Thermal, Chemical and pH Induced Unfolding of Turmeric Root Lectin: Modes of Denaturation

Curcuma longa rhizome lectin, of non-seed origin having antifungal, antibacterial and α-glucosidase inhibitory activities, forms a homodimer with high thermal stability as well as acid tolerance. Size exclusion chromatography and dynamic light scattering show it to be a dimer at pH 7, but it converts to a monomer near pH 2. Circular dichroism spectra and fluorescence emission maxima are virtually indistinguishable from pH 7 to 2, indicating secondary and tertiary structures remain the same in dimer and monomer within experimental error. The tryptophan environment as probed by acrylamide quenching data yielded very similar data at pH 2 and pH 7, implying very similar folding for monomer and dimer. Differential scanning calorimetry shows a transition at 350.3 K for dimer and at 327.0 K for monomer. Thermal unfolding and chemical unfolding induced by guanidinium chloride for dimer are both reversible and can be described by two-state models. The temperatures and the denaturant concentrations at which one-half of the protein molecules are unfolded, are protein concentration-dependent for dimer but protein concentration-independent for monomer. The free energy of unfolding at 298 K was found to be 5.23 Kcal mol−1 and 14.90 Kcal mol−1 for the monomer and dimer respectively. The value of change in excess heat capacity upon protein denaturation (ΔCp) is 3.42 Kcal mol−1 K−1 for dimer. The small ΔCp for unfolding of CLA reflects a buried hydrophobic core in the folded dimeric protein. These unfolding experiments, temperature dependent circular dichroism and dynamic light scattering for the dimer at pH 7 indicate its higher stability than for the monomer at pH 2. This difference in stability of dimeric and monomeric forms highlights the contribution of inter-subunit interactions in the former.


Introduction
Lectins are a special group of sugar-binding proteins of non immune origin able to agglutinate cells and/or precipitate glucoconjugates. They have been identified in nature from microorganisms to animals and function as cognate receptors to various cell surface glycoproteins, resulting in several important cellular mediated events ranging from mitogenic processes to plant defense mechanisms [1,2]. In recent years, plant lectins have become of increasing interest, mainly due to the discovery of their potent biological activities, and different lectins have been purified and characterized in some detail with respect to their biochemical properties and carbohydrate binding specificities [3,4]. In addition the N-acetyl-D-glucosamine (GlcNAc) and mannose specific lectins generally manifest potent insulin like activities [5]. It has also been reported that the anti-insect activity as well as binding and phagocytosis of microorganisms in animals has generally been observed in the case of GlcNAc and mannose specific lectins [6,7,8,9]. Consequently, to aid in the understanding of their function and in designing mimics, there is also an increasing interest in the molecular structure of plant lectins in relation to their functions [10].
The vast majority of all currently known plant lectins can be classified into seven families of evolutionary and structurally related proteins [11]. The lectin families are distinguished namely (i) the legume lectins [12], (ii) the chitin-binding proteins containing hevein domain(s) [13], (iii) the type 2 ribosomeinactivating proteins [14], (iv) the monocot mannose-binding lectins [15], (v) the Cucurbitaceae phloem lectins, (vi) the jacalin family and (vii) the Amaranthaceae lectins. The first lectin structures to be determined derived from two phylogenetically conserved families-the Leguminoseae [12] and the Gramineae [13]. The legume lectins display diverse sugar-binding specificities [16] and differences in their quaternary structures; however, they share a common polypeptide fold known as the 13-sandwich (or jelly roll) motif, consisting of three sets of antiparallel b-sheets connected by loops of diverse size [17]. The Gramineae family lectins belonging to a super family of chitin-binding proteins, have also been studied extensively [13]. These dimeric proteins have a most distinctive subunit fold that consists of four independently folded and helically assembled 43-residue segments. These lectins have evolved by gene duplication of an ancestral disulfide-rich domain [18] lacking significant amounts of regular secondary structure [19]. The first three monocot lectin structures determined were of Galanthus nivalis, family (iv) of b-prism II fold [20]; jacalin, of family (vi) of b-prism I fold [21]; and Amaranth caudatus seeds of family (vii) of b-trefoil fold [22] respectively, their folds stabilized by conserved hydrophobic side chains which line the core of the b-barrel. All three are relatively small proteins of 12-16 kD per subunit or domain, and possess subunit structures that display pseudo threefold symmetry.
Curcuma longa, commonly called turmeric, is widely used as food, flavoring, coloring, spice agent, a traditional herb and belongs to the Zingiberaceae family. A purified dimeric Curcuma longa rhizome lectin (CLA) is a glycoprotein with about 5.2% covalently bound sugar. The lectin possessing antifungal, antibacterial and a-glucosidase inhibitory activities also showed apparent thermal stability for incubating temperatures below 333 K, as well as acid tolerance till pH 2, by measuring agglutination activity [23]. Despite this study [23] on the turmeric lectin, the molecular basis for its stability and assembly remains unknown. The relationship between protein stability and oligomerization of turmeric lectin (Zingiberaceae family) has been highlighted for the first time in our study.
Lectin unfolding was studied by differential scanning calorimetry (DSC) at different pH values, protein concentrations and scan rates. The oligomeric state of the lectin at different pH values were observed by size-exclusion chromatography and dynamic light scattering (DLS). The effect of pH on secondary and tertiary structures of the lectin was investigated by circular dichroism (CD) spectroscopy. Chemical unfolding of turmeric lectin by GdnCl was monitored by changes in the fluorescence characteristics and CD spectra of the lectin at different temperatures.

