ACAULIS5 controls Arabidopsis xylem specification through the prevention of premature cell death

Organ differentiation in eukaryotes typically requires the proper coordination of several different developmental processes. For example, the shift from vegetative to reproductive development in Arabidopsis thaliana involves a significant increase in the internode length of the inflorescence stem, which has to be coordinated with building of a functional vascular system, including the elongated structures of xylem. During the active elongation phase of the inflorescence stem, primary `protoxylem' vessel elements differentiate and develop spiral or annular secondary cell wall thickenings that allow longitudinal expansion of the cells and, therefore,elongation of the stem. When internode elongation ceases, primary `metaxylem'vessel elements with the more elaborate type of reticulate or pitted secondary cell wall thickenings differentiate. The primary vascular development is followed by the formation of vascular cambium and secondary growth, which, in addition to vessel differentiation, involves the formation of xylem fibers.

Plant hormones are involved in the control of all the different stages of xylem development. Physiological and pharmacological studies have demonstrated the important role of auxins and cytokinins in controlling the activity of the vascular cambium and the initiation of xylem development, while brassinosteroids, ethylene and gibberellins are important in the modulation of the cambial activity and the control of xylem differentiation (reviewed by Ye, 2002). Mutations in the various components of hormone synthesis, transport or signal transduction in Arabidopsis have largely confirmed the action of the various hormones and demonstrated the involvement of additional compounds, such as sterols(reviewed by Fukuda, 2004). Mutations affecting hormone transport and/or signaling provide evidence for the role of auxins in the initiation of the vascular meristem and the maintenance of vascular continuity(Gälweiler et al., 1998; Hardtke and Berleth, 1998; Hobbie et al., 2000), the role of cytokinins in phloem specification(Mähönen et al.,2000) and the inhibition of protoxylem differentiation(Mähönen et al.,2006), and the role of brassinosteroids in promoting xylem differentiation (Caño-Delgado et al., 2004). However, Arabidopsis mutants have demonstrated that there are as yet unknown signals that regulate xylem development (Koizumi et al.,2000; Parker et al.,2003). In addition, the signals controlling the maturation of xylem elements remain largely unknown.

Polyamines (PAs) are low molecular weight cationic molecules, the synthesis of which is initiated by decarboxylation of arginine or ornithine to produce putrescine, and sequential addition of two aminopropyl groups to putrescine through the activity of the aminopropyltransferases spermidine synthase and spermine synthase, to produce the triamine spermidine and the tetraamine spermine, respectively (Ikeguchi et al.,2006). Arabidopsis has two putative spermidine synthases(SPDS1 and SPDS2) and two putative spermine synthases (SPMS and ACL5)(Imai et al., 2004b; Panicot et al., 2002). Bacterially produced ACL5 was recently related to synthesis of thermospermine,which is an isomer of spermine (Knott et al., 2007), but this remains to be validated in planta. PAs have been shown to be involved in a variety of processes, such as cell proliferation and defense against both abiotic and biotic stresses, but they are also associated with the normal development of plants(Kumar et al., 1997; Walden et al., 1997). It has also been proposed that they participate in the control of vascular development based on their effect on cell division, interaction with other hormones and H₂O₂ produced during PA catabolism that could potentially affect processes such as vascular cambial activity, cell differentiation and cell death (Bais and Ravishankar, 2002; Møller and McPherson,1998; Sebela et al.,2001). Indirect support for this suggestion was provided by spermidine- and spermine-deficient transgenic plants, which exhibited a stunted phenotype (Kumar et al.,1996). In addition, an Arabidopsis mutant in the polyamine biosynthesis-related S-adenosylmethionine decarboxylase, with slightly reduced spermidine and spermine levels, was found to be stunted and to exhibit severely altered vascular development(Ge et al., 2006). However, a more direct involvement was demonstrated only recently, with the identification of ACAULIS 5 (ACL5) as a putative spermine synthase (Hanzawa et al.,2000). The Arabidopsis acaulis mutants were isolated on the basis of their severely impaired internodal elongation after the transition from the vegetative to the reproductive stage(Akamatsu et al., 1999), and acl5 was found to display overproliferation of xylem elements(Hanzawa et al., 1997). Furthermore, the thickvein mutant, harboring another loss-of-function allele of ACL5, was shown to display thicker veins and an increased number of vascular cells in the inflorescence stems(Clay and Nelson, 2005). Interestingly, changes to vascular development seem rather specific to the acl5 mutant, as the lack of spermidine synthesis in Arabidopsis is embryo lethal(Imai et al., 2004b), whereas mutations in SPMS, which encodes the major spermine synthase, do not affect plant development (Imai et al.,2004a). Recently, a mutation that allows higher production of a bHLH transcription factor, SAC51, was shown to suppress all the defects associated with the loss of ACL5 function(Imai et al., 2006), but as the suppressor mutant of bHLH was dominant it is not entirely clear whether its function is directly related to PA signaling. Therefore, the underlying mechanism for the action of PAs in plant growth and development remains unclear.

