The TUMOROUS SHOOT DEVELOPMENT2 gene of Arabidopsis encoding a putative methyltransferase is required for cell adhesion and co-ordinated plant development

Authors


(fax + 49 30 838 54345; e-mail tschmue@zedat.fu-berlin.de).

Summary

Mutations in the TUMOROUS SHOOT DEVELOPMENT2 (TSD2) gene reduce cell adhesion, and in strongly affected individuals cause non-coordinated shoot development that leads to disorganized tumor-like growth in vitro. tsd2 mutants showed increased activity of axial meristems, reduced root growth and enhanced de-etiolation. The expression domains of the shoot meristem marker genes KNAT1 and KNAT2 were enlarged in the mutant background. Soil-grown tsd2 mutants were dwarfed, but overall showed morphology similar to that of the wild-type (WT). The TSD2 gene was identified by map-based cloning. It encodes a novel 684 amino acid polypeptide containing a single membrane-spanning domain in the N-terminal part and S-adenosyl-l-methionine binding and methyltransferase domains in the C-terminal part. Expression of a TSD2:GUS reporter gene was detected mainly in meristems and young tissues. A green fluorescent protein-tagged TSD2 protein localized to the Golgi apparatus. The cell-adhesion defects indicated altered pectin properties, and we hypothesize that TSD2 acts as a pectin methyltransferase. However, analyses of the cell-wall composition revealed no significant differences of the monosaccharide composition, the uronic acid content and the overall degree of pectin methylesterification between tsd2 and WT. The findings support a function of TSD2 as a methyltransferase, with an essential role in cell adhesion and coordinated plant development.

Introduction

Plant morphogenesis involves a tight coordination of the spatial and temporal organization of cell division, cell expansion and cell differentiation. The varying rates of division and the specific differentiation accounting for the unique features of each organ must be precisely synchronized to perform a normal developmental program (Steeves and Sussex, 1989). Orchestration of these events requires the exchange of signals via signaling molecules, including phytohormones and macromolecules, and may depend on apoplastic routing or symplastic domains (Gisel et al., 1999; Haywood et al., 2002).

One important factor in plant development is the cell wall, which imposes numerous spatial and social constraints on the plant cell. The cell walls of higher plants provide mechanical strength, define the cell shape and the overall plant morphology, participate in cell–cell communication, and protect from attacks by pathogens and predators (Somerville et al., 2004). To correctly coordinate the planes of cell divisions and cell expansion in development, the ordered deposition of cell-wall material and its composition are important (Baskin, 2001). In addition to contributing to the mechanical aspects of plant development, the cell wall also possesses inherent signaling properties. Evidence is accumulating indicating that the plant cell wall is capable of triggering multiple signaling pathways through which growth responses can be coordinated or altered appropriately (reviewed by Pilling and Höfte, 2003). A number of genes involved in cell-wall formation have been identified using mutants with altered cell morphology and a defect in plant development (see for example Zuo et al., 2000; Scheible and Pauly, 2004, and references therein).

Plant cell walls are primarily composed of polysaccharides including cellulose microfibrils and matrix components (Bacic et al., 1988; McNeil et al., 1984). One of the matrix components is pectin, which is highly negatively charged and tends to form gel-like structures (Jarvis, 1984). Pectin is the most abundant and structurally most complex matrix polysaccharide in dicots, where it is largely restricted to the primary cell walls and is thought to have a major role in ensuring cell adhesion (Willats et al., 2001). Chemically, pectin is a mixture of heterogeneously branched and highly hydrated polysaccharides (reviewed by Ridley et al., 2001). The main pectic polysaccharide is linear homogalacturonan (HGA), which consists of linear α-1,4-linked d-galacturonic acid (GalA), which is variably methylesterified at C6 (Bacic et al., 1988). Pectic polysaccharides are synthesized in the Golgi apparatus (Delmer and Stone, 1988;Driouich et al., 1993), and a substantial portion of HGA is secreted in its methylesterified form (Lennon and Lord, 2000; Li et al., 1997, 2002). The degree of methylesterification of pectin is important in determining the adhesive properties of pectin. It is controlled by the activity of pectin methyltransferases, which incorporate methyl groups into pectin in the Golgi apparatus (Goubet and Mohnen, 1999b; Vannier et al., 1992), and pectin methylesterases, which demethylate pectins in the apoplast (Li et al., 2002). Several pectin methylesterase genes have been identified and investigated (Gaffe et al., 1997; Tieman et al., 1992; Wen et al., 1999). Pectin methyltransferase activity has been described for several plant species and the enzyme properties have been characterized (Goubet and Mohnen, 1999a; Goubet et al., 1998; Ishikawa et al., 2000; Kauss and Hassid, 1967), but no pectin methyltransferase gene has been identified to date.

We have previously performed a mutant screen to identify genes that are essential for coordinated plant development and normally prevent disorganized plant growth (Frank et al., 2002). Several mutants showing tissue de-differentiation and hormone-independent callus formation were identified. Shoots of the recessive tumorous shoot development (tsd) mutants frequently develop disorganized tumor-like tissue instead of organized leaves and stems. Here we report further analysis of the growth and development of the tsd2 mutant and identification of the TSD2 gene by map-based cloning. The TSD2 protein is predicted to function as a methyltransferase, and we hypothesize that it is involved in pectin modification. The results support the important role of the cell wall in plant development.

Results

Phenotypic classes of the tsd2 mutants grown in vitro

We have characterized the in vitro growth of mutants carrying the tsd2-1 allele (called tsd2 mutants hereafter) in more detail. tsd2 plants grown on standard MS medium showed large differences in expressivity of the phenotype, and progenies differed in terms of penetrance of the phenotype. Four phenotypic classes were distinguished. Plants of class I mutants were very small, and only leaf-like bulges were randomly and abundantly produced at the shoot apex (Figure 1a,b). Leaves did not develop further and finally a disorganized callus was formed. Accumulation of anthocyanins, degeneration of the cotyledons and eventually decreased viability were further characteristics of class I mutants. Class II mutants were more robust, with larger and curly leaf blades and short broad petioles (Figure 1a,c). Cell projections protruded from uneven leaf surfaces and led to callus formation around petioles and eventually on leaf blades (Figure 1c). These mutants also showed an accumulation of anthocyanins (Figure 1c). During later development, class II mutants tended to lose chlorophyll and turned brown at a time when the wild-type (WT) was still green (Figure 1a). Class III mutants nearly resembled WT plants (Figure 1a). The leaf blades developed relatively normally and the leaf proportions were similar to those of WT (Figure 1a,d). However, in contrast to WT, unorganized cell clumps formed on the petiole surface of the class III mutants (Figure 1d). Axillary shoot meristems were not visible in WT 14 days after germination (DAG) but they appeared in the leaf axils of class II and III mutants at this stage (Figure 1d,f). Accumulation of anthocyanin was less visible in this mutant class. The fourth class of mutant plants did not differ macroscopically from WT plants.

