Vestibuloocular Reflex of the Adult Flatfish. III. A Species-Specific
Reciprocal Pattern of Excitation and Inhibition
WERNER GRAF,1 ROBERT SPENCER,2,✠ HARRIET BAKER,3 AND ROBERT BAKER4
1
Laboratoire de Physiologie de la Perception et de l’Action, College de France—Centre National de la Recherche
Scientifique, F-75231 Paris Cedex 05, France; 2Department of Anatomy, Medical College of Virginia, Richmond, Virginia
23298; 3Department of Neurology, Burke Rehabilitation Center, White Plains 10605; and 4Department of Physiology and
Neuroscience, New York University Medical Center, New York, New York 10016
Received 10 December 1999; accepted in final form 13 April 2001
✠ Deceased 24 March 2001.
Address for reprint requests: R. Baker, Dept. of Physiology and Neuroscience, New York University School of Medicine, 550 First Ave., New York,
NY 10016 (E-mail: [email protected]).
1376
tions. Altogether these data provide electrophysiological, immunohistochemical, and ultrastructural evidence for reciprocal excitatory/
inhibitory organization of the novel vestibulooculomotor projections
in adult flatfish. The appearance of unique second-order vestibular
neurons linking the horizontal canal to vertical oculomotor neurons
suggests that reciprocal excitation and inhibition are a fundamental,
developmentally linked trait of compensatory eye movement circuits
in vertebrates.
INTRODUCTION
Flatfish are a natural paradigm to study a developmental
adaptation to a changing living situation. Larvae begin their
lives in an upright body posture with the eyes placed on the
left and right sides of the head (Fig. 1A). At this stage,
vestibuloocular reflex (VOR) eye movements elicited during
swimming involve detection of horizontal head rotation
about the vertical axis by the horizontal semicircular canals.
Horizontal conjugate eye movements are produced by the
lateral and medial rectus eye muscles in the earth-horizontal
plane (Graf and Baker 1985a) (Fig. 1A). The neuronal
circuitry mediating this reflex involves the classical threeneuron arcs along with the specialized abducens internuclear pathway (Fig. 1A).
During metamorphosis, flatfish rotate 90° about the longitudinal axis to become better adapted to an adult lifestyle on the
bottom. The eye ipsilateral to the side that now faces the sea
bottom migrates across the dorsal aspect of the head to the
upper side of the flatfish (Fig. 1B). The eyes, including the
extraocular muscle apparatus retain their original orientation
with respect to the environment; however, since the labyrinths
remain in their original position in the head (Fig. 1B), the
horizontal canals detect swimming motions that now occur in
the vertical plane. The appropriate compensatory eye movements, by contrast, are parallel backward and forward rotations
of the eyes produced by contraction of vertical eye muscles.
This behavior requires a principal VOR circuitry connecting
the horizontal canals to vertical extraocular muscles as shown
in Fig. 1B (Graf and Baker 1983, 1985b). The previous neuroanatomical experiments described the structural basis of the
The costs of publication of this article were defrayed in part by the payment
of page charges. The article must therefore be hereby marked ‘‘advertisement’’
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
0022-3077/01 $5.00 Copyright © 2001 The American Physiological Society
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
Graf, Werner, Robert Spencer, Harriet Baker, and Robert Baker.
Vestibuloocular reflex of the adult flatfish. III. A species-specific
reciprocal pattern of excitation and inhibition. J Neurophysiol 86:
1376 –1388, 2001. In juvenile flatfish the vestibuloocular reflex
(VOR) circuitry that underlies compensatory eye movements adapts
to a 90° relative displacement of vestibular and oculomotor reference
frames during metamorphosis. VOR pathways are rearranged to allow
horizontal canal-activated second-order vestibular neurons in adult
flatfish to control extraocular motoneurons innervating vertical eye
muscles. This study describes the anatomy and physiology of identified flatfish-specific excitatory and inhibitory vestibular pathways. In
antidromically identified oculomotor and trochlear motoneurons, excitatory postsynaptic potentials (EPSPs) were elicited after electrical
stimulation of the horizontal canal nerve expected to provide excitatory input. Electrotonic depolarizations (0.8 – 0.9 ms) preceded small
amplitude (⬍0.5 mV) chemical EPSPs at 1.2–1.6 ms with much larger
EPSPs (⬎1 mV) recorded around 2.5 ms. Stimulation of the opposite
horizontal canal nerve produced inhibitory postsynaptic potentials
(IPSPs) at a disynaptic latency of 1.6 –1.8 ms that were depolarizing
at membrane resting potentials around ⫺60 mV. Injection of chloride
ions increased IPSP amplitude, and current-clamp analysis showed the
IPSP equilibrium potential to be near the membrane resting potential.
Repeated electrical stimulation of either the excitatory or inhibitory
horizontal canal vestibular nerve greatly increased the amplitude of
the respective synaptic responses. These observations suggest that the
large terminal arborizations of each VOR neuron imposes an electrotonic load requiring multiple action potentials to maximize synaptic
efficacy. GABA antibodies labeled axons in the medial longitudinal
fasciculus (MLF) some of which were hypothesized to originate from
horizontal canal-activated inhibitory vestibular neurons. GABAergic
terminal arborizations were distributed largely on the somata and
proximal dendrites of oculomotor and trochlear motoneurons. These
findings suggest that the species-specific horizontal canal inhibitory
pathway exhibits similar electrophysiological and synaptic transmitter
profiles as the anterior and posterior canal inhibitory projections to
oculomotor and trochlear motoneurons. Electron microscopy showed
axosomatic and axodendritic synaptic endings containing spheroidal
synaptic vesicles to establish chemical excitatory synaptic contacts
characterized by asymmetrical pre/postsynaptic membrane specializations as well as gap junctional contacts consistent with electrotonic
coupling. Another type of axosomatic synaptic ending contained
pleiomorphic synaptic vesicles forming chemical, presumed inhibitory, synaptic contacts on motoneurons that never included gap junc-
EXCITATION AND INHIBITION IN FLATFISH VOR
novel horizontal canal-to-vertical eye muscle connections, and
the current work addresses the electrophysiology of the excitatory and inhibitory pathways.
In all vertebrates, VOR pathways exhibit a reciprocal excitatory and inhibitory innervation of extraocular motoneurons in
which a stereotyped, constant relationship is maintained between each of the three semicircular canals (anterior, posterior,
and horizontal) and one extraocular muscle pair (reviewed in
Evinger 1988; Graf 1988; Graf and Ezure 1986; Graf et al.
J Neurophysiol • VOL
1983). In particular, with regard to the vertical canals (anterior
and posterior), second-order VOR neurons exhibit a unilateral
innervation pattern of the oculomotor neurons (Graf and Ezure
1986; Graf et al. 1983, 1997). Excitatory VOR neurons activated from the anterior canal (AC) contact contralateral superior rectus (SR) and inferior oblique (IO) motoneurons,
whereas posterior canal (PC) VOR neurons terminate on contralateral superior oblique (SO; i.e., trochlear) and inferior
rectus (IR) motoneurons. Inhibitory pathways originating from
the same set of canals reach the motoneuron populations of the
antagonists of the above muscles via ipsilateral projections.
Thus each vertical oculomotor neuron receives disynaptic excitatory and inhibitory input from the appropriate contralateral
and ipsilateral semicircular canals, respectively. This representative VOR plan has been described in all vertebrates including
flatfish (Baker et al. 1973; Graf et al. 1997; Magherini et al.
1974; Precht and Baker 1972).
By contrast to the above-described horizontal canal reflex
circuitry, adult flatfish develop a “novel” set of second-order
vestibular neuron connections with extraocular motoneurons
hypothesized to perform a VOR role appropriate for the 90°lateral rotation postural adaptation (Graf and Baker 1983,
1985b). These previous experiments showed that vestibular
neurons activated from the left horizontal semicircular canal
(HC) nerve in the flatfish terminated bilaterally either in the
oculomotor nuclei in motoneuron pools containing SR and IO
motoneurons (Fig. 1B, yellow), or in the trochlear and oculomotor nuclei with SO and IR motoneurons (Fig. 1B, blue). The
first termination pattern was considered to be excitatory and the
second, inhibitory. The parent axons of both these prospective
excitatory and inhibitory vestibular neurons crossed the midline in the hindbrain, and axon collaterals then recrossed the
midline within the midbrain to contact oculomotor motoneurons on both sides.
