| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Neurobiology and Physiology, Northwestern University, Evanston 60201; 2 Physical Therapy Program, Midwestern University, Downers Grove, Illinois 60515
Submitted 21 May 2002; accepted in final form 17 April 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
; Guitton
1991; Schall 1997;
Sparks and Hartwich-Young
1989; Wurtz and Munoz
1995). In both structures, depending on the location of the
electrode site within the topographic map of eye movement space, low-threshold
microstimulation evokes contraversive saccades or suppresses saccades and
produces active fixation (Bruce et al.
1985; Burman and Bruce
1997; Munoz and Wurtz
1993b; Robinson
1972; Robinson and Fuchs
1969). In addition, reversible inactivation of the frontal eye
field or the superior colliculus interferes with the generation of saccades
and fixation (Dias and Segraves
1999; Hikosaka and Wurtz
1985,
1986;
Schiller et al. 1987;
Sommer and Tehovnik 1997).
Surgical ablation of only the frontal eye field or superior colliculus
produces mild to moderate deficits in oculomotor behavior while combined
lesions of the frontal eye field and superior colliculus produce profound
oculomotor deficits, suggesting that there are both serial and parallel
contributions from these areas to the control of eye movements
(Deng et al. 1986;
Schiller et al. 1987).
Elaborating on the relative contributions of serial and parallel inputs of the
frontal eye field and superior colliculus to the control of saccades, a recent
report indicates that, in spite of direct links between the frontal eye field
and oculomotor centers in the brain stem, it is the indirect, serial pathway
from frontal eye field to colliculus to brain stem oculomotor centers that is
the most important for the control of eye movements
(Hanes and Wurtz 2001).
Neurons within both structures exhibit a variety of eye-movement-related
activity. During voluntary, purposive saccades, frontal eye field neurons may
respond to visual stimuli within the neuron's receptive field as well as
produce movement-related activity associated with the generation of a saccade
of a preferred amplitude and direction
(Bruce and Goldberg 1985;
Mohler et al. 1973;
Schall 1991). In contrast to
brain stem oculomotor centers, which are involved in the direct control of
oculomotor neurons, the frontal eye field appears to be involved in a higher
level of control of eye movements. Neurons in the frontal eye field show
activity related to the process of saccade target selection during both visual
search tasks and scanning of natural images
(Bichot et al. 1996;
Burman and Segraves 1994;
Schall and Hanes 1993;
Schall et al. 1995;
Thompson et al. 1996). Within
the deep layers of the superior colliculus, saccade-related activity has been
characterized as including a prolonged anticipatory-like activity prior to the
onset of the saccade and/or a discrete burst of activity associated with the
onset of the saccade (Glimcher and Sparks
1992; Mays and Sparks
1980; Mohler and Wurtz
1976; Munoz and Wurtz
1995b; Sparks
1978; Sparks et al.
1976). In this report, we define deep layers as all collicular
layers located below the superficial layers (superficial gray and stratum
opticum), including the intermediate and deep gray layers. Both frontal eye
field and superior colliculus neurons have been shown to generate preparatory
set activity that is directly correlated with saccadic reaction time (Basso
and Wurtz 1997,
1998;
Dorris and Munoz 1998;
Dorris et al. 1997;
Everling et al. 1999;
Everling and Munoz 2000).
As part of the overall network responsible for the control of voluntary
saccades, the deep layers of the superior colliculus receive input from both
cortical and subcortical structures involved in this process
(Grantyn 1988;
Moschovakis et al. 1996). The
signals sent from the frontal eye field to the superior colliculus have been
characterized. It has been known for some time that frontal eye field neurons
with fixation and saccade-related activity project to the deep layers of the
superior colliculus, providing information to the superior colliculus
concerning the maintenance and release of fixation and the generation of the
saccade (Segraves and Goldberg
1987). More recently, this finding has been confirmed and extended
to include corticotectal neurons with visually driven activity located
primarily within lateral portions of the frontal eye field
(Sommer and Wurtz 2000). Thus
there appears to be little selectivity in the output of frontal eye field
activity that is sent to the superior colliculus. Frontal eye field input
provides collicular neurons with an upper level eye movement signal relaying
information concerning fixation, anticipatory activity prior to the appearance
of a target, the presence and selection of visual targets, and
movement-related activity for saccades of specified amplitude and direction
(Schlag-Rey et al. 1992;
Segraves and Goldberg 1987;
Sommer and Wurtz 2000).
Is there selectivity in the types of superior colliculus neurons that receive this frontal eye field input? This study attempts to answer this question by identifying superior colliculus neurons that receive direct excitatory frontal eye field input. Extracellular recordings were made simultaneously in the frontal eye field and superior colliculus of rhesus monkeys, and antidromic excitation of frontal eye field neurons from the colliculus was used to isolate interconnected cortical and collicular sites. Microstimulation at the frontal eye field site was then administered to orthodromically activate superior colliculus neurons that received short-latency frontal eye field input. The activity of collicular neurons receiving direct frontal eye field input was then characterized during visuomotor tasks. This identification and characterization of the types of superior colliculus neurons that receive excitatory frontal eye field input contributes to our understanding of how these components of the oculomotor network interact to control the generation of saccades. This work also provides insight into the involvement of the different functional types of superior colliculus neurons in the generation of voluntary saccades.