ACL5 is specifically expressed in the procambial and/or the provascular tissues during primary growth of the root(Birnbaum et al., 2003; Clay and Nelson, 2005), but it does not seem to have any major function at this stage(Clay and Nelson, 2005). To elucidate the function of ACL5 during vascular development, we therefore focused in the current study on the vascular tissues of the hypocotyl, which display extensive secondary growth during prolonged growth period. We demonstrate here that, in the hypocotyl as well as in the inflorescence stem, ACL5 is expressed not just broadly with respect to vasculature, as shown earlier (Clay and Nelson,2005), but specifically in the xylem vessel elements at a strictly defined developmental stage, suggesting direct involvement of ACL5 in xylem vessel differentiation. Furthermore, we show that the acl5 mutant displays severe overall inhibition of the secondary growth of the vascular tissues, dramatic alteration in the morphology of the vessel elements and complete lack of xylem fibers. Finally, we propose a mechanistic model for the function of ACL5 in xylem specification, based on experiments carried out in transgenic plants expressing a DT-A toxin gene under the control of the ACL5 promoter and in the Zinnia elegans tracheary element differentiation system.

The analyses of the Arabidopsis thaliana acl5 mutant were conducted using the mutant allele acl5-4 (in Ler background)or the SALK_028736 line, when comparisons were made with transgenic Pro_ACL5:DT-A, Pro_DR5:GUS, or Pro_XCP2:GUS lines, which were all in the Arabidopsis Col-0 background. Soil-grown plants were kept in a walk-in climate chamber (Kryo Service, Helsinki, Finland) under long-day conditions (18 hours light/6 hours darkness, 70% humidity, 21°C/18°C). The plants grown in vitro were kept, without selection, on MS plates (Duchefa)containing 10 g/l sucrose, in growth rooms under long-day conditions (18 hours light/6 hours darkness, 21°C/18°C).

2.48 kb of the ACL5 (At5g19530) sequence, upstream from the start codon,was amplified from Col-0 genomic DNA using the primers 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTCCACCATCGAATGGTATGC-3′ and 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTATCCAAGTTGAGGAGAAGAT-3′, which include recognition sequences for the Gateway recombination system(Invitrogen). Amplification conditions followed the manufacturer's instructions, and the promoter fragment was sequenced after the first cloning step in pDONR201.

