Transcript No Slide Title

Coordinated burst firing in the retina and lateral geniculate nucleus of the macaque
Lawrence C. Sincich, Ken F. Park, Daniel L. Adams, John R. Economides and Jonathan C. Horton
Beckman Vision Center, University of California San Francisco, San Francisco, CA 94143
Bursting activity in the retina and LGN
spike train
≥100 ms
≤4 ms
LGN spikes
Interspike intervals that are often used to identify when I T is active. From intracellular
studies in cats, the combination of these interspike intervals has been correlated with IT
current.
sweep
Timing criterion for LT bursts
sweep
The spiking activity of neurons in the lateral geniculate
nucleus (LGN) is often categorized into two modes: burst and
tonic. The bursting mode has been shown in cats and guinea pigs
to depend on activation of the low-threshold calcium current (IT).
Characteristically, all spikes but the first one in a burst do not
require additional synaptic input to occur because IT depolarizes
the cell, generating several INa action potentials.
Although bursting is commonly associated with sleep
states, it is also found under awake conditions. Bursting has been
proposed as mechanism for signaling novelty to neurons in V1. In
the macaque monkey, it is not known if IT is active when stimuli
are derived from natural scenes. It is also unclear how the firing
pattern in the retina influences the mode of firing in the LGN.
We have addressed several questions about firing activity
in the macaque LGN: (1) What is the dominant firing mode during
a stimulus paradigm that mimics the viewing of natural scenes?
(2) Are LGN bursts simply due to bursting activity in the retina?
(3) Does the LGN ignore retinal input during bursts because of IT
activation?
To answer these questions, we recorded the action
potentials of LGN neurons along with their retinal input in the form
of excitatory postsynaptic potentials (S-potentials) using a single
extracellular electrode. This method offers the advantage of
extended recording periods, wherein the LGN could switch
between tonic and burst modes, while providing access to the
firing patterns on both sides of the retinogeniculate synapse. LGN
spikes and retinally driven postsynaptic potentials have distinct
waveforms:
Identifying retinal input to the LGN
Growth of retinal ganglion cell S-potentials as microelectrode approaches an LGN
neuron. During this 15 min trace, the electrode was advanced 80 mm from the point the
unit was first recorded. The initial waveform (timepoint 1) was triphasic, characteristic of
an LGN somatic spike recorded at a distance. By timepoint 5, S-potentials began to be
observed in isolation and immediately preceding LGN spikes. After timepoint 8, the Apotential occasionally appears with S-potentials, but without the B-potential, indicating a
failure of propagation into the somatodendritic tree.
RGC spikes
Introduction
Merged S-potentials are identifiable in both tonic and burst LGN spikes. A template
LGN spike, stripped of its S-potential, was derived from a spike which had an long Spotential latency, but within the absolute refractory period of the cell (bottom left, in red).
This template was aligned at the half-height point of the A-potential of other spikes,
displayed in rank-order according to the S-potential latency (left column). After
subtracting the template, the remaining waveforms reveal S-potentials (middle column).
The distribution of S-potential latencies for 1,000 spikes show that most LGN spikes
occur within 0.5 msec of an S-potential (right column). From this set, 42 spikes were
categorized as secondary spikes in a burst, and their S-potential latencies had a similar
distribution (black circles). The presence of S-potentials in burst spikes suggests that
LGN bursting is inherited from retinal spike trains rather than the activation of IT.
Examples of LGN bursts during naturalistic visual stimulation. During episodes of burst
firing (interspike intervals < 4 ms), S-potentials can be seen in isolation and clearly
leading some of the LGN spikes, including those after the first spike. In other LGN
spikes, the S-potentials are challenging to identify without a subtraction procedure.
S-potentials precede and drive nearly all LGN spikes in a burst. (Top) A rastergram and
spike histogram of 190 LT bursts, aligned on the first LGN spike of each burst. The
preceding 100 msec spike-free epochs are thought to be definitive of IT activity.
(Bottom) Rastergram and histogram of the S-potentials accompanying each LT burst
shown above. The RGC is not silent during the pre-LGN burst interval, nor is its input
ignored during secondary spikes in a burst. Before nearly all LGN bursts, one “priming”
retinal input appears within the prior 20 ms, followed by the S-potential that generates
the LGN burst.
0.2 mV
1 sec
Anatomy of an LGN spike and the S-potential. Averaged traces of LGN spikes (black)
and isolated S-potentials (blue) in the macaque show stereotypic features in their
waveforms. The S-potential is the excitatory postsynaptic potential (EPSP) originating
from retinal input. It is monophasic and preceded by a smaller presynaptic potential
designated “T” (Wang, Cleland & Burke, Brain Res., 1985). LGN spikes have these
same features, although the S-potential partially merges with the rising phase of the
action potential, itself exhibiting two components. The “A” potential is due to the action
potential initiated at the axon hillock, and is followed by the “B” potential which is the
backpropagation of the action potential into the somatodendritic tree.
