FEBS Letters 588 (2014) 1430–1438

journal homepage: www.FEBSLetters .org

Review

Gap junction regulation by calmodulin

0014-5793/$36.00 � 2014 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.febslet.2014.01.003

⇑ Corresponding authors.E-mail addresses: charles.louis@ucr.edu (C.F. Louis), jenny@gsu.edu (J.J. Yang).

Juan Zou a, Mani Salarian a, Yanyi Chen a, Richard Veenstra b, Charles F. Louis c,⇑, Jenny J. Yang a,⇑a Department of Chemistry, Center for Diagnostics and Therapeutics, Georgia State University, Atlanta, GA 30303, United Statesb Department of Pharmacology, SUNY Upstate Medical University, Syracuse, NY 13210, United Statesc Department of Cell Biology and Neuroscience, University of California, Riverside, CA 92521, United States

a r t i c l e i n f o a b s t r a c t

Article history:Received 5 December 2013Revised 9 January 2014Accepted 10 January 2014Available online 16 January 2014

Edited by Michael Koval, Brant E. Isakson,Robert G. Gourdie and Wilhelm Just

Keywords:Ca2+

ConnexinGap junction regulationCalmodulin binding

Intracellular Ca2+ activated calmodulin (CaM) inhibits gap junction channels in the low nanomolarto high micromolar range of [Ca2+]i. This regulation plays an essential role in numerous cellularprocesses that include hearing, lens transparency, and synchronized contractions of the heart.Previous studies have indicated that gap junction mediated cell-to-cell communication wasinhibited by CaM antagonists. More recent evidence indicates a direct role of CaM in regulatingseveral members of the connexin family. Since the intracellular loop and carboxyl termini ofconnexins are largely ‘‘invisible’’ in electron microscopy and X-ray crystallographic structures dueto disorder in these domains, peptide models encompassing the putative CaM binding sites ofseveral intracellular domains of connexins have been used to identify the Ca2+-dependent CaM bind-ing sites of these proteins. This approach has been used to determine the CaM binding affinities ofpeptides derived from a number of different connexin-subfamilies.� 2014 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction

Modulation of the cytoplasmic Ca2+ concentration ([Ca2+]i) is anubiquitous mechanism by which cells transduce external signalsinto biological responses. The signaling cascade initiated by therise in [Ca2+]i is often mediated via Ca2+-binding proteins such ascalmodulin (CaM) [1,2]. CaM, a key multifunctional transducer ofCa2+ signals in eukaryotes, has four EF-hand Ca2+ binding motifsin two globular N- and C-domains that are separated by a flexiblelinker. Upon Ca2+ binding, CaM undergoes a large conformationalchange, exposing hydrophobic patches that are important in itsbinding to more than 300 target proteins in multiple cellular pro-cesses [3–5]. CaM uses either its N- or C-domains with differentCa2+ binding affinities to differentiate between local and global[Ca2+]i changes, and to regulate a diverse group of membrane chan-nels/pumps that include cyclic nucleotide-gated (CNG) channels[6], N-methyl-D-aspartate receptor (NMDA receptor) [7], ryanodinereceptors (RyR) [8], Ca2+-activated K+ channels of small or interme-diate conductance (SK or IK) [9,10], Trp family channels [11], Ca2+

channels, and gap junction channels [12–16].Gap junctions are formed from the docking of paired hexameric

connexons, which are also called hemichannels with each connex-on comprised of six monomeric connexins. Approximately 21connexin isoforms have been identified in a multitude of humancell types [17]. According to their gene structure, homology and

specific sequence motifs, connexins have been grouped into threemajor categories, termed the a, b and c connexin subfamilies[18]. The different connexin subfamily members all share the samestructural topology with four a-helical transmembrane domainsconnected by two extracellular loops and a cytoplasmic loop; bothN- and C-terminal face the cytosol. Three highly conserved Cys res-idues residing in each extracellular loop form intraconnexin disul-fide bonds. The transmembrane domains and the extracellularloops are highly conserved among different connexin families.Although the N-terminus is also relatively conserved, the cytoplas-mic loop and C-terminus exhibit significant variation among differ-ent connexins.

Connexins are almost universally expressed in vertebrate tis-sues with the exception of several highly differentiated cell typesthat include skeletal muscle, erythrocytes, and mature sperm cells[19]. Gap junctions facilitate the direct intercellular cytoplasmicconnections that allow the exchange of ionic current, secondarymessengers, and small metabolites up to 1 kDa between the cyto-plasms of adjacent cells. Gap junctions play important roles inmany biological processes [20–22], including development, differ-entiation, cell synchronization, neuronal activity, and the immuneresponse. The functional loss of gap junction may result in embry-onic lethality (e.g. Cx45) [23,24], perinatal death due to congenitalmalformations (e.g. Cx43�/�) [25], and plays a major role in thepathogenesis of several other diseases (e.g. Charcot–Marie–Toothdisease, Cx32; Oculodentodigital dysplasia, Cx43; Non-syndromicdeafness, Cx26, Cx30, Cx43; lens cataracts, C46, Cx50) [26,27].Connexin mutations may interfere with gene expression, affect

J. Zou et al. / FEBS Letters 588 (2014) 1430–1438 1431

protein or DNA stability, impede protein trafficking, translocationand assembly, or alter the regulation of gap junction.

Intercellular communication mediated by gap junctions can beregulated by multiple factors that include [Ca2+]i, H+, CaM, voltage,and phosphorylation of the connexin subunits [28,29]. The role ofCaM in the regulation of gap junction channels has been previouslyreviewed by Peracchia [30]. In this paper, we review currentunderstanding of the Ca2+–CaM regulation of gap junctions com-prised of different connexins that is effected by the direct bindingof Ca2+–CaM to specific sequences of these connexins (Fig. 1). Wewill first discuss the functional regulation of gap junctions byCaM, we will then discuss the molecular basis of such regulationincluding the CaM binding locations, relative binding affinities,and binding modes for the different connexins, and finally, a gatingmodel for the regulation of the a-subfamily of connexins by Ca2+–CaM is proposed.

2. Functional regulation of gap junction by CaM

Over the years, a significant body of data supports the involve-ment of CaM in the regulation of gap junctions comprised of thethree connexin subfamilies. Délèze [31] first showed that Ca2+ isessential for the healing process in mammalian heart muscle bypreventing longitudinal diffusion of molecules in cardiac fibers.Subsequently, Rose et al. [32] showed that in salivary gland cells,inhibition of cell-to-cell coupling occurs when the intracellularconcentration of Ca2+ was increased from 0.1 to 50 lM.

Recognizing that most Ca2+-induced phenomena are mediatedby CaM [33], Peracchia and colleagues [34] first tested the hypoth-esis that gap junctions were regulated by CaM by examining theeffect of the CaM inhibitor trifluoperazine (TFP) on the electricalcoupling of amphibian embryonic cells exposed to CO2 to lowerintracellular pH. While the role of CaM in mediating the action ofCa2+ on enzyme activation was well known in the 1980s, gap

Fig. 1. Reported CaM-binding sites in rat Cx32, sheep Cx44, human Cx43 and 50. Two Cbinding sites in a-subfamily connexins are in the second half of the intracellular loop andof a hydrophobic residue. (⁄, Hydrophobic residues).

junctions represented the first membrane channels shown to bemodulated by CaM; only a decade later was the next CaM regulatedchannel identified [6]. Peracchia and colleagues demonstrated thatTFP reversibly inhibits the CO2-induced electrical uncoupling inamphibian embryo cells by interfering with the mechanism whichcloses the cell-to-cell channels. Subsequently, more specific CaMblockers (calmidazolium and W7) were shown to prevent uncou-pling of Xenopus embryonic cells [35] and crayfish axons [36,37],indicating the generalized nature of this role for CaM in regulatinggap junctions. This hypothesis was strengthened by evidence thatthe gap junction protein Cx32 bound CaM in gel overlays [38–40].The suppression of CaM expression in oocytes can also inhibit CO2

induced electrical uncoupling and injection of CaM into oocytescan recover it [41]. Cx32 was also shown to colocalize with CaMusing immunofluorescence microscopy [42]. Later, Blodow et al. re-ported that CaM antagonists suppress gap junction coupling of Cx26in isolated Hensen cells of the guinea pig cochlea [43].