To create Pro_ACL5:DT-A plants, the 3′UTR of the ACL5 gene was amplified from Col-0 genomic DNA with the primers 5′-CACAGTCGACAGACGAACCGGTTTCAGTTTC-3′ and 5′-CACAGAATTCAGATTTGGTGTGGAGAAATAAG-3′, which include restriction sites for SalI and EcoRI, respectively(underlined). This was subsequently cloned into pBluescript SK II to produce pACL-3′UTR/SK. The sequence of the diphtheria toxin A chain (DT-A) was amplified from pEW3 (Nilsson et al.,1998) with primers 5′-GAGTCGACATGGATCCTGATGATGTTGTTG-3′ and 5′-CCACGTCCAGACGTCGAC-3′, including restriction sites for SalI, and was cloned into pACL-3′UTR/SK. Orientation and sequence fidelity were checked by sequencing. The DTA-ACL5 3′UTR cassette was transferred, as a KpnI/SacI restriction fragment, to the binary vector pMDC205(Curtis and Grossniklaus,2003), the GFP-coding sequence of which was simultaneously removed to create the construct DTA-utr/pMDC205. Finally, the 2.48 kb ACL5 upstream sequence was recombined into DTA-utr/pMDC205 using the Gateway LR reaction(Invitrogen), and the resulting vector was transformed into Col-0 plants by the method of Clough and Bent (Clough and Bent, 1998). Transformants were selected on MS medium containing 20 g/l sucrose and 50 μg/ml hygromycin. Homozygous and heterozygous plants were identified by PCR and segregation analysis.

To create the Pro_ACL5:GUS plants, the 2.48 kb ACL5 promoter fragment was cloned into pK2GWFS7.0(Karimi et al., 2002). This vector was transformed into Col-0 plants using the method described by Clough and Bent (Clough and Bent,1998). Transformants were selected on MS medium containing 20 g/l sucrose and 50 μg/ml kanamycin. A non-segregating line was crossed to each of the acl5 mutant and a heterozygous Pro_ACL5:DT-A line 4. The resultant progenies were screened for homozygosity of either the acl5 mutation or the Pro_ACL5:DT-A transgene in F2 by the seedling phenotype and for the Pro_ACL5:GUS transgene in F3 by antibiotic resistance.

Histochemical β-glucuronidase (GUS) activity assays were performed for whole seedlings grown in vitro or tissue pieces that were excised from plants grown in soil for indicates times. The chromogenic substrate 5-bromo-4-chloro-3-indoxyl-beta-D-glucuronide cyclohexylammonium (Gold Biotechnology) was used, according to the manufacturer's instructions, to detect GUS activity. The GUS-stained whole seedlings were examined directly by microscopy. The excised tissue pieces were mounted after GUS staining in LR white resin (TAAB Laboratories) containing 10% PEG400 and sectioned at 25μm. If needed, the sections were counterstained with 0.05% Ruthenium Red.

Localization of the ACL5 mRNA was determined, using 8 μm sections embedded in paraplast, by in situ hybridization, as described by Jackson (Jackson, 1991). Briefly, antisense and sense riboprobes were generated from the complete cDNA of ACL5 isolated from the PRL1 gene library(Newman et al., 1994), using SP6 and T7 RNA polymerases, respectively, and then hydrolyzed by carbonate hydrolysis into 100-200 bp fragments. Probes were labeled with digoxigenin and immunodetected with an alkaline phosphatase-conjugated antidigoxigenin antibody. Alkaline phosphatase was detected using the BCIP-NBT procedure. Photographs were taken under the bright field of a Nikon Eclipse microscope.

For anatomical characterization, hypocotyls were collected from plants that were grown in soil for indicates times, fixed in FAA (5% formaldehyde, 10%acetic acid, 50% ethanol) and embedded in LR white resin (TAAB Laboratories). Transverse and longitudinal sections were taken and stained with 0.05%Toluidine Blue (Merck).

For analysis of the individual xylem elements, hypocotyls were macerated from soil-grown 2-month-old plants that were stimulated for extensive secondary growth by continuous decapitation of the inflorescence stems. The tissues were incubated at 95°C for 4 hours in 3%H₂O₂/50% acetic acid, gently washed twice with distilled water and neutralized by the addition of solid Na₂CO₃ to the last wash. Samples were then disaggregated mechanically and stored at room temperature. The cell suspensions were stained with 0.01% CelluFluor(Polysciences) and examined under UV illumination with a light microscope. For each genotype, three plants were analyzed and 50-300 vessel elements per plant were scored for length, width and the pattern of secondary cell wall thickenings.

Light microscopy images were taken using a Zeiss Axioplan II microscope equipped with Zeiss AxioCam CCD camera (Zeiss, Oberkochen, Germany). Xylem vessel elements were measured using the microscope images and Zeiss Axiovision 3.1 software.

