J Neurophysiol 106: 895–904, 2011.
First published May 25, 2011; doi:10.1152/jn.01046.2010.
Receptive field center size decreases and firing properties mature in ON
and OFF retinal ganglion cells after eye opening in the mouse
Christopher L. Koehler,1 Nikolay P. Akimov,1 and René C. Rentería1,2
1
Department of Physiology and 2Center for Biomedical Neuroscience, University of Texas Health Science Center, San
Antonio, Texas
Submitted 3 December 2010; accepted in final form 24 May 2011
vision; retina; multielectrode array
visual system development, eye opening is
an important event because it represents both the onset of
patterned vision and a period of development in which activitydependent changes are observed. Immaturities are present at
this time in the various stages of the visual system, including
the visual cortex, subcortical visual areas, and retina. The 2 wk
after eye opening are a critical period for visual system development in the mouse cortex (Hensch 2005; Tagawa et al.
2005). Profound maturation in retinal response properties to
patterns of light during this time could account for at least part
of this visual system maturation, but knowledge of the functional development of the mouse retina is incomplete.
Changes in the areas of retinal ganglion cell (RGC) dendritic
fields are known to occur after eye opening for most RGC
classes. During development, RGCs expand and retract their
dendritic arbors both in the lateral and vertical dimensions of
the inner plexiform layer (IPL). These dendritic changes occur
while the retinal area itself expands as the eye grows. Retinal
growth decreases the amount of visual space seen by any given
area of the retina. For most RGC classes, the dendrites undergo
a period of rapid areal expansion from postnatal day (P)8 until
eye opening around P13, outpacing retinal growth. This is
followed by dendritic retraction until the adult dendritic area is
DURING MAMMALIAN
Address for reprint requests and other correspondence: R. C. Rentería, 7703
Floyd Curl Dr., MC 7756, San Antonio, TX 78229 (e-mail: renteriarc@uthscsa.
edu).
www.jn.org
reached by P30 (Ren et al. 2009). Together, this suggests that
the area of the retina providing input to individual RGCs is
changing during development.
During this time, the inputs to the RGCs undergo large
changes. Expansion and retraction of dendrites both in the
lateral and vertical dimensions of the IPL while the visual
system is functional and in use suggests that synaptic inputs are
being made and broken in the 2 wk after eye opening (Ren et
al. 2009; Xu and Tian 2007). Indeed, spontaneous synaptic
events recorded from mouse RGCs during this period increase
markedly before tapering down to the adult level (Tian and
Copenhagen 2001), while light-evoked excitatory and inhibitory currents decrease (He et al. 2011). At the same time, RGC
intrinsic properties are also maturing, and together these
changes result in more robust light-evoked spiking from RGCs
(He et al. 2011; Qu and Myhr 2008). Thus, changes to the
functional receptive field–the area of the retina that, when
stimulated with either increments or decrements of light relative to background, affects the firing rate of that RGC–may be
occurring during this period.
In the mouse, OFF and ON-OFF RGC receptive field centers
have been shown to increase in the first few days after eye
opening before decreasing, along with ON RGCs, over the next
week (Cantrell et al. 2010). Other studies of the development
of receptive field properties of RGCs in kittens, rabbits, and
turtles have indicated that excitatory centers and inhibitory
surrounds of some RGCs are still immature and larger at eye
opening (Hamasaki and Sutija 1979; Masland 1977; Rusoff
and Dubin 1977; Sernagor et al. 2001; Sernagor and Grzywacz
1995), although a significant percentage of young RGCs had
adult-like responses in both the cat and rabbit (Masland 1977;
Tootle 1993).
Understanding the changing spatial receptive fields and
response properties of RGCs due to these morphological,
synaptic, and intrinsic changes during the 2 wk after eye
opening is important because this time forms a critical period
for development in the higher visual centers, which depend on
RGCs for their visual input. Here, we determined the functional development of receptive field central areas and firing
properties of mouse ON and OFF RGCs after eye opening, 3
days later, and after the fourth postnatal week, when mouse
vision has reached a mature state. We found that average
receptive field center size decreases during development for
both ON and OFF RGCs and that this decrease is accentuated
by eye growth. Contrast sensitivity increases to the mature
level within 3 days of eye opening. Finally, in accordance with
previous studies, we found that RGC latency is immature at
eye opening and progressively matures. The first few days after
0022-3077/11 Copyright © 2011 the American Physiological Society
895
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
Koehler CL, Akimov NP, Rentería RC. Receptive field center
size decreases and firing properties mature in ON and OFF
retinal ganglion cells after eye opening in the mouse. J Neurophysiol 106: 895–904, First published May 25, 2011;
doi:10.1152/jn.01046.2010.—Development of the mammalian visual system is not complete at birth but continues postnatally well
after eye opening. Although numerous studies have revealed changes
in the development of the thalamus and visual cortex during this time,
less is known about the development of response properties of retinal
ganglion cells (RGCs). Here, we mapped functional receptive fields of
mouse RGCs using a Gaussian white noise checkerboard stimulus and
a multielectrode array to record from retinas at eye opening, 3 days
later, and 4 wk after birth, when visual responses are essentially
mature. Over this time, the receptive field center size of ON and OFF
RGC populations decreased. The average receptive field center size of
ON RGCs was larger than that of OFF RGCs at eye opening, but they
decreased to the same size in the adult. Firing properties were also
immature at eye opening. RGCs had longer latencies, lower frequencies of firing, and lower sensitivity than in the adult. Hence, the
dramatic maturation of the visual system during the first weeks of
visual experience includes the retina.
896
RECEPTIVE FIELD DEVELOPMENT OF MOUSE RGCs
eye opening are thus an important time of change in the
response properties of ON and OFF RGCs.
MATERIALS AND METHODS
Animals
C57Bl6 wild-type mice were obtained from the Jackson Laboratory
or from Charles River. These mice were bred up to three generations
in our animal facility. All procedures were approved by the Animal
Care and Use Committee of the University of Texas Health Science
Center and followed guidelines set forth in the Association for
Research in Vision and Ophthalmology Statement for the Use of
Animals in Ophthalmic and Visual Research.
Extracellular Recording of Light-Evoked Responses From Retinal
Neurons Using a Multielectrode Erray
Mapping RGC Receptive Field Center Area and Determining
Intrinsic Firing Properties
Receptive field maps. Receptive fields of RGCs were mapped by
presenting Gaussian white noise checkerboards (Anishchenko et al.
