Figure 2. Schematic representation of CIB2 cation sensitivity and structural transitions. (
a) In the absence of cations, WT CIB2 is in a molten globule state (left). The addition of a physiological concentration of both Mg
2+ and Ca
2+ (middle top panel) triggers the binding of Mg
2+ in EF3 and EF4. In the presence of physiological Ca
2+ and absence of Mg
2+ (middle bottom panel), Ca
2+ binding is less favorable. When extra (non-physiological) Ca
2+ concentrations are added, Ca
2+ replaces Mg
2+ in EF4 (upper right panel). (
b) E64D CIB2 is in a molten globule state under physiological conditions (left). A well-defined three-dimensional structure is acquired upon the addition of extra (non-physiological) Mg
2+ (upper right panel) or Ca
2+ (bottom right panel), with a more pronounced effect for Ca
2+. The Wi-Fi logo represents the strength of allosteric interactions between E64 and N121 residues.
While a similar conclusion was also previously drawn for CIB1
[22], there is an important difference between the two proteins. The apparent affinity for Ca
2+ of CIB2 is as low as 500 μM
[14] while significantly higher affinity was measured for Ca
2+ binding to CIB1, (K
d = 0.5 μM for EF4; K
d = 1.9 μM for EF3
[22][23]), which makes its Ca
2+ sensing capabilities similar to those of calmodulin
[24]. Moreover, Mg
2+ binding is limited to the EF3 loop in the case of CIB1 (K
d = 120 μM
[22][23]), but it appears to involve both EF3 and EF4 in the case of CIB2 (apparent affinity: 290 μM
[14]). The physiological consequence is clear when these values are considered in the context of intracellular levels of the two cations. Intracellular Ca
2+ oscillates in the 0.1–10 μM range
[25], and in specific cell compartments such as the outer segments of photoreceptors, where CIB2 has also been detected [
2], it reaches even lower values
[26][27]. On the other hand, free Mg
2+ is generally constant in most cells and ranges in the 0.5–1 mM interval
[28][29]. Overall, this suggests that CIB1 can indeed work as a Ca
2+ sensor under physiological Ca
2+ levels, while CIB2 would keep two Mg
2+ ions constitutively bound to both functional EF-hands, being substantially insensitive to intracellular Ca
2+ oscillations (
Figure 2). It is to be noticed that based on a TNS fluorescence (2-p-toluidinylnaphthalene-6-sulphonate) assay, Blazejczyk et al.
[4] reported a much higher affinity (K
d = 0.14 μM) for Ca
2+ for GST-fused CIB2. This value is in contrast to data obtained with untagged CIB2 from direct spectroscopic analyses performed by NMR and CD titrations and indirect competition assays with 5,5′-Br
2-BAPTA
[14] and could be affected by artifacts induced by the interaction with the bulky (26 kDa) fused GST moiety.
The relatively low affinity for Ca
2+ of CIB2, distinguishing it from CIB1, is not surprising if its primary structure is considered (
Figure 1). The structure of CIB1
[19] reveals an optimal pentagonal bipyramid geometry offered by oxygen-coordinating Ca
2+ in EF4 due to the presence of an Asn residue (N169) in the -X position of the coordinating loop and to a Glu residue in the -Z position (E172), which constitutes a bidentate ligand for Ca
2+ that is broadly conserved among EF-hand motifs
[30][31]. At odds, -X and -Z positions in CIB2 are occupied, respectively, by the side-chain-lacking G165 and by D168, which cannot act as a bidentate ligand due to the shorter side chain. This likely distorts the geometry of the whole loop and results in a lower affinity for Ca
2+[14]. Important differences are also found in EF3, which explains the lower affinity for Ca
2+ of this loop for CIB2 compared to CIB1. A negatively charged residue (D118) occupies the Y coordinating position in CIB1, while a neutral one (N118) is substituted in CIB2, thus preventing a favorable coulombic interaction; moreover, the ninth position (-X) in the EF-hand loop is occupied by an Asn (N124) in CIB1 and by the least frequently observed Cys residue (C124), known for destabilizing some EF-hands
[32]. The sequence of CIB2 is, therefore, not evolutionary optimized for Ca
2+ binding.
