The Arabidopsis AtSUC1 protein has previously been characterized as a plasma membrane H+-sucrose symporter. This paper describes the sites of AtSUC1 gene expression and AtSUC1 protein localization and assigns specific functions to this sucrose transporter in anther development and pollen tube growth. RNase protection assays revealed AtSUC1 expression exclusively in floral tissue, which was confirmed by analyses of AtSUC1 promoter-β-glucuronidase (GUS) plants. In situ hybridizations identified AtSUC1 expression in anther connective tissue, in funiculi and in fully developed pollen grains. Indirect immuno-fluorescence analyses with anti-AtSUC1 antiserum confirmed AtSUC1 protein localization in the connective tissue and funiculi. In mature pollen grains, however, despite high AtSUC1 mRNA levels no AtSUC1 protein was found. Only after pollination of stylar papillae was AtSUC1 protein detected inside the pollen and later inside the growing pollen tubes, suggesting a translation of pre-existing AtSUC1 mRNA after pollination. Pollen germination analyses underlined the important role of sucrose for pollen tube growth. The data presented suggest a role of AtSUC1 in the controlled dehiscence of Arabidopsis anthers. It is postulated that an important function of AtSUC1 is the cell-specific modulation of water potentials.
Most of these energy-dependent sucrose-H+ symporters have been localized in the companion cells (PmSUC2: Stadler et al. 1995 ; AtSUC2: Stadler & Sauer 1996) or the sieve elements (NtSUT1, LeSUT1 and StSUT1: Kühn et al. 1997 ) of the phloem. These transporters accumulate sucrose to concentrations of 0.5–1.5 molar inside the sieve element–companion cell complex thereby generating the osmotic driving force for water influx and assimilating flow within the phloem vessels. It has been shown that recombinant green fluorescent protein (GFP) from the jellyfish Aequorea victoria that has been expressed in Arabidopsis companion cells under the control of the AtSUC2 promoter ( Imlau et al. 1999 ) or exogenously supplied carboxy fluorescein ( Oparka et al. 1994 ) can travel with this stream of assimilates towards the different sink organs of Arabidopsis. Using these fluorescent compounds it has been shown that certain sink tissues, such as root tips or young leaves, are symplastically connected to the phloem ( Imlau et al. 1999 ; Oparka et al. 1994 ). Other sinks, such as the embryo or pollen grains, were shown to be symplastically isolated ( Imlau et al. 1999 ; Scott et al. 1991 ) and dependent on the import of carbohydrates via plasma membrane-localized transport proteins.
Examples for sink loading by sucrose carriers were obtained for Plantago major ( Gahrtz et al. 1996 ) and Vicia faba ( Weber et al. 1997 ), where the PmSUC1 and VfSUT1 sucrose transporter genes are expressed to high levels in distinct areas of developing seeds. Expression of the VfSUT1 sucrose transporter gene has been localized on the cellular level and VfSUT1 protein was found only in the epidermal cell layers of broad bean embryos ( Weber et al. 1997 ). The VfSUT1 sucrose uptake system develops during the switch to the storage phase. Developmentally younger and mitotically still active embryos seem to import monosaccharides by the VfSTP1 monosaccharide transporter after extracellular invertase hydrolysis ( Weber et al. 1997 ). This suggests a preference for the uptake of monosaccharides into growing and metabolically active cells and a preference for the uptake of disaccharides into cells in the storage phase.
In this paper we present data describing the localization of the Arabidopsis AtSUC1 sucrose transporter of the cellular level. The AtSUC1 gene is expressed in different cells and cell layers of flowers. All cells exhibiting AtSUC1 expression are fully developed and mitotically inactive. We postulate that the flower-specific AtSUC1 sucrose carrier functions primarily to modulate water potentials which may influence the flow of water or cause an intercellular signal.
AtSUC1 mRNA is found only in Arabidopsis flowers
The AtSUC1 protein has previously been characterized as a sucrose-H+ symporter ( Sauer & Stolz 1994). To determine the site of AtSUC1 expression, total RNA was isolated from different Arabidopsis tissues and used for RNase protection analyses with a radiolabelled AtSUC1 specific antisense probe. mRNA levels of the second Arabidopsis sucrose transporter AtSUC2 and levels of the 18S rRNA were analyzed in parallel. Northern blots describing the tissue-specific expression of the two AtSUC genes have been published previously ( Sauer & Stolz 1994), but we found later that the probes used in these experiments cross-reacted.