J Neurophysiol • VOL
Statistical Tests
Both the histograms of the distributions and statistical tests of the
variances (Bartlett’s test) of receptive field diameters indicated that
the samples were non-normally distributed (data not shown); we
therefore used the Dunn-Holland-Wolfe, Kolmogorov-Smirnoff nonparametric, and Spearman’s rank correlations tests (IgorPro). Statistical significance was defined as P ⬍ 0.05.
RESULTS
White Noise Stimulus Mapping Reveals the Development of
Spatial Receptive Fields
We mapped these receptive fields for many RGCs simultaneously by stimulating with a Gaussian white noise checkerboard image sequence and recording from retinas mounted in
vitro on a MEA. Retinas were recorded from three age groups:
P13–14, P16 –17, and P30 –39. These ages were chosen to be
when mouse eyes open, when the eyes have been open for
several days but visual system performance is still immature by
many measures, and when both RGC morphology and visual
cortical responses are at adult levels.
106 • AUGUST 2011 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
Action potentials were recorded from mouse retinas as previously
described with minor modifications (Rentería et al. 2006). Briefly,
animals were dark adapted for at least 30 min and were then killed
after brief CO2 inhalation by cervical dislocation. Both eyes were
dissected, and the front elements were removed under dim red
illumination at room temperature in extracellular recording saline
bubbled with 95% O2-5% CO2 mixed gas (carbogen, Praxair). The
neural retina was placed onto a piece of nitrocellulose paper and
positioned ganglion cell side down in a recording chamber. The retina
was typically placed on a multielectrode array (MEA) within a
millimeter of the optic nerve head. Morphological data have shown
that, in the mouse, dendritic field diameter does not correlate with
retinal eccentricity; in other words, the mouse does not have a visual
streak or fovea, like other mammals such as the cat or primate, where
eccentricity is an important determinant of receptive field size (Diao
et al. 2004). A manipulator (Cell Micro Controls, Norfolk, VA) was
used to hold the tissue down with slight pressure. The multielectrode
chambers (MEA1060 system, MultiChannel Systems, Reutlingen,
Germany) consisted of an array of 60 planar electrodes, each 10 ␮m
in diameter and spaced 100 ␮m apart in 8 rows. The electrodes were
opaque, but the connecting wires were made of transparent indium tin
oxide to minimize shadows during visual stimulation from below.
Retinas were perfused at room temperature for 30 min and at 30°C for
another 30 min before recordings began, and the temperature was
maintained at 30°C throughout the recordings. Temperature at the
array was measured during perfusion with a small thermistor (Physiotemp). The perfusion saline solution consisted of the following (in
mM): 124 NaCl, 2.5 KCl, 2 CaCl2, 2 MgCl2, 1.25 NaH2PO4, 26
NaHCO3, and 22.2 glucose. pH was maintained at 7.3–7.4 by bubbling with carbogen. Acquired voltage signals were band-pass filtered
at 0.1–3 kHz and sampled at 50 kHz (MC_Rack, MultiChannel
Systems).
Light stimuli presented from a monitor (Dell Ultrascan P780 CRT,
75-Hz vertical refresh) were imaged onto the retina. The monitor was
calibrated so that linear changes in pixel value yielded linear changes
in luminance with an intensity range of 0.5– 68 cd/m2. Stimulus
images were generated and presented on the monitor by custom
macros written for Matlab (Mathworks, Natick, MA) using Psychophysics Toolbox extensions (Brainard 1997; Pelli 1997); checkerboards were centered over the array. Trigger pulses were sent to the
acquisition computer by recording voltage pulses from a photodiode
that faced a portion of the monitor. Spikes were sorted and assigned
to individual RGCs using software (Offline Sorter, Plexon, Dallas,
TX) as previously described (Rentería et al. 2006).
2010; Cantrell et al. 2010; Chichilnisky 2001; Kerschensteiner et al.
2008). Random 32 ⫻ 32 checkerboard frames with a checker size of
120 ␮m were displayed at a rate of 75 Hz, equivalent to 1 frame every
13.33 ms, to the retinas being recorded on the array. A spatial map of
the receptive field was generated from the spike-triggered average
(STA) image of the frames preceding each spike for a total of 75
images (i.e., for the 1-s period preceding each spike). A two-dimensional Gaussian fit to the peak frame of the STA (i.e., the frame
containing the largest deviation from the mean) was used as a spatial
representation of the receptive field (Cantrell et al. 2010). Receptive
field diameter was chosen as the average diameter of the ellipse
defined by the iso-sensitivity contour of the Gaussian at 1 SD. A
temporal receptive field profile was generated by measuring the
average contrast of the pixels covering the receptive field center at
each frame in time. These pixel locations were determined from the
frame with the maximum contrast deviation of the series of STA
images, where chosen pixels were 5 SD above the background, with
a minimum of three pixels being used. For each image, the average
value of these pixels was plotted versus time to yield the STA time
course for each RGC. This time course can be considered to be a
linear filter that the RGC applies to any 1-s sequence of checkerboard
images to determine how stimulatory that image sequence is for that
cell. Principal component analysis (PCA) of linear filters was performed using Matlab. For both the P30 –39 and P16 –17 groups of
retinas, white noise frames were presented for 1 h; for the P13–14
group of retinas, the white noise stimulus was presented for 3 h
because of their lower stimulus-driven excitability. Our stimulus
checker size was chosen to accommodate this relative insensitivity of
young RGCs. The numbers of RGCs recorded and analyzed in these
experiments were (ON and OFF cells) n ⫽ 277 and 40 cells from 8
retinas at P13–14, n ⫽ 497 and 149 cells from 19 retinas at P16 –17,
and n ⫽ 291 and 326 cells from 17 retinas at P30 –39.
Intrinsic firing properties. Latency was measured as the time of
zero-crossing preceding the major contrast peak of the linear filter.
The maximum contrast of the STA was determined from the difference between the maximum and minimum contrasts of the linear filter.
ON and OFF RGCs were defined based on the contrast between the
background and the major STA peak, the peak nearest in time to the
spike, with a light increment indicating an ON cell and a decrement
indicating an OFF cell. Analyses were performed using built-in
functions and custom macros in Matlab and IgorPro (WaveMetrics,
Lake Oswego, OR).
