Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently

BBAPAP-39363; No. of pages: 11; 4C: 6 Biochimica et Biophysica Acta xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbapap

3Q1

Annegret Ulke-Lemée a, Hiroaki Ishida b, Mona Chappellaz a, Hans J. Vogel b, Justin A. MacDonald a,⁎

4 5

a

6

a r t i c l e

7 8 9 10 11

Article history: Received 29 January 2014 Received in revised form 14 May 2014 Accepted 24 May 2014 Available online xxxx

12 13 14 15 16 17 18

Keywords: SMTNL1 CaM IQ-motif CHASM Calmodulin binding domain Intrinsically disordered protein

i n f o

R O

Department of Biochemistry and Molecular Biology, University of Calgary, 3280 Hospital Drive NW, Calgary, AB T2N 4Z6, Canada Department of Biological Sciences, University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada

a b s t r a c t

T

E

D

P

The smoothelin-like 1 protein (SMTNL1) is a modulator of smooth and skeletal muscle contractility and can bind to calmodulin and tropomyosin. Calmodulin is the major calcium sensor of eukaryotic cells and it can cycle between calcium-free (apo-CaM) and calcium-bound (Ca-CaM) forms. Bioinformatic screening of the SMTNL1 sequence predicted a second CaM-binding region (CBD1) that is located N-terminal to the previously defined apo-CaM-binding site (CBD2). Pull-down assays, surface plasmon resonance, isothermal calorimetry and NMR techniques were used to determine that CBD1 associated preferentially to Ca-CaM while CBD2 bound preferentially to apo-CaM. Mutation of hydrophobic residues abolished Ca-CaM-binding to CBD1 while acidic residues in CBD2 were necessary for apo-CaM-binding to CBD2. The dissociation constant (Kd) for Ca-CaM-binding to a CBD1 peptide was 26 ∗ 10−6M while the value for binding to a longer protein construct was 0.5 ∗ 10−6 M. The binding of SMTNL1 to both apo-CaM and Ca-CaM suggests that endogenous CaM is continuously associated with SMTNL1 to allow for quick response to changes in intracellular calcium levels. We also found that the intrinsically disordered N-terminus of SMTNL1 can reduce binding to apo-CaM and increase binding to Ca-CaM. This finding suggests that an additional CaM-binding region may exist and/or that intramolecular interactions between the N-terminus and the folded C-terminus reduce apo-CaM-binding to CBD2. Intriguingly, CBD1 is located close to the SMTNL1 phosphorylation site and tropomyosin-binding region. We discuss the possibility that all three signals are integrated at the region surrounding CBD1. © 2014 Published by Elsevier B.V.

E

C

b

O

F

2Q2

Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently

1

35

R

39 37 36 38

1 . Introduction

41 42

The ubiquitous calcium regulatory protein calmodulin (CaM) is the primary cellular signal transducer that provides highly specific and localized responses to changes in the calcium concentration (reviewed in [1–3]). CaM has a dumbbell structure composed of two calciumbinding domains (N- and C-lobes) which are connected by a flexible

N C O

45

Abbreviations: aa, amino acids; bp, base pair; C-terminal, carboxy-terminal; CaM, calmodulin; Ca-CaM, calcium-saturated calmodulin; apo-CaM, calcium-free calmodulin; CBB, coomassie brilliant blue; CBD, aM-binding domain; CH, calponin homology; ΔCH, deletion of the calponin homology domain; CSP, chemical shift perturbation; Δ4K, deletion of KTKKK; GST, glutathione-S-transferase; HSQC, heteronuclear single quantum correlation; ITC, isothermal titration calorimetry; kDa, kilodalton; Kd, equilibration dissociation constant; MARCKS, myristoylated alanine-rich protein kinase C substrate; MLCK, myosin light chain kinase; N-terminal, amino-terminal; NaCl, sodium chloride; [NaCl], sodium chloride concentration; pxIDR, proximal intrinsic-disordered region; SMTNL1, smoothelin-like 1 protein; SPR, surface plasmon resonance; TMB, tropomyosin-binding; VGCC, voltage-gated calcium channel; WT, full-length wild-type SMTNL1 ⁎ Corresponding author at: Smooth Muscle Research Group at the Libin Cardiovascular Institute of Alberta, University of Calgary, Faculty of Medicine, 3280 Hospital Drive NW, Calgary, AB T2N 4Z6, Canada. Tel.: +1 403 210 8433; fax: +1 403 270 2211. E-mail addresses: [email protected] (A. Ulke-Lemée), [email protected] (H. Ishida), [email protected] (M. Chappellaz), [email protected] (H.J. Vogel), [email protected] (J.A. MacDonald).

U

43 44

R

40

linker region. Each lobe comprises two EF-hands that bind to calcium. Calcium binding leads to major conformational rearrangements in each lobe and the formation of large hydrophobic patches that accept bulky hydrophobic residues of target proteins [4]. In the most common protein–protein complex structures, calcium-saturated CaM (Ca-CaM) is wrapped around the CaM-binding domain (CBD) of a target protein. The association of CaM with a target protein enables selective alterations in enzymatic activities, intracellular localization and/or other biological events. Ca-CaM most often binds to a classical CBD that consists of a basic, amphipathic ~ 20 residue α-helix with large hydrophobic amino acids [4,5]. The regulation of cellular events is most commonly associated with Ca-CaM; however, it is also recognized that calcium-free CaM (apoCaM) has important signaling properties by associating with distinct apo-CaM-binding regions of target proteins. The usual apo-CaMbinding domain possesses an IQ-motif sequence that is named for the conserved IQ-residues that initiate the “IQXXXRGXXXR” consensus sequence (X, any amino acid) [6,7]. Various three-dimensional structures of CaM-target complexes highlight the enormous versatility of complex formation that is facilitated by CaM [2]. CBDs can be located close to phosphorylation sites (and/or other post-translational modification sites), protein–protein interaction domains or auto-inhibitory domains,

http://dx.doi.org/10.1016/j.bbapap.2014.05.011 1570-9639/© 2014 Published by Elsevier B.V.

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67

114

122 123

Mammalian CaM was expressed and purified from Escherichia coli strain BL21 (DE3) as described previously [21]. The CBD1 peptide (aa 307–329, Ac-RGPRAQNRKAIMDKFGGAASGPT-NH2) was synthesized by the University of Calgary Peptide Synthesis Facility (Calgary, AB), confirmed by mass spectrometry and shown to be N 95% pure by analytical HPLC. PreScission protease, glutathione-Sepharose 4B, CNBractivated CH Sepharose and pGEX-6P1 were purchased from GE Healthcare (Piscataway, NJ). All other chemicals were purchased from VWR Scientific (Edmonton, AB) or Sigma Chemical Company (St. Louis, MO).

124

2.2. Expression and purification of recombinant SMTNL1 proteins

125

Various SMTNL1 mutants derived from mouse Smtnl1 (GenBank ID: EDL27304.1) were utilized (see Fig. 1C for an overview): Full-length SMTNL1 (WT, base pairs (bp) 1–1377/amino acids (aa) 1–459) CHdomain (CH, lacking the intrinsically disordered domain, bp

93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109

115 116 117 118 119 120 121

126 127 128

C

91 92

E

89 90

R

87 88

R

85 86

O

83 84

C

81 82

N

79 80

U

77 78

2.3. Binding of CaM to immobilized GST-SMTNL1

154

F

2.1. Materials

75 76

O

113

74

R O

2 . Material and methods

72 73

129 130

P

112

70 71

1038–1380/aa 346–459), tropomyosin-binding fragment (TMB lacks the initial 194 amino acids, bp 583–1380/aa 195–459), deletion of CH-domain (ΔCH, bp 1–1038, aa 1–346) and proximal intrinsically disordered region (pxIDR, lacking the CH-domain, aa 583–1038/aa 195–346; termed ΔTBD/ΔCT in [14]) were described previously [9,13, 14]. Several novel constructs, including CBD1/2 (bp 879–1377/aa 293–459) that contains both CBD1 and CBD2 but lacks the intrinsically disordered N-terminus, were newly generated using standard PCR techniques. The Δ4K truncations that lack the terminal five amino acids of CBD2 (KTKKK) were generated using primers with a premature stopcodon. Various single and double amino acid point mutants of WTSMTNL1, TMBΔ4K and CH were generated with the QuickChange sitedirected mutagenesis kit (Stratagene, La Jolla, CA). All SMTNL1 variants were cloned into the pGEX-6P1 vector (GE Healthcare) to generate Nterminal glutathione-S-transferase (GST)-fusion proteins. Proteins were expressed in E. coli strain BL21 (DE3) and isolated using glutathione-Sepharose following the manufacturer's protocol. In most cases, the N-terminal GST was removed by cleavage with PreScission protease followed by MonoQ or glutathione-Sepharose chromatography. The purified SMTNL1 proteins all contain the cloning artifact “GPLGS” at the N-terminus. Purified proteins were concentrated with an Amicon centrifugal filter (Millipore, Billerica, MA) and the buffer exchanged if necessary for the ensuing experiments. Proteins were not aggregated as judged from native gel-electrophoresis and dynamic light scattering (Supplementary Fig. 1).