Figure 1.

 Penetrance and expressivity of the tsd2 phenotype in vitro.
(a) Development of various phenotypic classes of tsd2 compared to wild-type (WT) 21 days after germination (DAG). From left to right: WT, tsd2 class I (indicated by the arrow), tsd2 class II and tsd2 class III.
(b) Close-up of a tsd2 class I mutant 21 DAG.
(c) Callus formation around the leaf petiole of a tsd2 class II mutant.
(d) Cell clumps on the petiole surface of a tsd2 class III mutant.
(e,f) Axillary meristems of WT (e) and the activated meristems (indicated by arrows) of a tsd2 mutant (f) at 14 DAG.
(g,h) Penetrance of the tsd2 callus-like phenotype in homozygous progeny grown on hormone-free MS medium (g) or on MS medium containing cytokinin (15 μm iP) (h) at 35 DAG.
Scale bars = 5 mm (a,e,f), 1 mm (b–d) and 0.5 mm (g,h).

The penetrance of the callus-like phenotype (class I) varied between progenies. For example, among nine different F2 progenies originating from a single homozygous tsd2 mutant, between 65 and 100% of the seedlings developed the class I phenotype in vitro. Independent F3 progenies of these nine lines again showed variable penetrance, between 28 and 65%, of the class I phenotype. Penetrance and expressivity were influenced by several factors, including sucrose and hormones. For example, in one population, 49% (n = 97) of the mutants exhibited a callus-like class I and class II phenotype when grown on standard MS medium, while 96% (n = 129) of the mutants exhibited this phenotype on MS medium containing cytokinin (Figure 1g,h). In contrast to the highly proliferative callus-like mutants, the growth of WT-like mutants and WT seedlings was inhibited on medium containing cytokinin (data not shown). It is possible that the expressivity and penetrance of the tsd2 mutation may be influenced by as yet unknown additional external or internal factors.

Root development of the tsd2 mutant

In general, root development of the tsd2 mutant seems to be less affected than shoot development. In order to investigate this further, tsd2 root growth and the formation of lateral roots were recorded over a 14 day period after germination in a homozygous tsd2 line grown on standard MS medium in the light. Figure 2(a) shows that the root length of tsd2 mutant seedlings at 4 DAG was about 30% of the WT root length (P < 0.001). This difference remained constant over the whole test period. Additionally, the tsd2 mutant forms significantly fewer lateral roots than WT (P < 0.001; Figure 2b). At 11 DAG, tsd2 mutants had formed approximately one-third of the lateral roots formed by WT (tsd2 3.1 ± 2.7, n = 57; WT 9.6 ± 5.6, n = 60). In conclusion, the tsd2 mutation impairs growth of the primary root and the formation of lateral roots.

Figure 2.

 Root development of the tsd2 mutant.
(a) Elongation of the primary root of seedlings grown in vitro on MS medium at various days after germination (DAG).
(b) Number of lateral roots 11 DAG. Bars indicate SD (n > 55).

Development of the tsd2 mutant in the dark

The stunted stature of the tsd2 mutants grown in vitro and on soil (see below) indicates that the tsd2 mutant might have a defect in cell elongation. Therefore, tsd2 hypocotyl elongation was tested in the dark, where growth depends mainly on cell elongation (Gendreau et al., 1997). The length of tsd2 hypocotyls grown for 6 days in the dark on a sugar-free MS medium was only 18% (1.5 ± 0.4 mm, n = 48, P < 0.001) of the WT hypocotyl length (8.6 ± 1.7 mm, n = 22) (Figure 3a). The epidermal cells of the tsd2 mutant hypocotyls were significantly shorter and broader than those of the WT (Figure 3b,c, and data not shown). In addition, the cell-adhesion defect described by Frank et al. (2002) was visible on the hypocotyls of dark-grown mutants. Cells of the WT hypocotyls were tightly attached and arranged in an orderly manner (Figure 3b). In contrast, cells that were pulled apart from each other and had lost contact or were even peeled off the hypocotyl surface were observed in the tsd2 mutant seedlings (Figure 3c).

Figure 3.

 Development of the tsd2 mutant in the dark.
(a) WT (two seedlings on the left) and tsd2 (two seedlings on the right) grown in the dark on sugar-free MS medium at 6 DAG.
(b,c) Scanning electron microscopy of the hypocotyl of WT (b) and tsd2 (c) grown in the dark (4 DAG).
(d) WT (two seedlings on the left) and tsd2 (two seedlings on the right) grown in the dark on MS medium (1% sucrose) at 14 DAG.
(e) Magnification of a WT shoot shown in (d).
(f) Magnification of a tsd2 shoot shown in (d).
(g) Roots of the WT seedlings (left) and tsd2 mutants (right) shown in (d).
Scale bars = 5 mm (a,d,e), 30 μm (b,c) and 2 mm (f,g).

Further analysis of dark development revealed additional symptoms of enhanced de-etiolation in tsd2 mutants. Almost all tsd2 plants (98%, n = 48) formed no apical hook after 6 days of growth on sugar-free medium in the dark, while only 2% (n = 42) of WT plants exhibited the same feature. The de-etiolated phenotype of the tsd2 mutant was more pronounced when plants were grown on 1% sucrose MS medium (Figure 3d). At 14 DAG on this medium, WT seedlings showed a long hypocotyl, open unexpanded cotyledons with long petioles, and on most seedlings the first true unexpanded leaf was visible at the shoot apical meristem as a small projection with a light yellow color (Figure 3d,e and Table 1). In contrast, tsd2 mutants exhibit short hypocotyls, open cotyledons with short petioles, and, on average, three partially expanded true leaves had appeared at the shoot apical meristem (Figure 3d,f and Table 1). The roots of tsd2 seedlings grown in the dark were twice as long as WT roots and had formed nine times more lateral roots (Figure 3g and Table 1).

