Introduction

Auxinic herbicides have been widely used over the past few decades to control dicot weeds in cereal crops and are grouped into different classes based on their chemical nature: phenoxycarboxylic acids [e.g. 2,4-dichlorophenoxyacetic acid (2,4-D) and 4-chloro-2-methylphenoxyacetic acid (MCPA)], benzoic acids (e.g. dicamba), pyridines (e.g. picloram) and quinoline carboxylic acids (e.g. quinclorac and quinmerac). Auxinic herbicides, as their name suggests, mimic auxins. At low concentrations they induce cell division and elongation while at higher concentrations they have inhibitory effects on growth and development (Sterling and Hall 1997; Grossmann 2000; Zheng and Hall 2001). Some of the abnormalities observed with increasing concentrations of auxins and auxin-like herbicides include epinasty, leaf abscission, and abnormal elongation of root and aerial structures leading to senescence (Grossmann 2000). These features of auxin overdose have led to the adoption of synthetic auxins as herbicides in agriculture (Cobb 1992; Sterling and Hall 1997; Grossmann 1998).

Although the phenotypic response to auxin-like herbicides is well characterised, the biochemical and molecular basis of their action is less well understood. Recent studies have proposed that ethylene is induced in response to auxin-like herbicides (Grossmann 2000; Zheng and Hall 2001), and that ethylene in turn triggers abscisic acid (ABA) biosynthesis (Grossmann and Hansen 2001). This induction of ethylene is due to increased expression of the gene encoding 1-aminocyclopropane-1-carboxylic acid (ACC) synthase, which catalyses the rate limiting step in ethylene biosynthesis (Yang and Hoffman 1984; Woeste et al. 1999). Studies in cleavers (Galium aparine L.) support the proposal that ABA biosynthesis is triggered in response to auxin-induced ethylene biosynthesis (Hansen and Grossmann 2000), and this increase in ABA biosynthesis is mediated by 9-cis-epoxycarotenoid dioxygenase, a key regulator of the ABA biosynthesis pathway (Cutler and Krochko 1999; Qin and Zeevaart 1999; Hansen and Grossmann 2000). Further cell damage and death is thought to be due to the synthesis of cyanide, formed as a co-product of ethylene biosynthesis during oxidation of ACC by ACC oxidase (Grossmann 1996).

In summary, a scheme for the mode of action of auxinic herbicides was proposed from research with auxin-sensitive dicot cleavers; a cascade of reactions, initiated by an increase in ethylene levels, mediated via de novo synthesis of ACC synthase results in an increase of ABA levels. This increase in ABA is rapid, occurring within 5 h of root treatment, and that ABA accumulation results in growth inhibition and other morphological abnormalities (Hansen and Grossmann 2000). This phenomenon is sometimes referred to as “auxin overdose” (Grossmann 2000). Detailed studies of kinetics of the auxin response in shoots of treated G. aparine plants indicated increased levels of ACC synthase activity, with corresponding increases in ACC and ethylene 2 h after root treatment with 0.5 mM indole acetic acid (IAA; Hansen and Grossmann 2000) clearly indicating rapid signal transduction and genetic response in the plant after application of IAA to the roots.

In this study we report the results of a holistic approach to understand the effects of auxinic herbicides on plant gene expression. We have chosen to use the ATH1-121501 Affymetrix array to monitor global changes in gene expression. The data evaluate the expression of genes associated with the current model of auxinic herbicides as proposed by Hansen and Grossmann (2000). The data will be used to understand how these genome-wide changes may indicate the onset of senescence and plant death in response to 2,4-D application.

Materials and methods

Seed germination and experimental treatments

Arabidopsis thaliana ecotype Columbia plants were raised from surface-sterilised seeds in vitro on 30 ml half strength Murashige and Skoog medium supplemented with 1.5% (w/v) sucrose and vitamin and solidified with 1% phytagel (w/v) in deep petri dishes. Twelve plants per plate were maintained. Plates were kept in the dark at 4°C for 48 h and were grown in culture rooms at 22°C under a 16 h photoperiod. Plants were raised up to 14 days from day of sowing. At this stage the plants developed four rosette leaves measuring greater than 1 mm in diameter, which represents the principal growth stage 1.04 (Boyes et al. 2001). Plants at this stage were treated with 1 mM 2,4-D (pH 7.0) for a period of 1 h. Control plants were flooded with 1 ml distilled water.

