Hypoxia upregulates mitochondrial genome-encoded transcripts in Arabidopsis roots

Plants are frequently exposed to limitations in oxygen availability during their lifetime. During evolution, they have developed a number of physiological and morphological adaptations to tolerate oxygen and other stress conditions. These include regulation of growth by gene expression and ATP generation. The regulation of nuclear genes after hypoxia and anoxia is well studied; however, the regulation of mitochondrial genes in response to oxygen stress has not been characterized to date. Therefore, we have established an Arabidopsis mitochondrial genome-specific microarray that accommodates probes for all mitochondrial DNA-encoded genes and conserved open reading frames. Our analysis showed an up-regulation of mitochondrial transcripts in Arabidopsis roots after 48 h of hypoxia. Since no significant difference was detected in the expression of mitochondrial RNA polymerases or the mitochondrial DNA content per cell, we propose a transcriptional mode of induction of mitochondrial gene expression under hypoxia.


INTRODUCTION
Plants exhibit remarkable variation in their ability to tolerate low oxygen stress, including both hypoxia (a decline in oxygen availability) and anoxia (the complete absence of oxygen).Generally, limitations in oxygen availability result from a reduced supply of oxygen from the environment (Bailey-Serres and Voesenek, 2008) or as a consequence of diffusional restrictions in oxygen uptake (Armstrong et al., 2009).To tolerate oxygen stress, plants have developed numerous morphological and physiological adaptations (Kende et al., 1998;Drew et al., 2000;Bailey-Serres and Voesenek, 2008), in which they either elongate their internodes and leaf blades to reestablish gas exchange (Kende et al., 1998;Pierik et al., 2009) or limit their growth to save carbohydrate and energy reserves until oxygen becomes available again (Hinz et al., 2010).Plant responses to oxygen stress also include differential expression of a core set of genes (Klok et al., 2002;van Dongen et al., 2009;Mustroph et al., 2010).Genes whose transcripts are up-regulated are chiefly required to meet the energy needs of the cell through glycolysis, such as alcohol dehydrogenase, pyruvate decarboxylase and alanine aminotransferase (Ismond et al., 2003;Rocha et al., 2010).In contrast, genes whose expression is down-regulated are associated with ATP-consuming processes such as transport, signaling, lipid metabolism, secondary metabolism and redox regulation (Klok et al., 2002;Ismond et al., 2003;Loreti et al., 2005;van Dongen et al., 2009;Rocha et al., 2010).
As mitochondria are the sites of cellular respiration, ATP production and free radical generation, limitations in oxygen availability have a major impact on mitochondria (Logan, 2006).Given the potential threat from mitochondria during oxygen deprivation, cells execute stress responses that include changes in mitochondrial biogenesis and gene expression (Branco-Price et al., 2005;Loreti et al., 2005;van Dongen et al., 2009).Studies to date have focused mainly on exploring the response of nuclear-encoded genes to anaerobic conditions, and have revealed most dramatic changes in the expression of genes associated with processes taking place within the mitochondrion (Loreti et al., 2005;van Dongen et al., 2009).Conversely, mitochondrial-encoded genes which could also play a role in tolerating anaerobic conditions have not yet been explored.Therefore, to investigate changes in the steady-state transcript levels of mitochondrial-encoded genes, we have established an oligonucle-otide-based mitochondrial genome-specific microarray.This array contains probes for all the genes and conserved open reading frames of the Arabidopsis thaliana mitochondrial genome.For expression analysis, Arabidopsis seedlings were grown on vertical plates and subjected to 48 h of hypoxia (4% O 2 ).This analysis revealed significant up-regulation of mitochondrial transcripts in response to hypoxia.Since no change either in mitochondrial DNA content per cell or in the expression of mitochondrial RNA polymerases was detected, we propose that transcriptional control of mitochondrial gene expression occurs under hypoxia.

