Desiccation and zinc binding induce transition of tomato abscisic acid stress ripening 1, a water stress- and salt stress-regulated plant-specific protein, from unfolded to folded state.

Abscisic acid stress ripening 1 (ASR1) is a low molecular weight plant-specific protein encoded by an abiotic stress-regulated gene. Overexpression of ASR1 in transgenic plants increases their salt tolerance. The ASR1 protein possesses a zinc-dependent DNA-binding activity. The DNA-binding site was mapped to the central part of the polypeptide using truncated forms of the protein. Two additional zinc-binding sites were shown to be localized at the amino terminus of the polypeptide. ASR1 protein is presumed to be an intrinsically unstructured protein using a number of prediction algorithms. The degree of order of ASR1 was determined experimentally using nontagged recombinant protein expressed in Escherichia coli and purified to homogeneity. Purified ASR1 was shown to be unfolded using dynamic light scattering, gel filtration, microcalorimetry, circular dichroism, and Fourier transform infrared spectrometry. The protein was shown to be monomeric by analytical ultracentrifugation. Addition of zinc ions resulted in a global change in ASR1 structure from monomer to homodimer. Upon binding of zinc ions, the protein becomes ordered as shown by Fourier transform infrared spectrometry and microcalorimetry, concomitant with dimerization. Tomato (Solanum lycopersicum) leaf soluble ASR1 is unstructured in the absence of added zinc and gains structure upon binding of the metal ion. The effect of zinc binding on ASR1 folding and dimerization is discussed.

Tomato ASR1 is localized in both cytosol and nucleus compartments (Kalifa et al., 2004a). Fusion proteins of ASR homologs with reporter proteins were shown to be localized in nuclei (Cakir et al., 2003;Wang et al., 2003bWang et al., , 2003c. Tomato ASR1 (Kalifa et al., 2004a) and the grape (Vitis vinifera) ortholog VvMSA (maturation-, stress-, and ABA-induced protein; Cakir et al., 2003) were shown to posses DNA-binding activity. The zinc-dependent DNA-binding activity of tomato ASR1 can be obtained in a carboxy terminus fragment of ASR1 (Rom et al., 2006). Two zinc-binding sites were mapped at the amino terminus of the protein (Rom et al., 2006). 1 This work was supported by the Israel Science Foundation (to D.B.Z. and Y.G.) and in part by the Intramural Research Program of the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (to R.G.).
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Dudy Bar-Zvi (barzvi@bgu.ac.il). [W] The online version of this article contains Web-only data. [OA] Open Access articles can be viewed online without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.106.092965 Overexpression of tomato ASR1 protein in tobacco (Nicotiana tabacum) plants results in an increased salt tolerance and in the modulation of expression of other genes (Kalifa et al., 2004b). Overexpression in Arabidopsis (Arabidopsis thaliana) of the ortholog LLA23 gene from lily (Lilium longiflorum) increases the plant tolerance to drought and salinity (Yang et al., 2005). Maize (Zea mays) ASR1 was proposed as a candidate gene for quantitative trait locus for drought stress response (de Vienne et al., 1999). Moreover, the ASR orthologs VvMSA (Cakir et al., 2003) and LLA23 (Yang et al., 2005) were suggested to be involved in the expression of sugar-metabolizing genes and in ABA-signaling pathways, respectively.
Although it is a DNA-binding protein, ASR1 does not share sequence or structure homologies with other known DNA-binding proteins (see Kalifa et al., 2004a;Rom et al., 2006). ASR1, a low M r , highly charged protein, is predicted to be intrinsically unstructured. Recently, intrinsically unstructured/disordered or natively unfolded proteins have become a focus of scientific interest (Uversky, 2002(Uversky, , 2006Tompa, 2002;Fink, 2005). It is estimated that as many as 30% of eukaryotic proteins are either completely or partially disordered (Fink, 2005). Folding of intrinsically unstructured proteins is suggested to regulate the activity of these proteins. Folding can be induced by the binding of cofactor or by proteinprotein interaction (Dyson and Wright, 2002). Certain stress proteins have been shown to be intrinsically unfolded. For example, some dehydrin proteins that are expressed at high levels under abiotic stress conditions were shown to be unstructured (Eom et al., 1996;Lisse et al., 1996;Soulages et al., 2003;Mouillon et al., 2006). Moreover, dehydrin homolog from a desiccationtolerant nematode was also shown to be unfolded under physiological condition (Goyal et al., 2003).
In this study, we demonstrate both by biochemical and biophysical methods that recombinant, full-size, nontagged ASR1 is disordered under physiological conditions. The protein transitions to an ordered state upon the binding of zinc ions, as demonstrated spectroscopically and calorimetrically. Fourier transform infrared (FTIR) studies indicate that folding can also be induced by desiccation of the ASR1 preparation. The oligomeric state of ASR1 was studied by analytical centrifugation. ASR1 was found to be monomeric in the absence of zinc and dimeric in the presence of the metal ion. Chemical cross-linking confirmed the ability of ASR1 to form homodimers. Tomato leaf cytosolic ASR1 is highly sensitive to protease degradation indicative of its unstructured nature. Zinc binding results in a decreased susceptibility of leaf soluble ASR1 to protease activity in agreement with the higher degree of structure induced by its binding to the protein.

