Introduction

Sugars regulate major metabolic and developmental pathways in plants. In recent years, it has been suggested that sugar phosphorylation by hexose phosphorylating enzymes mediate sugar-sensing processes in plants, and it has been shown that Arabidopsis AtHXK1 mediates sugar sensing in Arabidopsis and tomato (Jang et al. 1997; Dai et al. 1999). Plants have several hexose phosphorylating enzymes encoded by independent genes whose specific roles in sugar sensing remain to be elucidated. Hexose phosphorylating enzymes are characterized by their affinities to various sugars. Hexokinases (HXKs) phosphorylate glucose and fructose, but their affinity for glucose is usually two orders of magnitude higher than for fructose (Menu et al. 2001; Dai et al. 2002b; Kandel-Kfir et al. 2006). Fructokinases (FRKs) have high affinities for fructose, and so phosphorylate only fructose (Renz and Stitt 1993; Kanayama et al. 1997, 1998; German et al. 2004). HXKs and FRKs are the only glucose and fructose phosphorylating enzymes that have been identified in plants (Dai et al. 2002b). A few HXK genes have been identified in various plant species (Veramendi et al. 1999, 2002; Giese et al. 2005; Cho et al. 2006). In tomato, four HXK genes, LeHXK1–4, were identified (Menu et al. 2001; Dai et al. 2002b; Kandel-Kfir et al. 2006). Two of these, LeHXK1 and LeHXK4, are the major HXK genes expressed in most tissues, including leaves and fruits (Menu et al. 2001; Dai et al. 2002b; Kandel-Kfir et al. 2006). The only two HXK isozymes observed in leaves and fruits of tomato plants following HPLC separation, correspond to the LeHXK1 and LeHXK4 genes (Dai et al. 2002b; Kandel-Kfir et al. 2006).

The HXK genes have been divided into two major categories, type A and type B, based on their intracellular localization (Olsson et al. 2003). Type A HXKs are characterized by a chloroplast transit peptide. Type B HXKs share a common hydrophobic membrane anchor domain and are probably associated with membranes. The first type A (plastidic) HXK was identified in the moss Physcomitrella patens (Olsson et al. 2003). More recently, type A HXKs have been identified in higher plants, specifically tobacco, rice and tomato (Giese et al. 2005; Cho et al. 2006; Kandel-Kfir et al. 2006). Analyses of the N-terminal domains of the tomato LeHxk3 and LeHxk4 suggest that LeHxk3 is a type B membrane-associated enzyme while LeHxk4 is a type A enzyme localized in plastids. The intracellular location of LeHxk3 and LeHxk4 enzymes were analyzed following the expression of LeHxk3::GFP and LeHxk4::GFP fusion proteins in tobacco protoplasts (Kandel-Kfir et al. 2006). LeHxk3 appeared in small particles associated with the mitochondria, while LeHxk4 was located in plastids and stromules, tubular extensions of the plastid membrane that allow transport of proteins between plastids (Kohler and Hanson 2000).

Fructokinase genes of plants have received less scientific attention than HXK genes. Although several FRK isozymes and an increasing number of novel FRK genes have been characterized (Taylor et al. 1995; Kanayama et al. 1997; Gonzali et al. 2001; German et al. 2002, 2004; Jiang et al. 2003; Zhang et al. 2003), the number of FRK genes in various species and their specific roles are still unknown. In tomato, four FRK genes, LeFRK1–4, have been identified. Three of them, LeFRK1–3, correspond to the three isozymes found in tomato fruits (Petreikov et al. 2001; Dai et al. 2002a; German et al. 2004). The fourth one, LeFRK4, is expressed only in stamens (German et al. 2002). The intracellular locations and the specific roles of each of these FRK isozymes are still unknown.

In order to obtain a broader understanding of the roles of HXKs and FRKs, it is necessary to study each of these enzymes within a single species. Tomato is the only plant species from which four HXK genes and four FRK genes have been isolated and characterized (Kanayama et al. 1997, 1998; Menu et al. 2001; Dai et al. 2002a, b; German et al. 2002, 2003, 2004; Odanaka et al. 2002; Kandel-Kfir et al. 2006). In this study, we determined the intracellular location of each of the known tomato HXK and FRK enzymes, in order to increase our understanding of their respective functions.

Materials and methods

Construction of GFP fusion proteins

LeHXK1,2 and LeFRK1,2,3,4 open reading frames, excluding the stop codons, were amplified with added XhoI and BamHI (for LeHXK1,2 and LeFRK1,3,4) or HindIII and BamHI (for LeFRK2) restriction sites. The primers used for the PCR amplification are shown in Table 1. The resulting fragments were fused in-frame, preceding the N terminal end of GFP in pART7-GFP.

