Using genomic DNA-based probe-selection to improve the sensitivity of high-density oligonucleotide arrays when applied to heterologous species
- John P Hammond†3,
- Martin R Broadley†2,
- David J Craigon1,
- Janet Higgins1,
- Zoe F Emmerson1,
- Henrik J Townsend1,
- Philip J White3 and
- Sean T May1Email author
© Hammond et al; licensee BioMed Central Ltd. 2005
Received: 03 August 2005
Accepted: 09 November 2005
Published: 09 November 2005
High-density oligonucleotide (oligo) arrays are a powerful tool for transcript profiling. Arrays based on GeneChip® technology are amongst the most widely used, although GeneChip® arrays are currently available for only a small number of plant and animal species. Thus, we have developed a method to improve the sensitivity of high-density oligonucleotide arrays when applied to heterologous species and tested the method by analysing the transcriptome of Brassica oleracea L., a species for which no GeneChip® array is available, using a GeneChip® array designed for Arabidopsis thaliana (L.) Heynh. Genomic DNA from B. oleracea was labelled and hybridised to the ATH1-121501 GeneChip® array. Arabidopsis thaliana probe-pairs that hybridised to the B. oleracea genomic DNA on the basis of the perfect-match (PM) probe signal were then selected for subsequent B. oleracea transcriptome analysis using a .cel file parser script to generate probe mask files. The transcriptional response of B. oleracea to a mineral nutrient (phosphorus; P) stress was quantified using probe mask files generated for a wide range of gDNA hybridisation intensity thresholds. An example probe mask file generated with a gDNA hybridisation intensity threshold of 400 removed > 68 % of the available PM probes from the analysis but retained >96 % of available A. thaliana probe-sets. Ninety-nine of these genes were then identified as significantly regulated under P stress in B. oleracea, including the homologues of P stress responsive genes in A. thaliana. Increasing the gDNA hybridisation intensity thresholds up to 500 for probe-selection increased the sensitivity of the GeneChip® array to detect regulation of gene expression in B. oleracea under P stress by up to 13-fold. Our open-source software to create probe mask files is freely available http://affymetrix.arabidopsis.info/xspecies/ and may be used to facilitate transcriptomic analyses of a wide range of plant and animal species in the absence of custom arrays.
KeywordsArabidopsis thaliana Brassica oleracea Brassicaceae cross-species microarray oligonucleotide phosphate phosphorus plant nutrition probe masking transcriptomics
High-density oligonucleotide (oligo) arrays are a powerful and widely used tool for large-scale gene-expression profiling . One extensively validated oligo array type is available as commercial GeneChip® technology (Affymetrix, Santa Clara, USA). GeneChip® arrays use probe-sets rather than single oligos per gene, comprising of between 11 and 20 probe-pairs to quantify abundance for each transcript. Each probe-pair consists of a perfect-match (PM) and a mismatch (MM) probe. The PM probe is a 25-base sequence complementary to the target transcript, whilst the MM probe is identical to the PM probe with the exception of a single mismatch at the 13th base. Transcript abundance can be calculated either from extrapolated hybridisation differences between PM and MM probes across a probe-set, or simply by using the PM data depending on the analysis used. GeneChip® arrays provide reproducible, accurate data at high throughput rates which can be easily stored and compared across experiments [2–4]. An example of the wide-scale adoption of GeneChip® technology can be seen within the plant sciences research community, where data from thousands of GeneChip® arrays under large numbers of experimental challenges on the model plant Arabidopsis thaliana (L.) Heynh. are publicly-available . Unfortunately, at present, GeneChip® arrays are available for only a few species of eukaryotes. Thus, in contrast to A. thaliana, the transcriptomes of most agriculturally- or ecologically-important plant species are less extensively studied, since extensive sequence information and the fabrication of expensive custom arrays would be required before experimentation can begin.
