Nectar formation and floral nectary anatomy of Anigozanthos flavidus: a combined magnetic resonance imaging and spectroscopy study.

Metabolic processes underlying the formation of floral nectar carbohydrates, especially the generation of the proportions of fructose, glucose, and sucrose, are important for understanding ecological plant-pollinator interactions. The ratio of sucrose-derived hexoses, fructose and glucose, in the floral nectar of Anigozanthos flavidus (Haemodoraceae) was observed to be different from 1:1, which cannot be explained by the simple action of invertases. Various NMR techniques were used to investigate how such an unbalanced ratio of the two nectar hexoses can be formed. High-resolution (13)C NMR spectroscopy in solution was used to determine the proportion of carbohydrates in vascular bundles of excised inflorescences fed with (13)C-labelled carbohydrates. These experiments verified that feeding did not affect the metabolic processes involved in nectar formation. In vivo magnetic resonance imaging (e.g. cyclic J cross-polarization) was used to detect carbohydrates in vascular bundles and (1)H spin echo imaging non-invasively displayed the architecture of tepal nectaries and showed how they are connected to the vascular bundles. A model of the carbohydrate metabolism involved in forming A. flavidus floral nectar was established. Sucrose from the vascular bundles is not directly secreted into the lumen of the nectary but, either before or after invertase-catalysed hydrolyses, taken up by nectary cells and cycled at least partly through glycolysis, gluconeogenesis, and the pentose phosphate pathway. Secretion of the two hexoses in the cytosolic proportion could elegantly explain the observed fructose:glucose ratio of the nectar.

Introduction

The chemical composition of nectar is of immense ecological importance, especially for plant–pollinator interaction. The concentrations and relative proportions of sugars, amino acids, and other nectar components are specifically adapted to the requirements of the pollinator type and therefore show substantial chemical diversity (Elisens and Freeman, 1988; Baker and Baker, 1990; Cronk and Ojeda, 2008). Moreover, the evolutionary origin and architecture of nectaries are as diverse as the nectar supply seems to be.

Despite a plethora of publications about nectaries, nectar formation and nectar composition (reviewed, for example, by Lüttge, 1961; Fahn, 1979, 2000; Pacini ; De la Barrera and Nobel, 2004), the mechanisms determining nectar composition are not yet fully understood. The ultrastructural and cell-specific localization of the processes involved in generating nectar is similarly unclear. Nectaries should not be assumed always to be capable of fine-tuning the chemical composition of the nectar; in some plants, they may function only as tissue secreting phloem sap. Relatively little is known about the genetic background of nectary differentiation or the regulation of nectar production. Only one gene (NEC1) thought to be involved in the nectary differentiation has been described (Ge ). Two nectar quantitative trait loci (QTL), that help control the amount of nectar produced, have been identified in Mimulus and Petunia (Bradshaw ; Stuurman ). In Petunia, Stuurman found another QTL; this controls the ratio of hexoses to sucrose, which may coincide with invertase activity. Invertase-catalysed hydrolysis of sucrose seems to be responsible for the ratio of hexoses to sucrose (Nicolson and Fleming, 2003) in the nectar. However, the simple action of invertase should result in a glucose-to-fructose ratio of 1:1, which is not always the case (Baker ; Chalcoff ; Wolff, 2006; Hölscher ). Hence, additional enzymes and transporters must be involved in generating the chemical composition of nectars. Acidic phosphatases have been reported to play a role in nectar secretion (Ziegler, 1956).

The nectar composition of many plants has been studied for more than 50 years using various chromatographic and spectroscopic techniques. In a recent study, nuclear magnetic resonance (NMR) spectroscopy was used for metabolic profiling of Anigozanthos floral nectar (Hölscher ). 13C NMR was used to determine the major carbohydrates quantitatively, which in this nectar are glucose and fructose. Sucrose, which usually is one of the major sugars in nectar, was detected in low concentrations, together with other minor nectar components such as amino acids only by using the much more sensitive 1H NMR spectroscopy. In addition to the very low sucrose level, interestingly, the Anigozanthos floral nectar was found to contain glucose and fructose in a ratio of 1.2:1.0 (Hölscher ).

