Rod and cone photoreceptors use specialized biochemistry to generate light responses that differ in their sensitivity and kinetics.
imaged fluorescent indicators of synaptic vesicle cycling and intraterminal Ca2?. We report that the Ca2?sensitivity of release is
Retinal rods and cones are specialized for transducing and trans-
mitting information about different ranges of light intensity.
Rods respond to dim light, generating a slow, prolonged signal,
whereas cones require brighter light, generating a rapid, more
transient signal (Dowling, 1987). The distinct electrophysiologi-
cal responses of rods and cones are attributable to differences in
the biochemical machinery of phototransduction in outer seg-
and Korenbrot, 2004). However, it is unclear whether there are
to the distinct messages transmitted by the two cell types.
tained activation of voltage-gated Ca2?channels, continual
Ca2?influx, and tonic neurotransmitter release. Increasing light
intensity induces a graded hyperpolarization that turns off these
events and suppresses neurotransmitter release. Because neuro-
transmitter is packaged in synaptic vesicles that are released in a
quantal manner, the precision of synaptic encoding is con-
strained by the ambient rate of release, which is maximal in the
the release rate. Rods respond to dimmer light which, because of
its quantal nature, includes fewer gradations of intensity. Hence,
the signal-to-noise ratio of rods is lower than cones. Therefore,
the rod light response conveys less information leading to the
prediction that fewer vesicles are necessary for it to be encoded
We previously measured dark release from photoreceptors in
the cone-only retina of anole lizards (Choi et al., 2005a). We
imaged the loss of the fluorescent synaptic vesicle dye FM1-43
and used electron microscopy (EM) to count synaptic vesicles in
the terminals to estimate the vesicle release rate. Here, we mea-
rod-only retina of geckos. We also estimate release rates from
rods and cones of tiger salamanders. For both interspecies and
intraspecies comparisons, we find that release in darkness is
much slower in rods than cones, in accord with phasic release
rods transmit information at a slower rate.
result from a lower intraterminal Ca2?and/or a less Ca2?-
sensitive release process. The Ca2?dependence of phasic release
from photoreceptors was evaluated by measuring over the first
few seconds the change in membrane capacitance that occurs
after elevating cytoplasmic Ca2?(Rieke and Schwartz, 1994,
1996; Kreft et al., 2003; Thoreson et al., 2004). However, capaci-
tance measurements are not ideal for studying tonic release be-
cause at longer times the capacitance increase from exocytosis is
dependence of tonic release is still unknown. Moreover, the dark
Ca2?concentration is unknown in rod or cone terminals. Here,
we show that the Ca2?dependence of tonic release is similar in
rods and cones, but in darkness, rods have a lower resting Ca2?
underlying their slower tonic release.
Procedures for the care and use of tokay gecko lizards (Gekko gecko),
anole lizards (Anolis segrei), and larval tiger salamanders (Ambystoma
TheJournalofNeuroscience,May9,2007 • 27(19):5033–5042 • 5033
cycle. All procedures were performed in darkness using infrared illumi-
nation and night-vision goggles. To obtain retinas, animals were dark-
adapted overnight (1 h for anoles) and quickly killed by decapitation.
After enucleation, eyes were hemisected and the retinas were removed.
Retinas were bathed in normal saline for gecko [containing (in mM) 160
NaCl, 3.3 KCl, 1.5 CaCl2, 1.5 MgCl2, 10 HEPES, 10 glucose, pH 7.4],
anole [containing (in mM) 149 NaCl, 4 KCl, 1.5 CaCl2, 1.5 MgCl2, 10
HEPES, 10 glucose, pH 7.4], or salamander [containing (in mM) 110
NaCl, 2 KCl, 2 CaCl2, 1 MgCl2, 10 HEPES, 10 glucose, pH 7.4). All dye
solutions were made in normal saline unless otherwise stated. Ca2?-free
saline replaced CaCl2with 1 mM EGTA. High-K?saline contained 50
mM KCl, iso-osmotically replacing NaCl.
FM1-43 dye loading and unloading
Measuring the FM1-43 unloading rate from gecko rod. Procedures for
with 1 mM Advasep-7 (CyDex, Lenexa, KS) for 10 min. FM1-43 unload-
in darkness, bright light (107photons/?m2/s), or high-K?saline. Fluo-
releasable vesicles, and remaining background fluorescence was sub-
tracted. Variability among data are expressed as mean ? SEM.
To test whether the two-photon scans altered the unloading rate of
placed in complete darkness. One piece of retina was taken out at t ? 0
min, t ? 10 min, and t ? 20 min and transferred to Ca2?-free saline to
stop unloading, and then imaged. For each piece of retina, FM1-43 flu-
Measuring FM1-43 loading rate from salamander photoreceptors. The
retina was bathed in 30 ?M FM1-43 for 0–15 min, transferred to Ca2?-
free saline, and dissociated as described previously (Rea et al., 2004). A
suspension of dissociated retinal cells was then transferred onto a cover-
slip for imaging. Total fluorescence from photoreceptor terminals was
Forty optical sections (focused 0.5?m apart) were obtained for each cell
and total FM1-43 fluorescence from each terminal was calculated. Vari-
ability among data are expressed as mean ? SEM.
Fluorescence recovery after photobleaching. FM1-43-loaded retinal
slices were prepared as described previously (Rea et al., 2004). A 4 ?m2
region of the photoreceptor terminal was photobleached by continually
scanning three times at high laser power. Images were acquired every 3 s
before and after photobleaching with a spatial resolution of 34 pixels per
micrometer. Variability among data is expressed as mean ? SEM.