One solution to this problem is to use GeneChip® arrays designed for model organisms to study closely related species [5–12]. Whilst this strategy is feasible, studies conducted to date have not accounted for the probability that inefficient hybridisation of transcripts from the target species to GeneChip® probes designed to the model species will attenuate the overall signal calculated across a probe-set . For example, Ji et al.  describe a scenario where expression data analysed in Microarray Analysis Suite (MAS Version 5.0; Affymetrix), is based on the calculation that the weight carried by a probe within a probe-set is inversely related to its distance from the mean value of all probes within the probe-set. Thus, probes not generating signals due to sequence polymorphisms with the target organism will reduce the quality of information available from experiments using GeneChip® arrays designed for a model species to monitor the transcriptome of a closely related species. Ji et al.  thus adopted an RNA-based probe-selection system, to study non-human mammalian transcriptomes with human GeneChip® arrays, using probe mask files to exclude probes which hybridised weakly to their target transcript. However, this technique reduced the number of probe-sets available for transcriptome analysis by biasing the analysis towards the measurement of abundant transcripts.
To address the potential problems of using GeneChip® arrays designed for one species to monitor the transcriptome of a closely related species, we have developed a novel technique to improve the sensitivity of high-density oligonucleotide arrays when applied to heterologous species. In contrast to Ji et al. , this technique is based on selecting probe-pairs based on the hybridisation efficiency of the PM oligonucleotide probe with genomic DNA from a target species for which the GeneChip® was not originally designed. We initially used the A. thaliana ATH1-121501 GeneChip® array to study the transcriptome of Brassica oleracea L., which represents several important crops, including cabbage, kale, broccoli, Brussels sprout, cauliflower and kohlrabi, and half of the B. napus (oil-seed rape) genome. Brassica oleracea was chosen because the ancestral lineage of Arabidopsis and Brassica (family Brassicaceae) diverged only 12 to 19 million years ago . The effects of genomic DNA-based probe-selection on estimates of transcriptional regulation in B. oleracea were quantified under an imposed physiological stress of phosphorus (P) deprivation. Probe mask files increased sensitivity of the ATH1-121501 GeneChip® array when detecting regulation of gene expression in B. oleracea under P stress by up to 13-fold. The response of B. oleracea to P stress was compared with that of A. thaliana [15–17]. The transcriptional response of plants to P stress can provide insights into nutrient signal pathways  and can inform efforts to improve fertiliser-use efficiency of crop production systems through breeding and nutritional diagnostics .
Results and Discussion
Genomic-DNA hybridisation and probe-selection
We used A. thaliana ATH1-121501 GeneChip® arrays to study the transcriptome of B. oleracea. Sequence polymorphisms between the two species are likely to result in an underestimate of transcript abundance if all probes are used within individual probe-sets . Therefore, we selected subsets of probe-pairs from each probe-set on the A. thaliana GeneChip® array based on the hybridisation efficiency of genomic DNA from B. oleracea to homologous A. thaliana GeneChip® array PM probes.
Genomic DNA from B. oleracea was biotin-labelled and hybridised to the A. thaliana ATH1-121501 GeneChip® array. Probe-sets were selected for subsequent transcriptome analyses if the probe-set was represented by PM probes with gDNA hybridisation intensities above a set threshold. Selection was performed using a .cel file parser script written in the Perl programming language (Xspecies Version 1.1, available at http://affymetrix.arabidopsis.info/xspecies/). The method was optimised empirically by generating 13 probe mask files with gDNA hybridisation intensity thresholds ranging from 0 (i.e. no probe-selection) to 1000. These 13 probe mask files were assessed in turn to investigate the quantification of the transcriptional response of B. oleracea to an imposed mineral nutrient (phosphorus; P) stress.
Testing DNA probe-selection: transcript abundance in control Brassica oleracea
Brassica oleracea were grown hydroponically using physiological techniques described elsewhere . Control plants were supplied with a full nutrient solution throughout the experiment; treated (P-starved) plants were supplied with a nutrient solution containing no P for 100 h. Transcriptional responses of B. oleracea to phosphate stress were determined by challenging A. thaliana ATH1-121501 GeneChip® arrays with total RNA extracted from control and treated plants. Following scanning of the GeneChip® arrays, raw cell intensity data files (.cel files) were loaded into GeneSpring (Silicon Genetics, CA, USA) using 13 probe mask files. For each of the 13 probe-selection conditions, data were prenormalised using Robust Multichip Average (RMA) algorithms . Subsequently, data from the eight GeneChip® arrays (four control and four P-starved samples) were treated as 13 individual 'experimental scenarios' to evaluate probe-selection stringency. Within each 'scenario', data were further standardised within GeneSpring; the signal value from each replicate P-starved plant was standardised to its corresponding control sample to give an expression ratio for each gene.
