A comparison of global gene expression measurement technologies in Arabidopsis thaliana

Experimental design, preprocessing, normalization and differential expression analysis of small RNA sequencing experiments

Prior to the advent of new, deep sequencing methods, small RNA (sRNA) discovery was dependent on Sanger sequencing, which was time-consuming and limited knowledge to only the most abundant sRNA. The innovation of large-scale, next-generation sequencing has exponentially increased knowledge of the biology, diversity and abundance of sRNA populations. In this review, we discuss issues involved in the design of sRNA sequencing experiments, including choosing a sequencing platform, inherent biases that affect sRNA measurements and replication. We outline the steps involved in preprocessing sRNA sequencing data and review both the principles behind and the current options for normalization. Finally, we discuss differential expression analysis in the absence and presence of biological replicates. While our focus is on sRNA sequencing experiments, many of the principles discussed are applicable to the sequencing of other RNA populations.

Expression quantitative trait loci analysis in plants

An expression Quantitative Trait Locus or eQTL is a chromosomal region that accounts for a proportion of the variation in abundance of a mRNA transcript observed between individuals in a genetic mapping population. A single gene can have one or multiple eQTLs. Large scale mRNA profiling technologies advanced genome-wide eQTL mapping in a diverse range of organisms allowing thousands of eQTLs to be detected in a single experiment. When combined with classical or trait QTLs, correlation analyses can directly suggest candidates for genes underlying these traits. Furthermore, eQTL mapping data enables genetic regulatory networks to be modelled and potentially provide a better understanding of the underlying phenotypic variation. The mRNA profiling data sets can also be used to infer the chromosomal positions of thousands of genes, an outcome that is particularly valuable for species with unsequenced genomes where the chromosomal location of the majority of genes remains unknown. In this review we focus on eQTL studies in plants, addressing conceptual and technical aspects that include experimental design, genetic polymorphism prediction and candidate gene identification.

In silico evaluation of predicted regulatory interactions in Arabidopsis thaliana

Background

Prediction of transcriptional regulatory mechanisms in Arabidopsis has become increasingly critical with the explosion of genomic data now available for both gene expression and gene sequence composition. We have shown in previous work [1], that a combination of correlation measurements and cis-regulatory element (CRE) detection methods are effective in predicting targets for candidate transcription factors for specific case studies which were validated. However, to date there has been no quantitative assessment as to which correlation measures or CRE detection methods used alone or in combination are most effective in predicting TF→target relationships on a genome-wide scale.

Results

We tested several widely used methods, based on correlation (Pearson and Spearman Rank correlation) and cis-regulatory element (CRE) detection (≥1 CRE or CRE over-representation), to determine which of these methods individually or in combination is the most effective by various measures for making regulatory predictions. To predict the regulatory targets of a transcription factor (TF) of interest, we applied these methods to microarray expression data for genes that were regulated over treatment and control conditions in wild type (WT) plants. Because the chosen data sets included identical experimental conditions used on TF over-expressor or T-DNA knockout plants, we were able to test the TF→target predictions made using microarray data from WT plants, with microarray data from mutant/transgenic plants. For each method, or combination of methods, we computed sensitivity, specificity, positive and negative predictive value and the F-measure of balance between sensitivity and positive predictive value (precision). This analysis revealed that the ≥1 CRE and Spearman correlation (used alone or in combination) were the most balanced CRE detection and correlation methods, respectively with regard to their power to accurately predict regulatory-target interactions.

Conclusion

These findings provide an approach and guidance for researchers interested in predicting transcriptional regulatory mechanisms using microarray data that they generate (or microarray data that is publically available) combined with CRE detection in promoter sequence data.

Quantitative miRNA expression analysis: comparing microarrays with next-generation sequencing

Recently, next-generation sequencing has been introduced as a promising, new platform for assessing the copy number of transcripts, while the existing microarray technology is considered less reliable for absolute, quantitative expression measurements. Nonetheless, so far, results from the two technologies have only been compared based on biological data, leading to the conclusion that, although they are somewhat correlated, expression values differ significantly. Here, we use synthetic RNA samples, resembling human microRNA samples, to find that microarray expression measures actually correlate better with sample RNA content than expression measures obtained from sequencing data. In addition, microarrays appear highly sensitive and perform equivalently to next-generation sequencing in terms of reproducibility and relative ratio quantification.

