Introduction

The plant cell wall is the world’s most abundant organic resource. Its most obvious function is morphology: it counteracts turgor pressure, forming, shaping and directing growth of the plant on a cellular level as well as the whole individual. Much of the cell wall biomass is derived from a common but cell wall-specific biochemical precursor, UDP-glucuronic acid (UDP-GlcA). In the model plant Arabidopsis thaliana about 50% of the cell wall biomass of mature leaves is derived from UDP-GlcA (Zablackis et al. 1995). The wood of many trees contains large amounts of hemicelluloses, in particular xylan, which is largely derived from UDP-GlcA. Given this importance, the biosynthesis of UDP-GlcA attracts our interest. As initially reported half a century ago, plants have established two pathways for the biosynthesis of UDP-GlcA (compare Fig. 1). It can be directly formed from UDP-glucose by the enzyme UDP-glucose dehydrogenase (UGD). This activity is encoded by a small gene family in Arabidopsis (Klinghammer and Tenhaken 2007). The second pathway starts with the biosynthesis of myo-inositol, which is subsequently converted to glucuronic acid by a unique type of oxygenase: myo-inositol oxygenase (MIOX; E.C. 1.13.99.1). The product requires the activity of glucuronokinase (Pieslinger et al. 2010) and a UDP-sugar pyrophosphorylase (Kotake et al. 2004) to finally end up in the pool of UDP-GlcA. The functionality of the MIOX pathway for plant cell wall biosynthesis has been shown in the past (Kroh and Loewus 1968; Wakabayashi et al. 1989; Loewus and Murthy 2000). After purification of the MIOX enzyme from Cryptococcus (Kanter et al. 2003), we cloned and characterized the corresponding Arabidopsis gene family for MIOX which has four active members.

Fig. 1
figure 1

Biosynthesis of UDP-glucuronic acid and related cell-wall precursors. Two branches lead to the cell wall precursor UDP-glucuronic acid: the myo-inositol oxygenation pathway with MIOX activity as committing step (right), and the nucleotide sugar oxidation pathway depending on UGD activity (left) (Seifert 2004; Kanter et al. 2005; Reiter 2008). ① Inositol-1-P-synthase; ② Phosphoglucomutase; ③ UDP-glucose pyrophosphorylase; ④ UDP-glucuronic acid decarboxylase; ⑤ UDP-arabinose 4-epimerase; ⑥ Inositolphosphate-phosphatase; ⑦ Glucuronokinase; ⑧ UDP-sugar pyrophosphorylase; ⑨ UDP-glucuronic acid 4-epimerase; ⑩ α-Galacturonosyl transferase

Full size image

Exploring public microarray databases (http://www.genevestigator.com) and MIOX-promoter::GUS lines (Kanter et al. 2005), we found a strong developmentally regulated gene expression of the different MIOX isoforms. Two developmental stages, young seedlings and flowers, in particular anthers and pollen, show the highest expression levels for MIOX genes. The role of the MIOX pathway to UDP-GlcA in Arabidopsis cell cultures has been addressed by Sharples and Fry (2007) by feeding radioactive precursors. They found a predominant role of UDP-glucose dehydrogenase for this type of tissue. However, these feeding studies are not feasible for all types of differentiated plant tissues.

MIOX genes can be found in almost all eukaryotes but are generally absent in prokaryotes. One of the few exceptions is the cyanobacteria Nostoc punctiforme, which has a single predicted MIOX gene based on bioinformatic homology searches. The amino acid sequences are highly conserved between species even from different phyla indicating a common evolutionary origin.

The MIOX protein from mouse was recently crystallized to get deeper insight into structure and enzymatic mechanism. The di-iron centre is coordinated by two His-residues for each of the two iron atoms which are additionally bridged by a water and an aspartate molecule (Brown et al. 2006). MIOX was found to have a valence-localized Fe(II)/Fe(III) pair as catalytically competent state, which sets it apart from other oxygenases. The catalytic cycle is unique in that it performs a full four-iron oxidation of the substrate, returning to its resting state without need for an external reductant (Moskala et al. 1981; Xing et al. 2006a, b, c). The terminal electron acceptor is oxygen, which is converted to water.

Myo-inositol is used by plant cells to synthesize a variety of low molecular weight compounds. Phytic acid, the hexakisphosphate ester of myo-inositol, accumulates in seeds as the major phosphor storage compound (Raboy 2001). Inositol also serves as a substrate for the formation of galactinol, the galactosyl-donor that plays a key role in the formation of raffinose family oligosaccharides (RFOs, raffinose, stachyose, verbascose) from sucrose. RFOs accumulate in plants under various stress conditions (Blackman et al. 1992; Kaplan et al. 2004; Peters et al. 2007).

Here we investigate the role of the MIOX enzyme family in knockout/knockdown mutant. To reduce the genetic redundancy of the pathway we have crossed single T-DNA insertion lines for each isoform to double, triple and finally a quadruple knockdown mutant hereafter referred to as the miox1/2/4/5-mutant. This allows the analysis of the function of this enzyme as the activity is strongly reduced or almost absent in the mutant depending on the tissue under investigation. One of the original questions was the contribution of the MIOX pathway to cell wall biosynthesis in plants and whether the enzyme UDP-glucose dehydrogenase could compensate for a loss in MIOX. We assume from our experiments that the latter assumption of complementation by UGD is correct.