The observation of a glucose-to-fructose ratio different from 1:1 prompted the present work. NMR spectroscopy and magnetic resonance imaging (MRI) were used to investigate both nectar formation and the anatomy of the nectary. In contrast to light and electron microscopy, the usual methods for studying the architecture of nectaries (Figier, 1968; Simpson, 1993), MRI is non-destructive. It requires only limited sample preparation and avoids artefacts. The cyclic J cross-polarization MRI method developed and demonstrated on castor bean seedlings by Heidenreich was used in the present study to observe the sugar composition and distribution in the vascular bundles of excised inflorescences of Anigozanthos flavidus. Another aim was to validate excised inflorescences as a model system, especially to verify if experimental manipulation, for example, feeding 13C-labelled carbohydrates, affects the metabolic processes involved in nectar formation. In this study excised inflorescences of A. flavidus were used for precursor feeding experiments and single flowers for MRI. A. flavidus, a Haemodoraceae native to Southwestern Australia, is interesting due to the accumulation of phenylphenalenones and other secondary metabolites with ecological implications.

Materials and methods

Plant material

Plants of A. flavidus DC. were raised from seeds (Chiltern Seeds, Bortree Stile, UK) and grown in soil in the greenhouse. The temperature was 20–24 °C during the day and 18–21 °C during the night. Relative air humidity was between 60% and 70%. The natural daily photoperiod was supported with 16 h illumination from Phillips Sun-T Agro 400 Na lights.

Labelled compounds

[U-13C]glucose (99% 13C) and [1-13C]fructose (99% 13C) were from Deutero GmbH, Kastellaun, Germany.

Light microscopy

Peduncles of A. flavidus were cut with a razor blade (Dovo Merkur, Solingen, Germany) to get extra-thin hand-made sections. The sections were made at ×40 magnification (objective Leica 506074 C Plan 4×/0.10) on a Leica DM6000B light microscopy system.

Feeding experiments

For nectar collection experiments, excised inflorescences of A. flavidus were allowed to take up [U-13C]glucose or [1-13C]fructose hydroponically from 20 mM and 25 mM aqueous solutions, respectively. Nectar was collected twice from each flower 24 h and 48 h after transfer of the excised inflorescences to the feeding solutions. For all other labelling experiments, cut A. flavidus flowers were fed overnight in 10–500 mM solutions of [U-13C]glucose or 5 mM solutions of [1-13C]fructose.

Nectar collection and sample preparation

In the feeding experiments, approximately 15 μl nectar per flower was collected with Eppendorf tips from inflorescences of A. flavidus. In the [U-13C]glucose feeding experiment, 30 μl nectar was dissolved in 500 μl D2O and subjected to 13C NMR. In the [1-13C]fructose feeding experiment 80 μl nectar was dissolved in 600 μl D2O and evaporated to dryness under a stream of N2 gas to remove H2O. This procedure was repeated once. Finally, the sample was subjected to NMR analysis in D2O. At the start of the feeding experiments, all nectar present in the flowers was collected, subjected to the same sample preparation as described above and used as a reference.

NMR

1H and 13C NMR data were recorded on a Bruker AVANCE 400 spectrometer operating at a 1H frequency of 400.13 MHz and a 13C frequency of 100.61 MHz, respectively. All spectra were recorded in D2O at 25 °C. 1H NMR spectra were measured with presaturation of the HDO signal. 13C NMR experiments were executed using power gated decoupling. 3-(Trimethylsilyl) 2,2,3,3-tetradeuteropropionic acid (TSP) was used as an internal standard for referencing 1H NMR spectra.

Magnetic resonance imaging/chemical shift imaging

All experiments were carried out on Bruker AVANCE 400 spectrometers equipped with Bruker microimaging systems. 1H spin echo images of flowers were acquired using a ‘Micro 5’ RT-probe with an exchangeable 5 mm coil insert. A 1H13C cyclic J cross-polarization image (Heidenreich ) of a 5 mm slice of a flower peduncle 18 mm above the cutting site was acquired using a cryogenic microimaging probe at the facilities of Bruker Rheinstetten. The excised inflorescence was placed in a 5 mm NMR tube with 50 μl of a 500 mM [U-13C]glucose solution. For the cross-polarization image the 1H13C cross signal (δ 5.23/93.2) of C-1 of glucose in the sucrose and of α-glucose was selected.