Materials
Ultrapure guanidinium chloride (GdnCl) and ammonium sulfate were purchased from Calbiochem. Concanavalin A sepharose, HiTrap Q HP column, Superdex 75 pg column and gel filtration molecular mass standards were purchased from GE Healthcare Biosciences Ltd. Biogel P-100 was purchased from BioRad. A protein molecular mass marker set was purchased from Fermentas.

Protein purification
Fresh rhizomes of C. longa were purchased from a local market, were grounded after peeling and cut into small pieces adding 20 mM Tris-HCl, 1 mM CaCl 2 , 1 mM MnCl 2 , 150 mM NaCl (pH 7.4) 1:5 (w/v) using a homogenizer. The extraction and purification procedure using Concanavalin A sepharose column was otherwise similar to the published procedure [23]. Final elution of the protein was carried out using 0.5 M methyl-a-Dglucopyranoside in the same buffer. Fractions eluted from Concanavalin A sepharose column having hemagglutination activity were pooled and dialyzed against 20 mM Tris-HCl (pH7.2). For measuring this activity, rabbit erythrocytes were washed 5 times with PBS (phosphate buffer saline, 10 mM potassium phosphate buffer pH 7.2 containing 150 mM NaCl). A 2% (v/v) suspension of the erythrocytes in the same buffer was prepared and assays performed [24]. The dialyzed sample was then loaded on a Hitrap Q HP column coupled to the AKTA prime plus (GE Healthcare) system, pre equilibrated with 20 mM Tris-HCl (pH 7.2). The protein was then eluted using a linear gradient of 0 to 0.35 M NaCl with a flow rate of 1 ml min 21 . The active fractions were collected and further dialyzed against 50 mM PBS. The dialyzed protein was then concentrated using Centriprep-10 (Millipore) and loaded on Superdex 75 pg column pre equilibrated with 0.2 M methyl-a-D-glucopyranoside in 50 mM PBS at a flow rate of 1 ml min 21 . The active fractions were collected and analyzed by SDS-PAGE.

Amino acid analysis and protein concentration determination
The amino acid compositions of experimental samples were determined using the HPLC-Pico-Tag method. Protein concentrations were determined spectrophotometrically using an extinction coefficient of 13,980 M 21 cm 21 at 280 nm.

pH treatment of CLA
To study the effect of pH on the secondary structure of CLA, the following buffers were used having 150 mM NaCl: 10 mM Glycine-HCl (pH 1-3), 10 mM Na-acetate (pH 4-5), 10 mM Naphosphate (pH 6-7). Protein was incubated for 5 h at room temperature at different pH values before recording various spectroscopic measurements, size exclusion chromatography, dynamic light scattering or differential scanning calorimetry as described below.

