Note: The nucleotide sequences reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession numbers AB052880 and AB052881. The structure reported in this paper has been submitted to the Protein Data Bank with accession number 1JU8, BMRB 5098.
Note: As the capability of binding of insulin and the 4-kDa peptide to the 43-kDa protein is similar, in our previous paper we named the 4-kDa peptide as leginsulin. But the 4-kDa peptide is not insulin, and one must discriminate between them. To avoid confusion, we use 4-kDa peptide as the name of the peptide instead of leginsulin in this paper.
H. Hirano, Yokohama City University, Kihara Institute for Biological Research/Graduate School of Integrated Science, Totsuka, Yokohama, 244–0813 Japan. Fax: + 81 45 8201901, Tel.: + 81 45 8201904, E-mail: firstname.lastname@example.org
Previously, we isolated a 4-kDa peptide capable of binding to a 43-kDa receptor-like protein and stimulating protein kinase activity of the 43-kDa protein in soybean. Both of them were found to localize in the plasma membranes and cell walls. Here, we report the physiological effects of 4-kDa peptide expressed transiently in the cultured carrot and bird's-foot trefoil cells transfected with pBI 121 plasmid containing the 4-kDa peptide gene. At early developmental stage, the transgenic callus grew rapidly compared to the wild callus in both species. Cell proliferation of in vitro cultured nonembryogenic carrot callus was apparently affected with the 4-kDa peptide in the medium. Complementary DNAs encoding the 4-kDa peptide from mung bean and azuki bean were cloned by PCR and sequenced. The amino-acid sequences deduced from the nucleotide sequences are homologous among legume species, particularly, the sites of cysteine residues are highly conserved. This conserved sequence reflects the importance of intradisulfide bonds required for the 4-kDa peptide to perform its function. Three dimensional structure of the 4-kDa peptide determined by NMR spectroscopy suggests that this peptide is a T-knot scaffold containing three β-strands, and the specific binding activity to the 43-kDa protein and stimulatory effect on the protein phosphorylation could be attributed to the spatial arrangements of hydrophobic residues at the solvent-exposed surface of two-stranded β-sheet of 4-kDa peptide. The importance of these residues for the 4-kDa peptide to bind to the 43-kDa protein was indicated by site-directed mutagenesis. These results suggest that the 4-kDa peptide is a hormone-like peptide and the 43-kDa protein is involved in cellular signal transduction of the peptide.
A 43-kDa protein found in the soybean seeds is a glycoprotein with sedimentation coefficient of 7S and isoelectric point ranging from 9.05 to 9.26 . This protein has been classified into the category of globulin, which is soluble only in high ionic strength of salt solutions . It consists of α and β subunits linked by disulfide bridge(s). There are a cysteine-rich domain in the N-terminal side of α subunit, a putative transmembrane domain in the β subunit , and a consensus sequence of ATP-binding site indispensable for protein phosphorylation activity . The 43-kDa protein has autophosphorylation activity and protein kinase activity about two thirds of tyrosine kinase activity of the rat insulin receptor . Immunocytochemistry has indicated that the 43-kDa protein is localized in the plasma membranes and the middle lamellae of cell walls , suggesting that it is a receptor-like protein. Western blotting and DNA cloning experiments revealed that these proteins are structurally similar to the 43-kDa protein and distribute in a number of legume species such as azuki bean, cowpea, French bean, lupin, mung bean and winged bean [5,6], and nonlegume species such as carrot .
The presence of this receptor-like protein has allowed us to predict that the physiologically active peptides, which are capable of binding to the 43-kDa protein, may also be present in plants. To isolate such peptides, affinity chromatography using Sepharose CL-4B column immobilized the 43-kDa protein as a ligand was conducted . By this chromatography, we purified a 4-kDa peptide from the fractionated extract of soybean radicles. Ligand blotting experiments using the radioiodinated 4-kDa peptide confirmed that this peptide is capable of binding to the 43-kDa protein . Maximum stimulatory effect was observed at relatively low concentration (1 nm) of the 4-kDa peptide, indicating possible involvement of the 4-kDa peptide and 43-kDa protein in some cellular signal transduction .