Zinc Binding Induces Dimerization of ASR1
Sedimentation equilibrium experiments were carried out to determine the oligomeric state of ASR1 protein in solution. At all rotor speeds, the sedimentation equilibrium profiles were consistent with the presence of at least two species having very distinct molecular masses. Analyses in terms of two ideal solutes lead to excellent data fits ( Fig. 1A) with buoyant molecular masses [M i (1 2 v i r)] of 3,100 6 210 and 41,500 6 2,200 D for the low and high molecular mass species, respectively. The predominant species is the smaller mass species whose buoyant molecular mass corresponds to an experimental mass of 11,900 6 840 D, indicating that in the absence of added Zn 21 , ASR1 (calculated mass, 13,129.7 D) is monomer in solution. The large aggregate, present as an impurity, had a molecular mass approximately 12 times that expected for the monomer. Similar observations were made in the presence of added 1 mM ZnCl 2 , except that [M 1 (1 2 v 1 r)] 5 6,030 6 200 D was measured. This value corresponds to a calculated mass of 23,100 6 770 D, implying that in the presence of Zn 21 , ASR1 forms dimers (n 5 1.8 6 0.06; Fig. 1B). In addition, lower amounts of a larger species having a molecular mass of 60,600 6 2,100 D, consistent with the presence of smaller ASR1 aggregates (n 5 4.6 6 0.6), were observed in presence of added Zn 21 . A similar analysis on ASR1D61 to 115 showed that in the absence of added Zn 21 , the truncated protein had a calculated mass of 7,600 6 260 D, indicating that it is monomeric in solution (calculated mass, 7,044.8 D). Addition of 1 mM ZnCl 2 to the protein solution led to the formation of a polydisperse system. The major ASR1 forms were monomers and dimers, although higher species were also noted (data not shown).

Apo-ASR1 Is Predicted to Be Intrinsically Unstructured
Full-size tomato ASR1 can be expressed in Escherichia coli as a water-soluble protein. Attempts to crystallize the purified protein using a large number (.600) of conditions were not successful. Primary amino acid sequence of ASR1 was thus analyzed using PONDR (Protein Disorder Predictor) VSL1 predictor (http:// www.pondr.com) and the FoldIndex program (Prilusky et al., 2005; http://bioportal.weizmann.ac.il/fldbin/ findex) based on the algorithm proposed by Uversky et al. (2000a). Figure 3 shows that according to these predictions, ASR1 protein is mostly unfolded under physiological conditions. Disordered proteins are relatively enriched in disorderpromoting amino acids (E, K, R, G, Q, S, P, and A) and depleted in order-promoting residues (I, L, V, W, F, Y, and C; Romero et al., 2001). Four amino acid residues contribute 61.7% of the total ASR1 protein residues. Three of these, Lys, Glu, and Ala (20, 18, and 15 residues, respectively), are disorder promoting and the fourth, His (18 residues), is neutral (other). Overall, 67 out of the 115 amino acid residues of ASR1 are disorder promoting. On the other hand, only 21 of the 115 amino acid residues belong to the order-promoting group. ASR1 also lacks any W and C residues, defined as order-promoting residues. Figure 4 shows that the percentage of disorder-promoting amino acid residues in ASR1 is higher than average tomato encoded proteins (http://bioinformatics.weizmann.ac.il/blocks/ help/CODEHOP/codons/tomato.codon.use). Likewise, order-promoting residues are more abundant in average tomato proteins than in ASR1.
To experimentally test this prediction, full-length ASR1 protein was expressed in E. coli and purified to homogeneity (Kalifa et al., 2004a). The M r of the expressed protein was confirmed by mass spectrometry (Rom et al., 2006). The purified 13.1-kD tomato ASR1 protein migrates in SDS-PAGE as an 18-kD protein (Kalifa et al., 2004a;Rom et al., 2006) due to its high content of charged residues. Migration anomaly in SDS-PAGE was demonstrated for other highly charged proteins, many of which are natively unfolded (e.g. Weinreb et al., 1996).