Size-exclusion chromatography
Size-exclusion chromatography was performed to determine the oligomeric state of CLA at different pH values using a Biogel p-100 column. The column volume 50 ml was determined by vitamin B 12 and void volume (V 0 ) of 16 ml by blue dextran. CLA was incubated in various buffers as already described. All standards (Conalbumin, 75 kD; Ovalbumin, 43 kD; Chymotripsinogen, 25 kD; Ribonuclease A, 13.7 kD) were run at pH 7. The standard curve was drawn by plotting the ratio of elution volume to void volume (V e /V 0 ) against logarithm of relative molecular mass.

Dynamic light scattering
DLS measurements were carried out at 830 nm with a Malvern Nano instrument equipped with a temperature-controlled micro sampler, using a protein concentration of 37.5 mM. Before measurement, all the solutions were spun at 12,000 rpm for 15 min and filtered through 0.22 mm filter. Protein was incubated at different pH values of 2 to 7. Data were analysed by using Dynamics 6.10.0.10 software at optimized resolution. The mean hydrodynamic radius (R h ) and polydispersity (Pd) were estimated on the basis of an autocorrelation analysis of scattered light intensity based on the translational diffusion coefficient, by Stokes-Einstein equation:

D~kT=6pgR h ð1Þ
Where R h is the hydrodynamic radius, k is the Boltzmann's constant, T is the absolute temperature, g is the viscosity of solvent and D is the translational diffusion coefficient. Hydrodynamic radius (R h ) as a function of temperature was determined by dynamic light scattering for pH 7 and pH 2.

Circular Dichroism Spectroscopy
Circular dichroism measurements were carried out at 298 K temperature with a J-800 JASCO spectrometer using a quartz cell of 1 mm path length. For far-UV analysis the spectra were recorded in the range of wavelengths 260-210 nm at a scan speed of 100 nm min 21 with a response time of 1s and slit width 1 nm. The lectin concentration used was 10 mM for all the samples. Each spectrum, including that measured for buffer alone for the purpose of baseline determination, was obtained as an average of 3 scanned ones. The spectra presented are after baseline subtraction.
Temperature-induced denaturation of the CLA secondary structures were monitored by following the loss of ellipticity at 227 nm, using CD spectroscopy. The temperature at midpoint of transition (T m ) was determined by plotting 0.5 unit of fraction native. The effect of temperature on the secondary structure of CLA was studied at pH 7 and 2 by increasing the temperature of the protein samples at the rate of 1 K min 21 in the temperature range of 298-368 K with the interval of 2 K. All data were corrected for respective buffer base line. Reversibility of protein unfolding was checked by return of the complete spectrum upon thermal unfolding followed by subsequent cooling to ambient temperature.
To study the tertiary structure of CLA at pH 7 and 2, the near-UV spectra were recorded at a protein concentration of 30 mM in the range of 250-300 nm using a cuvette of path length 5 mm.

Steady state Fluorescence Spectroscopy
Intrinsic fluorescence of the proteins were measured using a Hitachi F3010 spectrofluorimeter at a scanning speed of 240 nm per min in 1 cm path length quartz cuvette having slit widths of 2.5 and 5 nm for excitation and emission, respectively. The lectin solution (5 mM) was excited at 295 nm and emission spectra were recorded between 305-400 nm. To observe the effect of pH on fluorescence maximum, CLA was incubated at different pH values from 1 to 7 in the respective buffers for 5 h. Different spectra were analyzed by subtracting buffer spectra from protein spectra yielding the corrected spectra shown.
Fluorescence quenching measurements were carried out at pH 7 and pH 2 by adding small aliquots of acrylamide (5 M) by titrating to the protein (5 mM) samples. Protein samples were excited at 295 nm and emission spectra were recorded at 305-400 nm. The fluorescence intensities obtained were corrected for dilution. Quenching data were analyzed using the Sterne-Volmer equation:  where F 0 and F are the emission intensities in the absence and presence of the quencher, [Q] is the molar concentration of the quencher and K SV is the Sterne-Volmer quenching constant. For the ANS binding experiments, a stock of ANS was prepared in distilled water and protein to ANS molar ratio was maintained as 1 : 10. The excitation wavelength was set at 375 nm and emission spectrum was recorded at 430 to 550 nm.

