Transcripts regulated by the RNA-binding protein AtGRP7 are expected to have altered expression levels in transgenic plants constitutively overexpressing AtGRP7, as observed for endogenous AtGRP7 whose abundance is depressed by the elevated AtGRP7 protein level [20]. Therefore, differences in the mRNA complement of wt and AtGRP7-ox plants were analyzed on the Affymetrix ATH1 microarray. To filter out any line- or accession-specific effects, independent transgenic lines in the C24 background, line RS13 [20], and in the Col background, line G [6], were assayed. Transgenic lines and the wild types were grown in parallel in long days (16 h light, 8 h darkness) and harvested at the circadian maximum of AtGRP7 expression (zt12, zeitgeber time 12, 12 hrs after lights on).

The RankProduct algorithm was employed to compare transcript profiles of the AtGRP7-ox lines to the respective wild types [34]. Transcripts corresponding to 153 probe sets were present at an elevated level in AtGRP7-ox plants (Additional file 1) and 161 were present at a reduced level (Additional file 2) with a signal log ratio > 0.6 in all three experiments (p < 0.05). Among the transcripts most strongly reduced in AtGRP7-ox plants was AtGRP8 previously shown to be under negative control by AtGRP7 [20,24], validating the strategy.

To determine the proportion of rhythmic transcripts among the AtGRP7 targets, we compared the differentially expressed genes to a published dataset scoring 15.4% of the Arabidopsis genes as circadian-regulated [35]. In this experiment, eight-day-old seedlings were entrained in 12-hr-light/12-hr-dark cycles before transfer to constant light and harvested at 4-hr-intervals, starting 26 h after the last dark-light transition. Thus, subjective dawn corresponds to zt24 and zt48, respectively, and subjective dusk corresponds to zt36. AtGRP7, also named CCR2 (COLD AND CIRCADIAN REGULATED 2), peaked around zt36 in this experiment. Among the transcripts we found to be differentially expressed in AtGRP7-ox plants rhythmic transcripts are significantly enriched. 47 of the 153 probe sets with elevated levels in AtGRP7-ox, corresponding to 30.7% (p = 5.27E+7) (Additional file 1) and 61 of the 161 probe sets with reduced levels corresponding to 37.9% (p = 7.91E+13) (Additional file 2) are among those classified as rhythmic in the Edwards dataset. Edwards and coworkers have shown a nearly uniform distribution of the peaks across all time points (Figure 1A) implicating a complex network downstream of the core oscillator in conveying different phases upon clock-controlled output genes. Notably, the genes differentially expressed in AtGRP7-ox are biased toward specific circadian phases. A large fraction of transcripts with reduced level in AtGRP7-ox mainly peaks between zt30 and zt35, in the second half of the subjective day (Figure 1B). This suggests that AtGRP7 has a mostly negative effect on transcripts oscillating with a similar phase. Conversely, transcripts upregulated in AtGRP7-ox mainly peak between zt44 and zt50 in the Edwards dataset (Figure 1C), i.e. towards the end of the subjective night and thus in antiphase to AtGRP7.

To determine how AtGRP7 affects rhythmic downstream genes, oscillations of selected candidate target transcripts were compared between wt and AtGRP7-ox plants under free-running conditions. Plants were grown in long days for two weeks and subsequently transferred to continuous light and harvested at 3-h intervals for three days. SALT TOLERANCE HOMOLOGUE (STH) encoding a B-box zinc finger protein [36] which shows a mean SLR of 1.5 in all three AtGRP7-ox vs. wt comparisons at zt12 oscillates with a maximum around subjective dawn both in wt and AtGRP7-ox plants (Figure 2A). In the AtGRP7-ox plants, the STH peak is higher and broader while the phase is maintained. HY5-HOMOLOG (HYH) encoding a bZip transcription factor involved in phyB signalling [36] which shows a mean SLR of 1.4 in all three AtGRP7-ox vs. wt comparisons at zt12 (Figure 2B) oscillates at a higher level in AtGRP7-ox plants.

The observation that one third of AtGRP7 targets are clock-controlled is in line with a function of the AtGRP7 feedback loop in clock output but the fact that two thirds are not classified as rhythmic [35] points to additional processes AtGRP7 may be involved in. The distribution of genes into ontology categories corresponding to plant GOSLIM revealed that the categories "response to stress" and "response to abiotic or biotic stimulus" were significantly enriched among transcripts with reduced levels in AtGRP7-ox plants (p ≤ 1.1E-2 and p ≤ 1.1E-3, respectively) and were also prevalent among transcripts with elevated levels (Table 1, table 2). Further, significant enrichment among transcripts with reduced abundance in AtGRP7-ox was found for transcripts upregulated by the phytohormones methyl jasmonate (p ≤ 3.2E-8) and abscisic acid (p ≤ 1.8E-6).