Gel Filtration Analysis
The Stokes radius (R ST ) of purified ASR1 was determined using gel filtration chromatography using HiLoad 16/60 Superdex 200 column. The column was calibrated with Blue dextran (eluted at exclusion volume), bovine serum albumin (R ST 5 35.5 Å ), ovalbumin (R ST 5 30.5 Å ), lysozyme (R ST 5 19.8 Å ), and AMP (eluted at total volume). The column was equil-ibrated and eluted with 50 mM Tris-HCl, pH 7.5, 100 mM NaCl buffer at a flow rate of 0.5 mL/min. Elution times for the calibration standards were 64, 106, 116, 142, and 167 min, respectively. ASR1 protein eluted at 124 min. Including zinc in the elution buffer resulted in protein precipitation within the column. Using the elution volumes of ASR1 and standards, the R ST of 27.4 Å was determined for ASR1 (Table I). Uversky (2002) developed a set of equations for calculating the predicted R ST of polypeptides of a given M r , assuming different conformations of globular proteins. Using these equations for calculating the R ST of a 13,121-D polypeptide, our data suggests that without zinc, the ASR1 protein cannot be a globular protein but rather exists in an unfolded protein (Table I).

Far-Ultraviolet Circular Dichroism Analysis
Circular dichroism (CD) is often used for the assessment of the fraction of structural components within a Figure 1. Sedimentation equilibrium analyses of ASR1. Full-length ASR1 (0.2 mM) in buffer containing 20 NaPi pH 7.2 and 0.1 M NaCl buffer (A) or in buffer containing 1 mM ZnCl 2 (B) was brought to equilibrium at 4.0°C at three rotors speeds: A, 10,000 (red), 14,000 (green), and 18,000 rpm (blue); B, 10,000 (red), 12,000 (green), and 14,000 (blue). Protein concentrations were measured by UV absorbance. Best fits in terms of two ideal solutes are shown as solid black lines through the corresponding symbols.  Fig. S1) has a maximum negative ellipticity at 205 nm and low ellipticity around 215 to 225 nm, which indicates a heavily disordered protein structure (Zeev-Ben-Mordehai et al., 2003). Furthermore, no significant ellipticity was observed in the near UV, suggesting that unlike typical globular proteins, in ASR1 there is no hydrophobic core containing oriented aromatic residues.

Dynamic Light Scattering Studies
Dynamic light scattering (DLS) is used to determine the hydrodynamic radius of macromolecules. A hydrodynamic radius of 3.01 nm was determined for ASR1 without added zinc (Table II). This radius corresponds to a 44.3-kD globular protein (3.4 times the mass of monomeric ASR1) and is consistent with an unfolded 13.1-kD protein ( Table I). Addition of 1 mM ZnCl 2 resulted in an increase of the hydrodynamic radius of the protein to a value corresponding to globular protein of 100.2 kD. About 1.2% of the protein analyzed in the presence of zinc was aggregated with a calculated mass of about 9,000 kD. Similar hydrodynamic values were determined at pH 7 (data not shown). These results suggest that monomeric and dimeric ASR1 are not compact structures, because the hydrodynamic values determined for both forms are larger than those expected for a tightly packed globular protein.

