- Email: [email protected]
Frequency dependent hub role of the dorsal and ventral right anterior insula
Accepted Manuscript Frequency dependent hub role of the dorsal and ventral right anterior insula Yifeng Wang, Lixia Zhu, Qijun Zou, Qian Cui, Wei Liao, Xujun Duan, Bharat Biswal, Huafu Chen PII:
S1053-8119(17)30817-0
DOI:
10.1016/j.neuroimage.2017.10.004
Reference:
YNIMG 14382
To appear in:
NeuroImage
Received Date: 29 June 2017 Accepted Date: 2 October 2017
Please cite this article as: Wang, Y., Zhu, L., Zou, Q., Cui, Q., Liao, W., Duan, X., Biswal, B., Chen, H., Frequency dependent hub role of the dorsal and ventral right anterior insula, NeuroImage (2017), doi: 10.1016/j.neuroimage.2017.10.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
Running title: Sub-hubs of the rAI
Frequency dependent hub role of the dorsal and ventral right
RI PT
anterior insula
M AN U
Duan1, Bharat Biswal3,1, Huafu Chen1*.
SC
Yifeng Wang1, Lixia Zhu1, Qijun Zou1, Qian Cui2, Wei Liao1, Xujun
1 Key laboratory for NeuroInformation of Ministry of Education, School of Life Science and Technology and Center for Information in BioMedicine, University of Electronic Science and Technology of China,
TE D
Chengdu, 610054, China.
2 School of Political Science and Public Administration, University of Electronic Science and Technology of China, Chengdu 610054, China.
EP
3 Department of Biomedical Engineering, New Jersey Institute of
AC C
Technology, 607 Fenster Hall, University Height, Newark, NJ 07102, USA
* Correspondence: Huafu Chen Tele: +86-138-0800-3171; Fax: +86-28-83208238 Email: [email protected]. Or Yifeng Wang Email: [email protected] 1
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
Abstract
2
The right anterior insula (rAI) plays a crucial role in generating adaptive
3
behavior by orchestrating multiple brain networks. Based on functional
4
separation findings of the insula and spectral fingerprints theory of
5
cognitive functions, we hypothesize that the hub role of the rAI is region
6
and frequency dependent. Using the Human Connectome Project dataset
7
and backtracking approach, we segregate the rAI into dorsal and ventral
8
parts at frequency bands from slow 6 to slow 3, indicating the frequency
9
dependent functional separation of the rAI. Functional connectivity
10
analysis shows that, within lower than 0.198 Hz frequency range, the
11
dorsal and ventral parts of rAI form a complementary system to
12
synchronize with externally and internally-oriented networks. Moreover,
13
the relationship between the dorsal and ventral rAIs predicts the
14
relationship between anti-correlated networks associated with the dorsal
15
rAI at slow 6 and slow 5, suggesting a frequency dependent regulation of
16
the rAI to brain networks. These findings could improve our
17
understanding of the rAI by supporting the region and frequency
18
dependent function of rAI and its essential role in coordinating brain
19
systems relevant to internal and external environments.
20
Key words: anterior insula; frequency-specificity; functional separation;
21
hub; Human Connectome Project.
AC C
EP
TE D
M AN U
SC
RI PT
1
22 2
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 23
Introduction The human brain operates as a highly organized multi-scale system
25
consisting of multiple structural and functional networks (Bullmore and
26
Sporns, 2009; Chan et al., 2016). The execution of normal brain function
27
requires orchestration among networks such as the balance between
28
task-positive and task-negative networks (Fox et al., 2005). Inter-network
29
cooperation depends on the modulation of particular brain regions which
30
usually serve as functional hubs (Cole et al., 2013; Liu et al., 2015). The
31
right anterior insula (rAI) has been thought to be an important hub in
32
coordinating brain networks to produce adaptive behavior in various
33
cognitive processes (Chen et al., 2013; Sridharan et al., 2008).