131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153

Pull-down experiments with various GST-SMTNL1 proteins were performed to demonstrate CaM-binding. GST-SMTNL1 proteins or GST alone (200 μg) was immobilized on glutathione-Sepharose (20 μL) and then incubated for 2 h with purified CaM (100 μg) in buffer either containing 2 mM EDTA or 5 mM CaCl 2 plus 25 mM Tris, pH 7.2 and 50 mM NaCl. After extensive washing, bound CaM was eluted with SDS-PAGE loading buffer, subjected to SDS-PAGE and visualized by western blotting with anti-CaM antibody (05–173, Millipore). Equal loadings of SMTNL1 proteins onto GSTSepharose were verified by Coomassie brilliant blue (CBB) staining of SDS-PAGE gels run in parallel.

155 156

2.4. Binding of SMTNL1 mutants to CaM-Sepharose

166

We completed pull-down experiments with CaM-Sepharose which was prepared by covalently coupling CaM to CNBr-activated CH Sepharose (GE Healthcare) following the manufacturer's recommendations. Blank-Sepharose was generated as a negative control by immobilizing with Tris. Purified SMTNL1 proteins (GST-tag removed) were incubated with CaM-Sepharose or blank-Sepharose in 20 mM HEPES, pH 7.0, in the presence of 5 mM CaCl2 or 2 mM EDTA. After overnight incubation at 4 °C, the Sepharose was washed extensively. One-third of the resin was eluted with SDS-PAGE loading buffer and the remainder extensively washed in the same buffer plus 50 mM NaCl. Next, one-half of the resin was eluted as before and the remainder washed with the same buffer plus 150 mM NaCl and eluted. Equal volumes of all elutions were separated on SDS-PAGE, and SMTNL1 proteins were visualized by CBB staining. The gels were imaged with an LAS4000 luminescent image analyzer (GE Healthcare) and bands containing proteins quantified using Image Quant TL (GE Healthcare). To avoid inaccuracies introduced by variations in CBB staining, each band intensity was normalized to the sum of all detected protein. Statistical significance was calculated for binding to CaM-Sepharose, blank-Sepharose, presence or absence of calcium, as well as the binding dependency on [NaCl]. Binding to CaMSepharose was considered significant if it was different from blankSepharose values at the same [NaCl] and calcium level. GraphpadPRISM was used for all statistical calculations, using two-way ANOVA with Bonferroni post hoc test, p b 0.05, n ≥ 4, unless otherwise noted.

167

T

110 111

which offers an additional explanation for how CaM-binding can contribute to the regulation of the cellular activity of a diverse array of targets. The smoothelin-like 1 (SMTNL1) protein, originally termed calponin homology-associated with smooth muscle (CHASM), was discovered as a protein phosphorylated during calcium desensitization in the smooth muscle [8,9,15]. Several studies have implicated SMTNL1 in the modulation of vascular smooth muscle contractile activity as well as cardiovascular and skeletal adaptation to exercise, development and pregnancy [16–20]. A global SMTNL1 knock-out mouse has allowed the study of its physiological role [16]. The first investigations focused on the cardiovascular system and showed an exercise-adapted phenotype [16]. Newer data suggest that SMTNL1 may provide adaptive regulation of blood vessel contractile capacity through transcriptional/translational effects on protein abundance [18,19]. We have previously reported that SMTNL1 possesses an IQ-motif located within the terminal αhelix of its CH domain [12]. Calponin homology (CH)-domains are protein–protein interaction modules that are involved in cytoskeletal dynamics and signal transduction (for review, see [10,11]). The CHdomain of SMTNL1 is integral to several functional properties of the protein, including its association with apo-CaM [12]. The CH-domain together with a portion of the intrinsically disordered N-terminal region forms a tropomyosin-binding domain that was necessary for association with contractile filaments [13,14]. Furthermore, the SMTNL1 IQ-motif enabled the interaction with apo-CaM but not Ca-CaM. Interestingly, the interaction with the SMTNL1 IQ-motif was mediated by the two EF-hands in the C-lobe of apo-CaM rather than the flexible linker region. Many of the CaM side chains that also serve as calcium-binding ligands formed electrostatic contacts with SMTNL1. Thus we rationalized that binding of calcium to the EF-hand disrupts the interactions with the IQ-motif by requiring the same binding region. In the current study in silico analysis of the SMTNL1 sequence revealed a potential additional CBD that was expected to associate with Ca-CaM. This CBD is located upstream of the CH-domain and within the tropomyosin-binding region near the S301 phosphorylation site. In this work, we verified the binding activity of the newly identified CBD within SMTNL1. Using pull-down studies, isothermal calorimetry, surface plasmon resonance and NMR spectroscopy we demonstrate that SMTNL1 possesses two CBDs, a novel site that specific for Ca-CaM and the previously described IQ-motif that is specific for apo-CaM. The vast majority of proteins regulated by CaM bind either to its calciumfree or -saturated form, creating a calcium-dependent regulatory switch. Intriguingly, SMTNL1 has the ability to associate with both apo-CaM and Ca-CaM, suggesting a novel regulatory mechanism.

D

68 69

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx

E

2

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

157 158 159 160 161 162 163 164 165

168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190

3

R O

O

F

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx

191 192

196

2.5. Isothermal titration calorimetry (ITC) measurements

197

208

ITC experiments were performed on a MicroCal VP-ITC microcalorimeter (GE Healthcare) at 30 °C. Solutions of CaM (0.5 mM) in 20 mM HEPES, pH 7.0, 1 mM EDTA and 1 mM 2-mercaptoethanol were sequentially injected (4–7 μL steps) into a sample cell containing 1.43 mL of 20 μM CH in the same buffer. All titrations were fit to a one-site binding model with MicroCal Origin software to obtain dissociation constants (Kd) and free enthalpy (ΔH). For ITC, surface plasmon resonance and nuclear magnetic resonance, all protein concentrations were determined using the predicted molecular extinction coefficient (using Expasy ProtParam [22], in cm− 1·M−1): CaM ε280 = 2560, pxIDR ε280 = 5500 and TMB and its derivatives ε280 = 18,450, CH and its derivatives ε280 = 18,450.

209

2.6. Surface plasmon resonance (SPR)

210 211

CaM with an extra N-terminal Cys residue was immobilized by thiolcoupling onto a CM5 sensor chip (GE Healthcare) and analyzed with a Biacore X (GE Healthcare). The running buffer contained 10 mM Tris– HCl, pH 7.5, 150 mM KCl, 1 mM CaCl2, and 0.005% (v/v) Tween-20. Different concentrations of analyte (i.e., SMTNL1 protein) were injected at a flow rate of 20 μL min− 1 with a contact time of 2 min at 20 °C. BIAevaluation 3.2 Software (GE Healthcare) was used to process and analyze all the SPR data. The dissociation constant was calculated from the sensorgram. Binding affinities were defined as “not detected” when no signal was detected at the highest sample concentration tested (~10 μM), equivalent to Kd N 10−5 M.

202 203 204 205 206 207

212 213 214 215 216 217 218 219 220

C

E

R

R

200 201

N C O

198 199

U

193 194

T

195

For presentation, the binding to blank-Sepharose was subtracted and the bound protein was expressed as percentage of binding to apo-CaM in the absence of NaCl. Since the strength of binding to apo-CaM varies between the different SMTNL1 proteins used, only the binding strengths within one experiment were compared (see Supplementary Fig. 3).

