Establishment of an animal model of
unilateral spatial neglect with macaque
monkeys: Quantitative behavioral analysis
and functional imaging
Tsujimoto, Kengo
Doctor of Philosophy
Department of Physiological Sciences
School of Life Science
SOKENDAI (The Graduate University for
Advanced Studies)
Summary
Establishment of an animal model of unilateral spatial neglect
with macaque monkeys:
Quantitative behavioral analysis and functional imaging
Tsujimoto, Kengo
SOKENDAI (The Graduate University for Advanced Studies)
School of Life Science
Department of Physiological Sciences
Introduction
Unilateral spatial neglect (USN) is a characteristic failure to explore the side of
space contralateral to a brain lesion, which cannot be explained by primary sensory or
motor disorders. The neural mechanisms of USN involve the dorsal attention network
(DAN) and the ventral attention network (VAN). The most influential theory in recent
years proposed that USN is caused by damage to VAN. The theory also hypothesized
the neural mechanisms of USN as follows. The damage to VAN reduces the functional
connectivity of the ipsi-lesional DAN and enhances the functional connectivity of the
contra-lesional DAN. This imbalance causes the symptoms of USN. However, this
hypothesis has not been experimentally validated. The purpose of this study is (1) to
establish a monkey model of USN by testing monkeys with behavioral tasks and (2) to
elucidate the neural mechanisms of USN in the monkeys using functional magnetic
resonance imaging (fMRI) techniques. Previous anatomical and imaging studies suggest
that the homologous region of VAN in humans includes the superior temporal gyrus
(STG) in monkeys. Based on this knowledge, I made a surgical lesion in the right STG
of four monkeys and investigated the effects of the lesion using behavioral tasks and
functional imaging techniques.
Methods
Animals
Four Japanese macaque monkeys (Macaca fuscata) were used for behavioral tests
(Monkey A: 5.5kg, female, Monkey B: 8.2kg, male, Monkey C: 7.5kg, male, Monkey
D: 5.2kg, female). Structural MR images were acquired from all the monkeys before the
lesion to make sure that they had no structural abnormality in their brain.
Test for spatial neglect 1: food-choice task
The food-choice task was used to test Monkey A, Monkey B and Monkey C in their
cage. The monkeys were presented with six wells (three on their right side and three on
their left side) in front of their cage. A piece of apple was hidden in the two out of six
wells under a cover with a grating pattern. These ‘target’ wells were always presented
one on their right side and another on their left side. Other four wells are empty and
with homogeneous gray covers. The positions were randomly changed every trial.
Before starting the task, the front side of the cage was covered with an opaque black
plastic board to avoid the monkeys to see the target positions beforehand. Once the
targets were set, the cover was removed to start the task. If the monkeys picked up the
reward in both target wells, the next trial was started. If the monkeys failed to pick up
the rewards within 30 sec, the trial was aborted and the next trial was started. In each
daily session, the monkeys were tested with at least 9 trials (range: 9-36 trials).
Test for spatial neglect 2: target-selection task
The target-selection task was used to test Monkeys B, Monkey C and Monkey D on
the monkey-chair in a head-free condition. The visual stimuli were displayed on a 19-
inch LCD monitor with a touch panel (ET1989L, Tyco Electronics, Pennsylvania). The
distance from the monkey to the display was fixed for each monkey (around 30-40 cm).
To evaluate performance for their right hand and their left hand separately, the opposite
hand was gently restrained to permit use of only one hand during the task. The visual
stimuli consisted of one target and nine distractors. In this example, the target was a red
circle and the distractors were nine gray circles. Each task started from appearance of a
start cue (a red circle) on the center of the display. If the monkeys touched the start cue
within 2 sec, one target and nine distractors were presented on the display. If the
monkeys touched the target within 2 sec, the monkey got a juice reward (0.6 ml / trial).