13C NMR chemical shifts of glucose and fructose

13C NMR signals were assigned according to Kalinowski : δ 102.6 (C-2, β-fruf), δ 99.2 (C-2, β-frup), δ 97.0 (C-1, β-glcp), δ 93.2 (C-1, α-glcp), δ 81.7 (C-5, β-fruf), δ 77.0, 76.9 (C-3, C-5 β-glcp), δ 76.5 (C-3, β-fruf), δ 75.6 (C-4, β-fruf), δ 75.3 (C-2, β-glcp), δ 73.9 (C-3, α-glcp), δ 72.6, 72.5 (C-2, C-5, α-glcp), δ 70.8 (C-4, β-frup), δ 70.75 (C-4, α-glcp), δ 70.7 (C-4, β-glcp), δ 70.3 (C-5, β-frup), δ 68.7 (C-3 β-frup), δ 65.0 (C-1, β-frup), δ 64.4 (C-6, β-frup), δ 63.8 (C-1, β-fruf), δ 63.5 (C-6, β-fruf), δ 61.9 (C-6, β-glcp), δ 61.7 (C-6, α-glcp).

Results

Labelling experiments to observe carbohydrate conversion in vascular bundles

To gain a better understanding of the processes generating the relative proportion of the nectar carbohydrates, hydroponic feeding experiments with [1-13C]fructose and [U-13C]glucose, respectively, were carried out using excised inflorescences of A. flavidus. The nectar, collected 24 h and 48 h after the inflorescences were transferred into the feeding solutions with the labelled compounds, was dissolved in D2O and analysed without further treatment by 13C NMR unless otherwise noted. Signals of α- and β-D-glucopyranose (α-glcp, β-glcp), β-fructopyranose (β-frup), and β-fructofuranose (β-fruf) were assigned in compliance with the literature (Kalinowski ). Mixtures of known glucose-to-fructose ratios were used as references to determine the glucose-to-fructose ratio of floral nectar from untreated intact plants (Hölscher ), i.e. plants which were not subjected to feeding experiments. This method allowed errors due to different relaxation effects and NOE enhancement of different carbons to be avoided. In the [U-13C]glucose feeding experiment, the glucose-to-fructose ratio was determined based on the integrals of C-1 and C-6 signals of fructose and glucose. Overlapping multiplets of C-2 to C-5, caused by 13C-13C coupling, were not used for integration. All signals except that of C-1, from which some label may be lost due to decarboxylation in the pentose phosphate pathway, were used for integration in the [1-13C]fructose feeding experiments. In the [1-13C]fructose as well as in the [U-13C]glucose feeding experiments, a glucose-to-fructose ratio of approximately 1.2:1 was found which is identical with that observed in nectar from untreated plants. Sucrose was not detectable in the nectar using 13C NMR. In the [1-13C]fructose feeding experiment, the 13C abundance at C-1 in α- and β-glcp could be determined from the satellites of their well-resolved signals at δ 5.234 and δ 4.647 in the 1H NMR spectrum as 4.8% (label+natural abundance). The natural 13C abundance of the nectar glucose was determined as 1.2% using the nectar sampled at the start of the [1-13C]fructose feeding experiment. Analysing 13C NMR spectra gave a ratio of 13C label of C-1 to C-6 of 3.5:1 for α- and β-glcp and of 4.8:1 for β-fruf and β-frup. For positions C-2 to C-5, no significant incorporation of 13C label was observed. The fraction of 13C-labelled hexoses in the [1-13C]fructose feeding experiment is around 5%. In the hexoses obtained from the [U-13C]glucose feeding experiments, the C-1 signals of α- and β-glcp of 13C-labelled glucose appear as doublets due to 13C13C coupling (JC-1–C-2=45.6 Hz). In contrast, the C-1 signals of the unlabelled glucose appear as central singlets between the doublet lines (Fig. 1). The proportion of labelled and unlabelled glucose was determined from the integral ratio of the doublet and central singlet. Taking into account the natural abundance of 13C in the nectar hexoses of 1.2%, the fraction of 13C-labelled hexoses was determined to be approximately 8%.

