A K+ channel from salt-tolerant melon inhibited by Na+

Authors

  • Yi-Dong Zhang,

    1. Plant Science Institute, School of Agriculture and Biology, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai, 200240, People’s Republic of China
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Anne-Aliénor Véry,

    1. Biochimie et Physiologie Moléculaires des Plantes, UMR 5004 CNRS/UMR 386 INRA/SupAgro-M/UM2, Place Viala, 34060 Montpellier Cedex 2, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Li-Min Wang,

    1. Plant Science Institute, School of Agriculture and Biology, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai, 200240, People’s Republic of China
    2. Biochimie et Physiologie Moléculaires des Plantes, UMR 5004 CNRS/UMR 386 INRA/SupAgro-M/UM2, Place Viala, 34060 Montpellier Cedex 2, France
    Search for more papers by this author
  • Yang-Wu Deng,

    1. Plant Science Institute, School of Agriculture and Biology, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai, 200240, People’s Republic of China
    Search for more papers by this author
  • Hervé Sentenac,

    1. Biochimie et Physiologie Moléculaires des Plantes, UMR 5004 CNRS/UMR 386 INRA/SupAgro-M/UM2, Place Viala, 34060 Montpellier Cedex 2, France
    Search for more papers by this author
  • Dan-Feng Huang

    1. Plant Science Institute, School of Agriculture and Biology, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai, 200240, People’s Republic of China
    Search for more papers by this author

Author for correspondence:
Dan-Feng Huang
Tel: +86 21 34206943
Email: hdf@sjtu.edu.cn

Summary

  • The possible roles of K+ channels in plant adaptation to high Na+ conditions have not been extensively analyzed. Here, we characterize an inward Shaker K+ channel, MIRK (melon inward rectifying K+ channel), cloned in a salt-tolerant melon (Cucumis melo) cultivar, and show that this channel displays an unusual sensitivity to Na+.
  • MIRK expression localization was analyzed by reverse-transcription PCR (RT-PCR). MIRK functional analyses were performed in yeast (growth tests) and Xenopus oocytes (voltage-clamp). MIRK-type activity was revealed in guard cells using the patch-clamp technique.
  • MIRK is an inwardly rectifying Shaker channel belonging to the ‘KAT’ subgroup and expressed in melon leaves (especially in guard cells and vasculature), stems, flowers and fruits. Besides having similar features to its close homologs, MIRK displays a unique property: inhibition of K+ transport by external Na+. In Xenopus oocytes, external Na+ affected both inward and outward MIRK currents in a voltage-independent manner, suggesting a blocking site in the channel external mouth.
  • The degree of MIRK inhibition by Na+, which is dependent on the Na+/K+ concentration ratio, is predicted to have an impact on the control of K+ transport in planta upon salt stress. Expressed in guard cells, MIRK might control Na+ arrival to the shoots via regulation of stomatal aperture by Na+.

Introduction

Melon (Cucumis melo) is a commercially important crop, cultivated in all temperate regions of the world. The fruits are popular because their pulp is refreshing and sweet, with a pleasant aroma (Lin et al., 2004). Compared with other fruits and vegetables, it is particularly rich in minerals, especially (k) potassium and, to a lower extent, sodium (Na; Artés et al., 1993). Fruit K+ content has been reported to be positively correlated with the content of sugars and volatile acetate components (n-amyl acetate, 2-butoxyethyl acetate), which are expected to improve the fruit’s organoleptic quality (Lin et al., 2004). At the agronomic level, melon appreciates fertile soils with high potassium concentrations.

K+ is the most abundant cation in plant cells. It is involved in a number of basic functions linked together at the cellular or the whole-plant level, for example control of cell turgor, and thereby control of cell enlargement or guard cell movements. The molecular mechanisms of potassium uptake by plant roots and subsequent transport within the plant have been intensively studied during the last decade. A large number of K+ transport systems have been identified at the molecular level (Mäser et al., 2001). Among them, channels of the so-called Shaker family have been shown to be involved in wholesale and highly selective K+ uptake or secretion in several tissues and cell types. For instance, they have been shown to play important roles in, for example, K+ uptake from the soil solution, K+ secretion into the xylem sap, K+ redistribution via the phloem sap and K+ transport in guard cells during stomatal movements (Shabala, 2003; Véry & Sentenac, 2003; Gambale & Uozumi, 2006; Lebaudy et al., 2007).

A high degree of soil salinity is one of the most significant abiotic stresses for the growth and development of crops. Na+, which interferes with the transport and cytosolic functions of K+, is a major component of this stress (Qi & Spalding, 2004). The capacity to maintain a high K+ : Na+ concentration ratio in the cytosol plays an important role in salt tolerance. Interestingly, overexpression of an inward K+ channel in rice cells has been shown to result in improved salt tolerance (Obata et al., 2007). However, the possible roles of K+ channels in salt tolerance are not fully understood (Su et al., 2001; Fuchs et al., 2005; Zhang et al., 2006).

Melon, which is extensively cultivated in semi-arid regions in particular, can be exposed to salinity stress as a result of the intensive use of salt-containing irrigation water (Shannon & Francois, 1978; Mangal et al., 1988; Mendlinger & Pasternak, 1992; Rodríguez-López et al., 2000). Except when salt is introduced at precise (proper) developmental stages, a practice that can improve fruit quality (Sivritepe et al., 2003; Bustan et al., 2005), growth in high saline conditions generally results in reduced yield and fruit quality (Shannon et al., 1984; Mendlinger & Pasternak, 1992).

The mechanisms of K+ transport in melon are still poorly known. Here we report the cloning and characterization of a Shaker K+ channel, MIRK (melon inward rectifying K+ channel), in a salt-tolerant melon cultivar. MIRK is expressed in guard cells. Its activity is shown to be inhibited by external Na+. The possible role of this regulation in the melon salt tolerance is discussed.

Materials and Methods

Plant materials

Melon plants (Cucumis melo L. var. reticulatus Naud.), cv Chunli, were grown in 17 dm3 bags containing a mixture of peat, perlite and slag (1 : 1 : 1 v/v/v) in a plastic greenhouse. A half-Hoagland culture solution was applied through a fertigation system: 800 ml per bag once every 2 d before pollination, then 1000 ml per bag every day. The fruits were harvested 1 d after pollination.

