Neuron
Article
Molecular and Functional Asymmetryat a Vertebrate Electrical SynapseJohn E. Rash,1 Sebastian Curti,2,3 Kimberly G. Vanderpool,1 Naomi Kamasawa,4 Srikant Nannapaneni,2
Nicolas Palacios-Prado,2 Carmen E. Flores,2 Thomas Yasumura,1 John O’Brien,5 Bruce D. Lynn,6 Feliksas F. Bukauskas,2
James I. Nagy,6 and Alberto E. Pereda2,*1Department of Biomedical Sciences, Colorado State University, Fort Collins, CO 80523, USA2Dominick P. Purpura Department of Neuroscience, Albert Einstein College of Medicine, Bronx, NY 10461, USA3Laboratorio de Neurofisiologıa Celular, Departamento de Fisiologıa, Facultad de Medicina, Universidad de la Republica,
Montevideo 11800, Uruguay4Max Planck Florida Institute, Jupiter, FL 33458, USA5University of Texas Health Science Center, Houston, TX 77030, USA6Department of Physiology, University of Manitoba, Winnipeg, MB R3E 0J9, Canada
*Correspondence: [email protected]
http://dx.doi.org/10.1016/j.neuron.2013.06.037
SUMMARY
Electrical synapses are abundant in the vertebratebrain, but their functional and molecular complex-ities are still poorly understood. We report herethat electrical synapses between auditory afferentsand goldfish Mauthner cells are constructed byapposition of hemichannels formed by two homo-logs of mammalian connexin 36 (Cx36) and that,while Cx35 is restricted to presynaptic hemiplaques,Cx34.7 is restricted to postsynaptic hemiplaques,forming heterotypic junctions. This molecular asym-metry is associated with rectification of electricaltransmission that may act to promote cooperativitybetween auditory afferents. Our data suggest that,in similarity to pre- and postsynaptic sites at chem-ical synapses, one side in electrical synapsesshould not necessarily be considered the mirrorimage of the other. While asymmetry based on thepresence of two Cx36 homologs is restricted toteleost fish, it might also be based on differencesin posttranslational modifications of individual con-nexins or in the complement of gap junction-associ-ated proteins.
INTRODUCTION
While the physiological importance of electrical synaptic trans-
mission in cold-blooded vertebrates has long been established
(Bennett, 1977), progress over the last decade has also revealed
the widespread distribution of electrical synapses, and this
modality of synaptic transmission was reported to underlie
important functional processes in diverse regions of the
mammalian CNS (Connors and Long, 2004). Consequently, elec-
trical transmission is now considered an essential form of inter-
neuronal communication that, together with chemical transmis-
sion, dynamically distributes the processing of information
within neural networks. In contrast to detailed knowledge of
the mechanisms underlying chemical transmission, far less is
known about how the molecular architecture or the potentially
diverse biophysical properties of electrical synapses encoun-
tered in physiologically disparate neural systems govern their
function or impact on characteristics of electrical transmission
in those systems.
Electrical synaptic transmission is mediated by clusters of
intercellular channels that are assembled as gap junctions
(GJs). Each intercellular channel is formed by the docking of
two hexameric connexin hemichannels (or connexons), which
are individually contributed by each of the adjoining cells,
forming molecular pathways for the direct transfer of signaling
molecules and for the spread of electrical currents between
cells. As a result, electrical synapses are often perceived as
symmetrical structures, at which pre- and postsynaptic sites
are viewed as the mirror image of each other. Connexons
are formed by proteins called connexins that are the products
of a multigene family that is unique to chordates (Cruciani
and Mikalsen, 2007). Because of its widespread expression
in neurons, connexin 36 (Cx36) is considered the main
‘‘synaptic’’ connexin in mammals. In contrast to other connex-
ins, such as some found in glia (Yum et al., 2007; Orthmann-
Murphy et al., 2007), all pairing configurations tested so far
indicate that Cx36 forms only ‘‘homotypic’’ intercellular chan-
nels (Teubner et al., 2000; Li et al., 2004), where connexons
composed of Cx36 pair only with apposing Cx36-containing
connexons. Notably, the number of neuronal connexins is
higher in teleost fishes, which, as a result of a genome duplica-
tion (Volff, 2005), have more than one homolog gene for most
mammalian connexins (Eastman et al., 2006). This raised the
possibility of more complex configurations of neuronal con-
nexin coupling in teleost fish evolving in response to functional
demands.
Because of their experimental access, auditory ‘‘mixed’’
(electrical and chemical) synapses on the teleost Mauthner cell
(M-cell) (a reticulospinal neuron involved in tail-flip escape
responses; Faber and Pereda, 2011), known as ‘‘large myelin-
ated club endings’’ (CEs), constitute a valuable model for
studying vertebrate electrical transmission (Pereda et al.,
Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc. 957
Figure 1. Presence of Two Homologs of Cx36 at CEs
(A) Diagram of the M-cell. Auditory afferents terminate as CEs in the distal
portion of the lateral dendrite, formingmixed (electrical plus chemical) synaptic
contacts (dashed box).
(B–D) Laser scanning confocal projection of the distal portion of the lateral
dendrite. Double immunolabeling with a monoclonal Cx35 antibody (C, red)
and a polyclonal Cx34.7 (IL) antibody (D, green) shows a high degree of
colocalization at individual CEs, shown by red/green overlay (B, asterisks).
(E–G) Highmagnification of an individual CE showing intense punctate labeling
for Cx35 (F, red) and Cx34.7 (G, green) and high colocalization (E).
(H) Confocal line-scan imaging (blue line in inset) illustrates the high degree of
colocalization of Cx35 and Cx34.7.
(I) Plot represents the ratio of superposition of Cx35 to Cx34.7 labeling at
individual CEs (converted to percentage) against the ratio of Cx34.7 to Cx35
(n = 30).
See also Figures S1 and S2.
Neuron
Molecular Asymmetry and Electrical Rectification
958 Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc.
2004). We previously reported that connexin 35 (Cx35) (O’Brien
et al., 1998), a fish homolog of mammalian neuronal Cx36
(Condorelli et al., 1998), is present at CEs (Pereda et al., 2003).
Given the genome duplication that occurred in teleost fish, we
investigated the presence of an additional Cx36 homolog at
these terminals. Using immunofluorescence and ultrastructural
approaches, we show that a second homolog of Cx36, connexin
34.7 (Cx34.7) (O’Brien et al., 1998), is also present at CEs.
Strikingly, matched double-replica freeze-fracture immunogold
labeling revealed that Cx35 is restricted to presynaptic CE
hemiplaques, whereas Cx34.7 is restricted to postsynaptic
M-cell hemiplaques. Asymmetry in the molecular composition
of adjoining connexons was proposed to allow electrical
rectification at some GJs (Barrio et al., 1991; Phelan et al.,
2008). Consistent with this notion, our estimates of GJ resistance
in each direction revealed a near 4-fold difference in con-
ductance, favoring the spread of postsynaptic membrane
responses to presynaptic endings, which, by acting as a
mechanism of lateral excitation (Pereda et al., 1995; Curti and
Pereda, 2004), would facilitate the fish’s escape. Thus, molecu-
lar asymmetries in neuronal gap junctions can underlie complex
functional properties and suggest that the apposed sides of
electrical synapses are not necessarily the mirror images of
each other.
RESULTS
Cx35 and Cx34.7 Colocalize in Club EndingsThe large size and distinctive morphology of the M-cell (Fig-
ure 1A) allows the imaging of long stretches of membrane in a
single optical section. Because of their unusually large size,
CEs can be unequivocally identified on the distal portion of the
M-cell lateral dendrite using Cx35 labeling (Figures 1B and 1C;
Pereda et al., 2003; Flores et al., 2008). To determine whether
other teleost homologs of Cx36 are present at CEs, we investi-
gated whether Cx35 colocalizes with Cx34.7 at CEs by perform-
ing double immunofluorescence labeling using an anti-Cx35
antibody (Chemicon MAB3043) and an anti-Cx34.7 intracellular
loop (IL) antibody (see below; Experimental Procedures; Table
S1 available online). Both Cx35 and Cx34.7 antibodies showed
intense punctate staining and colocalization at contacts
between CEs and M-cells (Figures 1B–1D). We previously
showed that the number of anti-Cx35 fluorescent puncta at indi-
vidual CEs (Figures 1E–1G) was consistent with ultrastructural
demonstration of 63–243 closely spaced GJ plaques at these
terminals (Tuttle et al., 1986), suggesting that each punctum
represents an individual plaque (Flores et al., 2008). Accordingly,
confocal line-scan imaging illustrated the colocalization of label-
ing for both antibodies at single puncta, suggesting that both
connexins coexist at individual plaques (Figure 1H). We quanti-
fied the colocalization of Cx35 with Cx34.7 (and vice versa) using
confocal reconstruction of individual terminals, as identified by
shape and Cx35 labeling (Figure 1F). Averaged over individual
endings, 85.04% (±9.12 SD) of the area of Cx35 immunolabeling
also showed Cx34.7 labeling and 81.23% (±8.34 SD) of the area
of Cx34.7 labeling showed Cx35 labeling (n = 30) (Figure 1I).