Table 1.   Enhanced de-etiolation of the tsd2 mutant
 WTtsd2Percentage of WTP value
  1. Plants were grown for 14 days in the dark on MS medium containing 1% sucrose. To synchronize germination, sown seeds were kept for 3 days at 4°C in the dark and then exposed to light for 5 h in order to stimulate germination.

Root length (cm)1.8 ± 0.43.6 ± 1.2200<0.001
Number of lateral roots0.5 ± 0.74.5 ± 3.3900<0.001
Number of leaves0.6 ± 0.62.6 ± 0.8430<0.001
Hypocotyl length (cm)2.3 ± 0.30.4 ± 0.217<0.001

Expression of shoot meristem marker genes is altered in the tsd2 mutant

In order to explore which tissue contributes to the over-proliferaton of tsd2 shoots in vitro, we introgressed the shoot meristem marker genes KNAT1:GUS (Chuck et al., 1996) and KNAT2:GUS (Pautot et al., 2001) in the mutant background. Both marker genes stain the shoot apical meristem in WT seedlings, in particular cells of the L2 and L3 layers (Figure 4a,c) (Chuck et al., 1996; Pautot et al., 2001). tsd2 seedlings of classes I and II show an increased size of the apical expression domains of KNAT1:GUS and KNAT2:GUS, and the expression occasionally expanded into leaf petioles (Figure 4b,d). Enlargement of the zone containing cells with meristematic activity indicates that the spatial organization of their differentiation is disturbed in the tsd2 mutant. It is presumably the continued growth of these cells that contributes most to the tumorous phenotype of strongly affected mutants and their ability to grow continuously on hormone-free medium. However, the staining pattern also shows that tsd2 mutants have retained the ability to downregulate KNAT1 and KNAT2 gene expression, which is required for leaf formation (Byrne et al., 2000; Ori et al., 2000).

Figure 4.

KNAT1:GUS and KNAT2:GUS expression in the tsd2 mutant shows the increased size of the shoot meristematic zone.
(a) KNAT1:GUS expression in WT is seen mainly in the shoot apex.
(b) The KNAT1:GUS expression domain is enlarged in the tsd2 background (class I mutant).
(c) KNAT2:GUS expression in WT is seen mainly in the shoot apex.
(d) The KNAT2:GUS expression domain is enlarged in the tsd2 background (class II mutant). Seedlings were stained and photographed 25 days after germination. Scale bars = 0.5 mm (a,b) and 1 mm (c,d).

Phenotype of soil-grown tsd2 mutants

We were interested in studying whether the variability of the phenotype and penetrance as seen in vitro would be similar in soil-grown plants. At 7 DAG, a similar frequency of germination was seen for tsd2 seedlings on soil (95%, n = 210) and on MS medium (98%, n = 210). Lethality of the tsd2 mutant at the seedling stage was lower than expected from the penetrance of the class I callus phenotype detected in the same lines in vitro. The majority of tsd2 mutants grew on soil without difficulties, although tsd2 seedlings were typically smaller than WT (Figure 5a,b). About 10% of the seedlings did not grow beyond the cotyledon stage and eventually died (Figure 5c). Callus formation was occasionally observed among these seedlings (Figure 5c). Developmental defects were seen in 13% of the tsd2 seedlings, for example drying of cotyledons or growth retardation (Figure 5d). However, these plants were usually able to recover and to flower. At the flowering stage, tsd2 plants were shorter than their WT counterparts (Figure 5e).

Figure 5.

 Phenotype of soil-grown tsd2 mutants.
(a) WT seedling at 14 days after germination (DAG).
(b–d) Varying expressivity of the tsd2 phenotype in soil-grown seedlings at 14 DAG: (b) weak phenotype, (c) strong phenotype with callus formation on the cotyledons, (d) intermediate phenotypes; the inset is a magnification of a seedling that showed growth arrest after germination.
(e) Comparison of WT (left) and tsd2 plants at 42 DAG.
(f) Rapid dehydration of tsd2 leaves. Rosette leaves (approximately 200 mg) were collected from 15-day-old soil-grown seedlings, and their weight was determined at various time points after harvest. Bars indicate the SD (n = 4).
Scale bars = 2 mm (a–d).

During experimental manipulations, we noted rapid wilting of detached tsd2 rosettes. Figure 5(f) shows that detached rosettes of the tsd2 mutants lost water almost twice as fast as rosettes of WT plants. Three hours after detachment, the tsd2 rosettes had lost about 73% of their FW, while WT rosettes lost only 44% (Figure 5f). We conclude that tsd2 mutants have a decreased water-holding capacity.

Positional cloning of the TSD2 gene

Fine mapping using 1274 F2 individuals of a cross between tsd2 (in the Col-0 background) and Ler located the TSD2 gene on the two BAC clones T11I11 and F3F9, approximately 120 cM from the top of chromosome 1 (Figure 6a). The left border of the 34.9 kb mapping interval was defined by six recombinants detected with a marker on BAC T11I11, and the right border was defined by two recombinants detected on the F3F9 BAC clone. The final mapping interval of 21.5 kb contained three predicted genes. Sequencing analysis revealed a single nucleotide mismatch G→A at position 629 in one of these, At1g78240 (Figure 6a). The mutation creates a new restriction site for endonuclease FspBI, which enables easy identification of the mutant allele (Figure 6b). The mutation changes the codon for tryptophan at amino acid position 210 into a stop codon (Figure 6c). The TSD2 cDNA was amplified by PCR from an Arabidopsis seedling cDNA library. The sequence corresponds to an open reading frame of 2055 bp. Transformation of tsd2 mutants with a T-DNA harboring the TSD2 cDNA under the control of the CaMV 35S promoter fully complemented the mutant phenotype in more than 20 independent transgenic lines (Figure 6d, and data not shown). We therefore conclude that the isolated gene corresponds to TSD2.

Figure 6.