Cloning of MIRK full-length cDNA

A core fragment of MIRK cDNA was amplified using two degenerate oligonucleotide primers, MF1 5′-CATGCTGAGAA(C/T)(T/C)CAAGAGA(G/A)ATG-3′ and MR1 5′-GAAACCA(G/A)(C/T)TG(A/G)AA(G/A/T)AGGAAGT-3′, designed based on sequence analysis of conserved amino acid sequences in reported plant potassium channel genes of the Shaker family (Véry & Sentenac, 2003). The forward primer (MF1) hybridized immediately downstream from the pore selectivity filter and the reverse primer (MR1) upstream from the putative cyclic nucleotide-binding domain (cNBD). Reverse-transcription PCR (RT-PCR) was used to amplify this ‘core conserved’Shaker gene fragment from total leaf melon RNA. Rapid amplification of cDNA ends (RACE) was then employed to amplify the 5′ and 3′ ends of the MIRK channel cDNA.

For 3′-RACE, total RNA (5 μg) was used to synthesize the first-strand cDNA (3′-ready cDNA) according to the manual of the SMART™ RACE cDNA Amplification Kit (Clontech Laboratories, Mountain View, CA, USA). The 3′-RACE was performed with the Universal Primer A Mix (UPM, provided by the kit) as reverse primer and MF2 (5′-GCTTCACTTCTTATCTCATCGGTAACAT-3′), designed based on conserved regions of cloned fragments, as forward primer. A nested-PCR was performed with Nested Universal Primer A (NUP, provided by the kit) as reverse primer and n-MF2 (5′-ACCAGTTACCAGAGCGAATACAAGAT-3′) as nested forward primer.

For 5′-RACE, the first-strand cDNA (5′-ready cDNA) synthesis was performed according to the manual of the SMART™ RACE cDNA Amplification Kit (Clontech Laboratories) using the 5′-RACE CDS Primer (provided by the kit). Based on the cloned core region and the sequence of the 3′-RACE product of MIRK, the reverse specific primer MR2 (5′-TGCGACTAGTCCAATGAACGAC-3′) was designed. The 5′-RACE PCR was carried out with MR2 and the forward primer UPM using the 5′-ready cDNA as template. Then the nested PCR was performed with the NUP as forward primer and n-MR2 (5′-ACGAGATTGGTTATGTTACCGATGAG-3′) as reverse primer.

After in silico assembly of the sequences of the 3′-RACE and 5′-RACE products, the full-length cDNA of MIRK was amplified by a simple PCR with the 5′-ready cDNA as template using specific primers MF3 (5′-ATGCGAAGCTCTTGTTGTACAAAGC-3′) and MR3 (5′-GCACTCCTAAAAATTCCATAAATC-3′).

Phylogenetic analyses

Similarity-based analyses were carried out using BLAST on NCBI (http://www.ncbi.nlm.nih.gov/BLAST/). Multiple alignments were achieved by ClustalW (http://www.ebi.ac.uk/clustalw/) and the Genedoc package (Thompson et al., 1994). The phylogenetic tree was constructed using the full-length amino acid sequences of plant inwardly rectifying potassium channels with MEGA 2 (Kumar et al., 2001) using the neighbor-joining method. The bootstrap value was generated using 1000 repeats.

Analysis of MIRK expression pattern

Total RNA (2 μg per sample) from leaves, stems, roots, flowers and fruits of melon plants were extracted using Trizol Reagent (Invitrogen) according to the manufacturer’s instructions. For the analysis of MIRK expression in the leaf, total RNA was extracted from main veins, epidermis and guard cell-enriched epidermis using the RNeasy Mini Kit (Qiagen, Hilden, Germany). Epidermal fragments were obtained by blending in cold tap water 20 young melon leaves of which the main veins had been removed. For the guard cell-enriched preparation, surface hairs were previously abraded with sandpaper. Rinsed epidermal fragments were treated for 30 min in the dark in an enzyme solution at low osmolarity adapted from Fairley-Grenot & Assmann (1992) (0.7% (w/v) Cellulysin (Calbiochem, Alexandria, New South Wales, Australia), 0.1% (w/v) PVP-40 and 0.25% (w/v) BSA in 275 μM CaCl2, 275 μM MgCl2, 275 μM ascorbic acid, 5.5 μM KH2PO4, 2.75 mM Mes-Tris, pH 5.5, and 308 mM sorbitol) in order to remove attached mesophyll cells and break epidermal cells.

Semiquantitative RT-PCR was used for the analysis of MIRK expression according to the manufacturer’s instructions for the one-step RNA PCR kit (TaKaRa, Dalian, China). The MIRK-specific primer pairs MF4 (5′-ATGCGAAGCTCTTGTTGTACAAAGC-3′) and MR4 (5′-TGCGACTAGTCCAATGAACGAC-3′) or MF5 (5′-aacctgggcttcacttcttatctcatc-3′) and MR5 (5′-atattcttttcacccccttcaacctct-3′) were used. Actin was used as control in the semiquantitative RT-PCR. The primers for amplification of the actin cDNA, act-F (5′-GTGACAATGGAACTGGAATGG-3′) and act-R (5′-AGACGGAGGATAGCGTGAGG-3′), were designed according to the conserved sequences of the actin genes from different plant species. We estimated the relative amounts of MIRK and actin cDNA in the different organs by determining the ratio of the amount of amplified fragments for MIRK and actin for each PCR reaction at each cycle from the 15th to 25th cycles. Analyses indicated that the amplification was linear at least from cycle 15 to 25 (for both MIRK and actin fragments).

Expression in yeast

The Arabidopsis Shaker K+ channel KAT1 was used as a positive control. MIRK and KAT1 cDNA were subcloned into the yeast high copy expression plasmid pFL61 (Minet et al., 1992). MIRK cDNA, KAT1 cDNA and the empty plasmid (negative control) were transferred into the K+-uptake defective yeast strain WΔ6 (Mat a ade2 ura3 trp1 trk1Δ::LEU2 trk2Δ::HIS3; Haro & Rodríguez-Navarro, 2003). Transformed yeast colonies were screened on synthetic defined (SD) medium (1.7 g yeast nitrogen base, 5 g (NH4)2SO4, 40% glucose, 20 g bactoagar) supplemented with 20 mg l−1 adenine, 30 mg l−1 tryptophan and 50 mM KCl. Actual yeast transformation of selected colonies was checked by PCR. Growth tests were performed in liquid arginine phosphate (AP) medium (Rodríguez-Navarro & Ramos, 1984) supplemented with 20 mg l−1 adenin, 30 mg l−1 tryptophan and the indicated concentrations of KCl and NaCl.