Target preparation and hybridisation for microarrays

Total RNA was extracted from 100 mg plant material using the RNeasy plant mini kit (Qiagen cat. no. 74904). RNA (10 μg) was reverse transcribed to generate the first strand cDNA using Superscript II RT (Invitrogen cat. no. 18090-019) and HPLC purified T7-(dT)24 primer (Genset). Second strand synthesis was continued to generate double-stranded cDNA (ds cDNA). The ds cDNA was purified using the Phase Lock Gel (Eppendorf-5 Prime, cat. no. pl-188233). The purified ds cDNA was in vitro transcribed into cRNA using the BioArray High Yield RNA Transcript Labelling Kit (Affymetrix, Millenium Science, Australia, cat. no. 900182). Biotin label was incorporated during in vitro transcription. Labelled cRNA (20 μg) was cleaned using the RNeasy Plant Mini Kit (Qiagen cat. no. 74904) and fragmented. A hybridisation cocktail was prepared with the fragmented cRNA (target) and was spiked with alignment control oligo B2, eukaryotic hybridisation controls and background control bovine serum albumin (BSA). The target was hybridised on to probe ATH1-121501 for 16 h. The hybridised probe was stained with R-phycoerytherin streptravidin (SAPE, Molecular Probes, cat. no. S-866) and anti-strepatravidin antibody (goat) biotinylated (Vector Laboratories, cat. no. BA-0500). The stained probe was scanned using the Gene Array Scanner. All image and data analysis was performed using the Microarray Suite version 5.0 (MAS 5.0, Affymetrix). Micro DataBase version 3.0 (MicroDB 3.0) served as the interface; data mining was continued using the Data Mining Tool version 3 (DMT 3.0, Affymetrix). For biological duplication two sets of plants were separately treated and RNA extracted from each were separately used to generate cRNA. Results from both the hybridisations were used to verify the reliability of expression results.

Results and discussion

In the experiments discussed below, Arabidopsis plants raised in culture on half strength Murashige and Skoog (MS) medium for 14 days were exposed to 2,4-D by root irrigation to a final concentration of 1 mM. After exposure for 1 h RNA was extracted from treated seedlings and was used to assess changes in gene expression levels. The probe used was the Affymetrix Arabidopsis array ATH1-121501, which enabled a quantitative approach to the examination of changes in all transcripts within the Arabidopsis genome.

Biological duplication

Biological duplication of the treatment (1 mM 2,4-D; 1 h) was used to verify the consistency in expression pattern. The Affymetrix Microarray Suite version 5.0 (MAS 5.0) software was used for analysis. Comparison of array results indicated that 99.2% of the genes that were called present in both duplicates (13,480) showed a variation coefficient of less than 50% (Fig. 1) indicating an extremely high degree of reproducibility between the biological replicates, and scatter plot analysis of the signal values from treatment duplicates further support this, with 80% of probes on both arrays within a twofold difference (Fig. 2). This analysis supports the premise that there was low biological variability between the duplicates.

Fig. 1
figure 1

Variation coefficients of 13,480 genes called present in both hybridisations of 1 mM 2,4-Dichlorophenoxyacetic acid over 1 h

Full size image
Fig. 2
figure 2

Scatter plot of signal values from biological replicates of 1.0 mM 2,4-D represents reproducibility of transcript abundance measurements. Two RNA samples were separately prepared and targets generated from each of the RNA samples were individually hybridised to two arrays. Signal values from array 1 are represented on the x-axis and signal values from array 2 are represented on y-axis. The diagonal lines represent fold changes, which are indicated by number across the lines

Full size image

The selection of reliable signals was carried out in a systematic manner following the procedure outlined by Wang et al. (2003). Signal values were generated based on one-step Tukey’s biweight estimate and detection calls were made based on the Wilcoxon’s signed rank test (Affymetrix 2002). To detect those genes that increased significantly, genes that gave a detection call “P” (present) in both duplicated arrays were chosen. From the P set of genes, those that gave a change call “I” (increase) and a twofold change in both the arrays with average signal values greater than 100 in case of the treatment were considered to have significantly increased (Table 1). To detect genes that decreased in expression those that gave a P (present) call in the control were selected, and those that decreased in both duplicated arrays by a twofold difference and with signal value greater than 100 in the control were considered to have decreased significantly (Table 1). Figure 3 represents a scatter plot of signal values of the control versus average signal of the duplicated arrays of treatment.