Comparative expression profiling in grape (Vitis vinifera) berries derived from frequency analysis of ESTs and MPSS signatures

BACKGROUND

Vitis vinifera (V. vinifera) is the primary grape species cultivated for wine production, with an industry valued annually in the billions of dollars worldwide. In order to sustain and increase grape production, it is necessary to understand the genetic makeup of grape species. Here we performed mRNA profiling using Massively Parallel Signature Sequencing (MPSS) and combined it with available Expressed Sequence Tag (EST) data. These tag-based technologies, which do not require a priori knowledge of genomic sequence, are well-suited for transcriptional profiling. The sequence depth of MPSS allowed us to capture and quantify almost all the transcripts at a specific stage in the development of the grape berry.

RESULTS

The number and relative abundance of transcripts from stage II grape berries was defined using Massively Parallel Signature Sequencing (MPSS). A total of 2,635,293 17-base and 2,259,286 20-base signatures were obtained, representing at least 30,737 and 26,878 distinct sequences. The average normalized abundance per signature was approximately 49 TPM (Transcripts Per Million). Comparisons of the MPSS signatures with available Vitis species' ESTs and a unigene set demonstrated that 6,430 distinct contigs and 2,190 singletons have a perfect match to at least one MPSS signature. Among the matched sequences, ESTs were identified from tissues other than berries or from berries at different developmental stages. Additional MPSS signatures not matching to known grape ESTs can extend our knowledge of the V. vinifera transcriptome, particularly when these data are used to assist in annotation of whole genome sequences from Vitis vinifera.

CONCLUSION

The MPSS data presented here not only achieved a higher level of saturation than previous EST based analyses, but in doing so, expand the known set of transcripts of grape berries during the unique stage in development that immediately precedes the onset of ripening. The MPSS dataset also revealed evidence of antisense expression not previously reported in grapes but comparable to that reported in other plant species. Finally, we developed a novel web-based, public resource for utilization of the grape MPSS data [1].

A spatial dissection of the Arabidopsis floral transcriptome by MPSS

BACKGROUND

We have further characterized floral organ-localized gene expression in the inflorescence of Arabidopsis thaliana by comparison of massively parallel signature sequencing (MPSS) data. Six libraries of RNA sequence tags from immature inflorescence tissues were constructed and matched to their respective loci in the annotated Arabidopsis genome. These signature libraries survey the floral transcriptome of wild-type tissue as well as the floral homeotic mutants, apetala1, apetala3, agamous, a superman/apetala1 double mutant, and differentiated ovules dissected from the gynoecia of wild-type inflorescences. Comparing and contrasting these MPSS floral expression libraries enabled demarcation of transcripts enriched in the petals, stamens, stigma-style, gynoecia, and those with predicted enrichment within the sepal/sepal-petals, petal-stamens, or gynoecia-stamens.

RESULTS

By comparison of expression libraries, a total of 572 genes were found to have organ-enriched expression within the inflorescence. The bulk of characterized organ-enriched transcript diversity was noted in the gynoecia and stamens, whereas fewer genes demonstrated sepal or petal-localized expression. Validation of the computational analyses was performed by comparison with previously published expression data, in situ hybridizations, promoter-reporter fusions, and reverse transcription PCR. A number of well-characterized genes were accurately delineated within our system of transcript filtration. Moreover, empirical validations confirm MPSS predictions for several genes with previously uncharacterized expression patterns.

CONCLUSION

This extensive MPSS analysis confirms and supplements prior microarray floral expression studies and illustrates the utility of sequence survey-based expression analysis in functional genomics. Spatial floral expression data accrued by MPSS and similar methods will be advantageous in the elucidation of more comprehensive genetic regulatory networks governing floral development.

A comparison of microarray and MPSS technology platforms for expression analysis of Arabidopsis

Background

Several high-throughput technologies can measure in parallel the abundance of many mRNA transcripts within a sample. These include the widely-used microarray as well as the more recently developed methods based on sequence tag abundances such as the Massively Parallel Signature Sequencing (MPSS) technology. A comparison of microarray and MPSS technologies can help to establish the metrics for data comparisons across these technology platforms and determine some of the factors affecting the measurement of mRNA abundances using different platforms.

Results

We compared transcript abundance (gene expression) measurement data obtained using Affymetrix and Agilent microarrays with MPSS data. All three technologies were used to analyze the same set of mRNA samples; these samples were extracted from various wild type Arabidopsis thaliana tissues and floral mutants. We calculated correlations and used clustering methodology to compare the normalized expression data and expression ratios across samples and technologies. Abundance expression measurements were more similar between different samples measured by the same technology than between the same sample measured by different technologies. However, when expression ratios were employed, samples measured by different technologies were found to cluster together more frequently than with abundance expression levels.