Testing DNA probe-selection: gene regulation under P stress in Brassica oleracea
Estimates of fold-differences in gene expression (P-starved samples versus control samples) increased using probe mask files generated with gDNA hybridisation intensity thresholds up to 500 (Figure 3, 4c). 'Volcano' plots summarise these effects (Figure 3a–c). In volcano plots, the y-axis represents the log10 of the reciprocal of the one-way ANOVA P- value corrected for multiple testing using the BH-FDR; the x- axis represents the log2 of the fold-change in gene expression. There was a 13-fold increase in the number of genes identified as differentially regulated > ± 1.3-fold under P stress using a probe mask file generated with a gDNA hybridisation intensity threshold of 500 (Figures 3, 4c). The use of probe mask files generated with higher gDNA hybridisation intensity thresholds (up to 1000) resulted in a substantial loss of probe-sets available for transcriptome analysis (Figures 1, 3d). However, the number of genes identified as differentially regulated > ± 1.3-fold was still much greater when using the probe mask file generated with a gDNA hybridisation intensity threshold of 1000 than when probe-selection was not used, even though only 7.3 % of the total number of PM probes on the GeneChip® array were used (Figure 4c). Estimates of the number of genes significantly differentially regulated under P stress (at BH-FDR MTC P < 0.05) increased when probe mask files were used (Figure 4a). For example, using a probe mask file generated with a gDNA hybridisation intensity threshold of 500, 111 genes were estimated as significantly differentially regulated, compared to only eight genes regulated using no probe-selection.
An optimal probe-selection strategy for transcriptional analysis of B. oleracea is to use a probe mask file generated at a gDNA hybridisation intensity threshold of 200 or greater. Although estimates of fold-differences in gene expression in P-starved versus control samples were greatest using probe mask files generated with gDNA hybridisation intensity thresholds of up to 500 (Figure 4c,d), there was a significant loss of available probe-sets for transcriptome analysis with gDNA hybridisation intensity thresholds > 500. Further, estimates of the number of significantly regulated genes declined using probe mask files generated at DNA hybridisation intensity thresholds > 500.
Biological significance of genes regulated under P stress in Brassica oleracea
Following probe-selection using a gDNA hybridisation intensity threshold of 400, 99 genes were identified as significantly differentially regulated in the shoots of B. oleracea following the withdrawal of P from the nutrient solution (BH-FDR MTC P < 0.05; Additional file 1). Of these genes, 39 had higher transcript abundance in P-starved samples than in control samples and 60 had lower transcript abundance in P-starved samples than in control samples. Data for all genes, analysed with all probe mask files generated at gDNA hybridisation intensity thresholds from 0 to 1000, are presented in Supplementary Table I.
Sequence polymorphisms between A. thaliana and B. oleracea are likely to result in the identification of homologous genes or alternative members of gene families in addition to possible gene orthologues. For this reason, we compared previously published studies on the response of A. thaliana to P starvation [15–17] at the level of individual gene and also at the level of the gene family and functional category. Several homologous genes responded similarly to P starvation in B. oleracea and A. thaliana. For example, SQD2 (At5g01220), which is involved in sulpholipid biosynthesis, increases its expression in responses to P starvation in A. thaliana [15–17, 20–22]. A homologue of this gene had higher hybridisation signals in P-starved B. oleracea than in control samples. The signal values for SQD2, obtained using the probe mask file generated with a gDNA hybridisation intensity threshold of 400, were based on two probes, in contrast to 11 probes when no probe-selection was used. The use of probe-selection increased the significance and differences in transcript abundance between P-starved and control plants (SQD2, 1.60 ± 0.2, BH-FDR MTC P = 0.047; mean normalised signal ratio ± 1 S.D.) than in the absence of probe-selection (SQD2, 1.53 ± 0.58, BH-FDR MTC P = 0.382).