Table 1 Selection of genes with reliable expression
Full size table
Fig. 3
figure 3

Scatter plot of control signal values versus average signal of treatment (1 mM 2,4-D; 1 h). Control values on x-axis and treatment on y-axis. Fold change is represented in bold numbers (same as in Fig. 2)

Full size image

Gene expression and functional classification

From a total of 22,810 genes on the ATH1-121501 array, 13,101 (57%) were called P in the control and 13,480 (59%) were called P in both treatments arrays. These percentages and similarity indicated reliable signals. A significant increase or decrease in signal in response to the treatment was indicated by an I (increase) or D (decrease), respectively. Expression results identified that about 1.0% (233) of all of the genes represented on the array were regulated with a greater than twofold change and signal values greater than 100. Data mining identified 148 genes up-regulated (Table 2) and 85 genes down-regulated (Table 3) in response to treatment of the plant roots with 2,4-D (1 mM for 1 h).

Table 2 A representation of genes that were up-regulated
Full size table
Table 3 A representation of genes that were down-regulated
Full size table

Functions were assigned based on automatically derived functional categories maintained by the Munich Information Center for Protein Sequences (MIPS): http://mips/.gsf.de/proj/thal/tables/tables_func_frame.html. Some of the significant categories of the up-regulated genes included cell rescue, defence and virulence, metabolism, transcription, cellular communication/signal transduction, transport facilitation and subcellular localisation (Table 2). Down-regulated genes were grouped into the following major categories: transcription, metabolism, cellular communication/signal transduction, cell fate and cell rescue, defence and virulence (Table 3). Functional classification indicated that 25.3% of the regulated genes did not belong to any known functional category (Fig. 4).

Fig. 4
figure 4

Percentage representation of functional categories of genes regulated by 2,4-D 1 mM over 1 h

Full size image

Analysis of auxin, ethylene and abscisic acid associated genes

Auxins are known to induce a series of responses related to growth and development in plants and genes have been shown to be up-regulated within 5–60 min of exposure to auxins (Abel and Theologis 1996). These early auxin response genes fall into three major categories: auxin/indole acetic acid (Aux/IAA), SAUR (small up-regulated proteins) and GH3 (Hagen and Guilfoyle 1985). Members of the Aux/IAA multigene family are up-regulated by auxins (Reed 2001). Our results indicated that a number of these auxin-related genes were regulated in response to 2,4-D, confirming the auxin-like nature of 2,4-D when applied to the roots of the plants at 1.0 mM concentration. Genes encoding putative auxin-responsive proteins (IAA1), auxin-induced proteins (IAA5, IAA19), and an auxin-regulated protein (IAA13) were up-regulated in response to 2,4-D. However, gene expression of a putative auxin-induced basic helix-loop-helix transcription factor and an auxin-induced protein 10A were down-regulated in this study.

In response to auxinic herbicides and high levels of auxins, ethylene biosynthesis is induced, which in turn triggers the biosynthesis of ABA. ABA subsequently plays a role in adaptive stress responses, but may also trigger growth inhibition and herbicide-like symptoms. It has been observed that in G. aparine, in response to IAA, the activity of ACC synthase and the levels of ACC and ethylene show a rapid but transient increase, with ACC synthase activity showing a maximal activity 3 h post-IAA application with ACC and ethylene levels peaking 5 h post-application (Hansen and Grossmann 2000).

Hansen and Grossmann (2000) observed that the activity of ACC synthase, levels of ACC and ethylene increased but had declined 5 h after treatment with 0.5 mM IAA. Hansen and Grossmann (2000) also observed an increase in ABA levels after 5 h of exposure to 0.5 mM IAA. However, in this study within 1 h of exposure to 1 mM 2,4-D, NCED3 encoding 9-cis-epoxycarotenoid dioxygenase, a key regulator of ABA biosynthesis, was up-regulated but there was no change in expression of genes encoding ACC synthase or ACC oxidase.