Table 1 Primers used for the PCR amplification of tomato HXK and FRK genes
Full size table

In order to obtain fusion proteins of LeHXK1,2,3 without the membrane anchor domain and of LeHXK4 and LeFRK3 without the plastidic signal peptides, the open reading frames of these enzymes were amplified excluding the first 24, 24, 25, 31 and 30 amino acids residues, respectively. The resulting fragments were fused to pART7-GFP.

Protoplast isolation and electroporation

Mesophyll protoplasts were isolated from leaves of Nicotiana tabacum L. samsun NN plants grown under sterile conditions (Draper et al. 1988). Electroporation of 5 × 105 protoplasts was carried out in pre-chilled electroporation medium (Fromm et al. 1985). The final DNA concentration was adjusted to 5 μg plasmid DNA and 15-μg calf thymus DNA per 0.5 ml of electroporation solution. Following electroporation, protoplasts were transferred into growth medium (Draper et al. 1988), incubated in the dark at 27°C for 24 h and then analyzed.

Detection of GFP and Mitotracker

All microscope observations and image acquisitions were performed using the OLYMPUS IX 81(Japan) inverted laser scanning confocal microscope (FLUOVIEW 500) equipped with a 488 nm argon ion laser, 543 nm green helium/neon laser and 60 × 1.0 NA PlanApo water immersion objective. GFP was excited by 488-nm light and the emission was collected using a BA 515–525 filter. MitoTracker (CMXROS from Molecular Probs, final concentration 400 nM) was excited by 543-nm light and the emission was collected using a BA 560IF filter. A BA 660 IF emission filter was used to observe chlorophyll autofluorescence. The images were color coded green for GFP, red for Mito Tracker and blue (or red, when MitoTracker was not used) for chlorophyll autofluorescence. Confocal optical sections were obtained at 0.5 μm increments. 3D images were obtained using the FLUOVIEW 500 software supplied with the CLSM.

Results

Prediction and analysis of the intracellular locations of tomato HXKs

The intracellular locations of the tomato HXK enzymes were predicted using TargetP (http://www.cbs.dtu.dk/services/TargetP) (Emanuelsson et al. 2000). LeHxk1 and LeHxk2 share a membrane anchor domain highly identical to that of LeHxk3, which has been previously shown to be associated with the mitochondria (Kandel-Kfir et al. 2006), and to that of spinach SoHxk1, which is anchored in the outer membrane of spinach plastids (Wiese et al. 1999) (Fig. 1a). In order to determine the cellular locations of LeHxk1 and LeHxk2, we prepared GFP fusions of these proteins and introduced them into tobacco mesophyll protoplasts. After 24 h, the transformed protoplasts were loaded with the mitochondria-selective probe MitoTracker. Fluorescence from MitoTracker and GFP was visualized by confocal microscopy. The control, GFP protein alone, was evenly dispersed in the cytoplasm while LeHxk1::GFP and LeHxk2::GFP, appeared as discrete particles of 0.5–3 μm in diameter, similar to LeHxk3::GFP. These particles were mainly associated with the mitochondria (Fig. 1b). We concluded that three of the known tomato HXK enzymes, LeHxk1,2&3, are associated with the mitochondria.

Fig. 1
figure 1

Comparison of the N-terminal region of tomato LeHxk1,2&3 and spinach SoHxk1 and intracellular location of LeHxk::GFP fusion proteins. a The comparison of the N-terminal region was performed using TargetP (http://www.cbs.dtu.dk/services/TargetP) to predict the subcellular localization of the proteins (Emanuelsson et al. 2000). TargetP scores and predictions are shown to the right with the highest score for each protein enclosed in a box. Abbreviations: cTP chloroplast transit peptide, mTP mitochondrial targeting peptide, SP secretory pathway, C predicted chloroplast import, S predicted secretory pathway. Prediction of the membrane anchored segment was done using TopPred2 (http://www.sbc.su.se/∼erikw/toppred2). b Localization of LeHxk1,2::GFP fusion proteins in mesophyll protoplasts of N. tabacum in relation to the mitochondria stained by Mito Tracker. The autofluorescence of the chloroplast is artificially stained in blue. c. Localization of LeHxk1,2&3::GFP lacking the N-terminal membrane anchor domain (LeHxk1–24::GFP, LeHxk2–24::GFP, LeHxk3–25::GFP, respectively) and of LeHxk4 enzyme without the plastidic signal peptide (LeHxk4–31::GFP). GFP is a control. The scale bar represents 5 μm

Full size image

In order to determine whether the locations of the LeHxk1,2&3 enzymes are determined by their putative N-terminal membrane anchor domains, we prepared GFP fusions of the LeHxk1,2&3 enzymes lacking the membrane anchor domain. These truncated GFP fusion proteins, named LeHxk1–24::GFP, LeHxk2–24::GFP and LeHxk3–25::GFP, were evenly dispersed in the cytoplasm following expression in tobacco mesophyll protoplasts (Fig. 1c). This confirms the role of the N-terminal domains in the determination of their intracellular location. Similarly, the GFP fusion protein LeHxk4, located in plastids (Kandel-Kfir et al. 2006), was dispersed in the cytoplasm upon deletion of the plastidic signal peptide (Fig. 1c). This confirms its role in guiding LeHxk4 to plastids.