Materials and methods

Plant growth

Plants were grown in standard fertilized soil (ED73; Einheitserde; www.einheitserde.de) in a growth chamber (23°C; 150 μmol photons m−2 s−1). Arabidopsis thaliana seeds Col (cultivar Columbia; NASC, University of Nottingham, UK, stock no. N60000) and the vtc1-mutant were obtained from the NASC stock centre. The MIOX quadruple knockdown mutant miox1/2/4/5 was produced from cross-fertilization of the four T-DNA-insertion lines Gabi 450D10 (miox1), Salk 040608 (miox2), Salk 018395 (miox4) and Salk 112535 (miox5). First, two double mutants miox1/2 and miox4/5 were isolated and genotyped to isolate a homozygous lines (Siddique et al. 2009). The miox1/2 mutant was crossed with a homozygous miox4-mutant and homozygous triple mutants miox1,2,4 were isolated. The miox4,5 mutant was crossed with a homozygous miox2 line, selfed, and homozygous triple mutants miox2,4,5 were identified. Both triple mutants miox1/2/4 and miox2,4,5 were finally crossed, selfed, and siblings were characterized by PCR for homozygous T-DNA insertion in all four MIOX genes, resulting in the quadruple mutant miox1,2,4,5.

The CaMV35S::MIOX4 lines were kindly provided by the Nessler group (Department of Plant Pathology, Virginia Tech, Blacksburg, VA, USA) and are described in Lorence et al. (2004). The pgm mutant was a kind gift from Mark Stitt (Max Planck Institute for Molecular Plant Physiology, Golm, Germany).

PCR techniques

DNA was extracted from young leaves and the genetic identity of the plants determined by PCR technique. The mixture was supplemented with 0.6 M betaine (final concentration), otherwise only faint product bands were observed. The thermal profile was as follows: 92°C, 3 min; 32 × [92°C, 15 s; 58°C, 30 s; 72°C, 1 min]; 72°C, 3 min. The primers are given in Suppl. Table S1. T-DNA insertion sites were determined from PCR fragments spanning the insertion site. They were subcloned into pJET 1.2, amplified and sequenced for exact determination of the insertion site.

Quantitative PCR

RNA from different tissues was extracted with the NucleoSpin RNA Plant kit (Macherey-Nagel) and transcribed to cDNA using the RevertAid Moloney murine leukaemia virus reverse transcriptase kit (Fermentas).

For quantification of MIOX and UGD transcripts, qPCR was performed on a Stratagene MX3000 realtime cycler using a SybrGreen method. One reaction (30 μL) consisted of 1× PCR-buffer, a 1:200,000 dilution of SybrGreen stock (Roche), 200 nM primers each, and 1 U Taq polymerase (recombinant wild type). Primers are given in Suppl. Table S2. For amplification of MIOX transcripts, we added betaine to a final concentration of 0.6 M. The housekeeping gene EF1α was employed for internal correction.

The qPCR programme for MIOX cycles through 30 s at 92°C, 30 s at 59°C, and 15 s at 72°C for 40 times; the one for UGD through 30 s at 94°C, 20 s at 58°C, and 30 s at 72°C for 40 times; a conclusive melting curve indicates identity and homogeneity of the product.

For calculations, the reaction efficiencies of the individual wells were computed from the original data employing the LinRegPCR software version 7.4 and raised to the power of the corresponding threshold cycle value. Results were averaged over triplicates and normalized with the calculated amounts of EF1α transcripts.

Feeding of myo-inositol/labelling with 3H-myo-inositol: incorporation in seedlings

Surface-sterilized seedlings were suspended in 0.5 mL one-half-strength Murashige and Skoog medium with 2.5 mM Mes (adjusted to pH 5.6 with KOH), 0.1% sucrose, and 1 μM myo-[2-3H]inositol (specific activity 555 GBq mMol−1; Amersham/GE Healthcare). After 7 days, 25 seedlings (approx. 50 mg fresh weight) were rinsed first in 1 g L−1 inositol, then in water, carefully rid of their seed coats, blotted dry and frozen in liquid nitrogen. The samples were homogenized in a Retsch ball mill and subsequently resolved in 200 μL 0.2 M Na2HPO4, pH 7.4. This crude extract was split into pellet and supernatant by centrifugation (18,000g, 10 min); the pellet was washed several times according to Kanter et al. (2005) to obtain the “cell wall” fraction. The supernatant was mixed with 800 μL acetone and incubated over night at −20°C. Centrifugation for 15 min at 4°C resulted in a supernatant retaining soluble sugars while oligosaccharides and proteins precipitated. The pellet was further treated with α-amylase (Sigma A6380) and amyloglucosidase (Sigma A1602) to degrade starch from the pellet and solubilize glucose units. The enzymes are essentially free of other carbohydrate hydrolases.