GdnCl induced unfolding of CLA and its analysis
Isothermal GdnCl-iduced unfolding studies of native CLA were performed in the temperature range 288 K to 333 K and monitored using fluorescence spectroscopy and far-UV circular dichroism. GdnCl stock solutions, 8 M, were freshly prepared either in 10 mM Na-Phosphate (pH 7) or 10 mM glycine-HCl (pH 2) having 150 mM NaCl. CLA samples were brought to 0-7 M (for pH 7) and 0-6 M (for pH 2) in GdnCl with these buffers. Fluorescence and CD measurements were done under same experimental conditions as stated earlier, using 5 mM protein concentrations. For concentration dependent study 2 mM and 10 mM protein sample were used at the desire temperature.
Assuming that unfolding of CLA follows a two-state model, the fraction of unfolded protein molecules (f u ) was calculated from the following equation: Where X, X n , and X u represent the observed spectroscopic signal of the protein at any denaturant concentration, the spectroscopic signal of the protein in the completely folded state, and the spectroscopic signal of the protein in the completely unfolded state, respectively. C m (denaturant concentration at the midpoint of unfolding transition) values of the two-state pathways was determined by nonlinear fitting of the unfolding data to eq 4 using GraphPad Prism (GraphPad Software Inc.) as described [25] Where X and Y represents the concentration of unfolding agent and the fraction of unfolded protein molecules respectively.
The change in free energy of unfolding in water (DGu) is obtained by linear extrapolation model [26]. The relationship between the denaturant and DGu is approximated by the following equation: and Where m is the experimental measure dependence of DG; DGu is the free energy change in absence of denaturant, its concentration denoted by [d]. R is the gas constant (1.987 cal K 21 mol 21 ) and T is 298 K.
After obtaining DGu at each temperature, the values were fitted to the following equation to obtain DH g , DC p and T g [27] DG(T)~DH g (1-T=T g )zDC p ½T-T g -T ln(T=T g ) ð7Þ Differential scanning calorimetry DSC experiments were performed on a Microcal VP-DSC micro calorimeter equipped with two fixed cells, a reference cell and a sample cell. DSC experiments were carried out as a function of pH, protein concentration and scan rate. For DSC experiments, 40 mM protein solution was used, varying the pH from 2 to 7 using buffers described. Prior to all the measurements the buffers and protein solutions were degassed and protein was dialyzed against respective buffers overnight. The DSC scans were run between 298 K and 373 K. Respective reference scan was run under identical DSC setup conditions and subtracted from each sample scan.
The van't Hoff enthalpy (DH v) associated with the DSC thermograms were determined by the two-state fit of the thermogram using the Origin software supplied by MicroCal assuming N 2 «2U type transition, where N represents the native monomer and U the unfolded monomer, while the calorimetric enthalpy (DH c ) was determined by the area under the transition. T m and T p are the temperatures of half of the transition peak area and the temperature at which the transition peak is at its maximum, respectively.
The dependence of T p upon protein concentration for an oligomeric protein undergoing dissociation can be expressed by following equation [28] ln½CLA~- Where n is the number of subunits per oligomer, T p is the temperature corresponding to the transition peak maximum, [CLA] concentration of the protein and DH v (S) is the van't Hoff enthalpy obtained by the slope of ln[CLA] vs 1/T p .

Protein purification
Following extraction from fresh rhizomes, ammonium sulfate precipitation, the precipitate containing the lectin was purified as described and judged to be extremely pure (.98%) (Fig. 1).