Expression in Xenopus oocytes and voltage-clamp experiments

MIRK cDNA was subcloned into the pGEMXho vector (adapted from pGEMDG vector, D. Becker, Würzburg, Germany) between the BamH1 and Not1 sites, under the control of the T7 promoter between the 5′ and 3′ untranslated regions of the Xenopus beta-globin gene. Capped and polyadenylated cRNA were synthesized in vitro using the mMESSAGE mMACHINE kit (Ambion, Texas, USA). Oocytes were isolated as previously described (Véry et al., 1995), injected with 50 ng of MIRK cRNA or 50 nl of deionized water (control oocytes), then kept at 18°C in ‘ND96’ solution (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, 2.5 mM Na-pyruvate and 5 mM Hepes-NaOH (pH 7.5)), supplemented with 50 mg ml−1 gentamycin, until electrophysiological recordings.

Whole-oocyte currents were recorded using the two-electrode voltage-clamp technique, 3–5 d after cRNA injection. The voltage-clamp amplifier was an Axoclamp 2A (Axon Instruments, Foster City, CA, USA). Voltage-pulse protocols, data acquisition and data analyses were performed using pClamp9 (Axon Instruments, Foster City, CA, USA) and Sigmaplot8 software (Jandel Scientific, Erkrath, Germany). Both actual imposed membrane potential and current were recorded. Correction was made for voltage drop through the series resistance of the bath and the reference electrode by using a voltage-recording microelectrode in the bath close to the oocyte surface (potential difference between the local electrode and the reference subtracted from the potential of intracellular electrodes in the amplifier for real-time series resistance correction). Electrodes were filled with 1 M KCl. The solution bathing the oocyte was continuously perfused during the voltage-clamp experiment. All bath solutions contained a background of 1 mM CaCl2, 2 mM MgCl2, and either 10 mM Hepes-Tris when the pH was 6.0 or 7.5, or 10 mM Mes-Tris when the pH was 5. Except when the effect of external pH on MIRK currents was analyzed, the pH of the solution was 6.0. Monovalent cations were added as chloride salts. The ionic strength was kept constant in all solutions. When the concentration of KCl ([KCl]) was < 100 mM, LiCl or NaCl or a mixture of both was added such that [KCl] + [LiCl] + [NaCl] = 100 mM. When the salt used for ionic strength adjustment was not specified, LiCl was used. Current traces shown in figures are whole oocyte currents. MIRK currents presented in IV relationships were extracted from whole oocyte currents doing a leak subtraction (leak resistance > 1 MΩ).

Patch-clamp on melon guard cell protoplasts

Guard cell protoplasts were isolated from abaxial epidermal strips obtained from young leaves of melon plants (cv Chunli) grown in compost in a glasshouse, using the enzymatic digestion protocol described by Véry et al. (1998). Patch-clamp bath solutions contained 1 mM CaCl2, 4 mM MgCl2, 10 mM Mes-Tris (pH 5.8), and K-glutamate and Na-glutamate as indicated. The pipette solution contained 100 mM K-glutamate, 5 mM EGTA, 1 mM CaCl2 (45 nM free Ca2+), 0.5 mM MgCl2, 2 mM MgATP, 20 mM Hepes and 20 mM KOH (pH 7.2). The osmolarity of bath and pipette solutions was adjusted with d-mannitol to 480 and 520 mOsM, respectively. Patch-clamp recordings on guard cell protoplasts were carried out as described in Lebaudy et al. (2008). Liquid junction potentials at the pipette–bath interface were measured and corrected.

Results

Molecular cloning of the full-length MIRK cDNA

Homologs of the Arabidopsis KAT1 or KAT2 Shaker channel genes (Anderson et al., 1992; Pilot et al., 2001) were sought in melon. Performing RT-PCR experiments on leaf melon RNA using degenerated primers, a c. 400 bp core fragment was amplified, which displayed similarity with other plant Shaker potassium channel genes. A 1055 bp fragment of 5′ cDNA and a 1355 bp fragment of 3′ cDNA were obtained (RACE procedure) and used to design primers to amplify the full-length cDNA (2506 bp; Genbank accession number DQ116940), which contained an open reading frame (ORF) of 2103 bp encoding a 701 amino acid (aa) polypeptide, a 3′-untranslated region of 265 bp and a 5′-untranslated region of 139 bp. The cloned channel was called MIRK for melon inwardly rectifying K+ channel.

The MIRK protein molecular mass is predicted to be 80.7 kDa. MIRK polypeptide shares 73% identity with the grapevine group 2 Shaker channel SIRK (AAL09479), 62 and 60% identity with the Arabidopsis group 2 Shaker channels KAT1 (NP_199436) and KAT2 (NP_193563), respectively, 57% with the potato group 2 channel KST1 (CAA56175), and 55% with the Arabidopsis group 1 channel AKT1 (NP_180233) (Shaker channel groups; see Pilot et al., 2003b). TopPred analysis (Claros & von Heijne, 1994) provided a predicted structure consistent with the current structural model of plant and animal Shaker channels (Véry & Sentenac, 2003; Long et al., 2005): six transmembrane segments (TMS), named S1 to S6, a pore-forming domain between S5 and S6 and a long hydrophilic C-terminus (Fig. 1a). S4, the voltage sensor, displays positively charged residues (Fig. 1b). The pore-forming domain (P-domain) contains the hallmark TxxTxGYGD motif of K+-selective channels (Anderson et al., 1995; Nakamura et al., 1997) (Fig. 1b). The MIRK C-terminal hydrophilic region contains, like that of other plant Shaker channels, a putative cyclic nucleotide-binding domain (cNBD) and, at its very end, a region rich in acidic and hydrophobic residues (Kha domain) which may be involved in channel tetramerization or clustering in the membrane (Véry & Sentenac, 2003). No ankyrin domain was present in the MIRK C-terminal region, as in most other group 2 Shaker channels. Thus, the deduced MIRK polypeptide displays the structural domains typical of plant (group 2) Shaker channels, suggesting that MIRK encodes a functional potassium channel subunit in melon.

Figure 1.

