Metal binding affinity and structural properties of calmodulin‐like protein 14 from Arabidopsis thaliana

Abstract

In addition to the well‐known Ca²⁺ sensor calmodulin, plants possess many calmodulin‐like proteins (CMLs) that are predicted to have specific roles in the cell. Herein, we described the biochemical and biophysical characterization of recombinant Arabidopsis thaliana CML14. We applied isothermal titration calorimetry to analyze the energetics of Ca²⁺ and Mg²⁺ binding to CML14, and nuclear magnetic resonance spectroscopy, together with intrinsic and ANS‐based fluorescence, to evaluate the structural effects of metal binding and metal‐induced conformational changes. Furthermore, differential scanning calorimetry and limited proteolysis were used to characterize protein thermal and local stability. Our data demonstrate that CML14 binds one Ca²⁺ ion with micromolar affinity (K _d ∼ 12 µM) and the presence of 10 mM Mg²⁺ decreases the Ca²⁺ affinity by ∼5‐fold. Although binding of Ca²⁺ to CML14 increases protein stability, it does not result in a more hydrophobic protein surface and does not induce the large conformational rearrangement typical of Ca²⁺ sensors, but causes only localized structural changes in the unique functional EF‐hand. Our data, together with a molecular modelling prediction, provide interesting insights into the biochemical properties of Arabidopsis CML14 and may be useful to direct additional studies aimed at understanding its physiological role.

Abbreviations

AMM

apparent molecular mass

ANS

8‐Anilino1‐naphthalene‐sulfonic acid

Ca²⁺

calcium

CaM

calmodulin

CML

calmodulin‐like protein

EDTA

Ethylenediaminetetraacetic acid

EGTA

Ethylene glycol‐bis(2‐aminoethylether)‐N,N,N′,N′‐tetraacetic acid

NMR

nuclear magnetic resonance

HSQC

heteronuclear single‐quantum coherence

TOCSY

total correlation spectroscopy

IPTG

Isopropyl β‐d‐thiogalactoside

ITC

isothermal titration calorimetry

DSC

differential scanning calorimetry

R_s

Stokes radius

SEC

size exclusion chromatography.

The full biophysical and structural information acquired in this work on calmodulin‐like protein 14 (CML14) from Arabidopsis thaliana demonstrates that CML14 binds only one Ca²⁺ ion without undergoing significant conformational rearrangement. This information addresses the need to obtain a more in‐depth understanding of the biochemical and structural properties of plant EF‐hand proteins in order to determine whether they function as Ca²⁺ sensors.

Introduction

Calcium (Ca²⁺) ion is a crucial second messenger in plants that regulates various signaling pathways through transient changes in its intracellular concentration. Such Ca²⁺ variations, that differ in their spatio‐temporal patterns and therefore are considered as distinct Ca²⁺ signatures, are then converted into a wide variety of biochemical changes by Ca²⁺ sensors.1

The common structural feature of most proteins that serve as Ca²⁺ sensors is the presence of a special Ca²⁺‐binding motif, called the EF‐hand, consisting of a conserved helix‐loop‐helix structure that can bind a single Ca²⁺ ion. In the canonical EF‐hand, Ca²⁺ ion is bound through a pentagonal bipyramidal arrangement to residues of the loop region at positions 1, 3, 5, 7, 9, and 12. However, different non‐canonical EF‐loops are known to bind Ca²⁺ through different coordination mechanisms. In approximately 10% of known EF‐hands, for example, the loop has the same length as the canonical sequence but the presence of aspartate at position 12 is reported to decrease the binding selectivity of Ca²⁺ versus the chemically similar but smaller Mg²⁺ cation.2

In plants, many different Ca²⁺ sensor proteins have evolved. Besides calmodulins (CaMs), plants exclusively possess a group of calmodulin‐like proteins (CMLs) that are predicted to have specific functions in the cell and are characterized as sharing at least 16% identity with mammalian CaMs and possessing no other known functional motifs except the EF‐hands.1, 3 In contrast to CaMs, CMLs considerably vary in the length and number of EF‐hands (from 1 to 6), which might enable them to respond in different ways to Ca²⁺ signals.1, 3 Sequence alignments also show variations in the EF‐hand loop motif, which may confer unique ion‐ or target‐binding abilities. Some CMLs contain variable N‐ or C‐terminal extensions that could constitute targeting sequences for organelle sorting, or motifs for myristoylation.3 Growing research on CMLs functional roles has revealed their involvement in different developmental processes, as well as in the response to stresses.4, 5, 6, 7, 8, 9, 10 This information, together with the presence and diversification of CMLs across plant taxa, allowed speculation that CMLs have distinct functions and that their evolution is related to the necessity for more specialized Ca²⁺ sensors in the colonization of new terrestrial environment.11 Notwithstanding, our current knowledge of CML‐mediated signaling in plants remains incomplete, and further investigation is needed to clarify that these proteins may act as true Ca²⁺ sensors through better definition of their Ca²⁺ binding and structural properties, and downstream targets.