Amino acid analysis and protein concentration determination
The results of our amino acid analysis are summarized in Table 1. The protein seems to be rich in Asx, Glx, Gly, Ile and Leu, but no Cys residue was found. The number of Trp residues cannot be found by the pico-Tag method as it gets destroyed by the digestion protocol using high temperature. Therefore it was determined by an old spectroscopic method by measuring protein spectra in 0.1 N NaOH at 294.4 nm and 280 nm [29]. The calculated molecular mass of deglycosilated CLA is 15,021.28 Daltons, whereas the molecular mass determined by MALDI-TOF mass spectrometry of deglycosilated protein gave a molecular mass of 14,947.0 Daltons (data not shown). The extinction coefficient was calculated to be 13,980 M 21 cm 21 at 280 nm using the number of Trp and Tyr residues shown in Table 1.

Size-exclusion chromatography
Gel-filtration studies using a Bio-rad P-100 column were performed in the pH range 2-7 ( Fig. 2A). There was an appreciable change in the position of the CLA peak at pH 2 relative to higher pH measurements, and elution volume increased progressively from 22.44 ml to 30.63 ml in the range of pH 7-2. From a standard curve, the molecular mass was calculated to be 16 kD for that peak eluting ,31 ml. It is noticed that for the experiments at pH 4, 5 and 7, the maxima of the peak is ,22 ml ( Fig. 2A). At intermediate pH values 3 and 2.5, the presence of two peaks eluting ,31 ml and ,22 ml are noticed. After lowering pH to 2, only the 16 kD peak eluting ,31 ml was observed. This implies, upon lowering pH from 7 to 2, CLA converts from dimer to monomer.

Dynamic light scattering
Dimerization in CLA can be detected by DLS measurements where an increase of hydrodynamic radii (R h ) as well as elevation in percent polydispersity takes place. Fig. 2B shows the plot of hydrodynamic radius against pH for CLA. Under these conditions, polydispersity varies from 10% to 22% from pH 2 to pH 7. It was found that CLA has a hydrodynamic radius of 3.160.2 nm in the pH range 4 to 7. The size gradually starts changing below pH 4; finally, at pH 2 it attains a size of 2.060.2 nm. The higher polydispersity (22%) and R h (3.160.2 nm) in neutral pH conditions are more favorable for oligomerization. On the other hand, the lower polydispersity (10%) and R h (2.060.2 nm) in pH 2 apparently suggest monomeric units. The CLA size of 3.1 nm in the pH range 4 to 7 indicates that the species is normally a dimer. At pH 3, there is a drop in size to 2.8 nm. Temperaturedependent DLS (Fig. 3) was performed on both forms of the protein (i.e., pH 7 and 2) from 298 to 368 K. For pH 7, in the temperature range 298-333 K, the average hydrodynamic radius is about 3.1 nm; at 368 K it reaches 3.8 nm. At pH 2, the average hydrodynamic radius is about 2.0 nm in the range 298-313 K, beyond 313 K, R h increases indicating unfolding of monomer.

Circular Dichroism Spectroscopy
An earlier report suggested that its activity peaked in the pH 2-7 range [23]. There was almost no change in secondary structure observed by specific ellipticity in the range pH 2 to 7, but as pH drops below 2, CD signals start dropping in magnitude and at pH 1, it goes down to about 40% of its maximum value at 227 nm (Fig. 4A). Secondary structure prediction by CDNN program [30] using the spectrum at pH 7 indicates CLA is a b protein containing frequent turns with 11% a-helix, 40% b-sheet and 22% turns. The spectrum at pH 2 yielded identical results by CDNN. The near UV spectra at pH 2 and pH 7 are virtually identical, with minima around 282 nm (Fig. 4E).
The temperature dependent value of ellipticity at 227 nm was plotted for CLA at pH 7 and pH 2 (Fig. 5). The melting temperature was determined to be 327.5 K for pH 2 and 351.2 K for pH 7. Hence the thermal stability of CLA is higher at pH 7 relative to that at pH 2. To check the reversibility of thermally unfolded CLA at pH 7and pH 2, protein samples were brought to 298 K and rescanned. At pH 7 it was observed to be approximately 90% reversible (Fig. 5 inset B) but at pH 2 the transition was observed to be irreversible (Fig. 5