Electron microscopy images were taken of hypocotyls embedded in Spurr resin(Sigma) according to Rensing (Rensing,2002), and examined in a Hitachi H-7000 transmission electron microscope (Hitachi, Tokyo, Japan).

Confocal images of vessels were taken using a Leica TCS SP2 microscope(Leica Microsystems, Wetzlar, Germany) with an excitation wavelength of 568 nm(helium-neon laser) and an emission wavelength of 585 nm. Projections of the confocal data were exported using TCS software.

The first pair of leaves from 14-day-old seedlings of Zinnia elegans cv Envy (Hem Zaden BV, Venhuizen, Holland) were used to isolate mesophyll cells for xylogenic cell suspension cultures according to the method of Fukuda and Komamine (Fukuda and Komamine, 1980). Cells were cultured in an induction medium containing 0.1 mg/l α-naphthaleneacetic acid and 0.2 mg/l benzyladenine(Sigma-Aldrich).

For the pharmacological treatments, 1 ml of the differentiating cell culture was treated, in 12-well plates, with various amounts of spermine(Sigma-Aldrich), ranging from 0 to 200 μM final concentration, and monitored for 7 days. Quantification of the tracheary element (TE)differentiation efficiency, as well as the width, length and type of TEs, were recorded using microscope images for 50 cells from each replicate cell culture.

The expression of ACL5 was earlier localized into the procambial and/or provascular tissues of Arabidopsis plants(Birnbaum et al., 2003; Clay and Nelson, 2005). We detected with high-resolution analyses using in situ hybridization and resin-embedded sections of Pro_ACL5:GUS plants that, in all parts of the plant, ACL5 was expressed in the vascular tissues(Fig. 1). More specifically, ACL5 expression was confined to xylem vessel elements in the vascular bundles of the inflorescence stems (Fig. 1C,E,F), in the junction between the silique and the pedicel(Fig. 1A) and in the hypocotyl(Fig. 1H). Expression was apparent immediately after expansion of the vessel elements, but before the onset of the secondary cell wall deposition. Protoxylem cells did not express ACL5 but sometimes parenchymatic cells next to the protoxylem elements showed expression in the inflorescence stem(Fig. 1F) and the roots (data not shown). In line with the earlier reports, we also found expression of Pro_ACL5:GUS in the procambial and/or provascular tissues of young hypocotyls (Fig. 1G). To assess the function of ACL5 during vascular development, we chose to focus on the hypocotyls, where large numbers of the ACL5-expressing vessels are continuously being formed.

Consistent with the expression profile of ACL5, its loss of function caused severe defects in the vasculature of the hypocotyl. Seven-day-old acl5 seedlings looked normal and the diameter of the stele was indistinguishable from that of the wild type(Fig. 2A,B) although slight changes, such as asymmetry of the stele and of the whole hypocotyl, could be discerned. As with the overproliferation of xylem vessels in the inflorescence stems and leaf veins reported previously(Clay and Nelson, 2005; Hanzawa et al., 1997), the hypocotyls of 13-day-old acl5 seedlings had more xylem vessels than the wild type (Fig. 2E,F). However, during prolonged growth of the plants, acl5 did not display further secondary growth, and the hypocotyls of 35-day-old acl5seedlings were significantly thinner than those of the wild type(Fig. 2G,H). The lack of secondary growth was accompanied by complete lack of xylem fibers in the acl5 hypocotyls (Fig. 2H). In addition to the weak secondary growth of the hypocotyl,defects were visible in the secondary cell wall patterning of the acl5 mutant. Instead of having clearly defined spiral-type protoxylem and pitted-type metaxylem vessels, as in the wild type(Fig. 2C), acl5 seemed to form spiral-type vessels that were slightly reticulated, i.e. that had a few interconnecting strands between the spiral whorls(Fig. 2D).