A preliminary report of this study has been published
(Helminski and Segraves
1996).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Preoperative training
The monkeys were trained preoperatively in a simple visual fixation task with liquid reward. The preoperative training taught monkeys to get in and out of a primate chair and attend to a visual stimulus. Monkeys were seated in a primate chair positioned in front of a tangent screen. Each trial began when the monkey touched a metal bar in front of her, causing a spot of light to appear on the screen. When the light dimmed, the monkey was required to release the bar to obtain a liquid reward. Preoperative training was completed when the monkey was ready to progress to more difficult oculomotor tasks that required monitoring of eye position.
Surgery
Two surgeries were performed on each animal under aseptic conditions. Prior to surgery, the monkey was anesthetized and tranquilized with ketamine hydrochloride (10 mg/kg im) and methohexital sodium (11 mg/kg iv or to effect), and secretions were reduced by administration of atropine sulfate (0.05 mg/kg im). For the duration of the surgery, anesthesia was maintained with halothane inhaled through an endotracheal tube.
During the initial surgery, a subconjunctival wire coil was implanted in
the left eye to measure eye position using the magnetic search coil technique
(Judge et al. 1980;
Robinson 1963). Two trephine
holes were made through the skull to enable a microelectrode to be inserted
through the dura to record neuron activity from the frontal eye field and
superior colliculus. Slots were cut through the skull extending away from the
trephine holes. Stainless steel bolts were positioned in the slots and secured
with stainless steel nuts and washers. Thin, curved strips of stainless steel
were used to interconnect the bolts. The stainless steel bolts and strips
functioned to enhance the bond between the skull and the dental acrylic.
Recording cylinders were positioned over the left frontal eye field and
superior colliculus trephine holes and bonded to the skull with dental
acrylic. The frontal eye field cylinder was centered at stereotaxic
coordinates anterior +25.0 mm and lateral 20.0 mm. The orientation of the
frontal eye field cylinder was adjusted to allow penetrations to be made
roughly parallel to the bank of the arcuate sulcus. The superior colliculus
cylinder was tilted posteriorly, 38° from vertical in the sagittal plane
and oriented so that a line of projection through its center would pass
through a point on the midline 15 mm above and 1 mm posterior to the
inter-aural line. This location enabled us to record from both left and right
superior colliculi. A steel receptacle to fix the monkey's head during
recording sessions and a connector for the eye coil were fixed in place and
bonded to the skull with dental acrylic. For both monkeys, a right frontal eye
field recording cylinder was later added during a second surgery.
To prevent infection, the monkey was given the antibiotic cefazolin (25 mg/kg iv) prior to and immediately after surgery as well as twice a day for 1 wk after surgery (25 mg/kg im). The analgesic buprenorphine hydrochloride (0.01 mg/kg im) was administered to the monkey at the end of surgery and twice a day for 3–4 days after surgery.
Postoperative training
Postoperative training began 10–14 days after surgery. The monkey was seated in a primate chair with its head restrained and centered within horizontal and vertical magnetic field coils. All behavioral control and data collection were performed via a PDP-11/73 (Digital Equipment). With the room dimly lit, visual stimuli originating from light-emitting diodes as well as laser diodes were rear-projected onto a tangent screen in front of the monkey. Stimuli consisted of a stationary and a movable light where the movable stimulus was positioned by a pair of servo-controlled mirror galvanometers (General Scanner) driven by computer generated analog signals. Eye position was measured by the magnetic search coil system with a phase-sensitive detector (C.N.C. Engineering).
Visuomotor tasks
Monkeys were trained to perform visuomotor tasks for a liquid reward (Fig. 1). Visual stimuli were generated by using two independent light-emitting or laser diodes. The first was a stationary light located at the center of the tangent screen. This light was used as the initial fixation point for the monkey. The second was a movable light positioned by mirror galvanometers that was used as either a visual stimulus or a saccade target. Each visuomotor task began after the monkey had fixated the central light and maintained fixation of this light for a randomly varied interval of 100–400 ms.
|
The first two visuomotor tasks required the monkey to maintain fixation of the fixation point during the task.
FIXATION TASK (FIG. 1A). The monkey held fixation for the duration of the trial. During the trial, the fixation light was turned off for brief periods of time. This enabled the task to be used to distinguish between fixation-related activity and visual activity driven by the foveal stimulus. If a neuron's activity was primarily related to fixation behavior, we would expect a change in firing rate that was maintained throughout the period of time that the monkey fixated the central light. If instead, a neuron's activity was primarily related to sensory visual input, then we would expect to see changes in firing rate coinciding with the appearance and disappearance of the fixation light.
VISUAL NO-SACCADE TASK (FIG. 1B). The monkey held fixation while a single stimulus was briefly flashed in the periphery. The location of the stimulus on the screen could be varied within the monkey's visual field either by explicitly entering desired stimulus coordinates into the software controlling stimulus position or by joystick readout. The task was used to identify a neuron's visual receptive field by delineating the area of tangent screen where visual stimuli produced increased neuron firing.
The last four visuomotor tasks required the monkey to make saccades from the fixation point to a target location. For all but the gap task, once the center of a response field was identified, a single target positioned at the center of that response field was used.
VISUALLY GUIDED SACCADE TASK (FIG. 1C). After the monkey maintained fixation for 700–1,000 ms, the fixation point was turned off at the same time that a target appeared in the periphery. When the target appeared, the monkey looked at the target. The location of the target on the tangent screen could be varied within the monkey's visual field. The task was used to identify a saccade-related neuron's movement field and preferred movement vector.