Protein Structure Analysis by FTIR Spectrometry
Secondary protein structure can be estimated by monitoring the FTIR amide I band (1,600-1,700 cm 21 ). As distinct protein structural elements have specific peaks, the fraction of major structure elements in the protein can be determined by peak analysis of the obtained FTIR spectrum (Byler and Susi, 1986;Pribic et al., 1993). Freeze-dried ASR1 was dissolved in D 2 O to avoid interference of the HOH vibration with amid I band (Sieler and Schweitzer-Stenner, 1997). In the absence of zinc, the FTIR spectrum displayed a peak Figure 3. Prediction of the folding state of ASR1. A, ASR1 order prediction via the VSL1 program of PONDR using the default parameters. Values smaller and larger than 0.5 represent ordered and nonordered protein, respectively. B, ASR1 folding prediction via the FoldIndex program using the values predicted window 5 10 and step 5 1. Positive and negative numbers represent ordered and nonordered protein, respectively. C, The primary amino acid sequence of ASR1. Amino acids suggested in B as being ordered and nonordered regions are shown in regular and bold characters, respectively. at 1,642 cm 21 (Fig. 5A), which corresponds to the signature of a random coil (Byler and Susi, 1986;Krimm and Bandekar, 1986;Surewicz and Mantsch, 1988;Bandekar, 1992;Surewicz et al., 1993). Data analysis suggests that the peak assigned for a disordered structure represents over 50% of the protein ( Fig. 5A; Table III). Moreover, about 27% of the protein is assigned to be in turn structure (1,680 and 1,693 cm 21 peaks). The 1,664 cm 21 peak (13.2% of the total area) can be assigned to either turns (Byler and Susi, 1986) or to a 3 10 helix (Surewicz et al., 1993). No significant peak for b-strands was observed. Addition of zinc resulted in FTIR spectral changes for ASR1. The major (56.6%) 1,642 cm 21 and minor (13.2%) 1,664 cm 21 peaks corresponding to a disordered or a turn/ 3 10 helix structure, respectively, are diminished and replaced by peaks at 1,657 cm 21 (39.9%) and 1,626 cm 21 (16.6%) in the presence of zinc, corresponding, respectively, to a-helix and b-strands structures ( Fig.  5B; Table III). The relative intensities of the other fitted peaks revealed minor changes. Drying the ASR1 preparation resulted in similar changes of the observed spectrum to that observed for soluble ASR1 in the presence of added zinc (Fig. 5C). Drying ASR1 from a zinc-containing medium did not result in further changes (Fig. 5D).

Thermal Stability of ASR1
Differential microcalorimetry (DSC) was used to monitor temperature-induced changes in the structure of ASR1 plus and minus added zinc. No significant phase changes were observed in ASR1 solution without added zinc, over a temperature range of 4°C to 90°C (Fig. 6). On the other hand, zinc binding to ASR1 displayed structural heat denaturation at 76°C (Fig. 6), indicative of an ordered structure below this temperature.

Cytosolic ASR1 Is in Unstructured State
Unfolded proteins have increased sensitivity to proteases, and we showed that the susceptibility of puri-fied recombinant ASR1 to a number of proteases is decreased in the presence of zinc (Rom et al., 2006). Here, we analyzed the proteolytic susceptibility of cytosolic tomato ASR1 without or with added zinc. Leaf soluble ASR1 is rapidly degraded by endogenous tomato proteases and by added trypsin (Fig. 7A). Addition of zinc to the tomato leaf soluble protein fraction stabilized ASR1 and decreased markedly its proteolytic degradation. This is consistent with folding of the ASR1 protein. The protease effect is specific to ASR1, as the proteolytic activities had only a marginal effect on the overall electrophoretic protein pattern in the extract (Fig. 7B). Moreover, the concentrations of zinc used in this experiment did not inhibit the activity of trypsin (Rom et al., 2006).

ASR1 Is Present in Tomato Seeds and Pollen Grains
To address the possible physiological relevance of desiccation on inducing folding of ASR1 (Fig. 6C), we used anti-ASR1 antiserum to probe protein extracts prepared from tomato pollen grains and from fully developed tomato seeds. Seeds were isolated from fully ripened tomato fruit and rinsed thoroughly to remove residual loculus tissue, and pollen grains were collected from tomato flowers. Figure 8 shows that fully developed pollen grains and tomato seeds that contain very little water have high levels of ASR1 protein (comparable to that in salt-stressed leaf tissue on the basis of fresh weight). Oligomeric form of ASR1 is also observed in the leaf extract.

ASR1 as Hydrophilin
Hydrophilins are proteins defined by high hydrophilicity index and Gly content (.1.0% and .6%, respectively; Garay-Arroyo et al., 2000). Hydrophilins are found mainly in plants, bacteria, and yeast (Saccharomyces cerevisiae). They represent a small fraction (,0.2%) of the genome and are suggested to be a predictor for responsiveness to hyperosmosis (Garay-Arroyo et al., 2000). ASR1 meets with the hydrophilin criteria in that 7% of its amino acid residues are Gly, and the average protein hydrophilicity index is 1.17. Late embryogenesis abundant (LEA) proteins comprise the largest group of hydrophilins. Although  (Wise and Tunnacliffe, 2004), it shares similarities with some LEA proteins; for example, the expression of ASR1 and LEA is increased under osmotic stress. Moreover, some LEA proteins (e.g. proteins from LEA subgroups 1a, 2a, 3b, and 6) are DNAbinding proteins (for review, see Wise and Tunnacliffe, 2004).