Calcium imaging with fura-2
were sliced in Ca2?-free saline and transferred to Ca2?-free saline con-
taining 5 ?M fura-2 AM (Invitrogen) for 30 min. The double-labeled
scope. Fluorescence was detected with an Imago Sensicam (TILL Pho-
tonics, Planegg, Germany) via a 40 ? 0.7 numerical aperture water-
immersion objective (Olympus, Tokyo, Japan). The wavelength of
excitation light was controlled with a Lambda 10–2 filter wheel (Sutter
Instruments, Novato, CA). FM1-43 was excited with 490 nm light.
Fura-2 was excited with 350 and 380 nm light for ratiometric measure-
tems, Santa Clara, CA). Images of fura-2 and FM1-43 were acquired
every 10 s and binned 4 ? 4. The in vitro fura-2 calibration of the pho-
tometric system was performed with a fura-2 calcium imaging calibra-
tion kit (Invitrogen). Variability among data is expressed as mean ?
Measuring intraterminal Ca2?concentration in darkness. The retina
was bathed in 125 ?M fura-2 AM in darkness for 60 min. The retina was
then washed and mounted on filter paper for imaging. The fura-2 ratio
was obtained using two-photon excitation wavelengths of 700 and 760
nm and a 500–550 bandpass (BP) emission filter. To locate the photore-
with FM4-64 in darkness before fura-2 loading. OPL was first identified
with FM4-64 fluorescence using a two-photon excitation wavelength of
the two-photon imaging system was performed with a fura-2 calcium
imaging calibration kit (Invitrogen).
Gecko and salamander retinas were prepared and fixed for EM as de-
photoreceptor terminals was calculated from ultrathin EM sections (70
nm). The average terminal volume of gecko rods was calculated from
three-dimensional (3D) reconstructions of EM sections using Recon-
Boston, MA). The average terminal volume of salamander rods and
cones was calculated from 3D reconstructions of fluorescence images of
FM1-43-loaded terminals (Imaris software; Bitplane, Saint Paul, MN).
We measured synaptic release from the rod-only gecko retina by
analyzing the loss of FM1-43. We then compared this data with
previous measurements of dye release from anole cones (Choi et
al., 2005a). Gecko retina is advantageous because the rod termi-
nals are large (5–10 ?m diameter), facilitating optical measure-
ments from an individual cell, and they are also homogenous,
terminals in the OPL. Because the tokay gecko is nocturnal, its
visual system is normally operating only in the scotopic range of
ancestral cone photoreceptor and they still have some cone-like
properties, including cone-like photopigments and a large
pedicle-shaped terminal with multiple synaptic ribbons (Kojima
et al., 1992; Ro ¨ll, 2000). However, the intensity–response rela-
tionship and kinetics of the gecko rod light response are very
similar to those of rods from other species (Kleinschmidt and
Dowling, 1975; Rispoli et al., 1993), strongly suggesting that the
gecko rod synapse encodes light information in a manner that is
similar to rod synapses in other species.
Intact retinas were first loaded with FM1-43 in the dark,
which, in the OPL, results in selective labeling of photoreceptor
we used two-photon microscopy, which uses infrared light to
excite fluorescence while minimizing retinal phototransduction.
Because infrared light penetrates deeply into tissue, two-photon
imaging enables visualization of photoreceptor terminals
were imaged every 2–4 min in the dark. We found that FM1-43
was released much more slowly from rods than cones (Fig. 1A).
but with very different time constants (?69 min for rods and 10
min for cones) (Fig. 1B).
The different rates of release suggest that rods and cones have
sible that the infrared laser inadvertently excited photopigments
in the outer segments of rods and cones and triggered photo-
transduction, which would inhibit vesicle release. We had estab-
However, rods are 50–100 times more sensitive to light than
5034 • J.Neurosci.,May9,2007 • 27(19):5033–5042Shengetal.•SynapticVesicleReleaseandCa2?inRodsandCones
cones (Dowling, 1974), perhaps accentuating the problem of
stimulating them inadvertently. To test for this, we measured
FM1-43 release from preloaded pieces of gecko retina that were
To terminate unloading, the tissue was placed in Ca2?-free sa-
line, and only then was the tissue visualized with two-photon
microscopy. Dye loss measured with this protocol (Fig. 1B, R2)
the same as that measured with repeated two-photon scanning
(Fig. 1B, R1) ( p ? 0.001, Student’s t test). Hence, under these
the dark rate of FM1-43 release from rods.