Steady State Fluorescence Spectroscopy
The intensity and the maxima of the emission spectra are used to characterize protein molecular structures as tryptophan residues are highly sensitive to their surrounding environment. The emission l max of CLA for pH 2 to 7 is seen at 33062 nm; therefore the tryptophan environments of CLA do not change in this pH range (Fig. 4C). Moreover, the l max of the unfolded protein is close to that of free tryptophan and shows a substantial increase in fluorescence intensity. The unaltered tryptophan environment is also observed by fluorescence quenching measurements, in which the neutral acrylamide quenched about 50% of the protein intrinsic fluorescence at 0.425 M concentration (data not shown). The quenching profiles with acrylamide at pH 2 and pH 7 give similar results (Fig. 6 & Table 2). In the native state at pH 7, CLA does not bind to ANS, whereas at pH 2 substantial binding of ANS was observed (data not shown).

GdnCl induced unfolding of CLA
The denaturation profiles of dimeric and monomeric CLA are monitored by fluorescence and CD isothermal melts. The ellipticity at 227 nm was the most sensitive indicator as for different values of [GdnCl] it varied more than those at other wavelengths studied. The intrinsic fluorescence at 360 nm was used as a probe to monitor the change, as at this wavelength there was maximum difference between the intensity of the native and denatured species. The curve fitted well to a two-state model described by Eq. 4 and Eq. 6 (r-square value = 0.99), by the observation that fluorescence and CD plots can be superimposed (Fig. 7).
The C m of dimeric CLA increases at 298 K from 4.9660.06 M at 2 mM protein to 5.3860.02 M at 10 mM protein (data not shown), consistent with an N 2 «2U denaturation transition. The C m of monomeric CLA was found to be independent in the range of protein concentrations studied (2- (Table 3). From each of the isothermal melts, DGu of unfolding at each temperature for dimer (Fig. 8A, Table 4) was calculated and it can be clearly seen that the protein becomes less stable at temperatures above and below 298 K, particularly above 323 K (Fig. 8C, Table 4). The value of m, defined in Eq. 6, varies between 22.6 to 22.0 (Fig. 8D). The values of DH g , DC p and T g generated from the stability curve by fitting the values of DGu and T to Eq. 7 (Fig. 8B, Table 5).
To check the reversibility of chemical unfolding of CLA at pH 7 and pH 2, protein samples were unfolded by GdnCl, then samples were diluted 10 fold. Fig. 9 shows the refolding at two pH values, monitored by CD, fluorescence and gel filtration, respectively. At pH 2 and pH 7, reversibility was observed approximately at 70% and 90% levels respectively. The l max of the unfolded protein is significantly red-shifted at 350 nm with respect to that in the folded protein (both for monomer and dimer) (Fig. 9C, 9D).

Differential scanning calorimetry
In this study, the thermal stability of C. longa rhizome lectin has been investigated under different conditions by using highsensitivity DSC measurements. Effect of pH on the unfolding transition of CLA was studied by performing DSC scans on CLA samples in the pH range 2 to 7. However, thermodynamic parameters could only be calculated in the pH range 7-4 (Table 6)  Table 2. Acrylamide quenching constants (K sv ) of dimeric (pH 7), monomeric (pH 2) and denatured states (6 M GdnCl) of CLA. since the thermal denaturation in this range was almost 90% reversible since transition peak was also observed upon down scan of the sample. The unfolding was irreversible in the pH range 3 to 2. At pH 2, the apparent T m was found to be 327 K by two-state fitting. Hence, the irreversible nature of the thermal transitions prevented further analysis of the data. Fig. 10 shows a typical DSC thermogram of CLA corrected for buffer base line at pH 7 at a scan rate of 60 K h 21 along with the fit of the transition data to a two-state model. The enthalpy (DH v ) obtained as a two-state model fit at different pH (pH 7 to pH 4, dimeric range) scans, when plotted against the respective T m yields a slope of 3.2360.19 Kcal mol 21 K 21 (Fig. 11), is an estimate of the heat capacity (DC p ). DSC scans were performed at different scan rates. Though the T m values increase a bit with increase in scan rate (data not shown), the thermodynamic quantities DH c and DH v are independent of scan rate, hence the transitions are kinetically controlled and not thermodynamically controlled. As pH decreased from 7 to 2, T m also decreased indicating decreasing stability of the lectin. At pH 7, protein concentration had a significant effect on the T p value, i.e., oligomeric protein undergoing dissociation could be related by Eq. 8. (Fig. 12) but the T p value at pH 2 is independent of protein concentration, supporting a monomeric form.