Feeding of myo-inositol/labelling with 3H-myo-inositol: incorporation in leaf discs

Leaf discs of 8 mm in diameter were cut out of fully expanded leaves and immediately floated abaxial side down on inositol-free Murashige and Skoog medium (compare Endres and Tenhaken 2009) with 2.5 mM Mes buffered at pH 5.6 (KOH). myo-[2-3H]Inositol (specific activity 555 GBq mMol−1; Amersham/GE Healthcare) and unlabelled myo-inositol were added to a final concentration of 1 and 500 μM, respectively. After 3 h of incubation at room temperature and low light conditions, the discs were rinsed in water, blotted dry, and processed.

When feeding cold inositol, the final concentration of this compound was 1 g L−1 (ten times more than in regular Murashige and Skoog media). After allowing the leaf discs to take up inositol at room temperature and low light conditions for 2 h, they were carefully rinsed in pure water, blotted on filter paper, and transferred to inositol-free medium. Samples were measured at appropriate time points. Control samples were kept in inositol-free medium over the time course of the experiment. Samples were resuspended in Rotiszint eco plus cocktail (Roth) and counted in a Liquid Scintillation Analyzer TRI-CARB 2100TR (Packard).

Fractionation of labelled material

Monosaccharides from the TFA hydrolysate from seedlings grown in liquid one-half-strength MS supplied with 3H-myo-inositol were analyzed on a high performance anion exchange chromatography with pulsed amperometric dectection (HPAEC-PAD) using a Dionex ICS 3000 system, equipped with a CarboPac PA-20 column. The separation method is based on Dionex technical note 20 (http://www.dionex.com/en-us/webdocs/5023-TN20.pdf) with the following modifications: Buffer A, 200 mM NaOH; buffer B, 15 mM NaOH; buffer C, 50 mM NaOH with 500 mM sodium acetate; buffer D, 2 mM NaOH. All sugars (except xylose and mannose) were separated under the following conditions: flow rate 0.45 mL min−1; t0 to t15 100%B, t15 to t25 40%C, 60% B; t25-t28 100%B. The column was regenerated with 100%A for 10 min followed by equilibration with starting buffer B. Simultaneously, fractions were collected every 20 s (~150 μL) and scintillation counted. Xylose and mannose were separated in a separate run using buffer D. Quantification of the sugar signal was done with the Chromeleon software using standard curves (0.1–1 nMol of each sugar).

Metabolite analysis

Soluble sugar extraction

The extraction of soluble carbohydrates from leaves was performed as described previously (Endres and Tenhaken 2009). Seed sugar extraction followed the protocol given in Bentsink et al. (2000). Separation of the samples on a CarboPac MA1 analytical column (Dionex) allows identification of, myo-inositol, and galactinol. For raffinose (and stachyose) measurements, the same samples were loaded on a PA200 column (Dionex). Sucrose was detected on a PA20 column. Concentrations were determined by comparison to authentic standards. Quantification of ascorbic acid and dehydro-ascorbic acid was performed as described previously (Endres and Tenhaken 2009).

The HPLC method for phytic acid extraction described in Talamond et al. (2000) was adapted as follows: 20 mg of seeds was frozen in liquid nitrogen, homogenized in a Retsch ball mill and resuspended in 1 mL 0.5 M HCl. After 10 min incubation at 90°C, the samples were centrifuged for 10 min and 800 μL of the supernatant were transferred to a new tube. Another 114 μL of 7 M HCl were added, the samples were centrifuged for 5 min at 18,000g, and 600 μL of the supernatant were evaporated to dryness. The samples were dissolved in 600 μL water and analyzed on an AS11 analytical column (Dionex) by monitoring conductivity with activated suppressor.

Stress induction and ascorbic acid quantification

Plants adapted to low light (25 μmol photons m−2 s−1) for 3 days were transferred to high light conditions (250 μmol photons m−2 s−1) and the increase in total ascorbic acid content over time was monitored as detailed in Endres and Tenhaken (2009).

Seed mucilage analysis

Dry seeds (10 mg) were incubated in 300 μL 50 mM HCl for 40 min at 80°C. After centrifugation, the supernatant was transferred to a new tube, the pellet extracted with another 50 μL HCl, and the supernatants were pooled. Addition of 15 μL 1 M NaOH partially neutralized the extract prior to precipitation over night at −20°C with 1.2 mL ethanol. The sample was centrifuged for 10 min at 18,000g, the supernatant discarded and the pellet dried. Once the pellet was completely dry, it was hydrolyzed by autoclaving it for 2 h in 250 μL 2 M TFA and 10 μL 0.5% inositol. Once again the sample was dried, re-dissolved in 200 μL water, centrifuged to sediment particles, and prepared for HPLC analysis (see section Feeding of 3H-inositol).