In addition, electron micrographs of 3-week-old hypocotyls showed the presence of numerous vessels with altered patterns of secondary cell wall deposition and the lack of xylem fibers in acl5(Fig. 3E). Interestingly, the pattern of vessel maturation seemed to differ as well. Normal maturation of vessel elements involves extensive deposition of secondary cell wall material,which is terminated by the collapse of the central vacuole and the release of the vacuolar contents into the cytoplasm, leading into rapid cell death and concomitant autolysis of the vessel elements. Maturation can proceed after the death of the vessel elements, even though majority of the secondary cell wall material is normally deposited before the collapse of the vacuole. In the wild type, it was always possible to distinguish a few maturing vessel elements with an intact central vacuole and ongoing secondary cell wall deposition(Fig. 3B-D). However, we could not find any vessel elements in acl5 hypocotyls that had distinguishable secondary cell walls and that still had intact central vacuole(Fig. 3F-H). Therefore, the maturing vessel elements observed in acl5 must have surpassed the vacuolar collapse, and it seems that the vacuole collapses in acl5too early in relation to the progress of the cell wall deposition.

To get a more in depth insight into the morphological changes of the xylem elements, hypocotyls of wild-type and acl5 plants were macerated in an alkaline medium, and the size and secondary cell wall architecture of the individual xylem elements that remained intact after the procedure were examined. Hypocotyls were collected from two-month-old plants, which normally at this age exhibit extensive secondary growth. Marked differences were observed in the morphology of vessel elements that were classified as annular,spiral, reticulate or pitted according to their secondary cell wall pattern(Esau, 1977). The vessel elements of acl5 were mainly spiral, while only a very small portion of the vessel elements were of this type in the wild-type(Fig. 4A, Fig. 2J). Most strikingly, the pitted elements, that were dominant in the wild type during secondary growth,were completely missing in acl5(Fig. 4A, Fig. 2I). The length and the width of the vessel elements were also reduced in acl5(Fig. 4B,C). In addition, we were able to verify the absence of xylem fibers in the macerates of all acl5 plants examined (see Fig. 7E). Thus, our microscopic analyses showed that the absence of ACL5 function has a profound effect on xylem development. In particular, the pattern of xylem maturation was altered in a dramatic way, which raises the issue of whether it is causally related to the observed alterations in xylem specification.

To further examine the function ACL5 in xylem specification and maturation, acl5 was crossed to two different marker lines that are indicative of vascular development. Pro_DR5:GUS is a well-established marker for endogenous auxin levels(Ulmasov et al., 1997),normally showing highest expression in the root tips and no expression in the hypocotyls of young Arabidopsis seedlings(Fig. 5A). However, when expressed in the acl5 background, Pro_DR5:GUSshowed weak activity in the developing xylem vessel elements of the hypocotyls(Fig. 5B). As auxin is effective in stimulation of cambial activity, increase in the expression of an auxin marker is in line with the observed increase in the cambial activity and vessel differentiation during secondary growth of the young acl5seedlings.

Pro_XCP2:GUS is a marker for cell death in the xylem elements of Arabidopsis seedlings(Funk et al., 2002). The XCP2 promoter is isolated from a gene encoding a xylem-specific cysteine protease that is believed to function as an effector protease during autolysis of the cell contents (Zhao et al., 2000). In the wild-type background, Pro_XCP2:GUS was expressed in the maturing xylem vessels of the hypocotyls, but not in the immature vessel elements without secondary cell wall thickenings (Fig. 5C). A slightly different pattern was observed in the acl5 background; Pro_XCP2:GUS was expressed not only in the maturing vessels but also at an earlier developmental stage in the immature vessel elements(Fig. 5D). Therefore, the results are in accordance with the results of the electron microscopy analysis, suggesting premature onset of the cell death program in relation to the formation of the secondary cell walls in the vessel elements of acl5.

To investigate the function of polyamines in xylem maturation and cell wall patterning in a simplified system, we examined the effect of spermine in the in vitro Zinnia elegans xylogenic system. This system allows direct transdifferentiation of freshly isolated mesophyll cells of Zinniainto structures similar to xylem vessels, commonly called tracheary elements(TEs), within a time frame of 72 to 96 hours and in a semi-synchronous manner(Fukuda and Komamine, 1980). Concentrations of exogenous spermine were physiologically relevant as the endogenous levels of spermine were at a maximum ∼50 μM in the Zinnia cells (data not shown).

