Bloomfield SA (2009)

Bloomfield SA (2009)
This article was originally published in the Encyclopedia of Neuroscience
published by Elsevier, and the attached copy is provided by Elsevier for the
author's benefit and for the benefit of the author's institution, for noncommercial research and educational use including without limitation use in
instruction at your institution, sending it to specific colleagues who you know,
and providing a copy to your institution’s administrator.
All other uses, reproduction and distribution, including without limitation
commercial reprints, selling or licensing copies or access, or posting on open
internet sites, your personal or institution’s website or repository, are
prohibited. For exceptions, permission may be sought for such use through
Elsevier's permissions site at:
Bloomfield S A (2009) Retinal Amacrine Cells. In: Squire LR (ed.) Encyclopedia
of Neuroscience, volume 8, pp. 171-179. Oxford: Academic Press.
Author's personal copy
Retinal Amacrine Cells 171
Retinal Amacrine Cells
S A Bloomfield, New York University School of
Medicine, New York, NY, USA
ã 2009 Elsevier Ltd. All rights reserved.
The amacrine cells are the major interneurons in the
proximal retina of vertebrates. They form the most
diverse type of neuron in the retina, with 20–30
morphological subpopulations described in mammals, based on differences in somatic and dendritic
architecture (Figure 1). The dendritic branches of
these subtypes stratify within restricted levels of the
inner plexiform layer (IPL), indicating selectivity in
the synaptic connections made with bipolar cell axon
terminals as well as the dendritic processes of neighboring amacrine and ganglion cells. It is believed
that the structural diversity within the amacrine cell
population reflects an equally large range of physiological properties and the corresponding functional
roles played in formulating the retinal output signals
expressed by the postsynaptic ganglion cells.
Amacrine Cell Morphology
The amacrine cells were named by Ramón y Cajal
after a Greek derivation of ‘neurons lacking long
fibers,’ based on the typically circumscribed space
occupied by their dendritic arbors. The morphological classification of amacrine cells has been based on a
number of structural parameters, including (1) size,
position, and shape of the perikaryon; (2) size and
symmetry of dendritic arbors, including the shape
and caliber of branches and density of arborization;
and (3) the stratification pattern of dendrites within
the IPL. Beginning with the Golgi method, a wide
variety of techniques have been employed to elucidate
the different morphological subtypes of amacrine
cell in the retina. It is now clear that 20–30 amacrine cell subtypes exist. A significant number of
them are conserved across mammalian species and
even show correspondence to amacrine cell subtypes
in lower vertebrate retinas. Within a given retina, the
amacrine cell subtypes form regular mosaics, but
these show tremendous variability in intercellular
spacing, indicating differences in their sampling of
visuotopic space. It is interesting that any single subtype of amacrine cell forms only a small percentage
of the total population. While the AII subtype,
which subserves the pathway that carries rod information, constitutes about 13% of all amacrine cells,
no other subtype totals more than 5%. This division
of amacrine cell subtypes suggests that each plays
a separate, yet equally important, role in visual
The somata of amacrine cells lie mainly within the
proximal portion of the inner nuclear layer (INL), but
those of many subtypes are displaced to the ganglion
cell layer (GCL) or lie within the IPL. In some retinas,
displaced amacrine cells even outnumber the ganglion cells in the GCL. On the basis of dendritic
arbor size, amacrine cells can be divided generally
into small-field cells, thought to play a role in local
processing and higher acuity vision, and wide-field or
long-range cells, involved in more-global retinal processing. Some of these long-range amacrine cells have
very e xtensive dendritic arbors that span hundreds of
micrometers laterally across the retina. Further, a
number of amacrine cell subtypes, described recently
in mammals, have very long, narrow, axonlike processes distinct from a conventional dendritic arbor
(Figure 2). These fibers show the ultrastructure of
typical axons in the central nervous system (CNS),
including the expression of high-molecular-weight
neurofilament proteins. In contrast to ganglion cells,
these so-called polyaxonal amacrine cells bear multiple axonlike fibers that emerge from the soma or
dendritic processes and course for several millimeters
through the IPL but do not exit the retina via the optic
The synapses made by amacrine cells are confined
to the IPL. Chemically mediated synapses are made
with ganglion cell dendrites, bipolar cell axonal processes, or dendrites of other amacrine cells. The dendrites of amacrine cells can maintain both presynaptic
and postsynaptic specializations. These synapses can
occur juxtaposed along the same dendritic segment to
form complex circuits. For example, amacrine cell
processes that are postsynaptic to bipolar cells at
ribbon synapses can make nearby reciprocal synapses
back to the bipolar cell axon. This feedback synapse,
which is inhibitory, is thought to modulate the release
of neurotransmitter from the bipolar cell. A second
type of complex amacrine cell circuit is called a serial
synapse; in it, one amacrine cell makes contact with
a second amacrine cell, which, in turn, synapses onto
a third cell, all within a short distance.