Cell wall samples

Leaf material (100–150 mg) was frozen in liquid nitrogen, homogenized in a Retsch ball mill and suspended in 70% ethanol. Subsequent extractions with methanol/chloroform and acetone as detailed in Kanter et al. (2005) resulted in a cell wall pellet, which was dried and suspended in 800 μL 0.25 M NaAc (pH 4.0) solution. The sample was incubated at 80°C for 20 min, chilled on ice, and adjusted to a pH of 5.0 with 1 M NaOH. An aliquot of 0.01% NaN3 was added prior to incubation with 0.5 mg α-amylase and 0.5 μl pullulanase (Sigma) at 37°C over night. The following day, the sample was boiled for 10 min in a water bath, centrifuged at 18,000g for 5 min, and the supernatant discarded. To remove all residual free sugars, the pellet was washed with 1 mL water for four times. After two more washing steps with acetone, the pellet was dried. The weight of the sample was determined; the hydrolysis in TFA was performed as described above. The dry sample was re-dissolved in 1 mL distilled water per 3 mg of dry sample; 200 μL were diluted with 300 μL water and analyzed via HPLC (see section Feeding of 3H-inositol).

Treatment with methylviologen

Four-week-old plants grown in standard growth conditions were sprayed with 50 μM methylviologen in 0.1% (v/v) Tween 20 and transferred to a growth chamber where they were exposed to ~1,500 μmol photons m−2 s−1 light intensity at 15°C ambient temperature. PSII activity was monitored with a Mini PAM Chlorophyll Fluorometer (Walz, Effeltrich, Germany).

Freezing tolerance

The freezing experiments were conducted as detailed in Rohde et al. (2004). The method is based on an increased ion-leakage from cold-treated leaves, when cells do not exhibit freezing tolerance. We tested batches of plants grown under standard growth conditions in three independent experiments. The resulting data were combined and subjected to a sigmoidal fit performed by the Origin software 6.0 according to the Boltzmann-model. Details of the method are given in the supplement.

Sequence data

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers NP_172904.2 (MIOX1; At1g14520); NM_127538.3 (MIOX2; At2g19800), NM_118759.4 (MIOX4; At4g26260); NM_125047 (MIOX5; At5g56640).

Results

By cross-fertilization of the four T-DNA-insertion lines Gabi 450D10 (miox1), Salk 040608 (miox2), Salk 018395 (miox4) and Salk 112535 (miox5) we obtained a miox1/2/4/5 quadruple mutant. MIOX3 (At4g26255) is a transcribed but truncated and highly mutated pseudogene. The mRNA sequence has only partial homology to the neighbouring MIOX4-gene and contains numerous stop codons in all three reading frames. In the miox1/2/4/5-mutant a PCR-product of the predicted size was obtained for all four MIOX genes using the left border T-DNA (LB-T-DNA) insertion primer and a suitable MIOX isoform specific primer, consistent with the integration of a T-DNA in each MIOX gene (Fig. 2a). Homozygosity of the miox1/2/4/5-mutant was tested with MIOX gene specific primer pairs, spanning each T-DNA insertion site. Absence of the intact wild type genes can be proven for all four gene loci (Fig. 2a).

Fig. 2
figure 2

Genetic characterization of wild type and miox1/2/4/5-mutant. a Ethidium bromide strained PCR bands verifying the genetic identity of the quadruple knockout in miox1/2/4/5. b T-DNA insertion sites in the genes of the MIOX isoforms. The upper scheme shows the relation of MIOX isoforms (grey boxes exons) and the sites of T-DNA insertions. MIOX1: Gabi-Kat, MIOX2, 4, and 5 SALK. The arrow indicates the position of the translation start codon, the asterisks marks the stop codon. The lower part of the illustration shows the exact location of the T-DNA insertion. This information is derived from resequencing the junction site between the MIOX gene and the LB of the T-DNA. For MIOX1, the insertion site is ~30 bp upstream of a splice site; MIOX2 and 5 are even closer to splice sites; the SALK lines for MIOX2 and MIOX4 are located in an intron; the former again close to a splice site. c Expression profile of MIOX isoforms. Realtime PCR data of wild type and miox1/2/4/5-mutant material are displayed to depict the transcript distribution of MIOX relative to a housekeeping gene in three different stages of development: in seedlings, leaves, and in flowers (please note the different scale for flower transcripts indicating the far higher higher expression in this organ). Total relative amounts of transcripts were calculated and the ratio of miox1/2/4/5-mutant to wild type was formed. The level of MIOX transcripts is highly reduced in miox1/2/4/5-mutant. The experiment was performed in triplicates; shown are means and standard deviations

Full size image

Transcript profiling

RNA was isolated from wild type and miox1/2/4/5-mutant plants to compare the transcript levels of the four MIOX genes in both plant lines. Due to the highly transient nature of MIOX expression particularly in flowers as seen in GUS reporter gene plants (Kanter et al. 2005), we performed the analysis with different organs. The wild type data correspond well with data available at Genevestigator (http://www.genevestigator.com; data not shown).

In the miox1/2/4/5-mutant, we were able to detect transcripts of MIOX1, MIOX4 and MIOX5; leaving MIOX2 as the only isoform that is knocked out while the others are severely reduced in their transcript levels but not completely knocked out (Fig. 2c). Subcloning of the left border T-DNA insertion sites and ensuing sequencing revealed that the insertions are located close to splice sites or, in the case of MIOX2 and MIOX4, even in an intron sequence (see Fig. 2b). For the qPCR experiments (MIOX1, 2, 4) we used primer pairs spanning the T-DNA insertion sites. The residual transcripts (Fig. 2c) suggest the presence of residual full length mRNAs. Interestingly, we see that in the miox1/2/4/5-mutant MIOX4 is expressed in tissues where it is normally not expressed in the wild type (compare Fig. 2c). Still, the total amount of MIOX transcripts that can be rescued in the miox1/2/4/5-mutant is between 2 (flowers) and 14% (in leaves).