Thus, although not completely overlapping, the two proteins
exhibit a high degree of colocalization in CEs.
Neuron
Molecular Asymmetry and Electrical Rectification
Ultrastructural Analysis Reveals that Cx35 and Cx34.7Are Differentially Segregated to Pre- versusPostsynaptic SidesTo confirm that Cx35 and Cx34.7 colocalize at individual GJ
plaques, we performed conventional freeze-fracture replica
immunogold labeling (FRIL), which allows broad expanses of tis-
sues to be examined and facilitates unambiguous assignment of
specific connexin labeling to GJ hemiplaques in either of two
apposed cells (see Supplemental Experimental Procedures).
Four replicas of goldfish hindbrain contained CE synapses on
identified M-cells. The CE terminals were identified on confocal
grid-mapped M-cells that had been injected with Lucifer yellow
during in vivo recordings prior to tissue fixation as well as in one
set of matched double replicas prepared by SDS-FRIL (see
Supplemental Experimental Procedures). Samples were either
single-labeled with anti-Cx36 Ab298, which binds to both
Cx34.7 and Cx35 (see below), or double-labeled for Cx35 and
Cx34.7 IL. In a double-labeled replica of a positively identified
M-cell, labeling for Cx35 was found directly associated with GJ
plaques in presynaptic membranes of CEs (n = 20 GJs). In
contrast, labeling for Cx34.7 was only on identified M-cell post-
synaptic membranes (n = 53 GJs). Consistent with this distribu-
tion, anti-Cx36 Ab298, which recognizes both Cx35 and Cx34.7
(see next section and Table S1), was found to label both pre-
and postsynaptic membranes (data not shown, but see data in
Pereda et al., 2003). Such differential distribution to pre- versus
postsynaptic membranes was investigated further by double-
immunolabeling for Cx35 and Cx34.7 using matched double-
replica FRIL (DR-FRIL). Initially, a sample prepared for DR-FRIL
was fractured andmajor portions of bothmatching complements
were retrieved and labeled. In one of the two M-cell comple-
ments, more than 400 labeled GJs were found; 367 were viewed
toward theM-cell side of the junction (Figures 2A–2D), all ofwhich
were labeled for Cx34.7 and none for Cx35; and 79 were viewed
from the M-cell side of the synapse toward the CE (Figure 2E), all
of whichwere labeled for Cx35 and none for Cx34.7. A diagramof
that same cell is indicated in Figures 2F and 2G, illustrating the
two primary views seen in Figures 2D and 2E.
To further investigate this apparent GJ connexin asymmetry,
analysis was performed on matching complements of individual
M-cell/CE GJs in these same samples. However, because of
damage to one of the matching replicas, only about 30 M-cell/
CE GJs could be matched in the two complementary replicas
(Figure 3). Of those 30 matching complements, 100% had label-
ing for Cx35 (10 nm gold beads) within the CE plasma mem-
brane, without labeling for Cx34.7 IL, and 100% had labeling
for Cx34.7 IL (5 nm gold beads) within the postsynaptic M-cell
plasma membrane, with no labeling for Cx35. Thus, whether
examined in single replicas or in matched complementary
double replicas of the same GJ hemiplaques, Cx35 was
restricted to the CE side of GJs (presynaptic hemiplaques) and
Cx34.7 was present only in the M-cell side of GJs (postsynaptic
hemiplaques), unambiguously demonstrating that GJ channels
between CEs and the M-cell dendrite are heterotypic.
Specificity of Anti-Connexin AntibodiesBecause of substantial amino acid sequence identity of Cx35
and Cx34.7, the specificity of the antibodies used here is critical
for the accurate identification of these two connexin homologs.
Our previous studies on connexins at CEs focused largely on
Cx35 at these synapses, using either anti-Cx35 antibodies or
anti-Cx36 antibodies that were shown to recognize Cx35. In
the present study, HeLa cells transfected with Cx34.7 or Cx35
were used to confirm the quality and specificity of a set of anti-
Cx34.7 antibodies and to establish which of the previously
utilized as well as currently available anti-Cx36 antibodies either
do or do not cross-react with Cx34.7 or Cx35 (Table S1).
HeLa cells were found to readily express Cx34.7 upon trans-
fection, and robust immunofluorescence detection of this con-
nexin both intracellularly and at plasma membrane contacts
was obtained with anti-Cx34.7 IL (Figure S1A1). The same cul-
ture labeled with anti-Cx36 Ab39-4200 showed codetection
and subcellular colocalization of labeling (Figures S1A2 and
S1A3), indicating Ab39-4200 recognition of Cx34.7 and therefore
serving as a positive control for Cx34.7 expression. The anti-
Cx36 Ab298 previously shown in our earlier study to recognize
Cx35 (Pereda et al., 2003) also recognized Cx34.7 (Figure S1B1)
and produced labeling that corresponded with labeling pro-
duced by Ab39-4200 (Figures S1B2 and S1B3). We next tested
immunofluorescence detectability of Cx35 with anti-Cx34.7 IL
in HeLa cells transfected with Cx35-enhanced yellow fluores-
cent protein (eYFP). Clusters of HeLa cells with high transfection
efficiency displayed intense intracellular eYFP fluorescence as
well as detection of Cx35-eYFP at cell-cell contacts (Figures
S1C1 and S1E1). In these cultures, Cx35 was not recognized
by anti-Cx34.7 IL (Figures S1C2 and S1C3), indicating specificity
of this antibody for Cx34.7. In contrast, while Cx34.7-transfected
cells showed robust labeling of Cx34.7 with anti-Cx36 Ab39-
4200 (Figure S1D1), polyclonal anti-Cx36 Ab51-6300 did not
cross-react with Cx34.7 in this same culture (Figures S1D2 and
S1D3) but showed robust detection of Cx35 (Figures S1E2 and
S1E3). See Figure S2 for additional antibodies tested. In addi-
tion, we previously established that anti-Cx35 (Chemicon
MAB3043) antibody does not crossreact with Cx34.7 (Pereda
et al., 2003).
Electrical Transmission between CEs and the M-cell IsAsymmetricHeterotypic GJ channels have been associated with asymmetry
of electrical transmission (Barrio et al., 1991; Phelan et al., 2008).
While simultaneously recording a single CE afferent at the VIIIth
nerve root and the M-cell lateral dendrite (Figure 4A), we found
a dramatic difference between orthodromic and antidromic
coupling coefficients (CCs), calculated using the M-cell and CE
action potentials and their respective coupling potentials (CC =
coupling/action potential). The CCs averaged 0.009 ± 0.001
(SEM) in the orthodromic direction and 0.083 ± 0.009 (SEM) in
the antidromic direction (p < 0.0005; n = 36). The �9-fold
disparity indicates that electrical transmission is stronger in the
antidromic direction. This difference is observed in the simulta-
neous recording illustrated in Figure 4A and is more clearly
observed in the experiment of Figure S3A, where multiple CEs
terminating in the same lateral dendrite were recorded sequen-
tially while maintaining the dendritic recording electrode. There
was a dramatic difference for CCs in the antidromic direction
at eachCE (Figure S3B), indicating that the functional asymmetry
Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc. 959
Figure 2. Pre- versus Postsynaptic Con-
nexins
Low- to high-magnification images of a single
M-cell after double labeling with 5 nm gold beads
for Cx34.7 (using antibody 2930-2 IL) and with
10 nm gold beads for Cx35 using MAB3045.
(A–C) M-cell lateral dendrite (A, yellow overlay)
surrounded by CEs (pale green overlays) and by
additional unidentified axon terminals (not
colored). Boxed area is shown at higher magnifi-
cation in (B), and the boxed area in (B) is shown at
higher magnification in (C).