In the RNase protection analyses presented in Fig. 1, AtSUC1 mRNA was identified exclusively in floral tissue. In agreement with previous reports, the gene of the companion cell-specific AtSUC2 transporter ( Stadler & Sauer 1996) was shown to be expressed in all Arabidopsis tissues to various extents. This showed that the expression patterns of the two sucrose transporter genes are clearly different and that the AtSUC1 protein does not represent a second transporter involved in the phloem loading of sucrose. In fact, the localization of AtSUC1 gene expression in flowers suggested a role for AtSUC1 in the import of sucrose into floral sink cells.
The AtSUC1 promoter directs GUS activity to anthers and pistils of Arabidopsis
The site of AtSUC1 expression was analyzed more precisely using AtSUC1 promoter–GUS plants. As genomic AtSUC1 sequences were not available from publically available databases, we screened a λGEM11 Arabidopsis genomic library with a radiolabelled probe derived from the AtSUC1 cDNA clone pTF2011 ( Sauer & Stolz 1994). A 3 kb genomic BamHI fragment was identified, subcloned into pUC19 and the insert of the resulting plasmid pET110 was sequenced (EMBL accession number AJ001364). It contained 2581 bp of 5′-flanking sequence and 448 bp of AtSUC1 coding sequence. This fragment was used to generate a translational fusion with the GUS gene consisting of the 2581 bp flanking sequence and the coding sequence for the first 17 amino acids of AtSUC1 ( Fig. 2). Arabidopsis plants were transformed with this construct and several independent transformants were analyzed for GUS activity.
In agreement with the RNase protection analyses presented in Fig. 1, blue GUS staining was confined to the flowers of transgenic Arabidopsis plants ( Fig. 3a,b). No GUS activity was detected in any other tissue of flowering or non-flowering plants (data not shown). Weak GUS staining was identified in the stigma ( Fig. 3a) and much stronger staining was seen in the pollen inside the anthers ( Fig. 3a–c), suggesting that AtSUC1 protein may be important for sucrose import into developing pollen. Figure 3(b) shows that no AtSUC1 promoter-driven GUS activity is seen in the still developing anthers of the presented floral buds indicating that AtSUC1 expression may be confined to the late stages of pollen development.
Blue GUS staining is also seen in pollen grains released from the anthers. Figure 3(d) also reveals GUS staining in the growing pollen tubes penetrating the walls of the papillae. Stigmatic papillae with invaded pollen tubes have a blue appearance, whereas papillae without pollen or pollen tubes are white.
Less intense GUS staining was detected in the short style of AtSUC1 promoter–GUS plants ( Fig. 3a). As in the pollen grains, this staining was absent from the styles of younger ovaries ( Fig. 3b).
In situ localization of AtSUC1 mRNA
To confirm the cell and tissue specificity of AtSUC1 expression determined in AtSUC1 promoter–GUS plants, we tried to localize AtSUC1 mRNA in Arabidopsis wild-type plants by in situ hybridization. Cross-sections of Arabidopsis anthers from different developmental stages were analyzed with radiolabelled sense and antisense probes derived from an AtSUC1 cDNA fragment. As expected from the results obtained with AtSUC1 promoter–GUS plants, AtSUC1 mRNA was detected in pollen grains during the latest stages of pollen development ( Fig. 4a–d). In Fig. 4(a–d), two different stages of late anther development are presented with the tapetum still being present in the developmentally younger anther shown in Fig. 4(a,b) and with the tapetum being already degraded in the older anther shown in Fig. 4(c,d). Moreover, in the anther presented in Fig. 4(c,d), the septums between the individual locules have already been cleaved and the anther is ready to dehisce. Prior to this final step in anther development large amounts of AtSUC1 mRNA accumulate in a ring of cells surrounding the vascular tissue of the connective tissue ( Fig. 4c). This site of AtSUC1 expression had not been detected in AtSUC1 promoter–GUS plants, probably due to the heavy GUS staining in the anthers.