MATERIALS AND METHODS
Plant material and growth conditions Arabidopsis thaliana ecotype Col-0 seeds were surface-sterilized, washed 3-4 times with sterile deionized water and airdried, and were then grown in Petri dishes containing Murashige and Skoog (MS) medium (Sigma-Aldrich, Steinheim, Germany) supplemented with 1% sucrose and 0.7% Phyta agar.After 72 h of stratification at 4 °C, the seedlings were transferred to a growth chamber having a 12-h/12-h light/dark cycle, 200 μE m -2 s -1 light intensity and 22 °C/18 °C day/night temperature.After germination, the seedlings were layered onto the surface of vertical plates containing MS medium supplemented with 1% sucrose and 1.5% Phyta agar.The high percentage of agar was used to prevent the roots from penetrating into the agar medium, thus enhancing oxygen diffusion toward the roots (Supplementary Fig. S1).
Hypoxia treatment Five weeks after germination, vertical plates containing the seedlings were vertically stacked into a plastic box (60 cm length, 40 cm width and 18 cm height).Starting at the beginning of normal night, these boxes were aerated with normal air (21% oxygen) as a control or low oxygen (4% oxygen, 0.03% carbon dioxide, 96% nitrogen, mixed by a gas flow controller) as a treatment.During the treatment, the gas flow rate was maintained at 5 l/min and the oxygen concentration within the boxes was measured regularly using an oxygen sensor (Precision Sensing, Germany).After 48 h, roots from the control and treated samples were harvested in liquid nitrogen and stored at -80 °C.

Design and immobilization of mitochondrial microarray oligos
An oligonucleotide microarray containing all genes and conserved open reading frames of the A. thaliana mitochondrial genome was designed using published sequences (Unseld et al., 1997;Sugiyama et al., 2005).The length of each oligo was 70 nucleotides and the GC content was as close to 50% as possible.The final oligonucleotide sequences were also adjusted, if necessary, for RNA editing at defined sites (Table 1, bold underlined letters) (Unseld et al., 1997;Sugiyama et al., 2005).A total of 60 oligonucleotides were selected (Table 1).To ensure reliable quantification and comparability between arrays, negative controls (Neg1, Neg2) and a series of reference and calibration DNAs (Ratio1-8, Utility1-3 and Calib1-10) were also printed on the array.These reference DNAs covered a range of 1 pg to 10 ng of RNA for quantification.For spotting onto the microarray, HPLC-purified oligos and reference DNAs (Lucidea Universal ScoreCard, GE Healthcare) were diluted in 1x Spotting buffer (ArrayIt).Ten microliters of each oligo and reference DNA was pipetted into a 386-well microtiter plate following the printing scheme (Table 2).A Piezoarray robot device (PerkinElmer) was used for printing 100 pg of each oligonucleotide onto SuperEpoxy 2 DNA Substrates (ArrayIt).Each array pattern was printed twice on each glass slide to cover any technical imperfections during array handling.After spotting, the slides were dried at room temperature for 2-3 h in the spotter chamber.Spotted slides were then stored in a desiccator at room temperature.

cDNA synthesis for microarray analysis
To synthesize cDNA for microarray hybridizations, a Super-Script Indirect cDNA Labeling Kit (Invitrogen) was used.Briefly, 10 μg of total RNA was mixed with 5 μg of random hexanucleotide primers and 2 μl of spike RNAs (Lucidea Universal ScoreCard).The mixture was heated at 70 °C for 5 min and immediately placed on ice.For the incorporation of aminoallyl-and aminohexyl-modified nucleotides into the cDNA, 5x first-strand buffer, 0.1 M DTT, dNTPs mix (including amino-modified nucleotides), RNaseOUT (40 U/μl) and Superscript III-RT (400 U/μl) were mixed and placed at 46 °C for 3 h.The cDNA was purified by filtering through S.N.A.P. columns (Super-Script Indirect cDNA Labeling Kit) followed by ethanol precipitation.The cDNA was suspended in 5 μl of 2x coupling buffer.For Cy3 dye labeling, cDNA was mixed with the dye and incubated in the dark at room temperature for 1 h.The unincorporated dye was removed by filtering through a S.N.A.P. column.The volume of labeled cDNA was reduced by filtering through a Microcon YM-10 filter (Millipore).The labeled cDNA was mixed with 130 μl of microarray hybridization buffer (0.1% SDS, 25% formamide, 5x SSC, 1% BSA, 40 mM Na 4 P 2 O 7 ), denatured at 95 °C for 5 min and immediately added to a microarray slide mounted on the hybridization station (HybArray 12, PerkinElmer).

