A T-DNA activation tagging construct was made by cloning the pUC9 vector containing the 35S double enhancer, the Alfalfa Mosaic Virus (AMV) 5′ leader and the hygromycin resistance marker between the synthetic borders of the pBIN19 derived pSDM14 vector using the unique XhoI site. Arabidopsis thaliana (ecotype C24) was transformed with this activation tagging construct and transformants were screened for developmental mutants. Initial characterization of the T-DNA/vector insert was described by van der Graaff (1997). Seeds were sterilized by incubation during 1 min in 70% ethanol, then 20 min in 1% hypochlorite, followed by four rinses with sterile water. Seeds were imbibed on solidified 0.5 MS medium, in the dark, for 2–4 days at 4°C. The seeds were then placed in a culture chamber (21°C, 16 h of light/8 h of dark, 3000 lux) to germinate. The transgenic plants were selected by germination on 10 mg/l hygromycin-containing medium (Duchefa). Hygromycin resistant seedlings were scored 2 weeks after germination. Microscopic images of wild-type and sns-D plants were recorded with a Leica MZ12 stereomicroscope (Leica Microsystems), equipped with a DKC-5000 camera (Sony). Plant material was fixed in 2% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) for 8 h at room temperature, dehydrated through an ethanol series and embedded in epon. Cross-sections were stained with toluidine blue, mounted in epon, and analyzed with an upright compound microscope.

For analysis of the T-DNA-plant DNA border sequence, thermal asymmetric interlaced (TAIL) PCR was performed according to Liu and Whittier (1995), with specific primers HYGLB1, HYGLB2, and HYGLB3 and random primer AD3 (Table 1).

Flower buds were used as a source for nuclei preparations. Tissue fixation, cell spreading, screening for appropriate stages, and fluorescence in situ hybridization (FISH) have been described by Schubert et al. (2001). BAC F3F19 and BAC F13K23 are located adjacent to each other on chromosome 1 near the T-DNA/vector integration site and they were labeled with biotin-dUTP and digoxigenin-dUTP and detected with rhodamine-labeled streptavidin (magenta fluorescence) and fluorescein isothiocyanate-labeled antibodies against mouse-anti-digoxigenin (green fluorescence), respectively. The preparations were counterstained and mounted in 1 mg ml 4′,6-diamino-2-phenylindole (DAPI) in vectashield (Vector Laboratories). Preparations were analyzed with a microscope (model BH2-RFC, Olympus) equipped with a 100 W mercury lamp. Images were acquired with a cooled CCD camera (Astro-med).

One- and five-week-old wild-type and sns-D seedlings were fixed in a mixture of 2% (w/v) paraformaldehyde and 2.5% (v/v) glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.3, for 3 h at room temperature. Samples were dehydrated through a graded series of 50, 70, 90, 96, and 100% (v/v) acetone and then dried with a Bal-Tec CPD 030 critical point drier (Leica Microsystems). The samples were then mounted on stubs, coated with gold on a Polaron SEM coating unit E5100 and observed by use of a Jeol 6400 scanning electron microscope (Jeol).

Ten plants of about 5-week-old were ground in 10 ml ice-cold extraction buffer containing 100 mM HEPES (pH 7.2), 10% (w/v) sucrose, 0.1% (w/v) CHAPS, 5 mM DTT, and 10⁻⁶% (v/v) NP40 with an Ultra-turrax (three pulses at 24000 rpm of 30 s each with 30 s intervals, on ice). Subsequently, the homogenate was incubated on ice for 15 min. Then, the homogenate was centrifuged to pellet cell debris at 2000×g for 5 min and the supernatant was centrifuged to pellet insoluble material and the microsomal fraction for 10 min at 100000×g at 4°C. The lipid layer was removed from the surface by sucking with a needle and a syringe. The soluble protein extract was filtered over a 0.22 μm Millex syringe driven filter unit (Millipore). Protein concentrations were determined using the Bradford method (Bio-Rad) with BSA as the standard (Bradford, 1977). In a 96-well plate, 75 μl of soluble protein extract containing 5 μg of proteins were mixed with 25 μl of the synthetic fluorogenic caspase-3 or caspase-6 substrates (Ac-DEVD-AMC or Ac-VEID-AMC, respectively, 75 μM final concentrations; Calbiochem). The specificity was measured by addition of caspase-3 or caspase-6 inhibitor (Ac-DEVD-CHO or Ac-VEID-CHO respectively, 250 μM final concentration). During 2 h, the proteolytic activity was measured every 10 min at room temperature in triplicates per sample. Substrate cleavage was detected in a fluorescence spectrometer LS50B (Perkin Elmer, USA) at an excitation wavelength of 380 nm and an emission wavelength of 460 nm, with an excitation and an emission slit value of 5.0. Kinetics of substrate hydrolysis was tested to be linear throughout 2 h of reaction. The caspase-like activity was calculated as increase of fluorescence units per microgram of protein per hour. The data presented were representative of three independent experiments (n = 3) with three different extracts and the data were presented as mean ± SD.