Experiments in the right-side winter flounder, Pseudopleuronectes americanus, identified vestibular neurons by electrical
stimulation of the downside (left) horizontal canal nerve
branches (lHC) and injected horseradish peroxidase (HRP) for
subsequent morphological reconstruction (Graf and Baker
1983, 1985b). The pattern of oculomotor termination was
considered appropriate to mediate parallel compensatory eye
movements during head motion (Fig. 1B). In later experiments,
second-order vestibular neurons were labeled from the upside
(right) horizontal canal system of winter flounders (Graf and
Baker 1986; Graf et al. 1991). Together, these anatomical
reconstructions unveiled a bilaterally symmetric innervation
scheme of flatfish horizontal canal VOR pathways that would
be well suited for the requirements of compensatory eye movements in postmetamorphic animals within the conceptual
framework of reciprocal excitatory/inhibitory innervation of
the VOR.
The present studies therefore were aimed at corroborating
this reciprocal excitatory/inhibitory projection onto oculomotor motoneurons in adult flatfish with electrophysiology and
anatomy. Evaluating the pattern of excitatory and inhibitory
postsynaptic potential (EPSP and IPSP, respectively) allowed
the operational mode of this species-specific pathway to be
tested as well as a basis for structure/function comparison to be
established with the reciprocal excitatory/inhibitory organization of other vertebrate eye movement circuits (Graf and Ezure
1986; Graf et al. 1997). The physiology of the excitatory/
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
FIG. 1. Horizontal canal vestibuloocular reflex (VOR) circuitry in pre- and
postmetamorphic right-sided flatfish. A: in the premetamorphic flatfish, labyrinths (red arrows) and eye axes (green arrows) are parallel (0° difference) and
the horizontal canal VOR pathways function as in other vertebrates. Horizontal
semicircular canal activation during horizontal head movement (i.e., vertical
axis rotation) reciprocally excites and inhibits the horizontal eye muscles, the
lateral rectus and medial rectus. As illustrated, head movement to the left
(circular black arrows) would excite the left horizontal canal with centripetal
endolymph flow as indicated by the red arrow. Compensatory horizontal eye
movements would be conjugate to the right (curved arrows above the eyes)
produced by contraction of the left medial rectus (MR) and right lateral rectus
(LR; shown in yellow) with relaxation of the antagonistic muscles (shown in
blue). Excitatory (red) and inhibitory (blue) 3-neuron arc pathways constitute
the classical horizontal VOR reflex circuitry. Two special pathways innervate
medial rectus motoneurons (MR MNs) in the oculomotor nucleus (III) via the
ascending tract of Deiters (ATD) and the abducens internuclear neurons (INT)
located at the level of the abducens (ABD) nucleus (VI). B: in the postmetamorphic animal, labyrinths and eye axes are oriented 90° in respect to each
other. The now vertically oriented horizontal canals detect vertical head
movements during swimming. A head-down movement in adult right-sided
flatfish (circular black arrows next to the labyrinth) excites the left (down-side)
horizontal canal via ampullopetal endolymph current (red arrow). Compensatory eye movements are parallel upward rotations of both eyes produced by
contraction (yellow) and relaxation (blue) of bilateral vertically yoked eye
muscles via the novel flatfish-specific VOR circuitry linked to the horizontal
canals (Graf and Baker 1985b). Excitatory and inhibitory 3-neuron arc pathways are shown in red and blue. This flatfish-specific VOR-circuitry contacts
four bilateral sets of motoneurons located in the oculomotor/trochlear nuclei
via specific excitatory and inhibitory 2nd-order vestibular neurons. SR, superior rectus; IO, inferior oblique; IR, inferior rectus; SO, superior oblique; IV,
trochlear nucleus.
1377
1378
W. GRAF, R. SPENCER, H. BAKER, AND R. BAKER
inhibitory innervation pattern of VOR pathways in flatfish was
extended with immunohistochemistry and electron microscopy
for comparison to similar integrated approaches undertaken in
oculomotor and postural control in goldfish (Aksay et al. 2000;
Baker et al. 1998; Faber et al. 1989; Korn et al. 1992; Kumar
and Faber 1999; Straka and Baker 1998; Suwa et al. 1999).
Preliminary results of this study have been previously summarized in abstract form (Graf and Baker 1984; Graf et al.
1985; Spencer et al. 1986).
METHODS
Surgical procedures
Animals were mounted in a “vertical” position in a fish-holder
system (Graf and Baker 1985b), and cooled (18°C), aerated seawater
was continuously circulated over the gills by an electric pump through
a tube in the mouth. Animals were kept immersed as deeply as
possible in water to submerge the gills. All surgery and physiological
experiments were done under tricaine methanesulfonate (1:20,000
wt/vol) anesthesia. In addition, incisions were infiltrated with local
anesthetics (2% lidocaine with epinephrine). During recording, animals were immobilized with gallamine triethiodide (Flaxedil). Access
to the labyrinths was obtained by removal of the skull bone overlying
the fourth ventricle. Individual labyrinthine nerve branches to the
bilateral anterior, horizontal, and posterior canal ampullae, as well as
the utricular, saccular, and lagenar endorgans were dissected and
exposed. The bone overlying the midbrain was removed, so the
medial longitudinal fasciculus (MLF) could be reached through the
intertectal cleft. Bipolar electrodes were positioned on the ampullary
nerves for orthodromic activation of vestibular pathways to extraocular motoneurons. The stimulation electrodes were held in place by
Narishige microdrives. The horizontal canal nerves were selectively
isolated and stimulated in 19 experiments, the vertical canals in 5
cases, and the entire vestibular nerve in the 2 other experiments. In all
experiments, two silver ball electrodes were placed into the orbits of
both eyes to activate antidromically extraocular motoneurons.
Electrophysiology
Vestibular synaptic input to oculomotor motoneurons was recorded
with glass micropipettes containing 1 M KCl or 1 M KAcetate
(approximately 20 –30 M⍀ resistance, 0.5–1.0 ␮m tip size). The
recording electrodes were positioned and advanced by a three-axis
micromanipulator (Narishige, Canberra type). Motoneurons were penetrated in the oculomotor and the trochlear nucleus. These nuclei were
reached through the intertecal cleft that was widened by cutting the
commissural fibers for direct visualization of the third ventricle.
Earlier neuroanatomical studies had indicated the location of the
various motoneuron pools (Graf and Baker 1985a), which were identified by their antidromic potentials following electrical activation (70
␮A) via the stimulation electrodes located in the orbits. Subsequently,
synaptic potentials were elicited by electrical stimulation from the
labyrinthine electrodes (20 ␮A). Each penetrated motoneuron could
be identified by its location within the oculomotor nuclei with respect
to the eye from which it was antidromically activated. The electroJ Neurophysiol • VOL
Neurotransmitters
Protocols for GABA immunohistochemistry were described in detail in Graf et al. (1997) and are briefly summarized here. Animals
were perfused with fixative solution containing 4.0% paraformaldehyde in 0.1 M phosphate buffer with 0.002% calcium chloride (pH
7.2). Vibratome sections were obtained through the abducens, trochlear, and oculomotor nuclei (25–50 ␮m) and subsequently processed
for immunohistochemical localization of GABA using an antibody
generated in rabbit against GABA conjugated to bovine serum albumin (Immunonuclear). Following primary and secondary antibody
incubation, the tissue was incubated in an avidin/biotin-HRP complex
(Vector). Addition of the chromogen 3,3⬘-diaminobenzidine and hydrogen peroxide produced a brown diffuse reaction product. The
material was mounted for light microscopic examination using brightfield or Normarski differential interference contrast optics.