OVERLAP TASK (FIG. 1D). During the overlap task, the fixation point and the target were illuminated together for 500 or 1,000 ms. The monkey maintained fixation as long as the fixation point was on, a period of 1,500 ms. To begin the task, the fixation point was turned on. After 500 (or 1,000) ms, the target was turned on. Then, at 1,500 ms, the fixation point was extinguished, and the monkey was required to look at the target. This task was used to distinguish between visual- and saccade-related neuronal activity when the monkey was required to make a saccade to a target.
GAP TASK (FIG. 1E). The monkey began this task by fixating the central light. The fixation light was kept on for 1,000 ms and then extinguished. At 400 ms after the disappearance of the fixation light, a target light was turned on, and the monkey was rewarded after making a saccade to this light. For this task, the target would appear at one of two positions with equal likelihood. The first target position was located at the center of the cell's response field. The second target position was positioned at a location with identical radius but with angle rotated 180° from the first target. This task reduces saccadic latency by removing the visual stimulus for active fixation prior to the appearance of the peripheral target light.
MEMORY-GUIDED SACCADE TASK (FIG. 1F). In this task, the monkey was required to make a saccade to a remembered target location. After the first 600 ms of a 1,000-ms fixation period, a peripheral target stimulus was turned on for a period of 300 ms. Then 100 ms after the target stimulus was turned off, the fixation point was extinguished, and the monkey was rewarded for making a saccade to the location on the tangent screen where the target had appeared. The separation that this task provided between when the visual target stimulus was present and when the monkey was allowed to make a saccade made it possible to distinguish between visual- and saccade-related neuronal activity.
In this study, the location of visual stimuli and targets are expressed in polar coordinates, radius and angle, where radius is equal to the distance, in degrees of arc, of the target or stimulus from the central fixation point. An angle of 0° describes a rightward horizontal direction, and a 90° angle describes an upward vertical direction.
In all tasks, the error window for comparison of eye position to fixation point and target position was ±3–5° in the horizontal and vertical directions.
Microelectrode recording and stimulation
Single-unit recordings were obtained using epoxy-insulated tungsten
microelectrodes (A-M Systems) with wire diameters of 0.203 and 0.254 mm for
the frontal eye field and 0.127 and 0.203 mm for the superior colliculus.
Electrode impedance ranged from 0.6 to 1.5 M
measured at a frequency of
1 kHz. Hydraulic microdrives (Narishige) were used to advance the electrodes,
and the isolation of the action potentials of single neurons was accomplished
using a computer-based waveform discriminator (Signal Processing Systems,
Prospect, Australia).
Electrodes were introduced into the brain through guide tubes held within a
plastic grid fastened within the recording cylinder
(Crist et al. 1988). The use
of guide tubes penetrating the dura allowed for the use of finer electrodes
that could not penetrate the dura mater alone. In addition, guide tubes helped
to increase electrode stability and facilitated the making of repeated
penetrations through the same neural sites. When a guide tube was not being
used, a stainless steel wire, coated with antibiotic (3% tetracycline HCL) was
inserted into it to prevent infection.
To identify the frontal eye field, electrode penetrations were made in the
region of the arcuate sulcus. During each penetration, unit activity was
characterized during visuomotor tasks at intervals of 50–100 µm in
depth, and electrical stimulation was applied every 200–250 µm in
depth. The frontal eye field (FEF) was physiologically defined as the area of
cortex located primarily on the rostral bank as well as the fundus of the
arcuate sulcus where saccades could be electrically evoked with stimulus
thresholds of <50 µA (Bruce and
Goldberg 1985; Bruce et al.
1985). Electrical stimuli consisted of 70-ms trains of biphasic
pulses (–0.2-ms pulse followed by +0.2-ms pulse) applied at a frequency
of 330 Hz. The output of the stimulator was connected to the electrode through
constant-current optical isolators. Maximum current intensity used was 75
µA. Threshold was defined as the current intensity at which saccades were
evoked in response to 50% of the stimulus trains. In these experiments, all
active sites that had FEF-like activity and low thresholds for evoking
saccades were tested for the effects of orthodromic stimulation on the
ipsilateral superior colliculus.
Projections from the FEF to the superior colliculus connect matching sites
within each topographical representation
(Komatsu and Suzuki 1985;
Segraves and Goldberg 1987;
Stanton et al. 1988b). To map
the topography of the superior colliculus, a microelectrode was advanced and
neuron activity was characterized during visuomotor tasks at intervals of
50–100 µm. Visual receptive fields and movement fields were
determined at each depth. These were compared with visual receptive field
(Cynader and Berman 1972) and
motor maps (Robinson 1972) to
determine the location of the electrode within the collicular topography. Once
the relationship of collicular topography to recording grid location was
understood, we positioned a guide tube at a site that corresponded to the
current recording site in the FEF. A comparison of the saccade amplitude for
response fields of pairs of FEF and superior colliculus sites where
antidromically activated cells were isolated yielded a correlation coefficient
of 0.80 (Pearson's product moment).
Antidromic excitation
Interconnected FEF and superior colliculus sites were identified using
antidromic excitation (Fig. 2).
All pairs of interconnected FEF and superior colliculus sites were ipsilateral
to one another. With the cortical electrode positioned at the surface of the
FEF and the collicular electrode positioned at a depth where both visual and
saccaderelated activity were observed, the FEF electrode was advanced in 50-
to 100-µm increments. At each depth, neuronal activity was characterized
during visuomotor tasks to monitor the progress of the penetration through the
cortex, and the superior colliculus was stimulated in an attempt to
antidromically excite FEF neurons. Stimuli used for antidromic excitation were
single biphasic pulses (–0.1-ms pulse followed by +0.1-ms pulse) and had
a maximum negative pulse intensity of 1 mA. The amplitude of the
positive-going component of the biphasic pulse was always less than the
negative component and was adjusted to produce the minimum stimulus artifact.