Furthermore, the two microarray technologies were more consistent with each other than with MPSS. We also investigated probe-position effects on Affymetrix data and tag-position effects in MPSS. We found a similar impact on Affymetrix and MPSS measurements, which suggests that these effects were more likely a characteristic of the RNA sample rather than technology-specific biases.

Conclusion

Comparisons of transcript expression ratios showed greater consistency across platforms than measurements of transcript abundance. In addition, for measurements based on abundances, technology differences can mask the impact of biological differences between samples and tissues.

Mechanical stress induces biotic and abiotic stress responses via a novel cis-element

Plants are continuously exposed to a myriad of abiotic and biotic stresses. However, the molecular mechanisms by which these stress signals are perceived and transduced are poorly understood. To begin to identify primary stress signal transduction components, we have focused on genes that respond rapidly (within 5 min) to stress signals. Because it has been hypothesized that detection of physical stress is a mechanism common to mounting a response against a broad range of environmental stresses, we have utilized mechanical wounding as the stress stimulus and performed whole genome microarray analysis of Arabidopsis thaliana leaf tissue. This led to the identification of a number of rapid wound responsive (RWR) genes. Comparison of RWR genes with published abiotic and biotic stress microarray datasets demonstrates a large overlap across a wide range of environmental stresses. Interestingly, RWR genes also exhibit a striking level and pattern of circadian regulation, with induced and repressed genes displaying antiphasic rhythms. Using bioinformatic analysis, we identified a novel motif overrepresented in the promoters of RWR genes, herein designated as the Rapid Stress Response Element (RSRE). We demonstrate in transgenic plants that multimerized RSREs are sufficient to confer a rapid response to both biotic and abiotic stresses in vivo, thereby establishing the functional involvement of this motif in primary transcriptional stress responses. Collectively, our data provide evidence for a novel cis-element that is distributed across the promoters of an array of diverse stress-responsive genes, poised to respond immediately and coordinately to stress signals. This structure suggests that plants may have a transcriptional network resembling the general stress signaling pathway in yeast and that the RSRE element may provide the key to this coordinate regulation.

Mechanical stress induces biotic and abiotic stress responses via a novel cis-element

Plants are continuously exposed to a myriad of abiotic and biotic stresses. However, the molecular mechanisms by which these stress signals are perceived and transduced are poorly understood. To begin to identify primary stress signal transduction components, we have focused on genes that respond rapidly (within 5 min) to stress signals. Because it has been hypothesized that detection of physical stress is a mechanism common to mounting a response against a broad range of environmental stresses, we have utilized mechanical wounding as the stress stimulus and performed whole genome microarray analysis of Arabidopsis thaliana leaf tissue. This led to the identification of a number of rapid wound responsive (RWR) genes. Comparison of RWR genes with published abiotic and biotic stress microarray datasets demonstrates a large overlap across a wide range of environmental stresses. Interestingly, RWR genes also exhibit a striking level and pattern of circadian regulation, with induced and repressed genes displaying antiphasic rhythms. Using bioinformatic analysis, we identified a novel motif overrepresented in the promoters of RWR genes, herein designated as the Rapid Stress Response Element (RSRE). We demonstrate in transgenic plants that multimerized RSREs are sufficient to confer a rapid response to both biotic and abiotic stresses in vivo, thereby establishing the functional involvement of this motif in primary transcriptional stress responses. Collectively, our data provide evidence for a novel cis-element that is distributed across the promoters of an array of diverse stress-responsive genes, poised to respond immediately and coordinately to stress signals. This structure suggests that plants may have a transcriptional network resembling the general stress signaling pathway in yeast and that the RSRE element may provide the key to this coordinate regulation.