Immunocytochemical studies revealed that a small amount of 4-kDa peptide is localized around the plasma membranes and cell walls . The subcellular localization of 4-kDa peptide is similar to that of the 43-kDa protein, suggesting that the 4-kDa peptide is localized at the site suitable for interaction with the 43-kDa protein.
The present study is performed to understand the physiological function of the 4-kDa peptide, and suggests that this peptide is involved in the regulation of callus growth and cell proliferation. Tertiary structure of the 4-kDa peptide has revealed that this peptide is a T-knot scaffold containing three β-strands, and the specific binding activity to the 43-kDa protein and stimulatory effect on the protein phosphorylation, which could be attributed to the spatial arrangements of the hydrophobic residues at the solvent-exposed surface of the two-stranded β-sheet of 4-kDa peptide. The site-directed mutagenesis suggests the importance of these residues in binding it to the 43-kDa protein.
Transformation of the 4-kDa peptide gene
Seeds of carrot (Doucus carota L., cvs. Benibijin and Harumakisanzun) were surface-sterilized in 2.5% (v/v) hypochlorite solution containing 0.02% (v/v) Tween 20 with shaking for 20 min. After washing with deionized water, the seeds were planted on Murashige & Skoog (MS) medium containing 30 g·L−1 of sucrose and 3 g·L−1 of Gelrite, and incubated at 25 °C under continuous illumination. After 1–2 weeks, hypocotyls of the developed seedlings were sliced into 3 mm segments and placed on MS medium containing 30 g·L−1 of sucrose, 2 mg·L−1 of 2,4-dichlorophenoxy acetic acid (2,4-D). After 2 days, the explants were transferred to MS medium containing 30 g·L−1 of sucrose, 3 g·L−1 of Gelrite without 2,4-D, and cultured for 10 days. Agrobacterium tumefaciens strains LBA4404  and the binary vector pBI 121  obtained from Clontech, CA were used for the transformation. pBI 121 contains a udi coding region of the Escherichia coliβ-glucronidase reporter gene (udiA) under the control of the cauliflower mosaic virus (CaMV) 35S promoter and a polyadenylation signal of nopaline synthetase gene (nos) region. The 4-kDa peptide gene was inserted, in either sense (545 bp DNA) or antisense directions, between the 35S promoter and udiA gene of pBI 121 plasmid. The pBI 121 was introduced into A. tumefaciens LBA 4404 by triparental mating with E. coli pRK 2013 as a helper strain. A. tumefaciens strain LBA4404 containing pBI 121was grown for 2 days on LB medium (10 g·L−1 of Bacto tryptone, 5 g·L−1 of Bacto yeast extract, 10 g·L−1 of NaCl) containing 100 mg·L−1 of kanamycin and 50 mg·L−1 of rifampicin. The hypocotyl segments of carrot described above were immersed in the bacterial suspension for 2 h and then transferred to MS medium containing 100 mg·L−1 of kanamycin, 100 mg·L−1 of cefotaxime and 2 mg·L−1 of 2,4-D, incubated at 25 °C and subcultured at 25 °C at 2-week intervals. After eight weeks, the callus regenerated from hypocotyl on the first selection medium was explanted for induction adventitious embryo on MS medium containing 100 mg·L−1 of kanamycin, 250 mg·L−1 of cefotaxime. The induced embryogenic callus was subcultured on the same medium to develop the plantlets.
Transformation was also performed in bird's-foot trefoil (Lotus cornialatus) by the methods as described above except that cotyledons were used as materials and 1 mg·L−1 of benzyladenine instead of 2,4-D was added into the medium. Histochemical and fluorometric assays for GUS activity were performed as described .
Carrot cell culture
The carrot nonembryogenic cells gifted by S. Satoh  were grown at 25 °C in MS liquid medium containing 30 g·L−1 of sucrose and 2 mg·L−1 of 2,4-D. The suspension was subcultured at two-week intervals. Three days after the final transplanting, the cells were precipitated by centrifugation (100 g, 5 min) and resuspended in the same medium containing different concentrations (0.1 pm, 1, 100 nm, 1, 10 and 100 µm) of the 4-kDa peptide at a density of 0.5 × 105 cells·mL−1. After 3, 7, 10 and 14 days, the cells were harvested to determine the density. Experiments were repeated six times.