Based on a review by Ranck
(1975) of several studies of
mammalian CNS stimulation, the passive spread of current from a 1-mA source
could excite cell bodies and myelinated axons at a distance of 
2 mm from
the electrode tip. The criteria for antidromic identification were a constant
response latency to antidromic excitation and the ability to collide
spontaneous and antidromically evoked spikes
(Bishop et al. 1962;
Fuller and Schlag 1976).
Antidromic latency was defined as the time interval between the onset of the
stimulus and the appearance of the antidromic spike. Threshold was defined as
the current intensity at which antidromic spikes were obtained in response to
50% of the stimulus pulses. Once an antidromically excited neuron was
identified, its activity was characterized during visuomotor tasks.
|
Current-source-density analysis
Extracellular field potential and current-source-density analysis were used to examine the vertical distribution of FEF terminations in the superior colliculus. A FEF electrode was positioned at a site where cells projecting to the superior colliculus had been localized by antidromic excitation, and a collicular electrode was positioned with its tip just above the dorsal surface of the superior colliculus. The collicular electrode was then advanced in 100-µm increments. After each advance, an electrical stimulus was applied through the FEF electrode and evoked field potentials were recorded from the collicular electrode. The cortical stimulus was a single biphasic pulse where a–0.1-ms pulse was followed by +0.1-ms pulse. As was the case for the stimuli used to search for antidromically excitable neurons, the maximum amplitude of the negative pulse was 1 mA with the positive pulse adjusted <1 mA to make the stimulus artifact as small as possible.
A current-source-density analysis was used to summarize the activity of
many individual neurons enabling us to assess the global pattern of activity
at a recording site. This analysis helped us to localize the site of dominant
synaptic input within the vertical distribution of field potentials providing
better localization of neuronal events than would be possible with field
potentials alone. Based on established techniques
(Ferster 1990;
Freeman and Nicholson 1975;
Mitzdorf and Singer 1978;
Nicholson and Freeman 1975),
current-source-density was defined as being directly proportional to the
second spatial derivative of the field potential at a point (x, y, z)
 | (1) |
is the resistivity of the medium, V is the voltage, and
I is the conductance.
Two assumptions were made based on the findings of Nicholson and Freeman
(1975). The first was that
conductivity in the extracellular space did not change significantly with
superior colliculus depth. The second was that the spread of conductivity
occurs symmetrically in both the x and the y directions. In
the present experiment, the x and y directions ran parallel
to the plane of the surface of the superior colliculus. Variation only
occurred in the z direction, perpendicular to the superior colliculus
surface. Therefore Eq. 1 was simplified to
 | (2) |
Evoked potentials were measured at finite points. Thus the second
derivative was approximated by a finite-difference equation as described by
Freeman and Nicholson (1975)
 | (3) |
The expression
(V(z+n
· 
z) –
2V(z) +
V(z–n
· z)) could be determined from our experimental
measurements, and the denominator (n ·
z2) were assumed to be constant, therefore
(V(z+n
· z) –
2V(z) +
V(z–n
· z)) was directly proportional to current density.
Previous studies showed that when field potentials were recorded at 50-µm
increments in depth, "n" should be 5 for optimal spacing
to reduce noise and maximize the sinks and sources
(Mitzdorf and Singer 1978). In
our experiments, n was chosen to be 3 because the separation in
recording sites (z) was 100 µm. Therefore
V(z) was the field
potential recorded at depth (z), and
V(z+n
· z) and
V(z–n
· z) were the field potentials recorded at depths 300
µm dorsal and 300 µm ventral to depth (z).
Orthodromic excitation
Orthodromic excitation was used to identify superior colliculus neurons
which received FEF input (Fig.
3). A FEF microelectrode was positioned with its tip within a site
where corticotectal cells had been identified, and another microelectrode was
positioned just above the superior colliculus
(Fig. 3A). The
superior colliculus electrode was advanced in 50- to 100-µm increments
through the entire depth of the superior colliculus, including both
superficial and deep layers. At each depth, the FEF was stimulated in an
attempt to orthodromically excite collicular cells. The stimulus consisted of
one or two biphasic pulses, –0.1-ms pulse followed by +0.1-ms pulse,
applied 1–4 ms after the superior colliculus neuron fired spontaneously
or in response to a brief visual stimulus in its response field
(Fig. 3, B and
C). The twin pulse had a 3-ms separation between pulses.
Orthodromically driven neuronal spikes were recognized by their relatively
fixed response latency following the FEF stimulus. Threshold for orthodromic
excitation was defined as the current intensity at which orthodromic spikes
were obtained in response to 50% of the stimulus pulses.
|
Data analysis
During the experiments, behavioral and neuronal activity was sampled at 1 kHz and was stored in the laboratory computer. After the experiment, the data were transferred to a UNIX-based workstation (Sun Microsystems) for further analysis.
Neuron activity was displayed in rasters and histograms aligned to specific
events occurring within the behavioral paradigm to evaluate the relationship
between neuron activity and those events. Neuronal activity was also plotted
as continuous spike-density functions
(Richmond et al. 1987;
Sanderson and Kobler 1976). To
generate a spike density function for each trial, a Gaussian pulse of fixed
width ( = 10 ms) was substituted for each spike, and then all of the
Gaussians were summed together to produce a continuous function expressing the
expected spike frequency for any point in time during the trial. A Poisson
spike train analysis technique (Hanes et
al. 1995; Legéndy and
Salcman 1985) was used to identify statistically significant
changes in neuronal firing.
