Structure of gap junction channels

seminars in CEll BIOLOGY Vol 3, 1992: pp 17-20

Structure of gap junction channels Kathrin A. Stauffer and Nigel Unwin Isolated channels

Gapjunctions are regions of contact between adjacent cells, consisting of arrays of channels linking thecell interiors. The channels are formed by polypeptides called connexins; the amino acidsequences ofmarry different connexins are known, and they are thought to resemble each other closely in tertiary and quarternary structure. Single channels have recently been isolated andpurified, .and earlierevidence has been confirmed showing that they consist ofsix identical subunits arranged around the central pore. GapJunction channels. are known to open and close in response to changes in ligand concentrations and electrical potential; in this respect they are very similar to ligand-gated ion channels which act as receptors in themembranes ofexcitable cells. The similarity is shown toextend tostructuralftatures such as theamino acidresidues lining thepore, and perhaps the location ofthe actual gate.

Intact, individual channels have been isolated from tissues or cultured cells by first purifying whole gap junction plaques and then treating them with detergent, using the appropriate ionic environment. 1-3 Complete solubilization, leading to a homogeneous population of channels, has been obtained with long chain detergents such as lauryl dimethyl amine oxide or dodecyl maltoside.P Other detergents, such as cetyl glucoside.s or mixtures of digitonin and octyl glucoside, I have been used, but it is not clear if the predominant species solubilized under those conditions is the intact, single membrane channel. In addition, either 10w2 or high'' pH, high ionic strength and the use of reducing conditions have been found to promote disruption of the bonds linking the channels to one another. . Electron micrographs of isolated channels in negative stain or in ice show them to be symmetrical, doughnut-shaped particles of about 80 A diameter when viewed along the axis of the pore. These images also confirm directly earlier evidencet-' that the channels are built from six subunits, as they often show strong six-fold modulations, particularly when imaged in ice (Figure 1). The particles, when viewed perpendicular to the axis of the pore, are about 75 A long and asymmetric in shape. They show an accumulation of stain at the extracellular end, where the pore is about 25 A widc.f At the other end, and in the middle of the particle, no such accumulation of stain is visible and the pore in these regions must therefore be much narrower.

Key words: connexon / pore structure / electron image / power spectrum

THE CHANNELS that comprise gap junctions are connected pairwise across the two membranes of adjoining cells so that they can facilitate the flow of ions and small molecules between their interiors. These connections are specialized structures since they are designed specially to prevent leakage of ions into the extracellular space. In other respects, gap junction channels are like other membrane channels: they are constructed from a ring of membranespanning subunits which delineate a central pore, or pathway for the ions; and they open and close in response to changes in ligand concentration. Recently, gap junction channels have been isolated from whole tissue and from cultured cells, and purified to homogeneity. This makes them amenable now to detailed biophysical analysis. We review here what is known about their basic molecular architecture, and compare them to other ligandgated channels, some of which have been better characterized.

The polypeptides The polypeptides forming the subunits of the gap junction channel are called connexins, and the oligomer of six subunits is known as a connexon. In the past few years, a large number of connexin isoforms have been sequenced (for a recent review see ref 6). All of these amino acid sequences have been deduced from nucleotide sequences, and in most cases the proteins themselves have not been studied. What information is available on the

.From theMedical Research Council, Laboratory ofMolecular BIology, Hills Road, Cambridge CB2 2QH, UK ©1992 Academic Press Ltd 1013-16821921010017 + 04$5.0010 17

18

K.A. Stauffer and N. Unwin

o. 5

I

0.10

O.GS

I

6

,8

12

n a

I

2

b

Figure 1. (a) Electron micrograph of purified gap junction channels in ice, showing them as doughnut-shaped particles of about 80 A diameter, with the aqueous pathway down the centre. Arrows mark th e density modulations made by the six surrounding subunits. The scale bar represents 20 nm. (b) Rotational power spectrum of a channel illustrating dominance of the six-fold harmonic.

structure of gap junction channels has been obtained from studies of three members of the connexin family: JJl-connexin, or connexm 32, from mammalian livers, arconnexin, or connexin 43, from heart, and a3-connexin, or MP-70, from ovine lens. Of these three, 13[-connexin is by far the most extensively studied protein. There are a number of features in the connexin sequences which imply posttranslational modification (for a review see ref 7). 131-connexin contains a CAAX box at the C-terminus, suggesting the protein might be isoprenylated, as well as a cysteine residue (probably Cys 217) which is acylated. This residue is also found in some other connexins. All connexins contain one or several consensus sites for phosphorylation, which suggests possible regulatory mechanisms. In none of the sequences published to date are there any cons ensus sites for N-glycosylation in the presumed extracellular domains, and biochemical studies have confirmed this.f