Nucleotide sequence of DNAs encoding the 4-kDa peptide superfamilies
The genomic sequences coding for the 4-kDa peptide precursor polypeptides were amplified by PCR strategy using the azuki bean and mung bean genomic DNAs as templates and the synthetic primers legF1 (5′-AGCAGCAGATTGTAATGGTG-3′) and legR1 (5′-CAGCACTTCAGAATCAGAGTC-3′). PCR products were cloned on pT7Blue T-vector (Novagen, Darmstadt) and their nucleotide sequences were determined. The amino-acid sequences of 4-kDa peptides were deduced from the nucleotide sequences.
Tertiary structure of the 4-kDa peptide
Natural 4-kDa peptide purified from soybean radicles as described in  and chemically synthesized one were used for NMR studies. The reduced peptide obtained by solid-phase synthesis using Boc strategy was subjected to oxidative folding in 0.1 m AcONH4 buffer (pH 7.4) in a 50% (v/v) aqueous isopropyl alcohol solution containing 0.5 m guanidine hydrochloride at a peptide concentration of 10−5m, in the presence of reduced and oxidized glutathione (GSH/GSSG) as redox reagents at room temperature for 60 h. The molar ratio of peptide/GSH/GSSG was set to 1 : 100 : 10. The crude cyclic peptide was purified on preparative high performance liquid chromatography (HPLC) with a C18 column. The homogeneity of the synthesized product purified by reversed phase-HPLC was further confirmed by amino-acid analysis, ion-exchange-HPLC, capillary zone electrophoresis and matrix assisted laser desorption ionization time-of-flight mass spectrometry. As both natural and chemically synthesized 4-kDa peptides provided the same NMR spectra at concentration of 200 µm, the synthesized peptide was used for further detailed NMR analysis. The solution used for NMR structure determination contained about 4 mm synthesized 4-kDa peptide in 70% H2O and 30% CD3COOD at pH 1.8. We obtained no evidence for any conformational changes and aggregation of the 4-kDa peptide even at the higher peptide concentration. All NMR spectra were recorded at 25, 40 and 50 °C on a Bruker DMX750 spectrometer equipped with a x,y,z-shielded gradient probe. Complete sequence-specific assignments for all backbone and side-chain protons were obtained using two-dimensional DQF-COSY, HOHAHA and NOESY experiments.
Structures of the 4-kDa peptide were calculated using the hybrid distance geometry-dynamical simulated annealing protocol within x-plor. For structure calculations, we used 541 interproton distance restraints [comprising 229 intraresidue, 161 sequential (|i – j| = 1), 56 medium-range (1 < |i – j| < 5) and 95 long-range (|i – j| > 5)] obtained from NOESY spectra with a mixing time of 150 ms. In addition to the NOE-derived distance restraints, 16 distance restraints for eight hydrogen bonds and 55 dihedral angle restraints (20 φ, 14 ψ, 19 χ1 and 2 χ2) were included in the structure calculation. A peptide bond between Val12 and Pro13 was set to a cis configuration, i.e. ω≈ 0°, based upon observation of an extremely strong sequential NOE between the Val12 Hα and Pro13 Hα. Structure calculations were first carried out without restraints regarding disulfide bridges. Analysis of Cα–Cα and Sγ–Sγ distances between cysteines observed for the resultant structures led to identification of disulfide bond pairings of the 4-kDa peptide as Cys3–Cys20, Cys7–Cys22 and Cys15–Cys32. The disulfide bond between Cys15 and Cys32 was experimentally confirmed by amino-acid sequence analysis of several peptide fragments generated from hydrolysis of the natural 4-kDa peptide with 10% (v/v) phosphoric acid at 101 °C for 15 h. Hence, the final structure calculations included disulfide bond restraints in addition to the NMR-derived distance and dihedral angle restraints. A final set of 15 lowest-energy structures was selected from 100 calculations. None of them had NOE and dihedral angle violations > 0.05 nm and 5°, respectively. The average coordinates of ensembles of the final 15 structures were subjected to 500 cycles of Powell restrained energy minimization to improve stereochemistry and nonbonded contacts. Figures were generated using molmol.