It is interesting that the ratio of the number of
synapses that ganglion cells make with amacrine cells
and with bipolar cells varies across species. This ratio
can also vary for different ganglion cell subtypes
within the same retina. It has been suggested that
ganglion cells with simple, concentric receptive fields
receive synaptic inputs mainly from bipolar cells,
whereas those with complex response properties,
Encyclopedia of Neuroscience (2009), vol. 8, pp. 171-179
Author's personal copy
172 Retinal Amacrine Cells
ambient, background light conditions. The result is
that AII cell coupling is constantly changing to ensure
the highest fidelity of responses to maintain the very
sensitive rod signals in the proximal retina (see the
section titled ‘AII amacrine cells’). In support of this
notion, the sensitivity of postsynaptic ganglion cells is
reduced by about one log unit of illumination in the
connexin36 knockout mouse retina, in which the AII
amacrine cells are uncoupled.
Amacrine Cell Neurotransmitters
25 μm
Figure 1 Amacrine cell subtypes in the central primate retina.
Drawings are of Golgi-impregnated cells in the rhesus monkey
retina as they appear in flat mount. The cells designated s.a.1, 2,
and 3 are different subtypes of amacrine cells with unistratified
dendritic arbors in the IPL. The amacrine cells labeled as d.a.
display diffusely stratifying dendritic arbors. Note the overlapping
fields of the different amacrine cell subtypes as well as differences
in their intercellular spacing. From figure 96 in Boycott BB and
Dowling JE (1969) Organization of the primate retina: Light
microscopy. Philosophical Transactions of the Royal Society B:
Biological Sciences 255: 109–184, with permission.
such as direction or orientation selectivity, receive
more amacrine cell synaptic input. This suggests
that amacrine cell circuitry in the IPL underlies the
processing of complex visual responses seen in some
ganglion cells.
Electrical coupling via gap junctions occurs for
almost all the subtypes of amacrine cell in mammalian retinas (Figure 3). This includes homologous
coupling between adjacent amacrine cell neighbors
or heterologous coupling between amacrine cells
and ganglion cells or bipolar cells. These electrical
synapses are thought to play a variety of roles in
information processing. For example, AII amacrine
cells receive inputs from rod bipolar cells and thereby
process information during nighttime vision. The
AII amacrine cells are extensively coupled to one
another via gap junctions composed of connexin36
protein. The coupling between AII cells is believed to
sum synchronous signals and reduce background
noise, thus increasing the fidelity of the rod signals
they carry. Further, the conductance of these gap
junctions is affected by dopamine, which acts as a
light-mediated neuromodulator. Thus, the size of the
network formed by AII cells changes under different
With one exception, all amacrine cells are inhibitory
interneurons and typically use either g-aminobutyric
acid (GABA) or glycine as their neurotransmitter.
The responses of virtually all ganglion cells are
affected by exogenous application of these transmitters, either directly or through the complex circuitry of
the IPL described above. The GABA receptors appear
to be segregated, whereby GABAA and GABAB receptor subtypes subserve feedforward inhibition of amacrine cells onto other amacrine cells and ganglion
cells, and GABAC receptors mediate the feedback
inhibition onto bipolar cell axon terminals. The combination of GABA effects via the different receptors is
believed to influence the sustained or transient characteristics of amacrine and ganglion cell responses. In
general, glycine is the transmitter used by small-field
amacrine cells, whereas GABA is used by medium- and
wide-field cells. Acetylcholine-containing amacrine
cells, also called starburst amacrine cells, consist
of mirror-image pairs with somata in the GCL and
INL and dendrites monostratified in the inner and
outer IPL, respectively. It is interesting that starburst
amacrine cells are found in all vertebrate retinas,
suggesting that they play an important role in visual
processing (see the section titled ‘Starburst amacrine
cells’). The acetylcholine release from starburst cells
occurs by two methods: a small, tonic release that is
independent of calcium and light and a second, larger
acetylcholine release that is light dependent. It is interesting that starburst cells also release GABA, and so this
neuron has mixed excitatory and inhibitory effects.