Although 2,4-D did not appear to regulate genes involved in ethylene biosynthesis, the results did indicate regulation of genes further along the ethylene signalling cascade. The plant hormone ethylene plays a central role in plant growth, development and general stress response in Arabidopsis. Ethylene in turn is perceived by a group of histidine kinase receptors (ETR1, ETR2, EIN4, ERS1 and ERS2) and downstream to these receptors is a protein kinase constitutive triple response 1 (CTR1). Ethylene receptors and CTR1 are negative regulators of ethylene signalling: in the absence of ethylene the receptors are in an active state, CTR1 binds to the receptor and inhibits the ethylene responses (Gao et al. 2003; Huang et al. 2003). Our results indicted that 2,4-D significantly down-regulated both ERS and CTR1 to the same degree (signal log ratio of −1.20). This down-regulation of ERS and CTR1 indicates that 2,4-D may have triggered an ethylene response, even in the absence of increased expression of genes in the ethylene biosynthesis pathway.

Downstream of CTR1 in the ethylene response cascade there are two groups of transcription factors—ethylene insensitive/ethylene insensitive-like (EIN3/EIL) and the ethylene receptor factor (ERF)—which are activated by ethylene binding. DREB2A/ERF4 and AtERF8, members of the AP2 EREBP family of transcription factors, were up-regulated in response to 2,4-D (1.0 mM; 1 h). ERFs interact with the cis-acting element GCC, found in defence-related genes, and regulate the stress response (Fujimoto et al. 2000). Regulation of ERF in response to biotic and abiotic stress may occur in an ethylene-independent manner and the increase in expression of AtERF4 and AtERF8 in response to 2,4-D might have been due to the onset of stress. Members of the ERF family are known to be differentially regulated in response to ethylene and other abiotic stresses (Fujimoto et al. 2000). AtERF4 is a transcriptional repressor of stress response genes. Both AtERF4 and AtERF8 function as repressors of stress signalling. The up-regulation of AtERF4 and AtERF8 is possibly a result of stress induced by 2,4-D application (Fujimoto et al. 2000).

Hansen and Grossmann (2000) also reported that 5 h after IAA application both xanthoxal and ABA levels show marked elevation, with levels remaining high until 25 h post-incubation (Hansen and Grossmann 2000). The biosynthesis of ABA involves 9-cis-epoxycarotenoid dioxygenase, the enzyme which cleaves 9-cis-xanthophyll and 9-cis-neoxanthin precursors for ABA aldehyde and ABA. Results in this study indicated that NCED3, the gene encoding 9-cis-epoxycarotenoid dioxygenase, was strongly up-regulated in response to root application of 1 mM 2,4-D. This response was rapid, mRNA levels had increased by 1 h post-application, and the signal log ratio for NCED3 was one of the highest observed in the entire data set, indicating significant transcription relative to the control levels. The up-regulation of NCED3 within 1 h of application of 2,4-D may be due to the high concentration of 2,4-D applied, as Hansen and Grossmann (2000) observed an increase in ABA levels only after 5 h of exposure to 0.5 mM IAA. The increased level of transcription of this gene implies that ABA may accumulate beyond normal physiological levels, which may in turn cause a variety of physiological responses leading to cell death and tissue damage, consistent with the auxin-overdose hypothesis proposed by Grossmann (2000). The gene encoding 9-cis-epoxycarotenoid is also known to be regulated in response to other forms of physiological stress (Cutler and Krochko 1999; Qin and Zeevaart 1999) indicating that induction of this gene may be part of an overall stress response by the plant as a result of the high levels of 2,4-D application used in the present study.

Hansen and Grossmann (2000) also confirmed that there was no increase in activity of levels of violaxanthin, neoxanthin and β-carotene (xanthophyll cycle) in response to high levels of auxin application. Our results supported this observation, as there was no significant change in expression of genes involved in the xanthophyll cycle, suggesting that it is the increase in levels of NCED3 which leads to increased ABA formation.

A number of genes encoding ABA-induced proteins (Seki et al. 2002) indicated increased levels of gene transcription in response to 2,4-D (e.g. ABA induced protein phosphatase 2C and protein phosphatase ABI1, which are both involved in signal transduction). Recent studies by Hoth et al. (2002) using mutant abi1 have identified ABI1 as an ABA-responsive gene. Gene ABF3, an ABA-responsive-element (ABRE) binding factor that has been shown to play a role in ABA signalling in response to stress (Kang et al. 2002), was induced in this study. Another ABA-response-element binding factor induced as a response to 2,4-D was the homeobox-leucine ATHB-12 transcription factor, previously induced in response to ABA (Lee et al. 2001). The HVA22 gene, which encodes another ABA and stress-induced protein (Chen et al. 2002) and classified by MIPS as involved in subcellular localisation was shown to be up-regulated by 2,4-D, as was dehydrin Xero2 protein, classified as involved in cell rescue, defence and virulence.

