R.C., Guis, M., Mol, J., Davis, R.W., Schena, M., van Tunen, A.J. and O’Connell, A.P. 2000. Identification of the SAAT gene involved in strawberry flavor biogenesis by use of DNA microarrays. Plant Cell 12: 647–661. Alwine, J.C., Kemp, D.J. and Stark, G.R. 1977. Method for detection of specific RNAs in agarose gels by transfer to diazobenzyloxymethyl-paper and hybridization with DNA probes. Proc. Natl. Acad. Sci. USA 74: 5350–5354. Arabidopsis Genome Initiative. 2000. Analysis of the genome sequence of the flowering plant Arabidopsis thaliana. Nature 408: 796–815. Arimura, G.I., Tashiro, K., Kuhara, S., Nishioka, T., Ozawa, R. and Takabayashi, J. 2000. Gene responses in bean leaves induced by herbivory and by herbivore-induced volatiles. Biochem. Biophys. Res. Commun. 277: 305–310. Audic, S. and Claverie, J.M. 1997. The significance of digital gene expression profiles. Genome Res. 7: 986–995. Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A., Struhl, K., Albright, L.M., Coen, D.M. and Varki, A. 2001. Current Protocols in Molecular Biology. John Wiley, New York. Azpiroz-Leehan, R. and Feldmann, K.A. 1997. T-DNA insertion mutagenesis in Arabidopsis: going back and forth. Trends Genet. 13: 152–156. Bachem, C.W.B., van der Hoeven, R.S., de Bruijn, S.M., Vreugdenhil, D., Zabeau, M. and Visser, R.G.F. 1996. Visualization of differential gene expression using a novel method of RNA fingerprinting based on AFLP: Analysis of gene expression during potato tuber development. Plant J. 9: 745–753. Bachem, C.W.B., Oomen, R.J.F.J. and Visser, R.G.F. 1998. Transcript imaging with cDNA-AFLP: a step-by-step protocol. Plant Mol. Biol. Rep. 16: 157–173. Baldwin, D., Crane, V. and Rice, D. 1999. A comparison of gel-based, nylon filter and microarray techniques to detect differential RNA expression in plants. Curr. Opin. Plant Biol. 2: 96–103. Basarsky, T., Verdnik, D., Zhai, J.Y. and Wellis, D. 2000. Overview of a microarray scanner: design essentials for an integrated acquisition and analysis platform. In: M. Schena (Ed.) Microarray Biochip Technology, Eaton Publishing, Natick, MA, pp. 265–284. Bassett, D.E. Jr., Eisen, M.B. and Boguski, M.S. 1999. Gene expression informatics – it’s all in your mine. Nature Genet. Suppl. 21: 51–55. Bateman, A., Birney, E., Durbin, R., Eddy, S.R., Howe, K.L. and Sonnhammer, E.L.L. 2000. The Pfam protein families database. Nucl. Acids Res. 28: 263–266. Beier, M. and Hoheisel, J.D. 1999. Versatile derivatisation of solid support media for covalent bonding on DNA-microchips. Nucl. Acids Res. 27: 1970–1977. Bertioli, D.J., Schlichter, U.H., Adams, M.J., Burrows, P.R., Steinbiss, H.H. and Antoniw, J.F. 1995. An analysis of differential display shows a strong bias towards high copy number mRNAs. Nucl. Acids Res. 23: 4520–4523. Bevan, M., Mayer, K., White, O., Eisen, J.A., Preuss, D., Bureau, T., Salzberg, S.L. and Mewes, H.W. 2001. Sequence and analysis of the Arabidopsis genome. Curr. Opin. Plant Biol. 4: 105–110.