FM1-43 release from gecko rods was completely suppressed
by exposure to bright light (Fig. 1C), similar to results obtained
from anole cones (Choi et al., 2005a). Depolarizing rods with
min; n ? 5 retinas) (Choi et al., 2005a). Approximately 94% of
the dye in rods was released after 20 min in high K?, suggesting
that labeling of unreleasable (i.e., reserve) vesicles and/or other
Release from both rods (Fig. 1D) and cones (Choi et al., 2005a)
was blocked by removing extracellular Ca2?, consistent with
Ca2?-dependent exocytosis. Hence, given sufficient depolariza-
tion, both rods and cones are capable of faster release, indicating
that in the dark their release machinery is not operating at full
is approximately ninefold higher than the
dark rate, and for cones it is 1.6-fold
To quantify the rate of synaptic vesicle re-
and the number of vesicles in the releas-
able (active) vesicle pool. This pool size
total number of vesicles in the terminal
(which we determine by EM). In a phasic
synapse, only a small fraction of synaptic
released (15–25% in hippocampal syn-
apses) (Pyle et al., 2000; Richards et al.,
icles.” However, in a cone photoreceptor,
most of the vesicles are releasable and
there is little or no reserve. We knew that
?86% of vesicles in a cone terminal accu-
mulate FM1-43 within 10 min of loading
(Choi et al. 2005a), closely matching the
percentage of vesicles that remain mobile
in the cytoplasm (87%), as determined by
fluorescence recovery after photobleach-
ing (FRAP) measurements (Rea et al.,
To estimate the fraction of mobile ves-
icles in the gecko rod terminal, we again
prevent release, retinal slices were imaged
to find individual labeled rod terminals. A small area (4 ?m2) in
high power for 1 s (Fig. 2A). This bleached the FM1-43 such that
vesicles within this region were no longer fluorescent. Subse-
quent low-intensity scans showed that the fluorescence within
the bleached region rapidly recovered and that the fluorescence
in the remainder of the terminal decreased as vesicles redistrib-
uted. To assess the fraction of vesicles that are mobile, we mea-
sured the recovery of fluorescence in the bleached region com-
pared with the average fluorescence in the remainder of the
bleach) the FRAP index was 90%. This indicates that 90% of
those vesicles that had accumulated dye during endocytosis re-
main mobile in the cytoplasm, similar to the releaseable fraction
few, if any, vesicles failed to accumulate dye during the initial
loading of gecko rods. Indeed, neither rods nor cones contain
synapsin (Mandell et al., 1990), the key protein that immobilizes
vesicles and therefore holds them in reserve at conventional syn-
apses (Huttner et al., 1983; Ba ¨hler and Greengard, 1987; Benfe-
nati et al., 1989), suggesting the lack of a reserve pool in
We used EM to visualize and quantify synaptic ribbons and
synaptic vesicles in the gecko rod terminal. Photoconversion of a
fixable FM1-43 analog shows that the dye is localized to synaptic
vesicles in gecko rod terminals (data not shown), as demon-
?m. B, Time course of unloading from rod (R) and cone (C) terminals in darkness, measured by repeated two-photon scans.
(closed squares) is for unloading without scans (n ? 3). At the end of experiments, high-K?saline was added to release all
Gecko rods release FM1-43 slower than anole cones in darkness. A, Fluorescence images showing the gradual
Shengetal.•SynapticVesicleReleaseandCa2?inRodsandConesJ.Neurosci.,May9,2007 • 27(19):5033–5042 • 5035
strated previously in anole cone terminals (Choi et al., 2005a).
We calculated the total number of vesicles in gecko rods
(156,000) by measuring the average number of vesicles per unit
volume (1525 ? 69 per ?m3; n ? 5) and the average volume of
terminals (102 ? 15 ?m3; n ? 5). Taking into account the per-
cent that are mobile, gecko rod terminals contain ?140,000 re-
leasable vesicles (Table 1), similar to the 146,000 measured pre-
viously in anole cones (Choi et al., 2005a). By multiplying these
values by the release rate determined from the initial FM1-43
unloading rate, we estimate that, in darkness, gecko rods release
?35 vesicles/s, sevenfold fewer than we reported for anole cones
(246 vesicles/s). Gecko rods have fewer ribbons than anole cones
(14 vs 25), so the release rates per ribbon (2.5/s for rods and 10/s
for cones) differ by approximately fourfold. Nevertheless, in
darkness, the rods clearly release much more slowly than cones.
cones release at similar rates (?300–400 vesicles/s). The release
s). Apparently, both rods and cones are capable of rapid tonic
To confirm the difference between rod and cone release rates in
the same species, we turned to the duplex retina of the tiger
salamander. The terminals of salamander rods and cones are in-
terspersed in the OPL and adhere to one another, making it dif-
ficult to separately measure FM1-43 fluorescence in the two cell
types. We tried to avoid this problem by isolating rods and cones
with intact terminals by enzymatic dissociation (Townes-
Anderson et al., 1985), but their dye release in darkness was very
slow or absent (our unpublished observations), suggesting dis-
ruption of physiological function. Therefore, we chose an alter-
nate approach, measuring the rate of endocytosis (by FM1-43
loading), which at steady-state should equal the rate of exocyto-
sis. The advantage was that loading could be performed on the
intact retina and measurements of dye accumulation could be
done after retinal dissociation.
To determine the rate of dye loading, salamander retina was
incubated in FM1-43 for 0–15 min in the dark. The tissue was
then washed in Ca2?-free saline to prevent additional loading or
unloading. After enzymatic dissociation (Rea et al., 2004), indi-
nal is connected to the inner segment by a long thin axon,
whereas in cones, the terminal emanates directly from the inner
segment (Fig. 3A). Fluorescence images show that FM1-43 accu-
mulates in the rod and cone terminals. The dye subsequently
could be released after depolarization with high-K?saline (Fig.
3B), confirming that, in the terminals, the dye had specifically
accumulated in releaseable synaptic vesicles.