 Positional cloning of the TSD2 gene.
(a) Genetic map position of the TSD2 gene in relation to other known phenotypic and molecular markers. The TSD2 gene was localized to a small region on BAC clone F3F9 on chromosome 1. The genomic sequence of the TSD2 coding region comprises eight exons and seven introns. The tsd2-1 mutation causes a G→A transition near the end of the first exon (indicated by an arrow). Black boxes in the TSD2 gene diagram indicate exons. Lines between exons indicate introns, and lines beyond exons indicate untranslated regions. The numbers above the introns refer to the intron sizes. Chr, chromosome.
(b) Genotyping of the tsd2-1 mutant allele. The single nucleotide exchange caused by the mutation in tsd2-1 creates a cleavage site for the FspBI restriction endonuclease. Digestion of PCR-amplified DNA fragments with FspBI shows that the FspBI site is present in the tsd2 mutant but absent in the WT.
(c) Nucleotide and amino acid sequences adjacent to the tsd2-1 mutation. The tsd2-1 mutation changes the WT codon encoding Trp210 into a stop codon indicated by an asterisk.
(d) Complementation of the tsd2 mutant. Phenotype of dark-grown seedlings at 5 DAG. From left to right: WT, tsd2-1 mutant, and a transgenic homozygous tsd2-1 mutant harboring a 35S:TSD2 gene.

Analysis of a second tsd2 mutant allele (tsd2-2; Frank et al., 2002) revealed that this mutant was due to a deletion encompassing the TSD2 locus (data not shown). As tsd2-1 and tsd2-2 caused similar phenotypic changes, we conclude that the mutant phenotype described here is due to a null mutation.

Sequence analysis of the TSD2 gene and its encoded protein

Alignment of the cDNA with genomic DNA revealed the presence of eight exons and seven introns in the translated region (Figure 6a). The cDNA encodes a polypeptide of 684 amino acids, with a predicted molecular weight of 77.9 kDa and a pI of 7.34. The transmembrane hidden Markov model (TMHMM) prediction server, version 2.0 (http://www.cbs.dtu.dk/services/TMHMM-2.0/), predicted that TSD2 is a type II transmembrane protein that comprises a short cytoplasmic N-terminus followed by a single transmembrane helix and a long non-cytoplasmic C-terminus (Figure 7a,b). A search for structural domains revealed the presence of an S-adenosyl-l-methionine binding motif and a plant-specific putative methyltransferase domain DUF248 (domain of unknown function) (Figure 7a,b). This classifies TSD2 as a member of the superfamily of S-adenosyl-l-methionine-dependent methyltransferases.

Figure 7.

 TSD2 protein structure, sequence alignment and phylogenetic analysis.
(a) TSD2 protein structure. The dashed box indicates the predicted N-terminal transmembrane domain (TMD). The predicted protein topology, i.e. cytoplasmic part (cyt) and luminal part (lum), is shown. The putative methyltransferase domain (MT) is shown in gray, and the S-adenosyl-l-methionine binding domain (SAM) is shown in black.
(b) Alignment of the deduced amino acid sequence of the TSD2 protein with the three closest homologous proteins from rice (Oryza sativa; XP_467861) and Arabidopsis (At1g13860 and At2g03480). Residues that are identical in at least three sequences are shaded in black. Conservative substitutions at an amino acid position that occur in three or more of the four sequences are shaded in gray. Arrows indicate the position of the transmembrane domain of the TSD2 protein. The line above the sequence marks the putative methyltransferase domain, and the dotted part of the line marks the S-adenosyl-l-methionine binding domain. The tsd2-1 mutation is located at the beginning of the methyltransferase domain in a conserved amino acid as indicated.
(c) Phylogenetic consensus tree of 29 members of the TSD2 family in Arabidopsis. The tree was constructed using the phylip program package (Felsenstein, 1989). One hundred cycles of bootstrap resampling were used and the bootstrap support values are shown for each branch.

A BLAST search in Arabidopsis revealed a family of 29 proteins that show high similarity to TSD2. Phylogenetic analysis revealed that many members of this gene family originate from recent gene duplications as they are grouped in pairs (Figure 7c). However, TSD2 has no close homolog. Figure 7(b) shows the alignment of TSD2 with the most closely related proteins, comprising a protein in rice (Oryza sativa, XP_467861; 51% identity) and two proteins from Arabidopsis (At1g13860 and At2g03480; 48 and 47% identity, respectively). The latter two proteins show a higher degree of identity with each other (81%) than with TSD2, and have a smaller N-terminal domain than TSD2 (Figure 7b).

Expression pattern of the TSD2 gene

Analysis of the TSD2 gene expression pattern using the Genevestigator database (Zimmermann et al., 2004) indicated the presence of its mRNA in almost all plant tissues. To explore the expression pattern further, we constructed a promoter–GUS reporter gene construct comprising aprroximately 2.6 kb of the TSD2 promoter. Six independent transgenic lines showed similar expression patterns. High GUS activity was found in the vasculature of the young seedlings and in the shoot and root apex (Figure 8a). The hypocotyls of light-grown seedlings were hardly stained, but promoter activity was clearly seen in hypocotyls of dark-grown seedlings (Figure 8b). Shoot and root meristems showed strong GUS activity over the whole growth period (Figure 8c,d). TSD2:GUS expression was also detected in the root and shoot vasculature and young leaves. The basal part of developing leaves was more strongly stained than the apical part, reflecting the apical–basal gradient of cell division activity (Figure 8c). TSD2:GUS was active in young reproductive tissues as well, with peak expression in the pedicel, receptacle and gynoecium, in the filament between pollen sacs in older flowers (Figure 8e,f), and in the abscission zone of siliques after seed maturation (Figure 8g).

Figure 8.

 Expression analysis of a TSD2:GUS reporter gene.
(a) Seedlings expressing TSD2:GUS in the shoot apex, the vasculature and the root apical meristem at 6 DAG.
(b) TSD2:GUS is expressed in the hypocotyl of dark-grown seedlings (6 DAG), with the greatest activity observed in the upper hypocotyl.
(c) TSD2:GUS expression is detected in the vasculature of the shoot and growing leaves, displaying higher activity in the basal part of the leaves.
(d) In roots, TSD2:GUS activity is localized in the vasculature and emerging lateral roots, with highest activity in the meristems.
(e,f) TSD2:GUS is expressed in young floral tissue, with high expression in the pedicel, receptacle and gynoecium (e). The activity decreases in older flower organs, where it was mainly observed in the filament between pollen sacs (f).
(g) TSD2:GUS expression in the abscission zone of mature siliques.
The duration of staining was overnight. Scale bars = 0.5 mm (a–f) and 2 mm (g).