Bishop, J.O., Morton, J.G., Robash, M. and Richardson, M. 1974. Three abundance classes in HeLa cell messenger RNA. Nature 463: 199–204. Blanc, G., Barakat, A., Guyot, R., Cooke, R. and Delseny, M. 2000. Extensive duplication and reshuffling in the Arabidopsis genome. Plant Cell 12: 1093–1101. Blanchard, A.P., Kaiser, R.J. and Hood, L.E. 1996. High-density oligonucleotide arrays. Biosensors Bioelectronics 11: 687–690. Blohm, D.H. and Guiseppi-Elie, A. 2001. New developments in microarray technology. Curr. Opin. Biotechnol. 12: 41–47. Bohnert, H.J., Ayoubi, P., Borchert, C., Bressan, R.A., Burnap, R.L., Cushman, J.C., Cushman, M.A., Deyholos, M., Fischer, R., Galbraith, D.W., Hasegawa, P.M., Jenks, M., Kawasaki, S., Koiwa, H., Kore-eda, S., Lee, B.H., Michalowski, C.B., Misawa, E., Nomura, M., Ozturk, N., Postier, B., Prade, R., Song, C.P., Tanaka, Y., Wang, H. and Zhu, J.K. 2001. A genomics approach towards salt stress tolerance. Plant Physiol. Biochem. 39: 1–17. Bonaldo, M.F., Lennon, G. and Soares, M.B. 1996. Normalization and subtraction: two approaches to facilitate gene discovery. Genome Res. 6: 791–806. Bosch, I., Melichar, H. and Pardee, A.B. 2000. Identification of differentially expressed genes from limited amounts of RNA. Nucl. Acids Res. 28: e27. Brenner, S., Johnson, M., Bridgham, J., Golda, G., Lloyd, D.H., Johnson, D., Luo, S., McCurdy, S., Foy, M., Ewan, M., Roth, R., George, D., Eletr, S., Albrecht, G., Vermaas, E., Williams, S.R., Moon, K., Burcham, T., Pallas, M., DuBridge, R.B., Kirchner, J., Fearon, K., Mao, J.I. and Corcoran, K. 2000a. Gene expression analysis by massively parallel signature sequencing (MPSS) on microbead arrays. Nature Biotechnol. 18: 630–634. Brenner, S., Williams, S.R., Vermass, E.H., Storck, T., Moon, K., McCollum, C., Mao, J.I., Luo, S., Kirchner, J.J., Eletr, S., DuBridge, R.B., Burcham, T. and Albrecht, G. 2000b. In vitro cloning of complex mixtures of DNA on microbeads: physical separation of differentially expressed cDNAs. Proc. Natl. Acad. Sci. USA 97: 1665–1670. Breyne, P. and Zabeau, M. 2001. Genome-wide expression analysis of plant cell cycle modulated genes. Curr. Opin. Plant Biol. 4: 136–142. Bruce, W., Folkerts, O., Garnaat, C., Crasta, O., Roth, B. and Bowen, B. 2000. Expression profiling of the maize flavonoid pathway genes controlled by estradiol-inducible transcription factors CRC and P. Plant Cell 12: 65–79. Buntjer, J., de Ruiter, M., Vallins, B., Minarik, M., Mahtani, M., van Schaik, R., Vos, P. and van Eijk, M. 1999. High throughput analysis of AFLP fragments on the MegaBACE capillary sequencer. In: Plant & Animal Genome IX, January 13–17, 2001, San Diego, CA. Burks, C. 1999. Molecular biology database list. Nucl. Acids Res. 27: 1–9. Celis, J.E., Kruhoffer, M., Gromova, I., Frederiksen, C., Ostergaard, M., Thykjaer, T., Gromov, P., Yu, J., Palsdottir, H., Magnusson, N. and Orntoft, T.F. 2000. Gene expression profiling: monitoring transcription and translation products using DNA microarrays and proteomics. FEBS Lett. 480: 2–16. Chen, J.J., Rowley, J.D. and Wang, S.M. 2000. Generation of longer cDNA fragments from serial analysis of gene expression tags for gene identification. Proc. Natl. Acad. Sci. USA 97: 349–353. Cho, Y. and Walbot, V. 2001. Computational methods for gene annotation: the Arabidopsis genome. Curr. Opin. Biotechnol. 12: 126–130. Cho, Y.J., Meade, J.D., Walde, J.C., Chen, X., Gui, Z. and Liang, P. 2001. Multicolor fluorescent differential display. Biotechniques 30: 562–572.

94 Chu, S., DeRisi, J., Eisen, M., Mulholland, J., Botstein, D., Brown, P.O. and Herskowitz, I. 1998. The transcriptional program of sporulation in budding yeast. Science 282: 699–705. Cohen, B.A., Mitra, R.D., Hughes, J.D. and Church, G.M. 2000. A computational analysis of whole-genome expression data reveals chromosomal domains of gene expression. Nature Genet. 