Louis and Lurtz first showed that the gap junction mediated cell-to-cell transfer of dye between lens epithelial cells was inhibited byCa2+–CaM [44]. The rapid onset of this inhibition (within seconds)suggested that this inhibition was mediated by the direct interactionof CaM with one or more of the lens connexins rather than by the ac-tion of a CaM-dependent protein kinase. They subsequently demon-strated in Cx43-transfected HeLa cells that cell-to-cell dye transferwas inhibited by Ca2+–CaM [45]. Cell-to-cell communication washalf-maximally inhibited at�300 nM [Ca2+]i [46], and this inhibitionwas prevented by pre-incubation of lens cultures with CaM antago-nists [47]. In HeLa cells transiently expressing the CaM-binding-defi-cient mutants (Cx43K146E,R148E-EYFP and Cx43M147Q,L151E,I156E-EYFP),elevated [Ca2+]i was unable to inhibit cell–cell dye transfer, confirm-ing that residues 136–158 in the intracellular loop of Cx43 contain theCaM-binding site that mediates the Ca2+-dependent regulation ofCx43 gap junctions [48]. Direct gap junction conductance (Gj)measurements confirmed that increases in [Ca2+]i and decreases in

aM-binding sites are located at the N- and C-terminus of Cx32. The identified CaM-fit the 1–5–10 subclass binding mode, where each number represents the presence

1432 J. Zou et al. / FEBS Letters 588 (2014) 1430–1438

Cx43 Gj were temporally correlated and inhibitable by CaM antago-nists like calmidazolium, CaMKII CaMi peptide, or the Cx43p136–158

peptide (Fig. 2) [49]. We have further shown that inhibition of gapjunction conductance by intracellular Ca2+ and CaM can be reversedby 90% with the addition of 10 mM EGTA and removal of externalCaCl2 from the bath saline solution, but only if Ca2+-chelationcommenced prior to complete uncoupling [49]. On the otherhand, Cx40-transfected N2a neuroblastoma cell pairs were notuncoupled by intracellular Ca2+, such lack of intracellular Ca2+

regulation of Cx40 was consistent with the observation that Cx40does not contain the putative CaM binding site in the same cytosolicloop region [49].

The inhibition of Cx44-mediated cell-to-cell dye transfer byCa2+ was also shown to be regulated by CaM as this inhibitionwas prevented by prior incubation of Cx44-transfected cells witha CaM antagonist [50]. Cx50 was also shown to colocalize withCaM [51], and by measuring gap junctional conductance of Cx50-transfected N2a neuroblastoma cells, the reduction of Cx50-medi-ated junctional coupling was shown to be Ca2+-dependent. Junc-tional coupling mediated by either Cx50 or Cx43 was preventedby pre-incubation of transfected cells with a CaM inhibitor,indicating that the [Ca2+]i-dependent inhibition of Cx50 and Cx43was CaM mediated. Furthermore, the Ca2+-dependent inhibitionof gap junction permeability in Cx50-transfected cells wasprevented by intracellular injection of a synthetic peptide

-3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 120

100

200

300

N = 6 cell pairs

N = 65 cells 3 exps

+1 µM ionomycin (1.8 mM [Ca2+]o)

ControlSaline

ControlSaline

[Ca2+

] i (nM

)

Time (min)

A

-3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 120.0

0.4

0.8

1.2

Cx4

3 G

j

+1 µM ionomycin (1.8 mM [Ca2+]o)

0 4 8 12 16 200.0

0.4

0.8

1.2

p < 0.05

p < 10-5

Cx43-3 (n=6) Cx43-scr (n=6)

1 µM ionomycin+ 1.8 mM Ca2+

o

Cx43

Gj

Time (min)

1 µM Cx43 peptide

C

Fig. 2. Ca2+/CaM regulation of Cx43 gap junctions. (A) Separate fura-2 and two whole ceconductance (Gj) in Cx43–N2a cells illustrate the temporal decrease in Gj (upper panel) anPretreatment with 2 lM calmidazolium prevented the Ca2+–ionomycin induced reductionpeptide (D) in both patch pipettes at the indicated concentrations (1–2 Kd for CaM bindiXu et al. [49] with author’s permission from the American Physiological Society.

encompassing the CaM binding domain of Cx50, while the scram-bled peptide was without effect [52].

CaM regulation of the c-connexin Cx45 was reported by Perac-chia and coworkers by monitoring the sensitivity of Cx45 channelsto CO2, and inhibiting CaM expression in oocytes [53]. Mouse Cx36,perch Cx35 and Cx34.7 that form electrical synapses have alsobeen reported to bind CaM in a Ca2+-dependent manner usingsurface plasmon resonance assays and GST fusion proteinsharboring the carboxyl-domain of these connexins [54,55].

3. Molecular basis of CaM regulation of connexins

There are several challenges in developing an understanding ofthe molecular mechanism by which CaM regulates the functionalactivity of connexins. Thus, the crystallization of these membraneproteins for determination of their high resolution atomic struc-ture by X-ray crystallography is especially challenging. At thistime, the structure of only one connexin, Cx26 in a presumed openstate, has been resolved to 3.5 Å resolution [56,57], although lowerresolution electron microscopy crystallographic structures areavailable [58]. However, the intracellular loop and carboxyl termi-nus are largely ‘‘invisible’’ in these structures due to intrinsicdisorder in these domains. In addition, in whole cell experiments,it is difficult to unambiguously distinguish between the effects ofCa2+ directly on the channel and the effects of Ca2+ mediated via

0 4 8 12 16 200.0

0.4

0.8

1.2

N.S.

B

2 µM calmidazolium

1 µM ionomycin+ 1.8 mM Ca2+

o

Cx43

Gj

Time (min)

+ 15 minprewash N = 6

0 4 8 12 16 200.0

0.4

0.8

1.2

p < 10-5

1 µM ionomycin+ 1.8 mM Ca2+

o

Gj

Time (min)

Control Cx43 100 nM CaMKII

peptide (n=6)

DN.S.

ll patch clamp experimental measurements of [Ca2+]i and normalized gap junctiond increase in [Ca2+]i (lower panel) during 1 lM ionomycin perfusion (1 ml/min). (B)in Cx43 Gj. Inclusion of inhibitory CaM binding peptides Cx43p136–158 (C) or CaMKII

ng) also prevented the Ca2+–ionomycin induced reduction in Cx43 Gj. Adapted from

J. Zou et al. / FEBS Letters 588 (2014) 1430–1438 1433

CaM. CaM is a versatile and mobile trigger protein with manytarget proteins and is involved in both its direct and indirectregulations of its target proteins. To overcome these challenges,peptide models are often used to understand the molecular mech-anism by which CaM interacts with its target proteins. Typically anumber of questions are posed that include identification of thelocation(s) in the target proteins that bind CaM, whether theCaM binding to its target proteins is Ca2+-dependent or -indepen-dent, the consequences of the Ca2+ sensitivity of CaM on its inter-action with its target protein, and the CaM binding modes bywhich it interacts with its target proteins.