34
Specifically, it has been confirmed that the rAI modulates the default
35
mode
36
attention-demanding tasks (Sridharan et al., 2008; Wen et al., 2013).
(DMN)
and
cognitive
control
network
in
EP
network
TE D
M AN U
SC
RI PT
24
Recent studies have suggested that the rAI could be further divided
38
into dorsal and ventral functional sub-parts (Chang et al., 2013; Deen et
39
al., 2011). The right dorsal AI (rdAI) is a core region of the salience
40
network (SN) and is connected to cognitive control networks. Typically, it
41
is activated in almost all tasks involving goal-directed cognition (Chang
42
et al., 2013; Nomi et al., 2016). By contrast, the right ventral AI (rvAI) is
43
rich in Von Economo neurons and has been demonstrated to be involved
44
in interoception, emotion, and social awareness via connections with the
AC C
37
3
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
limbic system (Evrard et al., 2012; Uddin, 2015). Similar to the limbic
46
system, the DMN is primarily involved in internal thoughts, such as
47
self-reference processing, episodic memory, planning and so on
48
(Andrews-Hanna et al., 2010; Mason et al., 2007; Raichle, 2015),
49
indicating that the DMN is more likely to be connected with the rvAI.
50
Therefore, we surmised that the rdAI and rvAI may be prone to connect
51
with externally- and internally-oriented systems, respectively (Menon and
52
Uddin, 2010; Zabelina and Andrews-Hanna, 2016).
SC
Besides
M AN U
53
RI PT
45
the
functional
separation
hypothesis,
the
spectral
fingerprints hypothesis (Maris et al., 2016; Siegel et al., 2012) suggests
55
that disparate cognitive processes are implemented by different brain
56
networks embedded in specific frequency bands. For instance, neural
57
oscillations in beta and gamma bands have been found to play distinctive
58
roles in top-down and bottom-up information streams (Bastos et al.,
59
2015). The externally-oriented and internally-oriented systems have also
60
been reported to be dominated by delta-theta and alpha-beta bands,
61
respectively (Lakatos et al., 2016). The spectral fingerprints of these two
62
systems in the low frequency blood oxygen level dependent (BOLD)
63
signal fluctuations is undetermined yet, although frequency-specific
64
functions have been widely observed in this frequency range in cognitive
65
processes and clinical cases (Martino et al., 2016; Palva and Palva, 2012;
66
Wang et al., 2016).
AC C
EP
TE D
54
4
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
In the current study, we investigated whether the hub role of the rAI
68
is region and frequency dependent. Specifically, we used the Human
69
Connectome Project (HCP) 100 unrelated dataset to explore whether the
70
externally- and internally-oriented systems are connected to different
71
parts of the rAI and whether these connections are frequency dependent.
72
In reverse, whether the sub-parts of rAI are connected to other systems
73
other than the conventional externally- and internally-oriented systems
74
and whether these connections are frequency dependent.
75
Methods
76
Data acquisition
M AN U
SC
RI PT
67
The resting-state functional magnetic resonance imaging (rfMRI)
78
data of 82 subjects were used in this study. The sample was selected from
79
the HCP 100 unrelated dataset (https://db.humanconnectome.org) with
80
the criteria that respiratory, cardiac, and head movement data are
81
available
82
rfMRI_REST2_LR, and rfMRI_REST2_RL). REST1 and REST2 were
83
acquired on two different days. The HCP scanning protocol was approved
84
by the local Institutional Review Board at Washington University in St.
85
Louis. Informed consent was obtained from all subjects. All participants
86
were scanned on a customized Siemens 3-T connectome-Skyra scanner.
87
The imaging parameters used to collect the rfMRI data were as follows:
88
TR = 720 ms; TE = 33.1 ms; flip angle = 52°; 2 mm isotropic voxels
EP
TE D
77
all
runs
(rfMRI_REST1_LR,
rfMRI_REST1_RL,
AC C
in
5
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
(FOV = 208 × 180 mm; 72 slices); multiband factor = 8; echo spacing =
90
0.58 ms; bandwidth (BW) = 2290 Hz/px; volumes = 1200. Full details on
91
the HCP dataset could be seen in Van Essen et al.’s article (Van Essen et
92
al., 2013).