Microarray hybridization, washing and scanning
Microarray slides were washed with 2x SSC + 0.1% sarcosyl for 2 min, 2x SSC for 2 min, 100 °C dH 2 O for 2 min and ice-cold 100% ethanol for 2 min, and dried by centrifugation at 500 g for 5 min.They were then mounted onto the HybArray 12 station.The slides were first heated to 75 °C for 5 min and then cooled to 45 °C.Cy3labeled cDNA was added to the slides and hybridization was carried out for 15 h, during which the temperature was gradually decreased from 45 °C to 41 °C and then maintained for the next 12 h.After hybridization, nonspecifically bound probe was removed by washing the slides with microarray wash buffers I (2x SSC, 0.1% Nlauroylsarcosine sodium salt), II (0.2x SSC, 0.1% N-lauroylsarcosine sodium salt) and III (0.2x SSC) at 25 °C.
The slides were dried by centrifugation at 500 g for 10 min.Cy3 fluorescence was detected by scanning the slides at 10-μm resolution using an FLA-8000 scanner (Fujifilm).

Microarray data normalization and visualization
After scanning, the resulting images were analyzed using Genespotter software (Microdiscovery).Grids representing the spotting scheme were manually adjusted on the microarray image to ensure the accuracy of data.An example of how we determined the spot intensity is shown in Supplementary Fig. S2.Data obtained after local background correction were calibrated using the factor obtained from the slope of the Lucidea Universal ScoreCard controls (Supplementary Fig. S3).This made it possible to compare data obtained from independently hybridized arrays.The data thus obtained were multiplied by the difference in median of biological replicates.The mean of these biological replicates was calculated and expressed as relative expression per microgram of total RNA.Microsoft Excel and Sigma Plot were used for data analysis and visualization.

Isolation of nucleic acids
Total genomic DNA was extracted using the CTAB method (Doyle and Doyle, 1990).In short, frozen root material was ground using a steel ball mill (Retsch MM301) under liquid nitrogen, and then suspended in CTAB buffer (2% CTAB, 1.4 M NaCl, 20 mM EDTA, 100 mM Tris-HCl, pH 8.0, 100 mM βmercaptoethanol).The mixture was incubated at 65 °C for 30 min.Cellular debris was removed by centrifugation at 12,000 g for 10 min.DNA was purified by chloroform/isoamyl alcohol extraction followed by isopropanol precipitation (15,000 g, 30 min, 4 °C).To prepare DNA free of RNA, isolated DNA was treated with RNaseA at 37 °C for 15-30 min, followed by phenol/chloroform/ isoamyl alcohol extraction and ethanol precipitation.The final pellet was dissolved in DNase-free water.For RNA isolation, PeqGold TriFast (100-200 mg/1.5 ml) was added to the homogenized root material and mixed for 15 sec.The suspension was incubated for 3-10 min at room temperature and centrifuged at 12,000 g for 10 min at 8 °C.
The upper aqueous phase was collected and transferred to a fresh tube.RNA was precipitated with 0.5 volumes of isopropanol and collected by centrifugation at 12,000 g for 10 min at 4 °C.The pellet was washed with 70% ethanol and dissolved in RNase-free water.
Northern blot analysis For Northern blot analysis, 5 μg of total RNA was mixed with RNA loading buffer (60% formamide, 18% formaldehyde, 1.5x MOPS buffer, pH 7.0, 0.1 μg/μl ethidium bromide, 0.1% xylene cyanol, 0.1% bromophenol blue) and separated by electrophoresis in a formaldehyde-containing 1% agarose gel.The RNA was then transferred to Hybond nylon membranes (Amersham) matR rrn26 trnQ(uug) orf111c ---Hypoxia up-regulates root mitochondria transcripts by capillary blotting using 20x SSC buffer (3 M NaCl, 0.3 M tri-sodium citrate dihydrate, pH 7.0).The immobilized RNA was covalently bound to the membrane by UV irradiation (0.12 J/cm 2 , wavelength ~302 nm).To prepare gene-specific probes for hybridizations (Table 5), the Megaprime DNA Labeling System (GE Healthcare) was used.Briefly, [α-32 P]dCTP was incorporated into the PCR-amplified fragments by Klenow DNA polymerase.To remove unincorporated radioactive nucleotides, the mixture was filtered through a Sephadex-G50 column provided with the Megaprime DNA Labeling System.For hybridization, the membranes were incubated for ~1 h at 65 °C in Church buffer (1% BSA, 0.5 M Na 2 HPO 4 , 7% SDS, 1 mM EDTA, pH 8.0).The 32 P-labeled PCR probes were denatured at 95 °C for 5 min and immediately added to hybridization tubes containing the membranes.The tubes were incubated for 12 h at 65 °C, after which the membranes were washed once with Northern wash buffer I (2x SSC, pH 7.0, 0.1% SDS) and twice with Northern wash buffer II (0.5x SSC, pH 7.0, 0.1% SDS) for 15 min each at 65 °C.The membranes were exposed to storage Phosphor Screens (GE Healthcare) and hybridization signals were visualized using a Typhoon Trio (GE Healthcare) scanner.