3.1. CaM binds to Cx32 in b-connexin subfamily

Equilibrium binding studies using a fluorescent CaM deriva-tive identified two CaM-binding domains in Cx32, a site in theN-terminal tail, and a site in the C-terminal tail region [59]. Apeptide encompassing the N-terminal domain of Cx32 wasshown to bind a fluorescent CaM derivative in a Ca2+-dependentmanner with high affinity (Kd, 27 nM). A peptide encompassing aC-terminal region of Cx32 was shown to bind this fluorescentCaM derivative with a Kd of 1.2 lM [59]. The CaM bindingdomains of the N- and C-terminal regions of Cx32 were bestdefined as residues 1–21 (Cx32NT) and 216–227 (Cx32CT),respectively with the Cx32CT CaM binding region showinga-helical propensity indicating that CaM binding induced ana-helical structure in the Cx32CT and involved both theN- and C-lobes of this connexin. These data, showing separatefunctions of the N- and C-lobes of CaM in its interactions withCx32, suggests trans-domain or trans-subunit bridging by CaMas a possible mechanism of gap junction gating [60]. Interest-ingly, both the N-terminal and the C-terminal CaM-binding do-mains of Cx32 are located close to the membrane and arecontiguous with hydrophobic membrane-spanning sequences(Fig. 1). Furthermore, although the C-terminal CaM bindingdomain in Cx32 represents only 15% of the C-terminal tail ofCx32, the removal of the other 85% had little effect on junctionalpermeability and chemical gating [61–63]. Recently, Stauch et al.further demonstrated that CaM binding could convert the intrin-sically disordered C-terminal domain to an a-helical conforma-tion that may enable Cx32 to interact with the protein partnersynapse-associated protein 97 [64].

3.2. CaM binds to the c-connexin subfamily

Using surface plasmon resonance with rapid kinetics, CaM wasshown to bind in a concentration- and Ca2+-dependent manner topeptide fragments corresponding to a 10–30 amino acid segmentat the beginning of the C-terminal intracellular domain of mouseCx36, Cx35 from perch, and the related perch Cx34.7 [55]. Dissoci-ation was also very rapid; Kds for CaM binding at a high-affinitysite ranged from 11 to 72 nM (Table 1). No binding of CaM to the

Table 1CaM-binding affinities of peptides derived from different subfamilies of connexins.

Peptide Family Position Predicted score

hCx43p138–157 a CL 16shCx44p129–150 a CL 6hCx50p141–166 a CL 13hCx32p1–21 b NT 9rCx32p216–230 b CT 0mCx36p269–321 c CT 14

a Predicted by CaM Target Database.b Peptide helicity induced by 90% TFE.

intracellular loops of these connexins was observed. The micromo-lar K½s, and the rapid on and off rates suggest that this interactionmay change dynamically in neurons, and may occur transientlywhen [Ca2+]i is elevated to a level that would occur in the nearvicinity of an activated synapse [54,55].

3.3. CaM binds to the a-connexin subfamily

We have predicted that a CaM binding region resides in the cen-tral cytoplasmic loop of a-connexins that include Cx43 [48], Cx44(the ovine analog of rodent Cx46) [50] and Cx50 [52,65]. We usedvarious biochemical approaches including circular dichroism (CD)nuclear magnetic resonance (NMR), isothermal titration calorime-try (ITC) and fluorescence spectroscopy to demonstrate that CaMbinds to the cytoplasmic loop of Cx43 [48], Cx44 [50] and Cx50[52] in a Ca2+-dependent manner. Fig. 3A shows the far UV CDspectra of CaM in the absence and presence of a Cx43 cytoplasmicloop peptide. The addition of this Cx43 peptide increased the a-helical content likely due to the conversion of unstructured freepeptide to helical conformation upon complex formation withCaM. On the other hand, based on the use of peptide models, no di-rect interaction was observed for other regions in this connexinsuch as the N-terminal region of Cx43 [59] and the C-terminaldomain of Cx43 [66].

Studies in our laboratories identified CaM binding domains inCx43 (residues 136–158) [48], Cx44 (residues 132–153) [50] andCx50 (residues 141–166) [52]. NMR studies demonstrated all threeCx43-, Cx44- or Cx50-derived peptides (Cx43p136–158, residuesfrom 136 to 158; Cx44p132–153, residues from 132 to 153;Cx50p141–166, residues from 141 to 166) bind CaM with a 1:1 stoi-chiometry. Such interactions of CaM with Cx43- or Cx50-derivedpeptides can also be observed using surface plasmon resonanceand florescence anisotropy. Dansylated CaM florescence andhigh-resolution NMR studies demonstrated that the binding of aCx43-derived peptide Cx43p136–158 to CaM in the presence butnot the absence of Ca2+. The apparent dissociation constant bindingof this peptide to CaM in physiologic K+ was 860 ± 20 nM [48]. Onthe other hand, the respective Kds for the peptides derived fromCx44 (Cx44p132–153) and Cx50 (Cx50p141–166) are 49 ± 3.0 and4.9 ± 0.6 nM respectively in the presence of Ca2+. In the absenceof Ca2+, both peptides bound to CaM with significantly reducedaffinities (Kds greater than 5000 and 8000 nM, respectively). Thus,Cx50p141–166 exhibited a 10-fold higher binding affinity for CaMthan Cx44p132–153, which was approximately 20-fold higher thanCx43p136–158.

Since a ‘‘ball-and-chain’’ or ‘‘particle-receptor’’ hypothesisinvolving the pH-dependent intramolecular interaction betweenthe cytoplasmic domain and part of the intracellular loop of Cx43has been proposed to explain the low pH-induced closure of gapjunction [66,67], we have also determined the CaM binding affinityof the Cx44 and Cx50 peptides to CaM as a function of pH. BothCx44p132–153 and Cx50p141–166 peptides behaved similarly when

a Helicityb Kd (nM) Refs.

Ca2+ EGTA

55.6 860 nd [48]59.5 49 >5000 [50]95.3 4.9 >8000 [52]nd 27 nd [60]nd 2100 nd [60]nd 11 nd [55]

-2.0

-1.6

-1.2

-8.0

-4.0

0.0

4.0

200 210 220 230 240 250 260

Apo-CaMApo-CaM-peptide

Ca2+-CaM

Ca2+-CaM-peptide

Wavelength (nm)

Mol

ar E

llipt

icity

( X

103 d

eg c

m2 d

mol

-1 re

s-1)

-9.0

-6.0

-3.0

0.0

3.0

200 208 216 224 232 240 248 256

Free Cx43Bound Cx43

Wavelength (nm)

Mol

ar E

llipt

icity

( X

103 d

eg c

m2 d

mol

-1 re

s-1)

0

20

40

60

80

100

120

0 20 40 60 80 100

Cx44 PeptideCx43 PeptideCx50 Peptide

Cal

cula

ted

-hel

ical

con

tent

(%)

TFE (%)

A B

Fig. 3. Circular dichroism studies of the interaction between Cx43p136–158 and Apo/Holo CaM. (A) Far UV circular dichroism spectrum of CaM in the presence of 1 mM EGTA(white circle) or 1 mM CaCl2 (white square) and a 1:1 CaM/synthetic peptide mixture with 1 mM EGTA (black circle) or 1 mM CaCl2 (black square) after subtracting thecontribution from the buffer. Buffer conditions: 10 mM Tris, 100 mM KCl at pH = 7.4 at room temperature. The inset demonstrates the far UV circular dichroism spectra ofCx43p136–158 (dashed line) and the calculated difference spectrum (solid line) by subtracting the spectrum of Ca2+–CaM from that of the Ca2+–CaM–Cx43p136–158 mixture with1 mM Ca2+ in a buffer consisting of 100 mM KCl, 10 mM Tris–HCl, at pH = 7.4. (B) TFE induced a-helical content increase of Cx43 (triangle), Cx44 (circle) and Cx50 (square)peptides corresponding to CaM-binding sites. Adapted from Zhou et al. [48,50] and Chen et al. [52] with authors’ permission from the Journal of Biological Chemistry,Biophysical Journal and Biochemical Journal.