Electron microscopy
Protocols were identical to those described in Graf et al. (1997) and
are summarized here briefly. Animals were anesthetized and perfused
with 1% paraformaldehyde and 1.25% glutaraldehyde in 0.1 M phosphate buffer with 0.002 calcium chloride (pH 7.4). Serial coronal
sections (50 –75 ␮m) through the oculomotor nuclei were cut on a
vibratome. Sections containing the oculomotor nuclei were trimmed,
postfixed in 1% osmium tetroxide, stained en bloc with 0.5% uranyl
acetate, dehydrated, and embedded in resin between vinyl plastic
microscope slides and cover slips.
Selected loci containing oculomotor neurons were analyzed by
serial ultrathin sectioning and electron microscopy. Serial sections (5
␮m) through the plastic-embedded vibratome sections were examined
by light microscopy and remounted on blank BEEM capsules from
which ultrathin sections were cut and collected on single-slot Formvar-coated grids. Sections were then stained with uranyl acetate and
lead citrate, examined, and photographed with a Zeiss EM-10CA
electron microscope. The same procedure was followed in three
animals who had received, 1 day earlier, an intracellular injections of
HRP into second-order vestibular neurons of the horizontal canal
system. The brains were also treated for HRP histochemistry before
selected sections were prepared for electron microscopy (Graf and
Baker 1985a,b).
RESULTS
A total of 102 extraocular motoneurons was recorded in the
trochlear, oculomotor, and abducens nuclei. The latency for
antidromic invasion averaged 0.7 ms. Disynaptic depolarizing
synaptic potentials were observed in all motoneurons following
whole vestibular nerve branch stimulation with latencies between 1.2 and 1.8 ms. As expected, selective activation of
individual canal nerves also evoked disynaptic depolarizations
that were shown to be EPSPs in the appropriate populations of
motoneurons (Figs. 2– 4). However, stimulation of the presumed antagonistic canal nerve supposed to provide inhibitory
input, typically produced membrane depolarizations as well
(Figs. 2, C and D, and 4C). Hyperpolarizing IPSPs were
recorded in only nine motoneurons (i.e., 9% of all cases; Fig.
3, B and D), and these IPSPs were about equally distributed
between the vertical oculomotor subgroups (2 each in IR, IO,
and SR motoneurons, and 3 in SO motoneurons). The existence
of inhibition was supported by finding GABAergic contacts on
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
All experiments were performed at the Marine Biological Laboratory, Woods Hole, Massachusetts, in 26 adult (25–30 cm body
length), right-sided winter flounders, Pseudopleuronectes americanus. Neither left-sided examples of this species (Policanski 1982) nor
the indigenous left-sided summer flounder, Paralichthys dentatus
were examined in this study. Experimental flounders were wildcaught by the trawler of the institution in the Atlantic Ocean off Cape
Cod and kept in large indoor-outdoor tanks with cooled running
seawater of 18°C.
physiological data were recorded either on magnetic tape (Neurocorder: Data 6000), or on magnetic disks (Nicolet 4096), and later
printed out on a laser printer for analysis and quantification.
EXCITATION AND INHIBITION IN FLATFISH VOR
1379
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
FIG. 2. Synaptic potentials in 2 left-side trochlear motoneurons (lTro) following horizontal canal nerve stimulations. A: antidromic
identification of somatic recording showing IS-SD break and M-spike (top and bottom arrows, respectively). Top and bottom traces are AC- and
DC-coupled records, respectively. B: AC recordings of an excitatory postsynaptic potential (EPSP) following contralateral (right-side) horizontal
canal nerve (rHCe) stimulation with double shocks at 2 times threshold (2 ⫻ T). Averaged (n ⫽ 8) and single sweeps are shown in the top and
bottom traces, respectively. Electrotonic coupling appears at 0.8 ms and the onset of chemical depolarization at 1.6 ms. Action potentials are
shown in the lower DC-coupled record labeled lTRO Mn. An intra-axonal DC record from a VOR neuron following rHCe stimulation is shown
in the 3rd trace [VOR axon in medial longitudinal fasciculus (MLF)]. The vertical arrows point out synchronization between the 1st vestibular
action potential and the electrotonic potential (E). Calibrations for A are shown in B. C: comparison of synaptic potential latencies elicited from
stimulation of the excitatory (rHCe; top 3 traces) and the inhibitory sides (lHCi; bottom 3 traces). AC recordings are averaged (n ⫽ 8) with the
2nd and 3rd traces showing single AC and DC sweeps, respectively. EPSPs following rHCe stimulation (1.5 ⫻ T) exhibited electrotonic coupling
(E) at 0.8 ms, and a small chemical depolarization (C) at 1.6 ms before onset of the larger EPSP at 2.4 ms. Inhibitory postsynaptic potentials
(IPSPs) elicited from lHCi appeared as membrane depolarizations following stimulation at 1.5 ⫻ T. Onset of the chemical depolarization (C)
was at 1.8 ms. Vertical arrows illustrate the latencies of the respective synaptic potentials. D: AC recording (n ⫽ 8; 1st trace) and single AC and
DC sweeps (2nd and 3rd traces) showing EPSPs (electrotonic, E, and chemical, C, at 1.5 ⫻ T) and presumed IPSPs (chemical, C) following
rHCe and lHCi nerve stimulation, respectively. IPSPs appeared as membrane depolarizations at a latency of 1.8 ms following lHCi activation
at graded stimulus strengths (1.5–10 ⫻ T). Vertical arrows illustrate the latencies of the respective synaptic potentials.
J Neurophysiol • VOL
86 • SEPTEMBER 2001 •
www.jn.org
1380
W. GRAF, R. SPENCER, H. BAKER, AND R. BAKER
extraocular motoneurons (Fig. 5) and by ultrastructural anatomical features associated with inhibitory synapses (Figs. 7
and 8).
Neurophysiology
FIG. 3. Synaptic potentials in left- and right-side superior rectus motoneurons following horizontal canal nerve stimulation. A: antidromic identification
of somatic recording from a left-side motoneuron (lSR) with IS-SD break
(arrow) and M-spike. AC- and DC-coupled records (top and bottom traces,
respectively). B: AC-coupled records of EPSP and IPSPs. An EPSP was
elicited after ipsilateral (left-side) horizontal canal nerve (lHCe) stimulation
(top trace). The IPSP following contralateral (right-side) horizontal canal
nerve (rHCi) stimulation at 2 ⫻ T appeared as a membrane depolarization at
normal membrane resting potential (middle trace), but injection of depolarizing current (⫹5.0 nA) produced a hyperpolarizing IPSP (bottom trace). C:
antidromic identification of a right-side motoneuron (rSR) and collision block
of the antidromic action potential following direct activation of the neuron
through the microelectrode. Electrotonic potentials were still recorded (slanted
arrow) after the collision block with the respective stimulation artifacts indicated by small upward pointing arrowheads. Top and bottom traces are ACand DC-coupled records, respectively. D: AC-coupled records of EPSP and
IPSPs following double shock stimulation at 2 ⫻ T. EPSPs were elicited from
contralateral (left-side) horizontal canal nerve (lHCe) stimulation (top trace).
IPSPs following ipsilateral (right-side) horizontal canal nerve (rHCi) stimulation (bottom traces) appeared as depolarizations at membrane resting potential
and became inverted during depolarizing current injection (⫹4.0 nA). Amplitude calibrations and time scales are shown in A.
J Neurophysiol • VOL
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
Synaptic potentials recorded in extraocular motoneurons
could be compared with those recorded in goldfish (Graf et al.
1997) as well as interpreted with respect to the vestibulooculomotor pathways shown earlier in flatfish (Graf and Baker
1985b) (see also Figs. 1B and 9). To do so, it was necessary to
recognize the recorded motoneuron pools (Graf and Baker
1985a). Hence, in each experiment the trochlear nuclei (SO
motoneurons) were located first by the antidromic field potential profile initiated from the contralateral orbits (Fig. 4A)
(Baker et al. 1973; Graf et al. 1997). SR motoneurons were
identified in the caudal oculomotor complex following stimulation of the contralateral orbits. IR and IO motoneurons were
recognized by antidromic activation from the ipsilateral orbits,
while abducens (LR) motoneurons were recorded in the posterior brain stem. Motoneuron somata were located adjacent to
the midline (Fig. 5A), and antidromic intracellular records were
characterized by initial segment–soma dendritic (IS-SD) invasion followed by a large afterdepolarization typical of an
intrasomatic penetration (Figs. 2A and 3, A and C).