Herein, we focus on CML14, an acidic protein with 148 residues and with a molecular mass of 16,579 Da, which shares 50% identity in amino acid sequence with Arabidopsis CaM2 (a highly conserved CaM). CML14 possesses three EF‐hand motifs, but only one is predicted to be a functional Ca²⁺ binding site according to PROSITE (ProRule:PRU00448).12

In this work, we produced the recombinant protein CML14 with the aim to evaluate its metal (Ca²⁺ and Mg²⁺) binding properties and to investigate the effect of metal ions on conformation and stability of the protein. These analyses, together with a molecular model of CML14, intend to clarify whether CML14 possesses the features of the typical Ca²⁺‐sensors.

Results

Isothermal titration calorimetry analysis of metal binding affinities

ITC was used to assess the metal binding properties of CML14. Representative ITC curves and the derived binding isotherms for the binding of Ca²⁺ to apo and Mg²⁺‐bound CML14 are shown in Figure 1, and a summary of the thermodynamic parameters obtained using a one‐site binding fitting model is listed in Table 1. Titration of CaCl₂ into apo CML14 results in a binding isotherm that clearly indicates the presence of only one Ca²⁺ binding site [Fig. 1(A)] with a dissociation constant ( $K_{\mathrm{d},\text{Ca}2+}$) of ∼12 μM. The binding of Ca²⁺ appeared to be an exothermic process with favorable enthalpy (ΔH, −13 kcal/mol) and overall unfavorable change in entropy (ΔS, ∼ −21 cal mol⁻¹ K⁻¹).

Figure 1.

ITC analysis of the Ca²⁺ binding to apo (panel A) and Mg²⁺‐bound (panel B) CML14. Baseline corrected raw ITC titrations (top) and the derived binding isotherm (bottom) in the absence (A) and presence (B) of 10 mM MgCl₂.

Table 1.

Thermodynamic Parameters for the Binding of Ca²⁺ to CML14

	N	K a (103 M−1)	ΔH (kcal mol−1)	ΔS (cal mol−1 K−1)	ΔG° (kcal mol−1)
CML14	0.97 ± 0.05	84.1 ± 7	−12.8 ± 2	−20.5 ± 5	−6.7 ± 0.1
+ 10 mM Mg2+	0.67 ± 0.03	16.6 ± 3	−17.8 ± 1	−40.4 ± 3	−5.8 ± 0.1

Data were fitted using the one‐set of sites model.

The mean values from triplicate experiments and S.D. are presented.

The presence of Mg²⁺ influences the Ca²⁺ affinity of CML14 [Fig. 1(B)]. Since it was not possible to obtain direct measurement of the Mg²⁺ binding affinities because of the low heat released, the value of the Mg²⁺ dissociation constants, K _d,Mg2+ was obtained from the effect of Mg²⁺ on the Ca²⁺ affinity ( $K_{\mathrm{d},\text{Ca}2+}$ _{, app} at 10 mM MgCl₂ versus $K_{\mathrm{d},\text{Ca}2+}$ at 0 mM MgCl₂) using Eq. 4 reported in “Isothermal titration calorimetry” section.13 Mg²⁺ decreased Ca²⁺ affinity of CML14 by ∼5‐fold ( $K_{\mathrm{d},\text{Ca}2+}$ _{, app} ∼ 60 μM) and the K _d,Mg2+ was found to be ∼ 2.5 mM.

Monitoring metal binding by nuclear magnetic resonance (NMR) spectroscopy

The capability of apo CML14 to bind Ca²⁺ and Mg²⁺ ions was also assessed using NMR spectroscopy measurements. To this aim, ¹H‐¹⁵N heteronuclear single‐quantum coherence (HSQC) spectra of ¹⁵N‐CML14 were recorded before and after addition of 5 mM CaCl₂ or MgCl₂ [Fig. 2(A‐C)]. The spectra of both apo and Ca²⁺/Mg²⁺‐bound CML14 showed significant signal dispersion in both the proton (∼ 4 ppm) and nitrogen (∼ 25 ppm) frequency dimensions, indicative of folded proteins. In all spectra it was possible to count ∼ 120 peaks, a number slightly lower than that of non‐proline residues in the primary structure of the CML14; this difference can be attributed to exchange broadening caused by conformational heterogeneity in some protein regions or to rapid exchange with solvent protons. The protein appeared to undergo only minor conformational changes upon addition of saturating amounts of either Ca²⁺ or Mg²⁺, based on the observation that only a few peaks in the HSQC spectrum of the metal‐free form experienced chemical shift or intensity changes. In the HSQC spectra of both Ca²⁺ and Mg²⁺ bound CML14, one new peak appeared in the region 10–10.5 ppm (¹H^N) and 110–115 ppm (¹⁵N), and in particular at the position where the main NH groups belonging to homologous glycine residues located at the sixth position of the functional Ca²⁺ binding EF‐loops are generally found.6 In the same region, the resonance at 10.3 ppm (¹H^N), 112 ppm (¹⁵N) does not change position or disappear even after addition of EGTA concentrations higher than 5 mM. The analysis of a 3D ¹H‐¹⁵N TOCSY‐HSQC spectrum, enabling the identification of spin‐systems, confirmed that these two downfield peaks in the Ca²⁺‐bound form are characterized by a system of spin‐spin coupled resonances typical of glycine residues [Fig. 2(D)].