RECEPTIVE FIELD DEVELOPMENT OF MOUSE RGCs
897
Receptive Field Center Sizes Decrease After Eye Opening
For each ON and OFF RGC, the STA image with the
greatest variation in pixel intensities was fit with a twodimensional Gaussian function, and the average diameter of
the iso-sensitivity contour of that Gaussian at 1 SD was used as
a measure of the receptive field center size. Figure 1C shows
maps from the peak of the STA for ON and OFF cells near the
average at each of the three ages. The intensities of the pixels
were normalized to the peak of the image with the highest
contrast to highlight the difference in contrast of the STAs
across the different ages. These same maps are shown in Fig.
1D, with pixel intensities normalized to the peak intensity of
each image individually. This better illustrates the center sizes.
The average diameters of the receptive field centers for ON and
OFF RGCs (Fig. 1E, expressed as distance on the retinal
surface) decreased with age (means ⫾ SE): 194 ⫾ 2.0 and 179 ⫾
8.0 ␮m (P13–14), 171 ⫾ 1.6 and 172 ⫾ 4.0 ␮m (P16 –17), and
154 ⫾ 2.0 and 154 ⫾ 1.8 ␮m (P30 –39). Among the ON RGCs,
these values were significantly different at all ages. Moreover,
a significant amount of the total decrease in diameter (⬃60%)
to the adult value occurred within 3 days of eye opening.
Among the OFF RGCs, the decrease in diameter was not
significant between P13–14 and P16 –17 but was significant
between these two ages and P30 –39. Our results indicate that
maturation includes a decrease in the absolute receptive field
center size of the ON and OFF RGC populations, suggesting a
majority of individual RGCs follow the same pattern during
their development (see DISCUSSION).
J Neurophysiol • VOL
Fig. 1. Receptive field (RF) maps and linear filters of retinal ganglion cells (RGCs)
from retinas of mice at different ages revealed RF spatial and temporal development.
A, top: examples of a white noise image sequence, with time increasing left to right.
Middle and bottom, spike-triggered averages (STA) from an ON RGC (middle) and an
OFF RGC (bottom), each from an adult RGC. The image intensities are normalized to
the peak value to better show the relative changes; this accounts for the apparent
difference in the background intensities, which were actually very similar. a–c indicate
the images used for the labeled points of the linear filters shown in B. Checker size was
60 ␮m. B: linear filters for an ON RGC (left) and an OFF RGC (right). (Maps from
the same RGCs are shown in A.) a–c indicate data points generated from the similarly
labeled images in A. Point c was used to define peak STA contrast. C: STA images
from Gaussian white noise checkerboard stimuli of retinas from mice at postnatal days
(P)13–14 (top), P16–17 (middle), and P30–39 (bottom) revealing RGC RFs. The
intensity of the gray background lies midway between white and black. Checker size
is 120 ␮m. The largest intensity increment or decrement for all the images was found
and used to scale all the images similarly, with that value above and below background.
D: data are the same as in C, but each image was scaled to its own highest increment
or decrement of pixel intensity from mean gray. E: average RF diameters, calculated
as the average diameter (in ␮m) at 1 SD from the peak of the two-dimensional
Gaussian fit, of ON RGCs (light bars) and OFF RGCs (dark bars) for each age. Cells
were defined as ON or OFF based on the value of the peak of the STA (point
c in B), which was always the peak nearest in time to each spike. Error bars
indicate SEs. *P ⬍ 0.05.
106 • AUGUST 2011 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
Sequences of Gaussian white-noise checkerboard images
drove the spiking of RGCs. Figure 1A, top, shows an example
image sequence. The most effective stimulus was determined
by averaging the 1-s sequence of images preceding each spike
to yield what is known as the STA (see MATERIALS AND METHODS), which is a spatiotemporal map of the receptive field.
Figure 1A, middle and bottom, shows these maps for an ON
RGC and an OFF RGC. The average contrast of the pixels in
the central receptive field at each frame in time defines a curve
that describes the temporal response profile of the excitatory
center (Fig. 1B). The time course of the STA equates to a linear
filter that the RGC applies (by convolution) to any sequence of
white noise images to generate a signal that determines
whether or not the cell will respond to that particular stimulus
sequence.
ON and OFF RGCs are classified by their STA’s positive or
negative contrast peak that falls closest in time to the spike.
Although the average number of recorded and analyzable cells
per retina did not vary much across the ages, the balance of ON
and OFF RGCs recorded did differ. The percentage of OFF
RGCs recorded increased with age. The percentage of ON and
OFF RGCs recorded at each age were 87% and 13% at
P13–14, 77% and 23% at P16 –17, and 47% and 53% at
P30 –39. Since the mapping method averages image sequences
that cause spiking, ON-OFF RGCs, cells that receive input
from both ON and OFF bipolar cells, are not represented unless
one response or the other dominates. For ON-OFF RGCs, the
image sequences that drive spiking tend to 1) cancel each other
so the average STA contrast approaches the background level
or 2) yield linear filters with tri- or multiphasic shapes, which
we did not find in our data.
898
RECEPTIVE FIELD DEVELOPMENT OF MOUSE RGCs
Eyeball Growth and Interstitial Retinal Expansion
Accentuate the Decrease in Receptive Field Center Size
J Neurophysiol • VOL
Fig. 2. RF diameters of mouse RGC populations decreased during development after eye opening. A and B: histograms of RF diameters converted to
degrees of visual angle (see text) for ON RGCs (left, light gray) and OFF
RGCs (right, dark gray) at P13–14 (top), P16 –17 (middle), and P30 –39
(bottom). Bin amplitudes are expressed as the fraction of the total RGCs of that
type and that age recorded. C: average RF diameters for the same cells as in
A and B for ON RGCs (light gray bars) and OFF RGCs (dark gray bars) for the
indicated ages. Error bars indicate SEs. *P ⬍ 0.05.
populations and found that they were very similar to the
average values. The median receptive field diameters for ON
and OFF RGCs, respectively, were 8.74 and 7.96° of visual
angle (P13–14), 7.14 and 7.08° of visual angle (P16 –17), and
6.02 and 6.10° of visual angle (P30 –39).
Functionally Classifying RGCs
The heterogeneity of receptive field sizes in the histograms
shown in Fig. 2 is expected because RGC classes with varying
dendritic extents were most likely recorded. We sought to
classify RGCs by functional criteria based on features of their
linear filters and to correlate these features to the receptive field
106 • AUGUST 2011 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
A given area or distance of the receptive field as measured
in vitro on the retina will collect light from different amounts
of the visual field in vivo depending on the total retinal area.