Methods
Experiments were conducted in 6 adult male macaques
using procedures approved by the UCSF Committee on Animal
Research. Extracellular recordings of neurons in the LGN were
made under standard conditions, with isoflurane as anesthetic.
Paralysis with vecuronium bromide was used to eliminate eye
movements.
We searched for LGN neurons with associated S-potentials
arising from retinal ganglion cells (RGCs). From a total of 240
cells recorded, 19 S-potential/LGN recordings with sufficient data
quality were chosen for analysis. Each LGN cell’s receptive field
center was stimulated with a light spot produced by an LED of the
preferred color shining on the back of a translucent screen. The
receptive field surround was masked with black flocking paper.
The luminance of the LED was varied at 80 Hz according to a
naturalistic temporal frequency distribution (van Hateren, Vision
Res. 1997). The temporal power spectrum of this stimulus
followed a 1/f power law, often called “pink noise”.
LGN spikes and S-potentials were classified as either tonic
or burst. A burst was any group of spikes with an interspike
interval less than 4 msec. Bursts preceded by a silent period of at
least 100 msec were subclassified as low-threshold (LT),
reflecting the dynamics of the thalamic low-threshold calcium
current IT (Lu et al., J. Neurophysiol., 1992). All remaining spikes
were designated as tonic.
secondary burst spikes
cardinal burst spikes
Retinal ganglion cells recorded directly show burst firing behavior to naturalistic stimuli.
Two examples of retinal ON cells are shown bursting in response to a step increase in
luminance (red arrow). The top trace was recorded in the LGN and contained Spotentials, while the bottom trace was recorded in the optic tract from a different animal.
Both traces show LT burst firing in the retina, with strikingly similar latencies to stimulus
onset.
S-potentials recorded during visual stimulation. (Top) Luminance variation of a spot of
light centered on the receptive field of an parvocellular ON-center cell. (Middle) Raw
electrode trace acquired during a 10 sec stimulus sample. (Bottom) Expanded segment
of the recording shown above, illustrating the occurrence of S-potentials associated with
LGN A-potentials and somatodendritic B-potentials. S-potentials which occur at the
beginning of the B-potentials are often difficult to identify because the waveforms merge,
but in this example a T-potential notch is still visible.
Conclusions
1. During naturalistic stimulation, nearly all LGN spikes are driven
by retinal spikes, and the tonic mode predominates.
2. When the LGN fires in burst mode, it is slavishly responding to
burst firing in the retina.
secondary burst spikes
S-potentials can arise from a single retinal ganglion cell. (Top) Average of 71 isolated Spotentials recorded with no other events occurring within 20 msec during spontaneous
activity. Gray shading represents +/-1 standard deviation around the baseline noise.
(Bottom) Average of 9,416 S-potentials during visually driven activity. The larger
standard deviation before and after the S-potential reflects the occurrence of other
spiking events. During the 2 msec prior to the S-potential, the standard deviation is
identical to that seen in the upper trace, indicating an absolute refractory period.
S-potentials driving LGN spikes can be unmasked using muscimol. (Top) During
continuous stimulation with a light spot turning on and off at 3 Hz, a 0.5% solution of
muscimol is injected in the LGN to hyperpolarize the neuron by opening GABAA receptor
channels. After a few minutes, the LGN neuron begins to cease firing (left), while the
RGC continues to fire at the same rate (right). Thus, muscimol does not block
presynaptic GABA receptors in the macaque LGN. (Bottom) Expanded views of the
recording episodes marked by red and green arrows above. During the control trace, Spotentials as well as S-A and S-A-B waveforms are present. After muscimol application,
all but one of the A- and B-potentials have disappeared. This provides further evidence
that S-potentials initiate most LGN spikes.
cardinal burst spikes
During naturalistic stimuli, RGCs and LGN neurons fire predominantly in tonic mode.
(Top) Scatterplot of pre- versus postsynaptic spike intervals for a parvocellular ON
neuron reveals a wide range of interspike times (n=10,000). Less than 1% of these
spikes could be classified as LT burst spikes, based on timing criteria (red dots). The
“secondary” burst spikes are presumed to be caused by IT activation, theoretically
requiring no input from the retina. (Bottom) Scatterplot of the S-potentials for the same
LGN neuron. The proportion of bursting activity between RGCs and LGN neurons is
similar.
3. LGN bursts require a “priming” input from the retina to be
initiated.
4. Under our stimulus and recording conditions, the low-threshold
calcium current IT is not required to drive LGN burst firing.