(D) High magnification of boxed area in (C),
showing view from CE E-face (green overlay)
toward the M-cell P-face (yellow overlay), with
connexins in the M-cell plasma membrane labeled
only for Cx34.7 (5 nm gold beads), whether as
M-cell P-face connexons (blue overlay) or CE
E-face pits (blue-green overlay). Two 10 nm gold
beads that are not on a gap junction represent
nonspecific background or ‘‘noise’’ (indicated by
the ‘‘barred circle’’). Purple overlays are clusters of
P-face pits representing imprints of glutamate
receptors. (See glutamate receptor E-face IMPs
in E.)
(E) View from the M-cell E-face (yellow overlay)
toward the CE P-face (green overlay) at the
opposite end of the same obliquely fractured
M-cell as shown in (A)–(D). The postsynaptic
membrane contains clusters of 10 nm E-face
intramembrane particles (IMPs; purple overlays)
corresponding to glutamate receptor particles
(Tuttle et al., 1986; Pereda et al., 2003). After SDS
washing, connexons remaining in the CE plasma
membrane, beneath replicated CE P-face IMPs
(green overlay) and M-cell E-face pits (yellow-
green overlay), are labeled solely for Cx35 (10 nm
gold beads).
(F and G) Stylized diagrams (i.e., molecular versus
cellular components not to same scale) illustrating
the location of the fracture plane coursing diago-
nally through a M-cell (F, yellow). The resulting
labeled replica (diagrammed in G) illustrates that
connexons remain in the cell whose residual
cytoplasm is retained beneath the replica. The
apparent labeling of gap junction E-face pits
(which are devoid of connexin proteins and
therefore cannot be labeled) occurs because
intact but unreplicated connexons remain strongly
adsorbed to the platinum/carbon replica at the bottom of the E-face pits, where they remain for subsequent labeling (Fujimoto, 1995). Clusters of 10 nmglutamate
receptor IMPs are retained in the extraplasmic leaflet, providing a marker for recognizing the M-cell plasma membrane E-face, as previously demonstrated
(Pereda et al., 2003). Lavender circles are synaptic vesicles. Small black dots are gold beads for Cx34.7; larger black dots are gold beads for Cx35.
See also Figures S1 and S2.
Neuron
Molecular Asymmetry and Electrical Rectification
represents a general property of CEs likely operating under
physiological conditions, as it was observed using physiological
signals, such as action potentials.
Electrical Synapses between CEs and theM-cell RectifyThe strength of electrical transmission (amplitude of the coupling
potential) does not solely depend on the conductance of the GJ
channels but also on the passive properties determined by the
resistance (and capacitance under some conditions) of the
coupled neurons. The relatively smaller size of CEs indicates
960 Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc.
that their input resistance is likely higher than that of the M-cell
dendrite, thus contributing to the asymmetry between ortho-
dromic and antidromic CCs. To evaluate the contribution of
heterotypic GJ channels to asymmetric electrical transmission,
we investigated possible asymmetries in GJ resistance between
CEs and the M-cell. Rectification refers to the propensity of
some electrical synapses to display differential resistance to
current flow in one versus the other direction across the junction
between two coupled cells (Furshpan and Potter, 1959). While
properties of junctional conductance (inverse of resistance) are
Figure 3. Connexin Localization by DR-
SDS-FRIL
Matched double replicas of twoGJs from the same
replica as in Figure 2, with diagrams illustrating
labeling of the matched hemiplaques.
(A and B) Matched GJ hemiplaques viewed toward
the M-cell plasma membrane P-face (A, yellow
overlay), with 5 nm gold beads (small arrows) for
Cx34.7 present beneath both P-face IMPs (blue
overlay) of the M-cell and beneath E-face pits
(blue-green overlay) of the CE plasma membrane
(green overlay). (B) View toward the CE (light green
overlay), with 10 nm gold beads (large arrows)
labeling for Cx35 beneath E-face pits (yellow-
green overlay) of the extraplasmic leaflet of the
M-cell (yellow overlay) and beneath P-face con-
nexon IMPs (light green overlay) of the CE plasma
membrane (light green overlay). These matched
complementary replicas document that CE/M-cell
GJs are heterotypic/asymmetric.
(C) Diagram of the fracture plane through a gap
junction between a CE (left panel, top) and the
M-cell (left panel, bottom). The fracturing process
(middle panel) separates all connexons (blue in
M-cell; green in CE) at their points of contact in the
extracellular space, with all connexons of the CE
remaining with the upper tissue fragment and all
connexons of the M-cell remaining with the lower
fragment. Circular arrow and curved arrow indicate that the upper tissue fragment is inverted for labeling (right panel). Center panel reveals 5 nm gold beads
labeling only Cx34.7 in the M-cell cytoplasm, regardless of whether E-face pits of the CE or P-face IMPs of the M-cell are replicated. Right panel reveals 10 nm
gold beads labeling only Cx35 in the CE cytoplasm, whether beneath M-cell E-face pits or beneath CE P-face IMPs.
See also Figures S1 and S2.
Neuron
Molecular Asymmetry and Electrical Rectification
generally examined with simultaneous recordings from two cells
under voltage clamp configuration (Barrio et al., 1991), this
approach in our case would require simultaneous in vivo intrater-
minal and intradendritic recording, which is feasible (Pereda
et al., 2003) but not sufficiently stable for analysis of rectification.
Moreover, the resistance of the presynaptic electrode and
geometrical characteristic of the afferents make it impractical
to use the voltage clamp configuration to directly determine
junctional resistance. Therefore, we followed an established indi-
rect approach (Bennett, 1966; Devor and Yarom, 2002), where
junctional resistance can be estimated from measurements of
CCs in each direction (from pre- to post and vice versa), in com-
bination with measurements of the input resistances of the
coupled cells (see Experimental Procedures). While this
approach might also be challenging in some cell types (GJs
are dendrodendritic in most mammalian neurons), the M-cell
and the CEs offer several unusual anatomical and physiological
characteristics that make it possible to estimate these parame-
ters in vivo: (1) CE afferents terminate with a single contact and
are tightly segregated to the distal portion of the lateral dendrite
of the M-cell; (2) the M-cell lateral dendrite as well as both the
axons and terminals of CEs are accessible for intracellular re-
cordings; and (3) the M-cell and the CEs have comparable and
unusually fast membrane time constants, estimated to be
400 ms in the M-cell (Fukami et al., 1965) and 200 ms in CEs (Curti
et al., 2008), which allow the use of physiological signals, such as
action potentials, for measurements of CCs. Due to spatial con-
siderations, measurements of CCs during simultaneous record-
ings of CE afferents in the VIIIth nerve root and theM-cell dendrite
are useful to expose asymmetry of electrical transmission (Fig-
ures 4A and S3B) but not accurate enough for estimating GJ
conductance (see below). To overcome this problem, we calcu-
lated average values of CCs for the population of afferents, using
values obtained under various experimental arrangements that
maximize their accuracy (see below).
The ‘‘population CC’’ in the orthodromic direction (CE to
M-cell) for a number of CEs was estimated as the ratio between
the average amplitude of the electrical component (or coupling
potential) of the unitary postsynaptic potential and the average
amplitude of the presynaptic spike (CC, postsynaptic coupling
potential/presynaptic spike; Figure 4A). The orthodromic
coupling potential (recorded during paired recordings with intra-
dendritic recordings in the terminal field of CEs) averaged 0.73 ±
0.04 mV SEM (n = 76). (Because the strength of electrical synap-
ses between individual CEs varies dramatically [Smith and
Pereda, 2003], it was not possible to assign differences in the
amplitude of individual coupling potentials to their relative posi-
tion within the dendritic field and therefore correct for potential
electrotonic attenuation. Thus, although potentially slightly
underestimated, we believe the average amplitude of ortho-
dromic coupling potentials represents the most appropriate
value to use for calculating the CC in the orthodromic direction.)
During simultaneous recordings, the amplitude of the presynap-
tic spike evoked at the recording site with long (200 ms)
depolarizing pulses does not represent the spike that ultimately
generates coupling, as the spike recorded at the site of depolar-
ization regenerates in subsequent nodes and, finally, at the
presynaptic terminal (see Figure S4). More importantly, its
Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc. 961
Figure 4. Electrophysiological Recordings
for Estimates of Junctional Conductance
(A) Experimental arrangement. AD and pre- and
postsynaptic electrodes are indicated.
