Biological
Symposium Abstracts
size ¼ 56, voxel size 2x2x2). Those with vascular depression
showed a pattern of increased local hippocampal connectivity
(at corrected p < 0.05).
Conclusions: Individuals with a vascular depression show
pattern of greater hippocampal-MTL local connectivity as has
previously been found among individuals at high risk for
AD. Hyperconnectivity has been suggested as a marker of
excitotoxicity. Thus, these results suggest that one pathway
by which LLD contributes to dementia risk is through increased
medial temporal lobe connectivity and associated
excitotoxicity.
Supported By: RF1 AG025516; R01 MH076079
Keywords: Neurodegeneration, Geriatric Depression, Late-
Life Depression
SYMPOSIUM
Brain Initiative: New Tools for Neurostimulation
Chair: David McMullen
Co-Chair: Wayne Goodman
Acoustically Targeted Chemogenetics for Noninvasive
Control of Neural Circuits
Jerzy Szablowski1, Audrey Lee-Gosselin2, Brian Lue2,
Dina Malounda2, and Mikhail Shapiro2
1 points, and propose a mechanism for how network control
Rice University, 2Caltech
Background: Existing treatments for brain disorders aim to
modulate the activity of neural circuits, but are either not cell
type-specific, lack spatial targeting, or require invasive pro-
cedures. Previously, we introduced an approach to modulating
neural circuits noninvasively with spatial, cell-type, and tem-
poral specificity called acoustically targeted chemogenetics, or
ATAC. We use ultrasound to open the blood brain barrier to
transduce neurons at specific locations in the brain with virally-
encoded chemogenetic receptors. To show neuronal inhibition
expressed inhibitory DREADD (hM4Di) throughout the hippo-
campus and tested mice in a fear conditioning protocol. The
context fear test showed that mice treated with saline froze
significantly more than those who had received CNO (p<2E-5,
n¼7,11) during the training phase.
Methods: We used a 1.5 MHz annular array transducer (8
elements, 20mm focal length, f¼0.8) to open the BBB in mice.
We administered Definity microbubbles (3.2E7 per gram of
body weight) and AAV9 virus injected IV.
Results: In this pilot study we have identified 5 mutated AAV
strains that improve transduction efficiency following ultrasonic
BBB opening while reducing transduction of peripheral tissues.
Conclusions: ATAC builds on progress in the field to achieve
a new paradigm of fully noninvasive, spatially, cell-type and
temporally specific neuromodulation, and is applicable to use
in animals of all sizes. Here we report improvement in the gene
delivery efficiency which will aid use of ATAC and noninvasive
gene delivery in neuroscience applications.
Supported By: NARSAD, Jacobs Institute for Molecular
Engineering for Medicine, DARPA
Biological Psychiatry May 1, 2020; 87:S1eS105 www.sobp.org/journal
Psychiatry
Keywords: CHEMOGENETICS, Ultrasound, Animal Model,
AAV, Gene Delivery
Perturbation and Control for Human Brain Network
Dynamics
Danielle Bassett1
1
University of Pennsylvania
Background: The human brain is a complex organ charac-
terized by heterogeneous patterns of interconnections. New
non-invasive imaging techniques now allow for these patterns
to be carefully and comprehensively mapped in individual
humans, paving the way for a better understanding of how
wiring supports our thought processes. While a large body of
work now focuses on descriptive statistics to characterize
these wiring patterns, a critical open question lies in how the
organization of these networks constrains the potential
repertoire of brain dynamics.
Methods: In this talk, I will describe an approach for under-
standing how perturbations to brain dynamics propagate
through complex writing patterns, driving the brain into new
states of activity.
Results: Drawing on a range of disciplinary tools e from
graph theory to network control theory and optimization e I will
identify control points in brain networks, characterize trajec-
tories of brain activity states following perturbation to those
evolves in our brains as we grow from children into adults.
Conclusions: Finally, I will describe how these computa-
tional tools and approaches can be used to better understand
the brain’s intrinsic control mechanisms and to inform stimu-
lation devices to control abnormal brain dynamics, for example
in patients with medically refractory epilepsy.
Supported By: R01 from National Institute of Neurological
Disorders and Stroke
Keywords: Network Control Theory, Electrocorticography,
Diffusion Imaging Data, Stimulation, Theoretical Neuroscience
Neural Decoding and Control of Mood to Treat
Neuropsychiatric Disorders
Omid Sani1, Yuxiao Yang1, Morgan Lee1, Kristin Sellers2,
Heather Dawes2, Edward Chang2, and Maryam Shanechi1
1
University of Southern California, 2University of California
San Francisco
Background: Designing systems for closed-loop control of
neural activity with electrical stimulation input is important for
devising precisely tailored therapies for neuropsychiatric dis-
orders. This design requires novel decoding methods to track
mood states in real time based on distributed neural re-
cordings, which is a challenging problem. Further, it requires
new models to describe the effect of stimulation on neural
recordings from an individual, which are lacking.
Methods: We develop a novel dynamical modelling frame-
work for distributed activity to achieve mood decoding and
S95
Biological
Psychiatry
model how neural network activity responds to electrical
stimulation. To estimate these models, we stimulate the brain
with a novel stochastic electrical stimulation waveform whose
frequency and amplitude randomly switch between discrete
levels in time.
