Allocation of raffinose family oligosaccharides to transport and storage pools in Ajuga reptans: the roles of two distinct galactinol synthases


*For correspondence (fax +41 1634 82 04; e-mail


Raffinose family oligosaccharides (RFOs) are important phloem transport and storage carbohydrates for many plants. Ajuga reptans, a frost-hardy evergreen labiate, ideally combines these two physiological roles and served as our model plant to study the regulation and importance of RFO metabolism. Galactinol is the galactosyl donor for the synthesis of raffinose (RFO-trisaccharide) and stachyose (RFO-tetrasaccharide), and its synthesis by galactinol synthase (GolS) is the first committed step of the RFO biosynthetic pathway. Two cDNAs encoding two distinct GolS were isolated from A. reptans source and sink leaves, designated GolS-1 and GolS-2, respectively. Warm- and cold-grown sink and source leaves were compared, revealing both isoforms to be cold-inducible and GolS-1 to be source leaf-specific; GolS-1 expression correlated positively with GolS activity. Conversely, GolS-2 expression was comparatively much lower and its contribution to the total extractable GolS activity is most probably only minor. These observations, together with results from phloem exudation and leaf shading experiments suggest that GolS-1 is mainly involved in the synthesis of storage RFOs and GolS-2 in the synthesis of transport RFOs. Furthermore, in situ hybridization studies showed GolS-1 to be primarily expressed in the mesophyll, the site of RFO storage, and GolS-2 in the phloem-associated intermediary cells known for their role in RFO phloem loading. A model depicting the spatial compartmentation of the two GolS isoforms is proposed.


Plants are naturally exposed to changing environmental conditions imposing constraints on growth and development. To buffer such fluctuating periods of carbon supply and demand and to sustain the growth of heterotrophic organs, plants have the ability to store and translocate some of the fixed carbon. Classically, they do this by using starch for storage and sucrose for translocation. However, a great variety of alternative carbohydrates exists in the plant kingdom, with the RFOs being the most prominent ones ( Kandler & Hopf 1982; Keller & Pharr 1996).

Chemically, the RFOs may be considered extensions of sucrose to which varying numbers of α-galactosyl residues are attached. Their biosynthetic pathway is well established. Galactinol is the galactosyl donor for the biosynthesis of raffinose (degree of polymerization [DP] = 3) and stachyose (DP = 4), from sucrose and raffinose, respectively. The enzymes catalyzing these reactions are raffinose synthase (RS, EC and stachyose synthase (STS, EC; Kandler & Hopf 1982; Keller & Pharr 1996; Lehle & Tanner 1973). High-DP (DP > 4) RFO biosynthesis proceeds by galactinol-independent galactosyltransferase reactions employing the novel enzyme, galactan:galactan galactosyl transferase (GGT, Bachmann et al. 1994 ; Gilbert et al. 1997 ). GGT has been shown to use at least raffinose, stachyose and verbascose (DP = 5) as substrates (C. Inan Haab and F. Keller, manuscript in preparation). Therefore, RFO biosynthesis requires, directly or indirectly, the presence of galactinol, which is formed by galactinol synthase (GolS, EC from UDP-galactose and myo-inositol. To date, galactinol has not been assigned any function in plants other than acting as galactosyl donor for RFO synthesis ( Keller & Pharr 1996). GolS can therefore be rightly regarded as a key enzyme of RFO biosynthesis. This is supported by several studies showing a positive correlation between GolS activity and the onset of RFO accumulation in a variety of plants and tissues (for review see Keller & Pharr 1996).

GolS has mainly been characterized biochemically from legume seeds and cucurbit leaves ( Keller & Pharr 1996) and a few cDNAs encoding GolS have been isolated ( Kerr et al. 1993 ; Liu et al. 1998 ). Regulation of GolS expectedly includes de novo synthesis and/or post-transcriptional regulation, such as metabolite or effector control and covalent enzyme modifications. However, experimental data are rare ( Handley & Pharr 1982; Liu et al. 1998 ). GolS is localized in the cytosol together with RS and STS, whereas the subsequent galactinol-independent RFO chain elongation by GGT takes place in the vacuole starting most probably from stachyose ( Bachmann & Keller 1995; R. Braun, C. Inan Haab and F. Keller, unpublished observations).