 Identification of MIRK (melon inward rectifying K+ channel), a melon Shaker channel. (a) Topology of the MIRK polypeptide. Hydropathy analyses indicate that MIRK hydrophobic regions comprise six putative membrane-spanning domains, S1–S6. A pore-forming domain, P, is located between S5 and S6. S4, the voltage sensor, contains positively charged residues (R and K). The hydrophilic C-terminal region contains a putative cyclic nucleotide-binding domain (cNBD) and a domain rich in acidic and hydrophobic residues (KHA). (b) Alignment of MIRK, VvSIRK and KAT2 sequences in the voltage sensor (S4) and pore (P) regions. (c) Phylogenetic tree performed with the whole protein sequences of MIRK and other Shaker potassium channels. Protein accession numbers are SKT1, CAA60016; LKT1, CAA65254; AKT1, NP_180233; ZMK1, NP_001105480; TaAKT1, AAF36832; MKT1, AAF81249; SPIK, NP_180131; AKT6, NP_194976; MIRK, AAZ66349; SIRK, AAL09479; KST1, CAA56175; KAT1, NP_199436; KAT2, NP_193563; KCT2, AAX19659; ZMK2, NP_001105120; NpKT1, BAA84085; SKT2, CAA70870; PTK2, CAC05489; AKT2, NP_567651; SPICK2, AAD39492; VfK1, CAA71598; SPICK1, AAD16278; AtKC1, NP_194991; KDC1, CAB62555.

MIRK phylogenetic relationships with other plant Shaker channels

A phylogenetic tree was constructed using the entire amino acid sequences of MIRK and of other plant Shaker channel subunits (including all inward subunits from Arabidopsis thaliana) belonging to the AKT1-like, KAT-like, AKT2-like and AtKC1-like subfamilies (Fig. 1c). The results indicated that MIRK belongs to the Shaker channel group 2 (KAT-like subfamily) (Fig. 1c). Among identified Shaker channel subunits, SIRK from vine (Pratelli et al., 2002) was MIRK’s closest relative.

MIRK expression pattern

MIRK expression could be detected (RT-PCR experiments) in leaves, stems, flowers, and fruits (1 d after pollination), but not in roots (Fig. 2a), a feature shared with other characterized Shaker genes of the KAT subfamily (Nakamura et al., 1995; Müller-Röber et al., 1995; Pilot et al., 2001; Pratelli et al., 2002; Su et al., 2005). MIRK transcripts were the most abundant in leaves, flowers, and fruits.

Figure 2.

 Expression of MIRK (melon inward rectifying K+ channel) in melon plants. (a) Top panel: relative abundance of MIRK transcript in different organs, analyzed by semiquantitative reverse transcription-PCR (RT-PCR) with actin transcript abundance (lower panel) used as control. R, root; L, leaf; S, stem; Fl, flower; Fr, 1-d-old fruit. Results are the average of three independent replicates. Error bars represent the standard deviation. (b) Expression of MIRK in the leaf. Top panel: ‘non treated’ epidermis fragment (e; left) and enzyme-treated epidermis for enrichment in guard cells (gc; right) used for MIRK expression analysis. Lower panel: comparison of MIRK transcript abundance in different leaf tissues, analyzed by RT-PCR. e, epidermis; gc, guard cell-enriched epidermis; gDNA, genomic DNA.

A second set of experiments was aimed at investigating MIRK expression in different leaf tissues. Total RNA was extracted either from intact whole leaves or from tissue preparations enriched in leaf veins, leaf epidermis or guard cells (Fig. 2b, upper panels). RT-PCR analyses (Fig. 2b, lower panels) revealed that MIRK is preferentially expressed in guard cells and in veins.

Electrophysiological analyses of MIRK in Xenopus oocytes

Expressed in Xenopus oocytes, the MIRK channel displayed inward rectification with a current activation threshold of c.−110 mV when oocytes were bathed in K+-containing solutions at pH 6 (Fig. 3a,b). The magnitude of MIRK currents was strongly dependent on the external concentration of K+ in the range 0.1–100 mM, confirming that MIRK is a K+-permeable channel (Fig. 3b).

Figure 3.

 Functional properties of MIRK (melon inward rectifying K+ channel) in Xenopus oocytes. (a) Example of current traces recorded in an oocyte expressing MIRK. External K+ concentration ([K+]) was 10 mM. The holding potential was −20 mV. The membrane potential was clamped from 0 to −150 mV in steps of −10 mV. (b) Effect of external K+ concentration on MIRK currents. External [K+] was successively 100 mM (K100), 10 mM (K10), 1 mM (K1) and 0.1 mM (K0.1). In solutions where [KCl] was < 100 mM, LiCl was added such that [KCl] + [LiCl] was 100 mM. (c) MIRK channel apparent affinity for K+. MIRK macroscopic inward conductance was estimated at infinitely negative voltages performing fits of the MIRK current–voltage relationships with a Goldman–Hodgkin–Katz equation coupled to a Boltzmann equation (solid lines in b) (Véry et al., 1995): I = (GE(1 – exp(F(EErev)/R/T))/(1 – exp(FE/R/T))) × (1/(1 + exp(zgF(E– Ea50)/R/T))). G (inward conductance at infinitely negative voltages, i.e. with all MIRK channels of the oocyte membrane able to open in an open state), zg (equivalent gating charge) and Ea50 (half-activation potential) are adjustable parameters. Erev is the current reversal potential through MIRK, Faraday Constant (F), Gas Constant (R), Temperature (kelvin) (T) have their usual meanings. In each oocyte, the resulting G values, obtained for external concentration of K+ ranging from 0.1 to 100 mM, were normalized by the G value obtained in 10 mM K+ to suppress variability caused by differences in oocyte levels of expression. Normalized G values were then plotted against the concentration of K+, which resulted in a hyperbolic shape indicative of a saturation phenomenon. The Michaelis–Menten formalism was used to fit normalized G values. A mean KM value (concentration resulting in half saturation) ± SE was derived performing such hyperbolic fits on conductance values obtained in each oocyte. The solid line shows a fit on mean normalized G values. (d) MIRK selectivity among monovalent cations. Currents were recorded in bath solutions containing successively 100 mM of either KCl (K100), RbCl (Rb100), NaCl (Na100), LiCl (Li100), CsCl (Cs100) or NH4Cl (NH4100). (e) MIRK high selectivity for K+ over Na+ and Li+ determined by reversal potential analysis. The bath solutions contained either a mixture of Na+ and K+ or of Li+ and K+, the concentration sums [Na+] + [K+] and [Li+] + [K+] being 100 mM. (f) Voltage dependence of MIRK channel activation. Po : Pomax values were obtained from deactivation (tail) current analysis and plotted against the membrane potential. The data were fitted with a Boltzmann equation to extract the gating parameters zg (equivalent gating charge) and Ea50 (half-activation potential). Data are from a representative experiment obtained on a single oocyte. External [K+] was successively 100, 10 , 1  and 0.1 mM. Mean zg and Ea50 gating parameters obtained from six oocytes, are, respectively 2.9 ± 0.06 and −158 ± 3 mV. (g) Effect of external pH. External [K+] was 10 mM. Half-activation potentials (Ea50 values) at the different pH values were obtained by fitting MIRK current–voltage relationships with a Goldman–Hodgkin–Katz equation coupled to a Boltzmann equation (cf. above; solid lines). (h) Blockage by external Cs+. External [Cs+] was 0, 1 or 5 mM, in the presence of 10 mM [K+]. Data in (b)–(e), (g) and (h) are shown as means ± SE; n = 6 in (c) and (d), n = 5 in (b) and (e), and n = 4 in (g) and (h). Vm, imposed membrane voltage of activation step.