E

D

P

Fig. 1. Bioinformatic analysis of the SMTNL1 sequence identifies two CaM-binding domains. An alignment of SMTNL1, smoothelin and SMTNL2 in the region surrounding the predicted CaM-binding domains (A). The CH-domain is underlined in gray and its helices (gray boxes) numbered. Calmodulin binding domains (CBD1 and CBD2) are labeled and underlined in black. The SMTNL1 phosphorylation site is marked with an arrow (▼) and * indicates 10 amino acid divisions. SMTNL1, NP_077192, aa 289–459; smoothelin, NP_001152756, aa 742–921; SMTNL2, NP_808444, aa 283–456. The two predicted CBDs in SMTNL1 and their location within the SMTNL1 sequence (B). Φ (F,Y,W,M,I) and x, any amino acid. SMTNL1 protein variants used in this study are shown in (C). The two CBDs are indicated. The amino acids of each truncation are given on the right. WT, wild-type; TMB, tropomyosin-binding; CBD1/2, CaM-binding domain 1/2; pxIDR, proximal intrinsically disordered region; CH, calponin homology domain; ΔCH, calponin homology domain deletion.

2.7. Nuclear magnetic resonance

221

The uniformly 15N-labeled CaM protein was produced by E. coli grown in M9 medium containing 0.5 g 15NH4Cl. All NMR spectra were obtained at 30 °C on a Bruker Avance 500 MHz NMR spectrometer equipped with a triple resonance inverse Cryoprobe with a single zaxis gradient. NMR samples contained 0.5 mM 15N-CaM, 3 mM CaCl2, 20 mM Bis-Tris, pH 6.8, 100 mM KCl, 5 mM dithiothreitol and 0.5 mM 2.2-dimethyl-2-silapentane-5-sulfate. [1H, 15N] HSQC spectra of CaM were obtained in the presence of increasing amounts of unlabeled CBD1 peptide or SMTNL1 proteins (CaM/CBD1 peptide ratio: 0, 0.5, 1.0, 1.2; CaM/TMB protein ratio: 0, 0.5, 1.15 and CaM/pxIDR protein ratio: 0, 0.75, 1.5). Chemical shifts were referenced using 2.2-dimethyl-2-silapentane-5-sulfate [23]. All spectra were processed using NMRPipe [24] and analyzed using the NMRView software [25]. The resonance assignments for the [1H, 15N]-HSQC spectrum of CaM were performed as described in [26]. The chemical shift perturbations (CSP) were evaluated as a weighted average chemical shift difference, between unbound CaM and peptide-saturated CaM, of 1H and 15N resonances, using the equation CSP = (ΔHN)2 + (ΔN/5)2 [27].

222 223

3 . Results

240

3.1. Bioinformatic analysis predicts an additional CaM-binding site

241

Our previous study demonstrated that the CH-domain of SMTNL1 contained an IQ-motif that binds apo-CaM but did not exhibit CaCaM-binding [12]. The “Binding Site Search” function of the annotated Calmodulin Target Database was used for further in silico analysis of the SMTNL1 sequence [27] (Supplementary Fig. 2). This analysis identified another putative CaM-binding domain located just N-terminal of the CH-domain and only 10 amino acids C-terminal of the S301 phosphorylation site (Fig. 1A) [28]. We termed this putative CaM-binding domain CBD1 (aa 310–325) and the previously characterized IQ-motif CBD2 (439–457) (Fig. 1B). The presence of two CaM-binding domains

242

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239

243 244 245 246 247 248 249 250 251

261 262 263 264 265 266 267 268 269 270 271 272 273

3.2. Pull-down experiments confirm CBD1 as a CaM-binding site

275

As a first approach we captured GST-tagged full-length SMTNL1 (WT) and ΔCH (aa 1–346; lacking the CH-domain) proteins onto glutathione-Sepharose and then added CaM in the presence or absence of calcium (Fig. 2). GST alone was used as a negative control. Eluted protein was resolved by SDS-PAGE and retained CaM detected by western blotting (Fig. 2A). The GST-WT and GST-ΔCH proteins bound both CaCaM and apo-CaM whereas there was significantly less association of CaM with GST protein (Fig. 2B). The presence of the CH-domain was

C E R R O

282

C

280 281

N

278 279

U

276 277

286

T

274

We completed pull-down experiments with CaM-Sepharose to achieve a more sensitive analysis, better control the amount of bound protein and avoid the non-specific binding of GST to CaM [31]. An example of full-length SMTNL1 (WT, aa 1–459) binding to Ca-CaMSepharose, apo-CaM-Sepharose or blank-Sepharose is shown (Fig. 3A). Eluted WT SMTNL1 was identified by CBB stain, bands were quantified and binding to CaM-Sepharose was considered significant if the protein amount (band density) was greater than that found for blankSepharose. The binding of SMTNL1 variants to blank-Sepharose was subtracted, and all values were normalized to binding to CaMSepharose in the absence of calcium and [NaCl] (Fig. 3B). WT-SMTNL1 bound to both apo-CaM and Ca-CaM, when compared to the binding of blank-Sepharose. When more stringent washing conditions were employed (i.e., increasing [NaCl] to 50 mM or 150 mM), the binding to apo-CaM was reduced to blank-Sepharose levels, while a significant amount of Ca-CaM-binding remained. This experiment was repeated with various SMTNL1 mutants to further assess CaM-binding properties. As reported previously [12], the CH-domain of SMTNL1 (CH, aa 346–459; containing only CBD2) could bind apo-CaM at low [NaCl] (Fig. 3C). Intriguingly, small but significant (compared to blankSepharose) amounts of CH were recovered with pull-down assays but CaM-binding was lost at higher [NaCl]. Similar amounts of CBD1/2 (aa 293–459), the shortest protein containing both CBDs, were recovered from CaM-Sepharose regardless of calcium levels (Fig. 3D). At high [NaCl] the binding to apo-CaM is reduced to the level recovered from blank-Sepharose. Thus CBD1/2 can bind both Ca-CaM and apo-CaM with slightly higher affinity to Ca-CaM. The pxIDR protein (aa 195–346), which includes CBD1 but lacks CBD2, exhibited robust binding to Ca-CaM and very weak binding to apo-CaM when compared to blank-Sepharose (Fig. 3E). Finally, the TMB protein (aa 195–459) contains both CBD1 and CBD2 and parts of the disordered N-terminus. At higher [NaCl], TMB bound significantly to Ca-CaM, similar to WTSMTNL1 and pxIDR (Fig. 3F). All SMTNL1 proteins bound significantly stronger to Ca-CaM at high [NaCl]. The exception is CBD1/2 which shows similar binding to apo-CaM and Ca-CaM at 50 mM [NaCl]. This surprising observation suggests that the additional N-terminal sequence included in WT and TMB, but lacking in CBD1/2 affects CaMbinding in an unknown manner. That is, the ratio of binding to CaCaM versus binding to apo-CaM at 50 mM [NaCl] is approximately 7, 2 and 1 for WT-SMTNL1, TMB and CBD1/2, respectively. The relative ability to bind Ca-CaM increased as the length of N-terminal disordered sequence was increased, suggesting that either binding to Ca-CaM is increased or binding to apo-CaM decreased. Taken together these data confirm that SMTNL1 contains at least two CaM-binding domains: the CBD1 binds preferentially to Ca-CaM whereas the CBD2 binds to apoCaM. Further, the N-terminal region upstream of CBD1 increased binding to Ca-CaM and/or reduced binding to apo-CaM. Technical limitations of the pull-down method (e.g., CBB staining of eluted SMTNL1 proteins varies with molecular weight and protein structure) prevent the direct comparison of the Ca-CaM affinities of each of the different SMTNL1 variants. Therefore we used the sensitivity to elution with increasing [NaCl] as a measure of relative CaM-binding affinity. However, we attempted to directly compare CaM-binding affinity by normalizing the CBB staining of each SMTNL1 variant to that of the molecular weight marker (Supplemental Fig. 3 and Supplementary Table 1). Using this approach, we were able to confirm that the presence of the N-terminal intrinsically disordered region played a role in CaMbinding, as the affinity to apo-CaM was reduced by its presence. Furthermore, the deletion of the KTKKK sequence in CBD2 reduced binding to apo-CaM with a smaller effect on Ca-CaM-binding. Lastly, the isolated CH-domain was found to be the weakest Ca-CaM-binding variant. The

F

259 260

285

O

258

3.3. CaM-Sepharose is a sensitive method for measuring CaM-binding

R O

256 257

found to significantly increase the amount of CaM recovered on the 283 SMTNL1-Sepharose. 284

P

254 255

within one protein suggests a complex calcium-dependent regulatory mechanism for SMTNL1. The CBD1 motif is strictly conserved in SMTNL1 from different organisms and shows similarity to smoothelins and SMTNL2 (smoothelin-like 2, an uncharacterized protein) (Fig. 1A). CBD1 does not align to any other known CaM-binding domains and shows no sequence similarity to other proteins outside of the smoothelin/Smtnl-family. This is in contrast to the CBD2 motif, a sequence which is conserved in the smoothelins but absent from SMTNL2. CBD1 is an unclassified CaM-binding domain that is predicted to associate with Ca-CaM rather than apo-CaM (Fig. 1B). Its hydrophobic moment (0.6 according to the Eisenberg formula with Kyte-Doolittle values [29]), percentage hydrophobic residues and propensity for αhelix formation (1.1 according to Chou-Fasman [30]) are in line with many Ca-CaM-binding domains [28]. CBD1 is predicted to be αhelical; indeed it is located at the beginning of a region of SMTNL1 that was suggested to have secondary structure whereas the remaining N-terminal sequence of SMTNL1 is thought to be intrinsically disordered [13]. Circular dichroism spectroscopy indicated no increase of α-helical content upon CaM-binding (data not shown). Based on the prediction of two CBDs in the SMTNL1 protein, we designed various truncations to test CaM-binding to each CBD (either singly or in tandem) in the presence or absence of calcium (Fig. 1C).