Dye accumulated more rapidly in cone than in rod terminals
(Fig. 3C). Over the initial 1 min of loading, cones incorporated
dye approximately fourfold faster. Even after dye loading had
vesicle captures the same amount of fluorescent dye, rods must
endocytose synaptic vesicles more slowly than cones.
first assumed that prolonged incubation of retina in high-K?
saline resulted in the loading of all active vesicles in rods and
5036 • J.Neurosci.,May9,2007 • 27(19):5033–5042Shengetal.•SynapticVesicleReleaseandCa2?inRodsandCones
more vesicles (Table 1), supporting the notion that all of the
internalized dye remained localized to the terminals and was re-
stricted to synaptic vesicles. Over a brief initial time period (1
min), the fractional accumulation of fluorescence represents the
percentage of vesicles that were endocytosed during this time.
Thus, each minute in darkness, cone and rod terminals endocy-
tose 15.5 and 9.5% of their active vesicle pools.
terminal volume of rods (32.82 ? 6 ?m3; n ? 6) and cones
(82.07 ? 12 ?m3; n ? 6) from z-stacks of dye-loaded terminals,
and the average density of vesicles within rods (2714 ? 300 ves-
icles/?m3; n ? 5) and cones (2628 ? 271 vesicles/?m3; n ? 5),
determined from EM analysis. These calculations indicate that
rods and cones have, respectively, 89,100 and 215,600 vesicles.
This is consistent with the smaller volume of the rod terminal.
Assuming that the fraction of mobile vesicles is similar to that in
vesicle pool in salamander rods and cones is 80,200 and 194,000,
respectively (Table 1).
Multiplying the cycling pool by the rate of endocytosis, we
estimate that, in darkness, the synaptic vesicle uptake rate is
30,070 vesicles/min (501 vesicles/s) in cones and 7619 vesicles/
min (127 vesicles/s) in rods, only 25% of the cone release rate.
Our estimation of rod release rate is lower than previous esti-
mates [?400 vesicles/s (Rieke and Schwartz, 1996)], but these
relied on capacitance measurements that report transient rather
than tonic release. Hence, within a species as well as across spe-
cies, rods exhibit slower tonic release than cones.
To understand what mechanisms cause rods to release vesicles
more slowly than cones, we considered the role of Ca2?. The
release rate in darkness depends both on the relationship be-
tween Ca2?and release and the dark concentration of intra-
To determine which of these parame-
ters differ between rods and cones, we re-
turned to the gecko and anole retinas to
measure the Ca2?dependence of release.
This required simultaneous measurement
indicator fura-2) and vesicle release (with
FM1-43). Our strategy was to monitor
vesicle release while “clamping” intracel-
lular Ca2?at different levels by applying
the Ca2?ionophore ionomycin with dif-
ferent extracellular Ca2?concentrations.
Gecko or anole retinas were loaded
with FM1-43 and then placed in Ca2?-
free saline to prevent unloading. They
which entered cells, where it was con-
verted into fura-2. To separately monitor
Ca2?and vesicle release, we used excita-
tion wavelengths selective for fura-2 (350
and 380 nm) and FM1-43 (490 nm). In
Ca2?-free saline, the fura-2 ratio was low
in both gecko rod and anole cones, indi-
cating low intraterminal Ca2?, but the
FM1-43 signal was high, indicating many
labeled vesicles (Fig. 4A). We then treated
the slices with ionomycin and added ex-
tracellular Ca2?. This increased intracellular Ca2?throughout
facilitated Ca2?entry caused synaptic vesicle release from both
rod and cone terminals.
Figure 4B shows recordings of Ca2?and release, using the
gecko rods as an example. Each measurement was from a differ-
ent retinal slice that was preincubated with the same concentra-
with different concentration of extracellular Ca2?. The fura-2
ratio and the FM1-43 unloading rates were measured after inter-
nal Ca2?had reached steady state (?3 min after perfusion). A
higher fura-2 ratio was always correlated with faster FM1-43
By plotting data from many such experiments we could ob-
serve the relationship between the fura-2 ratio and the FM1-43
release rate (Fig. 4C). For both rods and cones, the release rate
accelerated with increasing intracellular Ca2?. Least-squares fits
to the data show that the slopes of the Ca2?versus release plots
types having similar release machinery. Hence, the difference in
the dark release rate of rods and cones is not caused by a differ-
ence in the Ca2?dependence of exocytosis.
reside. Hence, ionomycin-
We next compared the dark concentration of Ca2?in rod and
by using a nonratiometric Ca2?dye that can be excited with
is excited by 800 nm light). However, the amount of dye loading
varied greatly among terminals in the intact retina, leading to
variability in fluorescence. To circumvent this, we returned to
fura-2, which can be imaged ratiometrically, thereby correcting
for differences in dye loading. We used two-photon excitation of
Shengetal.•SynapticVesicleReleaseandCa2?inRodsandConesJ.Neurosci.,May9,2007 • 27(19):5033–5042 • 5037
fura-2 at 700 and 760 nm instead of the
380 nm). This confined excitation to a
narrow focal plane within the OPL and
topigment in outer segments, which are
?50 ?m deeper into the flat-mount
Treatment of gecko retina with fura-2
terminals than in surrounding structures,
allowing clear identification of the termi-
image showed that rod terminals have
slightly lower Ca2?than the surrounding
tissue. In contrast, anole cones accumu-
late much less dye than the neighboring
Mu ¨ller glia cells (Fig. 5B), which appear
bright when illuminated with a single
wavelength (700 or 760 nm). This made it
difficult to identify cone terminals to se-
lectively measure their fluorescence. To
visualize cone terminals selectively, we
preloaded the retina with FM4-64, a syn-
aptic vesicle dye whose peak excitation
and emission wavelengths do not overlap
with those of fura-2. The ratio image
shows that the cone terminals identified
by FM4-64 have a higher 700/760 ratio
and, therefore, a higher Ca2?concentra-
tion than the surrounding tissue. Analysis
of 41 gecko rods and 24 anole cones
showed that rod terminals have signifi-
cantly lower Ca2?by 50% than cone terminals (157 ? 5 nM in
rods, 317 ? 13 nM in cones; p ? 0.05, Student’s t test). In fact,
there is almost no overlap in the ranges of Ca2?concentrations
observed in the two cell types.