Real-time quantitative PCR analysis
Real-time quantitative PCR was used to amplify and simultaneously quantify targeted DNA molecules.PCR was performed in an optical 384-well plate using an ABI PRISM 7900 HT Sequence Detection System (Applied Biosystems).The resulting data were analyzed using SDS 2.3 software (Applied Biosystems).For expression analysis, CT values for alcohol dehydrogenase 1 (ADH1), pyruvate decarboxylase 1 (PDC1) and mitochondrial RNA polymerase genes (RpoTm and RpoTmp) were normalized to the average CT value of the housekeeping genes ubiquitin-10 (UBQ10) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Table 3).The relative expression between treated and non-treated samples was obtained from the formula (1 + E) ΔΔCT , where ΔΔCT represents ΔCT treated minus ΔCT non-treated and E is the PCR efficiency.In experiments where the mitochondrial DNA content per cell relative to nuclear DNA was determined, CT for mitochondrial genomic regions (nad3, cox1, atp1, rpl5, rps4 and matR) was normalized to the CT of the nuclear RPOTm gene to obtain the ΔCT value (Table 4).The ΔCT values thus obtained were negative; therefore, the 40-ΔCT method was used to view the relative content of mitochondrial DNA per cell.

Design of an Arabidopsis mitochondrial genomespecific microarray
In this study, we have designed an oligonucleotide-based microarray covering all genes and conserved open reading frames of the A. thaliana mitochondrial genome (Unseld et al., 1997;Sugiyama et al., 2005).In total, 60 oligos were selected with 70 bp length and ~50% GC content (Table 1).For 53 genes, we were able to select a discrete oligo without an editing site(s) included, but for nad3, atp9, ccb203, ccb206, ccb256, ccb382 and ccb452 it was not possible to avoid an editing site(s); therefore, for these genes the final oligo sequences were selected after adjusting for RNA editing at the defined position(s) (Table 1, underlined bold letters) (Unseld et al., 1997;Sugiyama et al., 2005).Furthermore, because the sequences of trnM(cau), trnS(gga) and trnW(cca) encoded by the mitochondrial genome were found to be highly similar (93%) to those of plastidencoded trnM(cau), trnS(gga) and trnW(cca) (Duchene and Marechal-Drouard, 2001), the data for these tRNAs, although presented here, should be treated with caution.First, hybridization conditions were optimized to obtain a linear relationship between signal intensity and RNA concentration (Supplementary Fig. S3).For this, we printed Lucidea Universal ScoreCard controls on the arrays.These include negative controls (Neg1, Neg2) and a series of reference and calibration DNAs (Calib1-10).During analysis, we observed no signals for the negative controls, while detectable signals were obtained for all mitochondrial-encoded genes.The signal intensities of the calibration spots were then plotted against the corresponding RNA amounts to generate a calibration curve (Supplementary Fig. S3).The factor thus obtained was used to normalize the signal intensities from various slides.This helped to reduce the variation between samples resulting from differences in the efficiency of cDNA synthesis and/or hybridization.The data showed that the signal intensity is linearly correlated with the amount of RNA template.The number of array blocks per slide was also optimized to control differences in signal intensity resulting from any technical variation.To do so, the first set of spotting and hybridization experiments was performed with only one array block spotted per slide.The data obtained were highly variable due to an insufficient number of technical replicates per slide.To overcome this, we increased the number of array blocks per slide from one to two, which gave highly reproducible signals from all the oligos spotted on the array.