26: 183–186. Datson, N.A., van der Perk-de Jong, J., van den Berg, M.P., de Kloet, E.R. and Vreugdenhil, E. 1999. MicroSAGE: a modified procedure for serial analysis of gene expression in limited amounts of tissue. Nucl. Acids Res. 27: 1300–1307. DeRisi, J.L., Iyer, V.R. and Brown, P.O. 1997. Exploring the metabolic and genetic control of gene expression on a genomic scale. Science 278: 680–686. Desprez, T., Amselem, J., Caboche, M. and Höfte, H. 1998. Differential gene expression in Arabidopsis monitored using cDNA arrays. Plant J. 14: 643–652. Diatchenko, L., Lau, Y.F., Campbell, A.P., Chenchik, A., Moqadam, F., Huang, B., Lukyanov, S., Lukyanov, K., Gurskaya N., Sverdlov, E.D. and Siebert, P.D. 1996. Suppression subtractive hybridization: a method for generating differentially regulated or tissue-specific cDNA probes and libraries. Proc. Natl. Acad. Sci. USA 12: 6025–6030. Diehn, M., Eisen, M.B., Botstein, D. and Brown, P.O. 2000. Largescale identification of secreted and membrane-associated gene products using DNA microarrays. Nature Genet. 25: 58–62. Duggan, D.J., Bittner, M., Chen, Y., Meltzer, P. and Trent, J.M. 1999. Expression profiling using cDNA microarrays. Nature Genet. Suppl. 21: 10–14. Durrant, W.E., Rowland, O., Piedras, P., Hammond-Kosack, K.E. and Jones, J.D.G. 2000. cDNA-AFLP reveals a striking overlap in race-specific resistance and wound response gene expression profiles. Plant Cell 12: 963–977. Edman, C.F., Raymond, D.E., Wu, D.J., Tu, E., Sosnowski, R.G., Butler, W.F., Nerenberg, M. and Heller, M.J. 1997. Electric field directed nucleic acid hybridization on microchips. Nucl. Acids Res. 25: 4907–4914. Eggers, M. 2000. High-throughput microarray technology. Innov. Pharmaceut. Technol. 6: 36–44. Eisen, M.B. and Brown, P.O. 1999. DNA arrays for analysis of gene expression. Meth. Enzymol. 303: 179–205. Eisen, M.B., Spellman, P.T., Brown, P.O. and Botstein, D. 1998. Cluster analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. USA 95: 14863–14868. Emmert-Buck, M.R., Bonner, R.F., Smith, P.D., Chuaqui, R.F., Zhuang, Z., Goldstein, S.R., Weiss, R.A. and Liotta, L.A. 1996. Laser capture microdissection. Science 5289: 998–1001. Ermolaeva, O., Rastogi, M., Pruitt, K.D., Schuler, G.D., Bittner, M.L., Chen, Y., Simon, R., Meltzer, P., Trent, J.M. and Boguski, M.S. 1998. Data management and analysis for gene expression arrays. Nature Genet. 20: 19–23. Ewing, R.M., Kahla, A.B., Poirot, O., Lopez, F., Audic, S. and Claverie, J.M. 1999. Large-scale statistical analyses of rice ESTs reveal correlated patterns of gene expression. Genome Res. 9: 950–959. Fang, Y., Xing, W., Santo, G.E., Volkmuth, W., Shapiro, A., Miyamoto, S., Salazar, A. and Donson, J. 2000. Identification of target genes through microarray analysis of transcripts from diverse experiments. In: Arabidopsis Genomics, 7–10 December 2000, Cold Spring Harbor, NY. Ferea, T.L. and Brown, P.O. 1999. Observing the living genome. Curr. Opin. Genet. Dev. 9: 715–722.

Ferguson, J.A., Boles, T.C., Adams, C.P. and Walt, D.R. 1996. A fiber-optic DNA biosensor microarray for the analysis of gene expression. Nature Biotechnol. 14: 1681–1684. Gillespie, D. and Spiegelman, S. 1965. A quantitative assay for DNA-RNA hybrids with DNA immobilized on a membrane. J. Mol. Biol. 3: 829–842. Girke, T., Todd, J., Ruuska, S., White, J., Benning, C. and Ohlrogge, J. 2000. Microarray analysis of developing Arabidopsis seeds. Plant Physiol. 124: 1570–1581. Goldberg, R.B. 2001. From cot curves to genomics. How gene cloning established new concepts in plant biology. Plant Physiol. 125: 4–8. Goldberg, R.B., Hoschek, G. and Kamalay, J.C. 1978. Sequence complexity of nuclear and polysomal RNA in leaves of the tobacco plant. Cell 14: 123–131. Goodman, N. 2001. Hungry for gene-expression analysis solutions? The microarray sofware bill of fare now offers a delectable assortment of products for your data analysis pleasure. Genome Technol. 03.01: 40–50. Guigó, R., Agarwal, P., Abril, J.F., Burset, M. and Fickett, J.W. 2000. An assessment of gene prediction accuracy in large DNA sequences. Genome Res. 10: 1631–1642. Gygi, S.P., Rochon, Y., Franza, B.R. and Aebersold, R. 1999. Correlation between protein and mRNA abundance in yeast. Mol. Cell. Biol. 19: 1720–1730. Haag, E. and Raman, V. 1994. Effects of primer choice and source of Taq DNA polymerase on the banding patterns of differential display RT-PCR. Biotechniques 2: 226–228. Habu, Y., Fukada-Tanaka, S., Hisatomi, Y. and Iida, S. 1997. Amplified restriction fragment length polymorphism-based mRNA fingerprinting using a single restriction enzyme that recognizes a 4-bp sequence. Biochem. Biophys. Res. Commun. 