might have triggered phototransduction and therefore lowered
intraterminal Ca2?, especially in rods. Our standard ratio image
tion might cause, we repeatedly line-scanned an individual rod
terminal with 700 or 760 nm light for 2 s. We found that the
a single terminal, whereas during a two-dimensional scan, the
light is spread over ?100 terminals, with each receiving a small
50% difference of intraterminal Ca2?that we observed between
gecko rods and anole cones.
To test whether unhydrolyzed fura-2 AM was sequestered in
we substituted Ca2?with Mn2?(1.5 mM), a quencher of fura-2
fluorescence. We applied high-K?saline to stimulate Mn2?in-
flux into terminals. We found that fura-2 fluorescence was re-
ing nearly complete hydrolysis and consistent with results from
salamander (Szikra and Krizaj, 2006).
Finally, we compared the dark level of Ca2?in salamander
rods and cones. We used FM4-64 to identify photoreceptor ter-
terminals are spherical or ovoid and smaller than the adjacent
cone terminals (Fig. 6A, top). The 700/760 fura-2 ratio was con-
sistently lower in rod terminals, which appear as relatively dark
spots in pseudocolored ratio images (Fig. 6A, bottom). Analysis
of 36 rods and 19 cones showed that the intraterminal Ca2?
concentration is significantly lower by ?29% in rods than in
cones (236 ? 10 nM in rods, 332 ? 13 nM in cones; p ? 0.05,
Student’s t test) in the dark. We tested the extent of inadvertent
phototransduction by continuously line scanning individual ter-
minals with 700 or 760 nm light for 2 s (Fig. 6C). For rods, the
and ?10% over 2 s, but the ratio decreased by the same extent in
cones. Hence, inadvertent phototransduction decreased Ca2?to
Fluorescent indicator dyes enable the functional imaging of vi-
sual system activity in vitro (MacLean et al., 2006) and in situ
(Ohki et al., 2005). However, optical imaging of the retina pre-
sents a problem: light used to excite fluorescent indicators and
duction in rod and cone outer segments, altering activity in their
synapses and in downstream neurons. Two-photon microscopy
helps alleviate this problem. First, it uses infrared light, which is
relatively ineffective in triggering phototransduction. Second,
two-photon excitation is confined to a narrow focal plane (?1
5038 • J.Neurosci.,May9,2007 • 27(19):5033–5042Shengetal.•SynapticVesicleReleaseandCa2?inRodsandCones
Despite these advantages, two-photon scanning can trigger
inadvertent phototransduction, as reported in previous studies
monitoring fast Ca2?transients in response to light stimuli in
ganglion (Denk and Detwiler, 1999) and amacrine cells (Euler et
al., 2002). Therefore, it was important to gauge how much inad-
vertent phototransduction might confound our measurements.
Eliminating exposure to two-photon illumination in rods (Fig.
1B) or changing its duration in cones (Choi et al., 2005a) had no
effect on vesicle release rate, measured with FM1-43. However,
(Figs. 5D, 6C). If we had wanted to monitor Ca2?continuously,
for example, to track transients in response to light flashes, this
would have been a problem. However, because we were mon-
itoring only steady-state Ca2?and tonic vesicle release, the
scan could be applied infrequently, or even only once. By
sacrificing temporal resolution, we obtained reasonably accu-
rate measurements with little interference from inadvertent
ing phototransduction in outer segments, we used the novel ap-
proach of two-photon ratiometric fura-2 imaging. Our results
provide information about average levels of Ca2?in the termi-
nals. However, because the Ca2?channels are localized near the
synaptic ribbons (Nachman-Clewner et al., 1999; Morgans et al.,
2005), a gradient of Ca2?is likely in darkness. Observing this
gradient and accurately measuring sub-
membrane Ca2?at the release sites are
technical challenges, but such measure-
ments could help elucidate the mecha-
nisms underlying dark release.
Our studies involve an interspecies
cones and an intraspecies comparison be-
tween rods and cones of tiger salamander.
Presumably, two-photon FM1-43 and
malian photoreceptors, but the small size
them difficult to unambiguously distin-
we found that isolated mouse and rat ret-
inas begin to deteriorate during the long
time course required for dark loading of
FM1-43 into rods, such that subsequent
release is impaired, even in high K?. Ge-
et al., 1998; Bozza et al., 2004; Li et al.,
study release from mammalian rod and
toreceptor terminals is determined by the
chinery and the dark concentration of
Ca2?. Rods and cones have the same sub-
types of exocytotic and endocytotic pro-
teins, although there are differences in
their relative abundance (Sherry and Hei-
and not cones (Wang et al., 2003), but its function is unclear.
Despite these differences, we find the same Ca2?dependence of
release from rods and cones.