Subcellular localization of the TSD2 protein

The TargetP 1.1 and signalp programs indicated the absence of a cleavable N-terminal signal sequence and excluded localization of TSD2 in the chloroplast, mitochondria, nucleus, vacuole, the ER and the secretory pathway. The Golgi prediction program (http://ccb.imb.uq.edu.au/golgi/golgi_predictor.shtml) suggested post-Golgi localization of TSD2. In order to analyze the subcellular localization experimentally, we fused green fluorescent protein (GFP) to the N-terminus of TSD2 and expressed the fusion protein in Arabidopsis. In hypocotyl cells of transgenic plants, the GFP–TSD2 signal was localized to small punctate structures, typical of the Golgi apparatus (Figure 9a), which were highly motile in the cytoplasm. To exclude the possibility that the punctate fluorescence was of mitochondrial origin, the mitochondria were labeled with MitoTracker Red. Mitochondria and GFP–TSD2 fluorescence did not co-localize (Figure 9b). To test further the identity of these fluorescent structures as Golgi stacks, we investigated their sensitivity to brefeldin A (BFA). BFA is known to have a variety of effects on the morphology of the Golgi apparatus, causing either clustering or vesicle formation, depending on the conditions and tissue studied (Klausner et al., 1992; Satiat-Jeunemaitre and Hawes, 1992, 1994; Staehelin and Driouich, 1997). Following treatment of samples with BFA, the pattern of GFP fluorescence within the cells changed substantially. Instead of the numerous small punctate structures noted in the absence of BFA, a few large aggregated fluorescent structures were observed (Figure 9c). Localization of the GFP–TSD2 fusion protein in a BFA-sensitive compartment strongly supports Golgi localization. To further confirm this finding, we transiently expressed the GFP–TSD2 protein together with the known Golgi marker GONST1–YFP (Baldwin et al., 2001; Handford et al., 2004) in tobacco epidermal cells. Both GFP–TSD2 and GONST1–YFP showed a punctate pattern in tobacco cells (Figure 9d,e). Overlay analysis of GFP–TSD2 and GONST1–YFP revealed co-localization of their fluorescence (Figure 9f), which is in accordance with Golgi localization of the GFP–TSD2 protein. As a control, we prepared a fusion protein (GFP–Δ104TSD2) that lacks the N-terminal TSD2 sequence containing the predicted membrane anchor. It has been shown that the transmembrane domain of Golgi proteins is essential for retention in the Golgi apparatus (Gleeson, 1998; Munro, 1995). Figure 9(g) shows that, in transgenic plants expressing the truncated chimeric protein, the GFP signal was detected in the cytoplasm. This result indicates that the deleted protein part of TSD2 contains information necessary for localization to the Golgi.

Figure 9.

 Subcellular localization of a GFP–TSD2 fusion protein in the Golgi apparatus.
(a–c,g) Subcellular localization of GFP–TSD2 fusion proteins in hypocotyls of stably transformed Arabidopsis plants.
(d–f) Co-localization of a GFP–TSD2 fusion protein and the Golgi marker GONST1–YFP in Nicotiana benthamiana.
(a) Optical section through an Arabidopsis hypocotyl cell expressing GFP–TSD2 shows its localization in punctate subcellular structures.
(b) Comparison of GFP–TSD2 localization and mitochondria stained with MitoTracker Red shows that GFP–TSD2 is not localized in the mitochondria.
(c) Clustering of GFP–TSD2 fluorescence after 30 min of brefeldin A treatment suggests its localization in the Golgi.
(d–f) Confocal scanning micrographs of lower epidermis cells of N. benthamiana collected from the same optical section: (d) GFP–TSD2 fluorescence, (e) fluorescence of Golgi-localized GONST1–YFP, and (f) merged image of (d) and (e) showing extensive overlap (yellow) of protein localization shown in (d) and (e), indicating that GFP–TSD2 is localized to the Golgi.
(g) Deletion of the N-terminal part of TSD2 in the GFP–Δ104TSD2 fusion protein causes localization to the cytoplasm.
Cells were visualized with a confocal laser scanning microscope. Scale bars = 10 μm (a–c,g) and 5 μm (d–f).

Cell-wall composition of the tsd2 mutant

The phenotype of the tsd2 mutant suggested that reduced cell-wall adhesion might be the primary cause of the mutant phenotype. In the cell wall, pectins have a principal function in cell adhesion, and cell-adhesion defects have been observed in several pectin mutants (Bouton et al., 2002; Iwai et al., 2002). Importantly, the degree of pectin methylesterification influences its adhesive properties (Bush and McCann, 1999; Liners et al., 1994; Willats et al., 2001). As the sequence analysis suggested that TSD2 may function as a methyltransferase, we investigated whether TSD2 is involved in methylesterification of pectin polysaccharides. We tested first whether the cell wall of tsd2 mutants differed in its sugar composition and/or methylester content. We choose hypocotyls of dark-grown seedlings for our analysis as this tissue shows a distinct phenotype (Figure 3) and expresses the TSD2 gene (Figure 8b). Cell-wall material was prepared from dark-grown hypocotyls of WT and tsd2 plants, and their monosaccharide composition determined (Table 2). Rhamnose, arabinose, xylose and galactose were the predominant neutral sugars, with lesser amounts of fucose, mannose and glucose. There was no significant difference in the relative sugar content between the WT and the mutant. Analysis of the methyl ester content of the hypocotyl wall material showed on average an approximately 26% lower content in the tsd2 mutant. However, because of the large variance, the reduction was not statistically significant (α = 0.05) (Figure 10).

Table 2.   Relative content of neutral sugars and uronic acids in hypocotyl cell walls
 WTtsd2
  1. The relative monosaccharide composition including uronic acids of cell-wall material from hypocotyls of 6-day-old dark-grown WT and tsd2 seedlings given as mol% (±SD, n = 3). Seeds were germinated and grown for 6 days in the dark on sugar-free MS medium before collection of hypocotyls. Rha, rhamnose; Fuc, fucose; Ara, arabinose; Xyl, xylose; Man, mannose; Gal, galactose; Glc, glucose; UA, uronic acids.

Rha8.0 ± 0.87.3 ± 1.1
Fuc0.7 ± 0.20.6 ± 0.4
Ara13.8 ± 1.615.2 ± 1.2
Xyl10.7 ± 0.99.4 ± 0.6
Man1.8 ± 0.21.5 ± 0.2
Gal20.8 ± 2.521.5 ± 3.7
Glc4.4 ± 1.15.3 ± 1.1
UA+39.7 ± 4.639.2 ± 5.0
Figure 10.