234: 516– 521. Harmer, S.L., Hogenesch, J.B., Straume, M., Chang, H.S., Han, B., Zhu, T., Wang, X., Kreps, J.A. and Kay, S.A. 2000. Orchestrated transcription of key pathways in Arabidopsis by the circadian clock. Science 290: 2110–2113. Hegde, P., Qi, R., Abernathy, K., Gay, C., Dharap, S., Gaspard, R., Hughes, J.E., Snesrud, E., Lee, N. and Quackenbush, J. 2000. A concise guide to cDNA microarray analysis. BioTechniques 29: 548–562. Heyer, L.J., Kruglyak, S. and Yooseph, S. 1999. Exploring expression data: identification and analysis of coexpressed genes. Genome Res. 9: 1109–1115. Hihara, Y., Kamei, A., Kanehisa, M., Kaplan, A. and Ikeuchi, M. 2001. DNA microarray analysis of cyanobacterial gene expression during acclimation to high light. Plant Cell 13: 1–15. Hubank, M. and Schatz, D.G. 1994. Identifying differences in mRNA expression by representational difference analysis of cDNA. Nucl. Acids. Res. 25: 5640–5648. Hughes, T.R., Marton, M.J., Jones, A.R, Roberts., C.J., Stoughton, R., Armour, C.D, Bennett, H.A., Coffey, E., Dai, H., He, Y.D., Kidd, M.J., King, A.M., Meyer, M.R., Slade, D., Lum, P.Y., Stepaniants, S.B., Shoemaker, D.D., Gachotte, D., Chakraburtty, K., Simon, J., Bard, M. and Friend, S.H. 2000. Functional discovery via a compendium of expression profiles. Cell 102: 109–126. Hughes, T.R., Mao, M., Jones, A.R., Burchard, J., Marton, M.J., Shannon, K.W., Lefkowitz, S.M., Ziman, M., Schelter, J.M., Meyer, M.R., Kobayashi, S., Davis, C., Dai, H., He, Y.D., Stephaniants, S.B., Cavet, G., Walker, W.L., West, A., Coffey, E., Shoemaker, D.D., Stoughton, R., Blanchard, A.P., Friend, S.H. and Linsley, P.S. 2001. Expression profiling using microar-

95 rays fabricated by an ink-jet oligonucleotide synthesizer. Nature Biotechnol. 19: 342–347. Ikonomov, O.C. and Jacob, M.H. 1996. Differential display protocol with selected primers that preferentially isolates mRNAs of moderate- to low-abundance in a microscopic system. Biotechniques 6: 1030–1042. Ito, T., Kito, K., Adati, N., Mitsui, Y., Hagiwara, H. and Sakaki, Y. 1994. Fluorescent differential display: arbitrarily primed RTPCR fingerprinting on an automated DNA sequencer. FEBS Lett. 2: 231–236. Jackson, A.O. and Larkins, B.A. 1976. Influence of ionic strength, pH, and chelation of divalent metals on isolation of polyribosomes from tobacco leaves. Plant Physiol. 57: 5–10. Jones, C.S., Davies, H.V. and Taylor, M.A. 2000. Profiling of changes in gene expression during raspberry (Rubus idaeus) fruit ripening by application of RNA fingerprinting techniques. Planta 211: 708–714. Kafatos, F.C., Jones, C.W. and Efstratiadis, A. 1979. Determination of nucleic acid sequence homologies and relative concentrations by a dot hybridization procedure. Nucl. Acids. Res. 6: 1541– 1552. Kane, M.D., Jatkoe, T.A., Stumpf, C.R., Lu, J., Thomas, J.D. and Madore, S.J. 2000. Assessment of the sensitivity and specificity of oligonucleotide (50mer) microarrays. Nucl. Acids Res. 28: 4552–4557. Kanehisa, M. and Goto, S. 2000. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucl. Acids Res. 28: 27–30. Kawasaki, S., Borchert, C., Deyholos, M., Wang, H., Brazille, S., Kawai, K., Galbraith, D. and Bohnert, H.J. 2001. Gene expression profiles during the initial phase of salt stress in rice. Plant Cell 13: 889–906. Kreps, J.A. and Kay, S.A. 1997. Coordination of plant metabolism and development by the circadian clock. Plant Cell 9: 1235– 1244. Kuhn, E. 2001. From library screening to microarray technology: strategies to determine gene expression profiles and to identify differentially regulated genes in plants. Ann. Bot. 87: 139–155. Lee, M.L.T., Kuo, F.C., Whitmore, G.A. and Sklar, J. 2000. Importance of replication in microarray gene expression studies: statistical methods and evidence from repetitive cDNA hybridizations. Proc. Natl. Acad. Sci. USA 97: 9834–9839. Liang, P. and Pardee, A.B. 1992. Differential display of eukaryotic messenger RNA by means of the polymerase chain reaction. Science 257: 967–971. Linskens, M.H., Feng, J., Andrews, W.H., Enlow, B.E., Saati, S.M., Tonkin, L.A., Funk, W.D. and Villeponteau, B. 1995. Cataloging altered gene expression in young and senescent cells using enhanced differential display. Nucl. Acids Res. 16: 3244–3251. Lipshutz, R.J., Fodor, S.P., Gingeras, T.R. and Lockhart, D.J. 1999. High density synthetic oligonucleotide arrays. Nature Genet. Suppl. 