Therefore, the results obtained via analysis of the entire transcriptome of Arabidopsis suggest that exposure to high concentrations of auxinic herbicides (1.0 mM 2,4-D) regulated genes involved in auxin response (IAA1, IAA5, IAA13, IAA19), ethylene signalling (ERS, CTR1, AtERF4, AtERF8), ABA biosynthesis (key regulatory gene NCED3) and ABA signalling and response (ABF3, ABI1). A schematic representation of the key changes in gene expression in Arabidopsis as a response to application of 1.0 mM 2,4-D to plant roots is presented in Fig. 5. Our results identified that 2,4-D down-regulated ERS and CTR1, negative regulators of ethylene response, and up-regulated AtERF4 and AtERF8, repressors of GCC-motif-containing stress response genes. The up-regulation of AtERF4 and AtERF8 may possibly be due to stress. We also found that 2,4-D induced NCED3, a key regulator of ABA biosynthesis. The previously observed increase in ABA levels may also be a response to stress. The study indicated up-regulation of ABF3, associated with ABA signalling and also observed the regulation of ABA response genes.

Fig. 5
figure 5

A schematic representation of gene expression in response to 2,4-D (1.0 mM; 1.0 h). Thick arrows indicate up-regulation and thin arrows indicate down-regulation of genes indicated in the boxes. Dotted arrow indicate the possible path of regulation. AtERF4 and AtERF8 are repressors of GCC-motif-containing stress response genes. Up-regulated genes (+); down-regulated genes (−)

Full size image

In response to stresses such as drought or salinity, it has been observed that plants tend to accumulate the osmolytes proline, glycine betaine or sugar alcohols (Yoshiba et al. 1999). Microarray analysis indicated regulation of genes encoding enzymes involved in osmolyte and osmoprotectant biosynthesis or modulation in response to root application of 2,4-D. The enzyme delta-1-pyrroline-5-carboxylate synthetase had been previously shown to play a role in ABA and salt stress-induced proline accumulation in Arabidopsis (Abraham et al. 2003). The results in this study, consistent with the study by Abraham et al. (2003), indicated that the gene P5CS1 encoding delta-1-pyrroline-5-carboxylate synthetase, the rate limiting enzyme in proline biosynthesis, was up-regulated by 2,4-D. Concomitant with this was the observed down-regulation of a gene encoding proline oxidase. The gene encoding raffinose synthase involved in the synthesis of osmoprotectants, which has been previously identified to be induced during senescence (Buchanan-Wollaston et al. 2003), was up-regulated by 2,4-D. The results indicated an up-regulation of AtPIP5K1 encoding 1-phosphatidylinositol-4-phosphate 5-kinase, also induced under water stress and in response to ABA (Mikami et al. 1998).

The expression results indicated the up-regulation of genes encoding glutathione transferases, known to be involved in detoxification (Prade et al. 1998) and which are also auxin response genes (Abel and Theologis 1996). Cytochrome P450 enzymes (Cyt P450) constitute the largest group of plant proteins that determine herbicide tolerance and selectivity (Werck-Reichhart et al. 2000), and are also involved in the biosynthesis of defence-related compounds, signalling of gibberellins and jasmonates (Maughan et al. 1997) and pigment biosynthesis. Results from this study identified the up-regulation of genes encoding cytochrome P450 and cytochrome P450 monooxygenase (CYP91A2) and the down-regulation of cytochrome P450 90A1 (CYP90A1). CYP90A1 is involved in brassinosteroid biosynthesis (Bancos et al. 2002). Further studies on the role of these cytochrome P450s may suggest an involvement in the mode of action of 2,4-D.

Jasmonic acid (JA) and salicylic acid have well documented roles in plant defence and signalling (Liechti and Farmer 2003). Oxylipin-12-oxophyto-dienoic acid (OPDA) is involved in signal transduction and is a precursor in jasmonic acid biosynthesis, and has been reported to control a number of cellular processes (Schaller and Weiler 1997; Schaller et al. 2000). The three isozymes of 12-oxophytodienoate reductase (OPR1, OPR2, OPR3) are involved in the conversion of OPDA to JA. Brassinosteroids are known to up-regulate the expression at the OPR3 gene (Mussig et al. 2002) and the genes OPR1 and OPR2 are up-regulated in response to wounding and UV light (Biesgen and Weiler 1999). Expression of the gene encoding 12-oxophytodienoate reductase (OPR2) was induced in response to 1.0 mM 2,4-D with 1 h application.