 Methylester content of cell-wall material derived from 6-day-old dark-grown hypocotyls.
White bar, WT; black bar, tsd2 mutant.

Discussion

TSD2 is required for normal plant development

Mutation of the TSD2 gene has revealed its functions in cell adhesion and plant development. Strongly affected tsd2 mutants showed an increase in the size of the shoot apical meristematic region, an apparent difficulty in properly organizing the formation of leaves, and ectopic cell divisions, indicating an important role for TSD2 in the coordination of cell division and differentiation in the young shoot tissue. This role was particularly evident in plants grown in vitro, where strong phenotypes resulted in tumor-like growth. It could be that once the shoot apex is disorganized, it becomes increasingly difficult to reorganize a properly functioning meristem, leading to continued disorganized growth. This may be enhanced by cell division factors such as cytokinin, which increased the penetrance of the phenotype. Interestingly, individuals that were able to overcome the difficulties during the initial growth phase continued to grow more or less normally. Later occurrence of disorganized growth was never observed, indicating that TSD2 is most important during early development, possibly during the transition from embryonic to vegetative growth. It could also be that compensatory changes are activated that mask later consequences of tsd2 mutation. Contrasting with the changes in the shoot meristem, organization and function in the root meristem do not seem to be strongly affected.

A cell-wall defect with reduced cell adhesion is a main characteristic of the tsd2 phenotype in hypocotyls, the shoot apex and young leaves (Figures 1 and 3) (Frank et al., 2002), and is probably the cause of the developmental defects. The shoot apex requires a particularly strong intercellular attachment to ensure functioning of the complex communications between cells required to maintain the meristem and to coordinate the formation of new cells and their correct development fates. Reduced cell adhesion will disturb communication between cells, which occurs through the apoplast and cytoplasmic connections, the plasmodesmata. For example, transcription factors such as KNOTTED and SHR, which are important in organ patterning, move between cells through plasmodesmata (Lucas et al., 1995; Nakajima et al., 2001; Zambryski and Crawford, 2000). Thus the maintenance of cellular connectivity is of fundamental importance to coordinate position-dependent behaviour of cells and is an important function of cell walls. Disruption of intercellular communication could cause the mis-oriented cell divisions that were seen in the L2 and L3 layers of the tsd2 shoot meristems (Frank et al., 2002) and that finally lead to failure of proper development.

Reduced cell adhesion may also alter mechanical restraints on cell expansion and the direction of cell growth, which may cause perturbance of the concerted action of many cells. Finally, it should be noted that biologically active oligosaccharides derived from cell-wall polysaccharides can act as short-range signals (reviewed by Aldington et al., 1991; Coté and Hahn, 1994; Ridley et al., 2001). Thus changes in signalling may be the consequence of an altered cell-wall structure or composition and may be involved in establishing the developmental defects of the tsd2 mutant.

The TSD2 protein is a putative methyltransferase

Sequence analysis of the TSD2 protein revealed a Pfam profile of a putative methyltransferase and an S-adenosyl-l-methionine binding domain. Methyltransferases have numerous substrates including DNA, proteins, various metabolites and pectic polysaccharides. The molecular function of TSD2 as a methyltransferase has not yet been demonstrated and its putative substrate is not known. Importantly, a GFP–TSD2 fusion protein was localized in a BFA-sensitive compartment and shown to co-localize with the known Golgi marker GONST1–YFP, strongly indicating that TSD2 is localized in the Golgi (Figure 9). In this compartment methyltransferase pectins were detected (Zhang and Staehelin, 1992) and pectin methyltransferase activity was found in several species (Vannier et al., 1992; Goubet et al., 1998; Goubet adn Mohnen, 1999b).

Several known cell-adhesion mutants have been linked to altered pectin metabolism. Similar to tsd2, qua1 seedlings show a variable phenotype and symptoms of de-differentiation in vitro, although to a lesser degree than WT, and form shorter hypocotyls in the dark. The QUA1 gene codes for a member of the GAUT1-related superfamily (Bouton et al., 2002; Sterling et al., 2006). Another member of this superfamily, GAUT1, has recently been shown to possess HGA galacturonosyltransferase activity (Sterlin et al., 2006). The qua1 mutant has a reduced content of HGA, and it was proposed that QUA1 is involved in HGA synthesis. Interestingly, TSD2 expression was well correlated with that of the QUA1 gene (Pearson’s correlation coefficient r2 = 0.541) (Toufighi et al., 2005). Another cell-adhesion-defective mutant is the tobacco nolac-H18 (non-organogenic callus with loosely attached cells) mutant, which has a defective rhamnogalacturonan II structure and carries an insertion in a putative glucuronosyltransferase. The inability of this mutant to produce shoots corroborates the relevance of the pectin network in shoot meristem functioning (Iwai et al., 2002). Antisense inhibition of a pectin methyl esterase, which removes methyl groups from pectin, affected tissue integrity during ripening of the tomato fruit (Tieman and Handa, 1994), and antisense repression of a specific root cap-expressed pectin methylesterase of pea lead to enhanced cell adhesion, a defect in the separation of so-called border cells, and arrest of root elongation (Wen et al., 1999).

Although the phenotypic analysis and comparison with other mutants suggests pectin modification as a primary function of TSD2, we have been unable to obtain a clear proof of its molecular function. Significant changes in monosaccharide composition, uronic acid and methylester content in hypocotyls of dark-grown seedlings and flower tissue were not found (Figure 10, and data not shown). Analysis of the qua2 mutant, which carries a mutation in the same gene as tsd2, has been published recently (Mouille et al., 2007). Analysis of the cell wall of qua2 revealed a 13% reduction of total galacturonic acid, and further analysis of purified pectin fractions showed that qua2 is specifically deficient in HGA. Mouille et al. propose that QUA2 functions as a pectin methyltransferase and synthesizes, together with a galacturonosyl transferase, such as QUA1, a specific portion of HGA. Consistently, the degree of methylesterification of the remaining HGA fraction in qua2 was found to be unchanged (Mouille et al., 2007).

Taken together, TSD2 is a newly characterized gene that is required for cell adhesion and coordinated plant development. The tsd2 mutant phenotype, the subcellular localization of the protein and the sequence information of the cloned gene are consistent with the proposal that TSD2 is a putative methyltransferase that modifies the pectin structure, and it is thus a missing link in the understanding of pectin biosynthesis. However, the exact nature of TSD2 function still needs to be elucidated. The mutant provides an opportunity to study further the links between cell wall and developmental processes.