Ribonucleases, phosphatases and phosphodiesterases are involved in the recycling of P in plants during P starvation [16, 23, 24]. The hybridisation intensity of the ribonuclease RNS2 (At2g39780) was higher in P-starved samples compared to control samples, and is thought to be involved in the release of P from internal sources during P starvation . The signal value for RNS2, obtained using the probe mask file generated with a gDNA hybridisation intensity threshold of 400, was based on six probes in contrast to 11 probes when no probe-selection was used. The use of probe-selection increased normalised signal ratios in P-starved plants compared to control plants to a greater amount (RNS2, 1.92 ± 0.43, BH-FDR MTC P = 0.06, mean normalised signal ratio ± 1 S.D.) than in the absence of probe-selection, where the up-regulation of the gene was not detected (RNS2, 1.07 ± 0.06, BH-FDR MTC P = 0.43).
Higher hybridisation signals in P-starved plants compared to control samples were also observed for genes belonging to gene families or functional groups which have previously been observed responding to P starvation in A. thaliana, although not significantly [15–17]. These include genes involved in the transport of phosphate and phosphate-containing compounds and carbohydrates, indicating a change in the internal economy of P use in response to P starvation (Additional file 1). For example, at a gDNA hybridisation intensity of 400, two sucrose-phosphate synthases, and a malic enzyme had higher hybridisation intensities for in P-starved compared to control samples, whilst lower hybridisation intensities for genes involved in alternative glycolosis reactions that conserve P during P starvation conditions  were noted. There are other notable similarities in the P starvation response of B. oleracea with published responses of A. thaliana to P starvation at the level of the gene family. These include transcription factors from the zinc finger, F-box, bHLH and myb families, genes involved in flavanoid biosynthesis and genes encoding proteins from the cytochrome P450, glycosyl hydrolase, and peroxidase families.
Since GeneChip® arrays are available for only a small number of species, we have developed a novel probe-selection method to study the transcriptome of a plant or animal species for which GeneChip® arrays are not available. Genomic DNA from B. oleracea was labelled and hybridised to the A. thaliana ATH1-121501 GeneChip® array. Perfect-match A. thaliana probes which hybridised to the B. oleracea genomic DNA above selected hybridisation intensities were selected for subsequent B. oleracea transcriptome analysis using probe mask files generated using a .cel file parser script. Software to create probe mask files is freely available http://affymetrix.arabidopsis.info/xspecies/ and is designed to facilitate the analysis of the transcriptomes of a wide range of other species in the absence of custom arrays.
Probe-selection was tested by quantifying the transcriptional response of B. oleracea to a mineral nutrient (phosphorus; P) stress using probe mask files generated at gDNA hybridisation intensity thresholds ranging from 0 (i.e. no probe-selection) to 1000. A probe mask file generated at a gDNA hybridisation intensity threshold of 400 masked > 68 % of all of the A. thaliana probe-pairs from the subsequent transcriptome analysis whilst retaining 96.4 % of the total available probe-sets. Increasing the gDNA hybridisation intensity thresholds for probe mask file generation increased the sensitivity of the GeneChip® array to detect regulation of gene expression in B. oleracea under P stress by up to 13-fold. Ninety-nine genes were significantly regulated in the shoots of B. oleracea under P stress (BH-FDR MTC P < 0.05). Confirmatory analyses using quantitative PCR and reporter-genes fused to nutrient responsive promoters have already been conducted for several P stress regulated genes in A. thaliana [15, 16] and detailed confirmatory and functional molecular analysis of these genes is now clearly required in B. oleracea to resolve the biological significance of the P stress response. Understanding P stress responses in plants is likely, (i) to hasten the development of more nutrient efficient varieties of crops, (ii) to improve our understanding of nutrient signalling pathways, and (iii) to yield rapid advances in nutritional diagnostics, in particular if unit costs for microarray analyses continue to decrease [16, 18]. We appreciate that the resolution and clarity of the probe mask method will be influenced by gene duplication and will necessitate validation of the potential