The in situ hybridization analyses also detected AtSUC1 mRNA in female reproductive organs. In agreement with the GUS data ( Fig. 3a), mRNA accumulation was detected in the style, when radiolabelled AtSUC1 antisense RNA was hybridized to longitudinal sections of the pistil ( Fig. 4e) or to cross-sections through the style ( Fig. 4f). In addition, strong hybridization signals with AtSUC1 antisense RNA were found in the funiculi ( Fig. 4e,g), where AtSUC1 mRNA levels seem to be very high in the epidermal cells. Less strong signals were also obtained in the epidermal cell layer of the septum ( Fig. 4e,g).
Immunohistochemical localization of AtSUC1 protein in pollen and anther connective tissue
AtSUC1 protein was localized by indirect immunofluorescence using anti-AtSUC1 antiserum and visualized with anti-rabbit IgG-fluorescein isothiocyanate (FITC)-isomer 1-conjugate-labelled second antibodies. The antiserum used against recombinant AtSUC1 protein from transgenic baker’s yeast has previously been described and does not cross-react with the protein of the companion cell-specific sucrose transporter AtSUC2 ( Stadler & Sauer 1996).
In a first analysis cross-sections of Arabidopsis anthers from different developmental stages were analyzed ( Fig. 5a–d). As expected from the AtSUC1 promoter–GUS plants, no AtSUC1 protein was identified in sections from Arabidopsis anthers with still developing pollen (data not shown). However, sections from fully developed, still closed anthers also showed no FITC-dependent labelling after incubation with anti-AtSUC1 antibiodies ( Fig. 5a,b). In these anthers the tapetum was already degraded, but the pollen grains were still enclosed within the locules; cleavage of the stomium and the septum between the pair of locules had already commenced. This was in contrast to what had been shown with AtSUC1 promoter–GUS plants exhibiting strong GUS activity in mature pollen grains still enclosed within the anthers ( Fig. 3a–c). It was also in contrast to the results obtained with in situ hybridizations, where AtSUC1 mRNA had been identified in the pollen grains even at an earlier timepoint of anther development, when the tapetum was still present ( Fig. 4a). Immunohistochemical analyses of pollen grains that were still attached to the surface of already dehisced anthers showed that after the release of the pollen from the locules no AtSUC1 protein could be identified within the pollen grains ( Fig. 5c,d).
AtSUC1 protein in pollen grains was first detected during pollen germination on the papillae of the pistil ( Fig. 5e–h). With respect to AtSUC1 protein content and localization, different types of pollen grains could be distinguished on pollinated papillae: (i) pollen grains with no apparent AtSUC1 protein ( Fig. 5e); (ii) pollen grains with green FITC fluorescence confined to intracellular pollen compartments during germination ( Fig. 5f,g); and pollen grains with the green fluorescence localized at the surface of the pollen grain, putatively the plasma membrane ( Fig. 5h). Eventually, AtSUC1 protein was also detected in the pollen tubes of germinated pollen ( Fig. 5i).
These data suggest a role for AtSUC1 in the import of sucrose into growing Arabidopsis pollen. Analyses of pollen germination on media with different carbon sources ( Fig. 6) showed that sucrose is the only carbohydrate that supports growth of Arabidopsis pollen. Growth of pollen was not detected on media containing glucose, fructose or mannitol (control). We also tested the possibility that pollen tube growth on sucrose-containing medium results from extracellular sucrose hydrolysis by cell-wall bound invertases and the simultaneous availability of equimolar amounts of glucose and fructose. However, no pollen tube growth was detected on medium containing equimolar amounts of these monosaccharides.
AtSUC1 protein was also found at sites where it was predicted from the in situ hybridization experiments ( Fig. 4e,f). During the last stages of anther development, i.e. during or after anther dehiscence, AtSUC1 protein was also localized in few subepidermal cell layers surrounding the vascular bundle of the connective tissue ( Fig. 5c,d). In agreement with the in situ hybridization analyses, this localization of AtSUC1 protein is not detected at earlier times during anther development and is confined to a period where pollen development is complete, suggesting that AtSUC1 gene expression in the anther connective tissue is not needed for the channelling of sucrose towards developing pollen grains.