Dopaminergic amacrine cells form a distinctive
subtype that is also found in all vertebrate retinas.
Dopaminergic amacrine cells in the mammal have
polyaxonal morphology, consisting of a relatively
small dendritic arbor and numerous long axonal processes that extend more than 1 mm. Most processes
stratify in the IPL, where they encircle the AII amacrine
cells. As discussed below, dopaminergic amacrine cells
control the electrical coupling between AII amacrine
cells. The axons of some dopaminergic cells, particularly
in nonprimate species, extend outward past the INL to
terminate in the outer plexiform layer. The dendrites of
Encyclopedia of Neuroscience (2009), vol. 8, pp. 171-179
Author's personal copy
Retinal Amacrine Cells 173
Figure 2 Morphology of a subtype of polyaxonal amacrine cell in the rabbit retina. (a) Photomicrograph showing the extensive tracer
coupling pattern following injection of Neurobiotin into a type I polyaxonal amacrine cell (asterisk). (b) Somata of tracer-coupled type
I polyaxonal amacrine cells and the overlapping thick dendritic processes and thin axonal processes. Arrowheads indicate axonal
processes emerging from proximal dendrites. (c) Camera lucida drawing providing a flat-mount view of a type 1 polyaxonal amacrine
cell. The dendritic arbor is presented in black, and the axonal arbor and somata of tracer-coupled amacrine cells are shown in gray. Scale
bars ¼ 100 mm (a), 25 mm (b), 200mm (c). From figures 1 and 2 in Völgyi B, Xin D, Amarillo Y, and Bloomfield SA (2001) Morphology and
physiology of the polyaxonal amacrine cells in the rabbit retina. Journal of Comparative Neurology 440: 109–125.
dopaminergic amacrine cells receive most if not all the
excitatory synaptic input from bipolar cells, whereas
the release of dopamine occurs at varicosities within
the axonal arbor. Thus, there is a polarity to the structure of dopaminergic amacrine cells in terms of their
input and output synaptic circuits.
Another well-studied subtype is the indoleamineaccumulating or serotonergic amacrine cell. In the
mammal, the serotonergic amacrine cells maintain
very long and thin dendritic processes with numerous
varicosities. Computational models suggest that varicosities act to electrically isolate individual dendritic
segments, thereby allowing for independent processing of synaptic inputs. These cells receive input
mainly from rod bipolar cells, but rather than transmit the signals forward to ganglion cells, the serotonergic amacrine cells make reciprocal synapses with
the same rod bipolar axon terminals. These amacrine
cells are thus thought to modulate the release of
bipolar cell neurotransmitter under scotopic light
conditions. Little is known about the effects of serotonin on ganglion cells, but application of serotonin
antagonists appears to reduce the responses of oncenter ganglion cells and to increase the responses of
off-center cells.
In addition to conventional neurotransmitters,
amacrine cells express a large number of neuropeptides, including substance P, enkephalin, somatostatin, neurotensin, glucagon, vasoactive intestinal
peptide, neuropeptide Y, and cholecystokinin. The
peptidergic amacrine cells have been best studied in
the bird retina, in which they show differences in
dendritic structure, particularly in the IPL strata
they occupy. In the mammal, peptidergic amacrine
cells are mainly subtypes of polyaxonal cells, suggesting that peptides are released over large areas of the
retina and serve a neuromodulatory role. Many of
these peptidergic amacrine cells accumulate or express
more than one peptide and/or express the conventional inhibitory transmitters GABA and glycine.
Individual amacrine cells may thereby have multiple
roles in both local and global visual processing.
There have been relatively few studies of the effects
of peptides on the responses of retinal neurons, but they
appear to have general excitatory or inhibitory actions.