Since synaptic potentials recorded in all vertical oculomotor
motoneurons exhibited similar profiles, the electrophysiology
is thus illustrated for four key somatic recording sites (2
left-side trochlear motoneurons, Fig. 2, and 1 left-side and 1
right-side SR motoneuron, Fig. 3) and one intradendritic recording from a right-side trochlear motoneuron (Fig. 4). The
cases selected were stable penetrations with resting potentials
between ⫺50 and ⫺60 mV. Membrane depolarization was
elicited from stimulation of the right horizontal canal labyrinthine nerve branch (rHCe) predicted to provide excitatory input
to trochlear motoneurons (Figs. 1B and 9). Stimulation of the
prospective left horizontal canal inhibitory input (lHCi; Fig.
1B, blue pathway) did not produce significant membrane hyperpolarization at this level of resting potential. In fact, in
93/102 cases, the stimulus that should have produced a hyperpolarizing IPSP induced membrane depolarization (Fig. 2, C
and D). In only nine cells could hyperpolarizing IPSPs be
demonstrated without the application of extrinsic current following stimulation of the appropriate canal nerves (Fig. 3, B
and D). Depolarizing current injection between ⫹5 and ⫹15
nA, in the absence of action potentials, showed that stimulation
EXCITATION AND INHIBITION IN FLATFISH VOR
1381
of the vestibular nerve branch expected to provide inhibitory
input, produced IPSPs associated with a membrane conductance. This observation suggested that the IPSP equilibrium
potential might be close to this motoneuron’s resting potential.
In these cases, the synaptic potentials could also be re-reversed
to depolarization by injection of hyperpolarizing current ranging from ⫺4 to ⫺11 nA. Injection of chloride ions also
significantly displaced the IPSP equilibrium potential in some
motoneurons to a level equal to that of the contralateral EPSP
(not illustrated). However, in contrast to goldfish motoneurons,
use of depolarizing current and chloride injection to change the
membrane resting potential and/or IPSP equilibrium potential
did not reveal either robust hyperpolarizing IPSPs or modified
EPSP amplitudes (Graf et al. 1997).
Chemically elicited EPSPs were always preceded by a shortlatency electrotonic potential at a latency of 0.8 – 0.9 ms (Figs.
2, B–D, 3B, and 4C). The ubiquitous presence of electrotonic
coupling indicates distributed gap junction connections
throughout the extraocular motoneuron pools that was corroborated by electron microscopy (Figs. 6– 8). Electrotonic coupling could be demonstrated by collision block of the action
potential elicited from direct activation with that from antiFIG. 4. Dendritic recording from a right-side trochlear motoneuron neuron
(rTro) with a KCl electrode. A: antidromic response with ⫺2.0 nA membrane
hyperpolarization. Arrow indicates the somadendritic all-or-none action potential. Inset illustrates the antidromic field potential elicited from contralateral
orbit stimulation. B: EPSPs recorded after ipsilateral (right-side) horizontal
canal nerve (rHCe) activation at 2 ⫻ T. Top traces are 2 single sweep
AC-coupled records, and the bottom trace, a DC-coupled record of the EPSP
producing action potentials. C: comparison of synaptic potentials elicited from
the excitatory (rHCe) and inhibitory sides (lHCi). The top and the middle
traces are averages of the AC recording (n ⫽ 8). The bottom trace is a single
AC sweep. Arrows indicate onset of electrotonic coupling (E) at 0.9 ms and
chemical depolarization (C) at 1.5 ms for stimulation of the excitatory (rHCe)
horizontal canal. The IPSP following stimulation of the lHCi at 2 ⫻ T (C)
arrived at 1.8 ms, but was reversed due to chloride leakage and shift of the
equilibrium potential. Note the different latencies for the evoked excitatory
electrical potential (E) at 0.5 ms, excitatory chemical potential (C) at 1.5 ms,
and inhibitory chemical potential (C) at 1.8 ms. Amplitude calibrations and
time scale for B are shown in A.
J Neurophysiol • VOL
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
FIG. 5. GABA immunohistochemistry in adult flatfish. Bright field photomicrographs of coronal sections through the oculomotor nucleus (A and B)
and the caudal midbrain (C and D). A and B: low- and high-power magnification showing GABAergic terminals on somata of oculomotor neurons (arrows). C and D: low- and high-power magnification showing reaction product
in MLF axons that course perpendicular to plane of section (arrows). cC,
cerebellar commissure; BC, brachium conjunctivum; LL, lateral lemniscus;
mC, mesencephalic-cerebellar tract; MLF, medial longitudinal fasciculus; OC,
oculomotor nucleus; VC, valvula cerebelli; IIIn, oculomotor nerve.
1382
W. GRAF, R. SPENCER, H. BAKER, AND R. BAKER
the left horizontal canal nerve produced EPSPs in superior
rectus and inferior oblique motoneurons on both sides,
whereas IPSPs were elicited from the right horizontal canal
(cf. Figs. 1B; 3, B and D; and 9). Similarly, the bilateral
trochlear (superior oblique) and inferior rectus motoneurons
received excitatory input from the right horizontal canal
nerve and inhibitory input from the left horizontal canal
nerve (Figs. 1B; 2, B–D; 4, B and C; and 9). Synaptic
responses from the anterior and posterior canals (not illustrated) were qualitatively similar to those observed in mammals and goldfish (Graf et al. 1997).
In all animals, averaging of synaptic potentials following
single shock stimuli was sufficient to determine latency and
polarity (Figs. 2, B–D; 3B; and 4C); however, multiple
shock stimulation was required to elicit synaptic potentials
large enough to activate spikes in the penetrated motoneurons (Figs. 2B and 4B). This phenomenon, in part, may have
been due to either anesthetic (MS222 and Flaxedil) and/or
surgical effects on central vestibular excitability. Alternatively, the small synaptic potentials, especially IPSPs, after
single electrical stimuli could be due to a larger than normal
electrotonic load provided by the dense, bilateral terminal
arborization (Figs. 1B and 9; see also DISCUSSION) (Graf and
Baker 1985b).
FIG. 6. Electron microscopic reconstruction of flatfish 2nd-order vestibular axon terminal identified after lHC activation and
stained by intracellular injection of HRP. A: low-power electron micrograph of an inferior oblique motoneuron (IO MN) on the left
side showing a labeled axosomatic synaptic contact in the bottom right corner below the label of soma. B–D: serial reconstruction
of the labeled terminal shown in A. B and C: chemical (arrows) and electrical (arrowheads) axosomatic synaptic contacts (B). D
and E: axospinodendritic contacts established only chemical contact zones (arrows). Synaptic vesicle morphology and number are
obscured by the dense HRP reaction product inside the axon terminal. Calibrations are 5 ␮m in A, and 0.5 ␮m in B–E. d, dendrite;
sp, spine.
J Neurophysiol • VOL
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
dromic invasion (Fig. 3C). Following electrical activation of
the horizontal canal nerve, small-amplitude chemical depolarizations were recorded with latencies of 1.2–1.6 ms (Figs. 2,
B–D, and 3B). Larger excitatory depolarizations, presumed to
be EPSPs, occurred from 2.2 to 2.6 ms (Fig. 2, B–D). Stimulation of a vestibular nerve branch expected to produce IPSPs,
in general, also produced small membrane depolarizations beginning at a latency of 1.6 –1.8 ms that gradually increased in
amplitude resulting in larger depolarizations at 2.6 ms (Figs. 2,
C and D, and 3B).
Intra-axonally recorded action potentials in the MLF (Fig.