The retina expands in area during development as the eyeball
grows. Therefore, we converted the measured receptive field
diameters to the amount of visual field they would sample in
vivo at each age. The length of an image formed on the retina
in vivo can be converted to visual angle by generating a
conversion factor of micrometer per degreee. To do so, we
approximated the retina as a hemisphere with a radius equal to
the posterior nodal distance (PND). The distance (in ␮m) along
the sphere of a single degree of visual angle is then ␲(PND)/
180°, which can be used as a conversion factor. Remtulla and
Hallett (1985) found a PND for the adult mouse of 1.756 mm
at an eye axial length of 3.379 mm. This value gives a
conversion factor of 31 ␮m/°, which is what they reported. We
estimated axial lengths at different ages from published eye
diameters for C57Bl6 mice (in mm) using 2.50 for P13, 2.66
for ⬃P16, and 2.88 for ⬃P35 (from Fig. 1 of Ren et al. 2009).
We used a ratio of axial lengths to scale the adult mouse PND
and, using the equation above, calculated the distance on the
retina representing a single degree of visual angle for the
different age groups. These conversion factors were 22.67
␮m/° at P13–14, 24.12 ␮m/° at P16 –17, and 26.12 ␮m/° at
P30 –39.
We used these values to convert receptive field diameters
from micrometer into degrees of visual angle. Histograms of
these values across the ages revealed a decrease in overall
receptive field diameter with age as well as a significant
difference in receptive field size between ON and OFF RGCs
(Fig. 2, A and B, for ON and OFF RGCs, respectively). The
average receptive field diameters for ON and OFF RGCs,
respectively (Fig. 2C), were (means ⫾ SE) 8.56 ⫾ 0.10 and
7.88 ⫾ 0.36° of visual angle (P13–14), 7.08 ⫾ 0.08 and 7.14 ⫾
0.16° of visual angle (P16 –17), and 5.90 ⫾ 0.06 and 5.88 ⫾
0.08° of visual angle (P30 –39). Thus, ON RGC receptive field
center diameters decreased on average by 2.66° of visual angle
(31%), whereas OFF RGC centers decreased on average by
2.00° of visual angle (25%) in the 2 wk after eye opening. At
P16 –17, over half of this decrease had already occurred in the
ON population, whereas only one-third of the decrease occurred in the OFF population. Moreover, at eye opening, the
average OFF size was 9% smaller than ON size, whereas at
P16 –17 and P30 –39, ON and OFF average sizes did not
statistically differ.
The histograms also revealed that the diameters exhibited a
wide range of values at each age. For both ON and OFF cells
at P13–14 and P16 –17, a few RGCs with large receptive fields
existed that either were no longer present in the P30 –39 data
(e.g., were no longer sensitive to the white noise stimulus) or
had undergone a large decrease in receptive field diameter. A
group of ON and OFF RGCs with relatively small receptive
fields was also apparent at each age. This group also showed a
decrease in receptive field diameter with age, but the timing of
this decrease differed between ON and OFF cells. For ON
cells, the decrease occurred after P16, whereas for OFF cells
the decrease appeared complete by P16. Because the distributions exhibited these tails, we examined the medians of the
RECEPTIVE FIELD DEVELOPMENT OF MOUSE RGCs
Light Response Properties Develop After Eye Opening
The receptive field center defines the area of the visual field
to which an RGC is responsive. The characteristics of an
RGC’s spike train generated in response to center stimulation
are defined by a mix of parameters both intrinsic and extrinsic
to the RGC, including resting potential, threshold, spiking
patterns, peak spike rates, and inhibition. Each of these parameters undergoes maturation during development (Qu and Myhr
2008; Tian and Copenhagen 2001). To determine if the firing
properties revealed by RGCs during white noise checkerboard
stimulation were immature at eye opening, we further examined their white noise spike trains and linear filters.
Latency decreases after eye opening. The linear filter defines
the stimulus time course that best drives RGC spiking. We
J Neurophysiol • VOL
Fig. 3. RF diameter was not correlated with RGC functional type at P13–14 or
P16 –17 and weakly negatively correlated at P30 –39. RGCs were segregated
based on linear filter characteristics using the time of the major peak for ON
cells (point c in the ON cell linear filter shown in Fig. 1B) and the amplitude
of the light increment that precedes the light decrement in the normalized STA
for OFF cells (point b in the OFF cell linear filter shown in Fig. 1B). Left: ON
RGCs. Right: OFF RGCs. Top, data from P13–14 mice; middle, data from
P16 –17 mice; bottom, data from P30 –39 mice. Each open circle represents a
single recorded RGC; the lines show linear correlation best fits for each data
set. The correlations at P13–14 and P16 –17 were not statistically significant,
whereas the weak negative correlations at P30 –39 were (P ⬍ 0.05). Note that
the axis scales are different to more clearly show the scatter within each group.
determined latency to spiking as the zero-crossing time just
before the linear filter reached peak STA contrast, a time of
maximum response in the linear firing model (Anishchenko et
al. 2010). Latencies were longest at eye opening, became
significantly shorter by P16 –17, and became shorter still by
P30 –39 (Fig. 4). In OFF RGCs, much of this decrease in
latency occurred by P16 –17. Whereas the distribution shape of
latencies for ON cells remained constant from P13–14 to
P30 –39 (Fig. 4A), the distribution for OFF cells showed a
wider range at young ages that progressively narrowed (Fig.
4B). The average latencies for ON and OFF RGC responses,
respectively, were (means ⫾ SE) 322 ⫾ 3.1 and 448 ⫾ 10 ms
(P13–14), 272 ⫾ 1.7 and 313 ⫾ 5.0 ms (P16 –17), and 259 ⫾
2.4 and 265 ⫾ 2.9 ms (P30 –39) (Fig. 4C). Thus, our data
shows that latency quickly shortens and matures in both ON
and OFF RGCs after the eyes open.
STA contrast increases after eye opening. The amplitude of
the linear filter (in units of percent contrast) is a measure of the
excitability of the neuron, of its gain, and of the strength of the
106 • AUGUST 2011 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
center diameter. An RGC’s linear filter predicts which features
of the stimulus will cause the cell to fire. Classification of these
filters has been used to segregate RGCs into functional types.