(B) Simultaneous pre- and postsynaptic re-
cordings between CEs and the M-cell lateral
dendrite allow measuring orthodromic and anti-
dromic coupling potentials in the same terminal.
Here and elsewhere, unless indicated, traces
represent the average of at least ten single re-
sponses. Left panel: Estimating dendritic Rn.
Excitatory postsynaptic currents (EPSC) evoked
by extracellular stimulation of the posterior VIIIth
nerve (population response) obtained while
clamping at resting potential (�76 mV) using
single electrode voltage clamp (SEVC). Top
right: Currents evoked by command pulses
of +10, +5,�15, and�25mV from resting potential
(arrowhead indicates the time at which currents
were measured). Bottom right: voltage-current
relationship plotted for a larger range of command
pulses. Conductance (G) was calculated from the
slopes of the fitted line.
(C) Intraterminal recordings (club ending) obtained
after an initial recording of the antidromic spike (AD
spike) in the M-cell lateral dendrite (M-cell, left),
showing antidromic coupling (center). Recordings
are also illustrated scaled (right) (modified from
Curti and Pereda, 2004; Smith and Pereda, 2003).
(D and E) Spatially controlled sequential intra-
dendritic recordings of the AD spike obtained after
an initial extracellular field recording of the AD
spike in the M-cell axon cap (B).
(F) Graph plots the decay of the AD spike with
distance. Data were fit with a single exponential
(r2 = 0.99). Orange bar indicates the location of the
CEs terminal field in the lateral dendrite.
See also Figure S4.
Neuron
Molecular Asymmetry and Electrical Rectification
amplitude is affected by the pulse depolarization. Therefore, for
values of presynaptic spike amplitude, we used short depolariz-
ing pulses, at which spikes initiate from resting potential and are
likely representative of those normally occurring at the contact,
averaging 87.6 ± 0.9 mV SEM (n = 203). These measurements
yielded an orthodromic CC of 0.008. The input resistance of
the M-cell lateral dendrite was directly measured under single-
electrode voltage-clamp configuration during intradendritic re-
cordings (see Experimental Procedures) and found to be, on
average, 1.32 ± 0.3 MU SEM (n = 9; Figure 5B).
The population antidromic CC (M-cell to CE) was calculated as
the ratio between the amplitude of the antidromic (AD) coupling
potential (the coupling of the antidromic spike of the M-cell in the
CE) and the amplitude of the antidromic M-cell spike (AD spike)
962 Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc.
recorded in the dendrite (Figure 5C;
CC, AD coupling potential/AD spike).
Because the AD coupling potential is
greatly reduced by electrotonic attenua-
tion when recorded at the VIIIth nerve
root during simultaneous recordings, we
estimated its average value by performing
intraterminal recordings in the vicinity of
the M-cell lateral dendrite. This recording position allows mea-
surement of the true amplitude of the AD coupling without the
effect of attenuation by electrotonic axonal propagation (Fig-
ure 4C, bottom right). The coupling averaged 5.07 ± 0.31 mV
SEM (n = 24) but was subsequently corrected to 1.85 ±
0.11 mV SEM to take account for the amplification of the AD
coupling produced by a persistent sodium current (INa+P), which
is present in these afferents. (The correction was based on a
predicted amplification of 63.6% of the average AD coupling
amplitude from previous correlations of percent INa+P amplifica-
tion versus AD coupling amplitude at resting potential; see
Experimental Procedures; Curti and Pereda, 2004.)
We next considered the AD spike amplitude that is, on
average, most representative of that ‘‘seen’’ by the population
Figure 5. Voltage Dependence of Electrical Transmission
(A) The AD coupling is voltage dependent.
(B) Superimposed traces show the AD coupling potential recorded at resting potential (�73 mV) and near the threshold of the cell (–64 mV).
(C) Relationship between the antidromic coupling potential (AD coup, ordinates) and the presynaptic membrane potential (membrane potential, abscissa). The
dramatic increase in AD coupling with depolarization was blocked by QX-314 (Curti and Pereda, 2004), revealing a second voltage-dependent mechanism. See
also Figure S5.
(D) Only the QX-314-resistant mechanism is observed at the end of a 200 ms pulse.
(E) Superimposed traces show the AD coupling potential recorded at resting potential (–73 mV) near the threshold of the cell (–66 mV) and at �80 mV.
(F) Traces are illustrated scaled.
(G) Amplitude of AD coupling (obtained at the end of the 200ms pulse, red circles) and the afferent’s input resistance (blue boxes) versus membrane potential in a
representative experiment. Changes in amplitude of AD coup (K1) occur in the absence of changes in the afferent axon’s input resistance.
(H) Summary for ten experiments (red dots are AD coupling; blue squares are input resistance).
Neuron
Molecular Asymmetry and Electrical Rectification
of CEs. We reasoned that the amplitude of the AD spike at the
center of the terminal field of CEs in the lateral dendrite would
yield a good approximation. Because the amplitude of the M-
cell AD spike decays along the lateral dendrite (the M-cell spike
is generated at the axon initial segment and neither the soma nor
dendrite have active properties; Furshpan and Furukawa, 1962)
and because the precise location of the electrode in the dendrite
cannot be controlled, this AD spike amplitude varies between
experiments (10–20 mV). Therefore, to estimate the amplitude
of the AD spike at the center of the terminal field of CEs, where
most CEs terminate (Lin et al., 1983), we performed multiple
sequential recordings along the M-cell dendrite (Figure 4D).
Initial extracellular recordings were made in the M-cell axon
cap, which served as a ‘‘spatial calibration’’ marker, as the
distinctive field amplitude denotes the proximity of the electrode
to the initial segment (Furshpan and Furukawa, 1962). The elec-
trode was then moved at regularly spaced intervals along the
lateral dendrite for multiple recordings, during and after which
no changes were observed in the electrical properties of the
M-cell (Figure 4E). The amplitude of the AD spike decayed
Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc. 963
Table 1. Values Used for Estimates of Gap Junctional Resistance
Presynaptic
Spike
Postsynaptic
Coupling)
Population Coupling
Coefficient
Postsynaptic Cell
Input Resistance
Junctional
Resistance
Orthodromic (CE to M-cell) 87.6 ± 0.9 mV,
n = 203
0.73 ± 0.04 mV,
n = 76
0.008 1.32 ± 0.3 MU, n = 9 168.3 MU
Antidromic (M-cell to CE) 10.6 mV 1.85 ± 0.11 mV,
n = 24
0.175 8.05 ± 0.74 MU, n = 20 39.8 MU
Mean ± SEM.
Neuron
Molecular Asymmetry and Electrical Rectification
exponentially (r2 = 0.99) with a space constant of�300 mm and a
predicted amplitude of 10.6 mV at the center of the terminal field
of CEs (which start�200 mm from the initial segment; Figure 4F).
These measurements yielded an antidromic CC of 0.175. The
input resistance of CEs was directly measured with current
pulses during intracellular recordings, with a resulting average
of 8.05 ± 0.74 MU SEM (n = 20).
Using these measurements and the equation described in the
Experimental Procedures, we obtained values of junctional
resistance of 168.3 MU in the orthodromic direction and of
39.8 MU in the antidromic direction (Table 1). This more than
4-fold difference between orthodromic and antidromic junctional
resistance indicates that electrical synapses at CEs rectify in a
way that enhances transmission of signals from the M-cell
dendrite into presynaptic afferents. While calculations were
based on values that we consider are the most accurate mea-
sures of the signals involved, the asymmetry in junctional resis-
tance was observed for a wide range of values, including the
average AD spike amplitude obtained during paired recordings
(which averaged 15.9 ± 0.48 mV SEM; n = 18) and presynaptic
spikes’ amplitudes recorded at the terminal (Figure S4), therefore
providing a high degree of confidence in the conclusion that GJs
between CEs and the M-cell rectify. In other words, electrical
rectification is sufficiently large to be detected by our indirect
experimental method. Accordingly, despite less favorable
experimental conditions for calculating accurate antidromic
CCs (and therefore for revealing GJ asymmetries), calculations
of GJ resistance obtained for each of the CEs illustrated in Fig-
ure S3, using the values of presynaptic spikes and coupling
potentials recorded at each of the afferents, still reveal an asym-
metry of GJ resistance (Figure S3C). Thus, the asymmetry of
electrical transmission observed between CEs and the M-cell
is supported by two contributing factors, an asymmetry of input
resistances between the coupled cells and an asymmetry of GJ
resistance (rectification).