Increasing external K+ concentration, a saturation of MIRK conductance was observed (Fig. 3c). Half saturation occurred with an external K+ concentration close to 10 mM, indicating that MIRK’s apparent affinity for K+ was in the range of that reported in other plant inward Shaker channels (Véry et al., 1995; Mouline et al., 2002). The cationic selectivity of the MIRK channel was addressed by replacing K+ by other monovalent cations in the bath solution (Fig. 3d). This experiment revealed that MIRK is also permeable, however to a lower extent, to Rb+ and NH4+ (Fig. 3d), but is not significantly permeable to Li+, Na+ and Cs+. Permeability ratios PRb : PK, PNH4 : PK and PLi : PK, deduced from current reversal potential analysis in solutions containing successively the different cations, were 0.37 ± 0.02, 0.05 ± 0.002 and 0.0015 ± 0.0001 (n = 5) (not shown), respectively. Chord conductance ratios GRb : GK and GNH4 : GK at −140 mV were 0.16 ± 0.03 and 0.08 ± 0.03, respectively (n = 6; Fig 3d). Further analyses of MIRK cation selectivity in solutions containing K+ and either Na+ or Li+, confirmed that among the latter three cations, only K+ is transported by MIRK, since reversal potential of current in these conditions followed predictions for a purely K+-selective channel (Fig. 3e).

Analysis of MIRK sensitivity to voltage revealed that the gating of MIRK, like that of other Shaker channels, was voltage-dependent. The voltage dependence of the relative open probability (Po : Pomax) was independent of the external K+ concentration in the range 0.1–100 mM (Fig. 3f), a feature classically reported in inwardly rectifying plant Shaker channels (Véry & Sentenac, 2003). The equivalent gating charge, close to 3 (2.9 ± 0.06; n = 6), was similar to that reported in the MIRK homologs KAT2 and VvSIRK (Pilot et al., 2001; Pratelli et al., 2002). A rather negative half-activation potential (−158 ± 3 mV; n = 6) (Fig. 3f) was also a feature shared with KAT2 and VvSIRK. MIRK was activated by acidic external pH, acidification in the pH range 7.5 to 5 positively shifting the voltage threshold of activation by c. 10 mV per pH unit (Fig. 3g). This pH sensitivity was also reminiscent of that reported in KAT2 and VvSIRK (Pilot et al., 2001; Pratelli et al., 2002). Finally, also like KAT2 and VvSIRK, MIRK displayed a nonvoltage-dependent inhibition of K+ currents by external Cs+ (Fig. 3h).

MIRK is inhibited by external Na+ in Xenopus oocytes

Surprisingly, MIRK inward currents recorded in a solution containing 1 mM K+ and 99 mM Na+ were 50% smaller than those recorded on the same oocytes in a solution containing the same concentration of K+ but 99 mM Li+ instead of Na+ (Fig. 4a). As both Na+ and Li+ are negligibly permeant through MIRK (Fig. 3d,e), the smaller current intensity in the Na+-containing solution suggested that external Na+ exerted an inhibitory effect on MIRK channel activity. Outward deactivation currents (transient currents observed upon voltage clamp back to −60 mV after a channel activation step at a potential more negative than the channel activation threshold) were also smaller (by c. 60%) in the Na+-containing solution (Fig. 4a). Decreasing the external K+ concentration, the inhibitory effect of external Na+ increased. For instance, in the presence of 0.1 mM K+, substituting Na+ for Li+ at the same concentration (99.9 mM) in the external medium resulted in an almost complete inhibition (by 85–100%) of MIRK currents (Fig. 4b,c,d), whereas at a higher K+ concentration (10 mM), the current reduction was only 5–20% when Li+ was replaced by Na+ (at the same concentration, 90 mM) (Fig. 4b,d). Hence, the inhibitory effect of Na+ seemed to be dependent on the Na+ : K+ external concentration ratio. Further analysis of the Na+ effect on MIRK channel activity revealed that the reduction of inward currents in the presence of external Na+ was not the result of a modification of MIRK gating properties, since the equivalent gating charge and the half-activation potential determined in Na+-containing solutions (when inhibition was not complete) and in Li+-containing solutions were very similar (Fig. 4b). Furthermore, external Na+-induced inhibition of MIRK currents was not voltage-dependent (Fig. 4b,c,d), suggesting that Na+ did not inhibit the currents by plugging the channel in the narrow region of the selectivity filter deep inside the channel pore, but more probably by binding to one or more sites at the external mouth of the channel, outside the transmembrane electric field. Finally, the inhibitory effect of external Na+ ions was observed on both inward currents and outward tail currents (Fig. 4d). This suggested that the binding of Na+ rendered the channel nonconductive.

Figure 4.