The wild-type DNA sequence of 4-kDa peptide was amplified from the soybean 4-kDa peptide cDNA by polymerase chain reaction (PCR) using the following oligonucleotide primers: the 4-kDa peptide N-terminal primer: 5′-AACCATGGCTAAAGCAGATTGTAATGGTGCATGT-3′; the 4-kDa peptide C-terminal primer: 5′-AAGAATTCTTATTATCCAGTTGGATGTATGCAGAA-3′. The amplified sequence was cloned into the plasmid pKF18 via the EcoRI and SalI restriction sites in the multicloning site. This plasmid was designated as pKF18/LEG.
Site-directed mutagenesis was performed using pKF18/LEG as a template using the commercial kit of oligonucleotide-directed dual amber method (Mutan-Super Express Km, Takara Biochemicals, Osaka) . Arg16, Val29 and Phe31 in the 4-kDa peptide were singly replaced by Ala with pKF18/LEG, selection primer included in the Mutan-Super Express Km and the following oligonucleotide primers (mismatches are underlined): Variant R16A 5′-CCACCGTGCGCCTCACGTGATTG-3′, Variant V29A 5′-GGACTATTTGCTGGTTTCTGC-3′, Variant F31A 5′-CTATTTGTTGGTGCCTGCATACATC-3′. All variants were verified to be correctly constructed by dideoxy sequencing. Each DNA fragment of the 4-kDa peptide variants was removed by the EcoRI and SalI restriction enzymes and recloned into pET-32a(+). The 4-kDa peptide and its variants were prepared by the Escherichia coli protein expression system.
The assay of binding activity of the mutant 4-kDa peptides to the 43-kDa protein was carried out by ligand blotting. The 43-kDa protein was separated by SDS/gel electrophoresis and electroblotted onto a poly(vinylidene difluoride) membrane. The poly(vinylidene difluoride) membrane was soaked in Tris buffered NaCl/Pi (Tris/NaCl/Pi) for 5 min, and in Tris/NaCl/Pi containing 1% (w/v) skimmed milk for 1 h, then in Tris/NaCl/Pi for 10 min. The poly(vinylidene difluoride) membrane was packed in the plastic bag with 5 µg of the 4-kDa peptide or mutant 4-kDa peptides in 2 mL of Tris/NaCl/Pi. The membrane was incubated overnight at 4 °C and washed twice with Tris/NaCl/Pi for 5 min. Then, rabbit anti-(4-kDa peptide) Ig in 5 mL of Tris/NaCl/Pi was added. The membrane was incubated for 1 h at 4 °C, and washed twice with Tris/NaCl/Pi for 5 min. Goat anti-(rabbit IgG) Ig conjugated with alkaline phosphatase in 5 mL of Tris/NaCl/Pi was then added. The membrane was incubated for 1 h at 4 °C, and washed twice with Tris/NaCl/Pi for 5 min. Finally, the cross-reacted bands were detected with alkaline phosphatase substrate (Moss, Maryland).
Results and discussion
Possible physiological function of the 4-kDa peptide
To investigate the physiological functions of the 4-kDa peptide, we transiently expressed the peptide in the cultured carrot cells transfected with pBI 121 plasmid containing the 4-kDa peptide gene and GUS gene as a reporter gene using Agrobacterium transformation system. The presence of 4-kDa peptide gene in the transgenic plants was confirmed by Southern blotting (Fig. 1). The GUS activity was constitutively detected in the roots and leaves of the transgenic plants (Fig. 1). The transgenic plant has two integration sites for the 4-kDa peptide gene, as two bands were detected when the DNA was digested with HindIII. As shown in Fig. 1, at early developmental stage, the transgenic callus rapidly grew compared with the wild callus. Three weeks after transplanting, the growth ratio of the callus was 162.8 ± 84.6 for the transformant to 15.3 ± 5.3 for the nontransformant (control). However, the phenotype of intact transgenic plants regenerated from the calli was not noticeably different from that of the wild plant. The 43-kDa protein has been detected in the wild carrot cells , but not the 4-kDa peptide. This result suggests that the 4-kDa peptide, which was synthesized by the transfected gene, stimulated protein kinase activity of the carrot 43-kDa protein, and the signal transduction pathway was activated for the regulation of growth of callus. We transfected the carrot cells with pBI 121 plasmid containing the antisense 4-kDa peptide gene, but could not detect any significant effect of the antisense gene on the development of callus, probably because there was no 4-kDa peptide-like peptide gene which could interact with the anisense gene in carrot.