The presence of physiological targets is known to significantly tune the metal affinity of Ca
2+ sensors, often increasing it, such as in the case of myristoylated recoverin, a prototypical NCS protein in which the co-presence of the GRK1 target and the membrane is required to bring the Ca
2+ sensitivity into the physiological range, thus enhancing over 100-fold the apparent Ca
2+ affinity with respect to the isolated protein
[33]. In the case of CIB2, the presence of a membrane-proximal peptide from the cytoplasmic domain of the α7B integrin was shown to double the apparent affinity for Ca
2+ (K
dapp = 200 μM vs. 500 μM in the absence of the target)
[14]. While significant per se, this change does not support the physiological role of CIB2 as a Ca
2+ sensor, unless the microenvironment of specific cell compartments and the co-presence of other factors such as myristoylation, membrane, or other supramolecular complexes will prove otherwise by enhancing the apparent affinity of CIB2 for Ca
2+.
2.3. An Inter-Domain Allosteric Switch Regulates the Conformational Transitions of CIB2
Although with a lower affinity than CIB1, CIB2 binds Mg
2+ and Ca
2+ (the latter under non-physiological conditions), and this significantly stabilizes the structure of the protein. Thermal denaturation studies performed by monitoring the protein secondary structure content by far-UV CD spectroscopy indeed proved that apo-CIB2 is rather unstable, with a melting temperature (T
m) of 35 °C, while the addition of Mg
2+ or Ca
2+ significantly increases the stability by enhancing the T
m of 11 and 8 °C, respectively
[14]. In the co-presence of both cations, the T
m resembles that of Mg
2+, confirming the minor stabilizing role of Ca
2+ compared to Mg
2+. The stabilizing effects of cations on the protein tertiary structure are even more apparent when analyzed by NMR, which also highlighted an important allosteric mechanism connecting EF3 with the non-functional EF1 motif. An inter-domain allosteric communication was detected between the EF3 binding loop and the residue E64, which is predicted to form an electrostatic interaction with R33, thereby contributing to the stability of the EF1 subdomain. This was clearly demonstrated by
1H-
15N HSQC NMR spectra, where the chemical shift of N121 in EF3 and E64 in EF1 were found to show the same pattern upon Ca
2+-titration experiments. The switch that allows CIB2 to acquire a functional conformation at physiological levels of Mg
2+ seems, therefore, to be finely regulated by an allosteric, long-range communication connecting EF1 with EF3 (
Figure 2a). Interestingly, E64 is mutated into an Asp (E64D) in a group of patients affected by Usher syndrome type 1 J
[2]. Vallone et al. demonstrated that the apparently conservative E64D substitution breaks up such inter-domain communication, resulting in a protein that is unable to bind Mg
2+, which is necessary to adopt the required physiological conformation, thus providing a first mechanistic explanation for the molecular basis of disease
[14].
2.4. CIB2 Is Monomeric, but It Could Dimerize in the Presence of a Target
Analytical size exclusion chromatography (SEC) is a valuable technique to study the hydrodynamic properties of proteins, and it can be used to assess the molecular weight (MW) of the eluting protein when a calibration curve is obtained in the same conditions for a number of globular proteins is available
[34]. The apparent MW of CIB2 obtained by SEC in the presence of Mg
2+ or co-presence of Mg
2+ and Ca
2+ was determined to be around 39 kDa, therefore significantly higher the theoretical MW of a monomer (22 kDa); this was initially interpreted as evidence of the dimeric nature of CIB2
[14]. A similar conclusion was supported by dynamic light scattering (DLS) measurements. A subsequent study employing a variety of mass spectrometry (MS)-based techniques, including native ESI-MS, MALDI-TOF-MS, and cross-linking/MS integrated with novel SEC and DLS experiments based on a more accurate selection of the heterogeneous components of the elution bands, demonstrated that CIB2 was monomeric under all tested conditions
[12]. A comparison with results obtained with recoverin and calmodulin suggested that the apparent MW, extrapolated by analytical SEC for a Ca
2+ sensor protein and based on the hydrodynamic radius, can significantly differ from the real value when the protein has a high hydrophobic solvent-accessible surface, such as in the case of recoverin and CIB2
[12]. Analytical SEC could then be driven to erroneous conclusions as to the oligomeric state of the protein
[14]. This can explain the apparent contradiction with previous studies that detected CIB2 dimers based on FRET and co-immunoprecipitation. In these studies, CIB2 was fused to GFP
[2][35] or tdTomato fluorescent protein
[2]; its uncommon hydrodynamic properties, together with a possible interference of the bulky fusion constructs, may have led to the erroneous detection of a dimer
[12]. The most recent and thorough dedicated investigation performed with untagged purified CIB2 seems to exclude the existence of CIB2 dimers under conditions mimicking the physiological ones
[12]. In this respect, the oligomeric state of CIB2 resembles that of CIB1. Ca
2+-bound CIB1 was indeed found to be monomeric in the crystallographic structure reported by Gentry et al.