Like sucrose, RFOs may be used for both phloem translocation and storage of carbon. It seems helpful therefore to distinguish between two carbohydrate pools in plants, a transport pool and a storage pool. In RFO translocating plants, such as those of the Cucurbitaceae, Lamiaceae, Oleaceae and Scrophulariaceae, the phloem-mobile RFOs (mainly raffinose and stachyose) are formed in intermediary cells (specialized companion cells). These closely surround the sieve elements and form a symplasm with the mesophyll and the sieve elements through which RFO phloem loading is thought to occur ( Beebe & Turgeon 1992; Turgeon 1996; van Bel 1993).

It is probable that most plants synthesize RFOs (at least raffinose) to some extent at some stage of their development, but many neither transport nor accumulate large amounts of them ( Keller & Pharr 1996). However, some plants store RFOs in concentrations of up to 25–80% of their dry weight in specialized storage organs such as tubers (e.g. Stachys sieboldii), seeds (e.g. soybean) or in photosynthesizing leaves (e.g. A. reptans) ( Bachmann et al. 1994 ; Handley et al. 1983 ; Keller & Matile 1985). In leaves, storage RFOs accumulate in the mesophyll, the largest leaf tissue ( Bachmann & Keller 1995), and RFOs can be induced to accumulate by cold treatment ( Bachmann et al. 1994 ).

We chose to work with A. reptans because it readily stores and translocates RFOs, suggesting that two spatially separated sets of enzymes responsible for RFO biosynthesis may be present. To decipher and understand the regulatory processes underlying RFO metabolism at the molecular level, we have started to clone and characterize genes encoding the enzymes involved in RFO metabolism. Here we report on the physiological characterization of two GolS isoforms from A. reptans leaves, GolS-1 and GolS-2, which are mainly involved in the synthesis of storage and transport RFOs, respectively.


Two distinct cDNAs encoding GolS were isolated from A. reptans leaves

Based on known GolS sequence data, degenerate DNA primers were designed and used for RT–PCR amplification of homologous sequences from A. reptans leaves. Initially, one cDNA fragment each from source (GolS-1) and sink (GolS-2) leaves were obtained and employed for RACE (rapid amplification of cDNA ends) experiments. For GolS-1, one 5′-cDNA end containing 54 nucleotides before a first ATG codon was obtained and, thereafter, a GolS-1 cDNA was directly amplified through RT–PCR. Starting at the first ATG codon, the cDNA clone comprised 1251 nucleotides including a poly(A)-tail and contained one open reading frame of 333 codons sequence-identical to the initially obtained GolS-1 cDNA fragments. For GolS-2, the obtained 5′- and 3′-cDNA fragments corresponded to partial cDNAs, not covering the whole 5′ coding region. The combined GolS-2 cDNA fragments contained one open reading frame, starting at GolS-1 codon position 40 and ending at position 333 ( Fig. 1).

Figure 1.

Comparison of deduced amino acid sequences from cDNAs encoding GolS.

GolS of Ajuga reptans source (Ar.GolS-1) and sink (Ar.GolS-2) leaves, Glycine max (Gm.GolS; Kerr et al. 1993 ), Cucurbita pepo (Cp.GolS; Kerr et al. 1993 ) and Oryza sativa (Os.Wsi76; Takahashi et al. 1994 ) . Numbers indicate codons of Ar.GolS-1. Amino acids in italics were used to design degenerate DNA primers. A putative serine phosphorylation site is marked by an arrow. Gaps, marked by dots, are included for a better match. Capital letters indicate identical amino acids. The sequences reported here have been deposited in the EMBL database (accession numbers: GolS-1: AJ237693; GolS-2: AJ237694).

The nucleotide and deduced amino acid sequences of GolS-1 and GolS-2 are highly homologous, showing an amino acid identity of 72% ( Fig. 1). Both are also highly homologous to known GolS from soybean, zucchini ( Kerr et al. 1993 ) and rice ( Takahashi et al. 1994 ). All the represented GolS show a putative serine phosphorylation site at GolS-1 codon position 263, and besides the high overall identity there is a characteristic hydrophobic pentapeptide (APSAA) at the carboxy terminal end of all known GolS. The GolS-1 cDNA was expressed in Escherichia coli yielding a functional GolS; no GolS activity was detectable in control bacteria (data not shown).