Experimental procedures

Plant material and growth conditions

Isolation of two tsd2 mutant alleles has been described previously (Frank et al., 2002). The tsd2-1 allele originated from an ethyl methanesulfonate-mutagenized Col-0 gl1 mutant, while the tsd2-2 allele was identified in a population of En-1 transposon insertion lines, but was found not to be tagged (Frank et al., 2002). All work described here was carried out with the tsd2-1 allele. Unless stated otherwise, plants were grown in a growth chamber on soil or in vitro on 1× MS medium (Murashige and Skoog, 1962) containing 3% sucrose and 0.9% agar, pH 5.7, at 24°C and under a 16 h light/8 h dark cycle.

Dehydration experiment

For the dehydration experiment, rosette leaves were collected from 15-day-old plants grown in the greenhouse. Dehydration was measured by recording the weight of detached rosettes every 30 min for 3 h after their detachment from the root and exposure on the bench at 22°C. Four independent samples per genotype with an initial weight of approximately 200 mg were assayed in parallel. The result was expressed in terms of the weight that was lost during the observation period as percentage of the initial weight.

Gene mapping

Heterozygous mutant plants were crossed with WT plants of accession Landsberg erecta (Ler). A total of 1274 tsd2 mutant plants were selected from the F2 population. DNA was extracted using the CTAB method (Lukowitz et al., 2000) from individual F2 mutant plants, and analysed for recombination events using known SSLP and CAPS markers (Bell and Ecker, 1994; Konieczny and Ausubel, 1993; Lukowitz et al., 2000). Additional markers that are polymorphic between Col-0 and Ler were generated from single nucleotide polymorphisms and insertion/deletion polymorphisms (In Dels) listed in the Monsanto collection (Jander et al., 2002). Genetic distances were calculated according to the Kosambi function (Koornneef and Stam, 1992). Genotyping of the tsd2-1 allele was performed using the primers 5′-GCGTGATCTCGAGTTTCAGA-3′ (forward) and 5′-AAGCTTCGTAGTTTGCTATGCAC-3′ (reverse) for PCR amplification. The resulting product was digested with the FspBI restriction endonuclease and separated in an agarose gel.

Plasmid construction

cDNA containing the entire TSD2 open reading frame was amplified from a cDNA library (Bürkle et al., 2005) using gene-specific primers with 12 additional bases of attB1 and attB2 (Gateway cloning technology, Invitrogen; http://www.invitrogen.com/) at their 5′ ends: 5′-AAAAAGCAGGCTTCATGTCAATGCCACTACAACG-3′ (forward) and 5′- AGAAAGCTGGGTCTCAGATTGATTGTCGCTTGG-3′ (reverse). Full-length attB sites were introduced using adapter primers according to the manufacturer’s instructions. The resulting PCR amplification product was recombined via the Gateway BP reaction with donor vector pDONR221 to create entry clone pDON221-TSD2. The final expression clone carrying the CaMV 35S promoter was created by recombination of the pDON221-TSD2 entry clone with destination vector pB2GW7 (Plant Systems Biology, University of Ghent, Ghent, Belgium) via a Gateway LR reaction.

To obtain a promoter–GUS fusion gene, the promoter sequence of TSD2 starting 10 bp upstream of the translational start codon was amplified by PCR from DNA of A. thaliana Col-0. Primers were as follows: 5′-GGGGAACGTCTAGATCGGTGTAACCCGTTTGAAT-3′ (forward) and 5′-GGGAACGCCCGGGGCTGCTATAAATCTAAGATCCCTGT-3′ (reverse). The length of the amplified sequence was 2648 bp. The promoter sequence was inserted into vector pCBC308 (Xiang et al., 1999) in the unique XbaI and SmaI sites upstream of the GUS reading frame.

The 35S:GFP–TSD2 translational fusion gene was constructed by recombination of the pDON221-TSD2 entry clone with destination vector pK7WGF2 (Plant Systems Biology, University of Ghent, Ghent, Belgium) using LR clonase (Invitrogen). To construct the GFP–Δ104TSD2 fusion protein, the N-terminal sequence of TSD2 encoding the first 104 amino acids including the putative membrane-spanning domain was deleted by using the following primers with 12 bases of attB1 and attB2 sites for amplification, with pDON221-TSD2 as a template: 5′-AAAAGCAGGCTTCATTTCCATTTCGACTTCTTCC-3′ (forward) and 5′-AGAAAGCTGGGTCTCAGATTGATTGTCGCTTGG-3′ (reverse). The full-length attB sites were introduced as described above. The resulting PCR amplification product was recombined via the Gateway BP reaction with donor vector pDONR221 to create entry clone pDON221-Δ104TSD2. The final expression clone carrying the GFP–Δ104TSD2 fusion protein was created by recombination of the pDON221-Δ104TSD2 entry clone with destination vector pK7WGF2 (Plant Systems Biology, University of Ghent, Ghent, Belgium) via a Gateway LR reaction.

The plasmid containing the GONST1YFP construct has been described previously (Handford et al., 2004) and was kindly provided by Professor Paul Depree (University of Cambridge, UK). All plasmids were transformed into Agrobacterium tumefaciens strain GV3101.

Plant transformation

Arabidopsis thaliana Col-0 plants and tsd2 mutants were stably transformed by the flower dip method (Bechthold et al., 1993). Transgenic plants harboring the 35S:TSD2 construct were selected after surface sterilization of seeds on MS medium (Murashige and Skoog, 1962) containing 12 mg l−1 phosphinotricin. Arabidopsis plants containing promoter–GUS fusion genes were selected on soil by spraying with 0.1% BASTA (Hoechst AG, http://www.sanofi-aventis.de). Transgenic plants harboring GFP constructs were selected on MS medium containing 50 mg l−1 kanamycin.

For co-localization analysis, GFP–TSD2 and GONST1–YFP (Baldwin et al., 2001; Handford et al., 2004) were transiently expressed in 5-week-old Nicotiana benthamiana plants according to the method described by Romeis et al. (2001) with minor modifications. In brief, overnight Agrobacterium cultures were harvested by centrifugation. Cells were resuspended in 10 mm MgCl2, 10 mm MES, pH 5.6, and 100 μm acetosyringone to an OD600 of 0.3, and infiltrated with a syringe into lower epidermal cells of tobacco leaves. Transformed leaves were analysed 72 h after infection using a confocal laser scanning microscope.