Plants are sessile organisms constantly challenged by a wide spectrum of biotic and abiotic stresses. These stresses cause considerable losses in crop yields worldwide, while the demand for food and energy is on the rise. Understanding the molecular mechanisms driving stress responses is crucial to devising targeted strategies to engineer stress-tolerant plants. To identify primary stress-responsive genes we examined the transcriptional profile of plants after mechanical wounding, which was used as a brief, inductive stimulus. Comparison of the ensemble of rapid wound response transcripts with published transcript profiles revealed a notable overlap with biotic and abiotic stress-responsive genes. Additional quantitative analyses of selected genes over a wounding time-course enabled classification into two groups: transient and stably expressed. Bioinformatic analysis of rapid wound response gene promoter sequences enabled us to identify a novel DNA motif, designated the Rapid Stress Response Element. This motif is sufficient to confer a rapid response to both biotic and abiotic stresses in vivo, thereby confirming the functional involvement of this motif in the primary transcriptional stress response. The genes we identified may represent initial components of the general stress-response network and may be useful in engineering multi-stress tolerant plants.

RL-SAGE and microarray analysis of the rice transcriptome after Rhizoctonia solani infection

Sheath blight caused by the fungal pathogen Rhizoctonia solani is an emerging problem in rice production worldwide. To elucidate the molecular basis of rice defense to the pathogen, RNA isolated from R. solani-infected leaves of Jasmine 85 was used for both RL-SAGE library construction and microarray hybridization. RL-SAGE sequence analysis identified 20,233 and 24,049 distinct tags from the control and inoculated libraries, respectively. Nearly half of the significant tags (> or =2 copies) from both libraries matched TIGR annotated genes and KOME full-length cDNAs. Among them, 42% represented sense and 7% antisense transcripts, respectively. Interestingly, 60% of the library-specific (> or =10 copies) and differentially expressed (>4.0-fold change) tags were novel transcripts matching genomic sequence but not annotated genes. About 70% of the genes identified in the SAGE libraries showed similar expression patterns (up or down-regulated) in the microarray data obtained from three biological replications. Some candidate RL-SAGE tags and microarray genes were located in known sheath blight QTL regions. The expression of ten differentially expressed RL-SAGE tags was confirmed with RT-PCR. The defense genes associated with resistance to R. solani identified in this study are useful genomic materials for further elucidation of the molecular basis of the defense response to R. solani and fine mapping of target sheath blight QTLs.

Deep and comparative analysis of the mycelium and appressorium transcriptomes of Magnaporthe grisea using MPSS, RL-SAGE, and oligoarray methods

Background

Rice blast, caused by the fungal pathogen Magnaporthe grisea, is a devastating disease causing tremendous yield loss in rice production. The public availability of the complete genome sequence of M. grisea provides ample opportunities to understand the molecular mechanism of its pathogenesis on rice plants at the transcriptome level. To identify all the expressed genes encoded in the fungal genome, we have analyzed the mycelium and appressorium transcriptomes using massively parallel signature sequencing (MPSS), robust-long serial analysis of gene expression (RL-SAGE) and oligoarray methods.

Results

The MPSS analyses identified 12,531 and 12,927 distinct significant tags from mycelia and appressoria, respectively, while the RL-SAGE analysis identified 16,580 distinct significant tags from the mycelial library. When matching these 12,531 mycelial and 12,927 appressorial significant tags to the annotated CDS, 500 bp upstream and 500 bp downstream of CDS, 6,735 unique genes in mycelia and 7,686 unique genes in appressoria were identified. A total of 7,135 mycelium-specific and 7,531 appressorium-specific significant MPSS tags were identified, which correspond to 2,088 and 1,784 annotated genes, respectively, when matching to the same set of reference sequences. Nearly 85% of the significant MPSS tags from mycelia and appressoria and 65% of the significant tags from the RL-SAGE mycelium library matched to the M. grisea genome. MPSS and RL-SAGE methods supported the expression of more than 9,000 genes, representing over 80% of the predicted genes in M. grisea. About 40% of the MPSS tags and 55% of the RL-SAGE tags represent novel transcripts since they had no matches in the existing M. grisea EST collections. Over 19% of the annotated genes were found to produce both sense and antisense tags in the protein-coding region. The oligoarray analysis identified the expression of 3,793 mycelium-specific and 4,652 appressorium-specific genes. A total of 2,430 mycelial genes and 1,886 appressorial genes were identified by both MPSS and oligoarray.

Conclusion

The comprehensive and deep transcriptome analysis by MPSS and RL-SAGE methods identified many novel sense and antisense transcripts in the M. grisea genome at two important growth stages. The differentially expressed transcripts that were identified, especially those specifically expressed in appressoria, represent a genomic resource useful for gaining a better understanding of the molecular basis of M. grisea pathogenicity. Further analysis of the novel antisense transcripts will provide new insights into the regulation and function of these genes in fungal growth, development and pathogenesis in the host plants.