1434 J. Zou et al. / FEBS Letters 588 (2014) 1430–1438

pH was varied with binding affinities decreasing at pH values low-er than 5.5 and greater than 8.5 [50]. However, over the pH range5.5–8.5, CaM exhibits affinities for both peptides at nanomolar orsubmicromolar range. CaM binding to the peptide Cx50p141–166

was �1 order of magnitude stronger than to the peptideCx44p132–153 likely due to difference in electrostatic interactions.

It is interesting to note that Ca2+–CaM may directly or indirectlyaffect cytoplasmic loop/C tail interactions in a Cx43 hemichannel,either through its binding to the cytoplasmic loop of Cx43 whichcorresponds to amino acids 119–144 in the C-terminal portion ofthe cytoplasmic loop of Cx43, or through a Ca2+-dependentphosphorylation/dephosphorylation processes [29,68,69]. It is alsopossible that phosphorylation or other regulatory mechanismssuch as pH could modulate this action of Ca2+–CaM. Our resultsprovide direct evidence that CaM binds to a specific region of theintracellular loop region of the gap junction proteins Cx43, Cx44and Cx50 in a Ca2+-dependent manner. Our data suggest a commonmechanism by which the Ca2+-dependent inhibition of the a-classof gap junction proteins is mediated by the direct association of anintracellular loop region of these proteins with Ca2+–CaM.

3.4. Key factors contributing to the CaM binding affinity of the a-subfamily of connexins

Our results have also revealed that the a-helicity of the CaMbinding peptide derived from connexin is an important factor inpredicting their CaM binding affinities (Table 1) although the bind-ing affinity of the full length connexin is not yet available. The dif-ference in CaM binding affinity of the Cx50 and Cx44 peptides mayoriginate from intrinsic sequences encoded in the CaM binding re-gions of these different connexins especially at conserved residuepositions 1 and 5 (Fig. 1); Cx43 peptide contains two flexible Glysinstead of the a-helical forming residues present in Cx44 and Cx50(Fig. 1). Such sequence variations, that affect the ability of thesepeptides to form a-helices, are likely important in determiningtheir ability to interact with CaM [65]. In aqueous buffer, thesepeptides are largely unstructured in the absence of CaM. 2,2,2-Tri-fluoroethanol (TFE) is an organic solvent which when added toaqueous buffer can reveal the intrinsic helical propensity of a

peptide [70]. We have shown that all these peptides form a-helicalstructures in the presence of TFE that follows the rank orderCx50p141–166 > Cx44p132–153 � Cx43p136–158 (Fig. 3B). The higherCaM binding affinity of Cx44 versus that of Cx43 may also bedue in part to the relative helical content of these two peptidesin solution indicating that the intrinsic helicity of the peptidemay also contribute to this binding. The negatively charged E151located close to R149 in Cx50 may stabilize the a-helical conforma-tion of this peptide, while Cx43 with a positive charged R at thesame location may destabilize the a-helix conformation of thisconnexin [48,52].

There are several positively charged residues such as K147,R149 and R156 in Cx50 that are conserved in the a-connexinsubfamily members Cx43, Cx44 and Cx50. Indeed, the observedpH dependence of CaM binding affinity to the cytosolic looppeptides of Cx44 and Cx50 suggests electrostatic interactionsfollowing the protonation of Asp and Glu, and the deprotonationof Lys, are important for the connexin–CaM complex formation[50,52].

It is also noteworthy that the binding affinity of CaM to theconnexin-derived CaM binding peptides derived from the cytosolicregion of three a-subfamily connexins does not agree with theirpredicted CaM binding scores based on the CaM Target Database[65]. These predicted scores are based on parameters includingaverage hydrophobicity, average hydrophobic moment, and aver-age propensity for a-helix formation and are not intended to definethe CaM binding affinity for the complex formation [65]. It is inter-esting to note that the reported CaM binding site in the N-terminalregion of Cx32 (residues 1–21) is predicted to have a high score bythe CaM target database. This region is likely a-helical since it ishomologous to the NTH motif (N-terminal helix) in the X-ray-determined structure of Cx26 [57]; it is believed to form thechannel ‘‘plug’’ in the b-subfamily of connexins [71]. However,the reported C-terminal CaM binding region of Cx32 is not pre-dicted by the CaM target database and is largely undefined in thedetermined Cx26 X-ray-derived structure. Therefore it is importantto perform detailed experimental studies to define the actual CaMbinding sites in its target proteins especially for membranechannel proteins such as the connexins.

J. Zou et al. / FEBS Letters 588 (2014) 1430–1438 1435

3.5. CaM binding modes to the a-connexin subfamily using a peptidemodel

One of the key aspects to understand the molecular mechanismby which CaM interacts with its target proteins is the CaM bindingmodes, by which it interacts with its target proteins. Currently,over 200 structures of CaM complexed with its receptors arereported in the protein data bank. CaM exhibits different targetrecognition modes that can be classified into two general bindingstyles, namely extended and collapsed [48]. In the extended mode,CaM has different recognition modes when it binds to different tar-get proteins such as the IQ motif of myosin V Ca2+ channels [48],anthrax toxin [50], SK channels (Small conductance Ca2+-activatedpotassium channels) [52], Ca2+ pump [40], and glutamate decar-boxylase [39]. In the collapsed binding style, CaM can be furtherdivided into different binding modes depending on the spacesbetween conserved hydrophobic residues at the CaM bindingregions. For example, CaM binds to N-methyl-D-aspartate receptors(NMDARs) in the C1 region of its NR1 subunit in a 1–7 bindingmode [72]. A 1–10 binding mode has been reported for CaM bind-ing to CaM Kinase II a [73] and the Ca2+ channel CaV1.2 [74]. A1–5–10 CaM binding mode was observed for the CaM complexwith myristoylated alanine-rich C-kinase substrate (MARCKS)and CaMKII [75]. A 1–14 binding mode was observed in the CaMcomplex with cyclic nucleotide-gated ion channels (CNG) [76],Endothelial nitric-oxide synthase (eNOS) [6], CaM kinase I [7], amyosin light chain kinase that is exclusively expressed in adultskeletal muscle (skMLCK) [8], and metabotropic glutamate recep-tors (mGluR7a) [9]. Ca2+/CaM-dependent protein kinase kinase(CaMKK) [7] has a 1–16 binding mode. The ryanodine receptor 1(RyR1) [10] has 1–17 binding mode. The CaM binding regions ofsuch CaM target proteins generally have a helical conformationon complex formation with CaM.