2B) from horizontal canal vestibular neurons were recorded at
a latency of 0.6 – 0.8 ms after stimulation (Fig. 2B). The first
vestibular action potential was coincident with the electrotonic
synaptic potential at 0.8 ms (Fig. 2, B and E), whereas the
second vestibular neuron action potential overlapped with both
an electrotonic and chemical depolarization at 1.6 ms (Fig. 2,
B and C). In these experiments, multiple HC stimuli were
required to produce a large enough EPSP to initiate oculomotor
motoneuronal action potentials (Fig. 2B, lTro Mn).
In all 102 motoneurons, the presence of excitatory and
putative inhibitory synaptic potentials correlated well with
the proposed innervation scheme of horizontal canal secondorder input to oculomotor neurons in flatfish. Activation of
EXCITATION AND INHIBITION IN FLATFISH VOR
1383
GABA immunohistochemistry
Corroborative evidence for vestibular inhibition of oculomotor neurons was provided by the immunohistochemical localization of GABA, which is the inhibitory transmitter in the
vertical VOR of higher vertebrate species (Precht et al. 1973;
Spencer et al. 1989, 1992). GABA immunohistochemistry was
used in goldfish to demonstrate the axonal trajectories of inhibitory vestibular neurons in the MLF, as well as terminations
on extraocular motoneurons in the trochlear and oculomotor
nucleus (Graf et al. 1997). In flatfish, axons of HC inhibitory
neurons were physiologically identified to course in the MLF
(Graf and Baker 1985b), and the GABA reaction product was
also found in the transversely cut axonal profiles in the MLF
(Fig. 5, C and D). Unlike in goldfish, these HC inhibitory axons
did not travel together in bundles, but were dispersed throughout the coronal sections of this pathway (compare Fig. 5D with
Fig. 5C of Graf et al. 1997). In the flatfish oculomotor nucleus,
GABA-positive boutons were observed on the motoneurons in
all subdivisions (Fig. 5, A and B). Terminal arborizations were
distributed largely on the somata and proximal dendritic trees
(Fig. 6, B and E).
Electron microscopy
To provide evidence for the presence of electrotonic and chemical excitation and chemical inhibition, the pattern and mode of
J Neurophysiol • VOL
terminal arborizations were studied in the oculomotor nuclei of
four flatfish after intracellular HRP labeling of physiologically
identified HC activated neurons (Graf and Baker 1985b) (Figs. 6
and 7). Axons of horizontal canal VOR neurons were injected in
the MLF for 3–5 min to minimize masking the cytoplasmic details
of the contacts onto oculomotor neurons. The somata of neurons
in individual subnuclei were identified first in thick plastic sections (e.g., Fig. 7A) that subsequently were thin sectioned for
electron microscopy (Spencer and Baker 1983). Cytoplasmic organelles typical of other vertebrates, e.g., Golgi apparatus and
mitochondria, were observed in the motoneuron somata (Figs. 6A
and 7B). In addition, cisternal arrays of granular endoplasmic
reticulum typical of mammalian oculomotor motoneurons were
well-developed (Spencer and Baker 1983).
A moderate to high density of axosomatic synaptic endings was
found in all oculomotor subdivisions and on trochlear motoneurons (e.g., arrow in Fig. 6B). Electron microscopic reconstruction
of synaptic contacts on oculomotor neurons showed axosomatic
and axodendritic synaptic endings that contained spheroidal synaptic vesicles and established chemical (presumed excitatory)
synaptic contacts characterized by asymmetrical pre/postsynaptic
membrane specializations (Fig. 8, B and D). Many HRP-labeled
HC axosomatic synaptic endings also exhibited gap junction contacts consistent with electrotonic coupling (Figs. 7F and 8, C and
D). Chemical synaptic contacts exhibited an intersynaptic space of
approximately 20 nm, while gap junction contacts displayed a
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
FIG. 7. Electron microscopic reconstruction of flatfish 2nd-order vestibular axon terminal identified after rHC stimulation and
stained by intracellular injection of HRP. A: light-microscopic image of an IO MN contacted by a labeled 2nd-order vestibular axon
collateral (pta) in the right oculomotor nucleus. B: low-power electron micrograph of the IO MN and preterminal axon (pta). C–F:
serial sections of axodendritic and axosomatic chemical synaptic junctions. C: the labeled synaptic ending containing pleiomorphic
vesicles forms a chemical axodendritic contact (arrowhead). Spheroidal synaptic vesicles in the unlabeled terminal (top left)
establish a chemical synapse on the IO MN soma. D: adjacent section showing the postsynaptic profile of the labeled ending
(arrowhead). E: contact of the unlabeled (arrowhead) and labeled ending with the dendrite. F: formation of an axosomatic chemical
contact with the dendrite (arrowhead) and an electrotonic contact (arrow) with the dendrite. Calibrations are 10 ␮m in A, 5 ␮m in
B, and 0.5 ␮m in C–F. d, dendrite; pta, preterminal axon.
1384
W. GRAF, R. SPENCER, H. BAKER, AND R. BAKER
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
J Neurophysiol • VOL
86 • SEPTEMBER 2001 •
www.jn.org
EXCITATION AND INHIBITION IN FLATFISH VOR
1385
DISCUSSION
Summary
In our previous morphological study, second-order VOR
neurons were identified by electrical stimulation of the left-side
(down-side) HC nerve, injected with HRP and their axonal
trajectories and termination patterns reconstructed in the oculomotor complex (Graf and Baker 1983, 1985b). These observations suggested the presence of reciprocal HC species-specific excitatory and inhibitory vestibular projections to vertical
extraocular motoneurons as illustrated in Fig. 9 (red and blue
pathways, respectively). The electrophysiology (Figs. 2– 4) and
anatomy (Figs. 5– 8) presented here strongly support this interpretation and, in particular, also confirm the presence of a
second-order excitatory and inhibitory pathways originating
from the right-side (up-side) HC nerve (Fig. 9, green and
orange pathways, respectively). As a result, either right-side or
left-side horizontal semicircular canal stimulation initiates a
pattern of electrotonic/chemical synaptic depolarization and
inhibition. This is consistent with the idea of equally weighted,
bilaterally symmetric, reciprocal excitatory/inhibitory pathways contacting the four vertical oculomotor motoneuron
pools. Since the ultrastructural and immunochemical synaptic
profiles in flatfish were similar to some aspects of those described in the vertical VOR pathways of goldfish (Graf et al.
1997), we suggest that the development, hindbrain location and
function of VOR neurons may be an unchanging trait in not
only teleost fish, but also possibly throughout vertebrate phylogeny.
Electrophysiology
In antidromically identified oculomotor motoneurons innervating either the bilateral SO/IR or IO/SR muscle pairs, electrical stimulation of either the rHC or lHC vestibular nerve
FIG. 9. Schematic illustration of the species-specific horizontal semicircular canal pathways in the vestibulooculomotor system of adult winter flounders. The diagram shows the approximate topographic location of the extraocular motor nuclei in the horizontal plane with motoneuronal pools
schematically illustrated (modified from Graf and Baker 1985b). Excitatory
(⫹) pathways are shown in red and green, and inhibitory (⫺) pathways in blue
and orange. Activation of the left horizontal canal excites the bilateral superior
rectus (SR) and inferior oblique (IO) motoneurons (red pathway) and inhibits
the bilateral inferior rectus (IR) and superior oblique (SO) motoneurons (blue
pathway). Activation of the right horizontal canal excites the bilateral IR and
SO motoneurons (green pathway) and inhibits the bilateral SR and IO motoneurons (orange pathway). All pathways cross the midline at least once and
project bilaterally to the respective motoneuronal pools. Note exclusive projection to vertical extraocular muscles and reciprocal excitatory/inhibitory
innervation. lHC, left horizontal semicircular canal; rHC, right horizontal
semicircular canal; III, oculomotor nucleus; IV, trochlear nucleus; VI, abducens nucleus; 1°, primary vestibular neuron; 2°, 2nd-order vestibular neuron.
produced short-latency disynaptic electrotonic EPSPs (0.8 – 0.9
ms) followed by disynaptic chemical depolarizations with latencies of 1.2–1.8 ms, respectively. In each motoneuronal
population, and for all experiments, the amplitude of the single
shock EPSPs were small (⬍1 mV), whereas the depolarization
was significantly enhanced in response to multiple stimuli.