D

252 253

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx

E

4

Fig. 2. Binding of SMTNL1 protein variants to CaM using glutathione-Sepharose pull-down assays. GST-tagged SMTNL1 (WT, aa 1–459), GST-ΔCH (ΔCH, aa 1–346) or GST was immobilized to glutathione-Sepharose and calmodulin was added in the absence (apoCaM) or presence (Ca-CaM) of calcium. Captured calmodulin was eluted and visualized with SDS-PAGE and western blotting (A). Molecular weight markers in kDa are indicated on the left. In (B), bands were quantified and expressed as percentage of GST-SMTNL1 (WT) binding to apo-CaM (gray) or Ca-CaM (white), respectively. a, Significantly different from GST; b, significantly different from GST-tagged SMTNL1 by two-way ANOVA with Bonferroni post hoc test, p b 0.05; n = 3.

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346

5

P

R O

O

F

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx

347 348

C

T

E

D

Fig. 3. Binding of SMTNL1 protein variants to CaM-Sepharose. Purified SMTNL1 protein variants were incubated with CaM-Sepharose or blank-Sepharose in the absence (apo-CaM, gray) or presence (Ca-CaM, white) of calcium. After washing with NaCl of the indicated concentration, bound protein was eluted and visualized by SDS-PAGE and Coomassie brilliant blue staining (CBB). In (A), an example is provided of wild-type SMTNL1 (aa 1–459) binding to CaM-Sepharose (CaM-S) or blank-Sepharose (Blank-S) in the absence (−) or presence (+) of calcium [Ca] after washing with increasing NaCl concentration [NaCl]. Molecular weight markers in kDa are indicated on the left. SMTNL1 is visible at ~97 kDa. Eluted SMTNL1 protein was quantified by densitometry, the binding to blank-Sepharose was subtracted and binding expressed as percentage of binding to apo-CaM at 0 mM [NaCl] (B). Additional SMTNL1 variants tested include CH (aa 346–459, C), CBD1/2 (aa 293–459, D), TMB (aa 195–459, E) and pxIDR (aa 195–346, F). Due to the nature of the experiment the absolute binding of different SMTNL1 mutants cannot be compared. a, significantly different from binding to blank-Sepharose; b, significantly different binding to apo-CaM compared to Ca-CaM. Two-way ANOVA with Bonferroni post hoc test, p b 0.05; n ≥ 4.

350

3.4. NMR studies demonstrate that CBD1 is a Ca-CaM-binding domain

351

We employed NMR to investigate the binding of SMTNL1 to 15Nlabeled Ca-CaM (Fig. 4A). The HSQC spectrum of 15N-labeled Ca-CaM was recorded in the presence of increasing amounts of CBD1 peptide (aa 307–329), pxIDR (aa 195–346; lacks the CH-domain) or TMB (aa 195–459; contains CBD1, CBD2 and a portion of the N-terminus). The residues of CaM involved in binding to TMB and pxIDR are similar and were confirmed by superposition of the spectra of TMB and pxIDR. In contrast, CBD1 peptide binding to CaM is slightly different. The chemical shifts detected are sensitive to the environment of the nuclei, hence chemical shift perturbation (CSP), in dependency of protein concentration, can be used to determine which nuclei are involved in interactions. Plotting the CSPs of CBD1 peptide-saturated Ca-CaM against residue number indicated that the C-lobe of CaM was mainly involved in the binding (Fig. 4B). The N-lobe remains largely free in solution or weakly bound. This might allow the Nlobe to bind to other parts of SMTNL1 not included in the CBD1 peptide, for example the intrinsically disordered region. Indeed, the CSPs from bound TMB showed that the N-lobe of CaM contributes to interaction with regions not present in CBD1 (Fig. 4B). Also, the magnitude of the CSPs for the CBD1 peptide is smaller for TMB. This is consistent with the idea that the region outside of CBD1 also contributes to binding via CaM's N-lobe and to the higher binding affinity. The residues of CaM with a large CSP are located in the hydrophobic patch of Ca-CaM, suggesting that the interaction is

356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374

R

R

N C O

354 355

U

352 353

E

349

pxIDR variant bound much weaker than expected, which could be attributed to technical limitations associated with quantifying CBB staining intensities.

predominantly hydrophobic and is similar to the “classic CaMtarget binding motif” [5]. Since the spectra of pxIDR and TMB are virtually identical, we expect that both proteins interact with Ca-CaM in a similar manner. The overlay also shows that the CH-domain and CBD2 included in TMB, but not in pxIDR, do not contribute strongly to Ca-CaM-binding. This agrees with our previous observation that CBD2 cannot bind Ca-CaM [12] and further confirms that CBD1 is sufficient for binding of SMTNL1 to Ca-CaM.

375 376 377 378 379 380 381 382

3.5. Quantification of Ca-CaM and apo-CaM-binding to SMTNL1 by surface 383 plasmon resonance (SPR) 384 We employed SPR to quantify CaM-binding to SMTNL1. For this analysis, a Cys-mutant of CaM was immobilized by thiol-coupling to the solid phase, and various SMTNL1 proteins were used in the liquid phase. The sensorgram showed that pxIDR (aa 195–346) robustly bound to the CaM-saturated sensor chip in the presence of calcium but no binding was detected in the absence of calcium (Supplementary Fig. 4). The dissociation constants were determined from the sensorgrams (Table 1). A Kd = 0.494 ± 0.10 μM was measured for the binding of pxIDR protein (aa 195–459) to Ca-CaM. We further measured Ca-CaM-binding of the CBD1 peptide and found Kd = 26.2 ± 16.9 μM, much higher than to the Kd for pxIDR. There appeared to be additional components within the pxIDR that reduced the Kd when compared to the shorter CBD1 peptide (aa 307–329). This agrees with our pull-down studies and NMR data that suggest the N-terminal region of SMTNL1 could increase binding to Ca-CaM. In the absence of calcium no binding of peptide with CaM was detected, again suggesting that CBD1 is specific for Ca-CaM (Supplementary Fig. 4). SPR further confirms CBD1 as a major Ca-CaM-binding site in SMTNL1.

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402

6

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx

A

Ca-CaM + CBD1

Ca-CaM + pxIDR 104

104

137

137

57

57

109 29

17

114

N ppm

134

29

134

17

114

19

15

19

15

N ppm

109

143

143

119

F

128

128

124

124 147

9.9

9.4

8.9 1

8.4

7.9

64

7.4

6.9

6.4

10.4

5.9

H ppm

104

104

137 57

109

9.4

8.9

8.4 1

7.9

7.4

6.9

6.4

5.9

H ppm

E

N ppm

29 17

114

E

128

124 147

9.4

8.9 1

7.9

7.4

6.9

124

6.4

5.9

0.4

9.9

9.4

8.9

8.4

H ppm

7.9

7.4

6.9

6.4

5.9

H ppm

CBD1

C

0.3

10.4

1

O

B

8.4

119

R

9.9

R

141 115 64

10.4

C

143

119

114

15

T

15

19

27

D

109 134

N ppm

9.9

147

Ca-CaM + TMB (red), Ca-CaM + pxIDR (black)

Ca-CaM + TMB

N

0.2 0.1 0.0 0.4

U

CSP

R O

64

10.4

141

27

P

27

115

115

141

O

119

TMB

0.3 0.2 0.1 0.0 1

13

25

37

N-lobe

49

61

73

85

Residue number

97

109

121

133

145

C-lobe

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx

t1:3

SMTNL1 variant

Kd (μM) Ca-CaM b

t1:4 t1:5 t1:6 t1:7 t1:8 t1:9

CBD1 peptide pxIDR TMBΔ4K-WT TMBΔ4K-I317D TMBΔ4K-M318D TMBΔ4K-F321D

26.2 0.49 0.47 0.66 1.57 2.08

t1:10 t1:11

a

± ± ± ± ± ±

16.9 0.10 0.14 0.25 0.29 0.22

Data were obtained at 30 °C in the presence of 1 mM CaCl2. Dissociation constants (Kd) were calculated by fitting data from different concentrations of SMTNL1 during incubation with the CaM sensor chip. Data are means ± S.E.M, n = 3 separate determinations.