Figure 2.

NMR spectra of ¹⁵N‐CML14 in different conditions. (A–C) Two‐ dimensional ¹H‐ ¹⁵N HSQC NMR spectra of: ¹⁵N‐CML14 in 10 mM Tris‐HCl, 25 mM KCl pH 7.5 buffer with the addition of (A) 5 mM EGTA, (B) 5 mM CaCl₂, and (C) 5 mM MgCl₂. (D) Selected regions from 3D ¹H‐¹⁵N TOCSY‐HSQC spectrum of Ca²⁺‐bound ¹⁵N‐CML14. Slices at ¹⁵N chemical shift of 113 ppm (A) and 111.9 ppm (C) showing ¹H^N‐¹H correlations. Slices at ¹H^N chemical shift of 10.4 ppm (B) and 10.3 ppm (D) showing ¹⁵N‐¹H correlations. Dashed lines represent ¹H^N or ¹⁵N spin coupling with the two ¹H protons of glycine residues.

We further investigated the effect of chemically different metal ions, Ca²⁺ and Mg²⁺ binding to CML14 by comparing the HSQC spectra recorded on the metal‐loaded forms (Fig. 3). The two spectra showed slight differences in peak positions, even if the same number of peaks was observed, suggesting a similar structural rearrangement upon addition of Ca²⁺ and Mg²⁺. Interestingly, the addition of Ca²⁺ to Mg²⁺‐saturated CML14 induces a chemical shift variation of one of the glycine peaks towards the position typical of the Ca²⁺‐bound form (Fig. 3).

Figure 3.

Comparison of the HSQC spectra recorded on the metal‐loaded forms. (A) Overlay of ¹H‐¹⁵N HSQC spectra recorded on ¹⁵N‐CML14 after the addition of Mg²⁺ (green), Ca²⁺ (blue), and Ca²⁺ into Mg²⁺‐saturated ¹⁵N‐CML14 (red). (B) Zoom of the spectra of the glycine downfield peaks.

Contribution of metal binding to protein thermal and local stability

The effects of Ca²⁺ and Mg²⁺ on the stability of CML14 were analyzed by differential scanning calorimetry (DSC) and limited trypsin digestion experiments. We performed DSC studies on apo, Ca²⁺‐, and Mg²⁺‐bound CML14 at different protein concentrations (40–110 µM). Figure 4(A) shows the overlay of representative concentration‐normalized baseline‐corrected thermograms. The peak maximum of metal‐free (apo) CML14 (transition temperature, T _m = 44.1 ± 0.4°C, ΔH = 154 ± 3 kJ mol⁻¹) is lower than the peak maximum of Mg²⁺‐bound (T _m = 66.3 ± 0.6°C, ΔH = 198 ± 8 kJ mol⁻¹) and Ca²⁺‐bound CML14 (T _m = 65.0 ± 0.4°C, ΔH = 189 ± 2 kJ mol⁻¹); thus, binding of Ca²⁺ or Mg²⁺ to CML14 causes an ∼ 20°C stability increase over that of the apo protein. Interestingly, in the DSC thermograms for apo CML14, Mg²⁺‐ and Ca²⁺‐bound, a single sharp transition peak was observed, thus suggesting that CML14 likely undergoes unfolding as a single cooperative unit. All the T _m were independent of protein concentration and the processes were reversible as confirmed by recovery of the enthalpy of unfolding upon cooling and subsequent reheating. Data were fitted assuming monomeric two‐state unfolding.

Figure 4.