We have also used peptide models to understand the structuralmechanisms driving CaM regulation of connexins with the advan-tage of obtaining site specific molecular recognition information.Several of our studies using various spectroscopic methods suggestthat CaM binds to Cx50p141–166, Cx44p132–153, or Cx43p136–158 witha 1–5–10 collapsed CaM binding mode as observed in theCaM–CaMKII complex. All three peptides are largely unstructuredin aqueous buffer in the absence of CaM. Far UV circular dichroismstudies have indicated that these CaM binding peptides likelybecome more helical upon binding CaM. Fluorescence spectros-copy revealed conformational changes of both the peptide andCaM following formation of the CaM:connexin-derived-peptidecomplex. Pulse-field gradient NMR studies demonstrated thehydrodynamic radius of CaM (22.6 ± 0.6 Å) was decreased by 23%upon formation of the CaM–Cx50p141–166 complexes. The deter-mined size of the Ca2+/CaM–Cx50p141–166 complex of 17.3 ± 0.6 Åis comparable to the hydrodynamic radii of CaM in complex withCaM–CaMKIIa (PDB code 1CDM) [77]. Furthermore, chemical shiftchanges of 15N-labeled CaM have been used to show that thesethree connexin peptides are able to induce structural changes inboth the N- and C-terminal domains of CaM as well as in the linkerregion. Using small angle X-ray scattering, Myllykoski et al. alsoobserved that on binding a Cx43 cytoplasmic loop peptide (resi-dues 144–158), CaM adopted a more globular conformation andcollapsed the target [78]. In summary, these studies suggest thatCaM likely embraces the peptides of a-connexins in a collapsedstructure that involves the unwinding of the CaM central helix.

Because of its abundance, the mammalian water-channel aqu-aporin-0 (AQP0, or MIP26), was long considered to be the gap junc-tion protein in the mammalian lens [79]. Like the connexins,aquaporin-0 channels were shown to close when bound with Ca/CaM [80–86]. More recent biochemical and NMR studies suggestthat one CaM binds two AQP0 CBD peptides in a stepwise manner

and two CaM molecules bind to a single AQP0 tetramer [83]. The3D computational reconstruction of full-length AQP0-CaM com-plex based on the electron microscopy map and the determinedX-ray structure of the plant glutamate decarboxylase (ptGAD)–CaM complex [87] suggest that two Ca2+–CaM are located directlyunder two of the AQP0 subunits and bind to two adjacent antipar-allel C-terminal helices from two subunits of AQP0. This CaM bind-ing mode to the water channel aquaporin-0 (AQP0) with a ratio of1:2 is different from our observed CaM binding mode of thea-connexin subfamily which using peptide models have a 1:1binding ratio. It would be interesting to observe how CaM is ableto regulate the full length connexins using the same action mode.However, in the fully assembled gap junction, these connexin cyto-plasmic loop CaM binding domains likely form a six-sided barrelnear the M3 (third transmembrane) interface that may stericallylimit the number of CaM molecules that may bind to this motif.The cytoplasmic loop and C terminal domains are thought todimerize to close Cx43 gap junctions by a pH-dependent chemicalgating mechanism [88]. Conversely, the six NT domains that foldinto the cytoplasmic pore of the gap junction channel thatcontribute to the formation of the ion permeation pathway andtransjunctional voltage (Vj) sensor apparently require the ‘‘gating’’of only one NT domain to induce partial closure of the channel viathe fast Vj-gating mechanism [57,89,90]. The connexin–CaMstoichiometry of binding required for gap junction channel closureremains to be determined, but is likely to require only 1–2connexin subunits, based on current knowledge of the fastVj- and slow chemical gating mechanisms.

3.6. Expanding the [Ca2+]i sensing range of CaM upon binding toconnexins

We have shown that when connexin peptides bind to CaM, theapparent Kds of Ca2+ for CaM decreased and the Hill Coefficients in-creased. Isothermal titration studies of the Cx44p132–153 peptidesuggest that its interaction with CaM is an exothermic event thatis both enthalpically and entropically driven in which electrostaticinteractions play an important role. The binding of Cx44129–150

peptide to CaM increases the CaM intradomain cooperativity, andfurther enhances the Ca2+-binding affinities of the C-domain ofCaM by slowing the rate of Ca2+ release from the complex [50].CaM is also able to decrease the Ca2+ affinity of the N-lobe ofCx50p141–166-bound CaM 2-fold, whereas the Ca2+ affinity of theC-lobe increased by �20%. CaM responds to subtle changes in[Ca2+]i over a broader range of [Ca2+]i concentrations as a resultof the binding of Cx50p141–166 [52].

3.7. Proposed Ca2+–CaM gating model

Results from our studies on the changes in Ca2+-binding kinetics,affinities, energies, dye-transfer and gap junction conductance be-tween cells support the following working model of gap junctioninhibition by Ca2+–CaM (Fig. 4). Initially, an increase in [Ca2+]i issensed by the C-domain of CaM which effects the interaction ofCaM with the cytoplasmic loop of the a-connexin subfamily. Thisinteraction enhances the efficiency and sensitivity of intracellularCa2+ sensing by CaM because of an increase in both the Ca2+-bindingaffinity and intradomain cooperativity of the C-domain of CaM. Thepartially saturated, connexin-bound CaM might serve as an inter-mediate state to prevent the free diffusion of CaM into the cyto-plasm. Further increase in [Ca2+]i to the near micromolar rangeallows the half-saturated CaM to rapidly respond to this further in-crease in [Ca2+]i such that it adopts a fully open conformation whichis now able to inhibit gap junction mediated intercellular communi-cation [48,50]. Inhibition of the interaction of the cytosolic loopregion with CaM by the addition of a CaM inhibitor (e.g. CDZ) or a

Fig. 4. A proposed gating model for the mechanism of the closure of Cx43 gap junction channels by direct Ca2+/CaM binding. In this model, connexin has one open (Mo) stateand one closed state (M1). The induction of a conformational change in the cytoplasmic loop (CL) of Cx43 through Ca2+/CaM binding, steric hindrance of CaM binding near thecytoplasmic opening of the transmembrane pore, or a combination of both steric hindrance and conformational gating mechanisms causes the closure of the channel.Addition of the Cx43-CL-derived peptides or a CaMi (CaM inhibitor e.g., CDZ), prevents the Ca2+/CaM–Cx43 cytoplasmic loop interaction which changes the state of thechannel from closed (M1) to the open (M0) state.

1436 J. Zou et al. / FEBS Letters 588 (2014) 1430–1438

Cx43 cytosolic loop peptide prevents the open (M0) to close (M1)state transition. The Ca2+–CaM induced gating response that closesthe Cx43 gap junction is likely effected via a conformational changein the cytosolic loop of Cx43 [48]. It is also possible that, as proposedearlier by Peracchia and colleagues [91], Ca2+–CaM binds to theN- and C-terminal domains of Cx32 to act as a ‘‘cork’’, or it inducesa conformational change in the cytosolic domain that occludes thecytosolic mouth of the gap junction channel, resulting in the restric-tion of the passage of current-carry ions [60,92]. Unwin and Ennisproposed an alternative ‘‘iris’’ gating hypothesis that the extracellu-lar Ca2+-induced closure of liver gap junctions is a result of adecrease in the tilt angle of six connexins within a connexon junc-tion channel [93]. This model relied on the presence of the ‘‘bulky’’phenylalanines in the third transmembrane domain (M3), the majorchannel lining domain. Unwin and Ennis’ idea was that the phenyl-alanine residues would gate the channel by obstructing its lumen, asthey would be brought into the lumen by the M3 rotation. However,combined mutations of Cx32’s phenylalanine145 and phenylala-nine149 residues (the most conserved phenylalanines among con-nexins) to valine (a much smaller hydrophobic residue) generatedchannels with gating sensitivities that were indistinguishable fromCx32 wild type channels [30].

In summary, due to the challenges associated with large oligo-meric membrane proteins and the ‘‘invisibility’’ of the cytosolicloop and C-terminal domains within a X-ray-derived structuresof these proteins [56,57], the mechanism by which CaM regulatesgap junction channels remains to be fully defined. However, sev-eral recent studies provide strong evidence for a direct role ofCaM in the regulation of several classes of connexin gap junctionchannels. While CaM was reported to bind to the N- and C-termi-nal regions of the b-subfamily Cx32, it interacts with the cytosolicloop region of three a-subfamily connexins Cx43, Cx44 and Cx50.