GolS-1 and GolS-2 were cold-induced and GolS-1 was source leaf-specific and probably involved in storage RFO synthesis

Warm- and cold-grown sink and source leaves were compared with regard to their GolS-1 and GolS-2 transcript levels, GolS activities, and carbohydrate concentrations ( Fig. 2). Under warm growing conditions, only source leaves showed considerable GolS activity, which correlated positively with the concentrations of galactinol, raffinose, stachyose and high-DP RFOs ( Fig. 2b–d) and with GolS-1 expression ( Fig. 2a). In contrast, GolS-2 expression was equally weak in all leaf types ( Fig. 2a). The myo-inositol concentrations were similar in sink and source leaves and twice as much sucrose was present in source leaves than in sink leaves ( Fig. 2c).

Figure 2.

GolS RNA levels, GolS enzyme activities and carbohydrate concentrations in sink and source leaves from warm- and cold-grown plants.

(a) Radiosignals of 32P-labelled GolS-1 and GolS-2 hybridized on RNA gel blots and ethidium bromide staining of the total RNA loaded. (b–d) Extractable GolS activity and carbohydrate concentrations of leaves from warm-grown plants. Very small, still rolled-up leaves and just unrolled leaves are designated as ‘sink 1’ and ‘sink 2’, respectively, while leaves larger than 3 cm2 are designated as ‘source’ ( Bachmann et al. 1994 ). (e–g) are the same as (b–d), but are of leaves from cold-grown plants. Data points are means ± SE of two replicates.

FW, fresh weight; GolS, galactinol synthase; Gol, galactinol; Ino, myo-inositol; Suc, sucrose; Raf, raffinose; Sta, stachyose; > Sta, RFOs larger than stachyose.

Under cold growing conditions, GolS-1 and GolS-2 transcript levels and GolS activities were considerably higher than under warm growing conditions ( Fig. 2a,b,e). The expression pattern of GolS-1 was similar to that of the GolS enzyme activity. GolS-1 transcript levels were highest in source leaves and decreased with decreasing leaf size, while GolS-2 transcript levels were equal in source and sink leaves ( Fig. 2a). Similarly, galactinol and RFO concentrations correlated positively with GolS-1 expression and GolS activity in source and sink 2 leaves, but not in sink 1 leaves ( Fig. 2a,e–g). Although only very little GolS activity was found in sink 1 leaves, galactinol and RFO concentrations were as high as in sink 2 leaves. myo-Inositol concentrations were similar in sink 2 and source leaves, but considerably lower than in sink 1 leaves. Sucrose concentrations were equally high in all cold-grown leaves ( Fig. 2f,g).

A leaf shading experiment was performed using terminal leaf pairs of excised runners. One leaf per pair was shaded and, thereby, transformed into a sink leaf, while the other leaf was kept in the light and remained a source leaf, presumably allocating carbon to the shaded leaf ( Fig. 3). In the illuminated leaves, GolS-1 expression decreased during the first 24 h, but increased again later. The raffinose and stachyose concentrations followed a similar pattern. Concomitant with the GolS-1 increase at 96 h, galactinol levels almost doubled ( Fig. 3a,b). Conversely, GolS-2 expression increased in the first 24 h and then decreased again ( Fig. 3a). In the shaded leaves, GolS-1 expression decreased during incubation, whereas GolS-2 remained expressed at a constantly low level ( Fig. 3a). Galactinol concentrations dropped by 50% and the RFOs disappeared almost completely ( Fig. 3c).

Figure 3.

GolS RNA levels and carbohydrate concentrations of illuminated and shaded leaves of terminal leaf pairs of excised runners.

(a) Radiosignals of 32P-labelled GolS-1 and GolS-2 hybridized on RNA gel blots and ethidium bromide staining of the total RNA loaded.

(b,c) Extractable carbohydrate concentrations in terminal leaf pairs of excised runners. One leaf of a pair was kept in the light (b), the other one was shaded (c). Data points are means ± SE of two replicates.

FW, fresh weight; Gol, galactinol; Suc, sucrose; Raf, raffinose; Sta, stachyose.

In summary, the positive correlation of extractable GolS activity with galactinol level, RFO accumulation and GolS-1 expression (as shown in Figs 2 and 3) suggests that GolS-1 is mainly involved in the synthesis of storage RFO.