In contrast, we show direct evidence for lower dark intrater-
minal Ca2?in rods. This could reflect differences in Ca2?entry
some differences in L-type Ca2?-channel subtypes (Morgans et
al., 2005), the voltage dependence of activation is similar in dis-
sociated rods and cones (Corey et al., 1984; Maricq and Koren-
brot, 1988; Barnes and Hille, 1989; Rieke and Schwartz, 1994;
Wilkinson and Barnes, 1996; Savchenko et al., 1997; Hosoi et al.,
is modulated by horizontal cell feedback (Verweij et al., 1996;
Cadetti and Thoreson, 2006), so in the intact retina, the voltage-
dependent activation of rod and cone Ca2?current may differ
substantially. Moreover, group III metabotropic glutamate re-
ceptors and exocytosed protons can modulate the Ca2?current
of cones, but not rods (Hosoi et al., 2005). L-type Ca2?channels
are the sole pathway for extracellular Ca2?entry in rods, but
cyclic nucleotide-gated channels also contribute to Ca2?influx
nounced and contributes to tonic release in rods, but not cones
(Krizaj et al., 2003; Suryanarayanan and Slaughter, 2006). Thus,
there are several differences in the Ca2?entry pathways and in
dye (e.g., boxed areas). Ratio image shows slightly lower Ca2?in the terminals than in surrounding tissue. B, Anole retina
FM4-64, imaged with 880 nm light. Merged imaged shows overlap between FM4-64 and fura-2 ratio images. C, Box plot of
Shengetal.•SynapticVesicleReleaseandCa2?inRodsandConesJ.Neurosci.,May9,2007 • 27(19):5033–5042 • 5039
the modulation of voltage-gated Ca2?
channels that could contribute to the
lower dark level of Ca2?in rods.
The main route of Ca2?extrusion
from photoreceptor terminals is through
the plasma membrane Ca2?-ATPase
(PMCA) (Krizaj and Copenhagen, 1998).
The most abundant isoform, PMCA1, is
2002). However, the PMCA2 isoform,
which has a higher Ca2?affinity, is ex-
This could contribute to rods having a
lower intraterminal Ca2?concentration
than cones and, therefore, a slower release
rate in the dark. Indeed, knock-out mice
sitivity to scotopic light (Duncan et al.,
2006), suggesting that low Ca2?in rod
terminals is essential for normal visual
to-noise ratio than bright light (Rose,
1973). We have shown that rods release
synaptic vesicles more slowly than cones
by fourfold to sevenfold. Release of vesi-
cles from rods may occur at regular inter-
vals (Schein and Ahmad, 2005), or by a
stochastic process that follows Poisson
statistics (van Rossum and Smith, 1998;
Berntson et al., 2004). In either case,
less precise in encoding increments of light intensity than the
cone synapse. We reported previously that the dark release rate
from anole cones allows reliable distinction of 10 steps of light
would allow distinction of only four steps. Thus, the slow release
from rods has evolved to match the precision of signal output to
cones types with higher absolute light sensitivity and, therefore,
less precise encoding of light information, to have a lower tonic
vesicle release rate. Selective genetic expression of a fluorescent
whether the correlation between high sensitivity and low release
rate is a general phenomenon.
Our estimate of the dark release rate from gecko rods (35
vesicles/s) is surprisingly slow, even when compared with
salamander rods (127 vesicles/s). Our only assumption in these
estimations was that ?90% of vesicles are available for release.
Both the lack of synapsin (Mandell et al., 1990) and our FRAP
results suggest that few vesicles are held immobilized in a reserve
pool. The presence of any vesicles that are unavailable for release
would result in an overestimate of the release rate; thus, 35 vesi-
rods. If transmission from gecko rods were similar to that from
mouse rods (Field and Rieke, 2002; Sampath and Rieke, 2004),
this slow rate of release would be sufficient to saturate glutamate
receptors on On-bipolar cells. Perhaps the particularly slow re-
reuptake to ensure saturation of metabotropic glutamate
they are capable of tonic release at an approximately ninefold
higher rate, given sufficient depolarization with high K?. Cones
have an excess release capability of ?1.6-fold higher than the
expand the temporal and spatial contrast sensitivity of synaptic
release capability of rods is unknown, but it coincides with other
aspects of rod physiology. A large excess of the phototransduc-
tion machinery, including the number of rhodopsin molecules
and cyclic nucleotide gated channels, is important for maximiz-
ing light sensitivity (Yau and Baylor, 1989). Likewise, the excess
magnitude of the Ca2?conductance, synaptic vesicle number,
and synaptic vesicle release rate may be crucial for maximizing
the speed and reliability of signal transfer at the rod synapse.
Ba ¨hler M, Greengard P (1987) Synapsin I bundles F-actin in a
phosphorylation-dependent manner. Nature 326:704–707.
Barnes S, Hille B (1989) Ionic channels of the inner segment of tiger
salamander cone photoreceptors. J Gen Physiol 94:719–743.
boxed regions (right) show rod terminals (R) brightly labeled with FM4-64, but with lower fura-2 ratios than adjacent cone
extremes are the range of the data. C, Time course of changes in the excitation ratio of fura-2-loaded rod and cone terminals
effects of scanning may be explained by the preferential sensitivity to long-wavelength light of L-type salamander cones over rods,
The dark Ca2?concentration is lower in terminals of salamander rods than in terminals of salamander cones. A,
5040 • J.Neurosci.,May9,2007 • 27(19):5033–5042 Shengetal.•SynapticVesicleReleaseandCa2?inRodsandCones
BenfenatiF,Ba ¨hlerM,JahnR,GreengardP (1989) Interactionsofsynapsin
I with small synaptic vesicles: distinct sites in synapsin I bind to vesicle
phospholipids and vesicle proteins. J Cell Biol 108:1863–1872.