While the surgical and anesthetic conditions likely contributed
to this reduced synaptic efficacy, an alternative, more physiological explanation might be related to the observation that
unitary action potentials do not completely invade large terminal arborizations as seen, for example, in tectobulbospinal
neurons (Grantyn and Grantyn 1982) and other neuronal models (e.g., Mackenzie and Murphy 1998; Toth and Crunelli
1998). Intra-axonal records from HC excitatory and inhibitory
vestibular neurons ascending in the MLF always responded
with a single or double action potential of 0.6 – 0.8 ms latency
after single shock stimulation of the appropriate HC nerve (Fig.
2B) (Graf and Baker 1985b). Multiple HC stimuli were re-
FIG. 8. High-magnification electron micrographs of the distribution of chemical and electrical synaptic contacts in the flatfish
oculomotor nucleus. A: axosomatic synaptic endings containing pleiomorphic (presumed inhibitory) synaptic vesicles establishing
single chemical contact zones (arrow). B: spheroidal (presumed excitatory) synaptic vesicles establishing single chemical contact
zones (arrows). C and D: synaptic endings exhibiting gap junction contacts (small arrows) in close spatial proximity to chemical
synaptic contact zones (large arrows). Calibrations: A–C, 0.5 ␮m; D, 0.25 ␮m.
J Neurophysiol • VOL
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
separation of approximately 2 nm between the pre- and postsynaptic membranes (Fig. 8D). Mixed chemical and electrotonic
axosomatic synaptic endings contained fewer spheroidal vesicles
than axosomatic endings establishing only chemical synaptic contacts (Fig. 8B). By contrast, axodendritic endings, including those
on spines, established only chemical contact zones (Fig. 7E).
When individual synaptic contacts were serially reconstructed,
gap junctions were associated with other unlabeled axodendritic
synaptic endings (Fig. 7F). Both types of contacts could be
observed in relation to different (Fig. 7E) or the same (Fig. 8, C
and D) postsynaptic profiles.
Other axodendritic and axosomatic synaptic endings contained pleiomorphic vesicles and established single chemical
synaptic contacts (Figs. 7, C and D, and 8A). Synaptic contact
zones exhibited symmetrical pre-/postsynaptic membrane specializations (Fig. 8A). These presumed inhibitory synaptic
endings were never observed to establish gap junctions and
were located on either somata or proximal dendrites (Fig. 7, B
and D).
1386
W. GRAF, R. SPENCER, H. BAKER, AND R. BAKER
J Neurophysiol • VOL
tential close to membrane resting potential to be as effective as
the much more negative equilibrium potential in mammalian
oculomotor neurons (Llinás and Baker 1972).
GABA immunohistochemistry and electron microscopy
GABAergic terminals were found clustered around cell somata and proximal dendrites in a similar pattern as described
for goldfish oculomotor and abducens nuclei (Graf et al. 1997).
GABA antibodies localized inhibitory neurons in the vestibular
nuclei (not illustrated) and, as in goldfish, axons headed toward
and into the contralateral MLF. Terminal arborizations were
observed to distribute primarily over the somata and proximal
dendritic trees of trochlear (SO) and oculomotor motoneurons.
Thus the presence, spatial distribution, and magnitude of
GABAergic inhibition from presumed GABAergic neurons of
the HC vestibular system in flatfish appears to be similar to the
vertical VOR system in goldfish and mammals (Graf et al.
1997; Spencer and Baker 1992).
In the vestibulooculomotor system of mammals, the inhibitory transmitter related to vertical canal reflexes is GABA; that
related to the horizontal canal reflex is glycine (Spencer et al.
1989). Thus an apparent paradox exists in the adapted vestibuloocular reflex circuitry of adult flatfish: our immunohistochemical data indicate the existence of a robust GABA termination in the oculomotor and trochlear nucleus, while the
inhibitory input to these vertical eye muscle motoneurons is
largely derived from HC second-order vestibular neurons. The
solution to this apparent paradox will bear directly on the
embryonic origin of the adapted VOR circuits, and several
scenarios may be envisioned. If the novel VOR circuitry originates from extant classical and functional horizontal canal
VOR neurons with terminals originally in the abducens nucleus
where glycine exists as an inhibitory transmitter (Spencer et al.
1989), the neurotransmitter content must switch from glycine
to GABA during metamorphosis when new contacts would be
made with vertical extraocular motoneurons. Alternatively,
these inhibitory neurons might arise from neurons formerly
related to the vertical VOR, and secondarily became recruited
by the horizontal canals. A more parsimonious hypothesis,
however, would be that these second-order inhibitory neurons
and their excitatory counterparts originate from entirely different embryonic progenitors. Preliminary results support the
latter conjecture because newly born vestibular neurons appear
at a time when the postmetamorphic connectivity formed (Graf
et al. 1991). Establishing the embryonic origin of these novel
VOR pathways will reveal the relevant hindbrain rhombomeres
and genes responsible for flatfish-specific neuronal adaptation
(Baker 1998).
Ultrastructure
The electron microscopic visualization of flatfish oculomotor and trochlear nuclei revealed axosomatic and axodendritic
synaptic endings containing spheroidal synaptic vesicles that
established chemical, presumed excitatory, synaptic contacts
characterized by asymmetrical pre/postsynaptic membrane
specializations. In addition, many axosomatic synapses also
showed gap junction contacts consistent with electrotonic coupling. The latter were similar in morphological profile to the
“mixed” synapses described in the goldfish abducens nucleus
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
quired, however, to produce additional action potentials that
clearly were more effective in producing transmitter release as
evidenced from the intracellular records in oculomotor neurons. The latter interpretation suggests that orthodromic invasion of the multiple branch points requires a short presynaptic
interspike interval and/or a background of spontaneous activity
likely present under natural conditions (Graf and Baker 1990).
In the appropriate motoneurons, stimulation of the vestibular
nerve branch expected to be inhibitory produced small-amplitude depolarizations beginning at a latency of 1.6 –1.8 ms that
gradually increased in amplitude to peak at 3.0 ms. Direct
evidence for disynaptic inhibitory input to extraocular motoneurons in the form of hyperpolarizing IPSPs was only seen
in 9/102 motoneurons (i.e., 9%). Postsynaptic inhibition in
teleost fish largely appears as a shunting membrane conductance at the soma and proximal dendrites to reduce the input
resistance of the cell and thereby short-circuiting excitation
arriving at, and from, the more distal dendrites (Faber and Korn
1987; Graf et al. 1997). The location of inhibitory (GABAergic) terminals overlying the somata and proximal dendrites
supports a similar interpretation in flatfish (Figs. 5, 7, and 8A).
Manipulation of membrane potential by current injection
through the microelectrode demonstrated the short-latency polarization to be associated with a membrane conductance exhibiting an equilibrium potential slightly more negative than
resting potential (Fig. 3, C and F). However, IPSP reversal by
injected current appeared to be independent of either intrasomatic or intradendritic recording sites and, surprisingly, in
neither case did the injection of chloride ions significantly
displace the equilibrium potential.
Efficacy of the inhibitory synaptic conductance could not be
assessed in flatfish as easily as in either goldfish extraocular
motoneurons (Graf et al. 1997) or Mauthner cells (Faber and
Korn 1975, 1987). The steep voltage dependence of chloride
channels envisioned to increase the magnitude and duration of
the inhibitory response in goldfish motoneurons (Graf et al.
1997) was not seen in flatfish motoneurons. Hyperpolarizing
IPSPs also were not found in the VOR pathways of the puffer
fish, a marine teleost (Korn and Bennett 1972, 1975). Perhaps
sharp electrode intracellular recordings from hindbrain neurons
in teleosts may not reflect sufficiently well the extent of inhibitory conductance because reconstructed cells exhibit an extensive cable-like appearance arguing against a uniform change
in potential throughout all compartments (Graf and Baker
1985a; Highstein et al. 1992; Pastor et al. 1991; Straka and
Baker 1998; Zottoli and Faber 1980). Nonetheless, in view of
the voltage dependence of glycine-activated Cl⫺ channels in
which the kinetics of the synaptic response are enhanced in the
face of excitation (Faber and Korn 1988), activation of the
inhibitory HC canal nerve would be expected to reduce the
firing frequency of oculomotor neurons. Indeed, in some isolated cases, when spontaneous activity was present in the
recorded motoneuron, stimulation of the putative inhibitory
vestibular nerve markedly reduced the firing rate.