Exogenous spermine delayed the time of TE appearance and reduced the rate of TE differentiation in a dose-dependent manner(Fig. 6A). Spermine also altered TE type by stimulating differentiation of the more elaborate,metaxylem-type TEs (characterized by their reticulated or pitted secondary cell walls). While control TE cultures contained about 25% spiral and 75%reticulated cells, the addition of 50 μM spermine caused the appearance of pitted cells amounting to about 10% of the total(Fig. 6B). This trend was enhanced with higher concentrations of spermine: at 100 μM spermine, the proportion of reticulated and pitted cells was roughly equivalent(Fig. 6B). The same was true for 200 μM spermine (Fig. 6B,E), although very few TEs differentiated under these conditions(Fig. 6A). Concentrations above 200 μM were lethal for the cell culture. The length and the width of the TEs were also strongly affected, with a dose-dependent increase of over threefold for the width and twofold for the length, compared with the control values (Fig. 6C). In summary,these results show that spermine: (1) delays the time of TE appearance, (2)increases TE size and (3) favors pitted-type TE differentiation. It was recently reported that, instead of spermine synthase, ACL5 might encode an enzyme for synthesis of thermospermine, which is a more rare tetraamine (Knott et al.,2007). Our conclusions on the function of ACL5 in vessel specification, which is analogous to what we observed here for spermine,indicate that if thermospermine is the product of ACL5 activity it has at least to some extent, if not completely, equivalent role to spermine in the xylem vessels.

The evidence presented above is compatible with a model in which ACL5 prevents premature cell death in developing xylem vessel elements in order to allow proper xylem specification and especially formation of the elaborate type of vessels. One prediction based on this model is that the induction of premature death of the ACL5-expressing subset of cells should render a phenotype similar, in terms of xylem differentiation, to the one caused by the loss of ACL5 function. To test this, we constructed transgenic Arabidopsis plants expressing the diphtheria toxin A chain(DT-A) under the control of the ACL5 promoter. DT-Ahas been previously used in plants as a tool to ablate specific cell types by directing the production of this cell-autonomous toxin to those target cells(Nilsson et al., 1998).

The general phenotype of Pro_ACL5:DT-A plants resembled that of the acl5 mutant, with the severity depending on the gene dose(Fig. 7). Heterozygous plants were more slender and slightly smaller than wild-type plants, while homozygous plants had a greatly reduced rosette leaf size and stem length, resembling an extreme acl5 phenotype (Fig. 7A). It was evident that the expression of DT-A from the ACL5 promoter delayed the onset of xylem differentiation by a few days in in vitro grown seedlings, but 6 days after germination a broad cambium was already present and numerous vessels were being produced in the hypocotyl(Fig. 7L). This pattern was similar to the one observed in acl5, even though minor differences in the general anatomy of these two genotypes could be discerned(Fig, 7L,P). Further progression of the secondary growth was also similar to acl5:transverse sections of 2-month-old hypocotyls revealed that the secondary growth was severely reduced in the homozygous Pro_ACL5:DT-Ahypocotyls compared with the wild type(Fig. 7B-D). Comparison of the xylem cell types present in hypocotyls of two independent transgenic lines,after maceration, confirmed the similarity between the Pro_ACL5:DT-A plants and the acl5 mutant in terms of xylem specification. The reticulate-type vessel elements were predominant in the homozygous Pro_ACL5:DT-A lines, the pitted-type vessels were almost completely missing(Fig. 7F, Fig. 8A) and the vessel elements were significantly shorter and thinner than in the wild-type(Fig. 8B). Mature xylem fibers were not encountered in the macerates of the Pro_ACL5:DT-Aor acl5 hypocotyls (Fig. 7E,F), whereas this was the predominant type of xylem element produced in the wild-type hypocotyl at this stage(Fig. 7G). In conclusion,similarity in xylem specification and cell morphology of the Pro_ACL5:DT-A plants to that observed in acl5supports the proposed role of ACL5 in preventing premature death of the vessel elements.