The expression of a number of rescue, defence and virulence-related genes were also regulated in response to 2,4-D. The induction of dehydrin Xero2 gene is consistent with reports of induction of this gene by endogenous ABA (Rouse et al. 1996). The thionin gene, Thi2.2, associated with signalling, cell death and systemic acquired resistance (SAR) pathways (Nibbe et al. 2002) also showed induction after 2,4-D application. Also of note was the induction of ferritin genes, reported to show differential regulation in response to iron as well as a variety of stresses and senescence (Pic et al. 2002; Tarantino et al. 2003). The genes encoding a low temperature and salt responsive protein LT16A, low temperature-induced protein 78 and the heat shock protein HSP70 were also up-regulated as a response to 2,4-D. A gene encoding a leucine-rich repeat family protein involved in disease resistance was also up-regulated in this study. This study is the first to report the up-regulation of resistance/defence genes in response to auxinic herbicides. The results suggest that certain genes regulated by the herbicidal application of 2,4-D are also involved in response to stress. Cheong et al. (2002) described the overlapping gene expression patterns in response to wounding, pathogen, abiotic stress and hormonal response.

2,4-D and senescence

Senescence is induced prematurely in plants as a protective mechanism for survival against various forms of stress. Senescence is one of the mechanisms that plants have evolved for survival: to enhance chances of survival the plant triggers localized cell death. Some of the well-characterized metabolic processes observed to occur during senescence include chlorophyll, protein and lipid degradation (Buchanan-Wollaston et al. 2003). In this study, we have induced plant senescence by herbicidal applications of 1.0 mM 2,4-D. Genes encoding xyloglucan endotransglycosylases (meri5B and TCH4) and others were down-regulated in response to 1.0 mM 2,4-D. Xyloglucan endotransglycosylases catalyse the cleavage of xyloglucan leading to the loosening of cell walls and as a result promote growth. Xyloglucan endotransglycosylases (meri5B and TCH4) were induced by auxins and brassinosteroids (Xu et al. 1996). The gene meri5B, similar to BRU1, was up-regulated in response to brassinosteroids (Mussig et al. 2002) to promote cell elongation and growth. The results indicated down-regulation of meri5B and TCH4 at 1.0 mM 2,4-D, suggesting inhibition of cell growth.

Senescence is also accompanied by cell wall degradation and membrane disintegration (Buchanan-Wollaston et al. 2003). Expansin proteins are associated with cell elongation and wall organisation (Goda et al. 2002) and the results of this study indicated down-regulation of a gene encoding a putative expansin in response to 2,4-D. The genes encoding pectinesterases, xylosidase and beta-xylosidase are involved in cell wall metabolism (Micheli 2001; Goujon et al. 2003) and were down-regulated in this study. At the same time there was up-regulation of genes encoding polygalacturonase, a well-characterized cell wall modifying enzyme (Mahalingam et al. 1999). The results suggest that 2,4-D (1.0 mM; 1.0 h) represses cell elongation and growth.

Lipids are a major component of the cell membrane and are degraded by lipases during senescence (Buchanan-Wollaston et al. 2003). The results indicated up-regulation of a gene encoding lipase, again consistent with previous studies investigating gene expression during senescence (Buchanan-Wollaston et al. 2003). Interestingly the results indicated that a gene encoding lipid transport protein pEARL1 was down-regulated. According to the MIPS classification the pEARL proteins are involved in control of cellular organisation. The herbicide 2,4-D, at 1.0 mM, has triggered the degradation of the membranes. Although 2,4-D (1 mM; 1 h) inhibited growth and triggered cell degradation there was significant down-regulation of the gene encoding the senescence-induced protein sen1 (Oh et al. 1996). Results also indicated the down-regulation of transcription factor WRKY4, which has been reported to be induced during senescence (Chen et al. 2002).