The reciprocal semicircular canal inhibitory and excitatory
pathways must play an important role in coordinated eye
movements. If such reciprocity were not present, vestibular
input would result in a counterproductive functional input to
the oculomotor system (Graf and Baker 1990). Our current
interpretation therefore is that the special nature of excitable
membrane properties in fish permits an IPSP equilibrium po-
EXCITATION AND INHIBITION IN FLATFISH VOR
Reciprocal excitatory/inhibitory innervation
The pattern of excitation and inhibition in the vestibulooculomotor system of flatfish is in agreement with previously
obtained morphological results and also supports the structural
requirements to produce compensatory eye movements during
swimming in the adult animals (Graf and Baker 1985b, 1986,
1990). In particular, HC innervation of the SR, SO, IR, and IO
extraocular motoneurons is quite unique, but its blueprint fits
well into the classical reciprocal excitatory/inhibitory innervation scheme of semicircular canal-related VOR function. In eye
movement plants of upright symmetrical vertebrates, e.g.,
mammals and goldfish, excitation arrives from the contralateral, and inhibition from the ipsilateral labyrinth. The flatfish
innervation pattern of the right-sided winter flounder can only
be interpreted in a context-specific manner. In such case, the
bilateral SR and IO motoneurons receive excitatory input from
the downside (left) horizontal canal and reciprocal inhibitory
input from the upside (right) horizontal canal. The motoneurons innervating the antagonists of these muscles, the bilateral
SOs and IRs, would be innervated with opposite-directed signals, i.e., excitation from the (upside) rHC, and inhibition from
the (downside) lHC. Unlike VOR organization in mammals
and goldfish, wherein excitatory and inhibitory VOR neurons
ascend on the contralateral and ipsilateral side, respectively, all
ascending excitatory and inhibitory horizontal canal VOR neurons in flatfish ascend on the contralateral side (Figs. 1 and 9).
Spatial coordination of eye movements
VOR connections are quite remarkable, if not coincident,
when classical co-contraction patterns of yoke eye muscles are
compared between flatfish and other vertebrates (Szentágothai
1943, 1950). The SR muscle of one eye and the IO muscle of
the other eye are defined as yoke muscles as in similar fashion
are the SO and contralateral IR counterpart. Flatfish co-conJ Neurophysiol • VOL
tract either the bilateral SR and IO for backward rotation (i.e.,
extorsion) or the bilateral SO and IR for forward rotation (i.e.,
intorsion) to produce compensatory eye movements during
downward or upward head movements, respectively (Graf and
Baker 1983, 1985b). In light of extraocular muscle kinematics
(Graf and Baker 1985a), this arrangement is biologically meaningful and not different from either other upright fish or lateraleyed vertebrates (Graf and Brunken 1984; Graf and McGurk
1985; Simpson and Graf 1985). However, the similar VOR
behavior in flatfish is brought about by a VOR innervation
scheme that is different from all other vertebrates. In the latter
case both anterior (vertical) canals provide excitatory signals to
the bilateral SR and IO motoneurons during downward head
movements. In flatfish, comparable excitation is transmitted
from the downside horizontal canal to the bilateral SR and IO
motoneurons. A similar scenario holds for posterior (vertical)
canal reflexes in upright vertebrates as well as that elicited by
the upside horizontal canal in flatfish. Inhibitory input from the
same canals in the above-mentioned cases reaches the antagonists of the respective muscles. Thus to implement the evolutionarily adapted state, a previously established innervation
scheme based on retained spatial geometry and orientation of
the eye muscles is employed in the postmetamorphic flatfish
VOR circuits. In upright vertebrates, the sensory information
signaling upward and downward head movement arises
from the four vertically oriented canals, whereas in flatfish, the
two, now vertically oriented, horizontal canals fulfill this role
(Fig. 1B).
Conclusion
The conceptual view of reciprocal excitatory-inhibitory organization of VOR circuits as necessary and essential for
compensatory eye movement production has been upheld by
electrophysiological, immunohistochemical, and ultrastructural evidence obtained from the species-specific expression
found in flatfish. These observations also suggest that secondorder vestibular neurons can use a novel developmental plasticity to achieve new structural and physiological requirements.
Thus the unique VOR organization in flatfish can be used as a
model to study the embryonic, endocrine, environmental, and
genetic mechanisms underlying this pre- to postmetamorphic
transformation.
This work was supported by National Institutes of Health Grants DC-00239,
EY-04613, EY-02007, and NS-13742.
REFERENCES
AKSAY E, BAKER R, SEUNG HS, AND TANK DW. Anatomy and discharge
properties of pre-motor neurons in the goldfish medulla that have eyeposition signals during fixations. J Neurophysiol 84: 1035–1049, 2000.
BAKER R. From genes to behavior in the vestibular system. Otolaryngol Head
Neck Surg 119: 263–275, 1998.
BAKER R, GILLAND E, GREEN A, STRAKA H, AND SUWA H. Rhombomeric
organization of horizontal optokinetic and vestiulo-ocular reflexes in goldfish. Soc Neurosci Abstr 24: 1411, 1998.
BAKER R, PRECHT W, AND BERTHOZ A. Synaptic connections to trochlear
motoneurons as determined by individual vestibular nerve branch stimulation in the cat. Brain Res 64: 402– 406, 1973.
EVINGER C. Extraocular motor nuclei: location, morphology and afferents. In:
Neuroanatomy of the Oculomotor System, edited by Büttner-Ennever JA.
Amsterdam: Elsevier/North Holland, 1988, p. 81–117.
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
(Graf et al. 1997; Sterling 1977). Other axosomatic synaptic
endings contained pleiomorphic synaptic vesicles establishing
chemical synaptic contacts on motoneurons. These presumed
inhibitory synapses were never observed to establish gap junctions. Comparison of flatfish oculomotor and trochlear nuclear
complex to that in goldfish showed that the relative placement
of chemical versus electrotonic contacts of excitatory secondorder vestibular neurons was different (Spencer et al. 1986). In
goldfish, chemical contacts were located at some distance from
gap junctions (Graf et al. 1997), whereas in the flatfish, gap and
chemical junctions were adjacent to each other (Fig. 8, C and
D). In flatfish, the spatial relationship of chemical and electrotonic contacts was closer, and thus more like the arrangement
in the goldfish abducens nucleus (Graf et al. 1997). The functional significance of this difference in chemical versus electrical junction placement is undetermined in both species as is
the significance of electrotonic coupling per se in VOR behavior. A prior interpretation argued that electrotonic transmission
played a more important role in goldfish VOR (Graf et al.
1997) than in synchronizing motoneuronal discharge during
either fast phases or saccades (Korn and Bennett 1971). Nonetheless, electronic coupling apparently is not essential for normal VOR behavior in all vertebrates, because gap junctions are
not found in mammals.
1387
1388
W. GRAF, R. SPENCER, H. BAKER, AND R. BAKER
J Neurophysiol • VOL
KORN H, ODA Y, AND FABER DS. Long-term potentiation of inhibitory circuits
and synapses in the central nervous system. Proc Natl Acad Sci USA 89:
440 – 443, 1992.
KUMAR SS AND FABER DS. Plasticity of first-order sensory synapses: interactions between homosynaptic long-term potentiation and heterosynaptically
evoked dopaminergic potentiation [In Process Citation]. J Neurosci 19:
1620 –1635, 1999.
LLINÁS R AND BAKER R. A chloride dependent inhibitory postsynaptic potential
in cat trochlear motoneurons. J Neurophysiol 35: 364 –385, 1972.
MACKENZIE PJ AND MURPHY TH. High safety factor for action potential
conduction along axons but not dendrites of cultured hippocampal and
cortical neurons. J Neurophysiol 80: 2089 –2101, 1998.