Rectifying electrical synapses exhibit voltage-dependent
behavior (Furshpan and Potter, 1959; Giaume et al., 1987). We
have previously shown that the AD coupling potential produced
by the retrograde spread of the AD spike from the postsynaptic
M-cell is dramatically enhanced by depolarization of the presyn-
aptic terminal (Figure 5A; Pereda et al., 1995; Curti and Pereda,
2004). This dramatic voltage-dependent enhancement of electri-
cal coupling upon depolarization does not represent a property
of the junctions themselves but rather the activation of an
INa+P present at presynaptic terminals that acts to amplify the
synaptic response (Curti and Pereda, 2004). The interplay of
this current with an A-type repolarizing K+ conductance (IA)
964 Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc.
generally reproduces the waveform of the coupling recorded at
resting potential (Figure 5B; Curti and Pereda, 2004), exhibits
an increased time to peak (Figure 5B), and the amplification is
blocked by both extracellular TTX and intracellular application
of QX-314 (changes occurred within a time window in which
the spikes of theCEs remained essentially unaffected; Figure 5C;
Curti and Pereda, 2004). Blockade of the INa+P reveals a
second, less prominent, voltage-dependent component that is
symmetrical relative to resting membrane potential. This second
voltage-dependent component can also be observed in the
absence of TTX and QX-314 at the end of a long (250 ms) depo-
larizing pulse (Figure 5D) when the above-mentioned conduc-
tances are no longer active, further indicating the existence of
two different voltage-dependent mechanisms (Curti and Pereda,
2004). Both components can also be isolated by curve fitting
(Figure S5). The QX-314-insensitive voltage-dependent behavior
had a slope of 0.094, equivalent to a change in AD coupling
amplitude of 3.81% per mV of membrane potential change,
which is symmetrical from resting potential, and unlike the
INa+P component, it does not modify the time to peak (Figure 5E)
nor the kinetics of the coupling potential (Figure 5F). We hypoth-
esized that the QX-314-insensitive voltage-dependent compo-
nent could correspond to either (1) a voltage-dependent
behavior of GJ channels or (2) a voltage-dependent behavior of
the cell’s membrane resistance, which could proportionally
modify the amplitude of the coupling potential. To distinguish
between these two possibilities, we measured both the ampli-
tude of the AD coupling potential and the CE’s input resistance
under different membrane potentials at the end of a 250 ms
pulse, where active conductances do not contribute to coupling
amplification. As illustrated in Figures 5G (single experiment) and
5H (n = 10), changes in amplitude of the AD coupling potential
were independent of the CE’s input resistance, which remained
constant through the full range of membrane potentials. As is the
case with other rectifying electrical synapses (Giaume and Korn,
1984), we found a difference between the resting potentials of
the coupled cells. The values averaged �71.7 ± 0.32 mV SEM
(n = 203) for CEs, where �74 mV was the most hyperpolarized
value, and �78.7 ± 2.5 mV SEM (n = 95; p < 0.01) for the
M-cell, where �85 mV was the most hyperpolarized value, sug-
gesting the existence of a transjunctional voltage of �10 mV, on
top of which electrical signals operate. Thus, we conclude that
electrical synapses at CEs exhibit voltage-dependence, where
depolarization of the presynaptic terminal enhances retrograde
electrical communication. By virtue of their electrical direction-
ality and voltage-dependence, heterotypic GJs in CEs act syner-
gistically with the presynaptic QX-314-sensitive component
Figure 6. Rectification Promotes Lateral ExcitationCartoon illustrates the lateral excitation of CE afferents on the M-cell lateral
dendrite and the contribution of electrical rectification to this phenomenon.
Dendritic synaptic potentials (yellow) evoked by suprathreshold electrical
stimulation of VIIIth nerve afferents (orange labeled CEs) spread to neighboring
subthreshold terminals (yellow labeled CEs). Electrical rectification favors the
retrograde transmission of dendritic signals toward CEs in a higher resistance
pathway (>Rn), counteracting the leak of currents toward the soma following a
pathway of low resistance (<Rn). Because of its voltage-dependent properties,
electrical transmission acts as a coincidence detector, facilitating the
recruitment of CEs that are already depolarized, such as during the invasion of
an incoming action potential, whose depolarization travels ahead several
nodes (bottom left CE; note that the arrow is bigger and the yellow in the CE
more intense, denoting the increase in coupling produced by presynaptic
depolarization).
Neuron
Molecular Asymmetry and Electrical Rectification
(INa+P) to promote cooperativity between afferents via lateral
excitation of neighboring terminals (Figure 6). This voltage-
dependence is also likely to affect anterograde transmission.
Heterotypic channels formed by expression of Cx35 and
Cx34.7 in oocytes exhibited voltage-dependent rectification of
instantaneous current (O’Brien et al., 1998), with properties
consistent with those observed at CE/M-cell contacts, although
the magnitude of the rectification was significantly smaller
(�30%). We re-examined the properties of Cx34.7/Cx35 junc-
tions by expressing these connexins in Rin cells (Figure S6A).
These heterotypic junctions showed both instantaneous and
steady-state properties (Figure S6D) similar to those reported
in oocytes (O’Brien et al., 1998). Such disparity between in vivo
and in vitro behaviors suggested, in addition to the molecular
asymmetry, a possible contribution of cell-specific factors to
generate rectification at either the CEs or M-cell, which could
include the association of ions and charged molecules with con-
nexin-specific residues in one of the hemichannels. We recently
reported that changes in free intracellular [Mg2+] modify the
properties of Cx36 GJ channels, both in cell expression systems
and native electrical synapses (Palacios-Prado et al., 2013). To
test the ability of Cx35 and Cx34.7 hemichannels to promote
rectification in the presence of [Mg2+], we asked if modifications
of free [Mg2+] in only one of the coupled cells (a reduction, in this
case, from 1 mM to 25 mM) could lead to asymmetry of electrical
coupling. This manipulation led to dramatic rectification of both
instantaneous and steady-state conductance-voltage relations
(Figures S6E and S6F). Both Cx34.7 and Cx35 sides were sensi-
tive to changes in free [Mg2+], but remarkably, theywere differen-
tially affected, both qualitatively and quantitatively (compare
instantaneous and steady-state responses in Figures S6E and
S6F). Although Mg2+ is unlikely to be the factor that enhances
rectification under physiological conditions at CE/M-cell synap-
ses, these findings demonstrate that (1) asymmetry of cytosolic
factors can induce rectification and that (2) molecular differences
in heterotypic junctions might contribute to a differential sensi-
tivity of each hemichannel to induce electrical rectification.
Thus, molecular asymmetry may be required but might not be
sufficient to generate strong rectification, and interactions with
cytosolic soluble factors could endow electrical synapses with
complex rectifying properties.
DISCUSSION
Heterotypic Channels Formed by Two Teleost Homologsof Cx36 Mediate Electrical Transmission at CEsWe have previously reported the presence of Cx35 at CEs and
suggested that intercellular channels were likely homotypic but
specifically noted the possible presence of other connexins at
these junctions (Pereda et al., 2003). We report here the
presence of Cx34.7, a second teleost homolog of Cx36 (O’Brien
et al., 1998), at CE/M-cell contacts. Our earlier results are never-
theless consistent with the detailed characterization of
antibodies we report here, which indicates that some of the
Cx36 antibodies previously used (i.e., Ab298) recognize both
Cx35 and Cx34.7, therefore labeling both pre- and postsynaptic
hemiplaques. Members of the connexin protein family can be
permissive or nonpermissive for forming functional intercellular
channels with each other. Heterotypic channels are especially
prominent among glial cells (Rash, 2010) and are found in various
tissues (Elenes et al., 2001), where they provide diversity for
intercellular communication (Rackauskas et al., 2007; Pala-
cios-Prado and Bukauskas, 2009). Heterotypic junctions at
CEs are somewhat unconventional, in that they are formed by
two teleost homologs of a connexin that is normally not permis-
sive for forming intercellular channels with any other connexins.
In tests of the capacity of Cx36 to form channels with ten other
connexin family members, Cx36 was permissive for channel for-
mation only with itself (Teubner et al., 2000). The limited amino
acid sequence difference between Cx34.7 and Cx35 appear
not to have caused sufficient structural changes to render these
connexins incompatible, and indeed, our data show that adult
CE/M-cell GJs gap junctions are formed exclusively from hetero-
typic coupling of these two connexins.