GolS-2, but not GolS-1, appeared to be involved in transport RFO synthesis

Phloem exudation experiments were performed to study the roles of GolS-1 and GolS-2 in RFO phloem transport. Excised and EDTA-treated leaves from warm-grown plants were either kept in the warm or transferred to the cold ( Figs 4 and 5). When kept in the warm, the leaves readily decreased GolS-1 expression ( Fig. 4a). This decrease did not appear to be due to a general decrease in the amount of total RNA per leaf dry weight, because equal amounts of RNA could be extracted per dry weight of leaf material (data not shown). GolS-2 transcripts were barely detectable, but in contrast to GolS-1, they remained relatively constant ( Fig. 4a). In parallel, these leaves readily exuded stachyose, raffinose and sucrose after as little as 8 h of incubation in EDTA ( Fig. 4b).

Figure 4.

GolS RNA levels and phloem carbohydrate exudation of excised leaves from warm-grown plants incubated in the warm or transferred to the cold. Leaves were EDTA-treated to enable phloem exudation.

(a) Radiosignals of 32P-labelled GolS-1 and GolS-2 hybridized on RNA gel blots and ethidium bromide staining of the total RNA loaded.

(b,c) Exuded carbohydrate pools of leaves kept in the warm (b) or transferred to the cold (c). Data points are means ± SE of three replicates.

FW, fresh weight; Suc, sucrose; Raf, raffinose; Sta, stachyose.

Figure 5.

GolS enzyme activities and carbohydrate concentrations of the excised leaves of Fig. 4.

(a–d) Extractable GolS activity and carbohydrate concentrations in excised leaves from warm-grown plants either incubated in the warm (a,b) or the cold (c,d) are presented. Data points are means ± SE of three replicates.

FW, fresh weight; GolS, galactinol synthase; Gol, galactinol; Suc, sucrose; Raf, raffinose; Sta, stachyose.

When warm-grown leaves were transferred to the cold, GolS-1 expression remained unchanged whereas GolS-2 expression was transiently induced after 8 h ( Fig. 4a). Interestingly, these leaves exuded only small amounts of sugars as compared to leaves kept in the warm ( Fig. 4b,c). A similar negative correlation between GolS-1 expression and RFO phloem exudation was found when leaves from cold-grown plants were either kept in the cold (not exuding) or transferred to the warm (exuding; data not shown).

The decrease of GolS-1 expression in the leaves that were kept in the warm ( Fig. 4a) did not result in a marked change of GolS activity and carbohydrate concentration with the possible exception of sucrose levels which increased slightly ( Fig. 5a,b). When the leaves were transferred to the cold, their GolS activities increased considerably and their galactinol concentrations remained constant ( Fig. 5c). They accumulated primarily sucrose, already within the first 4 h of incubation, while raffinose and stachyose concentrations increased only slightly ( Fig. 5d).

The possibly distinct roles of GolS-1 and GolS-2 in storage and transport RFO synthesis, respectively, were further studied by in situ hybridization experiments with leaf cross-sections. Visual inspection of 30 sections of 5–10 leaves revealed that GolS-1 was primarily expressed in the mesophyll (in over 85% of the mesophyll cells; Fig. 6a,b), while GolS-2 transcripts were primarily found in intermediary cells, which are grouped around the sieve elements of the phloem ( Fig. 6c,f). No expression signals were detected in control sections ( Fig. 6c,f).

Figure 6.

In situ hybridization of GolS-1 and GolS-2 in warm-grown source leaves.

(a,b,d,e) Light micrographs of leaf cross-sections showing GolS-1 and GolS-2 expression, respectively, through in situ hybridization with digoxygenin-labelled antisense GolS-1 and GolS-2 RNA, respectively, and detection using alkaline phosphatase-coupled anti-digoxygennin antibodies. (c,f) Light micrograph of control leaf cross-sections treated like (a,b,d,e), but using sense GolS-1 and GolS-2 RNA, respectively. The vascular tissue is surrounded by bundle-sheath cells (BS), and is composed of the vascular parenchyma (VP), tracheary elements (T), sieve elements (SE) and intermediary cells (IC).

M, mesophyll cell; E, epidermis cell. Bar = 10 μm.

In summary, the negative correlation between GolS-1 expression and RFO phloem exudation and the GolS-2 expression observed in the intermediary cells (and not in the meophyll or epidermis) suggests that GolS-2 rather than GolS-1 is involved in transport RFO synthesis.