F

b

420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447

C

418 419

E

416 417

R

414 415

R

412 413

N C O

410 411

U

408 409

452 453 454 455 456 457 458 459 460

463

O

Ca-CaM-binding domains rely on hydrophobic amino acids that interact with hydrophobic pockets formed within CaM upon calcium binding. We mutated the three hydrophobic amino acids (I317, M318 and F321) of CBD1 to aspartate and tested the binding to Ca-CaM with pull-down assays and SPR experiments (Fig. 6 and Table 1). The mutations within CBD1 were combined with the Δ4K deletion in an attempt to create an SMTNL1 variant that lacked all CaM-binding properties, despite retaining most of the CBD1 and CBD2 sequences. Mutations were introduced into the TMB background (aa 195–459), yielding the TMBΔ4K-I317D (aa 195–455), TMBΔ4K-M318D and TMBΔ4K-F321D proteins. Furthermore, experiments were completed at higher [NaCl] in order to avoid the weak interaction of CBD2 with Ca-CaM. TMBΔ4K-I317D, -M318D and -F321D all showed significant (~70%) reductions in binding to Ca-CaM when compared to TMBΔ4K at 50 mM [NaCl] (Fig. 6A). At physiological [NaCl] only F321D failed to bind to Ca-CaM whereas I317D and M318D still bound significantly when compared to blank-Sepharose, albeit at a weaker level than TMBΔ4K. SPR experiments performed with the same proteins confirmed these data; the binding constants increased for the three mutations with the highest found for the F321D mutant (Kd = 2.08 μM, Table 1). Moreover the binding constant for TMBΔ4K (Kd = 0.47 μM) was similar to that of pxIDR (aa 195–346, Kd = 0.49 μM), confirming that CBD2 did not contribute to Ca-CaM-binding at higher [NaCl]. Taken together the hydrophobic residues within CBD1 were necessary for Ca-CaM-binding with F321 playing the most important role. This is in line with the requirement for bulky hydrophobic residues as an anchor in many Ca-CaMbinding sites [2]. Finally, the TMBΔ4K-F321D variant failed to bind either Ca-CaM or apo-CaM due to the mutation of F321D within CBD1 and the deletion of KTKKK in CBD2, respectively. Thus, these investigations reveal the importance of both CBD1 and CBD2 in Ca-CaM and apoCaM-binding, respectively. As a final assessment we verified that CBD1 and CBD2 could confer CaM-binding to full-length SMTNL1 protein (aa 1–459, WT). The Δ4K and F321D mutations were introduced and apo-CaM and Ca-CaMbinding potentials were assessed by pull-down studies. The WTF321D-Δ4K (aa 1–455) variant showed ~ 70% drop of binding to apoCaM; however, the binding to Ca-CaM was only reduced by ~ 50% when compared to WT (Fig. 6B). We thus added a double mutation in CBD1 to yield WT-I318D/F321D-Δ4K (Fig. 6C). The additional mutation of I318D further attenuated Ca-CaM-binding (~ 80% reduction) when compared to WT. Of note, the affinity of CBD1 for apo-CaM was clearly detectable and was reduced by the I318D/F321D double mutation.

R O

After confirming that SMTNL1 has two binding sites for CaM, we further defined the contribution of specific amino acids within CBD1 and CBD2. To define residues important in the function of CBD2, we deleted the last five amino acids (KTKKK) of CH to create CHΔ4K (aa 346–455). The relevance of these amino acids in the association with apo-CaM was suggested by high CSP values from previous NMR studies [12]. Deleting a larger portion of the C-terminal region of CBD2 yielded insoluble protein, likely due to the disruption of the hydrophobic core of the CHdomain (data not shown and [12]). The KTKKK deletion abolished CHΔ4K binding to apo-CaM-Sepharose in pull-down assays at low [NaCl], showing no increase over blank-Sepharose (Fig. 5A, compare to Fig. 3C). Interestingly a weak association with Ca-CaM was still apparent at lower [NaCl]. Further experiments revealed that deletion of the KTKKK-terminus from CBD1/2 also resulted in a significant reduction of apo-CaM-binding capacity at 50 mM [NaCl] (CBD1/2-Δ4K, aa 293–455, Fig. 5B, compare to Fig. 3D). The apo-CaM-binding activity that remained likely results from weak interaction with CBD1, an event we observed previously with the binding of pxIDR to apo-CaM (Fig. 3E). Minimal effects on binding were observed when KTKKK was deleted from the TMB construct (TMBΔ4K, aa 195–455, Fig. 5C, compare to Fig. 3F). This is likely because the N-terminus already suppresses apoCaM-binding, thus no further suppression is seen at higher [NaCl] with deletion of KTKKK. The significant difference between binding to apoCaM and Ca-CaM at low [NaCl] observed with TMB is lost in TMBΔ4K under the same conditions. The apparent decrease of apo-CaMbinding is likely stemming from loss of association with CBD2 in the TMBΔ4K mutant. Our observations are in line with the observed reduction of apo-CaM-binding by the presence of the N-terminal region upstream of CBD1 (Fig. 3). Taken together our results suggest that the last five amino acids KTKKK of CBD2 are required for apo-CaMbinding. We also confirmed that CBD2 has weak affinity for Ca-CaM which is not affected by deletion of KTKKK. This binding is likely without physiological significance since it was lost at higher salt concentrations. Thus we have identified residues within CBD2 that abolish apo-CaMbinding while leaving the weak Ca-CaM-binding unaffected. Based on the CSP values from previous NMR experiments, K435, E443, R446, K451 and K455 residues in CBD2 were indicated to be involved in apo-CaM-binding [12]. We thus mutated these amino acids to alanine and measured binding by ITC and CaM-Sepharose pulldown assays (Table 2 and Fig. 5D). An exothermic process was observed for the binding of CH (aa 346–459) to apo-CaM (Supplementary Fig. 5). The integrated data was fitted to a 1:1 binding model and resulted in an estimated Kd = 2.7 μM, similar to previously reported values [12].

450 451

461 462

P

405

448 449

3.7. Mutational studies define amino acids in CBD1 important for Ca-CaM-binding

D

3.6. Mutational studies define amino acids in CBD2 important for apo-CaM-binding

T

403 404

406 407

K435A, Δ4K and the double mutant K435A/R446A abolished apoCaM-binding (i.e., no quantifiable heat release was observed) while the R446A and E443A mutations had no effect on the measured Kd. The amount of heat generated (ΔH, free enthalpy) of K455A and K451A is only ~30% to 40% of CH, thus K455 and K451 also contribute to apo-CaM-binding. This is further reflected in their slightly increased Kd values. In addition to ITC data, we investigated these mutants using CaM-Sepharose pull-downs (Fig. 5D). K435A, the double mutant K435A/R446A and, surprisingly, R446A all failed to associate with apoCaM-Sepharose. According to our previous Haddock model, K435 and R446 both form salt bridges with the acidic residues of apo-CaM [12]. Thus we can conclude that basic residues within CBD2 are necessary for binding to apo-CaM.

Table 1 Surface plasmon resonance determination of SMTNL1 mutants binding to Ca-CaMa.

E

t1:1 t1:2

7

Fig. 4. NMR studies of Ca-CaM-binding to SMTNL1 verify that CBD1 is a major binding site. In (A), an HSQC spectrum of 15N-Ca-CaM with titration of the CBD1 peptide (aa 307–329, CaM/ CBD1 peptide ratio: 0, 0.5, 1.0, 1.2), pxIDR (aa 195–346, CaM/pxIDR ratio: 0, 0.75, 1.5) or TMB (aa 195–459, CaM/TMB ratio: 0, 0.5, 1.15). The residue numbers of some representative chemical shift perturbations (CSP) of CaM are indicated. Also shown is the overlay of HSQC's of Ca-CaM saturated with TMB (red) and pxIDR (black). The spectra were obtained at 30 °C in the presence of 0.5 mM 15N-CaM, 3 mM CaCl2, 20 mM Bis-Tris, pH 6.8, 100 mM KCl, 5 mM dithiothreitol and 0.5 mM 2.2-dimethyl-2-silapentane-5-sulfate. In (B), the CSPs elicited by CBD1 peptide (top) and TMB (bottom) binding to Ca-CaM are plotted as a function of the residue number of CaM. The red line indicates CSPs N0.1, and residues associated with the N-lobe and Clobe of CaM are indicated.