Fig. 1.

Partial 13C NMR spectrum of Anigozanthos flavidus nectar obtained from the [U-13C]glucose feeding experiment, showing the C-1 signal of β-D-glucopyranose. Due to 13C13C coupling JC-1–C-2=45.6 Hz (indicated above the signal) the C-1 signal of [U-13C]glucopyranose appears as a doublet (a+a′), whereas the C-1 signal of unlabelled β-D-glucopyranose appears as a central singlet (b). Integration of these signals leads to a ratio of 13C-labelled to unlabelled hexoses of (a+a′):b = 88:12.

The finding that the carbohydrate composition, i.e. the glucose-to-fructose ratio, of Anigozanthos nectar is not affected by 13C-labelled glucose or fructose feeding raised the question, what happens during the uptake and transport of the carbohydrates in cut inflorescences? To answer this, overnight feeding experiments of excised Anigozanthos inflorescences with 5 mM [1-13C]fructose or different concentrated solutions of [U-13C]glucose (10, 100, and 200 mM), respectively, were performed. After discarding 5 mm of the bottom end of the peduncle, which was immersed in the feeding solution, segments approximately 15 mm long from different positions in the peduncle were ground in liquid nitrogen, extracted with methanol, and subjected to 13C NMR. In segments taken from directly below the flower, the only carbohydrate signals in the spectrum of 5 mM [1-13C]fructose feeding were that of the glucose (δ 93.0) and fructose (δ 62.1) moieties of sucrose (Fig. 2). The signals at δ 72.7, δ 67.6, and δ 67.0 were assigned to shikimic acid using two-dimensional 1H13C heterocorrelation NMR spectroscopy, especially the heteronuclear single quantum coherence (HSQC) experiment, which unambiguously reveals the connectivity between 13C and 1H resonances. In the spectra of 10 mM and 100 mM [U-13C]glucose feeding, only signals of sucrose (split due to 13C13C couplings between isotopically enriched carbon positions, thereby making the spectrum more complex) and of shikimic acid were present. An additional signal in the spectrum of 200 mM glucose feeding at δ 97.0 (C-1, β-glcp) and a tiny signal at δ 99.2 (C-2, β-frup) showed that free glucose and fructose were present. From these data it was concluded that for fructose (5 mM) and glucose concentrations up to 100 mM, the two hexoses are completely transformed to sucrose during uptake into the peduncle and transport to the flower.

Fig. 2.

13C NMR spectra of methanol extracts of peduncles (15 mm-long segments positioned directly below the flower) after Anigozanthos inflorescences were fed overnight with 5 mM [1-13C]fructose and 10, 100, and 200 mM [U-13C]glucose, respectively. A spectrum of sucrose is shown as a reference. In the spectra of fructose feeding and 10 mM and 100 mM glucose feeding, only signals of sucrose and shikimic acid (marked by asterisks) are present. Additional signals at δ 99.2 (very small) and δ 97.0 in the spectrum of 200 mM glucose feeding are attributable to C-2 of β-D-fructopyranose and C-1 of β-D-glucopyranose, respectively.

Extracts of peduncle segments taken from different positions below the flower were used to investigate the relationship between hexose transformation and the distance of the cutting site. As distance from the cutting site increased, the C-1 signal of β-glcp (δ 97.0) decreased in feeding experiments with 100 mM (Fig. 3) and 200 mM [U-13C]glucose. In feeding experiments with 100 mM glucose 40 mm away from the cutting site (corresponding to the top segment below the flower), no signals of β-glcp but all signals of sucrose were detectable in the 13C NMR spectrum.

Fig. 3.

13C NMR spectra of methanol extracts of different sections of Anigozanthos peduncles after inflorescences were fed overnight with 100 mM [U-13C]glucose. Before extraction with methanol, the 70 mm-long peduncle was divided into three segments (top 0–30 mm, middle 30–52 mm, and bottom >52 mm below the flower). A spectrum of unlabelled glucose is shown at the bottom. From bottom to top the signal at δ 97.0 (C-1, β-D-glucopyranose) decreases. The absence of the C-1 signal of β-D-glucopyranose in the spectra of the top section indicates that all glucose is converted to sucrose during the uptake into and the passage through the peduncle.