21: 20–24. Liu, A.Y., True, L.D., LaTray, L., Ellis, W.J., Vessella, R.L., Lange, P.H., Higano, C.S., Hood, L. and van den Engh, G. 1999. Analysis and sorting of prostate cancer cell types by flow cytometry. Prostate 40: 192–199. Lockhart, D.J., Dong, H., Byrne, M.C., Follettie, M.T., Gallo, M.V., Chee, M.S., Mittmann, M., Wang, C., Kobayashi, M., Horton, H. and Brown, E.L. 1996. Expression monitoring by hybridization to high-density oligonucleotide arrays. Nature Biotechnol. 13: 1675–1680. Lockhart, D.J. and Winzeler, E.A. 2000. Genomics, gene expression and DNA arrays. Nature 405: 827–836. Luo, L., Salunga, R.C., Guo, H., Bittner, A., Joy, K.C., Galindo, J.E., Xiao, H., Rogers, K.E., Wan, J.S., Jackson, M.R. and Er-

lander, M.G. 1999. Gene expression profiles of laser captured adjacent neuronal subtypes. Nature Med. 5: 117–122. Macas, J., Nouzová, M. and Galbraith, D.W. 1998. Adapting the Biomek 2000 laboratory automation workstation for printing DNA microarrays. BioTechniques 25: 106–110. Maleck, K., Levine, A., Eulgem, T., Morgan, A., Schmid, J., Lawton, K.A., Dangl, J.L. and Dietrich, R.A. 2000. The transcriptome of Arabidopsis thaliana during systemic acquired resistance. Nature Genet. 26: 403–410. Marcotte, E.M., Pellegrini, M., Thompson, M.J., Yeates, T.O. and Eisenberg, D. 1999. A combined algorithm for genome-wide prediction of protein function. Nature 402: 83–86. Martin, K.J. and Pardee, A.B. 1999. Principles of differential display. Meth. Enzymol. 303: 234–297. Matsumura, H., Nirasawa, S. and Terauchi, R. 1999. Transcript profiling in rice (Oryza sativa L.) seedlings using serial analysis of gene expression (SAGE). Plant J. 20: 719–726. Matz, M.V. and Lukyanov, S.A. 1998. Different strategies of differential display: areas of application. Nucl. Acids Res. 26: 5537–5543. McCabe, P.F., Valentine, T.A., Forsberg, L.S. and Pennell, R.I. 1997. Soluble signals from cells identified at the cell wall establish a developmental pathway in carrot. Plant Cell 9: 2225–2241. McClelland, M., Mathieu-Daude, F. and Welsh, J. 1995. RNA fingerprinting and differential display using arbitrarily primed PCR. Trends Genet. 6: 242–246. McGonigle, B., Keeler, S.J., Lau, S-M.C., Koeppe, M.K. and O’Keefe, D.P. 2000. A genomics approach to the comprehensive analysis of the glutathione S-transferase gene family in soybean and maize. Plant Physiol. 124: 1105–1120. Mekhedov, S., Martínez de Ilárduya, O. and Ohlrogge, J. 2000. Toward a functional catalog of the plant genome. A survey of genes for lipid biosynthesis. Plant Physiol. 122: 389–401. Meldrum, D. 2000. Automation for genomics, Part Two: Sequencers, microarrays, and future trends. Genome Res. 10: 1288–1303. Mittal, V. 2001. DNA array technology. In: J. Sambrook and D. Russell (Eds.) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Plainview, New York, pp. A10.1–A10.19. Mysore, K.S., Tuori, R., D’Ascenzo, M., Debbie, P., Gu, Y., Crasta, O., Folkerts, O. and Martin, G.B. 2001. Expression profiling of genes that are induced or suppressed during an incompatible plant-pathogen interaction in tomato. Plant & Animal Genome IX Conference, January 13–17, 2001, San Diego, CA. Nelson, S.F. and Denny, C.T. 1999. Representational differences analysis and microarray hybridization for efficient cloning and screening of differentially expressed genes. In: M. Schena (Ed.) DNA Microarrays: A Practical Approach, Oxford University Press, New York, pp. 43–59. Ohlrogge, J. and Benning, C. 2000. Unraveling plant metabolism by EST analysis. Curr. Opin. Plant Biol. 3: 224–228. Okamoto, T., Suzuki, T. and Yamamoto, N. 2000. Microarray fabrication with covalent attachment of DNA using bubble jet technology. Nature Biotechnol. 18: 438–441. Okamuro, J., Jofuku, D., Pennell, R., Chen, Z., Dang, V-D., Donson, J., Fang, Y., Volkmuh, W. and Flavell, R.B. 2000. A comprehensive analysis of gene expression from ovule to seed and from flower to fruit. In: Arabidopsis Genomics, 7–10 December 2000, Cold Spring Harbor, NY. Parinov, S. and Sundaresan, V. 2000. Functional genomics in Arabidopsis: large-scale insertional mutagenesis complements the genome sequencing project. Curr. Opin. Biotechnol. 11: 157–161.