Heterotypic GJs between CEs and the M-cell RectifyAlthough the experimental access did not allow us to perform a
detailed biophysical analysis, our data indicate that these
rectifying junctions are associated with voltage-dependent
properties having kinetics similar to those at the classic crayfish
rectifying synapse (Furshpan and Potter, 1959; Giaume and
Korn, 1984). (These results contrast with a previous report sug-
gesting that electrical synapses at CEs do not rectify [Lin and
Faber, 1988]. The discrepancy with our estimates mainly arises
from differences in the values of AD coupling and dendritic input
resistance used for the calculations of junctional resistance that
Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc. 965
Neuron
Molecular Asymmetry and Electrical Rectification
were critical for revealing the asymmetry.) Heterotypic channels
formed by recombinant Cx32 and Cx26 exhibit rectification
properties (Barrio et al., 1991; Rubin et al., 1992; Bukauskas
et al., 1995) that are reminiscent of those observed at rectifying
synapses in crayfish (Furshpan and Potter, 1959) and hatchetfish
(Auerbach and Bennett, 1969), indicating that molecular asym-
metry between hemichannels might constitute a principal
determinant of electrical rectification. Supporting Furshpan
and Potter’s hypothesis that junctional membranes behave as
a diode (electrical rectifier) rather than a simple electrical resistor
(Furshpan and Potter, 1959), biophysical modeling combined
with genetic analysis of heterotypic Cx32/Cx26 channels (Oh
et al., 1999) suggested that instantaneous rectification of electri-
cal coupling observed at these channels can be explained by the
separation of fixed positive (p) and negative (n) charges across
the junctional membrane, which results from the pairing of hemi-
channels with opposite charges at their channel, leading to the
formation of a diode or ‘‘p-n junction,’’ which was shown to be
capable of generating steep current-voltage relations (Oh
et al., 1999). We show here that, in addition to molecular asym-
metries, cytosolic-soluble cell-specific factors (such as Mg2+)
can contribute substantially to the generation of rectification in
electrical synapses (Figure S6). Furthermore, although both
Cx34.7 and Cx35 sides were sensitive to changes in [Mg2+],
they were differentially affected, indicating that molecular differ-
ences might contribute to a differential sensitivity of each hemi-
channel to soluble factors to enhance electrical rectification.
While Mg2+ is unlikely to be the factor creating rectification under
physiological conditions at CE/M-cell synapses, as yet undeter-
mined channel interacting cytosolic soluble factors (including
intracellular polyamines; Shore et al., 2001; Musa and Veenstra,
2003; Musa et al., 2004) may induce electrical rectification, either
because their concentrations are different on each side of the
junction (coupling in the M-cell occurs between two different
cell types and their intracellular milieus could be different) and/
or by preferentially interacting with hemichannels of one side of
the heterotypic junction. Finally, asymmetry could be also gener-
ated by differences in posttranslational modifications of the
apposing hemichannels, such as connexin phosphorylation,
which may contribute to rectifying properties by altering surface
charge or conformation of the proteins (Alev et al., 2008; O’Brien
et al., 1998).
Rectification Promotes Bidirectionality of ElectricalCommunicationAlthough closely associated with early evidence for electrical
transmission (Furshpan and Potter, 1959), electrical rectification
is an underestimated property of electrical synapses. Notably,
rectification is generally associated with unidirectionality of elec-
trical communication. Our results clearly separate the two
notions (rectification and directionality), as rectification in this
case acts to promote bidirectionality of electrical communica-
tion, which otherwise is challenged by the geometrical character-
istics and electrical properties of theM-cell andCEs.We suggest
that rectification, as in theM-cell, could alsounderlie bidirectional
communication between neuronal processes of dissimilar size
elsewhere, compensating for potentially challenging electrical
and geometrical conditions for the spread of currents.
966 Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc.
The M-cell network mediates auditory-evoked tail-flip
escape responses in teleost fish, and much data support CEs
as having a primary role in generating these responses (Faber
and Pereda, 2011). Because electrical synapses at CEs are
bidirectional, signals originating in the M-cell dendrite can influ-
ence CE excitability (Pereda et al., 1995). We propose that
retrograde transmission is relevant functionally based on the
following: (1) it allows CEs to be electrically coupled to each
other through the lateral dendrite of the M-cell (Figure 6; Per-
eda et al., 1995); (2) this coupling serves as a mechanism for
‘‘lateral excitation’’ (Pereda et al., 1995; Herberholz et al.,
2002; DeVries et al., 2002); (3) lateral excitation promotes the
coordinated activity of a population of CEs; and (4) the coordi-
nated activity likely increases efficacy of auditory input for the
initiation of an escape response. These events are likely
enhanced by electrical rectification, which favors the spread
of currents from the M-cell lateral dendrite toward the presyn-
aptic CEs. That is, because dendritic currents would encounter
a lower resistance to spread across these junctions than those
generated presynaptically, electrical rectification favors the
retrograde transmission of dendritic signals, counteracting the
leak of currents toward the soma, following a pathway of low
resistance (Figure 6). Moreover, given that coupling increases
with presynaptic depolarization, the voltage dependence of
electrical coupling we describe here acts as a ‘‘coincidence
detector,’’ promoting the recruitment of CEs that are already
depolarized, such as during the invasion of an incoming action
potential, whose depolarization (because of cable properties)
travels several nodes ahead without reaching threshold (Fig-
ure 6). Thus, although differences in input resistance signifi-
cantly contribute to the asymmetry of electrical transmission
between these cells, rectification plays a critical functional
role by directing currents toward the presynaptic endings.
Are Hemiplaques in Electrical Synapses the MirrorImages of Each Other?A generalized perception is that each side in most GJ plaques
represents themirror image of the other, as its formation requires
the symmetric arrangement of hemichannels. This view, how-
ever, is changing with the recognition that connexins associate
with a variety of proteins, resulting in the formation of macromo-
lecular complexes (Herve et al., 2012). Furthermore, electrical
synapses have been shown to be dynamic structures, where
connexins actively turnover (Flores et al., 2012) and exhibit activ-
ity-dependent regulation of their coupling strength (Yang et al.,
1990; Pereda and Faber, 1996; Landisman and Connors, 2005;
Cachope et al., 2007; Haas et al., 2011). These properties
suggest that each side in a GJ plaque must be supported by a
scaffold structure, similar to postsynaptic densities at chemical
synapses (Kennedy, 2000). While the detailed composition of
this scaffold is largely unknown, several molecules interact
with Cx36 (Li et al., 2004, 2009; Burr et al., 2005; Ciolofan
et al., 2006; Alev et al., 2008) and its teleost homologs (Flores
et al., 2008, 2010). Thus, molecular diversity in electrical synap-
ses might not only be endowed by the connexins present but
also potentially by differences in the ensemble of scaffold and
regulatory molecules associated with each side of the gap junc-
tions that form these synapses, which could be an additional
Neuron
Molecular Asymmetry and Electrical Rectification
means of creating molecular asymmetry, impacting on the func-
tional properties of these channels.
EXPERIMENTAL PROCEDURES
For full methodological details see the Supplemental Experimental
Procedures.
Immunohistochemistry
Goldfish brains were fixed and sections were incubated simultaneously with
rabbit polyclonal anti-Cx34.7 IL antibody (see Table S1) and mouse
monoclonal anti-Cx35 (Chemicon MAB3043) antibody, incubated with Alexa
Fluor 488-conjugated goat anti-rabbit and/or Alexa Fluor 594-conjugated
goat anti-mouse secondary antibodies, and coverslipped using n-propyl
gallate-based mounting media.
Confocal Microscopy and Image Processing
Sections were imaged using anOlympusBX61WI confocal microscope. Image
analysis was performed using Image J (National Institutes of Health [NIH]) and
MetaMorph software. Colocalization of Cx35 andCx34.7wasmeasured as the
percentage of the area labeled for Cx35 that was also labeled for Cx34.7 and
the converse.
Antibodies
The antibodies used are listed in Table S1 along with their species of origin,
designation, epitope recognition, source, and characteristics of detection of
either or both Cx34.7 and Cx35.