93
Data preprocessing
RI PT
89
The data with minimal preprocessing pipeline (Glasser et al., 2013)
95
was used. This pipeline includes artifact removal, motion correction and
96
registration to standard space. Standard preprocessing procedure was
97
further applied to the data according to Finn et al.’s study (Finn et al.,
98
2015) which included removal of the linear trend, removal of linear
99
components related to the six motion parameters and their first
100
derivatives, regression of respiratory and cardiac noises, the mean time
101
courses of white matter and cerebrospinal fluid as well as the global
102
signal, smoothing with a full-width half-maximum (FWHM) of 6 mm.
103
The time series was band-pass filtered with the ideal filter (including six
104
frequency bands: slow 6: 0.001-0.01 Hz, slow 5: 0.01-0.027 Hz, slow 4:
105
0.027-0.073 Hz, slow 3: 0.073-0.198 Hz, show 2: 0.198-0.5 Hz, and slow
106
1: 0.5-0.694 Hz). The higher frequency boundary was determined by the
107
Nyquist theorem ((1 / 0.72 s) / 2 ≈ 0.694 Hz), whereas the lower
108
boundary was determined by the length of data (1 / (1200 * 0.72 s) ≈
109
0.001 Hz). The filtered time series of each run was normalized by
110
subtracting the mean and dividing by the standard deviation. The
AC C
EP
TE D
M AN U
SC
94
6
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 111
normalized time courses were linked together, forming a series of 4800
112
time points.
113
The segmentation of the rAI with backtracking approach To investigate whether the task-positive and task-negative networks
115
are connected to different parts of the rAI, we first computed the
116
functional connectivity (FC) between regions of interest (ROIs) in these
117
two networks and the rAI. The FC analysis within predefined six
118
frequency bands was performed based on ROIs proposed by Fox et al.
119
(Fox et al., 2005). Task-positive regions were centered in the intraparietal
120
sulcus (-25, -57, 46), the frontal eye field (25, -13, 50), and the middle
121
temporal region (-45, -69, -2). Task-negative regions were centered in the
122
medial prefrontal cortex (-1, 47, -4), posterior cingulate cortex (-5, -49,
123
40), and angular gyrus (-45, -67, 36). The radius of each ROI was 6 mm.
124
The target region of FC was determined with the rAI template (Deen et
125
al., 2011). The voxel-wise temporal correlation between each ROI and the
126
target region was computed and transformed to the Fisher’s z value in
127
each frequency band. Of note, we regressed out the global signal to
128
improve the specificity of correlations for different ROIs as suggested by
129
Murphy and Fox (Murphy and Fox, 2017). Because the global signal
130
regression could induce negative correlation (Murphy and Fox, 2017), we
131
replicated the FC analysis using data without global signal regression to
132
check the influence of this bias.
AC C
EP
TE D
M AN U
SC
RI PT
114
7
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
In order to obtain the functional parcellation of the rAI, conjunction
134
analysis was performed in each frequency band. Following Fox et al.’s
135
study (Fox et al., 2005), the effective region of rAI was determined by
136
one sample t-test (p < 0.05, z > 3, cluster size > 17 voxels) at each
137
frequency band. If the voxels in the rAI have significant positive
138
correlation with seeds in the task-positive network or negative correlation
139
with seeds in the DMN, they were labeled as 1; while voxels have
140
significant negative correlation with seeds in the task-positive network or
141
positive correlation with seeds in the DMN were labeled as -1. The
142
voxels without significant correlation with any of the seeds were equals
143
to 0. The labeled voxels were then overlaid, resulting in regions with
144
labels from 0 to ±6. The subparts of rAI were defined as regions with
145
label ≥ 5 or ≤ -5 (Fox et al., 2005).