MAGHERINI PC, PRECHT W, AND SCHWINDT PC. Functional organization of the
vestibular input to ocular motoneurons of the frog. Pflugers Arch 349:
149 –158, 1974.
PASTOR AM, TORRES B, DELGADO-GARCIA JM, AND BAKER R. Discharge
characteristics of medial rectus and abducens motoneurons in the goldfish.
J Neurophysiol 66: 2125–2140, 1991.
POLICANSKI D. The asymmetry of flounders. Sci Am 246: 116 –122, 1982.
PRECHT W AND BAKER R. Synaptic organization of the vestibulo-trochlear
pathway. Exp Brain Res 14: 158 –184, 1972.
PRECHT W, BAKER R, AND OKADA Y. Evidence for GABA as the synaptic
transmitter of the inhibitory vestibulo-ocular pathway. Exp Brain Res 14:
158 –184, 1973.
SIMPSON JI AND GRAF W. The selection of reference frames by nature and its
investigators. In: Reviews of Oculomotor Research. I. Adaptive Mechanisms
in Gaze Control, edited by Berthoz A and Melville Jones G. Amsterdam:
Elsevier, 1985.
SPENCER R AND BAKER R. GABA and glycine as inhibitory neurotransmitters
in the vestibulo-ocular reflex. In: Sensing and Controlling Motion: Vestibular and Sensorimotor Function, edited by Cohen B, Tomko DL, and
Guedry F. New York: Ann. NY Acad. Sci., 1992, p. 602– 611.
SPENCER R, WANG SF, AND BAKER R. The pathways and functions of GABA
in the oculomotor system. In: GABA in the Retina and Central Visual
System, edited by Mize RR, Marc R, and Sillito AM. Amsterdam: Elsevier,
1992, p. 307–334.
SPENCER RF AND BAKER R. Morphology and synaptic connections of physiologically-identified second-order vestibular axonal arborizations related to
cat oculomotor and trochlear motoneurons. Soc Neurosci Abstr 9: 1088,
1983.
SPENCER RF, WEISER M, GRAF W, AND BAKER R. Vestibulo-oculomotor
connections in goldfish and winter flounder (Abstract). Biol Bull 171: 498,
1986.
SPENCER RF, WENTHOLD RJ, AND BAKER R. Evidence for glycine as the
putative inhibitory neurotransmitter of vestibular, reticular and prepositus
hypoglossi neurons that project to the cat abducens nucleus. J Neurosci 9:
2718 –2736, 1989.
STERLING P. Anatomy and physiology of goldfish oculomotor system. I.
Structure of abducens nucleus. J Neurophysiol 40: 557–572, 1977.
STRAKA H AND BAKER R. Morphological and physiological characterization of
oculomotor-related precerebellar neurons in the hindbrain of goldfish. Soc
Neurosci Abstr 24: 1411, 1998.
SUWA H, GILLAND E, AND BAKER R. Otolith ocular reflex function of the
tangential nucleus in teleost fish. Ann NY Acad Sci 871: 1–14, 1999.
SZENTÁGOTHAI J. Die zentrale Innervation der Augenbewegungen. Arch Psychiat Nervenkr 116: 721–760, 1943.
SZENTÁGOTHAI J. The elementary vestibuloocular reflex arc. J Neurophysiol
13: 395– 407, 1950.
TOTH TI AND CRUNELLI V. Effects of tapering geometry and inhomogeneous
ion channel distribution in a neuron model. Neuroscience 84: 1223–1232,
1998.
ZOTTOLI SJ AND FABER DS. An identifiable class of statoacoustic interneurons
with bilateral projections in the goldfish medulla. Neuroscience 5: 1287–
1302, 1980.
86 • SEPTEMBER 2001 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on June 18, 2017
FABER DS, FETCHO JR, AND KORN H. Neuronal networks underlying the escape
response in goldfish. General implications for motor control. Ann NY Acad
Sci 563: 11–33, 1989.
FABER DS AND KORN H. Inputs from the posterior lateral line nerves upon the
goldfish Mauthner cell. II. Evidence that the inhibitory components are
mediated by interneurons of the recurrent collateral network. Brain Res 96:
349 –356, 1975.
FABER DS AND KORN H. Voltage-dependence of glycine-activated chloride
channels: a potentiometer for inhibition. J Neurosci 7: 807– 811, 1987.
FABER DS AND KORN H. Unitary conductance changes at teleost Mauthner cell
glycinergic synapses: a voltage-clamp and pharmacologic analysis. J Neurophysiol 60: 1982–1999, 1988.
GRAF W. Motion detection in physical space and its peripheral and central
representation. Ann NY Acad Sci 545: 154 –169, 1988.
GRAF W AND BAKER R. Adaptive changes of the vestibulo-ocular reflex in
flatfish are achieved by reorganization of central nervous pathway. Science
221: 777–779, 1983.
GRAF W AND BAKER R. Synaptic potentials in antidromically identified oculomotoneurons in the winter flounder, Pseudopleuronectes americanus. Biol
Bull 167: 527, 1984.
GRAF W AND BAKER R. The vestibuloocular reflex in the adult flatfish. I.
Oculomotor organization. J Neurophysiol 54: 887– 899, 1985a.
GRAF W AND BAKER R. The vestibuloocular reflex in the adult flatfish. II.
Vestibulo-oculomotor connectivity. J Neurophysiol 54: 900 –916, 1985b.
GRAF W AND BAKER R. Primary and secondary vestibular neuron organization
in the winter flounder, Pseudopleuronectes americanus. Biol Bull 171:
493– 494, 1986.
GRAF W AND BAKER R. Neuronal adaptation accompanying metamorphosis in
flatfish. J Neurobiol 21: 1136 –1152, 1990.
GRAF W, BAKER R, AND BAKER H. Inhibition in the vestibulo-oculomotor
system of the winter flounder, Pseudopleuronectes americanus. Biol Bull
169: 551–552, 1985.
GRAF W AND BRUNKEN WJ. Elasmobranch oculomotor organization: anatomical and theoretical aspects of the phylogenetic development of vestibulooculomotor connectivity. J Comp Neurol 227: 569 –581, 1984.
GRAF W, EVANS B, AND GALLAGER S. Vestibular system development in the
flatfish, Pseudopleuronectes americanus. Soc Neurosci Abstr 17: 128, 1991.
GRAF W AND EZURE K. Morphology of vertical canal related second order
vestibular neurons in the cat. Exp Brain Res 52: 35– 48, 1986.
GRAF W, MCCREA RA, AND BAKER R. Morphology of posterior canal related
secondary vestibular neurons in rabbit and cat. Exp Brain Res 52: 125–138,
1983.
GRAF W AND MCGURK J. Peripheral and central oculomotor organization in the
goldfish, Carassius auratus. J Comp Neurol 227: 391– 401, 1985.
GRAF W, SPENCER RF, BAKER H, AND BAKER R. Excitatory and inhibitory
vestibular pathways to the extraocular motor nuclei in goldfish. J Neurophysiol 77: 2765–2779, 1997.
GRANTYN A AND GRANTYN R. Axonal patterns and sites of termination of cat
superior colliculus neurons projecting in the tecto-bulbo-spinal tract. Exp
Brain Res 46: 243–256, 1982.
HIGHSTEIN SM, CAREY J, BAKER R, AND KITCH R. Anatomical organization of
the brainstem octavolateralis area of the oyster toadfish, Opsanus tau.
J Comp Neurol 319: 1–18, 1992.
KORN H AND BENNETT MVL. Dendritic and somatic impulse initiation in fish
oculomotor neurons during vestibular nystagmus. Brain Res 27: 169 –175,
1971.
KORN H AND BENNETT MVL. Electrotonic coupling between teleost oculomotor neurons: restriction to somatic regions and relation to function of somatic
and dendritic sites of impulse initiation. Brain Res 38: 433– 439, 1972.
KORN H AND BENNETT MVL. Vestibular nystagmus and teleost oculomotor
neurons: functions of electrotonic coupling and dendritic impulse initiation.
J Neurophysiol 38: 430 – 451, 1975.