Cx35 and Cx36, and Cx34.7 are regulated by CaM in a Ca2+-depen-dent manner through binding to only their C-terminal domains[55]. Furthermore, all these different connexins are regulated byCaM in a Ca2+-dependent manner. The gated closure of the a-sub-family of gap junction channels by Ca2+ is effected by Ca2+–CaMbinding to connexins via an embracing CaM-binding mode withhydrophobic residues in the CaM-binding region at positions 1, 5,and 10. Electrostatic and hydrophobic interactions play an indis-pensable role in the intrinsic a-helicity of the CaM bindingdomains of a-subfamily connexins. Future studies are needed toaddress the stoichiometry of CaM binding to the full length conn-exin proteins, the intracellular Ca2+ affinity, and the connexin/CaMinduced conformational changes required to induce the chemicalgating mechanism. Furthermore, innovative approaches are re-quired to reveal the novel mode of CaM action at the membraneinterface via its association with flexible intracellular connexin re-gions resulting in the large conformational changes required forregulating the function of these membrane channels [94]. Sinceseveral spontaneous mutations have been identified in the CaMbinding site of the cytosolic loop of Cx43 (such as in oculodento-digital dysplasia), understanding the mechanisms by which intra-cellular Ca2+–CaM regulates gap junctions is an essential steptowards revealing the molecular bases of such connexin-linkeddiseases.

Acknowledgements

We thank Gina Sosinsky, and Yubin Zhou for their helpful dis-cussions and Kathy Meenach for her help. Juan Zou is a fellow ofBrain and Behavior Program at Georgia State University. This workwas supported, in part, by NIH Grants EY-05684 to C.F.L. and J.J.Y.,HL-042220 to R.D.V. and J.J.Y., and GM-081749 to J.J.Y.

J. Zou et al. / FEBS Letters 588 (2014) 1430–1438 1437

References

[1] Klevit, R.E., Dalgarno, D.C., Levine, B.A. and Williams, R.J. (1984) 1H-NMRstudies of calmodulin. The nature of the Ca2+-dependent conformationalchange. Eur. J. Biochem. 139, 109–114.

[2] Wang, C.L. (1985) A note on Ca2+ binding to calmodulin. Biochem. Biophys.Res. Commun. 130, 426–430.

[3] Bhattacharya, S., Bunick, C.G. and Chazin, W.J. (2004) Target selectivity in EF-hand calcium binding proteins. Biochim. Biophys. Acta 1742, 69–79.

[4] Kawasaki, H. and Kretsinger, R.H. (1995) Calcium-binding proteins 1: EF-hands. Protein Profile 2, 297–490.

[5] Ikura, M. and Ames, J.B. (2006) Genetic polymorphism and proteinconformational plasticity in the calmodulin superfamily: two ways topromote multifunctionality. Proc. Natl. Acad. Sci. U S A 103, 1159–1164.

[6] Hsu, Y.T. and Molday, R.S. (1993) Modulation of the cGMP-gated channel ofrod photoreceptor cells by calmodulin. Nature 361, 76–79.

[7] Ehlers, M.D., Zhang, S., Bernhadt, J.P. and Huganir, R.L. (1996) Inactivation ofNMDA receptors by direct interaction of calmodulin with the NR1 subunit. Cell84, 745–755.

[8] Rodney, G.G., Williams, B.Y., Strasburg, G.M., Beckingham, K. and Hamilton, S.L.(2000) Regulation of RYR1 activity by Ca2+ and calmodulin. Biochemistry 39,7807–7812.

[9] Xia, X.M. et al. (1998) Mechanism of calcium gating in small-conductancecalcium-activated potassium channels. Nature 395, 503–507.

[10] Fanger, C.M. et al. (1999) Calmodulin mediates calcium-dependent activationof the intermediate conductance KCa channel, IKCa1. J. Biol. Chem. 274, 5746–5754.

[11] Tang, J., Lin, Y., Zhang, Z., Tikunova, S., Birnbaumer, L. and Zhu, M.X. (2001)Identification of common binding sites for calmodulin and inositol 1,4,5-trisphosphate receptors on the carboxyl termini of trp channels. J. Biol. Chem.276, 21303–21310.

[12] Zuhlke, R.D. and Reuter, H. (1998) Ca2+-sensitive inactivation of L-type Ca2+

channels depends on multiple cytoplasmic amino acid sequences of thealpha1C subunit. Proc. Natl. Acad. Sci. U S A 95, 3287–3294.

[13] Schonherr, R., Lober, K. and Heinemann, S.H. (2000) Inhibition of human etherà go-go potassium channels by Ca2+/calmodulin. EMBO J. 19, 3263–3271.

[14] Bers, D.M. and Grandi, E. (2009) Calcium/calmodulin-dependent kinase IIregulation of cardiac ion channels. J. Cardiovasc. Pharmacol. 54, 180–187.

[15] Dai, S., Hall, D.D. and Hell, J.W. (2009) Supramolecular assemblies andlocalized regulation of voltage-gated ion channels. Physiol. Rev. 89, 411–452.

[16] Tadross, M.R., Dick, I.E. and Yue, D.T. (2008) Mechanism of local and globalCa2+ sensing by calmodulin in complex with a Ca2+ channel. Cell 133, 1228–1240.

[17] Sohl, G. and Willecke, K. (2004) Gap junctions and the connexin proteinfamily. Cardiovasc. Res. 62, 228–232.

[18] Eiberger, J., Degen, J., Romualdi, A., Deutsch, U., Willecke, K. and Sohl, G. (2001)Connexin genes in the mouse and human genome. Cell Commun. Adhes. 8,163–165.

[19] Nielsen, M.S., Nygaard Axelsen, L., Sorgen, P.L., Verma, V., Delmar, M. andHolstein-Rathlou, N.H. (2012) Gap junctions. Compr. Physiol. 2, 1981–2035.

[20] Levin, M. (2007) Gap junctional communication in morphogenesis. Prog.Biophys. Mol. Biol. 94, 186–206.

[21] Saez, J.C., Berthoud, V.M., Branes, M.C., Martinez, A.D. and Beyer, E.C. (2003)Plasma membrane channels formed by connexins: their regulation andfunctions. Physiol. Rev. 83, 1359–1400.

[22] Sarieddine, M.Z., Scheckenbach, K.E., Foglia, B., Maass, K., Garcia, I., Kwak, B.R.and Chanson, M. (2009) Connexin43 modulates neutrophil recruitment to thelung. J. Cell Mol. Med. 13, 4560–4570.

[23] Kruger, O. et al. (2000) Defective vascular development in connexin 45-deficient mice. Development 127, 4179–4193.

[24] Kumai, M., Nishii, K., Nakamura, K., Takeda, N., Suzuki, M. and Shibata, Y.(2000) Loss of connexin45 causes a cushion defect in early cardiogenesis.Development 127, 3501–3512.

[25] Reaume, A.G. et al. (1995) Cardiac malformation in neonatal mice lackingconnexin43. Science 267, 1831–1834.

[26] Lai-Cheong, J.E., Arita, K. and McGrath, J.A. (2007) Genetic diseases ofjunctions. J. Invest. Dermatol. 127, 2713–2725.

[27] Laird, D.W. (2008) Closing the gap on autosomal dominant connexin-26 andconnexin-43 mutants linked to human disease. J. Biol. Chem. 283, 2997–3001.

[28] Del Corsso, C., Iglesias, R., Zoidl, G., Dermietzel, R. and Spray, D.C. (2012)Calmodulin dependent protein kinase increases conductance at gap junctionsformed by the neuronal gap junction protein connexin36. Brain Res. 1487,69–77.