Magnetic resonance imaging of sugars in vascular bundles

MRI and chemical shift imaging (CSI) techniques were used to visualize vascular bundles of A. flavidus peduncles and the carbohydrates within them. In axial 1H gradient echo images of peduncles, bright round spots are clearly visible (Fig. 4A). Given the position and the shape of these spots, they probably depict vascular bundles. Light microscopic data from a cross-section of the same position of the same peduncle taken after MR imaging substantiated this interpretation (Fig. 4B). Due to magnetic susceptibility effects, the air containing parenchyma in the centre of the peduncle appears black in the gradient echo image. By applying spin echo sequences, this artefact could be avoided (Fig. 4C).

Fig. 4.

(A) 1H gradient echo image of an Anigozanthos flavidus peduncle (62×62 μm2 in-plane resolution, 1 mm slice thickness). The bright spots are the vascular bundles. (B) 1H image was overlaid with a light microscopic image from a slice of the same peduncle as in (A) prepared after 1H imaging. (C) 1H spin echo reference image of A. flavidus peduncle measured at a cryogenic probe and 400 MHz 1H frequency (32×32 μm2 in-plane resolution, 1 mm slice thickness) (D) Reference image of (C) overlaid with a cross-polarization image in false colour mode showing the distribution of the 1H13C cross-signal of C-1 of α-D-glucopyranose and the glucose moiety of sucrose (156×156 μm2 in-plane resolution, 5 mm slice thickness, acquisition time 4 h, the signal intensity increases from dark blue over light blue, green, and yellow to red).

1H13C cyclic J cross-polarization imaging (Heidenreich ) was used to study carbohydrate distribution in the peduncle. After an Anigozanthos flower was fed overnight with 500 mM [U-13C]glucose solution, the flower with a 40-mm peduncle was transferred to a 5 mm NMR tube. Since feeding flowers such a highly concentrated glucose solution led to a mixture of glucose and sucrose in the peduncle (Fig. 2), the overlapping 1H13C cross-signals of C-1 of α-glcp and the glucose moiety of sucrose were selected for imaging. These parameters made the experiment more sensitive and allowed the distribution of both sugars to be mapped in one experiment. A cross-polarization image of a 5 mm slice 18 mm above the cutting site was superimposed on a 1H spin-echo image of the same slice (Fig. 4D). The position of the sugar signals coincided with that of the vascular bundles, which together with the results from 13C NMR analysis of peduncle segments after feeding inflorescences labelled sugars, led to the conclusion that both glucose and sucrose are located in the transport system of the peduncle. No significant amounts of 13C-labelled sucrose and glucose from the feeding were stored in other peduncle parenchyma tissue.

Nectary anatomy studies using 1H magnetic resonance microimaging

The anatomy (and phylogeny) of septal nectaries of Haemodoraceae (Grassmann, 1884), including A. flavidus, was studied by Simpson (1993) using light microscopy and computer-assisted 3D reconstructions. In the present work, the anatomy of the septal nectaries and especially their connection to the vascular system non-destructively was investigated by applying 1H MRI. 1H spin-echo images were acquired at 400 MHz resonance frequency using a room temperature probe. Intact flowers were placed into a 5 mm NMR tube. 1H spin echo images of a younger (A–L) and an older flower (M–P) of A. flavidus are shown in Fig. 5. In the axial slices the main parts of the flower are clearly discernible. The lower part of the flower is surrounded by three tepals that have grown together to form a tube. Inside the tepals the vascular bundles appear as bright spots. In Fig. 5E and F the bifurcation of a vascular bundle is marked by asterisks. The lower part of the flower is divided by three septa, each containing one nectary. In the younger flower, the lumens of the nectaries appears bright, indicating that they are filled with nectar, whereas in the older flowers the lumens are dark but surrounded by a brighter layer. The nectaries open in the cavity at the style base near the apex of the ovary (Fig. 5L).

Fig. 5.