The activity of FEF neurons during visuomotor tasks was classified
according to the criteria of Bruce and Goldberg
(1985). The activity of
superior colliculus neurons during visuomotor tasks was classified using
published criteria as a basis (Glimcher
and Sparks 1992; Mays and
Sparks 1980; Mohler and Wurtz
1976; Munoz and Wurtz
1995a; Sparks
1978; Sparks et al.
1976). For both FEF and superior colliculus, visual activity was
defined as a fixed, short-latency response to a visual stimulus regardless of
whether or not the stimulus would become the target for a saccade. The visual
no saccade and memory-guided saccade tasks were the primary tasks used to
identify visual activity. For both regions, saccade-related burst activity was
identified in the memory-guided saccade task because of the advantage it
provides of separating the time when the target stimulus is flashed and the
time when the saccadic eye movement is allowed by 
100 ms. Saccade-related
burst activity was defined as a high-frequency burst of activity peaking near
the beginning of the saccade. For prelude/build-up activity in the superior
colliculus, we relied on the use of the gap task. During the gap task, neurons
with prelude/build-up activity increased their firing rates during the gap
period prior to the appearance of a target light. With target location
alternated randomly between two possible locations, one within the response
field of the cell and one in the opposite hemifield, we required that activity
be present during the gap period regardless of where the target would appear
in order for it to be classified as prelude/build-up.
Histology
On completion of all experiments, monkeys MK04 and MK05 were given the analgesic ketamine hydrochloride (10 mg/kg im). Three to four reference guide tubes were inserted into each recording grid through the dura and into the brain. Each monkey was given a lethal dose of pentobarbital sodium and was perfused transcardially with saline followed by 10% formalin. The brain was removed and photographed. For monkey MK04, the brain was frozen sectioned at 50 µm. The location of tracts left by the reference guide tubes were used to assist in aligning the plane of sectioning parallel to the trajectories of electrode tracts. Sections through the superior colliculus were made in the coronal plane. Sections through the arcuate sulcus were made in the plane running rostrocaudally, parallel to the principal sulcus. These sections helped to confirm that our penetrations were, in fact, through the FEF and superior colliculus. Although a complete histological processing was not performed on monkey MK05, gross observation of the brain's surface confirmed that the FEF penetrations were through the anterior bank of the arcuate sulcus.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Antidromic excitation of FEF neurons by superior colliculus
microstimulation was used to identify 15 pairs of interconnected FEF and
superior colliculus sites in two rhesus monkeys (see
Table 1). In the process of
identifying these pairs of interconnected sites, 22 FEF neurons were
antidromically excited by superior colliculus stimulation and their activity
was characterized with visuomotor tasks. Ninety-one percent (20/22) of the
corticotectal neurons exhibited saccade-related activity during the
memory-guided or overlap tasks. Of these neurons, 30% (6/20) also had visually
related activity. One of the corticotectal neurons had only visually related
activity, and one of the corticotectal neurons responded only to novel visual
stimuli and was unresponsive in standard oculomotor tasks, both of these
neurons were recorded from our most lateral recording sites where saccades
could be evoked with thresholds of <50 µA. At points 0.5 mm lateral to
these sites, the threshold for evoking a saccade exceeded the 50-µA
criteria for defining the FEF. The mean response latency to antidromic
excitation for all corticotectal neurons was 2.00 ± 0.43 ms (range:
1.5–3.0 ms), and the mean threshold for antidromic excitation was 223
± 187 µA (range 10–700 µA;
Table 3). These findings
regarding cell activity type, as well as latency to antidromic excitation and
threshold are consistent with earlier reports
(Everling and Munoz 2000;
Segraves and Goldberg 1987;
Sommer and Wurtz 2000). In
addition to finding the corticotectal cells types reported here, Sommer and
Wurtz (2000) found a
prevalence of corticotectal neurons with exclusively visual activity in more
lateral regions of the FEF. Our failure to isolate a high percentage of FEF
sites where visual corticotectal cells were prevalent suggests that our
stimulation sites were confined to the medial FEF.
|
|
Localization in depth of FEF input to the superior colliculus
Once an antidromically activated site was identified, orthodromic
excitation was used to determine the location in depth within the superior
colliculus that received excitatory FEF input. This was done for monkey
MK04 alone. Our findings were then compared with published results from
anatomical tracing studies. Orthodromically evoked superior colliculus field
potentials were recorded from two interconnected FEF and superior colliculus
sites in this monkey. Field potentials were averaged by calculating the mean
of 50 individual responses and are shown in series dorsoventrally through the
superior colliculus (Fig.
4A). An increase in the amplitude of evoked field
potentials occurred from 1,100–1,700 µm below the level where
visually related background activity was first observed (0 µm). To refine
the broad spatial distribution of field potentials that results from the
linear addition of overlapping potentials, we used a current-source density
analysis that provided a better localization of the site of FEF input
(Fig. 4B). Visual
inspection of this current-source-density analysis showed that maximum current
densities occurred from 1,200–1,500 µm below the level where
visually related background activity was first observed. When the patterns of
field potentials and current-source-density were compared with histograms of
neuronal activity during the memory-guided saccade task at each depth, it was
found that recording sites between 0 and 1,000 µm exhibited primarily
visually related activity, sites located from 1,100 to 1,700 µm exhibited
primarily saccade-related activity and recording sites located between 1,800
and 2,000 µm exhibited no saccade-related neuronal activity.