Expression analysis of hypoxia-responsive genes
The expression of known anaerobic genes [alcohol dehydrogenase 1 (ADH1) and pyruvate decarboxylase 1 (PDC1)] was tested to determine if Arabidopsis seedlings had indeed experienced low-oxygen stress.The analysis showed significant induction in the expression of both ADH1 (~2.5-fold) and PDC1 (~5-fold) (Fig. 1).This confirmed that the Arabidopsis seedlings had experienced low-oxygen stress and could be used for mitochondrial gene expression analysis.

Mitochondrial gene expression under hypoxia
To examine the effect of low oxygen on mitochondrial transcript levels in Arabidopsis roots, five-week-old seedlings were treated with normal air (control) and air containing only 4% oxygen (hypoxia).The analysis showed an up-regulation of mitochondrial transcripts, with most significant changes in the expression of oxidative phosphorylation, ribosomal protein, ribosomal RNA and cytochrome biogenesis genes (Fig. 2).However, for nad2, nad4L, nad7, cox3, atp6-1, atp9, rpl5, rps4, orfs (except orf106b) and the tRNAs [except trnC(gca), trnE(uuc), trnS(uga) and trnY(gua)], the change in transcript levels was not significant.To confirm the microarray data, cox1, atp9 and rps3 transcript levels were also analyzed by Northern blotting (Fig. 3).The mRNA accumulation patterns detected in Northern blots corresponded to those observed in microarray analyses, confirming that the microarray reliably detected changes in the transcript levels of mitochondrial-encoded genes.
Mitochondrial DNA content per cell Mitochondrial DNA copy number per cell is known to vary between leaves, flowers, stem and roots in Arabidopsis (Preuten et al., 2010).Therefore, we investigated if a relationship exists between mitochondrial transcript levels and the mitochondrial DNA content per cell in Arabidopsis roots.Fig. 1.Relative expression of hypoxia-responsive genes in Arabidopsis roots.To quantify the expression levels of alcohol dehydrogenase 1 (ADH1) and pyruvate decarboxylase 1 (PDC1) genes, ubiquitin-10 (UBQ10) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) expression was used as a reference.Values presented here are calculated using the 2 -ΔΔCT method.Standard deviations were calculated from three biological replicates.One-way ANOVA (P < 0.05 followed by Tukey's test) was used to test the significance of differences between the hypoxia-and air-treated samples.Significant changes are marked by an asterisk (*).
To quantify mitochondrial DNA, gene-specific primers were used against nad3, cox1, atp1, rpl5, rps4 and matR genes.We observed that comparable amounts of mitochondrial DNA were present in hypoxia-treated and control samples (Fig. 4).This indicates that the observed induction of mitochondrial transcripts under hypoxia is not caused by an increase in mitochondrial DNA content per cell.