To confirm that the observed alterations in vessel morphology of the Pro_ACL5:DT-A lines were caused directly by DT-A toxin production and not, for example, by induction of ectopic non-DT-A-expressing xylem elements resulting from early ablation of the xylem elements, we analyzed expression of the Pro_ACL5:GUS marker in the homozygous Pro_ACL5:DT-A line 4. As expression of both GUS and DT-A are driven by the ACL5 promoter, it can be assumed that GUS activity reveals the sites of toxin production in the Pro_ACL5:DT-A Pro_ACL5:GUS plants. Whole-mount staining of hypocotyls revealed that GUS activity was limited to the vasculature of these plants (Fig. 7K). Higher resolution images allowed definition of this activity into developing xylem with highest activity in the incipient vessel elements(Fig. 7L,M) and sometimes even in the maturing vessel elements (Fig. 7N). The expression domain of Pro_ACL5:GUS was broader both in Pro_ACL5:DT-A(Fig. 7K-N) and acl5seedlings (Fig. 7O-Q) compared with the wild type (Fig. 7H-J),which is most probably related to the initial increase in the cambial activity and overproliferation of vessel elements in the young Pro_ACL5:DT-A and acl5 seedlings. A similar increase in the expression of ACL5 in the acl5 background was earlier suggested to indicate negative-feedback regulation of ACL5 expression (Clay and Nelson,2005; Hanzawa et al.,2000). Together, our results using the Pro_ACL5:GUS marker support: (1) the production of DT-A toxin in the xylem vessel elements of the Pro_ACL5:DT-Aplants at a stage which allows further maturation of the cells and (2) that the toxin-induced premature death of the vessels is causally related to the altered xylem specification of Pro_ACL5:DT-A plants.

Our work establishes that ACL5 has a prominent role in the correct specification of xylem cells. This is supported by at least three observations: (1) the xylem-specific expression of ACL5(Fig. 1); (2) the predominance of the spiral-type vessels and the lack of pitted vessels as well as fibers in the acl5 mutant (Figs 2, 3 and 4); and (3) the misregulation of vascular-related markers in the acl5 mutant(Fig. 5). The earliest xylem defects of acl5 colocalise with the ACL5-expressing subset of cells, and we therefore conclude that xylem specification is the primary function of ACL5. Activity of ACL5 as a polyamine biosynthetic enzyme indicates a completely novel role for polyamines in control of xylem specification by preventing premature death of the xylem vessel elements. This conclusion is substantiated by our observations of the early onset of the cell death program in the vessel elements of acl5(Fig. 3, Fig. 5D) and by the fact that the xylem vessel defect of acl5 could be phenocopied by the induction of premature DT-A toxin-mediated cell death in the ACL5-expressing vessels (Figs 7 and 8).

That polyamines protect against premature cell death and senescence is not surprising. A large number of studies demonstrate the role of polyamines in protection from apoptotic cell death in animals, even though the opposite function has also frequently been reported(Seiler and Raul, 2005). In plants, it has been suggested that polyamines retard senescence by maintaining membrane stability and by reducing the quantity and effects of free radicals(Bais and Ravishankar, 2002). In addition, putrescine, spermidine and in particular spermine have been shown to block tonoplast cation channels(Brüggemann et al., 1998);this is believed to block ion leakage from the vacuoles and contribute to the regulation of osmotic potential in the cell and provide protection against salt stress (Yamaguchi et al.,2006; Yamaguchi et al.,2007). On the basis of our results, we cannot predict the exact molecular mechanism of the ACL5-mediated protection against cell death, but the fact that alterations in ion fluxes, especially across the tonoplast, are known to be central in the control of xylem cell death(Kuriyama, 1999) makes it tempting to speculate that ACL5 may function through regulation of tonoplast channels in the xylem elements. An analogous mechanism was proposed for polyamines in the regulation of mitochondrial integrity, which is central to apoptotic cell death in animals (Tassani et al., 1995). In this context, it is interesting to note that spermine has been shown to inhibit mitochondrial membrane permeability in oats(Curtis and Wolpert,2002).