Freeze-Fracture Replica Immunogold Labeling
Specimens were processed by single-replica FRIL and one additional spec-
imen by double-replica SDS-FRIL (sodium dodecyl sulfate-digested fracture
replica labeling, which we designate as DR-FRIL). For single-replica samples,
a gold ‘‘index’’ grid (aka ‘‘Finder’’ grid) was bonded to the coated surfaces
using Lexan plastic (polycarbonate plastic) dissolved in dichloroethane; the
samples were thawed and ‘‘grid-mapped’’ by confocal microscopy, with
which the location of the M-cell lateral dendrite was determined.
Double-Replica FRIL
A 150-mm-thick slice of goldfish hindbrain containing a Lucifer Yellow-injected
M-cell was cryoprotected and fractured at �105�C in a prototype JFD-2
freeze-etch machine equipped with a turbopump but lacking a liquid-nitro-
gen-cooled shroud. The opened double-replica ‘‘sandwich’’ was coated with
3–5 nm of carbon, 1.5 nm of platinum, and �20 nm of carbon.
Electrophysiology
Surgical and recording techniques were similar to those described previously
(Smith and Pereda, 2003; Curti and Pereda, 2004). Intracellular recordings
were obtained in vivo from the lateral dendrite; both current clamp and sin-
gle-electrode voltage clamp techniques were employed. Individual VIIIth nerve
afferents were penetrated, either at the posterior VIIIth root during simulta-
neous recordings with the M-cell’s lateral dendrite or less often, intracranially,
close to the dendrite.
Estimate of Junctional Resistance
Junctional resistance in each direction was estimated following Devor and
Yarom (2002). These estimates of the junctional resistance assume a simple
two neuron model with passive membrane properties coupled directly by a
single junction.
In Vitro Electrophysiological Measurements
Experiments were performed on Rin cells expressing Cx35 or Cx34.7
tagged with eYFP or DsRed, respectively. To adjust the concentration of
intracellular free Mg2+, we used pipette solutions containing different concen-
trations of MgCl2 and EDTA and the web-based Maxchelator software
(http://www.stanford.edu/�cpatton/webmaxcS.htm) to calculate free ionic
concentrations.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures,
six figures, and one table and can be found with this article online at http://
dx.doi.org/10.1016/j.neuron.2013.06.037.
ACKNOWLEDGMENTS
We thank Michael V.L. Bennett and Peter Sterling for their comments on
the manuscript. This research was supported by National Institutes of
Health grants S10RR05831, S10RR08329, NS044395, and NS044010 (to
J.E.R.), EY012857 (to J.O.), RO1NS072238 (to F.F.B.), Canadian Institute for
Health Research (to J.I.N.), and by NIH grants DC03186, DC011099,
R21NS055726, and NS0552827 (to A.E.P.).
Accepted: June 13, 2013
Published: September 4, 2013
REFERENCES
Alev, C., Urschel, S., Sonntag, S., Zoidl, G., Fort, A.G., Hoher, T., Matsubara,
M., Willecke, K., Spray, D.C., and Dermietzel, R. (2008). The neuronal con-
nexin36 interacts with and is phosphorylated by CaMKII in a way similar to
CaMKII interaction with glutamate receptors. Proc. Natl. Acad. Sci. USA
105, 20964–20969.
Auerbach, A.A., and Bennett, M.V. (1969). A rectifying electrotonic synapse in
the central nervous system of a vertebrate. J. Gen. Physiol. 53, 211–237.
Barrio, L.C., Suchyna, T., Bargiello, T., Xu, L.X., Roginski, R.S., Bennett, M.V.,
and Nicholson, B.J. (1991). Gap junctions formed by connexins 26 and 32
alone and in combination are differently affected by applied voltage. Proc.
Natl. Acad. Sci. USA 88, 8410–8414.
Bennett, M.V. (1966). Physiology of electrotonic junctions. Ann. N Y Acad. Sci.
137, 509–539.
Bennett, M.V.L. (1977). Electrical transmission, a functional analysis and com-
parison with chemical transmission. In Cellular Biology of Neurons, Volume I,
The Handbook of Physiology - The Nervous System I, E.R. Kandel, ed.
(Baltimore: Williams and Wilkins), pp. 357–416.
Bukauskas, F.F., Elfgang, C., Willecke, K., and Weingart, R. (1995).
Heterotypic gap junction channels (connexin26-connexin32) violate the para-
digm of unitary conductance. Pflugers Arch. 429, 870–872.
Burr, G.S., Mitchell, C.K., Keflemariam, Y.J., Heidelberger, R., and O’Brien, J.
(2005). Calcium-dependent binding of calmodulin to neuronal gap junction
proteins. Biochem. Biophys. Res. Commun. 335, 1191–1198.
Cachope, R., Mackie, K., Triller, A., O’Brien, J., and Pereda, A.E. (2007).
Potentiation of electrical and chemical synaptic transmission mediated by
endocannabinoids. Neuron 56, 1034–1047.
Ciolofan, C., Li, X.B., Olson, C., Kamasawa, N., Gebhardt, B.R., Yasumura, T.,
Morita, M., Rash, J.E., and Nagy, J.I. (2006). Association of connexin36 and
zonula occludens-1 with zonula occludens-2 and the transcription factor
zonula occludens-1-associated nucleic acid-binding protein at neuronal gap
junctions in rodent retina. Neuroscience 140, 433–451.
Condorelli, D.F., Parenti, R., Spinella, F., Trovato Salinaro, A., Belluardo, N.,
Cardile, V., and Cicirata, F. (1998). Cloning of a new gap junction gene
(Cx36) highly expressed in mammalian brain neurons. Eur. J. Neurosci. 10,
1202–1208.
Connors, B.W., and Long, M.A. (2004). Electrical synapses in the mammalian
brain. Annu. Rev. Neurosci. 27, 393–418.
Cruciani, V., and Mikalsen, S.O. (2007). Evolutionary selection pressure and
family relationships among connexin genes. Biol. Chem. 388, 253–264.
Curti, S., and Pereda, A.E. (2004). Voltage-dependent enhancement of electri-
cal coupling by a subthreshold sodium current. J. Neurosci. 24, 3999–4010.
Curti, S., Gomez, L., Budelli, R., and Pereda, A.E. (2008). Subthreshold sodium
current underlies essential functional specializations at primary auditory affer-
ents. J. Neurophysiol. 99, 1683–1699.
Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc. 967
Neuron
Molecular Asymmetry and Electrical Rectification
Devor, A., and Yarom, Y. (2002). Electrotonic coupling in the inferior olivary
nucleus revealed by simultaneous double patch recordings. J. Neurophysiol.
87, 3048–3058.
DeVries, S.H., Qi, X., Smith, R., Makous, W., and Sterling, P. (2002). Electrical
coupling between mammalian cones. Curr. Biol. 12, 1900–1907.
Eastman, S.D., Chen, T.H., Falk, M.M., Mendelson, T.C., and Iovine, M.K.
(2006). Phylogenetic analysis of three complete gap junction gene families re-
veals lineage-specific duplications and highly supported gene classes.
Genomics 87, 265–274.
Elenes, S., Martinez, A.D., Delmar, M., Beyer, E.C., and Moreno, A.P. (2001).
Heterotypic docking of Cx43 and Cx45 connexons blocks fast voltage gating
of Cx43. Biophys. J. 81, 1406–1418.
Faber, D.S., and Pereda, A.E. (2011). Physiology of the Mauthner Cell,
Functions. In Encyclopedia of Fish Physiology, From Genome to
Environment, A.P. Farrell, ed. (San Diego: Academic Press), pp. 73–79.
Flores, C.E., Li, X., Bennett, M.V., Nagy, J.I., and Pereda, A.E. (2008).
Interaction between connexin35 and zonula occludens-1 and its potential
role in the regulation of electrical synapses. Proc. Natl. Acad. Sci. USA 105,
12545–12550.
Flores, C.E., Cachope, R., Nannapaneni, S., Ene, S., Nairn, A.C., and Pereda,
A.E. (2010). Variability of distribution of Ca(2+)/calmodulin-dependent kinase II
at mixed synapses on the mauthner cell: colocalization and association with
connexin 35. J. Neurosci. 30, 9488–9499.
Flores, C.E., Nannapaneni, S., Davidson, K.G., Yasumura, T., Bennett, M.V.,
Rash, J.E., and Pereda, A.E. (2012). Trafficking of gap junction channels at a
vertebrate electrical synapse in vivo. Proc. Natl. Acad. Sci. USA 109, E573–
E582.
Fujimoto, K. (1995). Freeze-fracture replica electron microscopy combined
with SDS digestion for cytochemical labeling of integral membrane proteins.