RFOs may fulfil at least two major physiological roles, namely storage and translocation of carbon. In A. reptans leaves, two different RFO pools are present, a long-term RFO storage pool in the mesophyll and a transport pool in the phloem ( Bachmann & Keller 1995; Bachmann et al. 1994 ). The RFO transport pool mainly consists of raffinose and stachyose, which are synthesized by a galactinol-dependent pathway ( Kandler & Hopf 1982; Keller & Pharr 1996), while the storage pool additionally consists of high-DP RFOs, which are synthesized by a galactinol-independent pathway using the novel chain elongation enzyme, GGT ( Bachmann et al. 1994 ). In both cases, RFO biosynthesis can only be initiated and maintained by the availability of galactinol. Therefore, its anabolic enzyme, GolS, can be regarded to catalyze the key regulatory step for RFO biosynthesis. Here we show that at least two GolS isoforms are present in A. reptans leaves, GolS-1 and GolS-2, and propose a model depicting that these two isoforms are differentially recruited for their respective physiological roles, i.e. RFO synthesis for storage in the mesophyll (GolS-1) and for translocation in the phloem (GolS-2) ( Fig. 7).

Figure 7.

Current model depicting the compartmentation and physiological roles of GolS-1 and GolS-2.

The large oval depicts the central vacuole of a mesophyll cell. FW, fresh weight; GolS, galactinol synthase; Gol, galactinol; Ino, myo-inositol; Suc, sucrose; Raf, raffinose; Sta, stachyose.

GolS-1 expression levels and GolS activities ran parallel with the accumulation of storage RFOs in source leaves grown in the cold or the warm ( Figs 2–5). This indicates that the regulation of RFO biosynthesis is at least partly based on a de novo synthesis of GolS, a notion which was further substantiated by the finding that cycloheximide (2 μm) inhibited the induction of GolS activity in excised leaves (data not shown).

It is interesting to note that very small sink leaves (sink 1; rolled up) of cold-grown plants accumulated considerable amounts of RFOs without containing significant GolS activity. Nevertheless, a significant galactinol concentration was present and, furthermore, the myo-inositol concentration was up to twice that of all other leaves ( Fig. 2). These findings could be partly explained by the model recently proposed for RFO catabolism in sink tissues ( Madore 1995), which states that stachyose is first hydrolyzed by an alkaline α-galactosidase and the resulting raffinose is then degraded by the reverse reaction of RS producing sucrose and galactinol. Similarly, galactinol was suggested to be catabolized by the reverse reaction of GolS producing myo-inositol and UDP-galactose. The net reaction products of these three catabolic steps would, theoretically, be sucrose, myo-inositol, UDP-galactose and galactose. The measured carbohydrate pools of A. reptans sink leaves and the finding that both the forward and reverse reactions of RS can be measured in vitro (B. Schäfer and F. Keller, unpublished observations) are in line with such a model for RFO catabolism. It is most likely that, in cold-grown sink leaves, galactinol was not catabolized completely, but directly reused for RFO synthesis. Consequently, myo-inositol would accumulate due to raffinose and stachyose formation by RS and STS, and due to the absence of GolS activity which otherwise would reduce the myo-inositol pool by forming galactinol. In summary, such a scenario would make high activities of GolS superfluous for RFO accumulation in RFO importing leaves.

Cold growing conditions induced GolS-1 and GolS-2 expression and subsequently led to the accumulation of RFOs in leaves. In intact plants, accumulation of RFOs started after 3 days and was pronounced after 1 week of cold treatment ( Bachmann et al. 1994 ). Seedlings of a chilling-sensitive rice variety exposed to a short osmotic stress attained some degree of chilling resistance and concomitantly expressed an initially unknown gene named wsi 76 (wsi = water-stress-induced; Takahashi et al. 1994 ), which most likely encodes a GolS ( Fig. 1). The wsi 76 expression in rice seedlings was reported to be induced within 1 h of osmotic stress treatment. Similarly, GolS expression increased in Arabidopsis when exposed to 4°C for 8 h ( Liu et al. 1998 ). In excised A. reptans leaves, GolS-1 expression did not markedly increase within 72 h of cold treatment, whereas GolS-2 expression was clearly induced already after 8 h in the cold (9/3°C; day/night) ( Fig. 4). In intact plants, both GolS-1 and GolS-2 expressions increased only slightly in 8 days in the cold (data not shown), but were clearly induced after a cold treatment of at least 2 weeks ( Fig. 2). Several studies have indicated a general correlation between RFO accumulation and exposure to cold, salt or drought implying a role of RFOs in stress adaptation, but direct experimental evidence is still missing (for review see Keller & Pharr 1996).