Saccade-related activity was first observed at a depth of 700 µm, and at
1,100 µm, the peak frequency of saccade-related activity was greater than
the peak frequency of visually related activity. At 1,800 µm, only very
weak saccade-related background modulation was found.
|
Our field potential and current-source-density analysis of these recording
sites, helped us to identify the depths at which FEF input was concentrated
within the superior colliculus. The FEF sites that we sampled did not appear
to project to sites containing purely visual neurons in the superficial layers
of the superior colliculus, instead, they projected selectively to recording
sites containing saccade-related neurons in the deeper layers of the superior
colliculus, the stratum griseum intermediate and profundum. These findings
confirm previous reports from anatomical tracer studies that the FEF neurons
terminate primarily in the deeper layers of the superior colliculus
(Astruc 1971;
Fries 1984;
Huerta et al. 1986;
Komatsu and Suzuki 1985;
Kunzle and Akert 1977;
Kunzle et al. 1976;
Leichnetz et al. 1981; Stanton
et al. 1982,
1988b). Our next step was to
isolate and characterize individual superior colliculus neurons that were
orthodromically excited by FEF stimulation.
Orthodromically excited neurons
PHYSIOLOGICAL CLASSIFICATION. Orthodromic stimulation from all antidromically activated sites was used to identify superior colliculus neurons that received FEF input and the activity of the neurons was characterized with visuomotor tasks. A total of 83 orthodromically excited superior colliculus neurons were isolated and their activity characterized (Table 1). All of these neurons increased their discharge rates in association with saccadic eye movements made in the visually guided saccade task. All of the orthodromically excited superior colliculus neurons also exhibited activity aligned to the beginning of the saccade, in the memory-guided or overlap tasks, with or without activity aligned to the visual stimulus. None had visual activity alone.
For each identified orthodromically excited neuron, a visually guided saccade task was used to identify the neuron's response field and preferred eye movement vector. Once the preferred movement vector was determined, the memory-guided saccade task and overlap task were used to distinguish between visual and saccade-related activity. Next, the gap task was used to distinguish between two types of presaccadic activity: prelude/build-up activity prior to the onset of the saccade and a high-frequency burst of activity associated with the onset of the saccade. The gap task helped us determine the duration of the prelude/build-up of neuronal activity during the delay period between the times when the fixation point was extinguished and the target was illuminated. Target position was randomly alternated between two target locations, corresponding to the endpoint of the neuron's preferred movement vector as well as the end point of a vector of the same amplitude but in the opposite direction.
Burst activity. Most of the superior colliculus cells driven orthodromically from the FEF exhibited saccade-related burst activity during memory-guided and overlap tasks (93%, n = 77). In the gap task, these cells exhibited a high-frequency burst of activity associated with the onset of the saccade in the direction of the preferred movement vector and no significant burst activity for saccades made in the opposite direction.
Prelude/build-up activity. During the gap task, 25% (n = 21) of the orthodromically driven superior colliculus neurons exhibited prelude/build-up activity prior to the onset of the saccade in the direction of the preferred movement vector as well as for saccades in the opposite direction. This activity was often accompanied by a high-frequency burst of activity beginning prior to the onset of the saccade in the direction of the preferred movement vector but not for saccades made in the opposite direction. Seventy-one percent (15/21) of the neurons with prelude/build-up activity exhibited a high-frequency burst of activity for saccades made in the preferred direction.
An example of the activity of an orthodromically excited neuron with prelude/build-up activity followed by a saccade-related burst of activity is shown in Fig. 5. During the gap task, prelude/build-up activity was elicited for saccades in the direction of the preferred movement vector (Fig. 5A) as well as for the opposite direction (Fig. 5B). The peak of the high-frequency burst of activity occurred 84 ms prior to the onset of the saccade made in the preferred movement direction (Fig. 5A). No burst of activity was elicited for saccades made in the opposite direction (Fig. 5B).
|
Visually related activity. In contrast to the frequent occurrence of saccade-related burst as well as prelude/build-up activity in the population of orthodromically driven neurons, none of the orthodromically driven superior colliculus neurons exhibited exclusively visually related activity during the memory-guided or overlap tasks. With each penetration, an attempt was made to identify orthodromically driven neurons within the superficial layers of the superior colliculus. None was found. This was in agreement with our initial field potential and current-source-density findings that orthodromic stimulation produced little effect in the pattern of evoked activity within the superficial layers of the colliculus. FEF driven trans-synaptic potentials were first observed within the transition region between the superficial layers and the deeper layers of the colliculus. Although neurons with visually related activity were found in this transition region, these neurons always had an additional component of their activity that was primarily aligned to the time of saccade onset. Of the orthodromically driven superior colliculus neurons found, 30% (n = 25) exhibited visual activity in combination with prelude/build-up and/or burst activity. Of these neurons, 8% (n = 2) combined visual activity with prelude/build-up activity alone, 44% (n = 11) exhibited visual activity in combination with both prelude/build-up and burst activity, and 48% (n = 12) exhibited visual activity in combination with burst activity alone.
In summary, all orthodromically driven neurons exhibited prelude/build-up and/or burst activity. During the gap task, 25% of the orthodromically driven neurons had prelude/build-up activity while 93% of the orthodromically driven neurons had a high-frequency burst of activity associated with the onset of the saccade. No orthodromically driven neurons had purely visually related activity, although 30% had a combination of visual activity with prelude/build-up or burst activity.
Relationship of activity of orthodromically driven collicular neurons to saccade timing
The saccade-related neuron population within the superior colliculus can be
divided based on the timing of the fall of activity relative to the end point
of the saccade (Waitzman et al.