Bioinformatic analyses

The protein sequence analyses were performed using TargetP, version 1.0 (http://www.cbs.dtu.dk/services/TargetP/) (Emanuelsson et al., 2000), and SignalP (http://www.cbs.dtu.dk/services/SignalP/ (Bendtsen et al., 2004). Database searching for proteins homologous with TSD2 was performed using the BLAST2 algorithm (Altschul et al., 1997). Protein sequences were aligned using the clustal w 1.8 program (Thompson et al., 1994). Sequence data for the sequences shown in Figure 7 may be found in the Genbank/EMBL data libraries under the accession numbers: TSD2 (NP_177948), At2g03480 (NP_027543), At1g13860 (NP_172839) and XP_467861. The sequence of the TSD2 cDNA has the accession number NM_106474. The phylogenies were constructed using the phylip package of programs (http://evolution.genetics.washington.edu/phylip.html) (Felsenstein, 1989). The Dayhoff percentage accepted mutation weight matrix was employed for distance calculation. One hundred bootstraps were performed with neighbor-joining to obtain a consensus tree and to estimate the confidence of each tree clade. The treeview cladogram drawing program (Page, 1996) was downloaded from http://taxonomy.zoology.gla.ac.uk/rod/treeview.html and used for displaying phylogenies. The output tree was revised with coreldraw 10.

GUS staining

Histochemical analysis of the GUS reporter enzyme was performed essentially according to the method described by Jefferson et al. (1987). Sample tissues were fixed in 90% ice-cold acetone for 1 h and incubated overnight in reaction buffer. Endogenous pigments were destained with 70% ethanol, and the GUS staining pattern recorded under a stereomicroscope (Olympus SZX12; http://www.olympus-global.com/) or a microscope (Zeiss Axioskop 2 plus; http://www.zeiss.com/) equipped with an Olympus C-4040ZOOM photographic device.

Scanning electron microscopy

For scanning electron microscopy, hypocotyls of tsd2 and WT seedlings grown for 4 days in the dark were fixed in cold FAA (50 ml 95% ethanol, 5 ml glacial acetic acid, 10 ml 37% formaldehyde, 35 ml distilled water) for a minimum of 48 h and stored in FAA at 4°C. Hypocotyls were washed in 70% ethanol, dehydrated in a graded ethanol series and acetone, and subjected to critical point drying. After mounting and gold coating on aluminium stubs, images were taken with a LIO 430 scanning electron microscope (Zeiss).

Analysis of cellular localization of the TSD2–GFP fusion protein

Transgenic plants expressing a 35S:GFP–TSD2 fusion construct were pre-analyzed using an epifluorescence microscope before confocal imaging. Imaging of GFP and MitoTracker fluorescence was performed on a Leica TCS SP2 confocal laser scanning microscope (Leica Microsystems AG, http://www.leica-microsystems.de/) equipped with an argon/krypton laser. For mitochondrial staining, whole seedlings were incubated for 15–20 min at 37°C in 0.5× MS medium containing 50 nm MitoTracker Red CMXRos (Molecular Probes, http://probes.invitrogen.com/) prior to imaging. A stock solution of BFA (Sigma-Aldrich; http://www.sigmaaldrich.com/) was prepared by dissolving 5 mg of BFA in 1 ml DMSO. For BFA treatment, plants were incubated for 30 min in 0.5× MS medium containing 100 μg ml–1 Brefeldin A (BFA). Images were taken with a 63× oil objective. For GFP and MitoTracker Red, excitation was set at 488 nm and multi-channel emissions were obtained with filter sets. GFP emission was detected between 500 and 550 nm, and MitoTracker Red was detected between 590 and 630 nm. For GFP and YFP co-localization, excitation was set at 488 nm for GFP and 514 nm for YFP. Emission was detected using a 490–515 nm bandpass filter for GFP and a 560–615 bandpass filter for YFP. Images were recorded and displayed using leica lcs version 2.61 and adobe photoshop 6.0 software (Adobe Systems Inc., http://www.adobe.com/).

Chemical analysis of cell walls

Hypocotyls were grown in the dark at 22°C for 6 days after a cold treatment (4°C) in the dark for 2 days and subsequent light exposure at 22°C for 8 h to induce germination. The hypocotyls were placed in a screw-capped Eppendorf tube together with a metal ball, frozen in liquid nitrogen, and ground in a Retsch MM200 grinder (Retsch GmbH & Co. KG, http://www.retsch.de/) for 2 min at 20 Hz. The ground material was washed once in 70% aqueous ethanol, followed by chloroform:methanol (1:1 v/v) then acetone, and then dried.

The degree of methylesterification was determined after saponification of cell-wall material (2 mg) with 0.5 m NaOH for 1 h at room temperature, during which time methanol is released. The resulting methanol concentration in the supernatant was determined spectrophotometrically as described previously (Klavons and Bennett, 1986).

To determine the neutral sugar composition and uronic acid content, the pellet was hydrolyzed with 2 m TFA (trifluoroacetic acid) for 1 h at 121°C, with myo-inositol as the internal standard. The neutral sugar composition was determined by converting the resulting monosaccharides to alditol acetates (Englyst and Cummings, 1984). The analysis was performed by GC-MS on an SP-2380 fused silica capillary column (30 m × 0.25 mm, 0.2 μm film thickness, Supelco, http://www.sigmaaldrich.com/Brands/Supelco_Home.html) in an Agilent 6890N gas chromatograph coupled to an Agilent 5973 mass selective detector (Agilent Technologies, http://www.agilent.com/). The temperature program started at 160°C for 2 min, then increased to 200°C at 20°C min–1, held for 5 min, then increased to 245°C at 20°C min–1 and held for 12 min.

The amount of uronic acids was determined by the colorimetric m-hydroxydiphenyl assay with galacturonic acid as a standard (Filisetti-Cozzi and Carpita, 1991).

Acknowledgements

We thank Annette Nöh for skilful technical assistance, Professor W. Frey, Christine Grüber and Elisabeth Scherer (Free Universität of Berlin) for help with the scanning electron microscopy, and Bernhard Grimm and Christina Kühn (Humbolt University of Berlin) for access to the confocal microscope. This work was partly funded by the Deutsche Forschungsgemeinschaft by a grant to T.S. (Schm 814/18-3) and a grant from the Freie Universität Berlin to E.K.

Ancillary