Transmembrane organization All connexins are predicted from hydropathy plots to contain four membrane-spanning segments termed M1 to M4 (refs 9,10; see Figure 2), with both the Nand C termini on the cytoplasmic side. The most conserved regions of the sequences are the membrane-spanning portions and the two loops on the extracellular side. The cytoplasmic loops betwe en

M2 and M3 and the C-terminal tails are less homologous and the C-terminal tails, in particular, are highly variable in length. Many experiments have been performed to probe the accessibility of connexin regions to antibodies and proteases, and to test the validity of the transmembrane topology implied by the hydropathy plots. It turns out to be easiest to show the exposure of sites to the cytoplasmic surface of the membrane, since isolated gap junction plaques are double membrane structures, and in these the 'gap' region is not accessible to most water-soluble reagents. Thus it has been established that the N-terminus, the loop between M2 and M3, and the C-terminal domain beyond M4 all face the cytoplasm. The evidence for this is binding of antibodies to peptides corresponding to amino acids 6-17,10 98-124,11 111-125,10 and 217-234,10 in I3rconnexin. Moreover, there is a proteolytic cleavage site at Lys 124,12 and probably multiple sites C-terminally of M4. 12 In order to demonstrate the extracellular location of the M1-M2 and M3-M4100ps it has been necessary to split gap junctions. This can be partly accomplished by treatment with urea at alkaline pH,13 or hydrochloric acid,12 or by perfusion of intact tissues with hypertonic sucrose.I! although none of these procedures are very satisfactory. Still, it has been possible in this way to show that antibodies to amino acids 38-53 10 and 164-189 11 will bind to split but not intact gap junctions, indicating the location of these segments in the extracellular or 'gap' domain. The

19

Gapjunction channels intracellular

c

MI

M2

M3

M4

extracellular

T

W fW T

Vi V Sl V

V,

fFL L

'

EFVA

I

'F

M3 Figure 2. Membrane topology of ~I-connexin as established from antibody-labelling experiments. Both termini lie on the cytoplasmicside of the membrane. Four membran e-spanning segments have been proposed. This leads to an additional cytoplasmic loop and two extracel1ular loops which form the pairwise linkage of channels. The putative a-helix, M3 (below), is composed predominantly of hydrophobic residues, but contains a line of polar residues (dotted line sloping up to the right) which may be forming the wall around the pore. demonstration that there are intramolecular disulphide bonds between cysteine residues located in the same regions'v supports this assignment, since cysteins located in the cytoplasm of a cell would normally be expected to exist as free sulphydryls. Recently an attempt has been made to label hydrophobic portions of the connexon, and it was shown that the N-terminal 30 kDa-portion of a3 connexin (or MP70), but not the domain further towards the C-terminus, is accessible to hydrophobic reagents.I'' However, there is clearly much more work required along these lines in order to narrow down the topological information to smaller polypeptide segments or single amino acid residues. All of these data fit into the model shown in Figure 2 which proposes the location of both termini On the cytoplasmic side of the membrane, and four

membrane-spanning segments. This results in two extracellular loops that could constitute the region of the protein which is responsible for the pairwise interaction of connexons, and one additional cytoplasmic loop. The observation that the proposed extracellular regions of the protein are strongly conserved between different connexins while there is extreme variability in those regions predicted to lie On the cytoplasmic side of the membrane, appears to support this model; One can picture proteins which have different regulatory domains on the cytoplasmic side of th e membrane but a similar structure On the extracellular side, allowing them to form hybrid channels while conserving their individual regulatory properties. Based on ,the cross-sectional dimensions of the channel in three-dimensional maps obtained from ice-embedded gap junctions, 17 and On the particular pattern of amino acid residues , it was suggested that the membrane-spanning segments (Figure 2) might consist of a-helices packed together as in a four-ahelical bundle.I'' Features in the X-ray diffraction pattern from oriented pellets of gap junctions have been interpreted to indicate that the transmembrane structure may consist of {3-sheet4 or of a-helical bundles.If Circular dichroism measurements demonstrating that {31-connexin contains 50-60% ahelix'? favour the latter possibility. However, a definitive picture of the transmembrane organization is not yet available and will have to await a higher resolution structure determination.