For example, substance P excites retinal ganglion cells
in the fish retina, whereas enkephalin appears to inhibit
their activity in the amphibian. In mammals, ablation
of neuropeptide Y amacrine cells results in a reduction
of the receptive field size of certain ganglion cells that
Encyclopedia of Neuroscience (2009), vol. 8, pp. 171-179
Author's personal copy
174 Retinal Amacrine Cells
Figure 3 Tracer coupling patterns of amacrine cells in the rabbit retina. (a) Injection of Neurobiotin tracer into an alpha ganglion cell
revealed homologous coupling to a ring of nearest-neighbor alpha cells (arrowheads) as well as extensive heterologous coupling to an
extensive array of amacrine cells (arrows). (b) Injection of Neurobiotin into an AII amacrine cell (asterisk) reveals extensive homolgous
coupling to more than 100 neighboring AII cells. (c) Homologous tracer coupling between a subtype of polyaxonal amacrine cell.
(d) Injection of an ON subtype of directional selective ganglion cell (asterisk) reveals that it is extensively coupled to an array of amacrine
cells. These amacrine cells show separate dendritic and axonal systems, identifying them as a subtype of polyaxonal amacrine cell. Scale
bars ¼ 50 mm (a–d). Panel (c) from figure 4 in Völgyi B, Xin D, Amarillo Y, and Bloomfield SA (2001) Morphology and physiology of the
polyaxonal amacrine cells in the rabbit retina. Journal of Comparative Neurology 440: 109–125. Panel (d) from Figure 1 in Ackert JM, Wu
SH, Lee JC, et al. (2006) Light-induced changes in spike synchronization between coupled ON direction selective ganglion cells in the
mammalian retina. Journal of Neuroscience 26: 4206–4215.
respond preferentially to low acuity scenes. This peptide is thus thought to have a role in the circuitry
underlying ganglion cell spatial tuning.
Amacrine Cell Physiology
Although there are 20–30 subtypes of amacrine cell in
the mammalian retina, the physiology of fewer than
ten subtypes has been studied in detail. Nevertheless,
it is clear the amacrine cells show an extraordinary
variety of response properties. In general, amacrine
cells can be divided into transient and sustained categories on the basis of the temporal properties of their
responses. In lower vertebrates, the transient cells
usually show on–off responses, although transient
‘on cells’ and transient ‘off cells’ have been described
in a number of species. Most sustained amacrine cells
show the center-surround antagonistic receptive field
organization displayed by bipolar cells and most ganglion cells. In classic experiments, amacrine cells were
shown to be robustly excited by a windmill stimulus,
indicating their sensitivity to moving objects.
In addition to the relatively simple center-surround
receptive field of most amacrine cells, certain subtypes display complex physiology similar to that of
some ganglion cells. These include amacrine cell subtypes with orientation selectivity derived from the
spatial organization of excitatory and inhibitory
inputs or an asymmetry in their dendritic field architecture (Figure 4). Amacrine cells with a response
preference for the direction of stimulus motion have
been reported. This includes the starburst amacrine
cell, which shows a response preference for stimuli
moving centrifugally away from the soma (see the
section titled ‘Starburst amacrine cells’).
In contrast to other retinal interneurons, many
amacrine cells express voltage-gated sodium channels
and display action potentials (Figure 5). Spiking has
been reported in both transient and sustained amacrine cells and is often seen superimposed atop large
excitatory potentials. In addition to large-amplitude
somatic spikes, some amacrine cells display allor-none sodium-mediated spikes similar to microspikes thought to be generated within the dendritic
Encyclopedia of Neuroscience (2009), vol. 8, pp. 171-179
Author's personal copy
Retinal Amacrine Cells 175
Figure 4 Orientation-sensitive amacrine cells in the rabbit retina. (a) Responses of an orientation-selective amacrine cell to a
rectangular slit of light swept across the retina at four orientations. The cell excited when the stimulus was oriented at 90 , but was
inhibited when the stimulus was rotated to 0 . (b) Flat-mount view of orientation-selective cell injected with horseradish peroxidase. Same
cell whose responses are shown in Panel. (c) Responses of an orientation-biased amacrine cell to a rectangular slit of light swept across
the retina at four orientations. A stimulus oriented at 0 evoked the largest depolarization, whereas a stimulus at the orthogonal orientation
of 90 produces little response. Unlike orientation-selective amacrine cells, orientation-biased amacrine cells showed no inhibitory
responses. (d) Orientation-biased amacrine cell whose responses are shown in panel (c). Note the asymmetry in the dendritic arbor
with the major axis oriented at 0 matching the preferred orientation physiology. Scale bar ¼ 150 (b), 50 mm (c). From figures 1, 5, 8, and 10
in Bloomfield SA (1994) Orientation-sensitive amacrine and ganglion cells in the rabbit retina. Journal of Neurophysiology 71: 1672–1691,
with permission.
membrane (Figure 5(a)). In wide-field amacrine cells,
including the polyaxonal cells, spikes are actively
propagated centrifugally to terminal branches. In
this scheme, somatic spikes are believed to rapidly
invade the entire arbor to produce global signaling
and release of neurotransmitter. In contrast, dendritic
spikes provide only regional activation resulting in
local signaling to postsynaptic cells. In this way, the
multifocal impulse-generating capability of certain
wide-field amacrine cells provides for a complexity
in receptive field properties and integration of synaptic inputs.