Analysis of thermal stability and conformational change upon metal binding (A) DSC thermograms showing the unfolding of the CML14. DSC thermograms of apo (red line), Ca²⁺‐ (blue line) and Mg²⁺‐bound CML14 (green line), respectively, after baseline‐correction and concentration‐normalization. The fits (colored lines) obtained by using a two‐state transition model are superimposed to the experimental data (black short dotted lines). (B) and (C) SDS‐PAGE analysis of trypsin digestion of CML14 in the absence (i.e., in the presence of EGTA) (B) and presence (C) of Ca²⁺. The intensity of the untreated CML14 band in lane 2, was assumed to be 100%. CML14 bands in the corresponding gel were normalized to this ∼ 17 kDa band (indicated by an arrow). Lane 1: MW marker, lanes 2–10:trypsin digestion products obtained following incubation of CML14 with trypsin 1:500 (w/w) for 0, 5, 10, 20, 40, 60, 120, 180, and 240 min, respectively. (D) ANS‐fluorescence of CML14 in the presence of 5 mM CaCl₂ (black line), MgCl₂ (red line), or EGTA (blue line). The spectra of ANS alone (green line) is also shown. (E) Intrinsic fluorescence spectra of CML14 in the presence of EGTA (blue line), CaCl₂ (black line), and MgCl₂ (red line). (F) Electrophoretic analysis of CML14 mobility in the presence (lane 2) or in the absence (lane 3) of Ca²⁺ (i.e., in the presence of the chelator EGTA). Lane 1 represents a MW marker.

Further information on the structural features of apo and Ca²⁺‐bound CML14 was obtained by limited proteolysis experiments. The proteolysis patterns of CML14 in the absence and presence of Ca²⁺ are shown in Figure 4(B,C). The digestion reaction was quenched at different time points ranging from 5 min to 4 h and analysed by SDS‐PAGE. The proportion of protein cleaved by trypsin was followed by the decrease in intensity of the ∼ 17 kDa band stained with Coomassie blue, which corresponds to intact CML14. In the absence of Ca²⁺, CML14 is highly susceptible to proteolysis and more than 85% of the apo CML14 was cleaved within 60 min of trypsin treatment [Fig. 4(B)]. In the presence of Ca²⁺, the decrease in the CML14 band intensity after 60 min was only 20%, leading to the accumulation of a ∼ 15 kDa fragment, suggesting that Ca²⁺ binding protects the protein against trypsin cleavage [Fig. 4(C)]. Moreover, the finding that the 15 kDa band did not stain with an antibody raised against His tag, as revealed by western blot analysis, indicates that the cleavage occurs within the N‐terminal region (data not shown).

Analysis of conformational changes upon metal binding

Ca²⁺‐dependent binding of the hydrophobic fluorophore ANS, whose fluorescence increases substantially when it binds to hydrophobic area of proteins, is a characterizing feature of conformational change in Ca²⁺ sensors. Following addition of saturating concentrations of Ca²⁺ or Mg²⁺, CML14 showed little to no change in the ANS fluorescence intensity [Fig. 4(D)]. To investigate the local structural changes after Ca²⁺ or Mg²⁺ binding, we have also recorded the fluorescence spectra of the apo and holo proteins by selectively exciting the only Trp residue (Trp124) of CML14. As shown in Figure 4(E), the Trp fluorescence spectrum of apo CML14 has an emission maximum at 346 nm. The addition of 5 mM MgCl₂ or CaCl₂ results in very little change in emission intensity and no change in emission maximum wavelength. These spectra appear to be indicative of the absence of a large conformational change between apo and Ca²⁺ or Mg²⁺‐saturated CML14. Structural changes upon Ca²⁺ binding were further investigated via measurements of the apparent molecular mass (AMM) by size exclusion chromatography to provide information on the Stokes radius of CML14. Apo and holo CML14 showed a R _s value of 24.3 ± 0.03 Å and 23.4 ± 0.01 Å, respectively. Thus, Ca²⁺ binding to CML14 results in a negligible decrease in R _s value (∼ 0.9 Å versus 2.36 Å in CaM1 from Arabidopsis thaliana 14), even if the variation is in the same direction of other Ca²⁺ binding protein, that is, CaMs. Consistent with these results, the electrophoretic mobility of CML14 shows only a very minor change in the presence or in the absence of Ca²⁺, in comparison to the well documented large shift for CaMs and other Ca²⁺ sensors [Fig. 4(F)].

CML14 sequence analysis and molecular model

A detailed analysis of CML14 EF‐hands primary sequences [Fig. 5(A)] revealed that EF‐hands 2 and 3 are most likely non‐functional in terms of Ca²⁺ binding because of changes at critical positions in the Ca²⁺‐binding loops that would prevent the seven‐coordination bonds necessary for Ca²⁺ chelation. In EF3, the highly conserved aspartate and glutamate residues at positions 3 and 12 of functional Ca²⁺‐binding loops are replaced by glycine and aspartate, respectively. In the EF2 region, the conserved aspartate residues at position 3 and 5 are substituted by glutamate and threonine. On the contrary, the sequence of EF1 is recognized as typical for Ca²⁺‐binding EF‐hands by PROSITE prediction.12 Moreover, a molecular model of CML14 [Fig. 5(B)], sharing a sequence identity of 45% with its closest template (PDB accession code 3EVR), predicted that only the first of these EF‐hands can be occupied by a Ca²⁺ ion. Interestingly, the presence of Ala at position 9 (Ala30) in EF1 likely causes the loss of water molecule in the coordination, and suggests that only six oxygens of the EF1‐hand motif are coordinating the Ca²⁺ ion in the metal bound protein. However, the homology model visual inspection indicates that Asp24 could donate both of its oxygens to satisfy the seven‐coordination arrangement [Fig. 5(C)], compensating for the loss of water coordination, as observed for other metal‐binding proteins.15

Figure 5.