Analogy with other ion channels The nicotinic acetylcholine receptor, the glutamate receptor and synaptophysin belong to other families of ligand-gated ion channels which share some characteristics with the gap junction channel (for reviews see refs 20, 21). For example, hydropathy plots obtained from the amino acid sequences of these other channels suggest that they too may have four membrane-spanning segments. The acetylcholine receptor is the best studied channel of all, and in this case several lines of evidence support the view that at least part of the membrane-spanning segment, M2, is facing the pore. Small polar amino acid residues, which repeat in every fourth position along the M2 sequence, are implicated. These residues are where the pore is narrowest, at the level of the cytoplasmic leaflet of the lipid bilayer. 22 The same repeating pattern is found in equivalent membrane-

20

spanning segments of all ligand-gated channels, suggesting that a common principle is involved in constructing the wall around the narrow .part of a pore. Moreover, the segments M2 (or equivalent segments) are thought to be a-helices, so that the small polar amino acids line up along the axis of the channel when it is open, and perhaps are replaced by conserved flanking large hydrophobic residues when the channel is closed.I'' The transmembrane segment M3 of the gap junction channel has the same pattern of small polar amino acids and flanking hydrophobic residues that characterizes the other channels, and M3 may therefore correspond to an a-helix lining the gap junction pore. The analogy with the other channels suggests further that the narrowest part of the pore would be at the level of the cytoplasmic leaflet of the bilayer, a location consistent with the results oflow resolution structural studies.P The gate is likely to be in the narrowest part. An alternative location for the gate is outside the bilayer on the cytoplasmic side ,2~ although there is little conservation in the amino acid sequences between different connexins in this region.

References 1. Mazer F, MazetJL (1990) Re storat ion of gap junction-like structure after detergent solubili saton of the proteins from liver gap junctions. Exp Cell Res 188:312-315 2. Rhe e SK, Harris AL (1991) Affinity-purification of connexin 32 using a monoclonal antibody. Biochim Biophys Acta, in press 3. Stauffer KA, Kumar NM, Gilul a NB, Unwin N (1991) Isolation and purification of gap junction channels. J Cell BioI 115:141-150 4. Makowski L, Caspar DLD, Phillips WC, Goodenough DA (1977) Gap junction structures. 2. Diffraction. J Cell BioI 74:629-645 5. Unwin PNT, Zampighi G (1980) Structure of the junction between communicating cells. Nature 283:545 -549

K.A. Stauffer and N. Unwin 6. Kumar NM (1991) Gap junctions: a multigene family. Adv Struct BioI 1:209-278 7. Bennett MVL, Barrio LC, Bargiello TA, Spray DC, Hertzberg E, SaezJC (1991) Gap junctions: new tools, new answers, new questions . Neuron 6:305-320 8. Hertzberg EL, Gilula NB (1979) Isolation and characterisatio n of gap j unctions from rat liver. J Biol Chern 254:2138-2147 9. Unwin N (1986) Is there a common design for cell membrane channels? Nature 323:12-13 10. Milks LC, Kumar NM , Houghton R, Unwin N, Gilula NB (1988) Topology of the 32-kd liver gap junction protein determined by site-directed antibody localisations. EMBO J 7:2967-2975 11. Goodenough DA, Paul DL, Jesaitis L (1988) Topological distribution of two connexin 32 antigenic sites in intact and split rodent hepatocytegap junctions] Cell BioI107:1817-1824 12. Zimmer DB, Green CR, Evans WH, Gilula NB (1987) Topological analysis of the major protein in isolated intact rat liver gap junctions and gap junction-derived single ., membrane structures. J Cell BioI 262:7751-7763 13. Manjunath CK, Goings GE, Page E (1985) Proteolysis of 'cardiac gap junctions during their isolation from rat hearts, J Membr BioI 85:159-160 14. Goodenough DA, Gilula NB (1974) The splitting of hepatocyte gap junctions and zonulae occludentes with hypertonic disaccharides. J Cell BioI 61:575-590 15. Rahman S, Evans WH (1991) Topography of connexin 32 in rat liver gap junctions. J Cell Sci 100:567-578 16. Kistler J, Schaller J, Sigrist H (1990) MP38 contains the membrane-embedded domain of the lens fiber gap junction protein MP70. J BioI Chern 265:13357-13361 17. Unwin PNT, Ennis PD (1984) Two configurations of a channel-forming membrane protein. Nature 307:609·613 18. Tibbitts TT, Caspar DLD, Phillips WC, Goodenough DA (1990) Difraction diagnosis of protein folding in gap junction connexons. Biophys J 57: 1025-1036 19. Cascio M, Gogol E, Wallace BA (1990) The secondary structure of gap junctions. J BioI Chern 265:2358-2364 20. Unwin N (1989) The structure of ion channels in membranes of excitable cells. Neuron 3:665-676 21. Betz H (1990) Homology and an alogy in transmembrane channel design: lessons from synaptic membrane proteins. Biochemistry 29:3591-3599 22. Toyoshima C, Unwin N (1990) Three-dimensional structure of the acetylcholine receptor by cryoelectron microscopy and helical image reconstruction . J Cell BioI 111:2623-2635 23. Makowski L, Caspar DLD, Phillips WC, Goodenough DA (1984) Gap junction stru ctures. 5. Structural chemistry inferred from X-ray diffraction measurements on sucrose accessibility and trypsin susceptibility. J Mol BioI 174:449-481