ASR1 as Intrinsically Unstructured Protein
The ASR1 polypeptide chain is predicted to be intrinsically unstructured (Figs. 3 and 4). It is relatively poor in order-promoting hydrophobic amino acid residues and enriched in disorder-promoting charged amino acid residues. Uversky et al. (2000a) did a comparative study of known natively unfolded proteins. They showed that most natively unfolded proteins have low M r s with less than 150 amino acid residues. ASR1, a 115-residue protein, conforms to this size group. However, most natively unfolded proteins have either an acidic or basic isoelectric pH, whereas the calculated pI for ASR1 is almost neutral. According to Uversky et al. (2000a), 130 different nonhomologous proteins in the Swiss Protein database were predicted to be natively unfolded on the basis of length, low mean hydrophobicity, and high net charge. Tomato ASR1 was one of their predicted proteins.
Our data confirmed that Apo-ASR1 protein is nonstructured in solution, as demonstrated by gel filtration (Table I), DLS (Table II), CD (Supplemental Fig.  S1), FTIR (Fig. 5A), and microcalorimetry (Fig. 6). Thus, in the absence of added zinc, ASR1 may be added to the growing list of unstructured or poorly structured small proteins that possess biological roles related to water stress. For example, certain (though not all) dehydrins were shown to be disordered under native conditions (Eom et al., 1996;Lisse et al., 1996;Soulages et al., 2003;Mouillon et al., 2006). Nearinfrared spectroscopy analysis of a number of proteins of similar M r range with ASR1 suggested that freezedried proteins mostly maintain spectral characteristics of native structure (Izutsu et al., 2006). Thus, the freeze drying step in the preparation of ASR1 is not likely to induce any major structural changes. Possible minor changes resulting from freeze drying are expected to be reversible upon redissolving the dried protein. DSC scans of group 1 LEA protein from pea (Russouw et al., 1997) and of group 1 (Soulages et al., 2002) and group 2 (Soulages et al., 2003) LEA proteins from soybean (Glycine max) showed no detectable high temperature peak, suggesting that these proteins failed to undergo detectable heat denaturation. EBM-1, a group 1 LEA protein from carrot (Daucus carota), was suggested to be unstructured in solution (Eom et al., 1996). The native form of DSP16, a dehydrin-like protein from Craterostigma plantagineum, displayed no defined three-dimensional structure (Lisse et al., 1996). The nematode LEA protein homolog AavLEA1 was shown to be unfolded in solution (Goyal et al., 2003), and the D-7LEA protein from purified Typha latifolia pollen was shown to be highly unfolded in solution (Wolkers et al., 2001). Finally, a large fraction of the soluble ASR1 from tomato leaf exhibits an unfolded secondary structure as shown by increased protease sensitivity that could be reduced upon addition of zinc to the leaf homogenate (Rom et al., 2006).