CML14 sequence alignment and predicted structure. (A) Sequence alignments of EF‐hand motifs of CML14 and various EF‐hand proteins. Black, dark‐gray and light‐gray shading represent amino acid similarity conserved in seven, six and five proteins, respectively. Sequence analysis was performed using GeneDoc (version 2.7.000; http://www.nrbsc.org/gfx/genedoc). (B) Modelled structure of CML14 in the presence of Ca²⁺ bound to the first EF‐hand (shown as a green sphere). (C) Structure of the first EF‐hand showing the potential Ca²⁺ coordinating residues.

Discussion

An interesting feature, typical of plant genome, is the presence of multiple genes predicted to encode a large numbers of CaM isoforms and CMLs. However, due to the absence of empirical evidence regarding the role of Ca²⁺ binding to CMLs, only a very limited number of CMLs have been unambiguously shown to function as Ca²⁺ sensors.

Arabidopsis thaliana CML14, which is phylogenetically close to CaMs, is predicted to have only one functional Ca²⁺ binding domain despite the presence of three EF‐hand motifs.12 Our ITC results are consistent with this prediction and show that CML14 is able to bind only one Ca²⁺ ion. Based on sequence analysis and a homology model, it is likely that EF‐hand 1 is responsible for the binding of a Ca²⁺ ion to CML14 as EF2 and EF3 are missing fundamental residues to coordinate Ca²⁺.

Depending on their affinity and selectivity for Ca²⁺, Ca²⁺‐binding sites generally can bind also Mg²⁺ four to five orders of magnitude less strongly. Ca2 ⁺‐binding sites of proteins are usually classified into Ca²⁺‐specific (or regulatory) and Ca²⁺/Mg²⁺ sites (or structural sites).2, 16 Ca²⁺/Mg²⁺‐sites commonly bind Ca²⁺ with dissociation constants K _d < 10⁻⁷ M, while Ca²⁺‐specific sites bind Ca²⁺ with K _d > 10⁻⁷ M.2 Moreover, in contrast to Ca²⁺‐specific sites, Ca²⁺/Mg²⁺ sites possess a higher affinity for Mg²⁺ (at least several fold). However, because of the high physiological levels of free cytosolic Mg²⁺ (0.4–0.5 mM) compared to intracellular free Ca²⁺ level at resting (∼100 nM) in plant cells,17, 18, 19 many Ca²⁺ specific sites have sufficient affinity to bind Mg²⁺ in the resting cell.16

The values of CML14 dissociation constants for Ca²⁺ and Mg²⁺ obtained from ITC ( $K_{\mathrm{d},\text{Ca}2+}$ ∼12 μM for Ca²⁺ and $K_{\mathrm{d},\text{Mg}2+}$ ∼2.5 mM for Mg²⁺ binding) allow the functional EF‐hand of CML14 to be classified as a Ca²⁺‐specific site and not as a Ca²⁺/Mg²⁺ site.2 A hallmark of Ca²⁺ specific binding sites is also a difference in conformation between the Ca²⁺ and Mg²⁺ bound states of EF‐hand motif.2, 20 The Mg²⁺‐bound form shows a NMR spectral fingerprint distinct from that of the Ca²⁺‐bound state, reflecting a locally different conformation, consistent with the properties of the Ca²⁺‐specific sites and with the chemical features of the two metal ions. Further, in the EF hand 1, the 12^th loop residue is a Glu (Glu33) and not an Asp. Generally, the presence of an Asp residue at position 12 provides a single ligand to bind both Ca²⁺ and Mg²⁺, thus creating a smaller binding site that favors binding of Mg²⁺ and normally produces a Ca²⁺/Mg²⁺ binding site.21 Moreover, the Ca²⁺/Mg²⁺ NMR displacement experiments are consistent with the absence of additional Mg²⁺ binding sites.