Five-week-old wild-type or sns-D leaves were fixed in 2% (w/v) glutaraldehyde in 10 mM NaH₂PO₄, 120 mM NaCl, 2.7 mM KCl, pH 7.4 (phosphate-buffered saline, PBS) overnight at room temperature. After dehydration at room temperature in a graded series of 70, 90, 96, and 100% (v/v) ethanol, samples were embedded in Technovit (Heraeus). Sections (2 μm) were attached on Biobond (Biocell) coated slides. Terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay was done with an in situ cell death detection kit (Roche). Following TUNEL reaction, Sytox orange nuclear staining was performed as described (Maraschin et al., 2005). Samples were analyzed with a Zeiss Axioplan confocal microscope (Zeiss) with a MRC 1024 ES Bio-Rad module (Bio-Rad). An Argon/Krypton laser (488/568 nm) was used to visualize the TUNEL (Ex 488 nm, Em 522 DF 32) and the Sytox orange (Ex 568, Em 605 DF 32) signals.

Total RNA was isolated from entire wild-type and sns-D plants or leaves with an RNeasy Plant Mini Kit (Qiagen). Residual DNA was removed from the RNA samples with DNase I (Ambion) in the presence of the RNase inhibitor RNasin (Promega). cDNA was produced on 1 μg of RNA using iScript cDNA synthesis kit according to the manufacturer's instructions (Bio-Rad). Quantitative real-time PCR (Q-PCR) analyses were done in triplicate using the iQ™ SYBR^® Green Supermix (Bio-Rad). Specific fragments (about 200 bp) were amplified of the genes flanking the T-DNA insertion site: At1g21550 (primers SP364 and SP365) and At1g13020 (primers Q44 and Q45) using a DNA Engine Thermal Cycler (MJ Research) equipped with a Chromo4 real-time PCR detection system (Bio-Rad). The sequences of the primers are listed in Table 1. The cycling parameters were 95°C for 3 min, 40 cycles of (95°C for 1 min, 60°C for 40 s), 72°C for 10 min. Normalization of relative gene expression was based on expression of the housekeeping gene Roc1 (At4g38740; primers ROC3.3, ROC5.2) and were presented as relative expression ratios. The value of the wild-type was set on 1. Standard deviations (SD) for the expression levels in the sns-D mutant were shown as error bars.

For northern blot analysis, RNA samples (10 μg) were electrophoretically separated in a 1.5% (w/v) agarose gel containing 2% (v/v) formaldehyde, 20 mM 3-(N-morpholino) propanesulfonic acid, 8 mM sodium acetate, 10 mM EDTA, pH 7.0; blotted with 50 mM Na-phosphate buffer pH 6.5 and 5 mM EDTA onto hybond-N (Amersham Biosciences) membranes and hybridized with ³²P-labeled cDNA probes as described previously (Menke et al., 1999). To generate radioactively labeled [32P]dCTP probes DNA fragments were amplified for At1g13020 (primers GG7 and GG12) and Roc1 (primers ROC3.1 and ROC5.1). The last probe was used to check equal loading of RNA. The corresponding cDNA fragments were purified from agarose gels using Qiaquick Gel Extraction kit (Qiagen). cDNA probes were labeled with Prime a Gene labeling system kit (Promega) and purified by use of ProbeQuant G-50 micro columns (Amersham Biosciences). The bands on the blots were quantified using the software program Quantity One (Bio-Rad). Normalization of relative gene expression was based on band intensities of the housekeeping gene Roc1 and was presented as relative expression ratios. For each developmental stage the value of the wild-type was set on 1.