Discussion
It is difficult to purify large amounts of the turmeric lectin, as the yield is approximately 1 mg per gram of fresh root. The primary amino acid sequence and crystal structure of CLA are not available yet, so it is important to study CLA by biophysical methods under various conditions to characterize its assembly and stability. Disruption of the various non-covalent interactions that stabilize a protein structure, e.g. by thermal, acidic pH or chemical-induced denaturation, may result in partial or complete loss of structure and its biological activity, enriching our understanding of protein folding.
For most multimeric proteins, it is not possible to find conditions in which the folded monomer is stable with respect to the denatured state. This has given rise to the belief that inter-subunit contacts make a very large contribution to the stability of multimeric proteins, and that these interactions are important for subunit folding [31]. In the case of CLA, we are able to find a stable folded monomer at pH 2 ( Figs. 2A, 2B). Hence in this case, inter-subunit interactions are not required for subunit folding. In multimeric systems, folding is typically coupled with subunit association [31]. Partially/completely folded, monomeric forms of a multimeric protein have been observed previously in a few cases. The dimeric CcdB (controller of cell division or death B protein) protein dissociates to monomer at pH 4, has similar secondary and tertiary structure, although dimeric interface is primarily hydrophobic [32]. The dimer of porcine mitochondrial malate dehydrogenase can be dissociated by relatively mild changes in pH. However, unlike in the present case, the monomeric form has different fluorescence properties. The monomeric form of the malate dehydrogenase enzyme present at pH 5 was shown by hydrogen exchange to be folded into a conformation similar to that adopted by the native dimer at pH 7 [33]. The dimeric enzyme 4-aminobutyrate aminotransferase also dissociates into monomers at pH 5 [34]. The monomers have similar secondary structure to the dimer. The E. coli enzyme glycan amide ribonucleotide transformylase (GART) also undergoes pH-depen- Table 3. Thermodynamic parameters derive from GdnCl-induced unfolding at 298 K for the stability of CLA at pH 7 and pH 2.