Immunohistochemical localization of AtSUC1 protein in female reproductive organs
Analyses of longitudinal and cross-sections from ovaries treated with anti-AtSUC1 antiserum-FITC also confirmed the style-specific expression of AtSUC1 that had been detected by in situ hybridization ( Fig. 4e,f) and in AtSUC1 promoter–GUS plants ( Fig. 3a). Green FITC-fluorescence was detected in a ring of parenchymatic cells between the xylem vessels ( Fig. 7a,b). The tissue inside this ring represents the upper end of the transmitting tract and it can be seen in Fig. 4(e) and 7(a) that the pollen tubes grow towards this transmitting tissue. As predicted from the in situ analyses AtSUC1 protein was also localized in the funiculi ( Fig. 7c,d). In agreement with the in situ data, AtSUC1 protein in the funiculi is seen almost exclusively in the epidermal cell layers.
AtSUC1 is important for pollen tube growth
Only recently a homolog of a plant monosaccharide transporter gene has been shown to be expressed in the pollen of Petunia hybrida ( Ylstra et al. 1998 ), and a sucrose transporter-like protein has been found in the pollen of tobacco ( Lemoine et al. 1999 ). Analyses of Petunia pollen germination on media with different carbon sources showed that this pollen can use monosaccharides as a carbon source for tube growth ( Ylstra et al. 1998 ). Deshusses et al. (1981) showed that 14C-labelled sucrose and glucose can be incorporated by lily pollen. However, the non-metabolizable glucose analogs 3-O-methylglucose and 2-deoxyglucose were shown to inhibit 14C-sucrose uptake, and extracellular sucrose hydrolysis by cell-wall bound invertase could not be fully excluded ( Deshusses et al. 1981 ).
Here we show that Arabidopsis pollen germinates and grows only on media containing the disaccharide sucrose ( Fig. 6). The described localization of the AtSUC1 sucrose transporter suggests that AtSUC1 is responsible for sucrose import into germinating pollen grains and into growing pollen tubes. Interestingly, AtSUC1 mRNA but not AtSUC1 protein is present in mature pollen grains after release from the anthers ( Figs 3c, 4c and 5c). This shows that AtSUC1 is not needed for carbohydrate import into developing pollen. Most importantly, however, this indicates that the translation of an already pre-existing AtSUC1 mRNA is initiated after pollination. The different subcellular localizations of AtSUC1 protein in pollen grains shortly after the transfer to the stigmatic papillae may thus represent the time course of AtSUC1 protein synthesis in intracellular membranes, such as the endoplasmic reticulum ( Fig. 5f,g) and the targeting of the protein to the plasma membrane ( Fig. 5g,h).
It has been found previously that polysome formation in hydrating pollen is extremely rapid and it has been postulated that masked, non-translatable mRNA molecules may be present in dehydrated pollen ( Linskens et al. 1970 ). This model was confirmed later, when it was shown that proteins needed during and after pollen germination can indeed be synthesized from mRNA pre-existing in the dehydrated pollen ( Capkova et al. 1987 , 1988; Jackson & Linskens 1982; Lin et al. 1987 ). It has been postulated that this mechanism enables rapid pollen germination and a high rate of pollen tube growth, primary factors governing the competition between pollen in reaching and effecting fertilization ( Mascarenhas 1990).
What role does AtSUC1 play in rapid pollen germination and tube growth? Obviously, AtSUC1 could be responsible for the import of sucrose as a carbon source for pollen metabolism. Alternatively, however, sucrose may not or not only be imported for metabolism but rather to increase the turgor of the pollen tube. As an osmotically active storage compound, sucrose may decrease the water potential thereby representing the osmotic driving force for rapid extension growth of the pollen tubes.
AtSUC1 expression in the connective tissue may trigger anther dehiscence
The identification of AtSUC1 protein in the connective tissue during the final stages of anther development and immediately before anther dehiscence ( Figs 4c and 5c) implies a physiological role of AtSUC1 at this step. Again, the function may be explained by an increase of the osmotic pressure in the AtSUC1-expressing cells. At the time of AtSUC1 protein synthesis both the anthers and the pollen grains are fully developed and an increased carbon metabolism in the connective tissue is unlikely. We postulate that the physiological function of AtSUC1 is the accumulation of sucrose within a ring of parenchymatic cells surrounding the connective tissue ( Figs 4c and 5c). This decrease in the water potential will result in water uptake from the adjacent cells of the anther walls representing the physical signal for a genetically controlled anther dehiscence. The water flow from the anther walls to the connective tissue will cause a dehydration of the endothecium and result in an increased tension of its U-shaped wall thickenings that will finally cause anther opening ( Keijzer 1985; Keijzer 1987). It has been discussed previously ( Keijzer 1987) that the important process of pollen release may be triggered by the plant and not only determined by the relative humidity at the time of bud opening.