Kerschensteiner et al. (2008) reported two ON types and two
OFF types of mouse RGCs that accounted for 80% of the
RGCs they recorded from C57Bl6 mouse retinas on a MEA. In
the rat, these types accounted for 30 – 60% of the RGCs
recorded using a MEA (Anishchenko et al. 2010). In both
studies, PCA of linear filter parameters was used to identify
clusters of filters, which represented the RGC functional types.
In the published examples, the two ON cell types were segregated primarily by their time of positive contrast peak preceding the spike [i.e., a measure of the latency to spiking, position
c in the ON filter of Fig. 1B (see Fig. 2 of Kerschensteiner et
al. 2008)]. The two OFF cell types were distinguished primarily by their maximum amplitude of STA contrast (i.e., position
b in the OFF filter in Fig. 1B).
PCA of the filter features did not clearly segregate ON or
OFF filters into clusters (data not shown). Instead, we determined the values of these linear filter parameters for all our
recorded RGCs, seeking to follow these functional types
through development. We examined RGCs by two methods:
1) correlation of these parameters with receptive field center size
and 2) assigning RGCs to groups based on parameter values to
approximate the class scheme of Kerschensteiner et al. (2008).
Figure 3 shows scatter plots of receptive field center size
against the linear filter parameter for ON and OFF cells.
Mostly weak, statistically insignificant correlations were apparent. The only significant correlations were at P30 –39, for
both ON and OFF RGCs. At both younger ages, the filter
parameter for both ON and OFF cells was not correlated with
receptive field center diameter. We approximated the classification scheme of Kerschensteiner et al. (2008) by determining
the average receptive field diameters for RGCs in the lower and
upper thirds of the filter parameter distribution. These parts of
the distribution should contain RGCs of the “medium” and
“fast” (for ON cells) and “monophasic” and “biphasic” (for
OFF cells) types, respectively, as defined in that study. We
found that the average receptive field center diameter for each
group of RGCs classified in this way followed the same pattern
as the average diameter of the whole population of ON or OFF
RGCs, i.e., for each RGC group, the receptive field center
became smaller after eye opening (data not shown), indicating
that these major RGC classes all follow the same pattern of
development.
899
900
RECEPTIVE FIELD DEVELOPMENT OF MOUSE RGCs
average at eye opening than at later ages. The average STA
contrasts for ON and OFF RGC responses, respectively, were
(means ⫾ SE) 1.93 ⫾ 0.05% and 1.76 ⫾ 0.10% contrast
(P13–14), 3.84 ⫾ 0.09% and 3.34 ⫾ 0.14% contrast (P16 –17),
and 3.48 ⫾ 0.13% and 3.60 ⫾ 0.12% contrast (P30 –39)
(Fig. 5C). This difference in contrast can be seen in the
receptive field maps shown in Fig. 1C. These data agree
with our observation that acquiring receptive field maps
required longer periods of white noise stimulation at P13–14
than at P16 –17 and P30 –39.
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
Fig. 4. Latency to spiking decreased during development after eye opening. A
and B: latency to spiking was determined from the linear filters generated from
the STA images by finding the zero-crossing time before spiking of the major
contrast peak of the linear filter. Histograms of latencies (in ms) are shown for
ON RGCs (left, light gray) and OFF RGCs (right, dark gray) at P13–14 (top),
P16 –17 (middle), and P30 –39 (bottom). Bin amplitudes are expressed as the
fraction of the total RGCs of that type and that age recorded. C: average
latencies for the same cells as in A and B for ON RGCs (light gray bars) and
OFF RGCs (dark gray bars) for the ages indicated. Error bars indicate SEs.
*P ⬍ 0.05.
stimulus necessary to drive a response (Demb 2008). In one
sense, this filter is applied (by convolution) by the RGC to any
given series of white noise images that are presented. The
resulting value predicts the likelihood of the RGC firing to that
given stimulus. Thus, a large filter amplitude indicates that the
RGC fires in response to lower contrast stimuli, whereas a
small filter amplitude indicates that it requires higher contrast
stimuli to evoke firing.
We calculated the overall amplitudes of the filter (the STA
contrast) and found that they were significantly lower on
J Neurophysiol • VOL
Fig. 5. STA contrast increased during development after eye opening. A and
B: maximum STA contrast was determined by the difference in the maximum and
minimum contrast values of the linear filter (i.e., as the magnitude of the
contrast difference between points c and b indicated in the linear filters in Fig.
1B) for each RGC. Histograms of STA contrasts (in %) are shown for ON
RGCs (left, light gray) and OFF RGCs (right, dark gray) at P13–14 (top),
P16 –17 (middle), and P30 –39 (bottom). Bin amplitudes are expressed as the
fraction of the total RGCs of that type and that age recorded. C: average
STA contrast for the same cells as in A and B for ON RGCs (light gray bars)
and OFF RGCs (dark gray bars) for the ages indicated. Error bars indicate
SEs. *P ⬍ 0.05.
106 • AUGUST 2011 •
www.jn.org
RECEPTIVE FIELD DEVELOPMENT OF MOUSE RGCs
Repetitive Spiking in Young RGCs Alters the Linear
Filter Shape
At P13–14 and P16 –17, several retinas had linear filters with
substantial oscillations. Examples of these filters, along with
their fits used for parameter calculation, are shown in Fig. 6, A
and B, left. Autocorrelations of the spike trains for these cells
revealed large peaks falling at intervals of ⬃400 – 450 ms (Fig.
6, A and B, right). This means that a light-evoked spike tended
to be followed by another spike ⬃400 – 450 ms later. The 1-s
sequence of images preceding one of these following spikes
J Neurophysiol • VOL
Fig. 6. Oscillations in some linear filters at both P13–14 and P16 –17 are due
to repetitive spiking synchronized by light-driven activity. Oscillations in some
but not all linear filters were seen at P13–14 and P16 –17. A: linear filters with
oscillations from a P16 ON RGC (top left) and a P16 OFF RGC (bottom left)
along with the first second of their spike autocorrelations (top right and bottom
right). The prominent peaks (at ⬃400 ms) in the autocorrelations can account
for the oscillations (see text). The earliest peak represents spiking during the
light response. B: linear filters from a P30 ON RGC (top left) and a P30 OFF
RGC (bottom left). RGCs recorded from P30-P39 retinas did not have oscillations. The spike autocorrelations for these RGCs (top right and bottom right)
did not have the additional peaks after the initial light response peak that were
seen from RGCs with oscillations in their linear filters, such as those in A.
will thus include the images that caused the RGC to fire the
initial spike; these images will appear shifted in time by the
400- to 450-ms interval. When all of the image sequences for
one of these RGC’s spikes are averaged, these shifted images
will appear as a peak in the linear filter shifted earlier in time.