To validate a differential expression of these candidate targets, a suite of them was monitored in several independent AtGRP7-ox plants and the corresponding wt plants using RT-PCR and Realtime PCR on zt12 RNA. The transcripts encoding the pathogenesis-related proteins PR1, PR2 encoding a β-glucanase and PR5 encoding an antimicrobial thaumatin-like protein are present at elevated levels in AtGRP7-ox plants (Figure 3A, Additional file 3). Because two cold- and ABA-regulated transcripts, RD29A (COR78) and RAB18, were confirmed to have reduced levels in AtGRP7-ox plants (Figure 3A), we tested another cold-induced transcript, COR15A, which was also reduced in AtGRP7-ox plants (Additional file 3). Transcripts encoding the plant defensins PDF1.1 and PDF1.2a, RAP2.3/ERF72/AtEBP encoding a member of the ethylene response factor subfamily B2 of AP2-domain transcription factors implicated in PDF1.2a activation [37,38] and the antimicrobial thionin THI2.2 are present at reduced levels in AtGRP7-ox plants (Figure 3A, Additional file 3).

Because methyl jasmonate (MeJ)-induced transcripts were overrepresented among genes with reduced levels in AtGRP7-ox plants, we investigated a potential role of AtGRP7 in the response of the AtGRP7 target gene PDF1.2a to MeJ treatment. PDF1.2a accumulated 24 hrs after addition of MeJ both in wt and independent AtGRP7-ox plants (Figure 3B). PDF1.2a levels in MeJ-treated AtGRP7-ox plants still remained lower than in MeJ treated wt plants. Thus, overexpression of AtGRP7 does not prevent the response to MeJ but MeJ does not overcome the negative regulation by AtGRP7.

We have shown previously that site-specific mutation of a single arginine to glutamine within the RNA recognition motif impairs both in vitro binding of recombinant AtGRP7 to its pre-mRNA and in vivo function [22]. Therefore we investigated the steady-state abundance of selected putative AtGRP7 targets in transgenic plants constitutively overexpressing the mutant protein (AtGRP7-RQ-ox). Real time PCR showed that levels of PR1, PR2, THI2.2 and RD29A remained similar to wt levels in plants harvested at zt12 (Figure 3C). For HYH and STH with a morning peak, plants were harvested at zt3. Again, HYH and STH levels remained similar to wt levels (Figure 3C). The fact that the candidate targets were affected by high levels of the authentic AtGRP7 protein but not of the mutant protein indicates that the regulation is based on the AtGRP7 RNA-binding activity. Nevertheless, this does not unambiguously imply direct binding of AtGRP7 to these transcripts, as overexpression of regulatory RNA-binding proteins, similar to the overexpression of transcription factors leads to direct and indirect effects on the transcriptome. To begin to understand how AtGRP7 may influence the differentially expressed transcripts we assessed their steady-state abundance in the atgrp7-1 T-DNA insertion line that lacks AtGRP7 [4]. PR1 and PR2 transcript levels were reduced in atgrp7-1, suggesting that their expression closely correlates with the AtGRP7 level (Figure 3D).

Also HYH and STH levels were weakly reduced at the time of their circadian maximum. The levels of THI2.2 and RD29A remained mostly unchanged and levels of PDF1.1 and PDF1.2a were reduced in atgrp7-1, suggesting either that elevated AtGRP7 levels have a slightly negative effect but reduced AtGRP7 levels are not sufficient to cause their upregulation or that they are influenced indirectly.

Overall, changes in expression of most candidate target genes were moderate (Additional file 1, Additional file 2). Therefore we subjected the expression data to PAGE (parametric analysis of gene set enrichment) [39], an improved tool to analyze overrepresentation of groups of genes that employs predefined gene sets. It relies on the assumption that differential expression manifests itself more clearly at the level of coregulated genes than at the level of individual genes. Thus, PAGE is a complementary approach to uncover genes differentially expressed with a small fold change (below the cut-off level). Its significance comes from the possibility to detect entire gene sets, with narrower definition than GO categories, that are co-ordinately up- or down-regulated to a small degree.

Most prominent among gene sets upregulated in AtGRP7-ox plants were structural constituents of the ribosome and functions associated with ribosome biogenesis and assembly (Figure 4A, Table 3). The next categories were RNA binding and small nucleolar ribonucleoprotein complexes, followed by rRNA processing, nucleolus and RNA splicing including the SR (serine arginine rich) proteins RSZ22 and RSZ32. This may point to an involvement of the RNA-binding protein AtGRP7 in the modulation of RNA processing and translational activity. Two transcripts encoding proteins participating in pre-mRNA splicing, the snRNP core protein D1 and the U5 snRNP helicase (Additional file 1), were confirmed to be expressed at higher levels in independent AtGRP7-ox lines (data not shown). Transcripts with reduced abundance in AtGRP7-ox comprise functions associated with chloroplasts (Figure 4B, Table 4). This may relate to the observation of a slightly reduced chlorophyll content in AtGRP7-ox plants (Streitner, unpublished).