Berntson A, Smith RG, Taylor WR (2004) Transmission of single photon
signals through a binary synapse in the mammalian retina. Vis Neurosci
Bozza T, McGann JP, Mombaerts P, Wachowiak M (2004) In vivo imaging
in the mouse. Neuron 42:9–21.
Cadetti L, Thoreson WB (2006) Feedback effects of horizontal cell mem-
brane potential on cone calcium currents studied with simultaneous re-
cordings. J Neurophysiol 95:1992–1995.
Choi SY, Borghuis BG, Rea R, Levitan ES, Sterling P, Kramer RH (2005a)
Encoding light intensity by the cone photoreceptor synapse. Neuron
Choi SY, Sheng Z, Kramer RH (2005b) Imaging light-modulated release of
synaptic vesicles in the intact retina: retinal physiology at the dawn of the
post-electrode era. Vision Res 45:3487–3495.
Corey DP, Dubinsky JM, Schwartz EA (1984) The calcium current in inner
segments of rods from the salamander (Ambystoma tigrinum) retina.
J Physiol (Lond) 354:557–575.
Cornwall MC, MacNichol Jr EF, Fein A (1984) Absorbance and spectral
sensitivity measurements of rod photoreceptors of the tiger salamander,
Ambystoma tigrinum. Vision Res 24:1651–1659.
DenkW,DetwilerPB (1999) Opticalrecordingoflight-evokedcalciumsig-
nals in the functionally intact retina. Proc Natl Acad Sci USA
Dowling JE (1974) Synaptic arrangements in the vertebrate retina: the pho-
toreceptor synapse. In: Synaptic transmission and neuronal interaction
(Bennett MVL, ed), pp. 87–104. New York: Raven.
Dowling JE (1987) The retina: an approachable part of the brain. Cam-
bridge, MA: Belknap.
Duncan L, Yang H, Doan T, Silverstein RS, Murphy GJ, Nune G, Liu X,
Copenhagen D, Tempel BL, Rieke F, Krizaj D (2006) Scotopic visual
signaling in the mouse retina is modulated by high-affinity plasma mem-
brane calcium extrusion. J Neurosci 26:7201–7211.
EulerT,DetwilerPB,DenkW (2002) Directionallyselectivecalciumsignals
in dendrites of starburst amacrine cells. Nature 418:845–852.
Field GD, Rieke F (2002) Nonlinear signal transfer from mouse rods to
bipolar cells and implications for visual sensitivity. Neuron
Hosoi N, Arai I, Tachibana M (2005) Group III metabotropic glutamate
receptors and exocytosed protons inhibit L-type calcium currents in
cones but not in rods. J Neurosci 25:4062–4072.
Huttner WB, Schiebler W, Greengard P, De Camilli P (1983) Synapsin I
(protein I), a nerve terminal-specific phosphoprotein. III. Its association
with synaptic vesicles studied in a highly purified synaptic vesicle prepa-
ration. J Cell Biol 96:1374–1388.
Kefalov VJ, Fu Y, Marsh-Armstrong N, Yau KW (2003) Role of visual pig-
KW (2005) Breaking the covalent bond—a pigment property that con-
tributes to desensitization in cones. Neuron 46:879–890.
Kleinschmidt J, Dowling JE (1975) Intracellular recordings from gecko
photoreceptors during light and dark adaptation. J Gen Physiol
Kojima D, Okano T, Fukada Y, Shichida Y, Yoshizawa T, Ebrey TG (1992)
Cone visual pigments are present in gecko rod cells. Proc Natl Acad Sci
KreftM,KrizajD,GrilcS,ZorecR, (2003) Propertiesofexocytoticresponse
in vertebrate photoreceptors. J Neurophysiol 90:218–225.
Krizaj D, Copenhagen DR (1998) Compartmentalization of calcium extru-
sion mechanisms in the outer and inner segments of photoreceptors.
Krizaj D, Demarco SJ, Johnson J, Strehler EE, Copenhagen DR (2002) Cell-
specific expression of plasma membrane calcium ATPase isoforms in
retinal neurons. J Comp Neurol 451:1–21.
Krizaj D, Lai FA, Copenhagen DR (2003) Ryanodine stores and calcium
regulation in the inner segments of salamander rods and cones. J Physiol
cone phototransduction. Nature
Laughlin SB (1994) Matching coding, circuits, cells, and molecules to sig-
Li Z, Burrone J, Tyler WJ, Hartman KN, Albeanu DF, Murthy VN (2005)
Synaptic vesicle recycling studied in transgenic mice expressing synapto-
pHluorin. Proc Natl Acad Sci USA 102:6131–6136.
MacLean JN, Fenstermaker V, Watson BO, Yuste R (2006) A visual
thalamocortical slice. Nat Methods 3:129–134.
Mandell JW, Townes-Anderson E, Czernik AJ, Cameron R, Greengard P, De
CamilliP (1990) Synapsinsinthevertebrateretina:absencefromribbon
synapses and heterogeneous distribution among conventional synapses.
Maricq AV, Korenbrot JI (1988) Calcium and calcium-dependent chloride
currents generate action potentials in solitary cone photoreceptors. Neu-
Miesenbo ¨ck G, De Angelis DA, Rothman JE (1998) Visualizing secretion
and synaptic transmission with pH-sensitive green fluorescent proteins.