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx

R O

O

F

8

505

Ultimately we can conclude that SMTNL1 possesses two CBDs in distinct regions of the SMTNL1 protein.

507

4 . Discussion

508 509

Reports implicate SMTNL1 in the modulation of smooth and skeletal muscle contraction [8,9,16,17,20]. SMTNL1 biological functions are thought to be facilitated by its biochemical properties which enable its phosphorylation as well as interactions with tropomyosin of the thinfilament and apo-CaM. Since CaM-binding properties can provide highly specific and localized calcium signaling responses, we have investigated this property of SMTNL1 in more detail. Herein we report that SMTNL1 associates with Ca-CaM via CBD1 with Kd ~ 26.2 μM and with apo-CaM via CBD2 with Kd ~ 3 μM [12]. CBD1 is located in an intrinsically disordered region and primarily directs binding of Ca-CaM. CaM-binding to CBD1 could be greatly reduced by the mutation of specific hydrophobic amino acids to acidic residues. The role of the additional acidic residues in CBD1 remains to be determined. Large CSPs measured by NMR locate to the hydrophobic patch of Ca-CaM when bound to the SMTNL1. This suggests that the interaction is predominantly hydrophobic and is similar to the classic CaMtarget binding motif [5]. The association of apo-CaM to the IQ-motif of CBD2 on the C-terminal end of the CH-domain, is dependent on basic amino acids. Mutation of the basic K435A, R446A or deletion of the Cterminal KTKKK all resulted in loss of apo-CaM-binding. This verifies previous results obtained by NMR studies that these amino acids are critically involved in the formation of salt bridges with the negatively charged residues in two calcium-binding loops of the C-terminal lobe of apo-caM [12]. Our study developed two important reagents for future examination of SMTNL1: a SMTNL1 variant with three minimal alterations (i.e. SMTNL1-I318D/F321D-Δ4K) that shows greatly diminished binding to CaM under physiological conditions and a SMTNL1 variant (i.e., SMTNL1-Δ4K) that shows reduced binding to apo-CaM while leaving Ca-CaM association intact. These proteins will be indispensable for studying the interplay between CaM signaling and SMTNL1 function in situ.

518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538

C

E

R

R

O

516 517

C

514 515

N

512 513

U

510 511

The finding that SMTNL1 binds to both apo-CaM and Ca-CaM suggests that endogenous CaM could be continuously associated with SMTNL1 in vivo to allow for rapid responses to changes in intracellular calcium levels and to confer independence from the concentration of ‘free’ CaM in the cell. There is an intense competition between at least 300 intracellular CaM targets, thus apo-CaM-binding allows for localization and maintenance of CaM with its signaling partner [2,28]. The additional binding site (i.e. CBD1 in SMTNL1) becomes occupied after calcium saturation of CaM, thus ensuring rapid activation in response to intracellular signals that involve changes in the calcium concentration [32]. The constant association of CaM with its binding partners, using one site for apo-CaM-binding and a second site for Ca-CaMbinding, has been described for few other proteins. These include IQGAP, voltage-gated calcium channels (VGCC), calcium-activated potassium channels and cyclic nucleotide-gated ion channels, among others [33–36]. The tertiary structure and affinity for apo-CaM and/or Ca-CaM of these complexes vary widely. The regulation of VGCC by CaM is well described and serves as an example of the variety of CaM regulation (reviewed in [36]). VGCCs associate with a single CaM molecule that is constitutively associated via an IQ-motif [37]. Membrane depolarization leads to entry of calcium and bound CaM regulates the activity of the channel via a calciumdependent feedback-loop. CaM has a dual function in VGCCs, mediating inactivation as well as increasing channel opening. This versatility is conferred by variations of the CaM conformations, complexity of the CaM-binding sites and dependency on the calcium-loaded state. In addition, CaM-target sites do not have to be close as distant sites of the channels can be affected by CaM-binding. In the case of the VGCC CaV1.2, the C-lobe of apo-CaM and Ca-CaM bind to neighboring regions of an IQ-motif in the C-terminus of CaV1.2. This leaves the N-lobe free to interact with an additional domain in the CaV1.2 N-terminus with low affinity, bridging both termini and contributing to the fine-tuning of channel activity [37,38]. In the case of SMTNL1, we observed an association of the C-lobe of CaM with the two CBDs, at least when investigated individually. These data suggest that the C-lobe of CaM is associated with SMTNL1

T

506

E

D

P

Fig. 5. Mutations in SMTNL1 further define the apo-CaM-binding site. The last 5 amino acids of CBD2 were deleted to yield Δ4K truncations of the indicated SMTNL1 variants (A, CHΔ4K, aa 346–455; B, CBD1/2-Δ4K, aa 293–455 and C, TMBΔ4K, aa 195–455). The truncations were used in pull-down studies with CaM-Sepharose in the absence (gray bars) or presence (white bars) of calcium as described in Fig. 3. In (A), CHΔ4K binding to apo-CaM was not detectable, so binding was normalized to that of Ca-CaM-Sepharose at 0 mM [NaCl]. In (B−D), the binding was normalized to apo-CaM in 0 mM [NaCl]. In (D), alanine point mutations within CBD2 (white bars) in the CH-domain background (aa 346–459, black bar) were used for pull-down studies with apo-CaM-Sepharose in 0 mM [NaCl]. a, significantly different from binding to blank-Sepharose; b, significantly different binding to apo-CaM compared to Ca-CaM; c, significantly different binding to CH-domain. Two-way ANOVA with Bonferroni post hoc test, p b 0.05; n ≥ 4.

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx Table 2 Isothermal titration calorimetry of apo-CaM-binding to the CH-domaina.

t2:3

CH-domain

Kd (μM) apo-CaM b

ΔH (kcal/mol) c

t2:4 t2:5 t2:6 t2:7 t2:8 t2:9 t2:10 t2:11

WT K435A E443A R446A K451A K455A K435A/R446A ΔKTKKK

2.70 ± n.d.d 2.91 ± 4.48 ± 5.34 ± 6.80 ± n.d. n.d.

−10,170 ± 467 n.d. −9889 ± 617 −10,630 ± 1556 −4808 ± 484 −2844 ± 464 n.d. n.d.

586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622

F O R O P D

584 585

T

582 583

C

580 581

E

578 579

regardless of intracellular calcium levels. This agrees with the observation that the C-lobe often has higher affinity for its target proteins [2]. Comparing CSPs of apo-CaM-binding to the IQ-motif in CBD2 [12] with CSPs of Ca-CaM-binding to CBD1 (Fig. 4) clearly shows that distinct regions of the C-lobe are involved in each interaction. This finding is in line with the observation that the EF-hand loops of CaM are binding to CBD2; therefore, this conformation must be resolved for the calcium binding of CaM to occur [2]. The N-lobe of CaM was not detected by NMR as binding to CBD1, but interactions were detected with pxIDR (aa 195–346) or the longer TMB (aa 195–459, Fig. 4). Also, pxIDR showed much tighter binding to CaCaM (Kd ~ 0.5 μM) compared to CBD1 (Kd ~ 26 μM) when investigated by SPR and NMR. It is possible that the in silico CaM-binding site prediction underestimated the length of CBD1. L331 and F332 could also contribute to the CaM-binding domain, forming a basic 1-5-16 type of binding motif [28]. This extended CBD1, which was present in pxIDR and TMB but not in the CBD1 peptide, could explain the CSPs for the N-lobe of CaM as well as the lower Kd. Alternatively, the N-lobe could bind to distant elements of the SMTNL1 protein, similar to the N-lobe association observed with the N-terminus of CaV1.2 [37,38]. The Nterminal extension of SMTNL1 is intrinsically disordered, just as the Nterminus of CaV1.2 [13], and we found that inclusion of this intrinsically disordered region increased the affinity for Ca-CaM and/or decreased the affinity for apo-CaM. This suggests that an additional CaM interaction site could be located within the disordered region; however, additional work is necessary to distinguish these possibilities. Previous studies have found the pxIDR of SMTNL1 to be engaged in intramolecular contacts with the CH-domain and suggest that the Nterminus, CBD1 and CBD2 are located in close proximity [13]. This structural conformation could allow a single CaM molecule to alternate between the CBD1, CBD2 and any additional CaM-binding sites. Thus, it is conceivable that the disordered N-terminus of SMTNL1 could reduce apo-CaM-binding to CBD2 via the intramolecular contacts with the CHdomain. When the N-terminus is removed (as in the CBD1/2 construct used herein), apo-CaM could bind to both CBD1 and CBD2. In addition, the intramolecular contacts between the pxIDR and CH-domain were relieved by tropomyosin-binding [13]. So it is also possible that an interaction of tropomyosin with SMTNL1 could have effects on the CaMbinding at CBD1 and CBD2. Previous analyses of SMTNL1 with PSIBLAST and LigBase algorithms revealed a conservation of structural elements between it and the myristoylated, alanine-rich C-kinase substrate (MARCKS) family of proteins [13]. The MARCKS proteins possess large degrees of intrinsic structural disorder and are involved in actin-filament re-organization, cell migration, neuronal development and adhesion [39,40]. MARCKS proteins can also bind apo-CaM and Ca-CaM, but in a very different mode than VGCC. A complex interplay between phosphorylation, Ca-CaMbinding and membrane association allow MARCKS to act like a switch