Thus, oscillations in the linear filter arise from the fact that a
spike driven by the white noise stimulus image sequence was
likely to be followed by another spike at this interval. This
accounts for the multiple positive peaks in the linear filter
shown in Fig. 6, A and B. Removal of RGCs with such
oscillations did not alter the parameter distribution shapes or
106 • AUGUST 2011 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
At all ages, ON and OFF RGCs were recorded that were
relatively insensitive to white noise checkerboard stimulation
of their centers. The notable change with age in the STA
contrast distribution is that its width increased markedly from
P13–14 to P16 –17, indicating an increase in the number of
RGCs that responded to low contrasts. Although our data may
mask the possibility that groups of cells with high STA
contrasts at P16 –17 have lower contrasts at P30⫹ and vice
versa, we consider this not parsimonious. We think instead that
many cells increase their sensitivity or excitability, so that the
distribution reaches its mature state within 3 days of eye
opening, and that many RGCs with relatively low sensitivity at
eye opening likely remain insensitive throughout development.
Increases in RGC responsiveness can explain the STA contrast increase. The higher contrast linear filters of RGCs can be
explained by increased RGC responsiveness to receptive field
center stimulation, but we considered three other possibilities.
First, if younger RGCs had a high rate of spontaneous spiking,
a decrease in the linear filter’s overall contrast would occur
because the pixels viewed by the receptive field center in the
images preceding those spikes would average to the background mean gray level. This was not the case. Although we
cannot distinguish spontaneous spikes from light-driven ones,
we did find that total spiking decreased for ON cells (mean
rates: 1.44 Hz at P13–14, 1.30 Hz at P16 –17, and 1.25 Hz at
P30 –39) and increased for OFF cells (0.59 Hz at P13–14, 0.82
Hz at P16 –17, and 1.31 Hz at P30 –39), yet STA contrast
increased with age for both ON and OFF RGCs.
Second, larger numbers of pixels in the larger receptive field
centers at P13–14 could cause a lower contrast on average than
smaller numbers of pixels in smaller centers. The pixel intensities that make up each white noise checkerboard are random
and uncorrelated with one another, so it is less likely that high
contrast combinations of pixels will be presented to a large
center. For this to explain the lower STA contrast at P13, we
would expect STA contrast to be correlated to receptive field
diameter at any age, and this was not the case (data not shown).
Finally, if surround strengths markedly differ during this
developmental period, the effect of those surrounds on firing
due to center stimulation could depend on age and perhaps
explain the change in STA contrasts. Surrounds, however, are
poorly stimulated by white noise checkerboards because their
larger area encompasses many more stimulus pixels, making it
unlikely that sequences of these images would produce the
proper contrast profile for surround stimulation. (This weak
surround stimulation can be seen in the STA images shown in
Fig. 1D as a slight halo of opposing contrast surrounding the
center.) Thus, the higher contrast of the RGC linear filters
appears due to an increase in RGC responsiveness.
901
902
RECEPTIVE FIELD DEVELOPMENT OF MOUSE RGCs
lead to significantly different average receptive field diameters
among the RGCs (data not shown).
DISCUSSION
J Neurophysiol • VOL
106 • AUGUST 2011 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
Our results show that receptive field centers of RGCs in the
mouse retina decrease in size during the 2-wk period after eye
opening. We mapped the spatiotemporal receptive fields of
three postnatal age groups: at eye opening (P13–14), about 3
days later (P16 –17), and after the fourth postnatal week (P30 –
39). White noise stimulus mapping showed that the receptive
field centers of ON cells progressively declined in size during
this time. In the OFF cell population, the two younger ages had
similar sized average centers, and these were both significantly
larger than in the adult. At P13–14 and P16 –17, OFF RGC
receptive field centers were smaller overall than the ON cells,
but at maturity, ON and OFF fields were the same size when
averaged across the whole population. After we accounted for
retinal area expansion during development, the decrease in
receptive field diameter after eye opening was 31% for ON
cells and 25% for OFF cells in degrees of visual angle.
At eye opening, the intrinsic firing properties of RGCs in
many mammals have been found to be immature. In mouse
RGCs, both the resting potential and firing threshold hyperpolarize during development, and the cells become capable of
firing at higher rates (Qu and Myhr 2008; Rothe et al. 1999a,
1999b). At the same time, the strength of light-evoked inputs to
RGCs, both excitatory and inhibitory, decrease from eye opening through ⬃P30 (He et al. 2011). We found that latencies
were long at eye opening, particularly for OFF cells, and
decreased to near mature levels by P16 –17, similar to the
results of other studies (Anishchenko et al. 2010; Tian and
Copenhagen 2003). The increase in RGC responsiveness with
age is indicated by the increase in the maximum contrast of the
linear filters. Stronger stimuli are necessary to cause spiking at
eye opening than in the adult, suggesting that stimulus-driven
excitability of RGCs is lower, despite the greater size of
light-evoked currents at early ages. By P16 –17, the population
average and width of the contrast distribution reached adult
levels. However, RGCs at both P16 –17 and P30 –39 had linear
filters with contrasts as low as those found at P13–14, suggesting that some RGCs that are only sensitive to high contrast
stimuli already have their adult sensitivity at eye opening.
Linear filters from some retinas at the two younger ages, but
not at the mature ages, had additional peaks that were a result
of oscillatory activity at a frequency of ⬃2.5 Hz. Some retinas
at the younger ages thus exhibit a strong tendency to spike at
that interval in our recording conditions. Oscillatory release of
neurotransmitter from bipolar cells may be an intrinsic property of retinal circuitry (Petit-Jacques et al. 2005). Spontaneous
retinal activity occurs in the early postnatal retina through the
first week after eye opening (Blankenship et al. 2009; Meister
et al. 1991; Stafford et al. 2009; Wong 1999). While these
retinal “waves” are well characterized, many neurons in the
retina have been shown to exhibit spontaneous and light-driven
oscillatory activity over a large range of frequencies and ages
(Neuenschwander et al. 1999; Petit-Jacques and Bloomfield
2008; Petit-Jacques et al. 2005). We hypothesize that firing of
the light-evoked spike resets the mechanism that drives the
spontaneous, maintained discharge of the cells, which tended
to occur at ⬃2.5 Hz here, and that the following spike is the
resumption of that maintained firing.