96 Pavy, N., Rombauts, S., Dehais, P., Mathe, C., Ramana, DV., Leroy, P. and Rouze, P. 1999. Evaluation of gene prediction software using a genomic data set: application to Arabidopsis thaliana sequences. Bioinformatics 11: 887–899. Peters, D.G., Kassam, A.B., Yonas, H., O’Hare, E.H., Ferrell, R.E. and Brufsky, A.M. 1999. Comprehensive transcript analysis in small quantities of mRNA by SAGE-Lite. Nucl. Acids Res. 27: e39. Petersen, M., Brodersen, P., Naested, H., Andreasson, E., Lindhart, U., Johansen, B., Nielsen, H.B., Lacy, M., Austin, M.J., Parker, J.E., Sharma, S.B., Klessig, D.F., Martienssen, R., Mattsson, O., Jensen, A.B. and Mundy, J. 2000. Arabidopsis MAP kinase 4 negatively regulates systemic acquired resistance. Cell 103: 1111–1120. Prashar, Y. and Weissman, S.M. 1996. Analysis of differential gene expression by display of 3 end restriction fragments of cDNAs. Proc. Natl. Acad. Sci. USA 93: 659–663. Qin, L., Prins, P., Jones, J.T., Popeijus, H., Smant, G., Bakker, J. and Helder, J. 2001. GenEST, a powerful bidirectional link between cDNA sequence data and gene expression profiles generated by cDNA-AFLP. Nucl. Acids Res. 29: 1616–1622. Renner, C., Trümper, L., Pfitzenmeier, J.P., Loftin, U., Gerlach, K., Stehle, I., Wadle, A. and Pfreundschuh, M. 1998. Differential mRNA Display at the Single-Cell Level. Biotechniques 24: 720– 724. Reymond, P., Weber, H., Damond, M. and Farmer, E.E. 2000. Differential gene expression in response to mechanical wounding and insect feeding in Arabidopsis. Plant Cell 12: 707–719. Richmond, T. and Somerville, S. 2000. Chasing the dream: plant EST microarrays. Curr. Opin. Plant Biol. 3: 108–116. Ruan, Y., Gilmore, J. and Conner, T. 1998. Towards Arabidopsis genome analysis: monitoring expression profiles of 1400 genes using cDNA microarrays. Plant J. 15: 821–833. Sablowski, R.W.M. and Meyerowitz, E.M. 1998.: A homolog of NO APICAL MERISTEM is an immediate target of the floral homeotic genes APETALA3/PISTILLATA. Cell 92: 93–103. Sagerstrom, C.G., Sun, B.I. and Sive, H.L. 1997. Subtractive cloning: past, present, and future. Annu. Rev. Biochem. 66: 751–783. Salunga, R.C., Guo, H., Luo, L., Bittner, A., Joy, K.D., Chambers, J.R., Wan, J.S., Jackson, M.R. and Erlander, M.G. 1999. Gene expression analysis via cDNA microarrays of laser capture microdissected cells from fixed tissue. In: M. Schena (Ed.) DNA Microarrays: A Practical Approach, Oxford University Press, New York, pp. 121–137. Sambrook, J. and Russell, D.W. 2001. Molecular Cloning: A Laboratory Manual, 3rd ed.. Cold Spring Harbor Laboratory Press, Plainview, NY. Sargent, T.D. 1987. Isolation of differentially expressed genes. Meth. Enzymol. 152: 423–432. Sato, S., Nakamura, Y., Kaneko, T., Asamizu, E. and Tabata, S. 1999. Complete structure of the chloroplast genome of Arabidopsis thaliana. DNA Res. 5: 283–290. Schaffer, R., Landgraf, J., Accerbi, M., Simon, V., Larson, M. and Wisman, E. 2001. Microarray analysis of diurnal and circadianregulated genes in Arabidopsis. Plant Cell 13: 113–123. Schena, M. 1999. DNA Microarrays: A Practical Approach. Oxford University Press, Oxford. Schena, M. 2000. Microarray Biochip Technology. Eaton Publishing, Natick, MA. Schena, M., Shalon, D., Davis, R.W. and Brown, P.O. 1995. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270: 467–470.

Schena, M., Shalon, D., Heller, R., Chai, A., Brown, P.O. and Davis, R.W. 1996. Parallel human genome analysis: microarray-based expression monitoring of 1000 genes. Proc. Natl. Acad. Sci. USA 20: 10614–10619. Schenk, P.M., Kazan, K., Wilson, I., Anderson, J.P., Richmond, T., Somerville, S.C. and Manners, J.M. 2000. Coordinated plant defense responses in Arabidopsis revealed by microarray analysis. Proc. Natl. Acad. Sci. USA 97: 11655–11660. Schermer, M.J. 1999. Confocal scanning microscopy in microarray detection. In: M. Schena (Ed.) DNA Microarrays: A Practical Approach, Oxford University Press, New York, pp. 17–42. Schupp, J.M., Dick, A.B., Zinnamon, K.N. and Keim, P. 2001. Serial analysis of gene expression applied to soybean. Plant & Animal Genome IX Conference, January 13–17, 2001, San Diego, CA. Seki, M., Narusaka, M., Abe, H., Kasuga, M., YamaguchiShinozaki, K., Carinci, P., Hayashizaki, Y. and Schinozaki, K. 2001. Monitoring the expression pattern of 1300 Arabidopsis genes under drought and cold stresses by using a full-length cDNA microarray. Plant Cell 13: 61–72. Sheen, J., Hwang, S., Niwa, Y., Kobayashi, H. and Galbraith, D.W. 1995. Green-fluorescent protein as a new vital marker in plant cells. Plant J. 5: 777–784. Shimkets, R.A., Lowe, D.G., Tai, J.T.N, Sehl, P., Jin, H., Yang, R., Predki, P.F., Rothberg, B.E.G., Murtha, M.T., Roth, M.E., Shenoy, S.G., Windemuth, A., Simpson, J.W., Simons, J.F., Daley, M.P., Gold, S.A., McKenna, M.P., Hillan, K., Went, G.T. and Rothberg, J.M. 1999. Gene expression analysis by transcript profiling coupled to a gene database query. Nature Biotechnol. 