We assumed that a disruption of the MIOX pathway should have consequences for inositol metabolism and cell wall composition. This can be visualized when miox1/2/4/5-mutant and wild type plantlets are grown on either inositol-free MS (control) or on medium with 1 g L−1 inositol: The lines differ strongly; while the wild type can utilize inositol as additional carbon source and shows increased growth at the same age, miox1/2/4/5-mutant seedlings are significantly less developed and resemble the condition without inositol (Fig. 3). A small increase in growth of miox1/2/4/5-mutant on myo-inositol plates may be caused by the small residual MIOX-activity.

Fig. 3
figure 3

Utilization of inositol. Wild type seedlings can utilize inositol as additional carbon source for growth, miox1/2/4/5-mutant seedlings cannot. Wt seeds on the right side of each dish, miox1/2/4/5-mutant seeds on the left side. Left inositol-free MS medium, right MS medium supplied with 1 g L−1 inositol

Full size image

UDP-xylose, UDP-arabinose and UDP-galacturonic acid are derivatives of UDP-glucuronic acid. Therefore, a change in cell wall composition in miox1/2/4/5 lacking MIOX activity seemed possible. However, HPLC analysis of hydrolyzed cell wall material failed to exhibit the reduction in these sugar residues in miox1/2/4/5-mutant cell walls compared to the wild type (Fig. 4a). As the MIOX genes are predominantly expressed in flowers and also in developing seeds we also tested the composition of seed coat mucilage, which also shows no significantly altered sugar composition (Fig. 4b).

Fig. 4
figure 4

Sugar composition for cell wall (a) and seed mucilage (b). Even though MIOX activity is related to cell wall biosynthesis, composition of cell wall and seed mucilage in miox1/2/4/5-mutant does not deviate significantly from the wild type. Shown are means and standard deviations of four replicates

Full size image

Inositol feeding experiments in leaf discs

Experiments with leaf discs floating on inositol-enriched medium document rapid uptake of free inositol as described by Schneider et al. (2006). During the feeding situation, the miox1/2/4/5-mutant accumulated more free inositol after 2 h of feeding than the wild type (data not shown). The inositol concentration in both lines rapidly dropped down after the feeding of inositol was stopped. To be able to follow the fed inositol, we performed a similar feeding experiment with [3H]-inositol. Figure 5a shows that while all plant lines take up roughly the same total amount of label, wild type and L2 (CaMV35S::MIOX4 overexpressor line L2; Lorence et al. 2004) incorporate approximately 40% of it into the cell wall; in miox1/2/4/5-mutant the majority (~95%)of the 3H-label remains in the soluble fraction. We tried to further analyze the signal present in the aqueous phase: by evaporation experiments we were able to exclude the possibility of 3H2O as the source for the label. Incubation with anion or cation exchange beads (Bio-Rex MSZ 501(D) resin, Biorad), respectively, showed that no radioactive metabolite bound to the ion-exchange resin, which indicates that the fed [3H]-inositol is not present in a free ionic form. This suggests low amounts of inositol-phosphates in the aqueous phase under these conditions, which were not detectable under our conditions. Extraction with chloroform indicated that a small fraction of the total incorporated label is present in a hydrophobic form in all plant lines tested without showing quantitative differences (data not shown).

Fig. 5
figure 5

Incorporation of 3H-myo-inositol. a Distribution of label during short-term incubation of leaf discs. Leaf discs were floated for 3 h on MS supplemented with 3H-myo-inositol and chased for 1 h with cold inositol. The total amount of label found in the three lines is similar. The gain-of-function mutant line L2 expressing a CaMV35S::MIOX4 construct incorporates slightly more signal into the cell wall than the wild type; the majority of signal detected in the miox1/2/4/5-mutant is retained in the soluble fraction. b Distribution of label in 7-day-old seedlings. Seedlings were grown in liquid MS supplemented with 3H-myo-inositol. The crude extract was split into pellet and supernatant; the pellet was washed several times to obtain the “cell wall” fraction; the supernatant was mixed with acetone to retain soluble sugars in the supernatant while precipitating oligosaccharides. The amount of signal in the cell wall of the miox1/2/4/5-mutant is highly reduced compared to the wild type; instead, more signal is found as soluble carbohydrate or acetone precipitate. The experiment was performed in triplicates; shown are means and standard deviations

Full size image

Inositol feeding experiments in seedlings

To clarify the identity of the soluble label and the residual signal in the miox1/2/4/5-mutant cell wall fraction, we grew seedlings on medium with [3H]-inositol. The distribution of 3H-label between the soluble and insoluble fraction is shown in Fig. 5b. We included a second fractionation step of the aqueous supernatant, a precipitation with 80% acetone, to separate soluble inositol from precipitable galactinol. HPLC analysis confirmed that the radioactive signal in soluble fraction is attributed to inositol and galactinol (data not shown).

Next, we turned our attention to the cell wall fraction. To exclude the possibility of starch as signal source, we performed a digestion with α-amylase and amyloglucosidase. This procedure released roughly one-fifth of the signal from the insoluble fraction, but this observation was made in wild type as well as miox1/2/4/5-mutant samples. The necessary incubation in slightly acid buffer also releases part of the pectic fraction and possibly also some arabinogalactan proteins. TFA hydrolysis of the cell wall fraction was performed to investigate whether the pattern of labelled sugars is altered in miox1/2/4/5-mutant compared to wild type samples. Figure 6 shows the HPLC elution profile in the upper portion and the scintillation counts of the collected fractions in the lower portion of each graph. Note that the scale for the radioactivity counts (right Y-axis) is tenfold lower for the miox1/2/4/5-mutant than for the wild type, confirming the >90% reduction of myo-inositol conversion into cell wall sugars in the miox1/2/4/5-mutant. Signal is found for inositol and the GlcA derivatives GalA, Ara and Xyl. GlcA itself is probably present in too small a concentration to elicit a signal (compare the amount in Fig. 4a). An unidentified signal, presumably resulting from incompletely hydrolyzed sugars, is released from the column during the transition from solely alkaline buffer to a buffer solution that also contains sodium acetate (the corresponding electrochemical signal is seen in the HPLC elution profile as well).

Fig. 6
figure 6

Distribution of label among cell wall monomers. Labelled cell wall pellets of wild type (upper panel) and miox1/2/4/5-mutant (lower panel) were characterized via HPLC (line), fractionated, and scintillation counted (columns). 1 Fuc, 2 Rha, 3 Ara, 4 Gal, 5 Glc, 6 Xyl, 7 GalA. Asterisks signal from buffer change. Please note the different scale for radioactive metabolites, used for wild type and miox1/2/4/5-mutant plants

Full size image

As we found no changes in the cell wall composition in miox1/2/4/5-mutant compared to wild type despite the block via the MIOX enzymes, we measured the transcripts of UGD, the key enzyme acting in a parallel reaction pathway leading to UDP-GlcA. UGD-genes are up-regulated (~1.5-fold in seedlings and leaves) in the context of our miox1/2/4/5-mutant—to the greatest extent in the flower (~3-fold), where the lack of MIOX transcript must be most pronounced. The increase in UGD transcripts results in approximately 15% higher UGD activity in flowers. The lack of a visible phenotype in the miox1/2/4/5-mutant is likely explained by the compensatory effect of increased UGD activity and the residual activity of MIOX in the miox1/2/4/5-mutant.

Metabolite profiling

Investigation of the RFOs, biosynthetic pathway is shown in Suppl. Fig. S1, shows an increase in miox1/2/4/5-mutant leaves under standard conditions which can even be elicited to a greater extend when salt stress is applied. In L2 (MIOX overexpressor), where the level of free inositol is lower due to increased MIOX activity, the response to salt stress is somewhat weaker than in the wild type (Fig. 7).

Fig. 7
figure 7

Salt-stress treatment leads to an increase of inositol and derivatives. Soluble carbohydrates were extracted from young leaves of 6-week-old plants 5 days after watering them with 150 mM NaCl (“induced” samples). Wild type plants, the miox1/2/4/5-mutant and a CaMV35::MIOX4 overexpressor line (L2) are compared. As a control served a set of plants that received tap water instead (“not induced”). Accumulation for inositol (a), galactinol (b) and raffinose (c) is presented. Shown are means and standard deviation of three independent experiments

Full size image

In seeds, the changes are displayed even more intensely: approximately 50× more free inositol and 5× more galactinol is present in miox1/2/4/5-mutant seeds compared to the wild type. Raffinose content is the same, and stachyose is reduced. We also observed that the content of phytic acid, the inositol-hexakisphosphate which serves as ion chelator and phosphate-storage molecule, is higher in miox1/2/4/5-mutant than in wild type or an overexpressing line (Fig. 8a, b).

Fig. 8
figure 8

Seed sugar content. a Inositol and RFO content in seeds. The miox1/2/4/5-mutant shows highly increased levels of free inositol and the galactosylated inositol galactinol. b Phytic acid content in seeds of wild type, miox1/2/4/5-mutant and a CaMV35::MIOX4 overexpressor line (L2). The level of the fully phosphorylated inositol-derivative phytic acid is elevated in miox1/2/4/5-mutant and reduced in the MIOX4-overexpressing line L2. A t test shows significant (P < 0.05) differences between samples, despite the large SD. Shown are means and SDs of triplicates

Full size image

Seedling characterization

Based on these findings, we decided to conduct a series of germination experiments (the data is summed up in Suppl. Fig. S2). First, we measured the inositol content in seedlings grown on media with differing inositol contents. This clearly demonstrated that the miox1/2/4/5-mutant plantlets not only accumulate externally applied inositol, but that their intrinsic level of autonomously synthesized inositol is high even at conditions where no free inositol can be detected in the wild type or L2 (MIOX overexpressor). Second, we performed a basic stress treatment with two concentrations of NaCl (50 and 150 mM). miox1/2/4/5-mutant is more sensitive than the wild type. This is a bit surprising since inositol and RFOs are interpreted as osmolytes which should convey a greater resistance to such stresses—not the greater susceptibility we see in this experiment. Third, we incubated seeds on phosphate-free MS medium, assuming that the increased level of phytic acid seen in miox1/2/4/5-mutant seeds enables them to deal better with this situation. Indeed, we see a slightly higher growth rate in miox1/2/4/5-mutant seedlings compared to the wild type.

Adult plant characterization

The RFOs are discussed to be involved in stress tolerance. Nishizawa et al. (2008) found plants with high galactinol and raffinose contents less susceptible to oxidative stress as measured via chlorophyll quenching after treatment with methylviologen. We repeated this experiment with our miox1/2/4/5-mutant line, but could not detect a similar protective effect of its elevated galactinol and raffinose content compared to the wild type, neither in standard (not shown) nor in high light conditions (Suppl. Fig. S3a).

Rohde et al. (2004) saw a correlation between high raffinose levels and increased freezing tolerance. We assumed that the elevated level of raffinose even at standard growth conditions might convey a similar freezing tolerance to our miox1/2/4/5-mutant plants. However, determination of the percentage of ion leakage did not result in significantly differing values (Suppl. Fig. S3b).

In a third line of experiments, we investigated the correlation of MIOX activity and ascorbic acid content as postulated in Lorence et al. (2004). The authors stated that three independent lines expressing a CaMV35S::MIOX4 construct had an up to threefold elevated ascorbic acid content in foliar tissue. We have conducted experiments comprising wild type plants, the same MIOX overexpressing plants, our miox1/2/4/5-mutant and vtc1 plants with a published content of ~30% of wild type ascorbic acid levels (Conklin et al. 1997), serving as a control. We measured their intrinsic foliar ascorbate under standard growth conditions and monitored it while feeding excess inositol (this data is shown in Suppl. Fig. S4) and during application of light stress (Suppl. Fig. S3c). Under all conditions tested, neither the MIOX overexpressing line(s) (L2 is shown representatively) nor the miox1/2/4/5-mutant with only marginal MIOX activity deviated significantly from wild type levels. Only the vtc1 mutant defective in the mannose-pathway leading to ascorbic acid showed the expected decreased content.

We also performed analysis of starch and soluble sugar content in adult leaf tissue, indicating that the central carbon metabolism is not affected in miox1/2/4/5-mutant or L2 MIOX overexpressor plants, since starch, glucose and sucrose levels are the same as in the wild type (data is shown in Suppl. Fig. S5).

Discussion

The analysis of miox1/2/4/5-mutant with a knockdown in all four MIOX genes clearly shows a drastic reduction of myo-inositol conversion into cell wall polymers and, thus, confirms the contribution to UDP-GlcA biosynthesis. Initial feeding experiments of labelled myo-inositol (~1 μM) in otherwise inositol-free MS-medium clearly showed a strong reduction of label incorporation into cell wall polymers. The low amount of myo-inositol in conjunction with the observed rapid metabolism of it (Endres and Tenhaken 2009), minimizes the perturbation of the myo-inositol pool and, thus, indicates that under normal growth conditions of seedlings and leaf discs part of the myo-inositol is converted into cell wall precursors. The functionality of the MIOX pathway is also supported by the better growth of wild type seedlings on regular MS-medium (containing 0.1 g L−1 myo-inositol) compared to miox1/2/4/5-mutant, which is further enhanced on myo-inositol fortified media (1 g L−1). We conclude that inositol liberated from phytate degradation during seedling germination can efficiently be used for cell wall biosynthesis in the growing wild type seedling. The prerequisite for a contribution of the MIOX pathway to cell wall biosynthesis is, however, a sufficient supply of myo-inositol.

The myo-inositol pool is likely controlled by the biosynthesis and by the degradation rate. The strong increase in myo-inositol in miox1/2/4/5-mutant suggests an important role of the degradation pathway for controlling cellular inositol concentration. Smart and Flores (1997) showed a fourfold higher inositol concentration in Arabidopsis plants after overexpression of an inositol-3-phosphate synthase gene from Spirodela polyrrhiza under the control of the strong CaMV35S-promoter. Our miox1/2/4/5-mutant shows increased myo-inositol concentrations, though preliminary microarray experiments suggest that at least one of the three genes for myo-inositol biosynthesis (inositol-3-phosphate synthase; EC 5.5.1.4) is down-regulated, whereas the two others show no change in gene expression. This prompted us to look for correlated expression between the genes for inositol-3-phosphate synthases and MIOX using expression data from public databases with the Genevestigator tool (http://www.genevestigator.com). To our surprise, most tissues show no co-regulation of the two groups of genes, which we would have expected for a functional MIOX pathway for cell wall precursor biosynthesis. This suggests us that one important function of MIOX is a major contribution to balance myo-inositol concentrations in plant cells. Controlled levels of myo-inositol are also important in animal systems. Changes in myo-inositol concentration have been linked to common diseases. Nascimento et al. (2006), for instance, found an important cell protective role of myo-inositol for endothelial cells in diabetes.

MIOX role for ascorbic acid

The contribution of the MIOX-derived glucuronic acid to ascorbate biosynthesis was recently suggested by two studies, in which either the MIOX4 gene (Lorence et al. 2004) or a purple acid phosphatase (Zhang et al. 2008) was overexpressed. We found no change in ascorbate concentration in our loss of function miox1/2/4/5-mutant under ambient, high light or inositol-feeding conditions. The gain of function MIOX4 overexpressor line has an increased MIOX enzyme activity but also no increase in ascorbate concentration (Endres and Tenhaken 2009). Therefore, the role of MIOX for ascorbate in plants in analogy to the mammalian pathway is questionable. However, the possibility remains that the subsequent glucuronokinase is the channeling enzyme which guides glucuronic acid into cell wall precursors and thus prevents the conceivable conversion into ascorbate.

Metabolic consequences of MIOX loss

The miox1/2/4/5-mutant has elevated levels of myo-inositol and galactinol. This increase in galactinol in unstressed plants supports a model that the availability of myo-inositol is more important for galactinol biosynthesis than an increase in galactinol-synthase. Similar findings were reported for low phytate mutants from barley with increased galactinol levels, which correlates with the metabolite myo-inositol but not with the activity of galactinol-synthase (Karner et al. 2004).

The elevated concentration of galactinol results in an increase of the RFOs, which are typical abiotic stress-associated metabolites in plants.

Whereas Zuther et al. (2004) found no enhanced freezing/drought tolerance in plants with elevated RFO-levels, others like Taji et al. (2002) functionally linked the increased RFOs with membrane stabilization and drought or increased tolerance to reactive oxygen species (Nishizawa et al. 2008). We carefully repeated many of these experiments with the miox1/2/4/5-mutant and found no significant increase in freezing or paraquat (methylviologen) mediated ROS tolerance. This let us conclude that additional other unidentified metabolic changes contribute to the observed increase in stress tolerance of high level RFO-plants which may be absent in miox1/2/4/5-mutant.

MIOX interaction with the UGD-pathway

Surprising to us, the almost total loss in MIOX activity causes no morphological changes. The cell wall and seed mucilage composition remain similar in miox1/2/4/5-mutant and wild type plants. This is most likely explained by two factors: first, the limited flux of carbohydrates for cell wall precursors via the MIOX pathway in many developmental stages. These genetic data support precursor feeding studies in Arabidopsis cell cultures as reported by Sharples and Fry (2007). Second, the observed upregulation of UGD-genes in the miox1/2/4/5-mutant, which are 1.5- to 3-fold more expressed in miox1/2/4/5-mutant as in wild type plants, depending on the tissue. Interestingly, a reduction of UGD-activity causes strong morphological changes which are not compensated by enhanced MIOX activity (data not shown). This suggests a model that the concentration of the metabolite UDP-GlcA, or another UDP-sugar derived from it, might be a signal for UGD gene regulation but is of marginal influence on MIOX gene expression. Some hints for such a mode of regulation come from previous studies, which showed a strong gene expression of UGD::GUS reporter genes in etiolated seedlings, which is almost absent in light grown seedlings. A similar observation was made in developing leaves, in which stomata cells and basal cells surrounding epidermal trichomes show transient UGD::GUS expression during cell wall thickening.

A possible role for MIOX in pollen tube growth

MIOX genes are most prominently expressed in flowers, in particular in pollen. In former studies, Kroh and Loewus (1968) showed by feeding experiments that most of the pectin in Lily pollen cell walls was derived from the MIOX pathway. We were therefore surprised to find no reduced fertility in miox1/2/4/5-mutant. Obviously the remaining ~10% activity (in sum) of all MIOX isoforms is sufficient to allow proper pollen development and growth. Using light microscopy we could not detect visible differences. Sections of embedded pollen were stained with monoclonal antibodies directed against different pectic epitopes (LM5, LM6, Jim5 Jim7) but failed to reveal differences between the miox1/2/4/5-mutant and wild type. In contrast, a knockout in UDP-sugar pyrophosphorylase as the final enzyme of the MIOX pathway to UDP-GlcA is lethal during pollen development (Schnurr et al. 2006).

It will be interesting to find out whether a stronger knockdown of residual MIOX genes will also cause pollen abortion. Alternatively, the UDP-sugar pyrophosphorylase has an additional function beside the activation of GlcA-1-P to UDP-GlcA. To test this we will express artifical miRNAs targeted against MIOX genes in the miox1/2/4/5-mutant background. A further but long-term approach might be required to isolate better knockout alleles and restart the crossing process.

In summary the strong knockdown of the MIOX gene family does not lead to visible phenotypes in Arabidopsis or to changes in the composition of the cell wall. The main pathway for UDP-GlcA as important cell wall precursor, catalyzed by the enzyme UDP-glucose dehydrogenase, is upregulated in miox1/2/4/5 and likely compensates for the loss of MIOX activity. MIOX is responsible for myo-inositol breakdown and therefore involved in the homeostasis of this metabolite. The miox1/2/4/5 mutant shows an increase in myo-inositol and metabolites containing myo-inositol.