Natural genetic variation in whole-genome expression in Arabidopsis thaliana: the impact of physiological QTL introgression

A long-standing and fundamental question in biology is how genes influence complex phenotypes. Combining near-isogenic line mapping with genome expression profiling offers a unique opportunity for exploring the functional relationship between genotype and phenotype and for generating candidate genes for future study. We used a whole-genome microarray produced with ink-jet technology to measure the relative expression level of over 21,500 genes from an Arabidopsis thaliana near-isogenic line (NIL) and its recurrent parent. The NIL material contained two introgressions (bottom of chromosome II and top of chromosome III) of the Cvi-1 ecotype in a Ler-2 ecotype genome background. Each introgression 'captures' a Cvi allele of a physiological quantitative trait loci (QTL) that our previous studies have shown increases transpiration and reduces water-use efficiency at the whole-plant level. We used a mixed model anova framework for assessing sources of expression variability and for evaluating statistical significance in our array experiment. We discovered 25 differentially expressed genes in the introgression at a false-discovery rate (FDR) cut-off of 0.20 and identified new candidate genes for both QTL regions. Several differentially expressed genes were confirmed with QRT-PCR (quantitative reverse transcription-polymerase chain reaction) assays. In contrast, we found no statistically significant differentially expressed genes outside of the QTL introgressions after controlling for multiple tests. We discuss these results in the context of candidate genes, cloning QTL, and phenotypic evolution.

Bioinformatics and its applications in plant biology

Bioinformatics plays an essential role in today's plant science. As the amount of data grows exponentially, there is a parallel growth in the demand for tools and methods in data management, visualization, integration, analysis, modeling, and prediction. At the same time, many researchers in biology are unfamiliar with available bioinformatics methods, tools, and databases, which could lead to missed opportunities or misinterpretation of the information. In this review, we describe some of the key concepts, methods, software packages, and databases used in bioinformatics, with an emphasis on those relevant to plant science. We also cover some fundamental issues related to biological sequence analyses, transcriptome analyses, computational proteomics, computational metabolomics, bio-ontologies, and biological databases. Finally, we explore a few emerging research topics in bioinformatics.

Comparison of transcript profiling on Arabidopsis microarray platform technologies

To date there have been few systematic studies comparing the results of transcript profiling from different microarray platform technologies. We evaluated in detail two different Arabidopsis thaliana microarray platforms: our own Genomic Amplicon arrays and the Qiagen long oligonucleotide arrays designed by Operon; furthermore, we cross-validated these arrays against the Affymetrix AG and ATH1 GeneChips. Data were obtained from all three platforms in each of two separate experiments; (1) at 2 h and (2) 8 h following a salicylic acid treatment applied to both wild-type and npr1-3 mutant plants. A total of 20 hybridizations were performed, analyzing the expression of 26,814 unique locus IDs. We demonstrate that intensity rank is a key variable that affects both inter-platform and cross-platform reproducibility. Although general agreement between platform technologies is low, data derived from high signal intensities (90th percentile) can correlate as well between differing platforms as replicates within the same platform (r=0.4-0.7). We also show that the identification of differentially expressed genes by significance analysis of microarrays is influenced by signal intensity and that overlap between significant gene lists from different platform technologies was as high as 67% when low intensity values were removed. Validation of 41 genes by Northern blot hybridization showed that all platform technologies performed well, qualitatively confirming 83-100% of differential gene expression. Our results suggest that the potential for the broad integration of microarray data from different platforms and laboratories is promising.

A comparative analysis of transcript abundance using SAGE and Affymetrix arrays

A number of methods are currently used for gene expression profiling. They differ in scale, economy and sensitivity. We present the results of a direct comparison between serial analysis of gene expression (SAGE) and the Barley1 Affymetrix GeneChip. Both technology platforms were used to obtain quantitative measurements of transcript abundance using identical RNA samples and assessed for their ability to quantify differential gene expression. For SAGE, a total of 82,122 tags were generated from two independent libraries representing whole developing barley caryopsis and dissected embryos. The Barley1 GeneChip contains 22,791 probe sets. Results obtained from both methods are generally comparable, indicating that both will lead to similar conclusions regarding transcript levels and differential gene expression. However, excluding singletons, 24.4% of the unique SAGE tags had no corresponding probe set on the Barley1 array indicating that a broader snapshot of gene expression was obtained by SAGE. Discrepancies were observed for a number of "genes" and these are discussed.