[29] Bennett, M.V., Barrio, L.C., Bargiello, T.A., Spray, D.C., Hertzberg, E. and Saez, J.C.(1991) Gap junctions: new tools, new answers, new questions. Neuron 6,305–320.

[30] Peracchia, C. (2004) Chemical gating of gap junction channels; roles ofcalcium, pH and calmodulin. Biochim. Biophys. Acta 1662, 61–80.

[31] Deleze, J. (1965) Electrophysiology of the HeartCalcium ions and the healing-over in heart fibers, pp. 147–148, Pergamon Press, Elmsford, New York.

[32] Rose, B., Simpson, I. and Lowenstein, W.R. (1977) Calcium ion produces gradedchanges in permeability of membrane channels in cell junctions. Nature(London) 267, 625–627.

[33] Cheung, W.G. (1980) Calmodulin plays a pivotal role in cellular regulation.Science 207, 19–27.

[34] Peracchia, C., Bernardini, G. and Peracchia, L.L. (1983) Is calmodulin involvedin the regulation of gap junction permeability? Pfluegers Arch. 399, 152–154.

[35] Peracchia, C. and Bernardini, G. (1984) Gap junction structure and cell-to-cellcoupling regulation: is there a calmodulin involvement? Fed. Proc. 43, 2681–2691.

[36] Peracchia, C. (1984) Communicating junctions and calmodulin: inhibition ofelectrical uncoupling in Xenopus embryo by calmidazolium. J. Membr. Biol.81, 49–58.

[37] Peracchia, C. (1987) Calmodulin-like proteins and communicating junctions.Electrical uncoupling of crayfish septate axons is inhibited by the calmodulininhibitor W7 and is not affected by cyclic nucleotides. Pfluegers Arch. Eur. J.Physiol. 408, 379–385.

[38] Hertzberg, E.L. and Gilula, N.B. (1982) Liver gap junctions and lens fiberjunctions: comparative analysis and calmodulin interaction. Cold SpringHarbor Symp. Quant. Biol. 46 (Pt 2), 639–645.

[39] Van Eldik, L.J., Hertzberg, E.L., Berdan, R.C. and Gilula, N.B. (1985) Interactionof calmodulin and other calcium-modulated proteins with mammalian andarthropod junctional membrane proteins. Biochem. Biophys. Res. Commun.126, 825–832.

[40] Zimmer, D.B., Green, C.R., Evans, W.H. and Gilula, N.B. (1987) Topologicalanalysis of the major protein in isolated intact rat liver gap junctions and gapjunction-derived single membrane structures. J. Biol. Chem. 262, 7751–7763.

[41] Peracchia, C., Wang, X., Li, L. and Peracchia, L.L. (1996) Inhibition of calmodulinexpression prevents low-pH-induced gap junction uncoupling in Xenopusoocytes. Pfluegers Arch. 431, 379–387.

[42] Sotkis, A., Wang, X.G., Yasumura, T., Peracchia, L.L., Persechini, A., Rash, J.E. andPeracchia, C. (2001) Calmodulin colocalizes with connexins and plays a directrole in gap junction channel gating. Cell Commun. Adhes. 8, 277–281.

[43] Blodow, A., Ngezahayo, A., Ernst, A. and Kolb, H.A. (2003) Calmodulinantagonists suppress gap junction coupling in isolated Hensen cells of theguinea pig cochlea. Pfluegers Arch. 446, 36–41.

[44] Lurtz, M.M. and Louis, C.F. (2003) Calmodulin and protein kinase C regulategap junctional coupling in lens epithelial cells. Am. J. Physiol. Cell Physiol. 285,C1475–C1482.

[45] Lurtz, M.M. and Louis, C.F. (2007) Intracellular calcium regulation ofconnexin43. Am. J. Physiol. Cell Physiol. 293, C1806–C1813.

[46] Churchill, G.C., Lurtz, M.M. and Louis, C.F. (2001) Ca2+ regulation of gapjunctional coupling in lens epithelial cells. Am. J. Physiol. Cell Physiol. 281,C972–C981.

[47] Lurtz, M.M. and Louis, C.F. (2003) Calmodulin and Protein Kinase C regulategap junctional coupling in lens epithelial cells. Am. J. Physiol. Cell Physiol. 285,C1475–C1482.

[48] Zhou, Y. et al. (2007) Identification of the calmodulin binding domain ofconnexin 43. J. Biol. Chem. 282, 35005–35017.

[49] Xu, Q., Kopp, R.F., Chen, Y., Yang, J.J., Roe, M.W. and Veenstra, R.D. (2012)Gating of connexin 43 gap junctions by a cytoplasmic loop calmodulin bindingdomain. Am. J. Physiol. Cell Physiol. 302, C1548–C1556.

[50] Zhou, Y., Yang, W., Lurtz, M.M., Chen, Y., Jiang, J., Huang, Y., Louis, C.F. andYang, J.J. (2009) Calmodulin mediates the Ca2+-dependent regulation of Cx44gap junctions. Biophys. J. 96, 2832–2848.

[51] Zhang, X. and Qi, Y. (2005) Role of intramolecular interaction in connexin50:mediating the Ca2+-dependent binding of calmodulin to gap junction. Arch.Biochem. Biophys. 440, 111–117.

[52] Chen, Y. et al. (2011) Molecular interaction and functional regulation ofconnexin50 gap junctions by calmodulin. Biochem. J. 435, 711–722.

[53] Peracchia, C., Young, K.C., Wang, X.G. and Peracchia, L.L. (2003) Is the voltagegate of connexins CO2-sensitive? Cx45 channels and inhibition of calmodulinexpression. J. Membr. Biol. 195, 53–62.

[54] O’Brien, J., Bruzzone, R., White, T.W., Al-Ubaidi, M.R. and Ripps, H. (1998)Cloning and expression of two related connexins from the perch retina definea distinct subgroup of the connexin family. J. Neurosci. 18, 7625–7637.

[55] Burr, G.S., Mitchell, C.K., Keflemariam, Y.J., Heidelberger, R. and O’Brien, J.(2005) Calcium-dependent binding of calmodulin to neuronal gap junctionproteins. Biochem. Biophys. Res. Commun. 335, 1191–1198.

[56] Suga, M., Maeda, S., Nakagawa, S., Yamashita, E. and Tsukihara, T. (2009) Adescription of the structural determination procedures of a gap junctionchannel at 3.5 A resolution. Acta Crystallogr., D: Biol. Crystallogr. 65, 758–766.

[57] Maeda, S., Nakagawa, S., Suga, M., Yamashita, E., Oshima, A., Fujiyoshi, Y. andTsukihara, T. (2009) Structure of the connexin 26 gap junction channel at 3.5 Aresolution. Nature 458, 597–602.

[58] Oshima, A., Tani, K., Hiroaki, Y., Fujiyoshi, Y. and Sosinsky, G.E. (2007) Three-dimensional structure of a human connexin26 gap junction channel reveals aplug in the vestibule. Proc. Natl. Acad. Sci. U S A 104, 10034–10039.

[59] Torok, K., Stauffer, K. and Evans, W.H. (1997) Connexin 32 of gap junctionscontains two cytoplasmic calmodulin-binding domains. Biochem. J. 326 (Pt 2),479–483.

[60] Dodd, R., Peracchia, C., Stolady, D. and Torok, K. (2008) Calmodulin associationwith connexin32-derived peptides suggests trans-domain interaction inchemical gating of gap junction channels. J. Biol. Chem. 283, 26911–26920.

[61] Wang, X., Li, L., Peracchia, L.L. and Peracchia, C. (1996) Chimeric evidence for arole of the connexin cytoplasmic loop in gap junction channel gating.Pfluegers Arch. Eur. J. Physiol. 431, 844–852.