R

576 577

R

575

a Titration calorimetry was performed at 30 °C in the presence of 1 mM EDTA. CaM (0.5 mM) was sequentially injected into a sample cell containing 20 μM CH protein. Data are means ± S.E.M, n = 3 kinetic determinations. The experiment was repeated with a unique CH protein preparation, and identical results were obtained. b The dissociation constants (Kd) were obtained with a one-site binding model. c ΔH, free enthalpy. d n.d. — no heat detected.

N C O

t2:13 t2:14 t2:15

0.39 0.53 0.97 1.96

U

t2:12

0.30

E

t2:1 t2:2

9

Fig. 6. CaM-binding to SMTNL1 is suppressed by mutations within CBD1 and CBD2. In (A), hydrophobic amino acids within CBD1 were mutated to aspartate in the TMBΔ4K background (aa 195–455, white bars) and compared to TMBΔ4K (black bars). a, Significantly different from binding to blank-Sepharose; b, significantly different from TMBΔ4K binding at the same [NaCl]. In (B), single point mutations were combined with the CBD2 truncation (Δ4K) of SMTNL1 (WT, aa 1–459) and tested for binding to CaM-Sepharose as described in Fig. 3. Only CaM-binding in the presence of 50 mM [NaCl] is shown. In (C), double mutations within CBD1 were combined with CBD2 mutations to test binding to CaM in the presence of 50 mM [NaCl]. IF-DD and I317D/F321D mutations were made on the SMTNL1-Δ4K background (aa 1–455). In (B) and (C), statistical significance was assessed by one-way ANOVA with Newman−Keuls post hoc test, p b 0.05; n ≥ 3. c, Significantly different from corresponding WT value.

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

654

Acknowledgements

655 656

661

This work was supported by research grants from the Canadian Institutes of Health Research (MOP-97931) to J.A.M. and the Natural Sciences and Engineering Research Council to H.J.V. A.U-L is a recipient of a Heart & Stroke Foundation of Canada Fellowship. J.A.M. is a recipient of an Alberta Innovates-Health Solutions (AIHS) Senior Scholar award and a Canada Research Chair (Tier II) in Smooth Muscle Pathophysiology. H.J.V. holds a Scientist award from AIHS.

662

Appendix A. Supplementary data

663 664

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bbapap.2014.05.011.

665

References

651 652

657 Q3 658 659 660

666 667 668 669 670 671 672 673 674 675

C

649 650

E

648

R

646 647

R

644 645

O

642 643

C

636 637

N

634 635

U

632 633

F

653

In conclusion, SMTNL1 is a novel Ca-CaM- and apo-CaM-binding protein. The intrinsically disordered amino-terminus of SMTNL1 can reduce its binding to apo-CaM and/or increase its binding to Ca-CaM. The newly described CBD1 could be part of a larger, flexible “effector domain” that also confers interaction with tropomyosin and intracellular targeting after phosphorylation. Future studies will concentrate on the interplay among these three biochemical properties. We expect CaM association with CBD1 to alter the availability of SMTNL1 for phosphorylation and/ or association with tropomyosin, or conversely phosphorylation and/or tropomyosin-binding may hamper CaM association. Either way, this opens exciting possibilities for the integration of multiple pathways at SMTNL1, connecting thin-filament dynamics, transcriptional regulation and calcium-sensing in contractile cells.