Results: First, we recorded multi-day multisite intracranial
EEG (iEEG) activity and mood self-reports from 7 human epi-
lepsy subject. We could successfully decode mood state
variations in every individual (P < 0.05 in every patient).
The average correlation coefficient between the decoded and
true IMS was 0.75 (P < 0.0001). The framework also identified
the limbic regions and in particular orbitofrontal cortex critical
for decoding. Second, by stimulating the brain with the new
stochastic waveform, we estimated models for the effect of
electrical stimulation in 10 epilepsy patients. The models
accurately predicted the neural response; the correlation co-
efficient between the predicted and true iEEG response was
0.56 (P<0.00001). Finally, we successfully combined the
decoder and the stimulation model for closed-loop control in
stimulation hardware.
Conclusions: These results could facilitate future closed-
loop personalized electrical stimulation therapies for neuro-
psychiatric disorders.
Supported By: DARPA SUBNETS
Keywords: Neuromodulation, Deep Brain Stimulation,
Computational Neuroscience, Depression, Anxiety
Controlling Brain Networks Through Oscillatory
Synchrony
Alik Widge1, Nicole Provenza2, Meng-chen Lo1,
Ethan Blackwood3, Mark Schatza1, Sarah Olsen1,
Ishita Basu4, Mustafa Taha Bilge4, Darin Dougherty4, and
David Borton2
1 Harvard Medical School
University of Minnesota, 2Brown University, 3Massachu-
setts General Hospital, 4Harvard Medical School/
Massachusetts General Hospital
Background: Psychiatric disorders arise from dysfunctional
communication in distributed brain networks. We do not know,
however, what "dysfunctional communication" means at a
physiologic level, or how to reliably restore healthy communi-
cation patterns. One dominant theory argues for communication
through synchrony: that when large populations of neurons fire
with reliable (coherent) timing relative to each other, they are
more likely to propagate action potentials (information).
Methods: I will present three linked electrophysiologic
studies. First, in humans with epilepsy (n¼17), we applied a
neural decoding approach to identify network synchrony cor-
relates of effortful cognitive control. Second, in Long-Evans
rats (n¼6), we applied oscillation-locked stimulation to reliably
increase of decrease synchrony between brain regions. Third,
returning to humans, we applied multi-site desynchronizing
stimulation to cortex and striatum of patients with severe OCD
(n¼1; n¼2 anticipated by May) to break cortico-striatal hyper-
connectivity.
Results: Synchrony within a broad fronto-temporo-striatal
network reliably decoded cognitive control (p¼0.0026), but
S96 Biological Psychiatry May 1, 2020; 87:S1eS105 www.sobp.org/journal
Symposium Abstracts
power in the same bands and regions did not (all p>0.05).
Oscillation-locked stimulation reliably increased (Z>1.5) or
decreased (Z<3) inter-regional synchrony in rats depending on
the protocol used. Further, targeted disruption of cortico-
striatal synchrony produced symptom relief (5 YBOCS points)
in an otherwise-refractory OCD patient.
Conclusions: Brain network communication is physiologi-
cally linked to synchronized, oscillatory neural activity across
regions. Effective cognitive function can be decoded from
synchrony, and rationally designed neurostimulation can
change it. Synchrony control may be a clinically viable
approach to controlling brain networks.
Supported By: DARPA, NIMH, NINDS, Harvard Bipolar
Disorder Research Fund
Keywords: Brain Stimulation, Neural Oscillations, Coher-
ence, Synchronization, Brain Networks
SYMPOSIUM
Spontaneous Gamma Activity in Schizophrenia:
New Findings and Clues to Mechanisms
Chair: Kevin Spencer
Co-Chair: James McNally
Spontaneous Gamma Activity Indexes Synaptic E/I
Imbalance and Local Circuit Connectivity in
Schizophrenia
Kevin Spencer1, Yoji Hirano2, Naoya Oribe2,
Shogo Hirano2, Wadim Vodovozov3, and Robert McCarley4
1
VA Boston Healthcare System, 2Kyushu University, 3VA
Boston Healthcare System/Harvard Medical School,
4
Background: Spontaneous gamma (30-100 Hz) activity (SG)
in the EEG may reflect the balance of cortical excitation and
inhibition (E/I balance), which is disrupted in schizophrenia. I
will review new work from our laboratory concerning SG and its
implications for understanding cortical circuit abnormalities in
schizophrenia.
Methods: Scalp EEG recordings in healthy controls (HC) and
individuals with chronic and first episode schizophrenia (CSZ/
FESZ; N’s ¼> 18) utilized dense electrode arrays. Independent
component analysis, dipole source localization, and the Lap-
lacian transform were used to attenuate artifacts and increase
spatial resolution. 3T MRI was used to measure cortical gray
matter volume.
Results: SG power in auditory cortex was increased in
FESZ during auditory steady-state stimulation (p<.01) and
did not change over a 1-year period. Increased SG in CSZ
was also found during the performance of auditory but
not visual oddball tasks (p<.05), although SG during auditory
oddball performance increased with age in HC (p<.01)
but not CSZ. In auditory cortex, SG power and cortical
volume were inversely correlated in CSZ (p<.05). However,
coupling of SG power to low frequency phase was unaffected
in CSZ.