A basic GolS-1 expression was always present in source leaves. However, GolS-1 transcript levels clearly decreased whenever excised source leaves were confronted with a strong sink situation such as forced exudation of RFOs through the phloem. Conversely, GolS-2 expression remained rather constant in such situations. Leaves that could not export RFOs, either due to cold ( Fig. 4) or callose formation in the sieve elements (data not shown), showed a constant GolS-1 and a transiently induced GolS-2 expression. Moreover, shading of one leaf of a leaf pair led, in the illuminated leaf (source), to an initially increased GolS-2 and a decreased GolS-1 expression which were paralleled by a decreased RFO concentration ( Fig. 3). Assuming that the illuminated leaf (source) exported RFOs to the shaded leaf (sink), the observed GolS expression patterns suggest that GolS-2 is involved in the synthesis of RFOs used for transport. GolS-1 expression increased later in the source leaf when the RFOs started to accumulate again.

We had hypothesized earlier that in A. reptans leaves two RFO biosynthetic machineries were present, one in the mesophyll and one in the intermediary cells ( Bachmann & Keller 1995). Comparison of whole leaves with mesophyll protoplasts from A. reptans had shown that the bulk of the enzymes involved in RFO synthesis was located in the mesophyll, the site of RFO storage ( Bachmann & Keller 1995). Here we show that GolS-1 expression is by far more dominant than GolS-2 expression and therefore GolS-1 is most likely involved in the build-up of the large RFO storage pool. This notion is supported by the primary expression of GolS-1 in the mesophyll ( Fig. 6). Conversely, GolS-2 is mainly expressed in intermediary cells which, together with the results from other experiments presented here, strongly suggests that GolS-2 is responsible for the synthesis of transport RFOs. We would like to point out, however, that we do not totally exclude the possibility that GolS-2 may also participate, at least to some extent, in the synthesis of storage RFO for reasons implied by the observed cold induction of GolS-2 expression. Similarly, a participation of GolS-1 in the synthesis of transport RFO cannot be totally ruled out, but seems to be rather unlikely because of the observed negative correlation of its expression with phloem exudation.

In conclusion, we report here on two distinct GolS isoforms and provide experimental evidence that allows us to ascribe two distinct physiological roles to them in RFO synthesis, for carbon storage and transport, respectively ( Fig. 7). GolS-1 shows an expression pattern that is in line with its proposed role in storage RFO synthesis (in the mesophyll). GolS-2 expression pattern and localization in the phloem-associated intermediary cells suggest a role in transport RFO synthesis.

Experimental procedures

Plant material and experimental set-up

Ajuga reptans L. plants were cultivated hydroponically in a substrate of Perlite. Plants were kept in a growth chamber with a photoperiod of 12 h (90 μmole m−2 s−1). Warm-grown plants were cultivated at 20/15°C (day/night). For cold treatments, plants were transferred to a similar chamber kept at 9/3°C (day/night) for at least 2 weeks. Plants were fertilized weekly with a standard nutrient solution. Leaves were harvested 6 h into the photoperiod, frozen in liquid nitrogen, and stored at −80°C. Leaves were defined as ‘sink 1’ (rolled up; < 1 cm2); ‘sink 2’ (just unrolled; ≈ 1 cm2); and ‘source’ (> 3 cm2) ( Bachmann et al. 1994 ).

For shading experiments, runners with a terminal leaf pair were excised 4 h into the photoperiod and incubated in water for up to 96 h under warm growing conditions. One leaf of each pair was covered with aluminum foil. Leaves were harvested, frozen in liquid nitrogen and stored at −80°C.

For phloem exudation experiments, the EDTA method was used ( Bachmann et al. 1994 ; King & Zeevart 1974). Leaves of warm-grown plants were excised 2 h into the photoperiod, trimmed submerged in exudation buffer (EB, 10 m m Tris/HCl, pH 7.5, 2 m m Na2EDTA), incubated in EB for 30 min, and then transferred into test tubes containing EB (2 ml) for incubation under warm or cold growing conditions. Solutions in the test tubes were replaced with new EB every 8–14 h. Leaves were harvested, frozen in liquid nitrogen, lyophilized and stored in a dessicator. Exudates were stored at −20°C before analysis.