1991) with neurons the activity of which ends with the end of the
saccade referred to as clipped and neurons the activity of which continues for
a short period beyond the end of the eye movement called unclipped. These two
forms of cell activity have important roles in a number of models of
oculomotor control (for example, Arai et
al. 1994; Quaia et al.
1999; Van Opstal and Kappen
1993; Waitzman et al.
1991).
To determine whether or not the FEF projection to the superior colliculus was selective for one of these two subpopulations of saccade-related neurons, we compared the timing of spike discharges of orthodromically driven superior colliculus neurons to the timing of the beginning and end of the saccades. Calculations were made from plots of mean spike density during the memory-guided saccade task. The baseline activity was calculated as the mean of activity that occurred during the 200-ms period prior to the disappearance of the fixation point. We measured the time of peak activity relative to the onset of the saccade, the decrement of activity during the saccade, and the time of the return of activity to baseline level relative to the end of the saccade.
Sufficient data to perform this analysis were available for 33
orthodromically driven neurons (Fig.
6 and Table 2). We
found that of the neurons with burst activity, 30% had a >90% decrease in
activity prior to the end of the saccade. Of the remaining neurons with burst
activity, the neurons' activity did not return to <90% of peak levels until
70 ms after the end of the saccade. Of the five neurons with
prelude/build-up activity (3 had prelude/build-up activity combined with burst
activity, 2 had prelude/build-up activity alone), the activity of two neurons
ended prior to the end of the saccade, whereas the activity of three neurons
ended 0–100 ms after the end of the saccade. For a comparison of the
decrement of activity during the saccade to the time of peak activity relative
to the onset of the saccade (Fig.
6C), if the time of peak activity occurred prior to the
onset of the saccade the percent decrement of activity during the saccade was
greater than when the time of peak activity occurred after the saccade onset.
For this sample of 33 orthodromically driven neurons, the Pearson's product
moment correlation coefficient was calculated comparing percent decrement of
activity during the saccade to the timing of peak activity relative to the
onset of the saccade and was found to be significant using a t-test
(P < 0.01). Thus the FEF projection to the superior colliculus
does not appear to be selective for cells with saccade-related activity based
on the timing of the fall of activity relative to the end point of the
saccade.
|
|
Orthodromic latencies and thresholds
We compared the orthodromic response latencies and thresholds for the different types of cells activated in our study to reveal similarities or differences in these parameters between the different cell types (Table 3). A single pulse was used to evoke an orthodromic response in 61 superior colliculus neurons and a twin pulse was used to evoke a response in 22 neurons. A t-test was used to determine if there were significant differences in the mean response latencies and thresholds for the single- and twin-pulse groups. If the response latency was calculated from the onset of the second twin pulse, no significant difference was found between response latencies and thresholds evoked by single or twin pulses. Moreover, no significant differences were noted in mean response latencies and thresholds between orthodromically driven neurons with prelude/build-up activity or burst activity. This suggests that neurons with prelude/build-up and burst activity received the same type of excitatory connections from the FEF.
We divided orthodromic latencies into two groups, based on whether latency
was less than or >6 ms. The range (prelude/build-up: 2.8–4.7 ms;
burst: 2.0–5.8 ms) and mean response latencies for orthodromic
excitation of the group with latencies <6 ms were slightly longer than the
range of response latencies reported for antidromic excitation of FEF neurons
by collicular stimulation (range: 0.8–5.0 ms, mean: 2.49 ms,
Everling and Munoz 2000;
range: 1.0–6.0 ms, mean: 2.25 ms,
Segraves and Goldberg 1987).
For the 22 corticotectal neurons isolated in these experiments, antidromic
latencies ranged 1.5–3.0 ms. One would expect the latency of orthodromic
stimulation of a monosynaptic pathway to be 0.5 ms longer than for
antidromic stimulation due to the introduction of a synaptic delay in the
orthodromic direction. Thus it is likely that the group with orthodromic
latencies of <6 ms represents monosynaptic input to saccade-related neurons
in the superior colliculus. For the small number of neurons with mean
orthodromic latencies >6 ms (range: prelude/build-up: 7.0–10.5;
burst: 7.2–8.2), it is likely that these latencies represent activations
of the collicular cells by multisynaptic pathways.
The distribution of neurons with prelude/build-up and burst activity versus the latency of the orthodromically evoked response is shown for responses evoked with a single pulse (Fig. 7A) and twin pulse (Fig. 7B). The response latencies were unimodally distributed, suggesting a homogeneous form of excitatory input to these neurons.
|
The distribution of orthodromic thresholds for neurons with prelude/build-up and burst activity is shown for activity evoked with both single (Fig. 8A) and twin pulses (Fig. 8B). The low threshold for orthodromic excitation and the required precise topographical alignment of FEF stimulation site with collicular recording site (see METHODS) argue strongly for the view that our FEF stimulation excited a direct monosynaptic projection to the colliculus.
|
Location of orthodromically driven neurons within superior colliculus
The location in depth within the superior colliculus of 75 of the 83 orthodromically driven neurons is plotted (Fig. 9). The depth of each neuron is measured relative to the depth at which multiunit visual activity was first encountered when entering the superior colliculus. The radius plotted on the x axis for each neuron is equal to the amplitude of the neuron's preferred movement vector. The amplitude of the movement vector did not appear to influence the depth at which orthodromically driven cells were found. Overall, neurons with prelude/build-up activity were found at a slightly greater average depth than neurons with burst activity. In monkey MK04, at a depth between 2.0 and 2.5 mm below the onset of visually related activity, three of the six orthodromically driven burst neurons that were found exhibited a high, spontaneous discharge rate, a high-frequency burst of activity associated with the onset of the saccade in the direction of the preferred movement vector, a cessation of activity during the saccade in the opposite direction, and no prelude/build-up activity. In the course of electrode penetrations made through the superior colliculus, this type of neuron was consistently found below neurons with prelude/build-up activity.