1H spin echo images of Anigozanthos flavidus flowers at two different developmental stages (slice thickness 150 μm). In the false colour images black corresponds to low signal intensities and brighter colours to higher signal intensities. (B–L) Axial slices of a young flower (30×30 μm2 in-plane resolution). In (A) the slice positions are shown. (M–P) Axial slices of an older flower (78×78 μm2 in-plane resolution). Arrow heads indicate some vascular bundles. In (E) and (F) the asterisks mark the branching of a vascular bundle. In (F) and (G) closed arrows indicate the position of septa. One septal nectary in each of the slices (G–P) is marked by an open arrow. In (G–L) the nectaries are filled with nectar, whereas in the older flower shown in slices (M–P), the nectaries have a dark centre, indicating that they are empty.

Coming from the peduncle, the vascular bundles furcate between the tepals and the middle of the flower, where they reach the vicinity of the nectaries and the base of the ovary. In the septa themselves no vascular bundles are discernible.

Discussion

Whether nectar constituents are supplied by phloem or xylem or both has long been a matter of debate. Agthe (1951) showed that the nectaries of Euphorbia pulcherrima are connected only to the phloem but not the xylem and concluded that the secreted nectar is pure phloem sap. Later studies confirmed that nectaries are supplied only by phloem (Elias ; Razem and Davis, 1999). But there are some other examples: in Ranunculus sp. the nectaries are supplied by xylem and phloem, in Fritillaria imperialis exclusively by xylem (Frey-Wyssling, 1955). Nowadays it is generally accepted that nectar is derived from the phloem solution (De la Barrera and Nobel, 2004). This implies that sucrose, the major carbohydrate in the phloem, is converted at least partially to glucose and fructose to generate the species-specific glucose:fructose:sucrose proportion of floral nectar. Lüttge (1961) assumed that in species where glucose and fructose are present in the nectar, the nectar sugar is secreted as sucrose followed by an extracellular cleavage by invertase. This should result in a glucose:fructose ratio of 1:1. On the other hand, in many nectars the glucose:fructose ratio is quite different from a balanced proportion (Baker ; Chalcoff ; Wolff, 2006). In Anigozanthos manglesii a glucose:fructose:sucrose ratio of 11:8:1 was reported (Baker ). Possible explanations for a glucose:fructose ratio that differs from 1:1 include the alteration of nectar by micro-organisms (Lüttge, 1961) or the occurrence of oligosaccharides in the nectar (Lüttge, 1962). No signals of oligosaccharides could be observed in the NMR spectra of A. flavidus nectar, but microbial effects on nectar composition of greenhouse-grown plants cannot be excluded in the present study.

De la Barrera and Nobel (2004) showed that flowers in intact plants are supplied with water via the phloem because of the higher water potential of flowers compared with the leaves. Leaves in intact plants are a source for sugars, which are transported via the phloem. In excised flowers it is not clear if carbohydrates and/or other substances are transported from the cutting site to the flower via the xylem, the phloem, or both (Zimmermann, 1953). Therefore the sugar composition of the nectar was manipulated experimentally in feeding experiments with 13C-labelled glucose and fructose. No alterations of the nectar composition were observed, however. The 13C label proved that the fed sugars are transported to the nectaries and used as a source for nectar production. Although similar amounts and concentrations of 13C-labelled sugars were supplied in parallel feeding experiments, the portion of labelled sugars in the nectar varied greatly between 1% and 8%. Differences of the inflorescences such as age, number, and size of individual flowers are the most likely explanation for this variation.