Transcript analysis of RNA polymerases
In Arabidopsis, three phage-type RNA polymerase genes are expressed in the nucleus: RPOTm (mitochondria), RPOTmp (mitochondria and plastids) and RPOTp (plastids).The RNA polymerases localized in mitochondria (RPOTm and RPOTmp) are known to execute all steps of transcription, including promoter recognition, initiation and elongation, as a single-polypeptide enzyme (Kuhn et al., 2007).The expression levels of mitochondrial RNA polymerases were recorded to see if they correlated with those of mitochondrial transcripts.We found no significant difference in the transcript levels of RPOT genes between hypoxia-treated and control samples (Fig. 5).Changes in mitochondrial transcript levels are therefore independent of RPOTm and RPOTmp expression levels.

DISCUSSION
Mitochondrial microarray In the current study, we have developed an Arabidopsis mitochondrial genomespecific microarray to study changes in steady-state transcript levels.There were two reasons to establish our own microarray platform: mitochondrial genome-specific microarrays are not available commercially, and the Affymetrix array cannot be used for this purpose because it contains few mitochondrial genome-specific oligos (Loreti et al., 2005;van Dongen et al., 2009).Furthermore, Affymetrix reagents include oligo(dT) primers for cDNA synthesis; the poly(A) tail contributes to mRNA stability for nuclear genome-encoded transcripts, but acts as an mRNA degradation signal in plant mitochondria (Gagliardi and Leaver, 1999;Lupold et al., 1999;Gagliardi et al., 2001).Therefore, the Affymetrix array cannot be used to investigate changes in the expression of mitochondrial genes.
The designed microarray contained 60 oligos covering all the protein-coding genes and conserved open reading frames of the Arabidopsis mitochondrial genome.The length of each oligo was fixed as 70 bases, because arrays with 70-base oligonucleotides display up to four-fold higher sensitivity than those with 50-or 60-base oligonucleotides; furthermore, 70-mer oligonucleotide arrays can produce hybridization signals comparable to PCR amplicon-based arrays (Kane et al., 2000;Bozdech et al., 2003;Rhee et al., 2004;He et al., 2005aHe et al., , 2005b)).Therefore, to detect mitochondrial transcripts at the highest possible sensitivity, we used 70-mer oligonucleotide probes for our microarray.
The 60 oligos were designed to avoid RNA editing sites wherever possible, to detect both primary and posttranscriptionally modified transcripts.Since the number of editing sites is very high (> 500) in Arabidopsis mitochondria (Bentolila et al., 2008), for seven genes (nad3, atp9, ccb203, ccb206, ccb256, ccb382 and ccb452) it was not possible to select an oligo that did not include at least one RNA editing site (Table 1; bold underlined letters).For these genes, the oligos were selected after modification of the editing site(s) because proteins arising from unedited transcripts do not assemble into mitochondrial multi-subunit complexes (Begu et al., 1990;Grohmann et al., 1994;Lu and Hanson, 1994).