 MIRK (melon inward rectifying K+ channel) is inhibited by Na+. (a) Current traces recorded in 1 mM external K+ (K1) and 99 mM of either Li+ (Li99) or Na+ (Na99). The membrane potential was clamped from 0 to −160 mV in steps of −10 mV, and the holding potential was −60 mV. (b) Comparison of MIRK currents in the presence of Li+ or Na+ at different external K+ concentrations (0.1 (K0.1), 1 (K1) and 10 mM (K10)). The Li+ or Na+ concentration was chosen such that the concentration sum [K+] + [Li+] or [K+] + [Na+] was 100 mM. Data are means ± SE (n = 5). Current–voltage relationships were fitted with a Goldman–Hodgkin–Katz equation coupled to a Boltzmann equation (cf. Fig. 3b) (solid lines). Gating parameters (equivalent gating charge zg, half-activation potential Ea50) obtained in solutions containing Li+ or Na+ are given in the key. (c) MIRK outward current inhibition by Na+. Outward currents recorded upon return to the holding potential are plotted against the voltage of the preceding activation pulse (Vm). Outward currents were normalized in each oocyte by the current recorded after an activation pulse at −150 mV in 1 mM K+ + 99 mM Li+ to suppress variability resulting from the difference in expression level among injected oocytes. Data are means ± SE (n = 3). (d) Comparison of the effect of Na+ on inward (in) and outward (out) currents for different external [K+]. Data were extracted from (b) and (c). The analysis was performed at three activation potentials, −135, −145 and −155 mV. Vm, imposed membrane voltage of activation step.

A dose–response effect of external Na+ on MIRK currents was obtained at 100 μM external K+ (Fig. 5). Outward currents alone were analyzed, as inward currents in this external K+ condition occurred at very negative membrane potentials (more negative than −160 mV) where activation of endogenous chloride currents rendered analysis of MIRK currents difficult. Both steady-state outward currents recorded at membrane potential between the activation potential (c.−110 mV) and the current reversal potential (−160 mV), and deactivation currents recorded at −60 mV after an activation step were analyzed. Very similar inhibitions were observed in both cases (not shown), confirming that the effect of external Na+ on MIRK activity was not voltage-dependent. Fig. 5 (b,c) shows the analysis of MIRK deactivation currents. Increasing the external concentration of Na+ from 1 to 100 mM resulted in an inhibition of MIRK currents ranging from 20% to almost 100%. Fifty per cent inhibition occurred at 5 mM Na+, and 60% inhibition at 10 mM Na+ (Fig. 5b,c). These results further supported the hypothesis that MIRK inhibition by Na+ is dependent on the Na+ : K+ concentration ratio in the external medium, since the degree of current inhibition was the same with 10 mM K+ and 99 mM Na+ (Fig. 4) as with 0.1 mM K+ and 10 mM Na+. Moreover, these results indicated that the inhibition of MIRK channel activity by Na+ can occur in physiological conditions (see the Discussion section).

Figure 5.

 MIRK (melon inward rectifying K+ channel) inhibition by Na+ at physiological external [K+]. (a) Current traces recorded in 0.1 mM external K+ (K0.1) and either 0 mM Na+ (left) or 10 mM Na+ (right, Na10). The membrane was clamped from 0 to −150 mV in steps of −10 mV. The holding potential was −60 mV. (b) Effect of Na+ on MIRK outward deactivation currents (tail currents). Deactivation currents recorded at −60 mV are plotted against the potential of the preceding activation step (Vm). External K+ concentration was 0.1 mM, and Na+ concentrations were 0 (Na0), 1 (Na1), 10 (Na10)and 100 mM (Na100). The Li+ concentration was chosen such that the concentration sum [Na+] + [Li+] was 100 mM. Deactivation currents were normalized in each oocyte by the current recorded after an activation pulse at −160 mV in 0.1 mM K+ + 1 mM Na+ to suppress variability resulting from the difference in expression level. Data are means ± SE (n = 3). Solid lines are Boltzmann fits. (c) Degree of inhibition of MIRK currents by Na+ in the presence of 0.1 mM K+. zg, equivalent gating charge; Ea50, half-activation potential. Vm, imposed membrane voltage of activation step.

Functional expression in yeast confirms MIRK sensitivity to Na+

MIRK sensitivity to Na+ was further assessed by functional expression in a Saccharomyces cerevisiae mutant strain (WΔ6) defective for K+ uptake as a result of disruption of the K+ transporter genes Trk1 and Trk2 (Rodríguez-Navarro & Ramos, 1984; Haro & Rodríguez-Navarro, 2003). Such double mutant cells are unable to grow in low K+ media and require K+ concentrations higher than 10 mM to display normal growth, when compared with wild-type cells. Mutant cells were transformed with constructs (in pFL61 vector; Minet et al., 1992) allowing expression of either MIRK or the Arabidopsis KAT1 Shaker K+ channel, which is not sensitive to external Na+ (Su et al., 2005). Negative control cells were transformed with the empty plasmid. Transformed cells were grown in liquid AP medium (Rodríguez-Navarro & Ramos, 1984) containing either 50 mM or 1 mM K+. In the latter case, 100 mM Na+ was added or not into the medium. Cell culture growth was followed for 10 h. The data, shown in Fig. 6, can be summarized as follows. In the presence of 50 mM K+, cells transformed with the empty plasmid displayed a growth rate similar to that of cells expressing MIRK or KAT1, in agreement with the fact that high K+ media restore normal growth in trk1 trk2 mutant cells. In the presence of 1 mM K+ and the absence of added Na+, expression of MIRK, as with that of KAT1, strongly increased the cell culture growth rate, when compared with transformation with the empty plasmid. Such a positive effect on growth rate is classically taken as evidence that the heterologously expressed plant channel is functional at the yeast cell membrane and restores K+ uptake (Sentenac et al., 1992). The positive effect of MIRK expression on cell growth in the presence of 1 mM K+ was strongly inhibited by the addition of Na+, while that of KAT1 expression was not inhibited. Thus, taken as a whole, the data provided physiological evidence, in growing cells, both that MIRK can mediate K+ uptake and that this capacity is reduced by Na+.

Figure 6.

 MIRK (melon inward rectifying K+ channel) functional expression in yeast. Saccharomyces cerevisiae cells from the K+ uptake defective strain WΔ6 transformed with either pFL61 empty plasmid or the Arabidopsis KAT1 Shaker cDNA or MIRK cDNA were grown in arginine phosphate medium containing either 50 mM KCl (K50; insert) or 1 mM KCl (K1, main panel) or both 1 mM KCl and 100 mM NaCl (K1Na, main panel). Data are mean optical density values at 600 nm (OD 600 nm) ± SE obtained in independent experiments (n = 3).