Zinc Binding and Desiccation Induce Order and Dimerization in ASR1 Protein
FTIR spectral analysis suggests that in the presence of Zn 21 , ASR1 gains more a-helix and b-strand domains ( Fig. 5; Table III), which implies a more highly ordered polypeptide structure. Zinc-dependent ordered protein structure was also supported by microcalorimetry (Fig. 6), where the peak of heat absorbance was only observed in the presence of zinc. Nevertheless, the ordered ASR1 dimers have a rather noncompact structure as determined by DLS (Table II). The free concentration of Zn 21 in the reaction mixes is much smaller than that of the added ZnCl 2 probably due to sequestering of zinc ions by the buffer used (Dawson et al., 1986). The ability of bound zinc to affect more structural order to the ASR1 polypeptide monomer, apart from promoting homodimerization, is supported by the recent findings of Rom et al. (2006), where the presence of zinc reduced the sensitivity of ASR1 to protease digestion. Rom et al. (2006) also reported other zinc-binding phenomena relevance to ASR1-DNA interactions. The importance of zinc in mediating protein structure pertinent to its function has been further strengthened by studies on human prothymosin a, a protein characterized as natively un-folded (Uversky et al., 2000b) and also by zinc-driven folding and oligomerization of g-carbonic anhydrase (Simler et al., 2004). Folding of apo-metalloprotein has been reported to be induced by the binding of bivalent ions, including zinc (Ejnik et al., 2002), and the folding and stability of the nuclear hormone receptor DNAbinding domain was shown to be zinc dependent (Low et al., 2002). Moreover, the folding of a single zinc-finger domain was observed to be dependent on binding of zinc ion (Frankel et al., 1987;Parraga et al., 1988). The concentration of free zinc in the cells of biological organisms is not known (Outten and O'Halloran, 2001;Rutherford and Bird, 2004) and is estimated to be in the picomolar range. This is lower than the dissociation constant of zinc binding to ASR1 (Kalifa et al., 2004a;Rom et al., 2006). Thus, the extent of zinc binding is expected to be dependent on the free zinc concentration.
We found that desiccation of an ASR1 solution also induced increase of structure of the protein (Fig. 5;   Figure 6. DSC analysis of ASR1. Heat absorbance of a solution containing purified ASR1 (0.2 mg/mL) in 20 mM HEPES-NaOH, pH 7.5 buffer without (dashed line) or with 0.5 mM (solid line) ZnCl 2 was analyzed using MicroCal VP-DSC micro calorimeter.  Table III). Adding zinc had no further effect on the level of the obtained degree of order. In similar observations, the D-7LEA protein purified from T. latifolia pollen was shown to be highly unfolded in solution (Wolkers et al., 2001), but upon drying, the protein conformation exhibited a largely a-helix structure. Moreover, the nematode LEA protein homolog AavLEA1 was shown to be an unfolded protein in solution (Goyal et al., 2003), whereas desiccation of the protein induced folding and increased structure. We detected high levels of ASR1 in fully developed tomato seeds and in pollen grains (Fig. 8). The lily ASR1 ortholog LLA23 was isolated from desiccating pollen grains (Wang et al., 1996(Wang et al., , 1998. The developmental processes of both seed and pollen maturation involve desiccation. These results suggest that order induced by desiccation might be a general mechanism of action of dehydrin proteins that are involved in water-stress response and tolerance. Analytical ultracentrifugation showed that in the absence of zinc, ASR1 is a monomer (Fig. 1A). However, in the presence of zinc ions, the protein becomes homodimeric (Fig. 1B), suggesting that the increased order in the primary structure of the protein observed by FTIR (Fig. 5) promotes the dimerization of ASR1. ASR1 dimerization was also shown by chemical crosslinking (Fig. 2). Even the purified amino terminal domain (residues 1-60) of ASR1 formed dimers (Fig.  2). This truncated ASR1 portion bound two zinc ions (Rom et al., 2006). The zinc-dependent dimerization is also supported by the increase of the hydrodynamic radius measured using DLS (Table II). The dimerization was also observed by chemical cross-linking even in the absence of added zinc (Figs. 2). Although analytical ultracentrifugation experiments were carried out using higher protein concentrations than those used in the chemical cross-linking experiment, residual zinc contaminations might be the basis of dimerization under low protein concentrations. Indeed, low amounts of tightly bound zinc were detected in ASR1 preparations even in the absence of added zinc (Rom et al., 2006). Furthermore, it is known that chemical crosslinkers can be used to trap transient association between polypeptides (Melcher and Xu, 2001), suggesting that zinc stabilizes the interactions between the ASR1 monomers, which might associate only weakly in its absence. Zinc-dependent oligomerization of the amino terminal portion of ASR1 protein was also observed by analytical ultracentrifugation (data not shown), suggesting that the tightly bound pair of zinc ions (Rom et al., 2006) encourages homodimer formation.
Dimerization of protein disulfide isomerase is a zinc-dependent process (Solovyov and Gilbert, 2004). Furthermore, cadmium ions were shown to induce the folding and dimerization of a designed metalloprotein (Kharenko and Ogawa, 2004), and trimerization of an Ala-containing peptide was shown to be mediated by zinc ions (Liu et al., 2003). A unique zinc-binding site was demonstrated in the x-ray structure of homotrimeric Apo2L/TRAIL protein (Hymowitz et al., 2000). This homotrimer contains a single zinc ion buried at the trimer interface in a charge-shared coordination between the three monomers.  With regards to the observation of ASR1's zincdependent DNA-binding capability (Kalifa et al., 2004a), a dimeric protein structure is a commonly observed motif with DNA-binding proteins in general (see Burley and Kamada, 2002). For example, the folding of intrinsically disordered C terminus of a nucleoprotein from measles virus depends on this protein's binding to a second protein (Bourhis et al., 2004). In Kalifa et al. (2004a), certain oligonucleotides selected as binding partners for ASR1 in vitro contained two copies of the ASR1 consensus DNA-binding domain, supporting our experimental observation that active ASR1 is homodimeric.