For Southern blotting, 5 μg DNA of the sns-D mutant was digested with HindIII and separated in 0.7% agarose gel along a dig-labeled Lambda EcoRI/HindIII marker, blotted onto Hybond-N and hybridized in DIG easy hyb (Roche) supplemented with 50 μg/ml herring sperm DNA with a hpt probe, labeled in a PCR reaction with primers MC141 and MC142 (Table 1) using DIG-labeling mix (Roche). After 16–20 h, blots were washed twice with 2 × SSC; 0.1% SDS at room temperature and three times with 0.2 × SSC; 0.1% SDS at 65°C. Detection was performed using the DIG wash and block buffer set and CDP-star (Roche), according the manufacturers protocol.

cDNA was produced from total wild-type RNA and using primers Q28 and Q29 the At1g13020 coding region was amplified and cloned in pJET1.2 (Fermentas). After sequence confirmation the EcoRI–HindIII fragment was cloned in pHANNIBAL (Helliwell and Waterhouse, 2003) and the SacI–XbaI fragment from the resulting plasmid was cloned in pCambia1302 (http://www.cambia.org). The resulting binary vector was introduced in Agrobacterium tumefaciens strain AGL1 (Lazo et al., 1991) by electroporation and the strain was used for floral dip transformation to produce transgenic plants overexpressing At1g13020.

Significance of the differences in mean values of the specific caspase-3 or caspase-6 like activities and the number of stomata were tested with the Student's t-test.

A pool of transgenic Arabidopsis plants harboring the activator T-DNA construct pSDM1550 was previously screened for developmental mutants (van der Graaff and Hooykaas, 1998). One of the T-DNA tagged transgenic lines is the Arabidopsis sns-D mutant, which produces spontaneous necrotic spots on the rosette leaves, even under aseptic conditions. Southern analysis demonstrated that only one T-DNA was integrated (Figure 1). The sequences flanking the right border of the insert were obtained by plasmid rescue employing the carbenicillin resistance gene present in the pUC9 vector sequences or the kanamycin gene of the binary vector. Interestingly, sequence analysis revealed that the T-DNA integrated together with the complete binary vector sequence resulting from T-strand formation initiated on the left border, followed by right border skipping (van der Graaff et al., 1996; line 17 is the sns-D mutant). Analysis of the 2 kb of plant DNA that flanks the integrated binary vector sequence showed that the T-DNA and binary vector sequences inserted in the 3′UTR of the gene At1g13020 (Figure 2) that encodes the eukaryotic initiation factor eIF4B2. In order to obtain the plant DNA sequences flanking the left border of the T-DNA, the genomic DNA sequences from DNA of hygromycin resistant plants were amplified by TAIL PCR, cloned, and sequenced. This revealed that the left border of the T-DNA was linked to the region between the genes At1g21550 and AT1g21560, suggesting that some chromosomal rearrangement had occurred.

In order to obtain more evidence for chromosomal rearrangements, FISH technique was performed (Figure 3). Two adjacent BAC clones, BAC F3F19 (magenta signal) and BAC F13K23 (green signal), were used as probes, of which BAC F3F19 overlaps the integration site in At1g13020. Wild-type nuclei (Figure 3A) showed continuous single magenta and green signals on each of the chromosomes. However, sns-D mutant nuclei showed continuous magenta and green signals only on one chromosome and discontinuous magenta and green signals on the other chromosome (Figures 3B,D), suggesting that a chromosomal rearrangement took place. In pachytene chromosomes which are fully paired and display a well-differentiated pattern, we observed an extra magenta signal (Figures 3C,E), indicating that indeed a region hybridizing with BAC F3F19 had been rearranged. The proposed structure of chromosome 1 of the sns-D mutant is shown in Figure 3F.

The T-DNA contained a hygromycin phosphotransferase (HPT) gene that confers resistance to the antibiotic hygromycin. Segregation of hygromycin resistance in the next generation occurred at a ratio 2:1 (291 resistant: 144 sensitive). PCR analysis showed that all hygromycin resistant plants were heterozygous. Analysis of siliques produced by hygromycin resistant plants showed empty spaces representing aborted embryos in the siliques (Figure 4), suggesting that homozygous T-DNA inserts in the sns-D mutant plants are lethal.