The dopaminergic amacrine cells show a pacemaker activity of rhythmic, spontaneous bursts of
action potentials (Figure 5(b)). This pacemaker activity is generated intrinsically by voltage-gated sodium
channels that slowly depolarize the membrane to
threshold levels. Since the polyaxonal dopaminergic
amacrine cells act on distant cellular targets, the pacemaker activity likely ensures the tonic release of
the neuromodulator. In addition to suprathreshold
spiking, many amacrine cells show rhythmic subthreshold oscillations. This idea is supported by the
finding that bipolar cell axon terminals display
calcium-dependent spontaneous membrane oscillations, which may lead to pulsatile transmitter release
and rhythmic activity of postsynaptic amacrine cells.
In contrast, the oscillatory activity of some amacrine
cell subtypes survives cell isolation in culture and thus
must be generated intrinsically. The subthreshold
oscillatory activity of bipolar and amacrine cells can
produce periodic release of neurotransmitter that, in
turn, will produce oscillatory activity in postsynaptic
cells, particularly ganglion cells. As at other CNS loci,
Encyclopedia of Neuroscience (2009), vol. 8, pp. 171-179
Author's personal copy
176 Retinal Amacrine Cells
Examples of Specific Amacrine Cell
Functional Roles
While it is clear that amacrine cells must play a wide
variety of roles in visual processing, only a few of the
many subtypes have been studied in detail. The physiology and morphology of two subtypes, the AII and
starburst amacrine cells, have been extensively studied, and we now have a good understanding of their
specific roles in retinal processing of visual signals.
The results of these studies are summarized below.
0 nA
0.02 nA
AII Amacrine Cells
0.05 nA
10 mV
200 ms
0.15 nA
20 mV
Figure 5 Spontaneous large somatic and small dendritic spikes
displayed by large amacrine cells. (a) Application of extrinsic
current via the recording electrode differentially evokes the
small- and large-amplitude spikes in a wide-field amacrine cell in
the rabbit retina. Relatively low-amplitude current effectively
evokes the large somatic spikes, whereas the smaller dendritic
spikes are evoked only with larger currents, suggesting moredistal initiating sites. Asterisks indicate small spikes. (b) Rhythmic,
spontaneous bursts of action potentials displayed by a dopaminergic polyaxonal amacrine cell in the dark-adapted mouse retina.
Both large and small amplitude are seen, presumably generated
at somatic and dendritic sites, respectively.
subthreshold oscillations distributed among a network of cells can result in light-evoked synchronous
activity as cells will tend to reach spike threshold
together. The spontaneous oscillatory activity of amacrine cells may thereby serve to organize ensembles by
coordinating the activity when activated by appropriate visual stimuli. This synchronous activity may
serve to increase stimulus efficacy, encode additional
information, and/or bind information about local
visual features.
In the mammalian retina, rod and cone photoreceptors synapse onto largely different bipolar cells,
thereby segregating their signals into different vertical
streams. Whereas up to 11 different morphological
types of cone bipolar cells have been reported,
showing both on- and off-center physiology, only a
single type of rod bipolar cell exists. It is interesting
that the axons of rod bipolar cells do not directly
contact ganglion cells but instead contact mainly the
small-field, bistratified AII amacrine cell. In turn, AII
cells form sign-conserving electrical synapses with the
axon terminals of on-center cone bipolar cells and
verting glycinergic chemical synapses with the
axon terminals of off-center cone bipolar cells. In this
way, both on- and off-center scotopic signals utilize
the cone pathways before reaching the ganglion cells
and ultimately higher brain centers.