Mitochondrial gene expression under hypoxia
To investigate changes in mitochondrial transcripts, Arabidopsis seedlings were subjected to hypoxia treatment for 48 h.We chose hypoxia over anoxia (complete absence of oxygen) because the latter represents an extreme stress condition from which plants do not usually recover (Ellis et al., 1999).Moreover, after prolonged anoxic treatment it becomes very hard to distinguish changes caused directly by oxygen stress and those arising as a severe side-effect of oxygen deprivation.Furthermore, few genes are differentially expressed in response to anoxia as compared to hypoxia (van Dongen et al., 2009).We therefore selected hypoxia over anoxia to record changes in mitochondrial steady-state transcript levels.
Many mitochondrial transcripts were significantly induced in response to hypoxia treatment (Fig. 2).Whether the significant induction in mitochondrial Fig. 4. Mitochondrial DNA content per cell.Mitochondrial DNA content per cell was determined by calculating the difference between the CT of selected mitochondrial (nad3, cox1, atp1, rpl5, rps4 and matR) genomic regions and that of the nuclear RPOTm genomic region.RPOTm was used as a reference because it is a single-copy gene in the Arabidopsis nuclear genome.Bars in the figure represent standard deviations from three biological replicates.Significance was tested using oneway ANOVA (P < 0.05 followed by Tukey's test).None of the differences between treated and untreated samples were significant.Fig. 5. Expression analysis of mitochondrial RNA polymerase genes in Arabidopsis roots.The relative expression of RPOTm and RPOTmp was determined using quantitative real-time PCR.Data were normalized to cytoplasmic UBQ10 and GAPDH levels as internal controls.Gene expression is presented here as 2 -ΔΔCT values, and data are representative of three biological replicates.The significance test was conducted using one-way ANOVA (P < 0.05 followed by Tukey's test).Neither of the differences between treated and untreated samples were significant.mRNA accumulation after hypoxia is caused by enhanced mRNA stability or decreased degradation remains to be established.However, from our data it is clear that the induction of mitochondrial gene expression is transcriptional in nature, since it cannot be attributed to changes in the mitochondrial DNA content per cell or in the expression levels of mitochondrial RNA polymerases (Figs. 4 and 5).
It has been shown repeatedly that anaerobic genes as well as nuclear genes associated with respiration are strongly induced under hypoxia (Branco-Price et al., 2005;Loreti et al., 2005;van Dongen et al., 2009).Since none of the three nuclear-encoded organellar RNA polymerases are induced by low oxygen in Arabidopsis (Branco-Price et al., 2005, 2008;van Dongen et al., 2009) and no change in mitochondrial DNA copy number per cell was observed (Fig. 4), the induction in mitochondrial transcripts favors a transcriptional mode of activation of mitochondrial gene expression under hypoxia.The consensus resulting from previous studies and from ours is that limited oxygen availability leads to decreased oxygen consumption, depression in ATP-consuming biosynthetic processes and induction of nuclear and mitochondrial genome-encoded genes (Loreti et al., 2005).This coinduction of mitochondrial and nuclear genes under hypoxia leads us to propose that a shared regulatory mechanism exists for the regulation of both genomes.Moreover, re-entry of oxygen into hypoxic tissue is known to produce harmful oxygen radicals and toxic oxidation products (Biemelt et al., 1998;Branco-Price et al., 2008).The up-regulation of mitochondrial transcripts under hypoxia may be part of the antioxidant defense mechanism to prevent oxidative damage during re-oxygenation.

CONCLUSIONS
Taken together, our data reveal a general up-regulation of mitochondrial transcripts in response to hypoxic stress.The observed overall change in mitochondrial transcripts may be part of the complex signaling pathways that regulate both nuclear and mitochondrial genomes and/or the antioxidant defense mechanism to prevent oxidative damage during re-oxygenation.More investigations are needed to reveal the components of the shared signal transduction pathway(s) connecting gene expression in mitochondria to that in the nuclear genome.

Fig. 2 .
Fig.2.Mitochondrial RNA abundance in Arabidopsis roots.Mitochondrial transcript levels are presented as RNA abundance per μg of total RNA.Open columns represent samples treated with air as a control and filled columns represent samples treated with hypoxia (4% oxygen).These data are representative of three biological replicates.To identify significant changes, one-way ANOVA (P < 0.05) followed by Tukey's test was used.Significant changes are marked by an asterisk (*).

Fig. 3 .
Fig. 3. Northern blot analysis to confirm Arabidopsis mitochondrial microarray data.Northern blots were conducted for mitochondrial cox1, atp9 and rps3 genes to confirm mitochondrial transcript levels observed through microarray analysis.The analyzed samples are indicated above each lane.Equal loading of RNA was checked by nuclear 18S rRNA (Nucl-rrn18) hybridization.

Table 1 .
List of Arabidopsis mitochondrial genome-specific oligos for microarray analysis Bold underlined Ts in the oligonucleotide sequences for nad3, atp9, ccb203, ccb206, ccb256, ccb382 and ccb452 represent RNA editing sites in these oligos.

Table 2 .
Arabidopsis mitochondrial microarray printing scheme

Table 3 .
List of primers used for qRT-PCR analysis

Table 4 .
List of primers used to determine mitochondrial DNA content per cell