The sns-D seedlings displayed a specific phenotype first visible after 2 weeks of growth on 0.5 MS medium in the culture chamber, consisting of the spontaneous formation of necrotic spots on the rosette leaves. It was not possible to differentiate between wild-type and mutant seedlings at an earlier stage. As shown in Figure 5, neither necrotic spots nor differences in morphology were visible on cotyledons or the first leaves for 1-week-old seedlings. After 2 weeks, sns-D seedlings started to produce necrotic spots (Figures 5G,H) on their oldest leaves, whereas this was never observed for wild-type seedlings cultured under the same conditions (Figures 5E,F). In addition, sns-D plants remained smaller than wild-type plants. The severity of the sns-D mutant phenotype increased throughout development and the differences between wild-type (Figures 5I,J) and the sns-D mutant (Figures 5K–N) were more pronounced after 5 weeks of growth. Two classes were distinguished among the sns-D plants, which both produce necrotic spots on the leaves and are smaller than the wild-type: the weak phenotype with leaves which are still green (Figures 5K,L) and the strong phenotype consisting of small and coiled leaves, which are rather necrotic (Figures 5M,N). Many of the sns-D plants with the strong phenotype died before an inflorescence could be formed or they produced a very small inflorescence. The sns-D plants with the weak phenotype formed bigger flowers, but after fertilization the siliques were smaller (Figures 5O–R). To determine whether the severity of the sns-D phenotype was determined by the parental origin of the T-DNA, reciprocal crosses were performed with wild-type plants. Crosses performed with the sns-D mutant as pollen donor as well as those with the sns-D mutant as ovule acceptor produced progenies that segregated for the strong and weak mutant phenotype. Thus, the severity of the sns-D phenotype could not be explained by the parental origin of the mutation (father or mother). Nevertheless, the germination frequency was reduced to 7% when the sns-D mutation was transmitted through the female gamete.

To further characterize developmental alterations caused by the sns-D mutation, analysis of rosette leaves was performed with scanning electron microscopy. No differences between the wild-type and the sns-D mutant were observed in their early stages of development. The adaxial and abaxial sides of the cotyledons of 1-week-old wild-type (Figures 6A–C) and sns-D seedlings (Figures 6J–L) were similar. However, analysis of 5-week-old seedlings showed that patches of epidermal cells of the sns-D mutants (Figures 6M–R) were flattened on both sides of the leaves. This was not observed in wild-type seedlings (Figures 6D–I). In cross-sections of the leaves of sns-D seedlings the collapsed cells in the necrotic regions were clearly visible (Figure 7).

The sns-D mutant produced smaller leaves. The size of the pavement cells was reduced resulting in two times more cells per surface area (Figure 8). The number of stomata was also determined on the adaxial side of rosette leaves from 5-week-old plants (Table 2). Compared to the wild-type, the sns-D mutant produced 1.5 times more stomata per surface area on the adaxial side of the leaves (2.9 ± 0.5/10⁻⁸ m² and 4.5 ± 1.0/10⁻⁸ m², respectively). Thus, the total number of pavement cells and stomata remained nearly unaltered, but the cell size was reduced in the sns-D mutant leaves.

Analysis of the mutant leaves suggested that both the epidermal and the internal cell layers of the necrotic spots contain dead cells. Cell death can occur by necrosis or by PCD. The latter is characterized by an activation of caspase-like proteases and specific DNA cleavage. We assayed for caspase-3 and caspase-6 like activities in leaf extracts by use of the synthetic fluorogenic caspase-3 substrate Ac-DEVD-AMC and the synthetic fluorogenic caspase-6 substrate Ac-VEID-AMC, respectively. Using these substrates, caspase-like activities have previously been found in plant cells (del Pozo and Lam, 1998; Sun et al., 1999; Lam and del Pozo, 2000; Bozhkov et al., 2004; Chichkova et al., 2010; Xu and Zhang, 2010). Because of the increasing severity of the sns-D phenotype throughout development, caspase activity was determined in leaf extracts from 5-weeks-old plants. An increase of the caspase-3-like activity was measured in leaf extracts from the sns-D mutant (Figure 9, gray bars). The caspase-3 like activity was about two-fold higher in the weak phenotypic class and about four-fold higher in the strong phenotypic class as compared to the wild-type control. A similar observation was made for caspase-6 like activity (Figure 9, black bars). Caspase-6 like activity was about 1.5- and 4-fold higher in the weak and strong phenotypic class, respectively. Thus, sns-D plants exhibit higher caspase-3 and caspase-6 like activities, indicative of PCD, with the highest activities in the strongest phenotypic mutant class.