|
Preferred movement vectors of orthodromically driven neurons
The optimal saccade amplitudes and directions for the orthodromically
driven superior colliculus neuron population are shown in
Fig. 10. For monkey
MK04, the amplitude of the preferred movement vector for orthodromically
driven neurons with prelude/build-up activity
(Fig. 10A), or burst
activity (Fig. 10B),
ranged from 3 to 22° in the right hemifield and from 1 to 23° in the
left hemifield. For the sample of neurons from monkey MK05, the
amplitude of the preferred movement vector for neurons with prelude/build-up
activity (Fig. 10C)
and burst activity (Fig.
10D) ranged from 4 to 19° in the left hemifield. For
the single cell with burst activity whose movement field was in the right
hemifield, the amplitude of the preferred movement vector was 8° and angle
was 4°. This range of amplitudes occupies approximately the rostral half
of the superior colliculus motor map
(Robinson 1972). The
limitation in this range was primarily a function of FEF recording sites and
not the topography of the SC. In these experiments, we first found an FEF
site, then looked for the collicular site with corresponding topography.
Response field centers for all of our FEF antidromically activated neurons
ranged from 3 to 24°. Because FEF sites with activity field centers
>20° were not used as stimulation sites in our experiments, it is not
surprising that the range of amplitudes for orthodromically driven collicular
neurons did not extend beyond 20°.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
;
Segraves and Goldberg 1987;
Sommer and Wurtz 2000) and
parietal cortex (Pare and Wurtz
1997,
2001), connections between
posterior parietal cortex and FEF
(Ferraina et al. 2002) as well
as studies of multisynaptic pathways from the superior colliculus to FEF via a
thalamic waypoint (Sommer and Wurtz
1998,
2002). There are several
well-known classes of neurons within the layers of the superior colliculus
where FEF projections are known to terminate. In addition, each neuronal class
may fulfill multiple functions in the preparation of movements of the eyes as
well as the head. By identifying the classes of neurons that are the target of
the major FEF input to the colliculus, the present findings further our
understanding of the functional organization of this complex network at both
the cortical and subcortical level. We found that the excitatory effects of
FEF stimulation was not present in the region containing purely visual neurons
in the superficial layers of the superior colliculus. Instead, the FEF appears
to provide direct excitatory input to saccade-related neurons located within
the deep layers of the superior colliculus, including direct inputs to neurons
with prelude/build-up activity as well as those with saccade-related bursts of
activity. Are FEF projections to the colliculus directed to a subset of the neuron population?
The activities of neurons within the deep layers of the superior colliculus
have been divided into three basic functional types. These include fixation,
prelude/build-up, and burst activity. A number of recent reports and reviews
have described these activities in detail, and they are summarized briefly
here. Fixation activity is characterized by increased activity during fixation
followed by a pause in activity before the onset of a saccade and a resumption
of firing at the end of the saccade (Munoz
and Wurtz 1993a). Prelude/build-up activity consists of an
increase of activity beginning several hundreds of milliseconds before the
beginning of a saccade (Glimcher and
Sparks 1992; Mays and Sparks
1980; Munoz and Wurtz
1995a). It appears to be related to the anticipation of or
preparation for a saccade toward the cell's movement field. The level of
prelude/build-up activity is related to the likelihood that a saccade will be
made into the cell's movement field (Basso
and Wurtz 1998; Horwitz and Newsome
1999,
2001; Ratcliff et al. 2003),
and increased levels of prelude/build-up activity appear to facilitate the
initiation of a saccade, decreasing saccade latency
(Dorris and Munoz 1998;
Dorris et al. 1997). Besides
representing activity characteristic of the preparatory stages of generating a
saccade, it has been proposed that prelude/build-up activity is responsible
for the drive required to generate the saccade-related burst
activity—the third major form of activity seen within the
saccade-related layers of the superior colliculus
(Optican 1994). Burst activity
rises quickly to a peak of several hundred spikes per second at the start of
the saccade (Sparks 1978).
This activity falls rapidly during the eye movement. Burst activity that
reaches baseline activity by the end of the saccade has been referred to as
clipped activity by Waitzman and colleagues
(Waitzman et al. 1991), and
cells whose bursts end shortly after the end of the saccade have been called
unclipped. Both prelude/build-up and burst activities are tuned to saccades
within a restricted field of amplitudes and directions.
We found that saccade-related superior colliculus neurons that had
prelude/build-up activity and/or a burst of activity prior to the onset of the
saccade received excitatory FEF input. To determine whether there was a bias
in the types of neurons receiving FEF input, we compared the distribution of
activity types in the neuronal population receiving direct FEF input with the
distribution of activity types in the overall population
(Munoz and Wurtz 1995a) and
found the distribution of prelude/build-up and burst activities to be
essentially the same in the two populations
(Fig. 11). In addition, we
compared the relationship of the activity of orthodromically driven collicular
neurons to saccade timing. Our sample of neurons receiving direct FEF input
included neurons the activity of which fell to near baseline by the end of the
saccade as well as neurons whose activity did not reach baseline levels until
after the saccade had ended.