Condition
Unfolding  dent dimerization [35]. The soybean agglutinin that dissociates to monomeric form at pH 1.9 has similar secondary structure but shows different tertiary structure as well as fluorescence properties. The subunit association in soybean agglutinin is primarily due to ionic interactions rather than hydrophobic interactions at the interface [36]. In the present study we have been able to characterize the equilibrium stabilities of CLA, in both monomeric and dimeric forms. Far and near UV CD spectra, fluorescence emission maximum, showed no change in secondary or tertiary structure in the pH range 2 to 7 (Figs. 3A to 3E), while pH dependent DLS and size exclusion chromatography show dimer at pH range 7-4, going fully to monomers at pH 2 ( Figs. 2A, 2B). As with the CcdB (controller of cell division or death B protein) protein [32], soybean agglutinin [36] and wheat germ lectin [37],  Table 4. Free-energy changes upon unfolding for pH 7 at different temperature obtained from GdnCl-induced isothermal denaturation melts. the dissociation of the dimer into corresponding monomers occurs probably due to electrostatic repulsion among the individual subunits with the increase in positive charge in the system. A pHdependent ANS study shows there was no binding of ANS in the presence of CLA dimer at pH 7 but ANS binding was observed for monomer at pH 2 (data not shown). The change in the charge distribution over surface of the protein might cause the hydrophobic patches to cluster at acidic pH; a similar result was observed for CcdB (controller of cell division or death B protein) protein [32]. Taken together with dynamic light scattering and gel-filtration studies, the spectroscopic observations suggest that, at pH 2, CLA exists as folded monomer with very similar secondary  and tertiary structures. The CcdB (controller of cell division or death B protein) protein [32], Soybean lectin [36] and wheat germ lectin [37] show similar compactness in monomeric structure, whereas banana lectin [38,39], lentil lectin [40] form molten globule without dissociating to monomeric form. Acrylamide quenching showing similar results at pH 2 and pH 7 (Fig. 6) also support a compact monomeric structure. GdnCl-induced unfolding at pH 7 and pH 2 was also monitored by CD and fluorescence (Figs. 7A, 7B) and data analyzed by two-state models. The equilibrium unfolding of CLA in the presence of GdnCl was found to be a cooperative process in which protein molecule undergoes unfolding without formation of any partially unfolded intermediate at both pH values. By both CD and fluorescence, the C m for GdnCl-induced unfolding was higher at pH 7 than at pH 2 ( Table 3, Fig. 7). Similarly, DGu for GdnCl-induced unfolding was higher at pH 7 compared to that at pH 2 (Table 3, Fig. 7). These indicate that the native dimeric state (at pH 7) is more stable than the acid induced monomeric state (at pH 2). The C m of the denaturant-induced transition was concentration-independent at pH 2, as expected for a monomeric state of the protein. Whereas, in the dimeric state C m of the transition was highly concentration-dependent. One of the most important features of protein unfolding is the exposure of its hydrophobic core. This exposure is reflected in the positive change in the heat capacity of the protein during unfolding reaction. Thus, the change in heat capacity is directly proportional to the extent of unfolding and conversely the extent to which the apolar residues are buried in the native protein. The equilibrium thermodynamic parameters for the unfolding of dimer obtained from the fit of Eq. 7 are listed in Table 5. The small DC p for unfolding of CLA reflects a buried hydrophobic core in the folded dimeric protein.
Temperature-dependent DLS showed transitions of hydrodynamic radius starting above 313 K, showing the monomer (at pH 2) is stable up to this temperature and the dimer (at pH 7) up to 333 K (Fig. 4), correlating with thermal unfolding studied at the same temperature region by the change in CD signal (Fig. 5).
The thermal stability of CLA was also characterized as a function of pH using DSC scans. At all the pH values listed in Table 6, the thermal unfolding transitions of CLA could be fit to a two-state model as N 2 «2U at pH 7 to 4. The results presented in Fig. 2 and Table 6 reassert that CLA maintains its dimeric state over the pH range of 4-7. The correlation coefficient of 0.98 for the estimates of DC p from DH v versus T m plots (Fig. 11)  Figure 10. DSC scan of CLA. Excess heat capacity curves of 40 mM CLA at pH 7 was monitored by differential scanning calorimetric study at a scan rate of 60 K h 21 . The solid line represents the best fit of the DSC data to a two state transition model. doi:10.1371/journal.pone.0103579.g010 indicates that the DC p values are pH-independent over the selected pH range. The value of DC p obtained from this method is considered a better estimate [41] than that obtained directly as the difference between the native and unfolded baseline of a single DSC scan because of the errors associated with arbitrariness in determining the unfolded baselines [42]. According to equilibrium thermodynamics, any change in the oligomerization state of the protein during denaturation process should produce a concentration dependence of T m [28], seen by plotting ln[protein] vs 1/T p for pH 7 (Fig. 12A). The lack of concentration dependence (Fig. 12B) also proves the presence of monomeric form only at pH 2.

Conclusions
The turmeric lectin unfolds via two-state pathway in both monomeric and dimeric states. The stabilization of the dimer derives primarily from interactions along the monomer-monomer contact interface of the CLA dimer. These inter-subunit interactions are not required for subunit folding. From available crystal structures of different lectins, the sugar binding ability appears to reside within a subunit. So it is important to maintain proper folding of each subunit to maintain biological function. Oligomerization stabilizes a protein in such a manner so that they can carry out their biological function efficiently. Though the turmeric lectin in our study forms monomer under acid-induced condition, it is still capable of sugar binding as its secondary and tertiary structures remain intact as in the dimer. This property may make CLA highly suitable for lectin-mediated drug delivery [4].