146
The FC pattern of dorsal and ventral parts of the rAI
TE D
M AN U
SC
RI PT
133
To investigate the functional connectivity patterns (FCPs) of rdAI
148
and rvAI and their frequency effect, the voxel-wise FC was conducted
149
based on two seeds in the rAI in each frequency band. The region and
150
frequency effects were evaluated with the repeated measures analysis of
151
variance (ANOVA) using SPM12 (www.fil.ion.ucl.ac.uk/spm). The
152
frequency bands and seeds were served as within subject factors. Post hoc
153
analyses were conducted using paired-samples t-tests. All resulting maps
154
were corrected using family-wise error (FWE) method (p < 0.05) for
AC C
EP
147
8
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 155
multiple comparisons (Worsley et al., 1996).
156
The relationship between rdAI-rvAI correlation and the role of network
157
reconciliation We further asked whether the relationship between the rdAI and
159
rvAI influences their role of network reconciliation. To answer this
160
question, we conducted the Pearson’s correlation analysis (1) between the
161
rdAI-rvAI relationship and the anti-correlated systems associated with the
162
rdAI, and (2) between the rdAI-rvAI relationship and the anti-correlated
163
systems associated with the rvAI. The relationship between the rdAI and
164
rvAI and that between the anti-correlated systems were assessed with the
165
mean signal in each ROI. The ROIs of anti-correlated systems were
166
determined by regions with positive or negative effects in paired-samples
167
t-tests.
168
Results
169
The task-positive network and DMN are connected to different parts of
170
the rAI
SC
M AN U
TE D EP
AC C
171
RI PT
158
Because there is no prior evidence about the distinctive roles of
172
subparts of the rAI in modulating the task-positive network and DMN, we
173
cannot use predefined templates of rdAI and rvAI. Instead, we used the
174
backtracking approach to determine whether the rAI could be separated
175
into two different parts based on the FC with seeds in the task-positive 9
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
network and DMN. The conjunction analysis shows that seeds in the
177
task-positive network tend to be positively connected to the rdAI (Fig. 1,
178
yellow part) whereas those in the DMN tend to be positively connected
179
with the rvAI (Fig. 1, cyan part). The negative connections are opposite
180
to the positive connections. The dorsal-ventral segmentation can be seen
181
in frequency bands of slow 6, 5, 4, and 3, indicating the band-limited
182
segmentation of rAI. The dorsal-vental segmentation is independent of
183
global signal regression, as shown by almost the same distribution of
184
rdAI and rvAI with data without global signal regression (see Fig. S1).
185
Insert Fig. 1 about here
186
Effects of seed, frequency, and their interactions
M AN U
SC
RI PT
176
The ANOVA shows that the seed effect spreads over almost the
188
whole brain with interlaced distribution, except for part of the primary
189
visual and sensorimotor cortices (Fig. 2), indicating that the FCPs of rdAI
190
and rvAI cover almost the whole brain in the interphase way. The
191
frequency effect is mainly located in the sensory region (the visual,
192
auditory, olfactory, and somatosensory cortices) as well as the limbic
193
system (Fig. 2), indicating that communications between the rAI and
194
external- and internal-worlds are influenced by frequency. The interaction
195
of seed by frequency is primarily located in the AI, DMN, fronto-parietal
196
network and limbic system (Fig. 2), including both externally- and
AC C
EP
TE D
187
10
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
internally-oriented systems.
198
Insert Fig. 2 about here
199
Complementary FCPs of rdAI and rvAI
RI PT
197
We examined the resting-state whole brain functional connectivity
201
associate with rdAI and rvAI, respectively. Similar to previous findings
202
(Chang et al., 2013; Deen et al., 2011; Nomi et al., 2016; Touroutoglou et
203
al., 2012), the rdAI is positively related to the dorsal and ventral attention
204
networks and sensorimotor network (SMN), while the rvAI is positively
205
linked to the DMN and limbic system (Fig. 3). By contrast, the rdAI is
206
negatively connected to the DMN, while the rvAI is negatively associated
207
with the dorsal and ventral visual streams as well as the SMN. The
208
complementary pattern appears at frequency bands from slow 6 to slow 3.