The gap junctions formed between AII cells and the
on-center cone bipolar cells form nonrectifying electrical synapses across which the direction of signal
flow changes with stimulus intensity. As mentioned
above, rod signals generated under dim, scotopic light
conditions move from the AII cells to the cone bipolar
cells to be distributed to the ganglion cells. In contrast, under bright, photopic conditions, cone signals
move in the opposite direction, from cone bipolar
cells to the AII amacrine cells. The interconnecting
gap junctions thus do double duty as conduits for
both rod and cone signals.
The AII amacrine cells also form gap junctions
between one another, forming an extensive electrical
syncytium. Computational models suggest that this
coupling increases the signal-to-noise ratio of AII cell
responses by summing synchronous responses and
deceasing asynchronous noise. It is interesting that
the conductance of these homologous AII cell–AII
cell junctions is affected by light via modulation of
dopamine release by changes in dark-light adaptation
(Figures 6(a)–6(c)). Under dim, rod-mediated light
conditions, the relationship between coupling and
ambient light intensity has two phases: AII cells are
relatively uncoupled in the dark-adapted retina but
Encyclopedia of Neuroscience (2009), vol. 8, pp. 171-179
Author's personal copy
Retinal Amacrine Cells 177
Receptive field/tracer coupling size (μm)
Receptive field
Tracer coupling
Starburst Amacrine Cells
show a dramatic increase in coupling when dim background stimuli are presented (Figure 6(d)).
What is the function of the light-induced changes in
AII cell–AII cell coupling? One idea is that these
changes reflect the need for AII cells, as vital elements
in the rod pathway, to remain responsive throughout the scotopic-mesopic range. In this scheme, dark
adaptation is analogous to starlight conditions, under
which rods will only sporadically absorb photons
of light. The need, then, is for AII cells to preserve
these isolated signals above the background noise.
Accordingly, the AII cells are relatively uncoupled in
that there are few correlated signals to sum, and so
extensive coupling would serve to dissipate and
thereby attenuate the few isolated responses rather
than enhance them. As dawn approaches, more
photons become available. In turn, AII amacrine
cells show an increase in coupling, which provides
for summation of synchronous activity over a wider
area, thus preserving the fidelity of these rod-driven,
correlated signals at the expense of spatial acuity.
This transition in coupling between dark-adapted
retinas and those illuminated with dim background
light suggests two basic operating states for AII cells
under scotopic/mesopic light conditions: (1) responding to single photon events and (2) summing signals
over a relatively large area to sum synchronized
events above the background noise. Overall, the AII
cell coupling ensures that the most sensitive rod signals are maintained at the ganglion cell level and
transmitted to higher brain centers.
None −6.5 −6.0 −5.5 −5.0 −4.5
Background light log intensity
Figure 6 Light-induced changes in the coupling between AII
amacrine cells. (a) Tracer coupling pattern of AII cells after injection of one AII cell (asterisk) with Neurobiotin in the dark-adapted
rabbit retina. Plane of focus on a group of eight darkly labeled
AII cell somata and an outer ring of 10–15 lightly labeled AII cell
somata are visible. (b) Plane of focus on tracer-coupled somata
of ON cone bipolar cells that lie more distal in the INL. (c) Tracer
coupling pattern of AII cells after injection of one cell with Neurobiotin in retina exposed to dim background light. Size of tracercoupled group has increased dramatically with light exposure.
(d) Scatterplot illustrating the similar modulation of tracer coupling
and receptive field size of AII cells across a range of background
light intensities corresponding to the scotopic and mesopic levels
(rod operating range). Each data point illustrates the average and
standard error of multiple injections. Well dark-adapted retinas are
represented by data point corresponding to ‘none’ background
light intensity. Note that coupling increases dramatically with dim
light exposure and then reaches a plateau level over the operating
range of rods. Scale bars ¼ 25 mm (a–c). (a, b, d) Adapted from
figures 2 and 9 in Bloomfield SA, Xin D, and Osborne T (1997)
Light-induced modulation of coupling between AII amacrine cells
in the rabbit retina. Visual Neuroscience 14: 565–576, with
A unique subtype of ganglion cell found in the retinas
of many species is the direction-selective (DS) unit.
DS cells respond vigorously to stimulus movement in
the preferred direction yet show little or no activity
following movement in the opposite or null direction.