We tried to culture the soybean callus in vitro to use it for the 4-kDa peptide gene transformation. However, the shoots and roots were not easily differentiated from the callus, and consequently we could not obtain the transgenic plants. However, we could construct the transgenic ones in bird's-foot trefoil instead of soybean. In this case, the 4-kDa peptide showed a similar effect on growth of the callus in the transgenic bird's-foot trefoil to that of the transgenic carrot (data not shown).
On the other hand, we investigated the effects of 4-kDa peptide on proliferation of the carrot auxin-autotropic nonembryogenic cells, which lost embryogenic competence. When the 4-kDa peptide was added into the liquid culture medium containing 2 mg·L−1 of 2,4-D, the cell proliferation was stimulated depending on the 4-kDa peptide concentration. The optimum concentration for maximum cell proliferation was 1 µm in the culture medium (Fig. 2). This indicates that the 4-kDa peptide may also be involved in the regulation of carrot cell proliferation.
Tertiary structure of the 4-kDa peptide in solution
The 4-kDa peptide consisting of 37 residues contains 6 half-cystines in three disulfide bridges (Fig. 3). Disruption of the disulfide bridges leads to a complete loss of the stimulatory effect of 4-kDa peptide on the phosphorylation activity of 43-kDa protein , indicating that the disulfide bridges might play an important role in maintaining the correct three-dimensional structure of 4-kDa peptide required for its function. Complementary DNAs encoding the 4-kDa peptide from mung bean and azuki bean were cloned by PCR and sequenced. The amino-acid sequences deduced from the nucleotide sequences are homologous among legume species, particularly, the sites of cysteine residues are highly conserved (Fig. 3). This conserved sequence reflects the importance of intradisulfide bonds required for the 4-kDa peptide to perform its function.
To investigate the structural basis for the 4-kDa peptide function, we have determined its three-dimensional structure by 1H-NMR spectroscopy. In the present study, the NMR structure was determined at pH 1.8, as the purified and lyophilized 4-kDa peptide is soluble only at this pH. Similar to the 4-kDa peptide, the higher solubility of the purified animal insulin and invertebrate insulin-like peptides have been reported elsewhere [20,21]. No information of these peptides at higher pH, which might cause chemical sift related to structural rearrangements, is available. As shown in Fig. 4, the 4-kDa peptide was found as a T-knot scaffold containing 3 β-strands (βA: Ala6–Ser8; βB: Cys20–Pro24; βC: Gly30–His34). Two adjacent β-strands, βB and βC, connected by a distorted type-I β-turn around Gly26-Val29, make up a two-stranded antiparallel β-sheet which is stacked by a long N-terminal loop containing βA, 2 type-I β-turns around Ser8–Glu11 and Ser17–Cys20, and a cis proline at position 13. The stacked structure is stabilized by 3 disulfide bridges formed between the β-sheet and the N-terminal loop. It has been reported that the T-knot scaffold  is shared by several small, disulfide-rich proteins with diverse functions, such as potato CPI  and calcium channel blockers ω-conotoxin GVIA from the venom of cone snail and ω-agatoxin-IVB from the venom of funnel web spider  (Fig. 6C). The X-ray crystal structure of CPA–CPI complex has revealed that CPI recognizes the enzyme using the C-terminal tetrapeptide and residues at the solvent-exposed surface of strand βB .
By ligand blotting experiments using 125I-labelled 4-kDa peptide, Watanabe et al.  demonstrated that the 4-kDa peptide competes with insulin for binding to the 43-kDa protein. This suggests that the 4-kDa peptide and insulin bind to the same sites of the 43-kDa protein in a similar manner, although both peptides were considered to have totally different folds. Hence, we have assumed that the 4-kDa peptide and insulin may possess similar spatial arrangements of functional residues, and searched topochemical similarity in local structures between them.