In glucose and fructose feeding experiments with excised Impatiens flowers, whose nectar only contains sucrose, Zimmermann (1953) also did not observe any changes in the nectar composition. He claimed that the glucose and fructose were included in the normal intermediary metabolism. This also seems to be the case for cut A. flavidus flowers. For sugar concentrations up to 100 mM in the feeding solution, either all the glucose or all the fructose is transformed into sucrose before reaching the flower. As demonstrated in cyclic J polarization experiments, the 13C-labelled sugars are located in the vascular bundles. Using a cryogenically cooled imaging probe, an exceptionally large signal-to-noise ratio and a high in-plane resolution of 156×156 μm was achieved. Thus, compared to measurements with 110×876 μm in-plane resolution done with a conventional room temperature probe for monitoring dynamic changes of the 13C signal intensity in intact castor bean seedlings (Heidenreich ), considerable progress was made, although the experiments were optimized for different purposes and, in the present study, experimental time was extended to 4 h compared to 1.5 h used by Heidenreich . Nevertheless, because the diameter of the vascular bundles is smaller than 300 μm, phloem and xylem cannot be distinguished. To obtain resolved images of phloem and xylem, increasing the in-plane resolution in both dimensions by a factor of at least four would be necessary; doing so would increase the measurement time by a factor of 64 (more than a week), while decreasing the signal-to-noise ratio by a factor of four. Therefore the distribution of sucrose and the two hexoses between the phloem and the xylem, the uptake mechanisms, and the site of sucrose formation remain to be investigated. Kallarackal and Komor (1989), studying Ricinus communis seedlings, showed that at low glucose concentrations, glucose is transformed to sucrose before being transported into the sieve elements, but at higher concentrations, unchanged glucose was taken up. Normally the carbohydrate concentration in xylem sap is lower than 0.5%, but in some perennial plants sugar concentrations up to 8% are observed in the early spring (Öpik and Rolfe, 2005). Hence, the possibility that carbohydrates are transported via the xylem in the artificial system used in the present study cannot be excluded. The observed decrease of glucose during the transport through the peduncle (Fig. 3) can be explained by two processes. One is the simple metabolic consumption of glucose in the phloem symplasm, which contains an operational glycolytic pathway (Geigenberger ). Another is the leaking and retrieval of sugars during phloem transport (Ayre ) associated with sucrose formation in the parenchyma of the vascular bundles. As a result, for the low sugar concentrations used in the feeding experiments for nectar sampling, all the glucose or fructose was metabolized or transformed into sucrose before reaching the nectaries. Therefore no differences in the nectar composition of untreated intact plants and the nectar in the feeding experiments using excised flowers were observed.

Studying the position of the 13C label in the [1-13C]fructose feeding experiments allowed the pathways in which the fructose is involved to be estimated before the labels appear in sucrose in the vascular bundles and in hexoses in the nectar. In the peduncle extracts from directly below the flowers, only C-1 of the glucose (δ 93.0) and fructose (δ 62.1) moieties of sucrose are enriched with 13C (Fig. 2). This suggests that the fructose is directly phosphorylated, probably after uptake in the parenchyma cells; in addition, half of the fructose-6-phosphate seems to be converted to glucose-6-phosphate and further to UDP-glucose. Both fructose-6-phosphate and UDP-glucose are used to synthesize sucrose, without entering other pathways such as glycolysis and gluconeogenesis. If [1-13C]fructose enters glycolysis, gluconeogenesis and the pentose phosphate pathway, some of the label should be transferred into the C-6 position (Fig. 6). Randomization in carbohydrate metabolism, although in the different context of starch and sucrose biosynthesis, was previously observed in Solanum tuberosum and Vicia faba (Viola ).

Fig. 6.

Hypothetical carbohydrate metabolism in floral nectaries of Anigozanthos flavidus. In [1-13C]fructose feeding experiments of inflorescences, the transfer of 13C label from position one to six of hexoses after the transport through the vascular bundles was observed. The only explanation is that the sucrose delivered from the vascular bundle is hydrolysed followed by entering of fructose 6-phosphate and glucose 6-phosphate into glycolysis to generate energy and the pentose phosphate pathway for NADPH formation. The crucial step for the transfer from 13C label of C-1 in the hexoses (highlighted by a grey circle) to C-6 (highlighted by an open circle) is the isomerization between dihydroxyacetone phosphate and glyceraldehyde 3-phosphate and the exchange of [3-13C]glyceraldehyde 3-phosphate between glycolysis and the pentose phosphate pathway. The reaction of [3-13C]glyceraldehyde 3-phosphate with sedoheptulose 7-phosphate leads to erythrose 4-phosphate and [6-13C]fructose 6-phosphate. The latter can be isomerized to [6-13C]glucose 6-phosphate and both could be used as sugars for nectar secretion. The decarboxylation in the beginning of the pentose phosphate pathway leads to a loss of 13C label. This is an elegant explanation for the lower 13C ratio of C-1 to C-6 of glucose in comparison to fructose in the [1-13C]fructose feeding experiment.