The RGCs recorded by the MEA and analyzed in our study
were necessarily only those that responded to the white noise
checkerboard stimulus used here and that yielded linear filters.
These will be dominated by ON and OFF RGCs with concentric center-surround receptive fields. Ideally, each RGC functional type would be followed through development to determine if individual classes go through unique developmental
programs (Coombs et al. 2007). In other studies, a large
percentage of rodent RGCs responsive to white noise checkerboards were segregated into four functional classes by the
characteristics of their STA time course (Anishchenko et al.
2010; Kerschensteiner et al. 2008). Although RGC functional
groups could not be resolved by PCA in our study, we segregated RGC types into these four classes using ranges of linear
filter parameters that appear to account for most of the variance
that separates the groups (Kerschensteiner et al. 2008). Receptive field center diameter decreased with age for all of these
groups, although the lack of clusters in PCA suggests that these
groups were likely contaminated by RGCs of other types.
A recent study (Cantrell et al. 2010) of the mouse retina also
examined receptive field center size using white noise checkerboard stimuli using a technique capable of identifying ONOFF RGCs in addition to ON and OFF RGCs. Sizes decreased
from P18 to P25 for ON-OFF and ON RGCs. OFF RGC
receptive fields trended smaller but were not significantly
different at the two ages, which is different than our results but
is over a much shorter age span. The study confirmed significant immaturity near eye opening (at P15) but found that
receptive field center area increased from P15 to P18, in
contrast to what we observed from P13–14 to P16 –17. Correcting for increasing retinal area with growth would not
eliminate this discrepancy. Recording conditions and stimulus
settings differed, which may account for this. Our stimuli were
presented at a higher refresh rate with higher maximum and
mean luminances. It is possible that the brighter stimuli were
able to drive additional RGCs in retinas at eye opening because
many were relatively insensitive, making it difficult to compare
the populations of RGCs in the two studies. On the other hand,
our higher refresh rate makes our stimulus less likely to elicit
a response. This may be more important at eye opening
because RGCs at that age appear to have a much slower
temporal frequency response than RGCs at P16 or P30 (R. C.
Rentería and N. P. Akimov, unpublished observations). A
higher refresh rate might suggest the stimulus frames are
essentially being averaged by the RGCs, which, in turn, might
imply a smoothing of the stimulus. Nonetheless, such smoothing would not cause broadening of the mapped receptive field
because the stimulus pixels lack any correlations. Similarly,
differences in checker size do not appear to change the measured receptive field size (Cantrell et al. 2010). These results
support the idea that considerable changes occur in the short
time period of the few days after eye opening.
Because RGCs at P8 can be grouped into the classes seen
in adult with no cells of intermediate morphological stage
being observed after eye opening (Coombs et al. 2007; Diao
et al. 2004; Sun et al. 2002), individual RGCs most likely do
not change their morphological type across development; a
type observed at eye opening would be recognized as the
same type at later ages. This classification is when the RGC
dendrites are viewed in the plane of the retina. These dendrites,
however, are dynamic in the vertical plane, shifting layers
RECEPTIVE FIELD DEVELOPMENT OF MOUSE RGCs
ACKNOWLEDGMENTS
The authors thank the Computational Biology Initiative (University of
Texas Health Science Center and University of Texas, San Antonio, TX) for
providing access to computing resources.
Present address of C. L. Koehler: Div. of Biological Sciences, Univ. of
California-San Diego, 9500 Gilman Dr., La Jolla, CA 92093.
GRANTS
This work was supported by a University of Texas Health Science Center
Research Committee New Faculty Award, pilot project awards from the
Nathan Shock Institute, and University of Texas Health Science Center
Institute for the Integration of Medicine and Science Clinical and Translational
Science Award UL1-RR-025767 (to R. C. Rentería).
DISCLOSURES
No conflicts of interest, financial or otherwise, are declared by the author(s).
REFERENCES
Anishchenko A, Greschner M, Elstrott J, Sher A, Litke AM, Feller MB,
Chichilnisky EJ. Receptive field mosaics of retinal ganglion cells are
established without visual experience. J Neurophysiol 103: 1856 –1864,
2010.
Blankenship AG, Ford KJ, Johnson J, Seal RP, Edwards RH, Copenhagen
DR, Feller MB. Synaptic and extrasynaptic factors governing glutamatergic
retinal waves. Neuron 62: 230 –241, 2009.
Brainard DH. The psychophysics toolbox. Spat Vis 10: 433– 436, 1997.
Brown SP, He S, Masland RH. Receptive field microstructure and dendritic
geometry of retinal ganglion cells. Neuron 27: 371–383, 2000.
Cantrell DR, Cang J, Troy JB, Liu X. Non-centered spike-triggered covariance analysis reveals neurotrophin-3 as a developmental regulator of
receptive field properties of ON-OFF retinal ganglion cells. PLoS Comput
Biol 6: e1000967, 2010.
Chichilnisky EJ. A simple white noise analysis of neuronal light responses.
Network 12: 199 –213, 2001.
J Neurophysiol • VOL
Coombs J, van der List D, Wang GY, Chalupa LM. Morphological
properties of mouse retinal ganglion cells. Neuroscience 140: 123–136,
2006.
Coombs JL, Van Der List D, Chalupa LM. Morphological properties of
mouse retinal ganglion cells during postnatal development. J Comp Neurol
503: 803– 814, 2007.
Demb JB. Functional circuitry of visual adaptation in the retina. J Physiol 586:
4377– 4384, 2008.
Diao L, Sun W, Deng Q, He S. Development of the mouse retina: emerging
morphological diversity of the ganglion cells. J Neurobiol 61: 236 –249,
2004.
Hamasaki DI, Sutija VG. Development of X- and Y-cells in kittens. Exp
Brain Res 35: 9 –23, 1979.
He Q, Wang P, Tian N. Light-evoked synaptic activity of retinal ganglion and
amacrine cells is regulated in developing mouse retina. Eur J Neurosci 33:
36 – 48, 2011.
Hensch TK. Critical period plasticity in local cortical circuits. Nat Rev
Neurosci 6: 877– 888, 2005.