In the case of insulin, residues at the A-chain N-terminus (GlyA1–IleA2–Val13A) [24,25], the A-chain C-terminus (TyrA19 and AsnA21) , the B-chain central helix (ValB12 and TyrB16) , and the B-chain C-terminal β strand (PheB24–PheB25–TyrB26) [28–30] have been shown to be important for receptor recognition. In the solution structure of human insulin (Fig. 5) , which is considered as the locked and inactive state, most of these residues form an extensive hydrophobic core at the interface between the B-chain β-strand and the B-chain central helix as well as the A-chain N- and C-terminal regions. It is generally believed that receptor binding is accompanied by some degree of conformational change of insulin from the locked, inactive state to the active state. Although the receptor bound conformation of insulin has not yet been experimentally determined, a model has been proposed based on the solution structure of the biologically active insulin analog, [GlyB24]human insulin, where the orientation of the disordered B-chain C-terminal region relative to the rest of molecule is not well defined . The proposed model is described as the unlocked state where the B-chain C-terminal β-strand is detached from the rest of molecule, resulting in exposure of the insulin pharmacophore to the receptor. This model is further supported by the high potency of des-(B26-B30)-insulin amide in which the B chain from TyrB26 to the C-terminus is truncated but the resultant new C-terminal PheB25 is amidated . It is worthwhile mentioning that among the insulin pharmacophore, IleA2-ValA3 at the A-chain N-terminus, TyrA19 at the A-chain C-terminus, and ValB12 and TyrB16 at the B-chain central helix assume essentially the same spatial arrangements both in the locked, inactive state and in the unlocked state (Fig. 5). Therefore, we first searched a tetragonal arrangement made of hydrophobic residues, TyrA19, ValB12, TyrB16 and either one of IleA2 or ValA3 of insulin, for the 4-kDa peptide structure.
We found that only the 4 hydrophobic residues, Val23, Val29, Phe31 and Ile33, at the solvent-exposed surface of the 4-kDa peptide β-sheet could fulfill essentially the same tetragonal arrangement as TyrA19, TyrB16, ValB12 and ValA3 of insulin (Fig. 5). Furthermore, it was turned out that when the 4-kDa peptide was superimposed against the unlocked, active state model of insulin using these tetragonal arrangements, Leu27 and Phe28 of the 4-kDa peptide could occupy similar space as PheB25 and TyrB26 of the insulin pharmacophore. These results suggest that the specific binding activity of 4-kDa peptide to the 43-kDa protein and its stimulatory effect on the protein phosphorylation are attributed to the spatial arrangements of the hydrophobic residues at the solvent-exposed surface of the two-stranded β sheet.
The preliminary experiments using site-directed mutagenesis suggested that the substitution of the hydrophobic residues at the solvent-exposed surface of the two-stranded β sheet caused a significant change in its binding activity to the 43-kDa protein (Fig. 6). Although there was no great difference in the activity between the normal and the Arg16→Ala mutant 4-kDa peptides, the Val29→Ala and Phe31→Ala mutant 4-kDa peptides bound to the 43-kDa protein less strongly than the normal 4-kDa peptide. These results show the importance of these residues for the 4-kDa peptide to function.
There are several reports on hormone-like peptides in plants. Systemin is an 18 residue-peptide which can induce the transcription of the tomato protease inhibitor gene in response to insect damage . The systemin signal has been considered to be transduced through the octadecanoid signalling pathway. This signalling mechanism seems to be different from that of the 4-kDa peptide. Phytosulfokines A and B which are eight- and five-residue peptides, respectively, can stimulate the proliferation of asparagus and rice cultured cells . Recently, the putative receptors for phytosulfokine were identified in the rice plasma membrane , and RALF, a 5-kDa peptide from tobacco leaves was also reported to reduce the root growth and development of tomato and Arabidopsis. A potential receptor-like serine/threonine protein kinase CLAVATA 1 and its ligand CLAVATA 3, which regulates cell proliferation and differentiation at the shoot meristem , were identified in Arabidopsis. The finding of these hormone-like peptides including the 4-kDa peptide strongly suggest the presence of hormone peptides in plants and their function as signal transduction systems, which are similar to the animal systems.
This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, and National Project on Protein Structural and Functional Analyses to H. H., and a grant from the Bio-oriented Technology Research Advancement Institution, Japan to T. Y. We thank Nazrul Islam for his help in preparing the manuscript.