As Mg²⁺ is effectively able to compete for EF binding sites, it can influence their Ca²⁺ affinities. Both Ca²⁺ specific and Ca²⁺/Mg²⁺ sites in the presence of Mg²⁺ are characterized by a decrease in Ca²⁺ affinity due to a Mg²⁺ inhibitory effect on the kinetic on‐rate for Ca²⁺ when the association with Ca²⁺ is limited by the kinetic off‐rate of Mg²⁺.22 ITC data indicated that in the presence of Mg²⁺ the CML14 affinity for Ca²⁺ decreases by 5‐fold. The capability of Ca²⁺ to completely displace Mg²⁺ is supported also by NMR experiments. Indeed, the HSQC spectrum recorded on CML14 after the addition of Ca²⁺ to Mg²⁺ bound form is totally superimposable to the spectrum of the Ca²⁺ bound species. By decreasing the apparent Ca²⁺ affinity, the in vivo biological effect of Mg²⁺ competition could be the delay of EF‐hand proteins response, in particular of proteins with Ca²⁺‐specific EF‐hands, to an increased cytoplasmic Ca²⁺ concentration. Thus, Mg²⁺ binding to the EF‐hand proteins can be hypothesized to have an active role in Ca²⁺‐dependent regulation of cellular processes by shifting the target enzyme activation curve to a higher concentration of Ca²⁺ and drastically decreasing the activity of target enzymes at resting Ca²⁺ levels.

Our characterization of the Ca²⁺, Mg²⁺ and metal‐free states of CML14 suggested that metal binding to CML14 provides the protein with significant thermal stability. DSC experiments showed that the apo form of CML14 is rather stable (mid‐transition temperature 44°C) and that Ca²⁺ and Mg²⁺ association increases its thermal stability by about 20°C. Furthermore, the presence of a bound Ca²⁺ ion affects the susceptibility of CML14 to proteases, resulting in a decreased protein accessibility. The protective nature of Ca²⁺ against proteolytic degradation of CML14 seems to be a general function of Ca²⁺ binding to EF‐hand containing proteins.

Binding of Ca²⁺ to CML14 does not result in a measurably more hydrophobic protein surface as shown by ANS fluorescence and by the fact that the protein cannot be purified by Ca²⁺‐dependent hydrophobic interaction chromatography, a method widely used for the purification of Ca²⁺‐binding proteins.23 Additionally, it does not induce the large conformational rearrangement typical of Ca²⁺ sensors as collectively illustrated by gel mobility, SEC analysis, NMR, and intrinsic fluorescence. The number of peaks and therefore of amino acids, which differ in the NMR spectra of the proteins in metal‐free and Ca²⁺ bound forms are consistent with localized structural changes presumably occurring in the functional EF‐hand.

A Ca²⁺ sensor protein must be able to respond structurally within a biologically relevant range of intracellular Ca²⁺ concentrations, and the exposure of hydrophobic patch on Ca²⁺ binding typically serves as a binding surface for target molecules. The small conformational change observed upon Ca²⁺ binding would hence exclude a classical Ca²⁺ sensor role for CML14 and could suggest a Ca²⁺ buffering capacity. However, the presence of a single Ca²⁺ binding site and the low affinity for Ca²⁺ are unlikely to be compatible with a buffer function for the CML14 protein. Thus, if CML14 behaves neither like a classical Ca²⁺ sensor nor like a Ca²⁺ buffer, the mechanism by which the protein works is still unclear. Moreover, the magnitude of conformational change for physiological relevance is likewise not understood. In this scenario, our data could indicate a different role of Ca²⁺ considering target interactions of CML14 in contrast to the typical switch‐like role coordinated by exposure of the interfacial hydrophobic surface of the protein.

While at this stage we are unable to propose a possible function for CML14, a literature survey and exploration of public expression data from Arabidopsis (htpps://www.genevestigator.ethz.ch 24) allow some speculations about the involvement of CML14 in aspects of plant development and the stress response. The expression patterns of CML14 showed clear upregulation (log2 ratio > 2) during seed germination. Moreover, the CML14 gene (together with some other CaM/CML genes, that is, CaM7, CML49, CML3, CML16, etc.) shows significantly increased transcription levels during pollen tube growth25 which is, along with pollen germination, a crucial process for flowering plants reproduction. Taken together, these data raise the possibility that CML14 may be involved in plant development. On the other hand, we cannot exclude CML14 involvement in plant response to stresses as nuclear proteome analysis showed a two‐fold downregulation of the protein in the nucleus of Arabidopsis seedlings upon cold treatment.26 It is worth noting that CML14 transcript levels from the Genevestigator database are not significantly affected by biotic or abiotic stimuli as other CML genes whose expression profiles showed consistent upregulation by various abiotic or biotic stress treatments and for which an involvement in plant response to stress has been ruled out.8, 9, 10, 27 However, with these limited data we can only speculate if these patterns reflect a possible function for the CML14 protein.

In this context, future studies aimed at identifying interaction candidates of CML14 (i.e. by yeast two‐hybrid) and the use of homozygous T‐DNA insertional mutants and/or silenced‐plants for CML14 gene, should help in elucidating the precise mechanism by which the protein works and the possible involvement of CML14 in plant development and/or plant stress responses.