The monkeys were allowed to touch the screen more than once as long as it was before
the time limit (2 sec). To examine whether the symptom depends on the visual feature of
stimuli, four kinds of stimulus configurations, where the target was defined based on the
difference in color/luminance, shape, orientation or motion, were used to test the
monkeys. To test object-centered neglect (see discussions in Chapter 4), two kinds of
targets were used in the task in which the target was defined by shape. In this condition,
the target was a C-shape but the gap was either on the right side or on the left side.
Test for spatial neglect 3: free-viewing task
The free-viewing task was used to test Monkeys B, Monkey C and Monkey D on
the monkey-chair in a head-free condition. The monkeys freely viewed images on the
display. To reinforce the monkeys to view the display, a drop of juice reward was
applied randomly, without any contingency with the visual stimuli. Gaze positions on
the display and positions of eyes in the head in the world coordinates were measured by
an eye tracker (TX300, Tobii) with a sampling frequency of 300 Hz. The distance from
the monkey to the display was set at 65 cm. One trial consisted of 8 calibration images
and 8 test images. Fifty-six natural images and the horizontally flipped ones were used
as test images and were randomly presented to the monkeys. In order to obtain accurate
measurement of gaze positions, an image of monkeys placed on the corner was used for
calibration between gaze positions on the display and measured gaze positions.
MRI scans
MRI scan was performed using a 3T Allegra scanner (Siemens). A four-channel
receive-only primate head coil with volume transmit coil was used for all the
experiments. The functional images were obtained using single-shot T2*-weighted
gradient echo echo-planar imaging (EPI) sequence. EPIs comprised axial slices
covering the almost entire brain (28 slices, 64 x 64 in-plane matrix, TR/TE = 2000/30,
voxel size 1.25 x 1.25 x 1.60 mm3). For each subject, four scanning runs of 10 min’
duration (315 volumes for each run) were performed for resting state under anesthesia.
Furthermore, whole brain high-resolution 3D T1-weighted anatomical image were
obtained by using MPRAGE (magnetization prepared rapid acquisition gradient echo)
sequence (TR/TE/TI = 2500/4.38/1100 ms, 192 x 192 x 128 matrix, voxel size 0.5 x 0.5
x 0.5 mm3). T2-weighted turbo spin echo images with the same coverage of EPIs were
also acquired so that EPIs could be superimposed on them. MR sessions were
conducted 1 week before the lesion and 1, 2, 3, 4, 8 and 12 weeks after the lesion.
Results
Test for spatial neglect 1: food-choice task
Three monkeys (Monkey A, Monkey B and Monkey C) were tested with the food-
choice task. Percentage of correct choices was plotted across weeks before and after the
lesion in three monkeys. The percentage of correct choices in the ipsi-lesional side was
almost 100 percent after the lesion. On the other hand, the percentage of the correct
choices in the contra-lesional side was significantly decreased for two days after the
lesion in Monkey A and Monkey C (p < 0.05, Fisher's exact test with Bonferroni
correction).
Test for spatial neglect 2: target-selection task
Three monkeys (Monkey B, Monkey C and Monkey D) were tested with the target-
choice task. The percentage of correct choices was plotted across weeks before and after
the lesion in three monkeys. While the percentage of correct choices for the targets in
the ipsi-lesional side was relatively unchanged, the percentage of correct choices for the
targets in the contra-lesional side was markedly decreased after the lesion. The
percentage of correct choices for the targets in the contra-lesional side was significantly
lower than that for the targets in the contra-lesional side for two weeks in Monkey B,
for four weeks in Monkey C and for three weeks in Monkey D (p < 0.05, Fisher's exact
test with Bonferroni correction).
The mean reaction time was plotted across weeks before and after the lesion in
three monkeys. The data from four different kinds of stimuli (with the target defined in
term of color/luminance, shape, orientation and motion) were separately plotted.
Overall, the mean reaction time for the contra-lesional side was longer than that for the
ipsi-lesional side for more than two or three months after the lesion (p < 0.05, Student’s
t-test with Bonferroni correction).