17: 798–803. Shoemaker, D.D., Schadt, E.E., Armour, C.D., He, Y.D., GarrettEngele, P., McDonagh, P.D., Loerch, P.M., Leonardson, A., Lum, P.Y., Cavet, G., Wu, L.F., Altschuler, S.J., Edwards, S., King, J., Tsang, J.S., Schimmack, G., Schelter, J.M., Koch, J., Ziman, M., Marton, M.J., Li, B., Cundiff, P., Ward, T., Castle, J., Krowlewski, M., Meyer, M.R., Mao, M., Burchard, J., Kidd, M.J., Dai, H., Phillips, J.W., Linsley, P.S., Stoughton, R., Scherer, S. and Boguski, M.S. 2001. Experimental annotation of the human genome using microarray technology. Nature 409: 922–927. Singh-Gasson, S., Green, R.D., Yue, Y., Nelson, C., Blattner, F., Sussman, M.R. and Cerrina, F. 1999. Maskless fabrication of light-directed oligonucleotide microarrays using a digital micromirror array. Nature Biotechnol. 10: 974–978. Smith, R.D. 2000. Probing proteomes: seeing the whole picture? Nature Biotechnol. 18: 1041–1042. Sompayrac, L., Jane, S., Burn, T.C., Tenen, D.G. and Danna, K.J. 1995. Overcoming limitations of the mRNA differential display technique. Nucl. Acids Res. 23: 4738–4739. Sosnowski, R.G., Tu, E., Butler, W.F., O’Connell, J.P. and Heller, M.J. 1997. Rapid determination of single base mismatch mutation s in DNA hybrids by direct electric field control. Proc. Natl. Acad. Sci. USA 94: 1119–1123. Southern, E.M. 1975. Detection of specific sequences among DNA fragments separated by gel electrophoresis. J. Mol. Biol. 3: 503– 517. Spellman, P.T., Sherlock, G., Zhang, M.Q., Iyer, V.R., Anders, K., Eisen, M.B., Brown, P.O., Botstein, D. and Futcher, B. 1998. Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol. Biol. Cell 9: 3273–3297. Stears, R.L., Getts, R.C. and Gullans, S.R. 2000. A novel, sensitive detection system for high-density microarrays using dendrimer technology. Physiol. Genomics. 3: 93–99.

97 Steel, A., Torres, M., Hartwell, J., Yu, Y.Y., Ting, N., Hoke, G. and Yang, H. 2000. The Flow-Thru ChipTM : a three-dimensional biochip platform. In: M. Schena (Ed.) Microarray Biochip Technology, Eaton Publishing, Natick, MA, pp. 87–118. Steemers, F.J., Ferguson, J.A and Walt, D.R. 2000. Screening unlabeled DNA targets with randomly ordered fiber-optic gene arrays. Nature Biotechnol. 18: 91–94. Streicher, J., Donat, M.A., Strauss, B., Spörles, R., Schughart, K. and Müller, G.B. 2000. Computer-based three-dimensional visualization of developmental gene expression. Nature Genet. 25: 147–152. Sun, Y., Hegamyer, G. and Colburn, N.H. 1994. Molecular cloning of five messenger RNAs differentially expressed in preneoplastic or neoplastic JB6 mouse epidermal cells: one is homologous to human tissue inhibitor of metalloproteinases-3. Cancer Res. 5: 1139–1144. Sutcliffe, J.G., Foye, P.E., Erlander, M.G., Hilbush, B.S., Bodzin, L.J., Durham, J.T. and Hasel, K.W. 2000. TOGA: an automated parsing technology for analyzing expression of nearly all genes. Proc. Natl. Acad. Sci. USA 97: 1976–1981. Tamayo, P., Slonim, D., Mesirov, J., Zhu, Q., Kitareewan, S., Dmitrovsky, E., Lander, E.S. and Golub, T.R. 1999. Interpreting patterns of gene expression with self-organizing maps: methods and application to hemoatopoietic differentiation. Proc. Natl. Acad. Sci. USA 96: 2907–2912. Tavazoie, S., Hughes, J.D., Campbell, M.J., Cho, R.J. and Church, G.M. 1999. Systematic determination of genetic network architecture. Nature Genet. 3: 281–285. Ten Bosch, J., Seidel, C., Batra, S., Lam, H., Tuason, N., Saljoughi, S. and Saul, R. 2000. Validation of sequence-optimized 70-base oligonucleotides for use on DNA microarrays. TIGR Genome Sequencing and Analysis Conference, 12–15 September, Miami Beach, FL. Trenkle, T., Welsh, J., Jung, B., Mathieu-Daude, F. and McClelland, M. 1998. Non-stoichiometric reduced complexity probes for cDNA arrays. Nucl. Acids Res. 26: 3883–3891. Unseld, M., Marienfeld, J.R., Brandt, P. and Brennicke, A. 1997. The mitochondrial genome of Arabidopsis thaliana contains 57 genes in 366,924 nucleotides. Nature Genet. 1: 57–61. van den Berg, A., van der Leij, J. and Poppema, S. 1999. Serial analysis of gene expression: rapid RT-PCR analysis of unknown SAGE tags. Nucl. Acids Res. 27: e17. Velculescu, V.E., Zhang, L., Vogelstein, B. and Kinzler, K.W. 1995. Serial analysis of gene expression. Science 5235: 484–487. Vidal, M. 2001. A biological atlas of functional maps. Cell 104: 333–339. Visioli, G., Maestri, E. and Marmiroli, N. 1997. Differential display-mediated isolation of a genomic sequence for a putative mitochondrial LMW HSP specifically expressed in condition of induced thermotolerance in Arabidopsis thaliana (L.) Heynh. Plant Mol. Biol. 34: 517–527. Vision, T.J., Brown, D.G. and Tanksley, S.D. 2000. The origins of genomic duplications in Arabidopsis. Science 5499: 2114–2117. Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot, J., Peleman, J., Kuiper, M. and Zabeau, M. 1995. AFLP: a new technique for DNA fingerprinting. Nucl. Acids Res. 23: 4407–4414.

