Calcium-dependent binding of calmodulin to neuronal gap junction proteins

Abstract

We examined the interactions of calmodulin with neuronal gap junction proteins connexin35 (Cx35) from perch, its mouse homologue Cx36, and the related perch Cx34.7 using surface plasmon resonance. Calmodulin bound to the C-terminal domains of all three connexins with rapid kinetics in a concentration- and Ca2+-dependent manner. Dissociation was also very rapid. K(d)'s for calmodulin binding at a high-affinity site ranged from 11 to 72 nM, and K(1/2)'s for Ca2+ were between 3 and 5 microM. No binding to the intracellular loops was observed. Binding competition experiments with synthetic peptides mapped the calmodulin binding site to a 10-30 amino acid segment at the beginning of the C-terminal domain of Cx36. The micromolar K(1/2)'s and rapid on and off rates suggest that this interaction may change dynamically in neurons, and may occur transiently when Ca2+ is elevated to a level that would occur in the near vicinity of an activated synapse.

Figure 1. Calmodulin binding to Cx35 cytoplasmic domains

A. A family of SPR traces showing steady state binding responses of the Cx35 intracellular loop domain GST fusion protein (1070 RU immobilized) to calmodulin ranging from 0 to 10 μM in buffer containing 1 mM free Ca²⁺. Responses to GST alone have been subtracted. B. A similar family of SPR traces showing binding responses of the Cx35 C-terminal domain GST fusion protein (344 RU immobilized) to 0 to 10 μM calmodulin in buffer containing 1 mM free Ca²⁺. Responses to GST alone have been subtracted as in A. C. Steady state CaM binding data for 10 [CaM] from 1 nM to 10 μM for Cx35 CT, and for 6 [CaM] from 100 nM to 10 μM for Cx35 IL. The fits of a single-component ligand binding model to the Cx35 CT and IL data are shown by the dashed lines; the fit for a two-component model to the CT data is shown by the solid line. Parameters for the two-component fit are shown in Table 2.

Figure 2. Calcium dependence of CaM binding to the Cx35 CT domain

Specific binding responses of the Cx35 CT domain to injections of 2 μM CaM in solutions containing 0.1 μM to 1 mM free Ca²⁺. The curve fit to the data represents a Hill equation with a K_½ for Ca²⁺ of 4.9 ± 1.9 μM, saturation value of 0.33 ± 0.02 moles CaM/mole Cx, and a Hill coefficient of 0.69 ± 0.20. Data are means of 6 measurements ± 1 SD.

Figure 3. Calmodulin binding to Cx34.7 and Cx36 cytoplasmic domains

A. Steady state binding curves for the Cx34.7 CT and IL domains. Binding conditions, CaM concentrations, and data analysis are as in figure 1. Parameters for the two-component binding model fit to the Cx34.7 CT domain are shown in Table 2. B. Calcium-dependence of CaM binding to Cx34.7 CT domain. Conditions are as described in figure 2. The curve fit is a Hill equation with K_½ for Ca²⁺ of 3.4 ± 3.1 μM, saturation value of 0.33 ± 0.05 moles CaM/mole Cx, and a Hill coefficient of 0.52 ± 0.28. C. Steady state CaM binding curve for the Cx36 CT domain. Binding conditions, CaM concentrations, and data analysis are as in figure 1. Parameters for the two-component binding model fit to the Cx34.7 CT domain are shown in Table 2. D. Calcium-dependence of CaM binding to Cx36 CT domain. Conditions are as described in figure 2. The curve fit represents a Hill equation with K_½ for Ca²⁺ of 3.0 ± 0.6 μM, saturation value of 0.24 ± 0.01 moles CaM/mole Cx, and a Hill coefficient of 1.01 ± 0.18.

Figure 4. pH dependence of CaM binding to connexin CT domains

Calcium-dependence of steady-state binding of 1.6 μM CaM to the CT domains of Cx34.7 (A), Cx35 (B), and Cx36 (C) at three pH’s. Data shown are means of three measurements ± 1 SD.

Figure 5. Identification of a calmodulin binding site through binding competition

Steady state binding responses of Cx34.7 CT domain (A), Cx35 CT domain (B) and Cx36 CT domain (C) to injections of 1 μM CaM in 300 μM Ca²⁺ buffer in the presence of synthetic peptides matching portions of the Cx36 CT domain sequence. Peptide sequences are shown in Table 3 and are represented in the figure by the abbreviations P1–P4. The key in panel A applies to A–C. Binding responses are normalized to the response to 1 μM CaM in 300 μM Ca²⁺ buffer in the absence of peptide. IC₅₀’s calculated from the exponential curves fit to the data are shown in Table 3. D. Amino acid alignment of the full C-terminal domains of perch Cx35, mouse Cx36 and perch Cx34.7, beginning with the last three amino acids predicted to be within transmembrane domain 4. Symbols below the alignment indicate amino acids that are identically conserved (*) and those that have similar character (.). The positions of the peptides P1–P4 used for binding competition experiments are shown by bars above the alignment. A calmodulin binding site predicted by the binding competition experiment is indicated by the dashed line.