Materials and Methods

Materials

8‐Anilino1‐naphthalene‐sulfonic acid (ANS), isopropyl β‐D‐thiogalactoside (IPTG), trypsin, protease inhibitor cocktail, Bradford reagent, Chelex 100 sodium form and gel filtration molecular mass marker kit were obtained from Sigma. Chromatographic columns were from GE Healthcare, and synthetic oligonucleotides were from Invitrogen. pET12b vector was from Novagene. pUNI51‐U18871 vector was obtained from Arabidopsis Information Resource (TAIR: AT1G62820). All other chemicals were of the highest grade commercially available.

Protein production

The coding sequence for CML14 (At1g62820) was PCR‐amplified using pUNI51‐U18871 vector as template. The amplicon was cloned as N‐terminal His₆‐fusion construct in E. coli expression vector pET12b using NdeI and BamHI restriction sites. The coding region was checked by DNA sequencing. Protein expression was carried out by growing freshly transformed E. coli BL21(DE3) cells in LB medium at 37°C to an OD of 0.6 at 600 nm; induction was subsequently performed at 25°C with 0.4 mM IPTG for 16 hours. M9 minimal medium supplemented with ¹⁵NH⁴Cl (1 g L⁻¹) as a sole nitrogen source was employed to grow cells, in order to obtain ¹⁵N labelled CML14 for NMR studies. After centrifugation, cells were resuspended in 5 mM Tris‐HCl pH 7.5, 150 mM KCl, 10 mM imidazole, and lysed by sonication. The supernatant was loaded on a Ni‐affinity column equilibrated with 5 mM Tris‐HCl at pH 7.5, 150 mM KCl, and 10 mM imidazole. The imidazole concentration was increased stepwise to 500 mM to elute the protein. The purity of the purified protein, assessed by reducing 15% SDS/PAGE and Coomassie staining, was at least 95%. The protein concentration used for all experiments was checked with the Bradford assay.28

Decalcified buffer was prepared using a Chelex‐100 resin column according to manufacturer's instructions. The residual buffer Ca²⁺ concentration after decalcification was measured by BAPTA absorption assay as previously described29 and ranged between 0.8 and 1.3 μM.

Nuclear magnetic resonance spectroscopy

NMR spectra were recorded on a Bruker Avance III spectrometer (Bruker, Karlsruhe, Germany) operating at 600.13 MHz proton Larmor frequency, and equipped with a cryogenic probe. The spectra were recorded at 25°C, in 10 mM Tris‐HCl, 25 mM KCl pH 7.5 at protein concentration of 380 µM in the absence and presence of 5 mM EGTA, 5 mM MgCl₂, or 5 mM CaCl₂, respectively.

A standard ¹H‐¹⁵N heteronuclear single‐quantum coherence (HSQC) pulse sequence was used, with pulsed field gradients for suppression of the solvent signal and cancellation of artifacts. ¹H‐¹⁵N HSQC experiments were performed with a data matrix consisting of 2048 (F2, ¹H) × 256 (F1, ¹⁵N) complex data points, spectral windows of 7211.539 Hz (¹H) × 2128.626 Hz (¹⁵N), 4 transients, and 1.2 s relaxation delay.

3D ¹⁵N‐edited TOCSY‐HSQC spectrum was recorded using a standard sequence with 2048 (F3, ¹H) × 48 (F2, ¹⁵N) × 128 (F1, ¹H) complex data points, spectral windows of 9615.385 Hz (¹H) × 1702.898 Hz (¹⁵N) × 9615.385 Hz (¹H), 8 transients, 1.0 s relaxation delay, and 80 ms spin lock time.

All spectra were processed with Topspin3.2 (Bruker, Karlsruhe, Germany).

Isothermal titration calorimetry

ITC experiments were performed on a TA microcalorimeter (TA Instruments). Ca²⁺ binding to apo or Mg²⁺‐saturated CML14 was evaluated by titrating 2 μL of 1.5 mM CaCl₂ solution into the sample cell containing 130 μM protein dissolved in decalcified 5 mM Tris‐HCl, 150 mM KCl, pH 7.5. To obtain Mg²⁺‐saturation, apo CML14 solution was incubated 15 min at 25°C with 10 mM MgCl₂. Titrant and protein solutions were degassed prior to each titration. Experiments were performed at 25°C, with 25 injections at 300 s intervals. To minimize artefact signals from the buffer, we subtracted the buffer contribution by performing buffer control titrations. In this blank experiment, the ligand solution was titrated into buffer in the sample cell.