209
In other words, the positive connection of the rdAI is to an
210
externally-oriented system involving in cognitive and motor control,
211
whereas that of the rvAI is to an internally-oriented system involving in
212
interoceptive, emotion, and self-related processes. Conversely, the
213
negative connection of the rdAI is to an internally self-related system
214
whereas that of the rvAI is to an externally visual-motor system. The
215
complementary connective pattern of rdAI and rvAI is independent of
216
global signal regression (see Fig. S2).
217
Insert Fig. 3 about here
AC C
EP
TE D
M AN U
SC
200
11
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 218
The correlation between the rdAI and rvAI predicts the anti-correlated
219
systems associated with the rdAI Significant negative correlations are found between the rdAI-rvAI
221
relationship and the anti-correlated systems associated with the rdAI at
222
slow 6 and slow 5 (Fig. 4). There are no significant correlations between
223
the rdAI-rvAI relationship and other systems (-0.182 < r < 0.152, p >
224
0.101). In other words, the effective modulation of rdAI to the
225
anti-correlated systems dependents on the anti-correlation between the
226
rdAI and rvAI in a frequency dependent way.
SC
Discussion
M AN U
228
Insert Fig. 4 about here
TE D
227
RI PT
220
In this study, we demonstrated that the rAI could be divided into two
230
functional parts which are frequency dependent. Specifically, their
231
functional connections are focused on from the slow 6 to slow 3
232
frequency bands. The dorsal and ventral parts of rAI are not simply
233
connected to the cognitive- and emotional-related systems, or to the
234
externally- and internally-oriented systems. Instead, the rvAI is specially
235
connected to the limbic internal system whereas the rdAI and rvAI are
236
linked to output- and input-related external systems, respectively. These
237
findings may improve our understanding of the low frequency spectral
238
fingerprints of cognition and the two coordinated systems in our brain.
AC C
EP
229
12
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 239
The externally- and internally-oriented systems associated with the rAI The role of rAI has not been fully understood. Initially, the
241
cognition-emotion dichotomy has been used to describe the functional
242
separation of the dorsal and ventral anterior cingulate cortex and insula
243
(Bush et al., 2000; Chang et al., 2013). In line with this, Touroutoglou et
244
al. (Touroutoglou et al., 2012) suggested two SNs in the human brain
245
based on the FC of rAI. They found that the dorsal and ventral SNs are
246
related to attentional and affective functions, respectively. However, the
247
cognition-emotion dichotomy of rdAI and rvAI has become increasingly
248
challenged in recent years (Lindquist et al., 2012; Pessoa, 2008; Uddin et
249
al., 2014). On the contrary, the internal-external dichotomy has garnered
250
more support (Menon and Uddin, 2010; Nomi et al., 2016; Zabelina and
251
Andrews-Hanna, 2016). This perspective agrees with most findings about
252
the role of rAI (Uddin, 2015; Uddin et al., 2014). Accordingly, we
253
hypothesized that the task-positive network and DMN should be
254
connected to the rdAI and rvAI, respectively.
SC
M AN U
TE D
EP
AC C
255
RI PT
240
In fact, the results show two somewhat different externally- and
256
internally-oriented systems associated with the rAI. Specifically, the rdAI
257
is positively associated with the goal-directed system and negatively
258
related to the DMN. This result is merited because the rdAI is activated in
259
almost all task domains accompanied by the deactivation of the DMN
260
(Binder, 2012; Kurth et al., 2010). Similar to the evidence that the rdAI 13
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
coordinates the task-positive network and DMN (Chen et al., 2013; Wen
262
et al., 2012), we suggest that the rvAI devotes itself to internal processing
263
while inhibiting the input of external information. This role is supported
264
by plenty of Von Economo neurons in the vAI which have a central role
265
in interoception, emotional experience and social awareness (Craig, 2009).