LeFrk1, LeFrk2 and LeFrk4 are located in the cytosol while LeFrk3 is located within plastids

The intracellular location of the tomato FRK enzymes was also predicted using TargetP. LeFrk1 and LeFrk2 were predicted to be in the cytosol while LeFrk3 and LeFrk4 were predicted to be localized in plastids, based on their putative chloroplast transit peptides (Fig. 2a). In order to verify the cellular location of tomato FRKs, we prepared GFP fusions of the four FRK proteins and introduced them into tobacco mesophyll protoplasts. LeFrk1::GFP, LeFrk2::GFP and LeFrk4::GFP (despite its predicted plastidic location) were found in the cytosol (Fig. 2b), similar to the control GFP protein alone (Fig. 1c). LeFrk3::GFP was found in the chloroplasts, as predicted. It was also found inside stromules (Fig. 2c), similar to LeHxk4. This indicates that LeFrk3 is a stromal protein. Deletion of the predicted signal peptide, a 30 amino acid N-terminal sequence, from LeFrk3 led to the accumulation of this protein in the cytosol (Fig. 2b). This confirms the prediction that the 30 amino acid chloroplast transit peptide is responsible for determining the plastidic localization of LeFrk3.

Fig. 2
figure 2

Comparison of the N-terminal region of the four tomato FRKs and intracellular location of LeFrk::GFP fusion proteins. a TargetP scores and predictions are shown to the right with the highest score for each protein enclosed in a box. Abbreviations: cTP chloroplast transit peptide, mTP mitochondrial targeting peptide, SP secretory pathway, C predicted chloroplast import, S predicted secretory pathway, O other. b Localization of LeFrk1,2,3&4::GFP fusion proteins in mesophyll protoplasts of N. tabacum and localization of LeFrk3–30::GFP lacking the plastidic signal peptide. c Enlargement of LeFrk3::GFP staining in which stromules are observed. Stromules are marked by arrows. The scale bar represents 5 μm

Full size image

Discussion

At this time, tomato is the only plant species from which four HXK and four FRK genes have been cloned and characterized. Only one of the known tomato HXK isozymes, LeHxk4, is located within plastids (Kandel-Kfir et al. 2006). The other three tomato HXK enzymes, LeHxk1,2&3, are associated with the mitochondria. This association is made possible by their N-terminal membrane anchor domains. In Arabidopsis, a completely functional glycolytic ‘metabolon’, which includes HXK enzyme, is present on the outside of the mitochondrial membrane (Giege et al. 2003). It has been suggested that glycolysis in plants occurs in a highly organized multienzyme complex linked to the membrane of the mitochondria, which may allow pyruvate to be provided directly to the mitochondrion, where it is used as a respiratory substrate (Giege et al. 2003). This microcompartmentation of glycolysis is in agreement with the localization of LeHxk1,2,3::GFP in particles associated with the mitochondria. In animal systems, hexokinase is thought to interact with porins on the outer membranes of mitochondria (Wilson 1980; Nakashima et al. 1986). However, the mechanism by which glycolytic enzymes, including HXK, are associated with plant mitochondria has not yet been identified.

In addition to the stromal and to the mitochondria associated HXKs, another HXK anchored to the outer membrane of spinach plastids by an N-terminal anchor has been reported (Wiese et al. 1999). Surprisingly, the N-terminal anchor of this spinach HXK is highly identical to that of the tomato and arabidopsis membrane associated (type B) HXKs (Fig. 1a) and to other type B HXKs (Olsson et al. 2003). However, none of examined tomato HXKs is associated with the outer membrane of the chloroplast. Whether an HXK associated with the outer membrane of plastids, similar to that identified in spinach, exists in tomato plants remains to be determined.