Cloning and molecular analysis

Unless otherwise stated, standard protocols were used for molecular analysis ( Sambrook et al. 1989 ). Sequences were analyzed with the GCG software package, version 8.1 (1995).

RNA extraction

Leaves (frozen or lyophilized) were ground in a mortar to a fine powder and total RNA was extracted on ice according to Wadsworth et al. (1988) . The RNA extraction buffer contained 25 m m Na-citrate, pH 7, 4 m guanidinium isothiocyanate, 1.5% (w/v) Na-lauryl sarcosine, 100 m m 2-mercapthoethanol, 50 m m Na-ascorbate, 5% (w/v) polyvinylpyrrolidone (Polyklar AT). Approximately 200 mg fresh weight or 20 mg dry weight were extracted in 0.5 ml buffer.

Cloning and sequencing

Single-stranded cDNA (ss-cDNA) was synthesized from total RNA (2 μg) isolated from source or sink leaves as described above. Reverse transcription was done using the DNA primer oligo(T)-anchor (3 μm; Table 1) and 2.5 units AMV-reverse transcriptase (Promega, Madison, WI, USA), according to a standard protocol (Promega). Unless otherwise stated, 0.2 volumes of ss-cDNA were used for PCR using DynaEX DNA polymerase (Finnzymes, Espoo, Finnland) and the following protocol: 94°C for 2 min; 35 cycles of 94°C for 0.5 min, 52°C for 1 min and 72°C for 1.5 min; 72°C for 8 min. Degenerate DNA primers were designed according to conserved amino acid regions of known GolS sequences ( Fig. 1) and two Arabidopsis genes (F8A5.2, PID:g2462751 and T08I13.2, PID:g2275196), and 5′ DNA primers GolS_5a or GolS_5b were used together with degenerate 3′DNA primer GolS_3a (1 μm each) in a PCR (35 cycles; Table 1). This yielded one product each from source (GolS-1) and sink (GolS-2) leaves. These were ligated into the plasmid vector pGEM-T (Promega), sequenced and utilized to design specific DNA primers for rapid amplification of cDNA ends (RACE) PCR ( Table 1).

Table 1. . Sequences of DNA primers used for PCR amplifications
PrimerSequence a
  • a

    N is an A, C, G, T residue.

  • b

    Nucleotides in italics were included for subcloning.


For 5′-RACE of GolS-1, ss-cDNA was synthesized from source leaves as described, but using primer GolS_3a to initiate the reverse transcription. Resulting ss-cDNA was purified using a Centricon-30 microconcentrator (Amicon, Beverly, MA, USA) and subjected to a poly(A)-tailing reaction at 37°C for 20 min using 10 units terminal transferase (Amersham-Pharmacia, Uppsala, Sweden) in the presence of 0.25 μm dATP. After heating at 70°C for 5 min, the reaction was diluted five times for second strand cDNA synthesis using the primer oligo(T)-anchor. The cDNA was purified using a Centricon-30 microconcentrator and used for PCR (35 cycles) using the primers anchor and GolS_3a at an annealing temperature of 50°C. An aliquot (0.1 volumes) was subjected to a second PCR (35 cycles) using the primers anchor and GolS1–5ra at an annealing temperature of 60°C. The product was ligated into the plasmid vector pGEM-T (Promega) and sequenced. The resulting GolS-1 sequence served to design a specific DNA primer (GolS1_ADA) that covered a first ATG codon, which was used together with primer anchor to directly amplify a GolS-1 from ss-cDNA performing a PCR (35 cycles) at an annealing temperature of 55°C. The resulting cDNA was ligated into pGEM-T easy (Promega) and sequenced.

All GolS-2 specific PCR were done using ss-cDNA obtained from sink leaves. For 5′-RACE ss-cDNA was poly(A)-tailed as described and subjected to PCR (three cycles) with only the primer oligo(T)-anchor at an annealing temperature of 37°C. Thereafter, 0.05 volumes were subjected to PCR (35 cycles) using the primers anchor and GolS2–5ra at an annealing temperature of 65°C. For 3′-RACE of GolS-2 two subsequent PCR (35 cycles each) were performed. First, using the primers GolS2–3ra and anchor at an annealing temperature of 55°C and, second, an aliquot (0.01 volumes) was reamplified using primers GolS2–3rb and anchor. The products were ligated into pGEM-T (Promega) and sequenced.