Our model of a controlled anther opening is supported by the finding that after anther dehiscence only the anther walls dessicate whereas the filaments stay turgescent to expose the released anthers ( Keijzer 1987). Moreover, a male-sterile Arabidopsis mutant has been described ( Dawson et al. 1993 ) which produces fertile pollen gains that cannot be released because the anther walls do not open, even though both the stomiums and septums between the individual locules have been opened. In Arabidopsis, anthers dehisce after the opening of the floral bud ( Regan & Moffatt 1990). This process is paralleled by an onset of filament growth for optimal exposition of the pollen grains. At this time no growth has been detected in the connective tissue where AtSUC1 is expressed ( Smyth et al. 1990 ).
The possible function of AtSUC1 in the style and funiculi
The strong expression of AtSUC1 in a ring of cells surrounding the transmitting tissue of the style ( Figs 4f and 7a,c) and in the epidermal cell layers of the funiculi ( Figs 4e,g and 7c,d) is unlikely to reflect a special metabolic requirement of these cells for carbohydrates in contrast to all other cells of the ovaries. There are no reports on the symplastic isolation of these cells or on special biosynthetic activities. It can be seen from Fig. 4(e) and 7(a), and it has been described previously ( Bowman 1994; Kandasamy et al. 1994 ; Pruit & Hülskamp 1994), that pollen tubes grow towards the transmitting tissue and thus through the ring of AtSUC1-expressing cells ( Figs 4e,f and 7b). Kandasamy et al. (1994) showed that Arabidopsis pollen tubes after germination on the stigma surface converge into the stylar transmitting tissue as a dense bundle and that this growth pattern depends on the developmental stage of the pistil. It has also been shown previously that further down in the ovary, having emerged from the transmitting tissue onto the placenta, most pollen tubes travel up the funiculi to reach the micropyles ( Bowman 1994) and thus along the second site of strong AtSUC1 expression in the ovary. It is intriguing to speculate that AtSUC1 may be involved in pollen tube guidance, that the ring of AtSUC1 expression in the style may direct the pollen tubes into the transmitting tract, and that the high levels of AtSUC1 in the funiculi may guide the tubes towards the ovule.
One paper in favor of this possibility has been published recently ( Lush et al. 1998 ). These authors showed that the directional supply of water, i.e. the availability of water to the pollen tube, can establish a cue for the guidance of pollen tubes. The efficient accumulation of sucrose within the stylar ring cells and the epidermis of the funiculi could modulate the availability of water and could thus indeed represent a signal in pollen guidance.
Strains and growth conditions
Plants of Arabidopsis thaliana strain (L.) Heynh. C24 were grown in the greenhouse in potting soil or on agar in growth chambers under a 16 h light/8 h dark regime at 22°C and 55% relative humidity. Cloning in Escherichia coli was performed in strain DH5α ( Hanahan 1983). Plants were transformed with Agro- bacterium tumefaciens strain LBA4404 ( Ooms et al. 1982 ).
Isolation of AtSUC1 genomic sequences
Radiolabelled sequences of the AtSUC1 cDNA clone pTF2011 ( Sauer & Stolz 1994) were used to screen a λGEM11 genomic library of Arabidopsis thaliana Columbia. This library was made by J.T. Mulligan and R.W. Davis (Stanford, CA, USA) and was provided by the EEC Arabidopsis DNA stock center (Köln, Germany). The positive lambda clone λA8 contained 430 bp of 5′-flanking sequence which were used to rescreen the genomic library. A 3 kb BamHI fragment of the resulting clone λD was subcloned into pUC19. The resulting construct pET110 harbors 448 bp of AtSUC1 coding sequence and 2581 bp of 5′-flanking sequence. The EMBL accession number of the AtSUC1 promoter is AJ001364.
RNA isolation and RNase protection analysis
Arabidopsis tissue was collected and RNA was isolated as described previously ( Sauer et al. 1990 ). RNase protection assays ( Gahrtz et al. 1996 ; Ratcliffe et al. 1990 ) were performed with a radiolabeled 246 bp AtSUC1 antisense RNA probe. The hybridizing portion of this probe was 200 bp long. RNA was separated on 10% polyacrylamide gels.