Kerschensteiner D, Liu H, Cheng CW, Demas J, Cheng SH, Hui CC,
Chow RL, Wong RO. Genetic control of circuit function: Vsx1 and Irx5
transcription factors regulate contrast adaptation in the mouse retina. J
Neurosci 28: 2342–2352, 2008.
Landi S, Cenni MC, Maffei L, Berardi N. Environmental enrichment effects
on development of retinal ganglion cell dendritic stratification require retinal
BDNF. PLoS ONE 2: e346, 2007.
Liu X, Grishanin RN, Tolwani RJ, Rentería RC, Xu B, Reichardt LF,
Copenhagen DR. Brain-derived neurotrophic factor and TrkB modulate
visual experience-dependent refinement of neuronal pathways in retina. J
Neurosci 27: 7256 –7267, 2007.
Masland RH. Maturation of function in the developing rabbit retina. J Comp
Neurol 175: 275–286, 1977.
Meister M, Wong RO, Baylor DA, Shatz CJ. Synchronous bursts of action
potentials in ganglion cells of the developing mammalian retina. Science
252: 939 –943, 1991.
Neuenschwander S, Castelo-Branco M, Singer W. Synchronous oscillations
in the cat retina. Vision Res 39: 2485–2497, 1999.
Pelli DG. The VideoToolbox software for visual psychophysics: transforming
numbers into movies. Spat Vis 10: 437– 442, 1997.
Petit-Jacques J, Bloomfield SA. Synaptic regulation of the light-dependent
oscillatory currents in starburst amacrine cells of the mouse retina. J
Neurophysiol 100: 993–1006, 2008.
Petit-Jacques J, Volgyi B, Rudy B, Bloomfield S. Spontaneous oscillatory
activity of starburst amacrine cells in the mouse retina. J Neurophysiol 94:
1770 –1780, 2005.
Qu J, Myhr KL. The development of intrinsic excitability in mouse retinal
ganglion cells. Dev Neurobiol 68: 1196 –1212, 2008.
Remtulla S, Hallett PE. A schematic eye for the mouse, and comparisons with
the rat. Vision Res 25: 21–31, 1985.
Ren L, Liang H, Diao L, He S. Changing dendritic field size of mouse retinal
ganglion cells in early postnatal development. Dev Neurobiol 70: 397– 407,
2009.
Rentería RC, Tian N, Cang J, Nakanishi S, Stryker MP, Copenhagen DR.
Intrinsic ON responses of the retinal OFF pathway are suppressed by the ON
pathway. J Neurosci 26: 11857–11869, 2006.
Rothe T, Bahring R, Carroll P, Grantyn R. Repetitive firing deficits and
reduced sodium current density in retinal ganglion cells developing in the
absence of BDNF. J Neurobiol 40: 407– 419, 1999a.
Rothe T, Juttner R, Bahring R, Grantyn R. Ion conductances related to
development of repetitive firing in mouse retinal ganglion neurons in situ. J
Neurobiol 38: 191–206, 1999b.
Rusoff AC, Dubin MW. Development of receptive-field properties of retinal
ganglion cells in kittens. J Neurophysiol 40: 1188 –1198, 1977.
Sernagor E, Eglen SJ, Wong RO. Development of retinal ganglion cell
structure and function. Prog Retin Eye Res 20: 139 –174, 2001.
Sernagor E, Grzywacz NM. Emergence of complex receptive field properties
of ganglion cells in the developing turtle retina. J Neurophysiol 73: 1355–
1364, 1995.
Stafford BK, Sher A, Litke AM, Feldheim DA. Spatial-temporal patterns of
retinal waves underlying activity-dependent refinement of retinofugal projections. Neuron 64: 200 –212, 2009.
Sun W, Li N, He S. Large-scale morphological survey of mouse retinal
ganglion cells. J Comp Neurol 451: 115–126, 2002.
106 • AUGUST 2011 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
within the IPL after eye opening (Liu et al. 2007; Xu and Tian
2007).
This vertical shift in the IPL has profound functional implications for RGC function. RGCs collect excitatory input from
bipolar cell axon terminals, and these terminals stratify in
various sublayers of the IPL. RGC dendritic stratification
typically defines if the input will be from ON or OFF bipolar
cells and so determines which features of a visual stimulus will
drive the cell. Xu and Tian (2007) examined the morphology of
mouse RGCs of all types at P12 and P33. They found that a
majority of RGCs significantly shift their dendritic trees vertically through the layers of the IPL, making and breaking
synaptic contacts with bipolar cells. RGCs with dendrites in
both the ON and OFF layers of the IPL in young retinas
appeared to shift to a single IPL layer with development. That
ON-OFF RGCs decrease after eye opening is evident by
physiology and morphology in other studies as well (Cantrell et
al. 2010; Landi et al. 2007; Liu et al. 2007; Tian and Copenhagen 2003). Thus, some ON and OFF cells detected at P30
likely would not have been detected from the same retina at eye
opening, when they were ON-OFF responsive.
Overall, we found that receptive field center size decreases
and intrinsic firing properties mature in ON and OFF RGCs of
the mouse retina during development after eye opening. These
findings have implications for the developmental time course
observed for the rest of the mouse’s visual system and are one
source of that system’s immaturity at eye opening and its
subsequent critical period of development.
903
904
RECEPTIVE FIELD DEVELOPMENT OF MOUSE RGCs
Tagawa Y, Kanold PO, Majdan M, Shatz CJ. Multiple periods of functional
ocular dominance plasticity in mouse visual cortex. Nat Neurosci 8: 380 –
388, 2005.
Tian N, Copenhagen DR. Visual deprivation alters development of synaptic
function in inner retina after eye opening. Neuron 32: 439 – 449, 2001.
Tian N, Copenhagen DR. Visual stimulation is required for refinement of ON
and OFF pathways in postnatal retina. Neuron 39: 85–96, 2003.
Tootle JS. Early postnatal development of visual function in ganglion cells of
the cat retina. J Neurophysiol 69: 1645–1660, 1993.
Wong RO. Retinal waves and visual system development. Annu Rev Neurosci
22: 29 – 47, 1999.
Xu HP, Tian N. Retinal ganglion cell dendrites undergo a visual activitydependent redistribution after eye opening. J Comp Neurol 503: 244 –259,
2007.
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 17, 2017
J Neurophysiol • VOL
106 • AUGUST 2011 •
www.jn.org