Evidence for MIRK-like K+ channel activity in melon guard cells

Since RT-PCR experiments provided evidence that MIRK is expressed in guard cells, patch-clamp experiments were performed on guard cell protoplasts of melon cv Chunli in order to assess the contribution of MIRK-like channels to K+ uptake in this cell type (Fig. 7). Shaker-like inwardly-rectifying time-dependent K+ currents were typically recorded upon protoplast membrane hyperpolarization above a threshold of c.−150 mV in the presence of 1 or 10 mM K+ (Fig. 7a). When K+ was decreased to 0.1 mM, inwardly rectifying K+ currents became hardly detectable, but the outward exponentially decreasing deactivation currents recorded at −100 mV after the activation steps at more negative voltages could attest activity of Shaker-type inwardly rectifying channels (Fig. 7a,b). At low external K+ concentrations (1 or 0.1 mM), the effect of addition of 100 mM Na+ on these deactivation currents was analyzed, as in oocyte experiments (Figs 4, 5). Na+ was found to reduce the deactivation currents (Fig. 7b,c), the reduction being c. 20% when external [K+] was 1 mM (22 ± 24%, n = 3; not shown) and c. 40% (41 ± 10%, n = 3) when external [K+] was 0.1 mM (Fig. 7b,c). Thus, the Shaker-like inward K+ conductance in melon guard cells, like MIRK conductance in oocytes, is inhibited by external Na+. The sensitivity to Na+ is, however, lower in guard cells than in MIRK-expressing oocytes (Figs 4c, 7c). This suggests that MIRK-like channels coexist in guard cells with other Shaker channels insensitive to Na+. Since MIRK is expressed in melon guard cells (Fig. 2b), it is likely that the MIRK-like activity in guard cells is mediated by MIRK itself.

Figure 7.

 Inhibition by Na+ of MIRK (melon inward rectifying K+ channel)-like K+ channel activity in melon guard cell protoplasts. (a) Example of current traces recorded in guard cell protoplasts of melon in the presence of different external K+ and Na+ concentrations: 10 mM K+ (K10) in the absence of Na+ (top panel), 0.1 mM K+ (K0.1) in the absence of Na+ (middle) or 0.1 mM K+ + 100 mM Na+ (bottom, K0.1 Na 100). The voltage-clamp protocol consisted of a channel activation step, with applied voltages ranging from −60 to −240 mV, followed by a deactivation step at −100 mV. (b) Zoom on deactivation currents (Itail) recorded at −100 mV after the activation step in the presence of 0.1 mM K+ (K0.1) with either no Na+ (–Na) or 100 mM Na+ (+Na). (c) Deactivation (tail) currents plotted vs the activation potential (Vm). Solid lines are Boltzmann fits. Data in (a–c) are representative of three independent experiments. zg, equivalent gating charge; Ea50, half-activation potential. Vm, imposed membrane voltage of activation step.

Discussion

Identification of a K+ Shaker channel in a salt-tolerant melon species

Since the first cloning of KAT1 and AKT1 from Arabidopsis (Anderson et al., 1992; Sentenac et al., 1992), a number of Shaker-like potassium channel genes have been identified in different plant species. Nine members of this family are present in Arabidopsis and > 30 have been identified in other plant species (Shabala, 2003; Véry & Sentenac, 2003). The Cucurbitaceae family includes > 700 species worldwide and many of them are very important agronomically. To our knowledge, MIRK is the first potassium channel gene from this family cloned in the genera Cucumis.

Detailed characterization of MIRK expressed in Xenopus oocytes revealed that MIRK shares a number of functional properties with its homologs of the Shaker channel group 2 (‘KAT’ subfamily) (Fig. 3). Based on protein sequence comparison, MIRK displays the highest percentage of identity with SIRK from vine (73%). At the functional level, MIRK’s closest homologs are VvSIRK and KAT2 from Arabidopsis, which share with MIRK similar cation selectivity, voltage gating parameters and sensitivity to external pH and Cs+ (Pilot et al., 2001; Pratelli et al., 2002). KST1 from potato and KAT1, also in the ‘KAT’ subfamily, are functionally more distant: KST1 differed from MIRK in the sensitivity to Cs+ (Müller-Röber et al., 1995), and KAT1 in gating parameters, sensitivity to pH and sensitivity to Cs+ (Véry et al., 1995; Pilot et al., 2001). Interestingly, MIRK displays an original functional property shared with none of the plant Shaker channels so far characterized, which is an inhibition of activity by external Na+ (see the following section).

MIRK activity is inhibited by external Na+, a unique feature among plant Shaker channels

Salt stress is a major abiotic stress in many plant species. Selective K+ transport, control of K+ and Na+ concentrations in the cytosol, and regulated Na+ accumulation in shoots are of crucial importance in plant adaptation to soils displaying high Na+ contents (Yeo, 1998; Maathuis & Amtmann, 1999). Regulation by salt of transporters or channels mediating K+ transport has been reported (Su et al., 2001; Véry & Sentenac, 2003). Concerning K+ channels, such regulation has thus far mostly been documented at the transcriptional level, in the Shaker family (Golldack et al., 2003; Véry & Sentenac, 2003; Zhu, 2003; Fuchs et al., 2005; Zhang et al., 2006). Although the links with salt tolerance were not always fully understood, transcript abundances of a few inward (or weak inward) Shaker channel genes have been reported to be increased or decreased upon salt stress conditions, in both salt-sensitive and salt-tolerant species. For instance, in Arabidopsis, the AtKC1 gene, which encodes a Shaker subunit acting as a negative regulator of Shaker channel-mediated K+ uptake in roots (Reintanz et al., 2002; Duby et al., 2008), was up-regulated by salt stress (Kreps et al., 2002; Pilot et al., 2003a), suggesting a down-regulation of the membrane conductance to K+ in root periphery cells during salt stress. In ice plant (Mesembryanthemum crystallinum), a salt-tolerant species, root K+ uptake seemed also to be down-regulated during salt stress since both the transcript abundance of MKT1 (a homolog of the Arabidopsis root K+ uptake AKT1 channel gene) and the amount of MKT1 proteins at the plasma membrane were reduced (Su et al., 2001). In rice, the transcriptional regulation during salt stress of a homolog of AKT1 in salt-tolerant Na+-excluding varieties was in line with that observed in the ice plant, since the Shaker gene expression was down-regulated in the exodermis of salt-tolerant varieties (Golldack et al., 2003) while such regulation was not observed in salt-sensitive varieties.