Influence of 2,4-D on signal transduction and transcription factors

The results indicated that 8.4% of the genes regulated by 2,4-D are involved in cellular communication/signal transduction according to the MIPS classification. Genes that were regulated in this category mainly encoded protein kinases, serine/threonine kinases and protein phosphatases. Genes encoding the receptor protein kinase TMK1 and serine threonine protein kinase CTR1 were down-regulated. A possible role of the protein coded by TMK1 in transmembrane signalling has been proposed (Chang et al. 1992). In tobacco, the accumulation of NtTMK1 mRNA was stimulated by methyl jasmonate, wounding, fungal elicitors and other stress-related factors (Cho and Pai 2000). CTR1 is a member of the raf histidine kinase family and a negative regulator of ethylene signal transduction (Gao et al. 2003; Huang et al. 2003).

Auxin and ABA signalling pathways are calcium dependent and calcium-dependent kinases respond to various forms of stress (Poovaiah and Reddy 1993; Yang and Poovaiah 2000). Auxin signalling is calcium/calmodulin dependent, with calmodulin known to play a significant role in auxin signalling (Poovaiah and Reddy 1993). Auxins increase cytosolic levels of calcium and the calcium-calmodulin complex, which then triggers a variety of cellular responses (Yang and Poovaiah 2000). The results of this study indicated a gene encoding a calmodulin-like protein (NaCl-inducible calcium binding) involved in signal transduction was up-regulated. There was also significant up-regulation of genes encoding NAC domain proteins involved in auxin signalling (Xie et al. 2000).

Functional classification indicated that 13.1% of the genes regulated by 2,4-D, encoded transcription factors. It was the second largest functional category to be regulated by 2,4-D. Members of the WRKY-type DNA-binding protein family and basic region/leucine zipper motif (bZIP) transcription factors, previously shown to be involved in pathogen defence and senescence (Eulgem et al. 2000) and stress signalling, were shown by microarray analysis to be regulated by 2,4-D. Genes encoding WRKY-type DNA-binding proteins (WRKY4) were down-regulated. In contrast bZIP transcription factor and ABA-responsive element-binding factor (ABF3) transcription factors were up-regulated, as were C3H type transcription factors (zinc finger proteins, CCCH-type, RING-H2) involved in the lignin and phenlypropanoid pathways (Franke et al. 2002). AP2/EREBP transcription factors, which play a role developmental processes and stress responses to various types of biotic and environmental stress (Riechmann and Meyerowitz 1998), were up-regulated in response to 2,4-D. ATHB-12, and homeobox-leucine zipper proteins (e.g. HAT22) of HB type transcription factors, known to be induced in response to ABA, were also up-regulated in response to 2,4-D. A C2C2 GATA-type family member (GATA4) was down-regulated. Members of this family are characterized by their ability to bind to the light-specific promoter with GATA motifs (Teakle et al. 2002). Our results identified the regulation of transcription factors belonging to the AP2/ERBP, bZIP, C2C2 GATA, C3H, NAC and HB families in response to the 2,4-D treatment.

Conclusion

This is the first report of an investigation of the gene expression pattern across the whole genome of A. thaliana in response to the auxinic herbicide 2,4-D at a herbicidal concentration. The application of 1.0 mM 2,4-D regulated genes known to be involved in the auxin response, ethylene signalling, ABA biosynthesis, signalling and response. The results indicated that genes in both abiotic and biotic stress response detoxification (glutathione transferases, cytochrome P450s) were also modulated. The herbicide down-regulated the expression of genes involved in cell growth and elongation (meri5B and TCH4). Interestingly there was down-regulation of sen1 and WRKY4, known to be induced during senescence.

The data were consistent with the pathway proposed by Hansen and Grossmann (2000). Our results indicated the 2,4-D down-regulated the expression of ERS and CTR1 (negative regulators of ethylene signalling). Furthermore, there was a significant increase in expression of AtERF4 and AtERF8, which may be induced in an ethylene-independent manner due to stress; these genes are also active repressors of stress response genes. The gene expression results not only confirmed that 2,4-D triggered ABA biosynthesis, as indicated by the increased expression of NCED3, but further indicated the regulation of genes involved downstream in ABA signalling (ABF3) and the ABA response.

This study importantly indicated the induction of a number of genes, either as a primary or secondary response to 2,4-D application, which have less clearly defined functions. Many of these genes have not previously been identified to have direct involvement in auxinic herbicide action. As an increased understanding of the regulation of the entire Arabidopsis transcriptome in response to a wider variety of chemical, physical and physiological stimuli becomes available, the exact role of these genes in response to 2,4-D should be elucidated and with this will emerge a clearer understanding of the molecular processes that occur during the plant-herbicide interactions for this widely used compound.