Test for spatial neglect 3: free-viewing task
Three monkeys (Monkey B, Monkey C and Monkey D) were tested with the free-
viewing task. One week before the lesion, the gaze positions (magenta dots) were
evenly distributed in the contra-lesional and ipsi-lesional sides. On the other hand, one
week after the lesion the gaze positions were strongly biased toward the ipsi-lesional
side on the test images. To demonstrate the ipsi-lesional bias, horizontal gaze positions
recorded from all 112 test images were classified into six categories and plotted as
histograms. The horizontal gaze positions were not biased one week before the lesion.
On the other hand, the gaze positions were strongly biased toward the ipsi-lesional side
one week after the lesion.
Resting-state fMRI
To identify DAN, a seed was set on the center of the right FEF before the lesion and
calculated the functional connectivities in whole brain areas. Voxels with a higher
correlation value include contralateral FEF, bilateral STG, bilateral LIP, the medial
prefrontal cortex and the medial parietal cortex. One week after the lesion, the
correlation value in the right LIP was reduced while the correlation value in the left FEF
and the left LIP looked similar or even increased. Four weeks after the lesion, the
correlation values in the right LIP were recovered to the level of pre-lesion values.
To quantify the time course of functional connectivities in DAN, the seed was set
on the center of the right or left FEF. Then the peak correlation value in the ROI for the
right or left LIP was calculated for each session. Similarly, the peak correlation value in
the ROI for the right or left FEF was calculated when the seed was set on the center of
the right or left LIP. Then these values were averaged to plot across weeks before and
after the lesion. The functional connectivity between the ipsi-lesional FEF and the ipsi-
lesional LIP was significantly decreased for 3, 12, 3 weeks after the lesion in Monkey
B, Monkey C and Monkey D, respectively (p < 0.05, t-test after Fisher z-transformation
with Bonferroni correction). On the other hand, the functional connectivity between the
contra-lesional (left) FEF and the contra-lesional LIP was significantly increased for 1
week after the lesion in three monkeys (p < 0.05, t-test after Fisher z-transformation
with Bonferroni correction). The functional connectivity between the ipsi-lesional FEF
and LIP was significantly lower than that between the contra-lesional FEF and LIP for
2, 3, 2 weeks after the lesion in Monkey B, Monkey C and Monkey D, respectively (p <
0.05, t-test after Fisher z-transformation with Bonferroni correction).
The same analyses were repeated for inter-hemispheric interactions between FEF and
LIP. The functional connectivity between the ipsi-lesional FEF and the contra-lesional
FEF was significantly reduced for 2-3 months after the lesion in three monkeys (p <
0.05, t-test after Fisher z-transformation with Bonferroni correction).
Discussion
This study had two purposes: 1) to establish a monkey model of USN and 2) to
clarify neural mechanisms of USN. For these purposes, the right STG was ablated in
four monkeys and I tested whether the monkeys showed similar symptoms observed in
human USN patients and tested whether the functional connectivity of DAN was
affected in the monkeys after the lesion. Analysis of behavioral results suggests that the
monkeys with the right STG lesion showed similar behavioral deficits to those of
human USN patients. These results suggest that a monkey model of USN was
established in this study. Analysis of MRI results suggests that the acute stage of the
lesion can be characterized by an imbalance between the connectivity within the ipsi-
lesional DAN and the connectivity within the contra-lesional DAN. On the other hand,
the chronic stage of the lesion can be characterized by reduced inter-hemispheric
interaction between the ipsi-lesional DAN and the contra-lesional DAN. These results
suggest that neural mechanisms similar to those hypothesized to explain USN in human
patients were demonstrated in the current study.
Taken together, the current study established a monkey model of USN and
succeeded in elucidating the neural mechanisms of USN in the monkeys. The current
study will contribute to develop new rehabilitation strategies by incorporating the idea
that different networks are involved in different recovery stages of USN.