266
Based on these evidence, we expand the external-internal dichotomy to a
267
more detailed model and show it in Fig. 5. In this model, the rdAI is
268
engaged in the externally-oriented system. It coordinates the task-positive
269
and task-negative networks to produce adaptive behavior (Sridharan et al.,
270
2008; Wen et al., 2012). More importantly, we find the internally-oriented
271
role of the rvAI. The rvAI may be engaged in internal self-referential
272
processing (e.g., the somato-self associated with the limbic system and
273
the social-self associated with the DMN) while restraining the distraction
274
of external information. These roles are identical to the function of vAI in
275
homeostasis and the function of Von Economo neurons in self-awareness
276
(Cauda et al., 2013; Namkung et al., 2017).
SC
M AN U
TE D
EP
AC C
277
RI PT
261
Furthermore, the positive correlation between the rdAI and rvAI
278
could disturb the modulation of the rdAI to the anti-correlation between
279
the task-positive network and DMN, suggesting that the anti-correlated
280
systems are not solely determined by the rdAI. The relationship between
281
the rdAI and rvAI is of importance for the coordinating function of rdAI,
282
indicating the balance between functional integration and functional 14
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 283
separation.
284
Frequency specific functional systems of the rAI The frequency effect of the functional systems associated with the
286
rAI could be understood from three facets: (1) the functional connection
287
of the rAI is focused on the very low frequency range from slow 6 to
288
slow 3; (2) the frequency specific local connections within the low
289
frequency range; and (3) the frequency dependent modulation of the
290
rdAI-rvAI relationship to inter-network correlation.
M AN U
SC
RI PT
285
First, the FC of rdAI and rvAI appears only at frequency range from
292
slow 6 to slow 3. The lack of FC at slow 2 and slow 1 may be due to
293
decreased FC strength and increased noise as frequency increases,
294
although we have regressed out head motion parameters, respiration and
295
cardiac noises (Lin et al., 2015; Thompson and Fransson, 2015).
296
Alternatively, some studies have suggested that the higher frequency
297
range really has particular physiological meanings (Lin et al., 2015;
298
Thompson and Fransson, 2015). Compared with BOLD signal
299
fluctuations in the infra-slow frequency range (0.01-0.1 Hz), those in the
300
slow frequency range (0.1-1 Hz) has got much less attention, resulting in
301
inadequate understanding of its psychophysiological meanings which
302
warrants future investigations (Boubela et al., 2013; Gohel and Biswal,
303
2015; Yuan et al., 2014). The current results add new evidence to the
304
functional significance of low frequency BOLD signal fluctuations,
AC C
EP
TE D
291
15
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 305
indicating that the BOLD signal fluctuations in higher frequency range
306
deserve further research. Second, the frequency effect within the low frequency range is
308
primarily located in the sensory, motor, and limbic systems. These
309
regions are primarily involved in exogenous and endogenous information
310
processing. The lack of frequency effect in the association cortex may be
311
due to differentiated cytoarchitectures or functions of different brain
312
regions (Song et al., 2014). The natural or dominant frequency of the
313
brain is found to be region-specific in higher frequency range (Murray et
314
al., 2014; Rosanova et al., 2009), leaving its frequency characteristics in
315
the lower frequency range (< 1 Hz) largely unknown. Recently, the
316
frequency specificity of different brain regions has been constantly
317
reported (Baria et al., 2011; Song et al., 2014; Zuo et al., 2010),
318
indicating that frequency specificity is pervasive even in the very narrow
319
low frequency band (0.01-1 Hz). The results that frequency effect appears
320
in regions involved in information processing rather than in information
321
integration suggest that these two kinds of regions possess different
322
frequency characteristics. In previous studies, we observed responses in
323
the supplementary motor area in a wider frequency range than those in
324
the task control network (Wang et al., 2015; Wang et al., 2014). It,
325
therefore, seems that different brain regions can communicate with other
326
regions in distinctive frequency bands depending on spatial, time, or
AC C
EP
TE D
M AN U
SC
RI PT
307
16
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 327
cognitive factors (Siegel et al., 2012). These factors are warranted in
328
future
329
organizations of brain networks in the low frequency range.