Regulation by Na+ of plant K+ channel activity is less well documented. Electrophysiological analyses on protoplasts have revealed blockage of K+ efflux by high cytosolic Na+ in guard cells from different species upon membrane depolarization (Thiel & Blatt, 1991; Véry et al., 1998). High cytosolic Na+ was also reported to indirectly inhibit, via a signalling pathway, inward K+ currents (reminiscent of Shaker channel currents) in guard cells of a salt-tolerant Aster species, but not of a salt-sensitive species (Véry et al., 1998). Also, a blockage by external Na+ of inward K+ currents has been reported in Avena sativa mesophyll cells (Kourie & Goldsmith, 1992). On the other hand, a direct effect of external Na+ on a cloned plant K+ channel had never been reported. The inhibition by external Na+ of MIRK inward channel activity evidenced in this study is therefore, to date, a unique feature among cloned plant K+ channels.

MIRK current inhibition by external Na+ was immediately observed when Na+ was added to the bath and quickly reversed upon Na+ removal. This indicated that Na+ acted on MIRK in a direct manner from the external side and not via signaling events. Furthermore, MIRK was shown to be impermeable to Na+ (Fig. 3d,e) and the current inhibition by Na+ was independent of the applied voltage (Fig. 4d), suggesting that Na+ did not enter deeply inside MIRK’s channel pore. Several K+ binding sites have been evidenced in K+ channels from animals and bacteria, in the narrow region forming the selectivity filter, but also in the outer and inner vestibules of these channels (Doyle et al., 1998; Harris et al., 1998; Vergara et al., 1999; Thompson & Begenisich, 2001; Bernèche & Roux, 2003). Na+ might competitively bind to one such site in the external mouth of the MIRK channel and thereby obstruct the permeation pathway and prevent K+ translocation in both the inward and outward directions (Fig. 4d). The outer mouth of the plant Shaker channel is expected to be mainly determined by part of the P (‘pore’) domain, the S5-P linker and the P-S6 linker (Jiang et al., 2003). Molecular determinants of the inhibition by Na+ in the MIRK channel could be located in the S5-P linker, and certainly not (or not exclusively) in the P and P-S6 regions, since the MIRK sequence in these regions is identical to that in the VvSIRK channel, which does not share sensitivity to Na+ (Pratelli et al., 2002).

Possible role of MIRK regulation by Na+ in melon salt tolerance

Decreases in melon vegetative growth and fruit yield under saline conditions (electric conductivity of 7–14 dS m−1, corresponding to Na+ concentration in the range 50–120 mM) have been reported (Shannon et al., 1984; Mendlinger & Pasternak, 1992; Mavrogianopoulos et al., 1999). Melon is considered a rather salt-tolerant crop, although differences exist among cultivars (Shannon et al., 1984; Mangal et al., 1988; Mendlinger & Pasternak, 1992). The melon cultivar used in our study, Chunli, is quite resistant to salt. Indeed, growth tests have shown that its vegetative growth was even slightly stimulated when 50 mM Na+ was added to the growth medium. Growth was normal in the presence of 100 mM Na+, and significantly reduced only when Na+ concentration was higher than 150 mM (not shown). We have identified at the molecular level a melon K+ uptake channel and evidenced an inhibition by external Na+ of its activity. Could this functional property be linked to the melon cultivar’s salt tolerance?

Reverse transcription-PCR and patch-clamp experiments provide evidence that MIRK is expressed in guard cells (Figs 2b, 7), in agreement with the fact that MIRK belongs to the Shaker channel group 2 (Pilot et al., 2003b), which is characterized (so far) by expression of its members in guard cells (Müller-Röber et al., 1995; Nakamura et al., 1995; Pilot et al., 2001; Pratelli et al., 2002). There is little information on the concentration of K+ and Na+ in guard cell apoplast upon salt stress or even under normal conditions. K+ concentration values varying within more than two orders of magnitude (submillimolar values to several tens of mM) have been reported (Blatt, 1985; Bowling, 1987; Szyroki et al., 2001). For Na+, a concentration up to several tens of mM during salt stress can be expected. Hence, a regulation by external Na+ of K+ channel activity such as that evidenced in MIRK, with half-inhibition occurring with Na+ in the range 10–100 mM when K+ is in the range 0.1–1 mM (Figs 4, 5), can be predicted to be physiologically relevant.

Accumulation of Na+ instead of K+ in guard cells can lead to stomatal opening. However, stomata opened in the presence of high Na+ concentrations display impaired functioning, notably a reduced closure ability (Jarvis & Mansfield, 1980; Lebaudy et al., 2008). Several halophytes have been shown to decrease their stomatal aperture in response to an increase in apoplastic Na+ (Perera et al., 1994; Robinson et al., 1997). This particular response to high salt concentrations has been proposed to reduce Na+ delivery to the leaf apoplast via the transpiration stream and thereby to be part of the mechanisms involved in the salt tolerance of such halophytes. In one of these halophytes, Aster tripolium, high cytosolic Na+ concentrations have been shown to lead, probably via a signaling pathway involving Ca2+, to down-regulation of guard cell inward K+ channel activity, a regulation thought to explain the particular stomatal response to Na+ (Véry et al., 1998). The down-regulation of MIRK channel activity by Na+ relies on a mechanism different from that evidenced in the halophytic Aster, since external Na+ is involved in the regulation (Figs 4, 5) instead of cytosolic Na+. However, as in the Aster halophyte, the resulting down-regulation of inward K+ channel activity in guard cells (Fig. 7) is likely to reduce stomatal aperture upon salt stress and thereby to play a role in the salt tolerance of the melon cv Chunli.

Acknowledgements

We thank Prof. A Rodríguez-Navarro (Universidad Politécnica de Madrid) for kindly providing the yeast mutant strain. This research was funded by Key Projects of Science and Technology Research (108141, Ministry of Education of the People’s Republic of China), Shanghai international cooperation research program (072307011) and Shanghai fundamental research program (09JC1408500, Shanghai Science and Technology Committee). We also thank the Project ‘Arcus 2006 Languedoc-Roussillon/China’ and the Leading Academic Discipline Project (B209) of Shanghai Municipal Education Commission for financial support.

Ancillary