Walt, D.R. 2000. Bead-based fiber-optic arrays. Science 287: 451– 452. Wang, E., Miller, L.D., Ohnmacht, G.A., Liu, E.T. and Marincola, F.M. 2000a. High-fidelity mRNA amplification for gene profiling. Nature Biotechnol. 18: 457–459. Wang, R., Guegler, K., LaBrie, S.T. and Crawford, N.M. 2000b. Genomic analysis of a nutrient response in Arabidopsis reveals diverse expression patterns and novel metabolic and potential regulatory genes induced by nitrate. Plant Cell 12: 1491–1509. Warrington, J.A., Dee, S. and Trulson, M. 2000. Large-scale genomic analysis using Affymetrix GeneChip®probe arrays. In: M. Schena (Ed.) Microarray Biochip Technology, Eaton Publishing, Natick, MA, pp. 119–148. Welford, S.M., Gregg, J., Chen, E., Garrison, D., Sorensen, P.H., Denny, C.T. and Nelson, S.F. 1998. Detection of differentially expressed genes in primaryn tumor tissues using representational differences analysis coupled to microarray hybridization. Nucl. Acids Res. 26: 3059–3065. Welsh, J., Chada, K., Dalal, S.S., Cheng, R., Ralph, D. and McClelland, M. 1992. Arbitrarily primed PCR fingerprinting of RNA. Nucl. Acids Res. 20: 4965–4970. Wheeler, D.L., Church, D.M., Lash, A.E., Leipe, D.D., Madden, T.L., Pontius, J.U., Schuler, G.D., Schrimi, L.M., Tatusova, T.A., Wagner, L. and Rapp, B.A. 2001. Database resources of the National Center for Biotechnology Information. Nucl. Acids Res. 29: 11–16. White, J.A., Todd, J., Newman, T., Focks, N., Girke, T., Martínez de Ilárduya, O., Jaworski, J.G., Ohlrogge, J.B. and Benning, C. 2000. A new set of Arabidopsis expressed sequence tags from developing seeds. The metabolic pathway from carbohydrates to seed oil. Plant Physiol. 124: 1582–1594. Wisman, E. and Ohlrogge, J. 2000. Arabidopsis microarray service facilities. Plant Physiol. 124: 1468–1471. Yang, G.P., Ross, D.T., Kuang, W.W., Brown, P.O. and Weigel, R.J. 1999. Combining SSH and cDNA microarrays for rapid identification of differentially expressed genes. Nucl. Acids Res. 27: 1517–1523. Yazaki, J., Kishimoto, N., Nakamura, K., Fujii, F., Shimbo, K., Otsuka, Y., Wu, J., Yamamoto, K., Sakata, K., Sasaki, T. and Kikuchi, S. 2000. Embarking on rice functional genomics via cDNA microarray: use of 3 UTR probes for specific gene expression analysis. DNA Res. 7: 367–370. Zhang, J.S., Duncan, E.L., Chang, A.C.M. and Reddel, R.R. 1998. Differential display of mRNA. Mol. Biotechnol. 10: 155–165. Zhao, S., Ooi, S.L. and Pardee, A.B. 1995. New primer strategy improves precision of differential display. Biotechniques 5: 842– 846, 848, 850. Zhou, Y.X., Kalocsai, P., Chen, J.Y. and Shams, S. 2000. Information processing issues and solutions associated with microarray technology. In: M. Schena (Ed.) Microarray Biochip Technology, Eaton Publishing, Natick, MA, pp. 167–200. Zhu, T. and Wang, X. 2000. Large-scale profiling of the Arabidopsis transcriptome. Plant Physiol. 124: 1472–1476. Zong, Q., Schlummer, M., Hood, L. and Morris, D.R. 1999. Messenger RNA translation state: the second dimension of highthroughput expression screening. Proc. Natl. Acad. Sci. USA 96: 10632–10636.