DNA cleavage, another hallmark of PCD was visualized by the TUNEL assay. In wild-type rosette leaves TUNEL-positive nuclei were not detected (Figures 10A–D), whereas pre-incubation with DNase I prior to the TUNEL reaction induced DNA cleavage throughout the leaf (Figures 10E–H). The sns-D mutant showed TUNEL-positive nuclei in the cells located in the necrotic spots, whereas no TUNEL signal was detected outside the necrotic spots (Figures 10I–L). With transmitted light the necrotic spots appeared as a brown zone of collapsed cells. Sytox orange was used to detect all nuclei in cross-sections of sns-D leaves. Pre-incubation with DNase I prior to the TUNEL reaction induced DNA cleavage throughout the leaf (Figures 10M–P). To conclude, the in situ detection of DNA fragmentation by TUNEL analysis of the leaves showed that many cells in the necrotic spots contain fragmented DNA, one of the hallmarks of PCD.

Due to a chromosomal rearrangement of chromosome 1, the T-DNA and the binary vector were inserted flanking the genes At1g21550 and At1g13020 (eIF4B2). Since the T-DNA is an activator tagging construct that contains the strong 35S enhancer, it is possible that the sns-D mutant phenotype is caused by increased expression (activation) of one or more genes flanking this T-DNA insert. The expression levels of the flanking genes were determined with Q-RT-PCR in both phenotypic classes of sns-D plants and compared with the expression levels in wild-type plants (Figure 11A). The expression levels of both At1g13020 and At1g21550 were higher in sns-D plants. The level of expression of At1g13020 correlated with the severity of the phenotype, whereas At1g21550 expression did not. Northern blot analysis showed that the expression of At1g13020 was higher in 2- to 5-week-old sns-D seedlings than in wild-type seedlings of the same age. In addition, the expression level was higher in sns-D plants and leaves with the strong phenotype than in plants and leaves with the weak phenotype (Figure 11B). Thus, the Atg13020 expression levels correlated with the timing and severity of the developmental defects that were observed in the sns-D mutant plants.

To confirm that the sns-D mutant phenotype was indeed caused by the activation tagging of At1g13020, resulting in increased eIF4B2 expression levels, plants were produced with the eIF4B2 coding region under transcriptional control of the constitutive CaMV 35S promoter. Transgenic plants were obtained, some having necrotic spots on their leaves (Figure 12A) and some having a very severe phenotype with large necrotic regions on their leaves that died before setting seeds. Plants that produced seeds were further analyzed. In the next generation most plant lines showed different size classes, like the sns-D mutant (Figure 12B).

Progeny of four plant lines were analyzed by Q-RT-PCR to determine the expression level of eIF4B2. These lines showed a 2.5- to 4-fold increased expression level, which was somewhat lower than the 4.9- and 5.5-fold increased expression level in sns-D plants with the weak and strong phenotype, respectively (Figure 12C). Note that the expression level in the sns-D mutant with the weak phenotype was somewhat higher in this experiment compared to the experiment of Figure 11. Subsequently, the overexpression lines were grown on soil along with the wild-type and the sns-D mutant. The phenotypic characteristics were comparable with those of the sns-D mutants with a weak phenotype (Figure 13) and accordingly the severity of this phenotype increased in older plants. The rosettes were small and bushy and the leaves were curled with necrotic spots or areas. Inflorescences appeared late and were small with little flowers. Since cell size of the sns-D leaves was reduced and the number of stomata was increased, we determined the cell size and the number of stomata on the leaves of the overexpression lines (Table 2). Indeed, the size of the pavement cells was reduced (Figure 8) resulting in more stomata per surface area on the adaxial side of these leaves compared to the wild-type and the number was in between the number on the wild-type leaves and that seen on the sns-D leaves. Thus, increased eIF4B2 expression levels caused phenotypic alterations that mimicked those detected in the sns-D mutant, indicating that the sns-D mutant phenotype is caused by activation tagging of eIF4B2.