[19], and this finding has been supported by NMR diffusion, SEC, and sedimentation equilibrium experiments
[36][37]. The head-to-tail dimer reported by another crystallographic study could result from the specific conditions for crystal formation, which included a GSH moiety at the N-domain
[38].
Surface plasmon resonance (SPR) spectroscopy also excluded the dimerization of CIB2 over a broad range of conditions, including selective incubation with Mg
2+ and Ca
2+[12], but suggested an interesting mechanism of binding to integrin. The formation of a protein–peptide complex between CIB2 and a peptide from the α7B integrin was shown to possibly drive the binding of a second CIB2 molecule; the process seems to be kinetically favored in the sole presence of Mg
2+[12]. Although this hypothesis awaits confirmation, the mechanism of target-induced CIB2 dimerization would explain the 2:1 protein:peptide stoichiometry detected for the same interaction in a previous fluorescence study
[14], and it is tempting to speculate that it might play a role in the integrin signaling mediated by CIB proteins.
2.5. CIB2 Myristoylation
Myristoylation is a post-translational modification operated by the N-myristoyl transferase that, in vivo, covalently binds a 14-carbon saturated fatty acid (myristoyl moiety) to the N-terminal glycine of proteins harboring the consensus sequence MGXXXS/T
[39]. CIB1, CIB2, and CIB3 have the optimal consensus sequence (
Figure 1) and are likely to be myristoylated in vivo, while CIB4 is not. The binding of Ca
2+ ions to some NCS proteins triggers the so-called “myristoyl switch” mechanism, which extrudes the myristoyl moiety from a hydrophobic cleft in the protein milieu to a fully solvent-exposed state, thus allowing membrane-binding and permitting specific cell localization and target interaction
[40].
It is not clear whether CIB1 undergoes a Ca
2+-induced myristoyl-switch, as contrasting conclusions have been reported by different groups. Jarman et al.
[41] concluded that the myristoyl switch mechanism occurs in CIB1, and it is necessary to translocate the sphingosine kinase 1 (SK1) target to the membrane; in contrast, according to Blazejczyk et al., the myristoyl moiety of CIB1 is likely solvent-exposed rather than buried within the protein, regardless of its Ca
2+ binding state
[42]. Whether or not the Ca
2+-induced switching mechanism occurs, the myristoylation of CIB1 is still important for protein stability and membrane targeting
[43][44], as well as shuttling its binding partners to the membrane
[41][45]. Myristoylation-deficient CIB1 variants retain a high affinity for target proteins and peptides both in vitro and in vivo
[42][44][46][47], thus suggesting that the myristoyl group is not specifically involved in target recognition.
Much less information is available about the myristoylation of CIB2. Blazejczyk et al.
[4] expressed GFP-tagged CIB2 and a myristoylation-blocking variant (
G2ACIB2-GFP) in COS-7 cells in the presence of radioactive
3H-myristic acid and compared the effect of CIB2 myristoylation with that of VILIP1-GFP
[48], known to undergo the Ca
2+-induced myristoyl switch, thereby resulting in specialized membrane compartment localization. In contrast to what was observed for VILIP1-GFP, no Ca
2+-dependent translocation was detected for CIB2-GFP, which was interpreted as a lack of Ca
2+-induced myristoyl switch
[4]. Moreover, localization of CIB2 was essentially limited to the crude membrane fraction, regardless of the tag (CIB2-FLAG gave the same results) and, surprisingly, of myristoylation as the same behavior was detected for
G2ACIB2-GFP
[4]. This would exclude a direct association of CIB2 to the membrane by myristoyl anchoring and rather suggests that the binding is mediated directly by lipids or by an interaction with other proteins. Within COS-7 cells, CIB2 co-localizes with the Golgi apparatus and not with the nucleus, and the absence of myristoylation does not affect this pattern
[4].
An independent study by Zhu et al.
[49] confirmed that CIB2 does not undergo a Ca
2+-induced myristoyl switch and blocks the agonist-induced membrane translocation of SK1, at odds with the activity of CIB1. Experiments were performed with HA-tagged CIB2 in HEK293 cells labeled with
3H-myristic acid and showed that the interaction of CIB2 with SK1 was independent of the presence of Ca
2+ or Mg
2+. Further experiments are needed to elucidate the biological role of CIB2 myristoylation, and biophysical studies may clarify the structural and mechanistic aspects that remain somewhat unclear.