CONCLUSION
Apo-ASR1 protein is intrinsically disordered (Fig.  9A). ASR1 gain order upon binding of zinc, most likely to the two zinc-binding sites in the N-terminal domain ( Fig. 9B; Rom et al., 2006) or possibly upon desiccation. Zinc is also involved in the DNA-binding activity of ASR1, possibly at another binding site(s) localized at the central part of the polypeptide ( Fig. 9C; Rom et al., 2006). However, these ASR1 structural results also imply that prediction methods for disordered structures of proteins are too simplistic, because they do not take account of ligand binding to other domains or protein-protein interactions. The ASR1 protein was shown to be localized both in the cytosol and nucleus of tomato leaf cells (Kalifa et al., 2004a). We propose that the nuclear-located DNA-bound ASR1 is ordered, whereas the cytosol pool of the protein might be at either structure. The secondary and tertiary structures of ASR1 protein as a function of its location in each subcellular compartment have yet to be determined. Soluble tomato ASR1 is mainly unordered, because it is highly susceptible to protease degradation (Fig. 7). Upon zinc binding, the soluble ASR1 becomes more resistant to protease, indicating that a structural change has occurred. Zinc binding was shown to decrease the protease sensitivity of purified recombinant ASR1 (Rom et al., 2006). On the other hand, nuclear-located DNA-bound ASR1 is ordered, because DNA binding is dependent on zinc (Kalifa et al., 2004a). Alternating between unfolded and ordered structures may comprise a means of regulating activity of the ASR1 protein through the linkage between desiccation and zinc binding. We propose that the DNA-bound ASR1 is folded, because DNA binding is zinc dependent, whereas the cytosol pool comprised of unfolded protein (Fig. 7). We cannot rule out that a fraction of the cytosolic ASR1 protein is ordered.

Plant Materials
Tomato (Solanum lycopersicum) seedlings were grown in vermiculite, as previously described (Kalifa et al., 2004b). One-month-old seedlings were watered with fertilizer solution without or with added 0.15 M NaCl, at a volume 5 times larger than pot volume. Leaves were collected 2 d later. Fully developed seeds were collected from well-ripened tomato fruit, washed thoroughly to remove loculus tissue, and blotted dry on filter paper. Acid soluble proteins were extracted, as previously described (Kalifa et al., 2004b). Pollen grains were collected from greenhouse-grown tomato plants. A 23 SDS-PAGE sample buffer (Laemmli, 1970) was added (5 mL/g pollen). Samples were heated for 30 min at 70°C and centrifuged for 15 min at 12,000g at room temperature.

Protease Sensitivity Assay
Leaves of salt-treated tomato plants were homogenized in ice-cold 50 mM HEPES-NaOH, pH 7.5 buffer, using 5 10-s bursts of KINEMATICA POLYTRON (Brinkmann Instruments) homogenizer. The homogenate was filtered thorough three layers of Miracloth (Calbiochem) and centrifuged for 15 min at 12,000g at 4°C. The supernatant was divided into two portions. ZnCl 2 was added to one portion to a final volume of 0.6 mM. Then 0.1 aliquots were incubated at 37°C for 30 min in the presence of the indicated amounts of trypsin. Seventy microliters 53 SDS-PAGE sample buffer was added, and mixes were heated at 70°C for 30 min and loaded immediately onto polyacrylamide gels.

ASR1 Expression and Purification
Full-length nontagged ASR1 protein was expressed in Escherichia coli and purified to homogeneity by metal chelating chromatography, as described (Kalifa et al., 2004a;Rom et al., 2006).

Protein Electrophoresis and Western-Blot Analysis
Denaturating PAGE (15% [w/v] SDS-PAGE) was run in high concentrations of Tris for improved resolution of polypeptides with low M r (Fling and Gregerson, 1986). Gels were stained with Coomassie Blue or electroblotted onto nitrocellulose membranes. ASR1 protein was detected using the previously described anti-ASR1 antiserum (Amitai-Zeigerson et al., 1995).

Gel Filtration Chromatography
Purified ASR1 and the protein standard mixture were loaded onto a HiLoad 16/60 Superdex 200 column (Pharmacia). The column was preequilibrated and eluted with buffer containing 50 mM Tris-HCl, pH 7.0, and 100 mM NaCl at a flow rate of 0.5 mL/min. The elution volume (V e ) was monitored by A 280 . The V e for a particular molecular species was then converted to K av by the following equation: where V o and V t are exclusion and total volumes, taken as the elution volumes of dextran blue and AMP, respectively. R ST was estimated using a linear calibration plot of R ST versus (2log K av ) 1/2 (Siegel and Monty, 1966). The following standard proteins were used: bovine serum albumin (R ST 5 35.5 Å ), ovalbumin (R ST 5 30.5 Å ), and lysozyme (R ST 5 19.8 Å ).