The genotypes used were C24, RS13 (AtGRP7-ox in C24 background) [20], Col, D and G (AtGRP7-ox in Col background) [22], AtGRP7RQ-ox [22] and the T-DNA insertion line atgrp7-1 [4,6]. Seeds were surface-sterilised, stratified at 4°C for two days, germinated and grown on half-strength MS medium [54] supplemented with 0.5% sucrose and 0.5 g MES/l in long days (16-hr light/8-hr dark cycles) at 20°C. After about ten days seedlings of comparable size were transferred to MS plates without sucrose.

For JA treatment, seeds were germinated in liquid half-strength MS medium supplemented with 0.5% sucrose and 0.5 g MES/l and incubated on a rotary shaker in long days. After 12 days 50 μM MeJ was added and control samples were treated with 0.2% ethanol. Plants were harvested after 24 h.

The aerial part of plants with about eight true leaves were harvested at zt12, the time of the circadian maximum of AtGRP7 transcript abundance. RNA was isolated using Trizol. Total RNA was treated with DNaseI and further purified using the RNeasy kit (Qiagen, Hilden, Germany).

Synthesis of cDNA and biotinylated cRNA were performed as recommended by Affymetrix (Santa Clara, USA). Total RNA (20 μg) was used to prepare cDNA with SuperscriptII Reverse transcriptase (Invitrogen) according to the manufacturer's instructions with an oligo(dT)₂₄-T7 oligonucleotide (GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG(dT)₂₄). The cDNA was subjected to in vitro transcription in the presence of 2 mM of each biotin-11-CTP and biotin-16-UTP (ENZO Life Sciences, Farmingdale, NY) with the MegaScript High Yield Transcription Kit (Ambion, Austin, TX). After purification of the cRNA on RNeasy columns (Qiagen, Hilden, Germany), 15 μg of cRNA was fragmented in a volume of 40 μl, denatured for 5 min at 99°C and hybridized to the arrays for 16 h. Washing and detection of labelled cRNA with streptavidin-phycoerythrin were performed according to the manufacturer's instructions. The arrays were scanned using Affymetrix 3000 7G confocal scanner. Affymetrix Arabidopsis ATH1 GeneChips(r) were used throughout the experiment (Affymetrix, Santa Clara, CA). The exact list of probes present on the arrays can be obtained from the manufacturer's website http://www.affymetrix.com. Analysis was based upon annotations compiled by TAIR (http://www.arabidopsis.org, version 2007-5-2).

Raw data were processed with MAS (Microarray suite) 5.0 from Affymetrix. Signal values were derived from Affymetrix *.cel files using GCRMA [55]. All data processing was performed using the statistic language R (version 2.6.2) that is freely available at http://www.r-project.org/[56].

Coefficients of variation (cv) between replicates as a quantitative measure of data quality and consistency between replicates were calculated as described previously [57]. Differentially expressed genes were identified using the RankProd package in R [34] that inherently corrects for multiple testing. Probe sets were called significantly differentially expressed when q < 0.05. To enrich for biologically relevant changes, only probe sets with a minimal fold change of 1.5 were selected. Differentially expressed genes were grouped into collapsed functional gene ontology categories (GOSLIM, obtained from http://www.arabidopsis.org). Furthermore, enrichment of detailed GO categories (obtained from TAIR) was tested. In this case, multiple-testing correction was according to [58] with a critical p-value of 1E-2. Grouping according to preferred phase of circadian expression was based on data from Edwards et al. [35]. Grouping according to phytohormone regulation was based on data from Nemhauser et al. [59]. The significance of enrichment was estimated based on the hypergeometric test and multiple-testing correction according to Bonferroni. PAGE (parametric analysis of gene set enrichment) was performed as described [39] using a critical p-value of 1E-6 after multiple testing correction according to Benjamini and Hochberg [58].

Isolation of total RNA and hybridization of RNA gel blots were performed as described [60]. For semiquantitative RT PCR, retrotranscribed RNA was amplified with Taq Polymerase. To determine the linear range of amplification for each primer pair, samples were withdrawn after 24, 26, 28, 30, 32 and 34 cycles. PCR products were separated on agarose gels and either visualized by Ethidium-bromide staining or transferred to a nylon membrane and hybridized with radiolabeled cDNA probes.

For Real time PCR, duplicate samples were analysed in a MJ research Opticon DNA Engine. Total RNA was treated with DNaseI and reverse-transcribed using Superscript II (Invitrogen). 20 ng of retrotranscribed RNA was amplified with the Eppendorf Real MasterMix kit using an initial denaturation step of 2 min, followed by 45 cycles of 20 sec at 94°C, 30 sec at 60°C and 40 sec at 68°C. C_T values were determined and relative expression levels for the analyzed transcripts were calculated based on non-equal efficiencies for each primer pair [61,62]. Data were normalized to transcripts encoding the translation initiation factor eIF-4A-1 (At3g13920), PTB (At3g01150) and PPR (At5g55840) [63]. Shown are the mean expression levels +/- s.d.. The absence of amplification products from genomic DNA was confirmed in non-retrotranscribed controls. Primers are listed in Additional file 5.