Application to the immunogold labeling of intercellular junctional complexes.
J. Cell Sci. 108, 3443–3449.
Fukami, Y., Furukawa, T., and Asada, Y. (1965). Excitability changes of the
Mauthner cell during collateral inhibition. J. Gen. Physiol. 48, 581–600.
Furshpan, E.J., and Potter, D.D. (1959). Transmission at the giant motor syn-
apses of the crayfish. J. Physiol. 145, 289–325.
Furshpan, E.J., and Furukawa, T. (1962). Intracellular and extracellular re-
sponses of the several regions of the Mauthner cell of the goldfish.
J. Neurophysiol. 25, 732–771.
Giaume, C., and Korn, H. (1984). Voltage-dependent dye coupling at a recti-
fying electrotonic synapse of the crayfish. J. Physiol. 356, 151–167.
Giaume, C., Kado, R.T., and Korn, H. (1987). Voltage-clamp analysis of a cray-
fish rectifying synapse. J Physiol. 386, 91–112.
Haas, J.S., Zavala, B., and Landisman, C.E. (2011). Activity-dependent long-
term depression of electrical synapses. Science 334, 389–393.
Herberholz, J., Antonsen, B.L., and Edwards, D.H. (2002). A lateral excitatory
network in the escape circuit of crayfish. J. Neurosci. 22, 9078–9085.
Herve, J.C., Derangeon,M., Sarrouilhe, D., Giepmans, B.N., and Bourmeyster,
N. (2012). Gap junctional channels are parts of multiprotein complexes.
Biochim. Biophys. Acta 1818, 1844–1865.
Kennedy, M.B. (2000). Signal-processing machines at the postsynaptic den-
sity. Science 290, 750–754.
Landisman, C.E., and Connors, B.W. (2005). Long-term modulation of electri-
cal synapses in the mammalian thalamus. Science 310, 1809–1813.
Li, X., Olson, C., Lu, S., Kamasawa, N., Yasumura, T., Rash, J.E., and Nagy, J.I.
(2004). Neuronal connexin36 association with zonula occludens-1 protein (ZO-
1) in mouse brain and interaction with the first PDZ domain of ZO-1. Eur. J.
Neurosci. 19, 2132–2146.
Li, X., Lu, S., and Nagy, J.I. (2009). Direct association of connexin36 with zon-
ula occludens-2 and zonula occludens-3. Neurochem. Int. 54, 393–402.
Lin, J.W., and Faber, D.S. (1988). Synaptic transmission mediated by single
club endings on the goldfish Mauthner cell. I. Characteristics of electrotonic
and chemical postsynaptic potentials. J. Neurosci. 8, 1302–1312.
968 Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc.
Lin, J.W., Faber, D.S., and Wood, M.R. (1983). Organized projection of the
goldfish saccular nerve onto the Mauthner cell lateral dendrite. Brain Res.
274, 319–324.
Musa, H., and Veenstra, R.D. (2003). Voltage-dependent blockade of con-
nexin40 gap junctions by spermine. Biophys. J. 84, 205–219.
Musa, H., Fenn, E., Crye, M., Gemel, J., Beyer, E.C., and Veenstra, R.D. (2004).
Amino terminal glutamate residues confer spermine sensitivity and affect
voltage gating and channel conductance of rat connexin40 gap junctions.
J. Physiol. 557, 863–878.
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 define
a distinct subgroup of the connexin family. J. Neurosci. 18, 7625–7637.
Oh, S., Rubin, J.B., Bennett, M.V., Verselis, V.K., and Bargiello, T.A. (1999).
Molecular determinants of electrical rectification of single channel conduc-
tance in gap junctions formed by connexins 26 and 32. J. Gen. Physiol. 114,
339–364.
Orthmann-Murphy, J.L., Freidin, M., Fischer, E., Scherer, S.S., and Abrams,
C.K. (2007). Two distinct heterotypic channels mediate gap junction coupling
between astrocyte and oligodendrocyte connexins. J. Neurosci. 27, 13949–
13957.
Palacios-Prado, N., and Bukauskas, F.F. (2009). Heterotypic gap junction
channels as voltage-sensitive valves for intercellular signaling. Proc. Natl.
Acad. Sci. USA 106, 14855–14860.
Palacios-Prado, N., Hoge, G., Marandykina, A., Rimkute, L., Chapuis, S.,
Paulauskas, N., Skeberdis, V.A., O’Brien, J., Pereda, A.E., Bennett, M.V.,
and Bukauskas, F.F. (2013). Intracellular magnesium-dependent modulation
of gap junction channels formed by neuronal connexin36. J. Neurosci. 33,
4741–4753.
Pereda, A.E., and Faber, D.S. (1996). Activity-dependent short-term enhance-
ment of intercellular coupling. J. Neurosci. 16, 983–992.
Pereda, A.E., Bell, T.D., and Faber, D.S. (1995). Retrograde synaptic commu-
nication via gap junctions coupling auditory afferents to the Mauthner cell.
J. Neurosci. 15, 5943–5955.
Pereda, A., O’Brien, J., Nagy, J.I., Bukauskas, F., Davidson, K.G., Kamasawa,
N., Yasumura, T., and Rash, J.E. (2003). Connexin35mediates electrical trans-
mission at mixed synapses on Mauthner cells. J. Neurosci. 23, 7489–7503.
Pereda, A.E., Rash, J.E., Nagy, J.I., and Bennett, M.V. (2004). Dynamics of
electrical transmission at club endings on the Mauthner cells. Brain Res.
Brain Res. Rev. 47, 227–244.
Phelan, P., Goulding, L.A., Tam, J.L., Allen, M.J., Dawber, R.J., Davies, J.A.,
andBacon, J.P. (2008). Molecularmechanism of rectification at identified elec-
trical synapses in the Drosophila giant fiber system. Curr. Biol. 18, 1955–1960.
Rackauskas, M., Verselis, V.K., and Bukauskas, F.F. (2007). Permeability of
homotypic and heterotypic gap junction channels formed of cardiac connexins
mCx30.2, Cx40, Cx43, and Cx45. Am. J. Physiol. Heart Circ. Physiol. 293,
H1729–H1736.
Rash, J.E. (2010). Molecular disruptions of the panglial syncytium block potas-
sium siphoning and axonal saltatory conduction, pertinence to neuromyelitis
optica and other demyelinating diseases of the central nervous system.
Neuroscience 168, 982–1008.
Rubin, J.B., Verselis, V.K., Bennett, M.V., and Bargiello, T.A. (1992). Molecular
analysis of voltage dependence of heterotypic gap junctions formed by con-
nexins 26 and 32. Biophys. J. 62, 183–193.
Shore, L., McLean, P., Gilmour, S.K., Hodgins, M.B., and Finbow, M.E. (2001).
Polyamines regulate gap junction communication in connexin 43-expressing
cells. Biochem. J. 357, 489–495.
Smith, M., and Pereda, A.E. (2003). Chemical synaptic activity modulates
nearby electrical synapses. Proc. Natl. Acad. Sci. USA 100, 4849–4854.
Teubner, B., Degen, J., Sohl, G., Guldenagel, M., Bukauskas, F.F., Trexler,
E.B., Verselis, V.K., De Zeeuw, C.I., Lee, C.G., Kozak, C.A., et al. (2000).
Functional expression of the murine connexin 36 gene coding for a neuron-
specific gap junctional protein. J. Membr. Biol. 176, 249–262.
Neuron
Molecular Asymmetry and Electrical Rectification
Tuttle, R., Masuko, S., and Nakajima, Y. (1986). Freeze-fracture study of the
large myelinated club ending synapse on the goldfish Mauthner cell: special
reference to the quantitative analysis of gap junctions. J. Comp. Neurol. 246,
202–211.
Volff, J.N. (2005). Genome evolution and biodiversity in teleost fish. Heredity
(Edinb) 94, 280–294.
Yang, X.D., Korn, H., and Faber, D.S. (1990). Long-term potentiation of electro-
tonic coupling at mixed synapses. Nature 348, 542–545.
Yum, S.W., Zhang, J., Valiunas, V., Kanaporis, G., Brink, P.R., White, T.W., and
Scherer, S.S. (2007). Human connexin26 and connexin30 form functional het-
eromeric and heterotypic channels. Am. J. Physiol. Cell Physiol. 293, C1032–
C1048.
Neuron 79, 957–969, September 4, 2013 ª2013 Elsevier Inc. 969