With the exception of a single, type A, plastidic HXK previously identified in various plant species, almost all of the HXKs identified so far have been type B, membrane-associated HXKs. Two other HXKs, OsHXK7 and OsHXK8, both found in rice, lack both the plastidic transit peptide and the membrane anchor domain (Cho et al. 2006). GFP fusion experiments have shown that OsHXK7 is present in rice cytosol (Cho et al. 2006). There is also biochemical evidence to suggest the presence of a cytosolic HXK in maize (Galina et al. 1995). However, all of the HXKs, which have been identified in dicots, including all those postulated from the complete genome sequence of Arabidopsis, have either membrane anchor domains or plastidic signal peptides (Olsson et al. 2003). This suggests that dicots may lack cytosolic HXK and that, in these plants, hexose phosphorylation by HXKs outside of the chloroplast may take place mainly adjacent to membranes. It is worthwhile to note that the results of HXK extraction from tomato plants have been misleading with regards to the localization of LeHxk1 isozyme (Dai et al. 2002b). Following HXK extraction, LeHxk1 isozyme appears in the cytosolic fraction while GFP localization analyses indicate its association with the mitochondria. Since LeHxk1 is soluble in water, despite its membrane anchor domain, we assume that, in the course of protein extraction, LeHxk1 is partially dissociated from its anchoring membrane. This leads to the enzyme ending up in the cytosolic fraction, despite the fact that, in vivo, it is associated with the mitochondria.

LeFrk3 and LeHxk4 are located in plastids and stromules. The presence of both LeFrk3 and LeHxk4 in stromules suggests that these enzymes are transported between plastids and/or active within stromules. Our work indicates that LeFrk3 is probably the only gene encoding a plastidic FRK in tomato plants. The existence of FRK in plastids has been previously suggested by biochemical studies of isolated plastids from spinach, maize and Chlamydomonas (Schnarrenberger 1990; Singh et al. 1993; Wiese et al. 1999) and by proteomic studies involving Arabidopsis (Peltier et al. 2006). The presence of FRK in plastids is quite surprising since, unlike glucose, the source of fructose within plastids is not clear. Fructose could be either transported into plastids by an as yet unknown fructose (or hexose) transporter such as the recently identified plastidic glucose translocator (Butowt et al. 2003), or formed within plastids following cleavage of sucrose by either invertase or sucrose synthase. The presence of sucrose in plastids has been suggested (Gerrits et al. 2001), but no plastidic invertase or sucrose synthase have been reported. Nevertheless, the presence of FRK in plastids implies that fructose phosphorylation does occur in plastids.

Except for LeFrk3, the other three FRK enzymes, LeFrk1,2&4 are dispersed in the cytosol. The cytosolic localization of fructokinase isoforms has been previously suggested by the results of fractionation experiments (Schnarrenberger 1990). Unlike the case with LeHxk1, our GFP localization studies support these conclusions and suggest that, except for a single plastidic FRK, all tomato FRKs are present in the cytosol in vivo. The plastidic localization of LeFrk4 predicted by TargetP (http://www.cbs.dtu.dk/services/TargetP) (Fig. 2a) turned out to be incorrect, based on results of the GFP localization experiments.

While glucose can only be phosphorylated by HXKs, fructose can be phosphorylated by either HXKs or FRKs. However, the affinity of FRKs to fructose is usually one to two orders of magnitude higher than that of HXKs to fructose (Dai et al. 2002b; German et al. 2004). LeFRK1 and LeFRK2 are expressed in most tissues, including leaves and fruits. Furthermore, LeFRK2 is the major fructose phosphorylating enzyme expressed in all tissues of tomato plants (German et al. 2004). It is therefore likely that most of the fructose outside of the chloroplast is phosphorylated in the cytosol by FRKs rather than by the mitochondria-associated HXKs. Unlike fructose, the cytoplasmic glucose is probably phosphorylated mainly adjacent to the mitochondria. The biological significance of this predicted cytoplasmic spatial separation of glucose and fructose phosphorylation should be further explored.

Plant HXKs have drawn much attention due to their involvement in sugar signaling. Most studies on plant HXKs and sugar signaling have involved Arabidopsis AtHXK1 (Jang et al. 1997; Dai et al. 1999; Moore et al. 2003) and potato StHXK1 and StHXK2 (Veramendi et al. 1999, 2002). These are all type B, membrane associated HXKs (Olsson et al. 2003). Modified expression of type B HXKs (tomato LeHXK1 and LeHXK2, and the potato homologues StHXK1 and StHXK2) in tomato plants resulted in sugar signaling effects such as inhibited expression of photosynthetic genes and accelerated senescence (unpublished results). It is therefore likely that type B HXKs and, more specifically, the mitochondria associated HXKs are involved in sugar signaling. In contrast to the effects of overexpression of type B HXKs, overexpression of LeFRK1 and LeFRK2 in tomato plants yielded no detectable phenotypic effects (German et al. 2003, unpublished results), further supporting the potential connection between sugar signaling and membrane association. The question of whether the plastidic HXK and plastidic FRK are also involved in sugar sensing remains open for further exploration.