Unlike the sucrose detected in peduncles, in hexoses found in the nectar the 13C labels were indeed observed in C-1 and C-6 positions. The appearance of the label in C-6 clearly indicates that the sucrose after uptake from the neighbouring vascular bundles into the nectaries must be hydrolysed, so that the hexoses can enter into the glycolysis, gluconeogenesis, and pentose phosphate pathways (Fig. 6). The reversible isomerization of glyceraldehyde 3-phosphate to dihydroxyacetone phosphate catalysed by triose phosphate isomerase during glycolysis and gluconeogenesis explains the transfer of some of the label from position C-1 to C-6 in hexoses. Phosphofructokinase-1 and fructose 1,6-bisphophatase as key enzymes of glycolysis and gluconeogenesis are reciprocally regulated, so that only one of the two pathways is active. Therefore some of the glucose-6-phosphate probably enters the oxidative part of the pentose phosphate pathway, which produces NADPH and pentose-phosphate, whereas C-1 is removed as CO2. This elegantly explains the lower ratio of 13C-labelled C-1 to C-6 for glucose (3.5:1) compared with fructose (4.8:1) in the [1-13C]fructose feeding experiments. Some of the hexoses will enter glycolysis. During the regeneration of hexose-phosphate from pentose-phosphate, glyceraldehyde 3-phosphate from the glycolysis pathway (13C-labelled in C-3) can enter the non-oxidative part of the pentose-phosphate pathway which leads to C-6 labelled hexoses. In the present discussion, a so-called traditional model for the glycolysis and pentose phosphate pathways (Kleijn ) is used. The half reaction model (Kleijn ) explains the observed labelling pattern but does not affect the arguments presented in this work. Metabolic flux measurements (Eisenreich ; Ratcliffe and Shachar-Hill, 2006) would be useful to study the pathways involved in nectar formation quantitatively. Excised flowers of Anigozanthos are considered a good model for using this method, since in our feeding experiments the nectar formation is not disturbed.

These results indicate that, in A. flavidus, sucrose is not simply transported apoplastically from the vascular bundles into the nectar accompanied by the action of invertases and therefore is not the direct source of nectar sugars. It seems that sucrose, or glucose and fructose after cleavage through (cell-wall bound) invertase, is first taken up by the cells of the nectaries and, at least partly, cycled through the glycolysis, gluconeogenesis, and pentose phosphate pathways as shown in Fig. 6. Glucose and fructose are then most probably secreted through hexose transporters, so that the nectar's glucose-to-fructose ratio simply reflects the equilibrium of these two hexoses or their phosphates in the cytosol of the nectary cells.

Anatomy

In the 1H magnetic resonance microimages, good resolution allows different parts of the flower to be clearly distinguished. MRI does not reach the high resolution of ultrastructural electron microscopic studies (Fahn, 1979). However, it is a non-invasive method and therefore artefacts due to sample preparation can be excluded. By imaging serial 150 μm slices, we could visualize the course of the vascular bundles in the vicinity of the nectaries. The images shown in Fig. 5 support the possibility that the nectaries are directly connected to thin strands of phloem or xylem. These results strengthen the argument that the sugars used for nectar formation are delivered by the vascular bundles and have only a short distance, if any, to bridge, apoplastically or symplastically, when translocated to the nectaries. In old flowers of A. flavidus (Fig. 5M–P), the nectaries show a bright layer of tissue surrounding the lumen of the nectaries. This tissue may consist of three to five layers of differentiated cells with densely staining content as described by Simpson (1993) in his light microscopic studies of Haemodoraceae nectaries.

Conclusions

In summary, the present investigations demonstrate that magnetic resonance spectroscopy and imaging techniques are useful tools with which to study processes of nectar formation and anatomy of nectaries. The present study, has led us to establish a model in Anigozanthos of carbohydrate transport and metabolism, according to which sucrose is translocated from the vascular bundles to the nectar-secreting epithelial cells in floral nectaries. A portion of the sucrose is processed in these cells through pathways of the intermediary metabolism and hydrolysed to glucose and fructose, which are proportionally secreted to the lumen of the nectary. Peduncles and flowers of A. flavidus can be used as model systems to study nectar formation.