Data were fitted using the one set of sites model to obtain values for the binding constant (K _a) and the enthalpy variation associated with binding (ΔH). The variation in free energy (ΔG) and entropy (ΔS) were then calculated from ΔH and K _a through the following equations:

$$\text{ΔG}=-\text{RT} \text{ln}{K}_{\mathrm{a}}$$ (1)

$$\Delta \mathrm{G}=\Delta \mathrm{H}-\mathrm{T}\Delta \mathrm{S}$$ (2)

The dissociation constants were determined from the fitted K_a using:

$$K_{\mathrm{d}}=1/K_a$$ (3)

The reported values represent the mean ± SD of at least three independent titrations. The dissociation constant for Mg²⁺ was determined from the effect of Mg^{2 +} on the observed Ca²⁺ affinity, using the following equation which describes the competition of Mg²⁺ and Ca²⁺ for a single site:13

$$K_{d,{Mg}^{2+}}=\frac{\left(K_{d,{Ca}^{2+}}\left[{Mg}^{2+}\right]\right)}{(K_{d,{Ca}^{2+}, app}‐K_{d,{Ca}^{2+}})}$$ (4)

where $K_{d,{Ca}^{2+}}$ and $K_{d,{Ca}^{2+}, app}$ are the dissociation constants measured at 0 mM and 10 mM MgCl₂, respectively.

Differential scanning calorimetry

DSC experiments were conducted using a nano‐DSC (TA instrument). CML14 samples of 40–110 µM were dissolved in 5 mM Tris‐HCl, 150 mM KCl, pH 7.5 in the presence of 5 mM CaCl₂, 5 mM MgCl₂ or 5 mM EGTA, and were heated from 10°C to 130°C at a scan rate of 60°C/h. The samples were then cooled and rescanned to check if the thermal transitions are reversible. All the samples were degassed prior to analysis. After correcting for the control scan and subtracting a progress baseline, the data were modelled according to a two‐state model using NanoAnalyze software (TA instrument). Each experiment was performed at least in triplicate and reported values represent means ± S.D.

Gel shift assay

The electrophoretic mobility of CML14 was determined after 30 min of incubation at RT of 3 μg of protein with 5 mM CaCl₂ or 5 mM EGTA. Samples were then separated on a 15% SDS‐PAGE gel.

Fluorescence spectroscopy

Fluorescence spectra were collected on a Jasco FP8200 spectrofluorometer. The intrinsic fluorescence emission of the single tryptophan residue was measured using 10 µM CML14 in 5 mM Tris‐HCl pH 7.5, 150 mM KCl in presence of 5 mM EGTA, CaCl₂, or MgCl₂. Spectra were recorded between 300 and 500 nm, upon excitation at 295 nm, with 2.5 nm bandwidths on both sides. Blank spectra (i.e. of samples containing all components except the protein) were subtracted from spectra of samples containing the protein.

ANS fluorescence spectra were recorded as in Ref. 14.

Stokes radius determination by size exclusion chromatography

The Stokes radius (R _s) of apo and holo CML14 was determined by size exclusion chromatography using a Superose 12 column (10/300GL, GE Healthcare) as previously described.14 The column was calibrated by running standard proteins bovine serum albumin (66 kDa, R _s 35.5 Å), ovalbumin (43 kDa, R _s 30.2 Å), carboanhydrase (29 kDa, R _s 23.6 Å), myoglobin (17.8 kDa, R _s 20.2 Å), and cytochrome c (12.4 kDa, R _s 17 Å). The buffer used was 5 mM Tris‐HCl, 150 mM KCl, pH 7.5, with the addition of either 5 mM CaCl₂ or 5 mM EGTA. Elution position V _e of the standards, the void and the included volumes of the column were measured and Stokes radii were determined from a calibration plot of log R _s versus the partition coefficient K _AV. Each experiment was performed at least in triplicate and reported values represent means ± S.D.

Limited proteolysis

CML14 at 30 μM protein concentration was treated with trypsin (1:500 w/w) in 5 mM Tris‐HCl, 150 mM KCl pH 7.5 containing 5 mM CaCl₂, or 5 mM EGTA, respectively, at 25°C. At various times 15 μl aliquots were removed for electrophoresis or western blot analysis against the N‐terminal his‐tag (monoclonal anti‐polyhistidine peroxidase conjugate, Sigma). The trypsin reaction was stopped by boiling for 5 min and rapid addition of reducing sample buffer. After staining with Coomassie blue, band intensities were visualized and analyzed using Chemidoc Image Lab Software (Biorad).

Sequence analysis and molecular modelling

The sequence of CML14 was subjected to protein domain functional analysis using Prosite. A computational molecular model of the structure of CML14 was built using the SwissModel server (doi: 10.1093/nar/gku340) with default parameters. The best template, that is, Myosin light chain kinase from Aequorea victoria (PDB: 3EVR), was used to build the coordinates and afterwards, by optimal superposition, the coordinates of the bound Ca²⁺ ion were transferred to the model.