630 631

O

641

629

R O

5 . Conclusions

627 628

P

640

625 626

[9] M.A. Borman, T.A. Freed, T.A. Haystead, J.A. MacDonald, The role of the calponin homology domain of smoothelin-like 1 (SMTNL1) in myosin phosphatase inhibition and smooth muscle contraction, Mol. Cell. Biochem. 327 (2009) 93–100. [10] G.T. Rozenblum, M. Gimona, Calponins: adaptable modular regulators of the actin cytoskeleton, Int. J. Biochem. Cell Biol. 40 (2008) 1990–1995. [11] H.R. Kim, S. Appel, S. Vetterkind, S.S. Gangopadhyay, K.G. Morgan, Smooth muscle signalling pathways in health and disease, J. Cell. Mol. Med. 12 (2008) 2165–2180. [12] H. Ishida, M.A. Borman, J. Ostrander, H.J. Vogel, J.A. MacDonald, Solution structure of the calponin homology (CH) domain from the smoothelin-like 1 protein: a unique apocalmodulin-binding mode and the possible role of the C-terminal type-2 CHdomain in smooth muscle relaxation, J. Biol. Chem. 283 (2008) 20569–20578. [13] J.A. MacDonald, H. Ishida, E.I. Butler, A. Ulke-Lemee, M. Chappellaz, S.E. Tulk, J.K. Chik, H.J. Vogel, Intrinsically disordered N-terminus of calponin homologyassociated smooth muscle protein (CHASM) interacts with the calponin homology domain to enable tropomyosin binding, Biochemistry 51 (2012) 2694–2705. [14] A. Ulke-Lemee, H. Ishida, M.A. Borman, A. Valderrama, H.J. Vogel, J.A. MacDonald, Tropomyosin-binding properties of the CHASM protein are dependent upon its calponin homology domain, FEBS Lett. 584 (2010) 3311–3316. [15] A.A. Wooldridge, J.A. MacDonald, F. Erdodi, C. Ma, M.A. Borman, D.J. Hartshorne, T.A. Haystead, Smooth muscle phosphatase is regulated in vivo by exclusion of phosphorylation of threonine 696 of MYPT1 by phosphorylation of Serine 695 in response to cyclic nucleotides, J. Biol. Chem. 279 (2004) 34496–34504. [16] A.A. Wooldridge, C.N. Fortner, B. Lontay, T. Akimoto, R.L. Neppl, C. Facemire, M.B. Datto, A. Kwon, E. McCook, P. Li, S. Wang, R.J. Thresher, S.E. Miller, J.C. Perriard, T.P. Gavin, R.C. Hickner, T.M. Coffman, A.V. Somlyo, Z. Yan, T.A. Haystead, Deletion of the protein kinase A/protein kinase G target SMTNL1 promotes an exercise-adapted phenotype in vascular smooth muscle, J. Biol. Chem. 283 (2008) 11850–11859. [17] B. Lontay, K. Bodoor, D.H. Weitzel, D. Loiselle, C. Fortner, S. Lengyel, D. Zheng, J. Devente, R. Hickner, T.A. Haystead, Smoothelin Like 1 protein regulates myosin phosphatase targeting subunit 1 expression during sexual development and pregnancy, J. Biol. Chem. (2010) 29357–29366. [18] K. Bodoor, B. Lontay, R. Safi, D.H. Weitzel, D. Loiselle, Z. Wei, S. Lengyel, D.P. McDonnell, T.A. Haystead, Smoothelin-like 1 protein is a bifunctional regulator of the progesterone receptor during pregnancy, J. Biol. Chem. 286 (2011) 31839–31851. [19] S.R. Turner, J.A. MacDonald, Novel contributions of the smoothelin-like 1 protein in vascular smooth muscle contraction and its potential involvement in myogenic tone, Microcirculation (2013). Q5 [20] A. Ulke-Lemee, S.R. Turner, S.H. Mughal, M.A. Borman, R.J. Winkfein, J.A. MacDonald, Mapping and functional characterization of the murine smoothelin-like 1 promoter, BMC Mol. Biol. 12 (2011) 10. [21] K. Nakashima, H. Maekawa, M. Yazawa, Chimeras of yeast and chicken calmodulin demonstrate differences in activation mechanisms of target enzymes, Biochemistry 35 (1996) 5602–5610. [22] H.C. Gasteiger E., A. Gattiker, S. Duvaud, M.R. Wilkins, R.D. Appel, A. Bairoch (Eds.), Protein Identification and Analysis Tools on the ExPASy Server, Humana Press, 2005. [23] D.S. Wishart, C.G. Bigam, J. Yao, F. Abildgaard, H.J. Dyson, E. Oldfield, J.L. Markley, B.D. Sykes, 1H, 13C and 15 N chemical shift referencing in biomolecular NMR, J. Biomol. NMR 6 (1995) 135–140. [24] F. Delaglio, S. Grzesiek, G.W. Vuister, G. Zhu, J. Pfeifer, A. Bax, NMRPipe: a multidimensional spectral processing system based on UNIX pipes, J. Biomol. NMR 6 (1995) 277–293. [25] B.A. Johnson, Using NMRView to visualize and analyze the NMR spectra of macromolecules, Methods Mol. Biol. 278 (2004) 313–352. [26] J.L. Gifford, H. Ishida, H.J. Vogel, Structural characterization of the interaction of human lactoferrin with calmodulin, PLoS One 7 (2012) e51026. [27] S. Grzesiek, A. Bax, G.M. Clore, A.M. Gronenborn, J.S. Hu, J. Kaufman, I. Palmer, S.J. Stahl, P.T. Wingfield, The solution structure of HIV-1 Nef reveals an unexpected fold and permits delineation of the binding surface for the SH3 domain of Hck tyrosine protein kinase, Nat. Struct. Biol. 3 (1996) 340–345. [28] M. Ikura, in: I. Lab (Ed.), Calmodulin Target Database, vol. 2013, Ikura M, Ontario Cancer Institute, 2002. [29] J. Kyte, R.F. Doolittle, A simple method for displaying the hydropathic character of a protein, J. Mol. Biol. 157 (1982) 105–132. [30] P.Y. Chou, G.D. Fasman, Prediction of the secondary structure of proteins from their amino acid sequence, Adv. Enzymol. Relat. Areas Mol. Biol. 47 (1978) 45–148. [31] C.B. Klee, M.H. Krinks, Purification of cyclic 3′,5′-nucleotide phosphodiesterase inhibitory protein by affinity chromatography on activator protein coupled to Sepharose, Biochemistry 17 (1978) 120–126. [32] H. Sanabria, M.A. Digman, E. Gratton, M.N. Waxham, Spatial diffusivity and availability of intracellular calmodulin, Biophys. J. 95 (2008) 6002–6015. [33] A. Pelikan-Conchaudron, C. Le Clainche, D. Didry, M.F. Carlier, The IQGAP1 protein is a calmodulin-regulated barbed end capper of actin filaments: possible implications in its function in cell migration, J. Biol. Chem. 286 (2011) 35119–35128. [34] A.D. Maher, P.W. Kuchel, The Gardos channel: a review of the Ca2+-activated K+ channel in human erythrocytes, Int. J. Biochem. Cell Biol. 35 (2003) 1182–1197. [35] U.B. Kaupp, R. Seifert, Cyclic nucleotide-gated ion channels, Physiol. Rev. 82 (2002) 769–824. [36] D.B. Halling, P. Aracena-Parks, S.L. Hamilton, Regulation of voltage-gated Ca2 + channels by calmodulin, Sci. STKE (2006) (2006) er1. [37] I.E. Dick, M.R. Tadross, H. Liang, L.H. Tay, W. Yang, D.T. Yue, A modular switch for spatial Ca2+ selectivity in the calmodulin regulation of CaV channels, Nature 451 (2008) 830–834. [38] Z. Liu, H.J. Vogel, Structural basis for the regulation of L-type voltage-gated calcium channels: interactions between the N-terminal cytoplasmic domain and Ca(2+)calmodulin, Front. Mol. Neurosci. 5 (2012) 38.

T

638 639

whereby Ca-CaM-binding leads to structural rearrangement and downstream signaling effects [41,42]. All of these activities are regulated via an amphipathic effector domain that also harbors the MARCKS' CBD. The CBD found within MARCKS is unusual in that it (i) has bulky hydrophobic amino acids located in close proximity, (ii) both lobes are involved in binding and form a compact structure and (iii) displays no increase in α-helical content after Ca-CaM-binding [43]. Notably these properties are similar in nature to those identified for the CBD1 of SMTNL1. The data may suggest that in addition to some conservation of structural elements within the intrinsically disordered region of MARCKS and SMTNL1, there is also some overlap of function. Further, the MARCKS effector domain is subject to phosphorylation by PKG and PKC with several consequences, one of them is to block Ca-CaMbinding [44]. Since the serine 301 phosphorylation site of SMTNL1 is located within nine residues of the CBD1 region, we are currently investigating whether CaM-binding is also regulated by phosphorylation and the region surrounding CBD1 could be a similar effector domain.

D

623 624

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx

E

10

[1] D. Chin, A.R. Means, Calmodulin: a prototypical calcium sensor, Trends Cell Biol. 10 (2000) 322–328. [2] H. Ishida, H.J. Vogel, Protein–peptide interaction studies demonstrate the versatility of calmodulin target protein binding, Protein Pept. Lett. 13 (2006) 455–465. [3] D.M. Watterson, Calmodulin and Signal Transduction, 1st Edition Academic Press, New York, 1998. [4] M. Ikura, Calcium binding and conformational response in EF-hand proteins, Trends Biochem. Sci. 21 (1996) 14–17. [5] M. Ikura, G.M. Clore, A.M. Gronenborn, G. Zhu, C.B. Klee, A. Bax, Solution structure of a calmodulin-target peptide complex by multidimensional NMR, Science 256 (1992) 632–638. [6] R.E. Cheney, M.S. Mooseker, Unconventional myosins, Curr. Opin. Cell Biol. 4 (1992) 27–35. [7] M. Bahler, A. Rhoads, Calmodulin signaling via the IQ motif, FEBS Lett. 513 (2002) 107–113. [8] M.A. Borman, J.A. MacDonald, T.A. Haystead, Modulation of smooth muscle contractility by CHASM, a novel member of the smoothelin family of proteins, FEBS Lett. 573 (2004) 207–213.

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011

A. Ulke-Lemée et al. / Biochimica et Biophysica Acta xxx (2014) xxx–xxx

11

[43] E. Yamauchi, T. Nakatsu, M. Matsubara, H. Kato, H. Taniguchi, Crystal structure of a MARCKS peptide containing the calmodulin-binding domain in complex with Ca2+-calmodulin, Nat. Struct. Biol. 10 (2003) 226–231. [44] J.M. Graff, T.N. Young, J.D. Johnson, P.J. Blackshear, Phosphorylation-regulated calmodulin binding to a prominent cellular substrate for protein kinase C, J. Biol. Chem. 264 (1989) 21818–21823.

U

N C O

R

R

E

C

T

E

D

P

R O

O

F

[39] A. Arbuzova, A.A. Schmitz, G. Vergeres, Cross-talk unfolded: MARCKS proteins, Biochem. J. 362 (2002) 1–12. [40] M.I. Mosevitsky, Nerve ending “signal” proteins GAP-43, MARCKS, and BASP1, Int. Rev. Cytol. 245 (2005) 245–325. [41] J.H. Hartwig, M. Thelen, A. Rosen, P.A. Janmey, A.C. Nairn, A. Aderem, MARCKS is an actin filament crosslinking protein regulated by protein kinase C and calciumcalmodulin, Nature 356 (1992) 618–622. [42] C. Gallant, J.Y. You, Y. Sasaki, Z. Grabarek, K.G. Morgan, MARCKS is a major PKCdependent regulator of calmodulin targeting in smooth muscle, J. Cell Sci. 118 (2005) 3595–3605.

Please cite this article as: A. Ulke-Lemée, et al., Two domains of the smoothelin-like 1 protein bind apo- and calcium–calmodulin independently, Biochim. Biophys. Acta (2014), http://dx.doi.org/10.1016/j.bbapap.2014.05.011