Transformation of Arabidopsis
A 2.6 kb BamHI/XbaI fragment of pET110 covering the entire 5′-flanking sequence and 50 bp of AtSUC1 coding sequence was used to generate a translational fusion of AtSUC1 and GUS in pBI101 ( Jefferson et al. 1987 ) yielding the construct pBI101SUC1II. This construct was used for transformation of Arabidopsis as described previously ( Truernit & Sauer 1995).
Antibody preparation and immunochemical techniques
Anti-AtSUC1 polyclonal antibodies were raised against a fusion protein of the AtSUC1 C-terminus and β-galactosidase. The serum was purified by pre-adsorption to recombinant AtSUC1 protein isolated from transgenic Saccharomyces cerevisiae and tested for cross-reactions with AtSUC2 protein as described by Stadler & Sauer (1996).
For immunohistochemical analyses, Arabidopsis flowers were embedded in methacrylate and sectioned (1–3 μm) as previously described ( Stadler et al. 1995 ). Sections were treated with affinity-purified anti-AtSUC1 antiserum and immunostained with anti-rabbit IgG-fluorescein isothiocyanate (FITC)-isomer 1-conjugate labelled second antibody as published previously ( Stadler et al. 1995 ).
In situ hybridization with AtSUC1 probes
To generate radiolabelled antisense and sense probes for AtSUC1, a 1346 bp EcoRI/ScaI fragment of the AtSUC1 cDNA clone pTF2011 ( Sauer & Stolz 1994) was cloned into HincII/EcoRI digested pBluescript II SK-(Stratagene, La Jolla, USA). For the AtSUC1 antisense probe this construct was linearized with EcoRI and in vitro transcribed with T3 RNA polymerase using α35S-UTP. The resulting fragment covered 74 bp of the 5′-untranslated region and 1272 bp of the coding region of AtSUC1. For the AtSUC1 sense probe the construct was linearized with NotI and in vitro transcribed with T7 RNA Polymerase.
For in situ hybridization Arabidopsis flowers were embedded in paraffin and treated as described by Truernit et al. (1999) . Tissue sections were incubated with the labelled probes overnight at 50°C in 50% formamide, 10% dextransulfate, 0.3 m NaCl, 1 × Denhard’s solution ( Maniatis et al. 1982 ), 45 mm dithiothreitol, 10 mm Tris–HCl (pH 7.5), 1 mm EDTA. After hybridization sections were washed twice for 45 min at 45°C in 1 × SSC (= 0.15 m NaCl, 0.015 m sodium citrate), 50% formamide, 10 mm dithiothreitol. After an additional 5 min wash at room temperature in 1 × SSC, 10 mm dithiothreitol, the sections were treated with RNaseA [20 μg ml–1 in 10 mm Tris–HCl (pH 8.0)] for 30 min at 37°C. The sections were washed again twice for 45 min at 45°C in 1 × SSC, 50% formamide, 10 mm dithiothreitol and once for 5 min at room temperature in 1 × SSC, 10 mm dithiothreitol. Finally, the sections were dipped in Kodak NTB2 fotoemulsion (Integra, Fernwald, Germany) and exposed.
In vitro germination of pollen
In vitro germination of Arabidopsis pollen was performed on medium consisting of 0.4 mm CaCl2, 0.4 mm H3BO3 and 0.8% phytagar as described by Hülskamp et al. (1995) with modifications. After dissolving the agar by heating to 65°C, sucrose, glucose, fructose or mannitol were added to final concentrations of 10%, 5%, 5% or 5%, respectively (yielding almost equivalent molar concentrations). In one experiment 5% glucose and 5% fructose were added simultaneously.
Small amounts of the medium were pipetted onto microscopic slides and allowed to solidify. Pollen grains were applied to the medium by dipping flowers onto its surface. The pollen was allowed to germinate in a humid chamber at 22°C for 20 h. The percentage of germinated pollen from at least four independent flowers was determined by counting more than 200 pollen per flower.
We thank Gudrun Steinhäuser for skilful experimental help. The work was supported by a grant from the Deutsche Forschungsgemeinschaft (Sa 382 5–1).