Morgans CW, Bayley PR, Oesch NW, Ren G, Akileswaran L, Taylor WR
(2005) Photoreceptor calcium channels: insight from night blindness.
Vis Neurosci 22:561–568.
Nachman-Clewner M, St Jules R, Townes-Anderson E (1999) L-type cal-
in plasticity. J Comp Neurol 415:1–16.
Ohki K, Chung S, Ch’ng YH, Kara P, Reid RC (2005) Functional imaging
with cellular resolution reveals precise micro-architecture in visual cor-
tex. Nature 433:597–603.
Perry RJ, McNaughton PA (1991) Response properties of cones from the
retina of the tiger salamander. J Physiol (Lond) 433:561–587.
Pyle JL, Kavalali ET, Piedras-Renteria ES, Tsien RW (2000) Rapid reuse of
readily releasable pool vesicles at hippocampal synapses. Neuron
Rabl K, Cadetti L, Thoreson WB (2005) Kinetics of exocytosis is faster in
cones than in rods. J Neurosci 25:4633–4640.
ReaR,LiJ,DhariaA,LevitanES,SterlingP,KramerRH (2004) Streamlined
synaptic vesicle cycle in cone photoreceptor terminals. Neuron
Rebrik TI, Korenbrot JI (2004) In intact mammalian photoreceptors,
Ca2?-dependent modulation of cGMP-gated ion channels is detectable
in cones but not in rods. J Gen Physiol 123:63–75.
Richards DA, Guatimosim C, Rizzoli SO, Betz WJ (2003) Synaptic vesicle
pools at the frog neuromuscular junction. Neuron 39:529–541.
Rieke F, Schwartz EA (1994) A cGMP-gated current can control exocytosis
at cone synapses. Neuron 13:863–873.
Rieke F, Schwartz EA (1996) Asynchronous transmitter release: control of
exocytosis and endocytosis at the salamander rod synapse. J Physiol
Rispoli G, Sather WA, Detwiler PB (1993) Visual transduction in dialysed
detached rod outer segments from lizard retina. J Physiol (Lond)
Ro ¨ll B (2000) Gecko vision-visual cells, evolution, and ecological con-
straints. J Neurocytol 29:471–484.
Rose A (1973) Vision: human and electronic. New York: Plenum.
Sampath AP, Rieke F (2004) Selective transmission of single photon re-
sponses by saturation at the rod-to-rod bipolar synapse. Neuron
Savchenko A, Barnes S, Kramer RH (1997) Cyclic-nucleotide-gated
channels mediate synaptic feedback by nitric oxide. Nature
Schein S, Ahmad KM (2005) A clockwork hypothesis: synaptic release by
rod photoreceptors must be regular. Biophys J 89:3931–3949.
Sherry DM, Heidelberger R (2005) Distribution of proteins associated with
synaptic vesicle endocytosis in the mouse and goldfish retina. J Comp
Sherry DM, Bui DD, Degrip WJ (1998) Identification and distribution of
Stella Jr SL, Thoreson WB (2000) Differential modulation of rod and cone
calcium currents in tiger salamander retina by D2 dopamine receptors
and cAMP. Eur J Neurosci 12:3537–3548.
Shengetal.•SynapticVesicleReleaseandCa2?inRodsandConesJ.Neurosci.,May9,2007 • 27(19):5033–5042 • 5041
Sterling P (2004) How retinal circuits optimize the transfer of visual infor- Download full-text
mation. In: The visual neurosciences (Chalupa LM, Werner JS, eds).
Cambridge, MA: MIT.
Suryanarayanan A, Slaughter MM (2006) Synaptic transmission mediated by
SzikraT,KrizajD (2006) Thedynamicrangeanddomain-specificsignalsof
intracellular calcium in photoreceptors. Neuroscience 141:143–155.
Tachibanaki T, Tsushima S, Kahamura S (2001) Low amplification and fast
toresponses. Proc Natl Acad Sci USA 98:14044–14049.
ThoresonWB,RablK,Townes-AndersonE,HeidelbergerR (2004) Ahighly
Ca2?-sensitive pool of vesicles contributes to linearity at the rod photo-
receptor ribbon synapse. Neuron 42:595–605.
Townes-Anderson E, MacLeish PR, Raviola E (1985) Rod cells dissociated
peroxidase. J Cell Biol 100:175–188.
van Rossum MC, Smith RG (1998) Noise removal at the rod synapse of
mammalian retina. Vis Neurosci 15:809–821.
Verweij J, Kamermans M, Spekreijse H (1996) Horizontal cells feed back to
cones by shifting the cone calcium-current activation range. Vision Res
WangMM,JanzR,BelizaireR,FrishmanLJ,SherryDM (2003) Differential
distribution and developmental expression of synaptic vesicle protein 2
isoforms in the mouse retina. J Comp Neurol 460:106–122.
Wilkinson MF, Barnes S (1996) The dihydropyridine-sensitive calcium
channel subtype in cone photoreceptors. J Gen Physiol 107:621–630.
Yau KW, Baylor DA (1989) Cyclic GMP-activated conductance of retinal
photoreceptor cells. Annu Rev Neurosci 12:289–327.
5042 • J.Neurosci.,May9,2007 • 27(19):5033–5042Shengetal.•SynapticVesicleReleaseandCa2?inRodsandCones