homologues found in the target species, however this caution is generally true in the design of any new microarray. Most array users are primarily interested in transcript differences between samples as clues to further investigation of a gene or gene family. This technique enables such differential expression indicators for genes from unusual species and additionally provides putative model organism gene homologue ontologies via the standard Affymetrix gene annotation. As a pragmatic method based on gDNA hybridisation the technique requires investigation of a range of thresholds to systematically test probe-selection at the bioinformatics level. Re-optimisation of gDNA hybridisation thresholds is particularly important for new species given that gDNA quality and origin may affect overall absolute hybridisation intensities per experiment but should not affect relative ranking of intensities between hybridisations of the same genome. The technique therefore allows an economical analysis of any RNA hybridisations performed, since gDNA-based probe-selection can be optimised at any time without repeating the RNA work. The method also requires minimal setup outlay represented by one representative gDNA hybridisation and some straightforward bioinformatics time, in contrast to the costs of manufacturing even a simple single oligo array. More importantly, however, gDNA-based probe-selection could facilitate transcriptional profiling of a wide range of plant and animal species of agricultural, ecological and evolutionary importance, even in the absence of available genomic information.
Materials and methods
Genomic DNA hybridisation and probe-selection
Probe-pairs from the A. thaliana ATH1-121501 GeneChip® array (Affymetrix, Santa Clara, CA, USA) were selected for transcriptome analysis of B. oleracea using a gDNA-based probe-selection strategy based on the hybridisation of gDNA to the PM probe. Total genomic DNA was extracted from 5 g of B. oleracea var. alboglabra cv. A12DHd leaf tissue using a DNeasy Plant mini kit (Qiagen Ltd, Crawley, UK). Genomic DNA was labelled using the Bioprime DNA labelling System (Invitrogen, Paisley, UK) and subsequently hybridised to Affymetrix ATH1-121501 GeneChip® arrays for 16 h at 45°C using standard Affymetrix hybridisation protocols. Subsequently, the GeneChip® array was scanned on an Affymetrix G2500A GeneArray scanner and a cell intensity file (.cel file) was generated using Microarray Analysis Suite (MAS Version 5.0; Affymetrix). This .cel file contained the gDNA hybridisation intensities between B. oleracea genomic DNA fragments and all A. thaliana probes. Probe-pairs from the .cel file were selected for subsequent transcriptome analysis using a .cel file parser script (Xspecies Version 1.1) written in the Perl programming language http://www.perl.com. The Perl script was designed to create probe mask (.cdf) files compatible with a range of microarray analysis software packages. A probe-set was selected when it was represented by one or more PM probe-pair(s) per probe-set (i.e. a minimum of 25 bp identical probe sequence to A. thaliana was required for subsequent transcriptome analysis of B. oleracea). There was no a priori restriction of a suitable gDNA hybridisation intensity threshold for probe mask file generation in a target species. Thus, the algorithm was designed to allow a user-specified gDNA hybridisation intensity threshold for probe mask file generation to be set. Files (cdf.) were therefore generated using a range of gDNA hybridisation intensity thresholds (from 0 to 1000). The Perl algorithm and B. oleracea DNA .cel files are freely available from http://affymetrix.arabidopsis.info/xspecies along with similar files for other species. The Perl masking script and instructions for using the script are also available as Additional file 2 and Additional file 3.
Testing the gDNA-based probe-selection strategy: determining the transcriptional response of B. oleracea to phosphate stress
Brassica oleracea var. alboglabra cv. A12DHd were grown hydroponically from seed in a system described in Hammond et al., . Control plants were supplied with nutrient solution containing all nutrients throughout the experiment whilst treated plants were supplied with a nutrient solution containing no P for 100 h. Replicate samples (each replicate comprising 8–10 individual plants) were harvested from control and treated plants, mid way through the photo-period, and snap frozen in liquid nitrogen. All plants had approximately 6–8 leaves at harvest. Three biological replicates and one technical replicate were used for transcriptome analysis.
Tissue samples, previously stored at -70°C, were placed in liquid nitrogen before grinding. To each sample, 1 ml of TRIzol reagent (Invitrogen) was added and total RNA was subsequently extracted as described previously . Total RNA yield and purity were determined using an Agilent 2100 Bioanalyser (Agilent Technologies, Stockport, Cheshire, UK). Approximately 5 μg of total RNA was reverse transcribed at 42°C for 1 h to generate first strand cDNA using 100 pmol oligo dT(24) primer containing a 5'-T7 RNA polymerase promoter sequence, 50 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl2, 10 mM dithiothreitol (DTT), 10 mM dNTPs and 200 units SuperScript II reverse transcriptase (Invitrogen Life Technologies). Following first strand cDNA synthesis, second strand cDNA was synthesised using 10 units of Escherichia coli polymerase I, 10 units of E. coli DNA ligase and 2 units of RNase H in a reaction containing 25 mM Tris-HCl (pH 7.5), 100 mM KCl, 5 mM MgCl2, 10 mM (NH4)SO4, 0.15 mM b-NAD+ and 10 mM dNTPs. The second strand synthesis reaction proceeded at 16°C for 2 h before 10 units of T4 DNA polymerase was added and the reaction allowed to proceeded for a further 5 minutes. The reaction was terminated by adding 0.5 M EDTA. Double stranded cDNA products were purified using the GeneChip® Sample Cleanup Module (Affymetrix). The synthesised cDNAs were in-vitro transcribed by T7 RNA polymerase (Enzo BioArray High Yield RNA Transcript Labelling Kit, Enzo Life Sciences Inc., Farmingdale, NY, USA) using biotinylated nucleotides to generate biotinylated complementary RNAs (cRNAs). The cRNAs were purified using the GeneChip® Sample Cleanup Module (Affymetrix). The cRNAs were then randomly fragmented at 94°C for 35 minutes in a buffer containing 40 mM Tris-acetate (pH 8.1), 100 mM potassium acetate, and 30 mM magnesium acetate to generate molecules of approximately 35 to 200 bp. Affymetrix A. thaliana ATH1-121501 GeneChip® arrays were hybridised with 15 μg of fragmented labelled cRNA for 16 h at 45°C as described in the Affymetrix Technical Analysis Manual. GeneChip® arrays were stained with Streptavidin-Phycoerythrin solution and scanned with an Affymetrix G2500A GeneArray scanner.
Microarray Analysis Suite (MAS Version 5.0; Affymetrix) was used to generate .cel files for each of the biological RNA replicates by scanning and computing summary intensities for each probe without the use of probe mask files. These .cel files were loaded into GeneSpring (Agilent Technologies) analysis software package using the Robust Multichip Average (RMA) pre-normalisation algorithm (Irizarry et al., 2003). During .cel file loading and pre-normalisation, .cel files were interpreted using either, (1) the A. thaliana .cdf file (i.e. with no probe-selection used), or (2) using .cdf files generated from the gDNA .cel file with gDNA hybridisation intensity thresholds from 50 to 1000. Following RMA pre-normalisation and masking of individual probes, further standardisations were applied to the probe-set raw signal value in GeneSpring. For each replicate array, each probe-set signal value from treated (P-starved) samples was standardised to the probe-set signal value of its corresponding control sample to give gene expression ratios between the two conditions. Genes with differential hybridisation intensities between P-starved and control samples were identified using a one-way ANOVA, with a Benjamini and Hochberg false discovery rate (0.05) multiple testing correction applied, to identify genes that were significantly differentially expressed between the two conditions. Genes with Benjamini and Hochberg FDR multiple test corrected P-values less than 0.05 were considered to be differentially regulated under P starvation.
Seeds of Brassica oleracea var. alboglabra cv. A12DHd were kindly provided by Graham King (Warwick HRI, UK). J. Okyere (University of Nottingham, UK) and Santhoshi Bandla, Cheng Li (Department of Biostatistics, Harvard School of Public Health, US) provided helpful technical discussions in the early stages of the project. The work was supported financially by the UK Department for the Environment, Food and Rural Affairs (grants HH3501SFV and HH3504SPO), by the UK BBSRC Investigating Gene Function Arabidopsis award (IGF12422), and by a University of Nottingham New Lecturer's Award.
- Lipshutz RJ, Fodor SPA, Gingeras TR, Lockhart DJ: High density synthetic oligonucleotide arrays. Nature Genet. 1999, 21: 20-24. 10.1038/4447.View ArticlePubMedGoogle Scholar
- Harmer SL, Hogenesch LB, Straume M, Chang HS, Han B, Zhu T, Wang X, Kreps JA, Kay SA: Orchestrated transcription of key pathways in Arabidopsis by the circadian clock. Science. 2000, 290: 2110-2113. 10.1126/science.290.5499.2110.View ArticlePubMedGoogle Scholar
- Zhu T, Wang X: Large scale profiling of the Arabidopsis transcriptome. Plant Physiol. 2000, 124: 1472-1476. 10.1104/pp.124.4.1472.PubMed CentralView ArticlePubMedGoogle Scholar
- Craigon DJ, James N, Okyere J, Higgins J, Jotham J, May S: NASCArrays: a repository for microarray data generated by NASC's transcriptomics service. Nucleic Acids Res. 2004, 32: D575-D577. 10.1093/nar/gkh133.PubMed CentralView ArticlePubMedGoogle Scholar
- Chismar JD, Mondala T, Fox HS, Roberts E, Langford D, Masliah E, Salomon DR, Head SR: Analysis of results variability from high-density oligonucleotide arrays comparing same-species and cross-species hybridisations. Biotechniques. 2002, 33: 516-524.PubMedGoogle Scholar
- Enard W, Khaitovich P, Klose J, Zöllner S, Heissig F, Giavalisco P, Nieselt-Struwe K, Muchmore E, Varki A, Ravid R, Doxiadis GM, Bontrop RE, Pääbo S: Intra-and interspecific variation in primate gene expression patterns. Science. 2002, 296: 340-343. 10.1126/science.1068996.View ArticlePubMedGoogle Scholar
- Caceres M, Lachuer J, Zapala MA, Redmond JC, Kudo L, Geschwind DH, Lockhart DJ, Preuss TM, Barlow C: Elevated gene expression levels distinguish human from non-human primate brains. Proc Natl Acad Sci USA. 2003, 100: 13030-13035. 10.1073/pnas.2135499100.PubMed CentralView ArticlePubMedGoogle Scholar
- Higgins MA, Berridge BR, Mills BJ, Schultze AE, Gao H, Searfoss GH, Baker TK, Ryan TP: Gene Expression analysis of the acute phase response using a canine microarray. Toxicol Sci. 2003, 74: 470-484. 10.1093/toxsci/kfg142.View ArticlePubMedGoogle Scholar
- Becher M, Talke IN, Krall L, Krämer U: Cross-species microarray transcript profiling reveals high constitutive expression of metal homeostasis genes in shoots of the zinc hyperaccumulator Arabidopsis halleri. Plant J. 2004, 37: 251-268.View ArticlePubMedGoogle Scholar
- Khaitovich P, Weiss G, Lachmann M, Hellmann I, Enard W, Muetzel B, Wirkner U, Ansorge W, Pääbo S: A neutral model of transcriptome evolution. PLo S Biol. 2004, 2: 682-689.Google Scholar
- Uddin M, Wildman DE, Liu GZ, Xu WB, Johnson RM, Hof PR, Kapatos G, Grossman LI, Goodman M: Sister grouping of chimpanzees and humans as revealed by genome-wide phylogenetic analysis of brain gene expression profiles. Proc Natl Acad Sci USA. 2004, 101: 2957-2962. 10.1073/pnas.0308725100.PubMed CentralView ArticlePubMedGoogle Scholar
- Weber M, Harada E, Vess C, v. Roepenack-Lahaye E, Clemens S: Comparative microarray analysis of Arabidopsis thaliana and Arabidopsis halleri roots identifies nicotianamine synthase, a ZIP transporter and other genes as potential metal hyperaccumulation factors. Plant J. 2004, 37: 269-281. 10.1111/j.1365-313X.2003.02013.x.View ArticlePubMedGoogle Scholar
- Ji W, Zhou W, Gregg K, Yu N, Davis S, Davis S: A method for cross-species gene expression analysis with high-density oligonucleotide arrays. Nucleic Acids Res. 2004, 32: e93-10.1093/nar/gnh084. doi:10.1093/nar/gnh084.PubMed CentralView ArticlePubMedGoogle Scholar
- AGI (The Arabidopsis Genome Initiative): Analysis of the genome sequence of the flowering plant Arabidopsis thaliana. Nature. 2000, 408: 796-815. 10.1038/35048692.View ArticleGoogle Scholar
- Hammond JP, Bennett MJ, Bowen HC, Broadley MR, Eastwood DC, May ST, Rahn C, Swarup R, Woolaway KE, White PJ: Changes in gene expression in Arabidopsis shoots during phosphate starvation and the potential for developing smart plants. Plant Physiol. 2003, 132: 578-596. 10.1104/pp.103.020941.PubMed CentralView ArticlePubMedGoogle Scholar
- Hammond JP, Broadley MR, White PJ: Genetic responses to phosphorus deficiency. Ann Bot. 2004, 94: 323-332. 10.1093/aob/mch156.PubMed CentralView ArticlePubMedGoogle Scholar
- Wu P, Ma L, Hou X, Wang M, Wu Y, Liu F, Deng XW: Phosphate starvation triggers distinct alterations of genome expression in arabidopsis roots and leaves. Plant Physiol. 2003, 132: 1260-1271. 10.1104/pp.103.021022.PubMed CentralView ArticlePubMedGoogle Scholar
- Ticconi CA, Abel S: Short on phosphate: plant surveillance and countermeasures. Trends in Plant Science. 2004, 9: 548-555. 10.1016/j.tplants.2004.09.003.View ArticlePubMedGoogle Scholar
- Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, Speed TP: Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003, 4: 249-264. 10.1093/biostatistics/4.2.249.View ArticlePubMedGoogle Scholar
- Essigmann B, Güler S, Narang RA, Linke D, Benning C: Phosphate availability affects the thylakoid lipid composition and the expression of SQD1, a gene required for sulfolipid biosynthesis in Arabidopsis thaliana. Proc Natl Acad Sci USA. 1998, 95: 1950-1955. 10.1073/pnas.95.4.1950.PubMed CentralView ArticlePubMedGoogle Scholar
- Yu B, Xu C, Benning C: Arabidopsis disrupted in SQD2 encoding sulfolipid synthase is impaired in phosphate-limited growth. Proc Natl Acad Sci USA. 2002, 99: 5732-5737. 10.1073/pnas.082696499.PubMed CentralView ArticlePubMedGoogle Scholar
- Kobayashi K, Awai K, Takamiya K-I, Ohta H: Arabidopsis type B monogalactosyldiacylglycerol synthase genes are expressed during pollen tube growth and induced by phosphate starvation. Plant Physiol. 2004, 134: 640-648. 10.1104/pp.103.032656.PubMed CentralView ArticlePubMedGoogle Scholar
- Bariola PA, Howard CJ, Taylor CB, Verburg MT, Jaglan VD, Green PJ: The Arabidopsis ribonuclease gene RNS1 is tightly controlled in response to phosphate limitation. Plant J. 1994, 6: 673-685. 10.1046/j.1365-313X.1994.6050673.x.View ArticlePubMedGoogle Scholar
- van der Rest B, Boisson A-M, Gout E, Bligny R, Douce R: Glycerophosphocholine metabolism in higher plant cells. Evidence of a new glyceryl-phosphodiester phosphodiesterase. Plant Physiol. 2003, 130: 244-255. 10.1104/pp.003392.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.