[62] Werner, R., Levine, E., Rabadan-Diehl, C. and Dahl, G. (1991) Gating propertiesof connexin32 cell–cell channels and their mutants expressed in Xenopusoocytes. Proc. Biol. Sci. 243, 5–11.

1438 J. Zou et al. / FEBS Letters 588 (2014) 1430–1438

[63] Wang, X. and Peracchia, C. (1997) Positive charges of the initial C-terminusdomain of Cx32 inhibit gap junction gating sensitivity to CO2. Biophys. J. 73,C1743–C1749.

[64] Stauch, K., Kieken, F. and Sorgen, P. (2012) Characterization of the structureand intermolecular interactions between the connexin 32 carboxyl-terminaldomain and the protein partners synapse-associated protein 97 andcalmodulin. J. Biol. Chem. 287, 27771–27788.

[65] Yap, K.L., Kim, J., Truong, K., Sherman, M., Yuan, T. and Ikura, M. (2000)Calmodulin target database. J. Struct. Funct. Genomics 1, 8–14.

[66] Duffy, H.S., Sorgen, P.L., Girvin, M.E., O’Donnell, P., Coombs, W., Taffet, S.M.,Delmar, M. and Spray, D.C. (2002) PH-dependent intramolecular binding andstructure involving Cx43 cytoplasmic domains. J. Biol. Chem. 277, 36706–36714.

[67] Morley, G.E., Taffet, S.M. and Delmar, M. (1996) Intramolecular interactionsmediate pH regulation of connexin43 channels. Biophys. J. 70, 1294–1302.

[68] Chin, D. and Means, A.R. (2000) Calmodulin: a prototypical calcium sensor.Trends Cell Biol. 10, 322–328.

[69] De Vuyst, E. et al. (2009) Ca2+ regulation of connexin 43 hemichannels in C6glioma and glial cells. Cell Calcium 46, 176–187.

[70] Yang, J.J., Buck, M., Pitkeathly, M., Kotik, M., Haynie, D.T., Dobson, C.M. andRadford, S.E. (1995) Conformational properties of four peptides spanning thesequence of hen lysozyme. J. Mol. Biol. 252, 483–491.

[71] Oshima, A., Tani, K., Hiroaki, Y., Fujiyoshi, Y. and Sosinsky, G.E. (2008)Projection structure of a N-terminal deletion mutant of connexin 26 channelwith decreased central pore density. Cell Commun. Adhes. 15, 85–93.

[72] Saimi, Y. and Kung, C. (2002) Calmodulin as an ion channel subunit. Annu. Rev.Physiol. 64, 289–311.

[73] Toutenhoofd, S.L. and Strehler, E.E. (2000) The calmodulin multigenefamily as a unique case of genetic redundancy: multiple levels ofregulation to provide spatial and temporal control of calmodulin pools?Cell Calcium 28, 83–96.

[74] Carafoli, E. (1987) Intracellular calcium homeostasis. Annu. Rev. Biochem. 56,395–433.

[75] Klee, C.B. and Vanaman, T.C. (1982) Calmodulin. Adv. Protein Chem. 35, 213–321.

[76] Kakiuchi, S., Yasuda, S., Yamazaki, R., Teshima, Y., Kanda, K., Kakiuchi, R. andSobue, K. (1982) Quantitative determinations of calmodulin in thesupernatant and particulate fractions of mammalian tissues. J. Biochem. 92,1041–1048.

[77] Meador, W.E., Means, A.R. and Quiocho, F.A. (1993) Modulation of calmodulinplasticity in molecular recognition on the basis of X-ray structures. Science262, 1718–1721.

[78] Myllykoski, M., Kuczera, K. and Kursula, P. (2009) Complex formation betweencalmodulin and a peptide from the intracellular loop of the gap junctionprotein connexin43: molecular conformation and energetics of binding.Biophys. Chem. 144, 130–135.

[79] Benedetti, E.L., Dunia, I., Bentzel, C.J., Vermorken, A.J., Kibbelaar, M. andBloemendal, H. (1976) A portrait of plasma membrane specializations in eyelens epithelium and fibers. Biochim. Biophys. Acta. 457, 353–384.

[80] Girsch, S.J. and Peracchia, C. (1991) Calmodulin interacts with a C-terminuspeptide from the lens membrane protein MIP26. Curr. Eye Res. 10, 839–849.

[81] Peracchia, C. and Girsch, S.J. (1989) Calmodulin site at the C-terminus of theputative lens gap junction protein MIP26. Lens Eye Toxic Res. 6, 613–621.

[82] Reichow, S.L. and Gonen, T. (2008) Noncanonical binding of calmodulin toaquaporin-0: implications for channel regulation. Structure 16, 1389–1398.

[83] Reichow, S.L., Clemens, D.M., Freites, J.A., Nemeth-Cahalan, K.L., Heyden, M.,Tobias, D.J., Hall, J.E. and Gonen, T. (2013) Allosteric mechanism of water-channel gating by Ca2+-calmodulin. Nat. Struct. Mol. Biol. 20, 1085–1092.

[84] Peracchia, C. and Girsch, S.J. (1985) Permeability and gating of lens gapjunction channels incorporated into liposomes. Curr. Eye Res. 4, 431–439.

[85] Girsch, S.J. and Peracchia, C. (1985) Lens cell-to-cell channel protein: II.Conformational change in the presence of calmodulin. J. Membr. Biol. 83, 227–233.

[86] Girsch, S.J. and Peracchia, C. (1985) Lens cell-to-cell channel protein: I. Self-assembly into liposomes and permeability regulation by calmodulin. J.Membr. Biol. 83, 217–225.

[87] Yap, K.L., Yuan, T., Mal, T.K., Vogel, H.J. and Ikura, M. (2003) Structural basis forsimultaneous binding of two carboxy-terminal peptides of plant glutamatedecarboxylase to calmodulin. J. Mol. Biol. 328, 193–204.

[88] Hirst-Jensen, B.J., Sahoo, P., Kieken, F., Delmar, M. and Sorgen, P.L. (2007)Characterization of the pH-dependent interaction between the gap junctionprotein connexin43 carboxyl terminus and cytoplasmic loop domains. J. Biol.Chem. 282, 5801–5813.

[89] Purnick, P.E., Benjamin, D.C., Verselis, V.K., Bargiello, T.A. and Dowd, T.L. (2000)Structure of the amino terminus of a gap junction protein. Arch. Biochem.Biophys. 381, 181–190.

[90] Oh, S., Abrams, C.K., Verselis, V.K. and Bargiello, T.A. (2000) Stoichiometry oftransjunctional voltage-gating polarity reversal by a negative chargesubstitution in the amino terminus of a connexin32 chimera. J. Gen. Physiol.116, 13–31.

[91] Peracchia, C., Wang, X.G. and Peracchia, L.M. (2000) Behavior of Chemical andSlow Voltage-gates of Connexin Channels. The Cork Gating HypothesisGapjunctions: molecular basis of cell communication in health and disease, pp.271–295, Academic Press, San Diego.

[92] Peracchia, C., Wang, X.G. and Peracchia, L.L. (2000) Slow gating of gap junctionchannels and calmodulin. J. Membr. Biol. 178, 55–70.

[93] Unwin, P.N. and Ennis, P.D. (1983) Calcium-mediated changes in gap junctionstructure: evidence from the low angle X-ray pattern. J. Cell Biol. 97, 1459–1466.

[94] Gifford, J.L., Ishida, H. and Vogel, H.J. (2012) Structural insights intocalmodulin-regulated L-selectin ectodomain shedding. J. Biol. Chem. 287,26513–26527.