The ON–OFF variety of DS ganglion cell can be
divided into four subtypes, each preferring a direction
of movement roughly corresponding to the attachment points of the extraocular muscles. The mechanism underlying the computation of direction of
motion has been the subject of investigation for
more than 40 years. Pharmacologic studies indicate
that acetylcholine and GABA can modify the selectivity of DS ganglion cells. In particular, application of
GABA antagonists abolishes the direction selectivity
of DS ganglion cells, suggesting that asymmetric
inhibition resulting from null direction stimulus
permission. Reproduced (c) from figure 1 in Xin D and Bloomfield
SA (1999) Comparison of the responses of AII amacrine cells in the
dark- and light-adapted rabbit retina. Visual Neuroscience 16:
653–665, with permission.
Encyclopedia of Neuroscience (2009), vol. 8, pp. 171-179
Author's personal copy
178 Retinal Amacrine Cells
movement plays a crucial role. Taken together, these
data suggest a role for the starburst amacrine cells,
which release both acetylcholine and GABA. As mentioned earlier, starburst amacrine cells are of two
mirror-image subtypes. The starburst-a cell has a
perikaryon in the INL and monostratifies within
sublamina-a of the distal IPL, whereas the starburst-b
cell is displaced to the GCL and monostratifies in
sublamina-b. It is now clear that the dendrites of both
the starburst-a and -b subtypes costratify and are presynaptic to the bistratified processes of ON–OFF DS
ganglion cells.
Recent studies have provided compelling evidence
that starburst cells indeed play a key role in the
generation of direction selectivity. Specific ablation
of starburst cells, either through pharmacological
manipulation or genetic targeting, abolishes the selectivity of DS cells. Electrophysiological studies have
shown that starburst cells provide direct inhibition
to DS ganglion cells, suggesting that they provide the
GABAergic null inhibition crucial for the direction
selectivity of DS cells. Furthermore, the null inhibition
from starburst cells is itself direction selective, being
stronger for stimulus movement in the null direction.
Taken together, these data suggest that starburst
amacrine cells manufacture the direction-selective
responses in the retina. In particular, two intrinsic
properties of starburst amacrine cells are believed
crucial to the generation of DS ganglion cell
responses. First, computational and calcium-imaging
studies have indicated that the dendritic branches of
starburst cells are electrically isolated from one
another. This means that single starburst cell dendrites can perform computations independently
and simultaneously. Second, light stimulation in
the centrifugal direction (away from the soma) produces a greater voltage response than movement in
the centripetal direction. It is unclear what establishes these two properties of starburst cells, but
potassium channels, calcium channels, chloride ion
transporters, and membrane cable properties have
all been implicated.
As schematized in Figure 7, centrifugal movement
of light along one dendritic branch of a starburst
amacrine cell will result in a robust depolarization
and the release of GABA onto postsynaptic DS ganglion cells. This provides the null inhibition to the DS
cells. Stimulus movement in the opposite, or centripetal, direction will not produce a large response in the
starburst cell, and thus no GABAergic inhibition of
the postsynaptic DS ganglion cells will be evoked.
Thus, the preferred and null directions for a DS
Figure 7 Schematic showing how a starburst amacrine cell generates the direction selectivity of the four ON–OFF direction-selective
(DS) ganglion cell subtypes that show different preferred directions: temporal, nasal, superior, and inferior. The dendritic arbor of the
starburst amacrine cell (center) can be divided into four parts that are electrically isolated and thus function autonomously. Stimulus
movement in the centripetal direction evokes a depolarization of the starburst cell dendritic branch and a release of g-aminobutyric acid.
This produces the null inhibition of the postsynaptic DS ganglion cell. The preferred direction of the DS cell is opposite to the null or
centrifugal in the starburst cell dendritic branch. Each starburst cell dendritic branch synapses with a different subtype of ON–OFF DS
ganglion cell and thereby produces its preferred direction responses of upward, downward, leftward, and rightward.
Encyclopedia of Neuroscience (2009), vol. 8, pp. 171-179
Author's personal copy
Retinal Amacrine Cells 179
ganglion cell are established by the direction of stimulus movement along a starburst cell dendritic
branch. Since the dendritic branches of the starburst
cell are electrically isolated, each can provide a different directional preference expressed by the four subtypes of ON–OFF DS ganglion cells. In this scheme,
each subtype of DS ganglion cell must be postsynaptic
to different dendritic branches of a given starburst
amacrine cell. Indeed, recordings from DS ganglion
cells indicate that while starburst cells lying on the
null side of a DS cell provide inhibition, those on the
preferred side do not. Likewise, morphological data
indicate that DS ganglion cells selectively synapse
only with particular starburst amacrine cell processes
within their dendritic field.