Sequencing was carried out using a PRISMTM Ready Reaction DyeDeoxy Terminator cycle sequencing kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer‘s instructions and an ABI 373 DNA Sequencer (Applied Biosystems).

RNA gel blot analysis

Total RNA (approximately 10 μg per lane) was denatured for 15 min at 65°C, separated on a 1% (w/v) denaturing agarose gel and blotted onto a positively charged nylon membrane (GeneScreen plus, NENTM, Boston, MA, USA). RNA blots were probed with a cDNA fragment of the GolS-1 comprising the 5′ coding region to codon position 266 and with a cDNA fragment of the GolS-2 comprising the 5′ region to GolS-1 codon position 231 ( Fig. 1). The cDNA was labelled with α32P dCTP using a random primed labelling kit (Amersham-Pharmacia). Hybridization and washing was performed at 65°C using high stringency conditions. Radiosignals were visualized on X-ray film.

In situ hybridization

Leaves of warm-grown plants were fixed in 50% (v/v) ethanol containing 3.7% (v/v) formaldehyde and 5% (v/v) acetic acid. Embedding, hybridization and detection were carried out according to Schneitz et al. (1998) with minor adaptations. Specimens were digested in proteinase K for 20 min at 23°C. Labelling of GolS-1 and GolS-2 antisense and sense RNAs was done by run-off transcription using SP6 and T7 RNA polymerases and ribonucleotides containing digoxygenin-UTP (Boehringer-Mannheim, Mannheim, Germany) according to the manufacturer‘s instructions. Label was detected through alkaline phosphatase coupled to anti-digoxygenin antibody Fab fragments (Boehringer-Mannheim) and BCIP-NBT substrates. Sections were observed using a microscope equipped with differential-interference-contrast optics.

Enzyme extraction and assay

Ground frozen or lyophilized leaf material (200–300 mg fresh weight, FW; 10–20 mg dry weight, DW) was immediately suspended in ice-cold extraction buffer at a ratio of 2.5 ml per g FW or 25 ml per g DW. Extraction buffer was composed of enzyme buffer (25 m m HEPES/KOH, pH 7.5, 2 m m MgCl2, 10 m m DTT, 0.1% (v/v) Triton X-100, 0.1% NaN3) containing 0.1 m m benzamidine, 1 m m PMSF, 10 m m 2-mercaptoethanol, 50 m m Na-ascorbate and 2% (w/v) polyvinylpolypyrrolidone (PVPP). Samples were mixed and centrifuged at 15 000 g for 10 min at 4°C and supernatants desalted on a Sephadex G-25 column ( Bachmann et al. 1994 ). Aliquots of the resulting extracts were assayed for GolS activity in enzyme buffer containing 100 m mmyo-inositol and 5 m m UDP-galactose. Enzyme assays were either boiled immediately (control) or incubated at 30°C for 15 min and boiled thereafter for 3 min. Samples were processed for carbohydrate analysis as described below.

Carbohydrate extraction and analysis

Carbohydrates were either extracted by the cryo sap method (fresh material) ( Bachmann et al. 1994 ) or ethanol extraction (lyophilized material) ( Sprenger et al. 1995 ). Aliquots of such extracts or of boiled enzyme assays were deionized and analyzed using HPLC as described previously ( Bachmann et al. 1994 ). myo-Inositol was also quantified after separation on an anion exchange column (MA1; 4 × 250 mm; room temperature; Dionex, Sunnyvale, CA, USA) run at 0.4 ml min−1 with 0.6 m NaOH. Quantification was done by the external standard method using authentic myo-inositol, galactinol, sucrose, raffinose, stachyose and verbascose. RFO with DP > 5 were quantified by using the response factor of verbascose.


We kindly thank Antje Redweik and Robert Dudler for excellent technical assistance and support, Jean-Jacques Pittet for help with the illustrations, and Stefan Hörtensteiner and Markus Bachmann for critically reading the manuscript. This work was supported by the Swiss National Science Foundation.


  1. EMBL accession numbers GolS-1 ( AJ237693) and GolS-2 ( AJ237694).