The premature vessel death in acl5 and in the transgenic Pro_ACL5:DT-A plants is accompanied by smaller sized vessels and the formation of the simpler spiral- and reticulate-type secondary cell walls instead of the more elaborate pitted-type secondary cell walls that usually dominate (Figs 4 and 8). Consistent with these results, an increase in the duration of differentiation in the Zinniaxylogenic cultures induced the formation of large tracheary elements and a shift towards the more elaborate type of secondary cell walls(Fig. 6). Our results therefore support the importance of the duration of vessel differentiation in determining vessel specification and especially the formation of the pitted-type vessels. It is generally believed that longer duration of secondary cell wall formation allows increased deposition of the secondary cell wall material (Barnett,1981). It is therefore possible that ACL5 controls xylem specification by extending the secondary cell wall deposition phase of the vessels to allow formation of the most elaborate types of vessels that also require the most extensive secondary cell wall deposition. However, it is also possible that, instead of the duration of vessel maturation, ACL5 mediates an increase in the duration of vessel expansion, resulting in an increase in the size of the vessel elements, which in turn determines the complexity of the secondary cell wall patterning, as suggested by Roberts and Haigler(Roberts and Haigle,1994).

The molecular control of xylem cell specification is poorly understood. The NAC family transcription factors VND7 and VND6 have been suggested to control the differentiation of the protoxylem and metaxylem elements, respectively(Kubo et al., 2005). We found increased expression of VND6 and especially VND7 in acl5, suggesting that defects in xylem specification are not due to lack of expression of either of these genes (data not shown). Galactoglucomannans(Benová-Kákosová et al., 2006) and an arabinogalactan protein(Dahiya et al., 2006) have been related to control of protoxylem versus metaxylem type vessel elements,but no information exists about their mode of function. Recently, reduced cytokinin signaling was shown to result in high abundance of protoxylem vessels in Arabidopsis roots(Mähönen et al.,2006). However, the predominance of protoxylem-type vessels in acl5 does not seem to be due to impaired cytokinin signaling as acl5 roots were actually shown to display increased cytokinin sensitivity (Clay and Nelson,2005).

Although earlier studies were correct in concluding that acl5mutants exhibit enhanced vascular development with respect to the number of vessels in the inflorescence stems and leaves(Clay and Nelson, 2005; Hanzawa et al., 1997), this proliferation does not lead to enhanced overall size of the vasculature but rather to a dramatic decrease in the width of the vasculature and the whole stem in both the inflorescence stem (data not shown) and in the hypocotyl primarily because of the complete lack of fibers(Fig. 2). There could be several reasons for the lack of fibers, but the occasional presence of(immature) fibers in Pro_ACL5:DT-A plants suggests that fiber development is somehow dependent on the correct specification of vessels, and that differentiation of the pitted-type vessels (also occasionally observed in the Pro_ACL5:DT-A plants) is required for fiber differentiation. There are, to our knowledge, no other mutants that have the interfascicular fibers (see Hanzawa et al., 1997) but that completely lack the xylary fibers, and ACL5 therefore seems to control fiber development through a completely novel mechanism.

Our results, when considered together, suggest that ACL5 is required for correct xylem specification through regulation of the lifetime of the xylem elements. The shorter lifetime of the xylem vessels in the acl5mutant results in the development of only the simple type of xylem vessels and the complete lack of xylem fibers.

We thank Detlef Weigel for providing access to the pEW3 plasmid, Eric Beers for the gift of the Pro_XCP2:GUS seeds, Karin Ljung for assistance with the auxin-related experiments and Alan Marchant for valuable comments on the manuscript. The work was supported by the Swedish Research Council Formas, the Kempe foundation and the Swedish Foundation for Strategic Research. M.A.B. is an EMBO Young Investigator. E.G.M. has been funded by Fellowships from the Spanish Ministry of Education and the Spanish National Research Council I3P Program.