DLS
Lyophilized samples of ASR1 were dissolved in 20 mM Tris-HCl, pH 7.5, 100 mM NaCl buffer to a final concentration of 1 mg/mL. A 100-mM ZnCl 2 solution was added to aliquots of both solutions to the final concentration of 1 mM. All samples were filtered through 0.2-mm filters and centrifuged at 16,000g for 10 min immediately before measurements. The measurements were carried out using DynaPro-801 DLS instrument (Protein Solutions). The wavelength of the incident light was 780 nm. The autocorrelation function of the scattered light intensity is used to calculate the diffusion coefficient DT that is converted to the hydrodynamic radius using the Stokes-Einstein equation: where kB is the Boltzman's constant, T is the abolute temperature, and h is the solvent viscosity.

CD
Purified ASR1 protein was dialyzed against 20 mM HEPES-NaOH, pH 7.5. Mixtures contained ASR1 protein in 20 mM HEPES-NaOH, 20 mM NaCl, and the indicated concentrations of ZnCl 2 . pH values of the resulting solutions were readjusted to 7.5. CD spectra were recorded at room temperature using a Jasco Circular Dichroism Spectroscope (model J-715).

DSC
DSC studies were performed using a MicroCal VP-DSC micro calorimeter. Heat absorbance by purified ASR1 (0.2 mg/mL) in 20 mM HEPES-NaOH, pH 7.5, in the presence of the indicated concentrations of ZnCl 2 was scanned from 4°C to 90°C at a scanning rate of 0.5°C/min. The pH of samples containing zinc was corrected to 7.5.

FTIR Spectroscopy
Purified ASR1 was dialyzed against 20 mM HEPES-NaOH at pH 7.0, 7.5, or 8.0, freeze dried, and redissolved in D 2 O at one-third of the original volume. ZnCl 2 was added to specific samples. Samples were placed on the surface of ZnSe crystal, and spectra were recorded using a Bruker Equinox 55 spectrometer (Bruker Optics). Protein drying experiments were done by air drying the D 2 O-dissolved sample on spotted on the surface of ZnSe crystal at room temperature. The spectra shown are average values of triplicate runs, each composed of 120 measurements. Parallel preparations showed that addition of zinc to the protein samples reduced the pH by less than 0.5 pH units. Second derivative analyses of spectra and peak fitting were performed using PeakFit software (Systat Software).

Analytical Ultracentrifugation
ASR1 (0.2 mM) was dialyzed extensively against a buffer of 20 mM NaP i , pH 7.2, 0.1 M NaCl. The sample was divided in half and ZnCl 2 (0.1 M) was added to one portion to a final concentration of 1 mM. Sedimentation equilibrium analysis was conducted at 4°C with a Beckman Optima XL-A analytical ultracentrifuge. Samples (160 mL) were studied at different rotor speeds: 10,000 to 18,000 rpm. Data were acquired as an average of eight absorbance measurements at 280 nm and a radial spacing of 0.001 cm. Equilibrium was achieved within 24 h. Due to the presence of a small amount of aggregated protein, data collected at three different rotor speeds were analyzed simultaneously in terms of two noninteracting ideal solutes using SigmaPlot 8.0 (SPSS). Simultaneous, weighted, nonlinear least-squares fitting of the data sets at each loading concentration was performed using a mathematical model of the following form: A r 5 A 0;1 exp½HM 1 ð1 2 v 1 rÞðr 2 2 r 2 o Þ 1 A 0;2 exp½HM 2 ð1 2 v 2 rÞðr 2 2 r 2 o Þ 1 E; where A 0,1 and A 0,2 are the absorbance of species 1and 2, respectively, at a reference point r o , A r is the absorbance at a given radial position r, H represents v 2 /2RT, v is the angular speed in rads 21 , R is the gas constant, T is the absolute temperature, and E a small baseline correction determined experimentally by overspeeding. Residuals were calculated, and a random distribution of the residuals around zero (60.02) was obtained as a function of the radius. Values of the smaller mass, M 1 , were obtained from the buoyant molecular mass, given as M 1 (1 2 v 1 r), and calculated using densities, r, at 4°C obtained from standard tables. A value of v 1 of 0.7360 mL g 21 was calculated for ASR1 based on the amino acid composition using consensus data for the partial specific molar volumes of amino acids (Perkins, 1986). Molecular masses, M 2 , of the larger species or aggregate were calculated in a similar fashion.
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number L08255.

Supplemental Data
The following materials are available in the online version of this article.