Amacrine cells form an extensive, heterogeneous
group of retinal interneurons that are positioned to
modify the output signals carried by the postsynaptic
ganglion cells. Despite the large diversity of amacrine
cell morphology found in the retina, no one subtype
dominates, suggesting each plays a distinct important
role in visual processing. This idea is supported by the
extraordinary variety of physiological response properties found for amacrine cells, despite the fact that
only a few subtypes have been studied so far. These
properties include complex receptive fields, including
orientation and direction selectivity, somatic and
dendritic spiking, pacemaker activity, rhythmic subthreshold oscillations, and electrical isolation of dendritic segments. Further, amacrine cells partake in a
variety of complex circuitry that includes both chemical and electrical synapses. In fact, nearly every neurotransmitter, neuromodulator, and neuropeptide
found in the CNS has been identified in specific subtypes of retinal amacrine cells. Taken together, these
data indicate that the amacrine cells play diverse
yet specific roles in the integration of visual signals
passing through the retina.
See also: Activity in Visual Development; Fovea: Primate;
Retina: An Overview; Retinal Development: An Overview;
Vision: Light and Dark Adaptation.
Further Reading
Ackert JM, Wu SH, Lee JC, et al. (2006) Light-induced changes in
spike synchronization between coupled ON direction selective
ganglion cells in the mammalian retina. Journal of Neuroscience
26: 4206–4215.
Bloomfield SA (1994) Orientation-sensitive amacrine and ganglion
cells in the rabbit retina. Journal of Neurophysiology 71: 1672–
Boycott BB and Dowling JE (1969) Organization of the primate
retina: Light microscopy. Philosophical Transactions of the
Royal Society B: Biological Sciences 255: 109–184.
Bloomfield SA and Dacheux RF (2001) Rod vision: Pathways and
processing in the mammalian retina. Progress in Retinal and
Eye Research 20: 351–384.
Bloomfield SA, Xin D, and Osborne T (1997) Light-induced modulation of coupling between AII amacrine cells in the rabbit
retina. Visual Neuroscience 14: 565–576.
Dacey DM (1989) Axon-bearing amacrine cells of the macaque
monkey retina. Journal Comparative Neurology 284: 275–293.
Deans MR, Völgyi B, Goodenough DA, Bloomfield SA, and Paul DL
(2002) Connexin36 is essential for transmission of rod-mediated
visual signals in the mammalian retina. Neuron 36: 703–712.
Dowling JE (1987) The Retina: An Approachable Part of the Brain.
Cambridge, MA: Harvard University Press.
Euler T, Detwiler PB, and Denk W (2002) Directionally selective
calcium signals in dendrites of starburst amacrine cells. Nature
418: 845–852.
Feigenspan A, Gustincich S, Bean BP, and Raviola E (1998) Spontaneous activity of solitary dopaminergic cells of the retina.
Journal of Neuroscience 18: 6776–6789.
Fried SI, Munch TA, and Werblin FS (2002) Mechanisms and
circuitry underlying directional selectivity in the retina. Nature
420: 411–414.
MacNeil MA, Heussy JK, Dacheux RF, Raviola E, and Masland RH
(1999) The shapes and numbers of amacrine cells: Matching of
photofilled with Golgi-stained cells in the rabbit retina and
comparison with other mammalian species. Journal of Comparative Neurology 413: 305–326.
Mills SL and Massey SC (1995) Differential properties of two gap
junctional pathways made by AII amacrine cells. Nature 377:
Tauchi M and Masland RH (1984) The shape and arrangement of
the cholinergic neurons in the rabbit retina. Proceedings of the
Royal Society London B: Biological Sciences 223: 101–119.
Vaney DI (1990) The mosaic of amacrine cells in the mammalian
retina. Progress in Retinal Research 9: 1–28.
Völgyi B, Xin D, Amarillo Y, and Bloomfield SA (2001) Morphology and physiology of the polyaxonal amacrine cells
in the rabbit retina. Journal of Comparative Neurology 440:
Xin D and Bloomfield SA (1999) Comparison of the responses of
AII amacrine cells in the dark- and light-adapted rabbit retina.
Visual Neuroscience 16: 653–665.
Encyclopedia of Neuroscience (2009), vol. 8, pp. 171-179
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF