We have previously described the T-DNA null alleles map3kε1 and map3kε2 (Chaiwongsar et al., 2006). In that previous work we observed that homozygous single-mutant lines for either map3kε1 or map3kε2 did not display any obvious abnormal phenotypes when grown under standard laboratory conditions. We also found that the map3kε1;map3kε2 double-mutant combination caused pollen lethality (Chaiwongsar et al., 2006). These results suggested that there is functional redundancy between MAP3Kε1 and MAP3Kε2 during pollen development. Because map3kε1;map3kε2 pollen is non-viable, however, it is not straightforward to generate a plant that is homozygous double-mutant for these T-DNA null alleles. For this reason one cannot directly explore via genetic analysis the function of this pair of potentially redundant genes in the sporophyte.

In order to overcome this challenge we needed an experimental strategy that would allow map3kε1;map3kε2 double-mutant pollen to be viable, which would in turn allow a homozygous double-mutant sporophyte to be formed. The strategy that we chose was to place the wild-type MAP3Kε1 coding region under the transcriptional control of an alcohol-inducible promoter system and induce expression of MAP3Kε1 during pollen and embryo development so that these processes could proceed normally. After seed had been produced by these plants, the progeny could be grown under conditions that do not induce expression of the MAP3Kε1 transgenic construct, thereby allowing us to observe any phenotypic consequences of the reduced levels of MAP3Kε1 and MAP3Kε2.

We began by constructing a plasmid in which a YFP tag is translationally fused to the wild-type MAP3Kε1 coding sequence (CDS) under the transcriptional control of the ethanol-inducible alcA promoter (Figure 1A; Caddick et al., 1998; Roslan et al., 2001; Deveaux et al., 2003). We included the YFP tag on this construct so that we could monitor YFP-MAP3Kε1 expression at the protein level via fluorescence microscopy. This construct was then introduced via Agrobacterium-mediated transformation into plants that were homozygous for map3kε2 and heterozygous for map3kε1. PCR-based genotyping was used to identify map3kε1^-/+;map3kε2^-/- primary transformants that were also carrying the alcA:YFP-MAP3Kε1 construct. In order to determine if these plants carried a functional copy of the alcA:YFP-MAP3Kε1 construct, we monitored YFP-MAP3Kε1 protein expression using fluorescence microscopy. Twenty-four hours after treatment with a 1% ethanol spray, strong induction of YFP fluorescence was observed throughout the aerial tissue of the plants, including leaves and flowers (Figures 1B,C), indicating that the alcA:YFP-MAP3Kε1 construct present in these lines was performing as expected.

In order to rescue the lethal phenotype displayed by map3kε1;map3kε2 double-mutant pollen, developing flowers of map3kε1^-/+;map3kε2^-/-;alcA:YFP-MAP3Kε1 plants were sprayed with 1% ethanol on a daily basis throughout their life-cycle in order to activate expression of alcA:YFP-MAP3Kε1 and allow double-mutant pollen grains to survive. Progeny from these ethanol-treated plants were genotyped and several map3kε1^-/-;map3kε2^-/-;alcA:YFP-MAP3Kε1 individuals were identified. These results indicated that the alcA:YFP-MAP3Kε1 construct was able to rescue the map3kε1;map3kε2 pollen lethal phenotype, and that the YFP tag did not interfere with MAP3Kε1 function during pollen development.

Progeny derived from eight independently transformed map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 lines were screened using quantitative reverse-transcriptase PCR to find the line with the lowest background level of YFP-MAP3Kε1 expression. In the absence of ethanol treatment, the double-mutant line chosen for phenotypic analysis in this study expressed YFP-MAP3Kε1 at a level that was 10-fold lower than that of native MAP3Kε1 expressed in wild-type plants (Figure 1D). When treated with exogenous ethanol, seedlings carrying this construct displayed strong induction of the YFP-MAP3Kε1 gene up to a level ca. 150 times higher than that of native MAP3Kε1 expressed in wild-type plants(Figure 1D). When grown in soil in the absence of exogenous ethanol treatment, progeny of this map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 line had an overall reduction in the size of their rosette leaves and bolts (Figure 1E).

In order to determine if the reduced size of rosette leaves seen in map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants grown in the absence of ethanol was due to the low level of MAP3Kε1 expression in these plants relative to wild-type, we performed the following experiment. Two pots of map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants were grown side-by-side, along with wild-type controls. One pot of map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants was sprayed with 1% ethanol every 2 days throughout development, while the second was sprayed with water only. We also subjected wild-type Columbia plants to the same regimen of water and ethanol treatments to determine if ethanol treatment had any effect on the growth of wild-type plants. As seen in Figure 2A, treatment with ethanol did not affect the growth of wild-type plants. Spraying map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants with ethanol to induce YFP-MAP3Kε1 expression, however, resulted the partial rescue of the reduced-leaf-size phenotype (Figures 2B,C).

We also performed scanning electron microscopy of the leaf epidermis of map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants to observe the size and shape of the epidermal cells. This analysis revealed that the epidermal cells of untreated map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants were heterogeneous in size, with a majority of the cells being much smaller than those of wild-type (Figure 2D). Some relatively large cells were also observed in the untreated mutant lines, but these cells did not have the typical jig-saw shape of wild-type epidermal cells. These results indicate that MAP3Kε1/2 may be important for the normal expansion of leaf epidermal cells. The cellular phenotypes displayed by the untreated map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants were partially rescued when the plants were sprayed with ethanol every 2 days throughout development to induce expression of YFP-MAP3Kε1 (Figure 2C).

To investigate the role of MAP3Kε1/2 in root growth, seeds of the map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 mutant line as well as a wild-type control were sown on agar plates containing growth media. Root length was then measured after 5 days of growth on vertically oriented plates. We observed that the primary roots of the map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants were ca. 40% shorter than wild-type (Figure 3A). In order to further investigate the contribution of map3kε1 and map3kε2 to this phenotype, we performed a time-course experiment to measure root length over the course of 9 days of growth on vertical agar plates. As additional controls, this experiment included map3kε1and map3kε2 homozygous single-mutant lines as well as the map3kε1^-/-;map3kε2^-/-;alc-YFP- MAP3Kε1 line.

As shown in Figure 3B, the primary roots of map3kε1 homozygous single-mutant lines were significantly shorter than those of wild-type. This phenotype is fully rescued by the presence of an ectopic copy of the wild-type MAP3Kε1 genomic locus (Figure 3B). By comparison, the roots of map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants were found to be even shorter than those of the map3kε1 single-mutant.

We next attempted to use ethanol-induction to rescue the short-root phenotype of map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants. We tested a variety of ethanol-treatment procedures but were not able to generate substantial YFP-MAP3Kε1 expression in root tissue as determined by fluorescent microscopy, and no phenotypic rescue was observed.

To further explore the cause of the short-root phenotype displayed by map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants, epidermal cell length was measured in three different zones of 7-day-old roots. These zones correspond to the root meristem, the elonga- tion zone, and the mature zone where cell elongation has ceased (Cruz-Ramirez et al., 2004). Root epidermal cells of map3kε1^-/-; map3kε2^-/-;alc-YFP-MAP3Kε1 seedlings were shorter than wild-type in all of these zones. On average the root epidermal cells were ca. 20% shorter in the map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants when compared to wild-type, indicating that cell elongation is reduced (Figure 4A). Because the overall length of the primary root in map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants is ca. 40% less than wild-type (Figure 3), reduced cell elongation alone cannot fully account for the total reduction in root length. The mutant lines must also have fewer total cells in the primary root, suggesting a reduction in the rate of new cell production in the roots of map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants.

We also documented the expression pattern of MAP3Kε1 and MAP3Kε2 in Arabidopsis primary roots using YFP fusion constructs. YFP-MAP3Kε1 and YFP-MAP3Kε2 translational fusions under the transcriptional control of their respective native promoters were introduced into wild-type Arabidopsis plants via Agrobacterium-mediated transformation. Transgenic lines were generated and fluorescence microscopy was used to detect YFP-MAP3Kε1 or YFP-MAP3Kε2 expression. After seed germination, YFP-MAP3Kε1 was strongly expressed throughout the root apical meristem (Figure 4B). By contrast, YFP-MAP3Kε2 expression in the root apical meristem was largely restricted to vascular tissues (Figure 4C). These results indicate that MAP3Kε1 and MAP3Kε2 are both expressed in the root apical meristem region with overlapping, but distinct expression patterns.

map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants produce fewer seeds than wild-type when grown in the absence of exogenous ethanol (Figure 5B). The mutant plants produce an average of 15 seeds per silique, compared to 55 seeds per silique for wild-type (Figure 5B). Although map3kε1;map3kε2 pollen is not viable (Chaiwongsar et al., 2006), Alexander’s staining showed that pollen from map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants was viable in the absence of exogenous ethanol treatment, suggesting that the alcA:YFP-MAP3Kε1 construct is expressed in the absence of ethanol treatment in developing pollen (data not shown). This phenomenon might be due to the fact that the ethanolic fermentation pathway is normally activated during pollen development (Mellema et al., 2002), resulting in the production of acetaldehyde and ethanol, which can both act as inducers of the alcA promoter (Tadege and Kuhlemeier, 1997).

We have previously shown that map3kε1;map3kε2 double-mutant female gametophytes develop normally (Chaiwongsar et al., 2006). Therefore, the decrease in seed production of the map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants may be due to a disruption of seed development, rather than reduced pollen fitness. To test this possibility, map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants were grown and allowed to self-pollinate in the absence of exogenous ethanol treatment. Developing seeds from the map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants were then analyzed microscopically. Siliques from wild-type and map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants were removed from the main shoot and compared. During normal Arabidopsis seed development, the external appearance of immature seeds changes from white to green, and then to brown at ca. 15–17 days after flowering (DAF). At 7 DAF, we observed that a portion of the seeds from the map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants were smaller than wild-type and brown (Figure 5A). By 16 DAF, a majority of the seeds from map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 siliques appeared collapsed and shriveled (data not shown).

Immature seeds within a single wild-type silique normally develop at approximately the same rate (Sparkes et al., 2003). In wild-type plants grown in our laboratory conditions, the embryos reached the globular stage by 2 DAF and the torpedo stage by 5–6 DAF (Figures 5C,D). However, many of the embryos from map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants grown under the same conditions displayed delayed development. All of the map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 embryos reached the globular stage at the normal time of 2 DAF, but at 6 DAF development was stalled at the globular or transition stage in many of the embryos (Figure 5D). At 7 DAF, most of the map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 seeds had collapsed and the embryos were arrested at either the globular or heart stage (data not shown).

The arrest of embryo development observed with map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 plants suggested that MAP3Kε1, MAP3Kε2 play an important role in normal embryo development. If these proteins are directly involved in embryo development, then one would expect the proteins to be expressed in developing embryos. To directly explore this question we used the YFP-MAP3Kε1 and YFP-MAP3Kε2, translational fusion constructs described above to determine if these proteins are expressed in developing embryos. These constructs express the fusion proteins using the gene’s respective native promoters. YFP fluorescence was detected via fluorescence microscopy in the developing embryos of plants expressing YFP-MAP3Kε1 and YFP-MAP3Kε2 (Figure 6). YFP-MAP3Kε1 expression was observed throughout all the tissues of the embryo. By contrast, the expression pattern of YFP-MAP3Kε2 was noticeably concentrated in the developing vascular cells (Figure 6).

Arabidopsis thaliana (Columbia ecotype) were grown under continuous light at 22–25°C in soil. To activate expression of the alcohol-inducible alcA-YFP-MAP3Kε1 construct in plants growing in soil, the plants were sprayed with 1% (v/v) ethanol every 2 days throughout development.

In order to create the YFP-MAP3Kε1 fusion construct, a wild-type MAP3Kε1 genomic clone previously described (Chaiwongsar et al., 2006) was modified using site-directed mutagenesis to add recognition sites for the restriction enzymes AvrII and AgeI immediately after the start codon. The YFP CDS was then PCR amplified from a plasmid vector using PCR primers that added an NheI site to the 5′. end of the CDS and an AgeI site to the 3′. end. This PCR-amplified fragment containing the YFP CDS was then ligated to the modified MAP3Kε1 clone using sticky ends generated by AvrII, AgeI, and NheI cleavage. NheI sticky ends are identical to those produced by AvrII. The resulting construct contains the YFP CDS fused in frame to the 5′. end of the MAP3Kε1 coding region and retains the MAP3Kε1 native promoter. The same strategy was used to construct the YFP-MAP3Kε2 fusion construct. The resulting plasmids were introduced into plants using Agrobacterium-mediated transformation (Clough and Bent, 1998).

In order to construct the ethanol-inducible version of the YFP-MAP3Kε1 fusion protein, recognition sites for the restriction enzyme RsrII were introduced immediately downstream of the ATG start codon of the YFP-MAP3Kε1 coding region described above. A 3.7 kb fragment from the plasmid pBinSRNACATN (Caddick et al., 1998) was then PCR amplified using primers that added RsrII sites to the PCR product and cloned into the RsrII sites in YFP-MAP3Kε1. This 3.7 kb fragment contained an AlcR expression cassette as well as the AlcA promoter. The resulting plasmid contains the YFP-MAP3Kε1 gene under the transcriptional control of the AlcA promoter (Figure 1A). This vector was introduced into Arabidopsis plants via Agrobacterium-mediated transformation (Clough and Bent, 1998).

Seeds were surface sterilized with 95% (v/v) ethanol for 5 min, air dried, and sown on 0.5× Murashige and Skoog basal salt mixture media (pH 5.8; Sigma) with 0.7% agar (w/v). After sowing, seeds were allowed to imbibe in the dark at 4°C for 2 days and were then grown on vertically oriented plates under constant light at 22–24°C. To measure the root lengths, images were captured using a flat-bed scanner and analyzed with Adobe Illustrator software.

Confocal microscopy was used to collect images of root epidermal cells to measure cell length. For these experiments, cell walls were stained using propidium iodide (PI) and optical sections were obtained using a Zeiss LSM 510 Meta Confocal microscope (Zeiss, Thornwood, NY, USA). For PI staining, fresh plant tissue was transferred to a solution of 10 µg/mL of PI (Sigma) for 15–30 min. Imaging was performed using a 568-nm excitation line and an emission window of 585–610-nm. Epifluorescence images were captured digitally using an Olympus DP70 camera (Center Valley, PA, USA) attached to an Olympus BX60 epifluorescence microscope (Center Valley, PA, USA).

To detect YFP fluorescence in embryos, embryos were removed from the seed coats at various stages and viewed with an epifluorescence microscope. To study embryo development, embryos were fixed with 4% glutaraldehyde (in 12.5 mM cacodylate, pH 6.9) overnight. After fixation, the embryos were dehydrated through a 20, 40, 60, 80, and 100% ethanol series for 20 min per step. After dehydration, embryos were cleared in 1:1 (v/v) benzyl benzoate:benzyl alcohol for minimum of 2 h. Embryos were then mounted with glycerol and observed with a Zeiss LSM 510 Meta Confocal microscope with a 488-nm argon laser and an LP539 filter to detect autofluorescence.

In order to measure the expression level of YFP-MAP3Kε1 expression in the map3kε1^-/-;map3kε2^-/-;alc-YFP-MAP3Kε1 lines, total RNA was isolated from rosette leaf tissue using a Qiagen RNeasy Mini Kit (Qiagen, Valencia, CA, USA). First-strand cDNA was prepared using 100 µg of total RNA with the Super Script First-Strand synthesis system (Invitrogen, Carlsbad, CA, USA). The resulting first-strand cDNA was diluted 1:100 with water and used as a template for real-time, quantitative RT-PCR amplification using an iCYCLER PCR system (BioRad, Hercules, CA, USA). SYBR Green (Molecular Probes, Eugene, OR, USA) was used to detect RT-PCR product accumulation. Primers for the actin gene ACT2 were used as a control (Genbank Accession: U41998). To detect MAP3Kε1 expression, the predicted cDNA sequence of MAP3Kε1 was used to design the following pair of PCR primers: ε1-RT-A1, 5′.-AAAAACATTGTGAAGTATCTTGGGTCGTC-3′.; ε1-RT-A2, 5′.-GCTTCTTTACGAATTTCGCGAGAACGATC-3′. (Chaiwongsar et al., 2006).