studies
to
elucidate
mechanisms
of
frequency-specific
Third, the relationship between the rdAI and rvAI influences the
331
coordinated capability of the rdAI to the anti-correlated systems in a
332
frequency specific means. Of note, this is the first study to uncover the
333
frequency characteristics of the coordinated role of the subparts of rAI. It
334
seems that the coordinated role the rdAI dependents on the relationship
335
between the rdAI and rvAI in slow 6 and slow 5 frequency bands,
336
whereas in the higher frequency range, the rdAI and rvAI play their roles
337
independently. This result is in line with the perspective that more
338
complex and integrated functions are executed in lower frequency range
339
(Penttonen and Buzsáki, 2003). Although the dominant frequency of each
340
brain function is thought to be established by evolution and experience
341
(Schroeder and Lakatos, 2012), each brain region has been suggested to
342
have specific intrinsic frequency determined by the characteristics of
343
neural system (Li et al., 2015; Murray et al., 2014; Rosanova et al., 2009).
344
We show that the functional segregation and functional integration from
345
the point of frequency dependence, shedding new light on the functional
346
coordination and frequency characteristics of brain systems.
AC C
EP
TE D
M AN U
SC
RI PT
330
347
Overall, the frequency specific functional relationships are
348
consistent with recent viewpoint that different brain regions have 17
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
distinctive intrinsic frequency (Murray et al., 2014) and coherent with
350
other regions in different frequencies to fulfill various psychological
351
activities (Siegel et al., 2012). This underlying hypothesis may provide us
352
a frequency route to decoding complicated brain functions.
353
Conclusions
RI PT
349
In summary, we demonstrated for the first time that the externally-
355
and internally-oriented systems have different connection patterns with
356
the rdAI and rvAI. The frequency specificity is pervasive at the system
357
level. It influences not only the strength of inter-regional FC but also the
358
functional segregation and functional integration.
359
Acknowledgements
TE D
360
M AN U
SC
354
The work was supported by the 863 project (2015AA020505), the
362
Natural Science Foundation of China (61533006, 31600930, 31400901,
363
and 61673089), the Science Foundation of Ministry of Education of
364
China (14XJC190003), and the Fundamental Research Funds for the
365
Central Universities (ZYGX2016KYQD120 and ZYGX2015J141).
AC C
366
EP
361
367
Author contributions
368
Y.W., L.Z., and H.C. designed the experiment; Y.W., L.Z., and Q.Z.
369
analyzed the data; Y. F., L.Z., Q.Z., Q.C., W.L., X.D., B.B., and H.C.
370
wrote the paper. 18
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 371 372
Competing financial interests
373
The authors declare no competing financial interests.
RI PT
374 375
References
376 377
Andrews-Hanna, J.R., Reidler, J.S., Sepulcre, J., Poulin, R., Buckner, R.L., 2010. Functional-Anatomic
378 379
Baria, A.T., Baliki, M.N., Parrish, T., Apkarian, A.V., 2011. Anatomical and functional assemblies of
380 381 382
Bastos, A.M., Vezoli, J., Bosman, C.A., Schoffelen, J.-M., Oostenveld, R., Dowdall, J.R., De Weerd, P.,
383
Binder, J.R., 2012. Task-induced deactivation and the "resting" state. NeuroImage 62, 1086-1091.
384 385 386
Boubela, R.N., Kalcher, K., Huf, W., Kronnerwetter, C., Filzmoser, P., Moser, E., 2013. Beyond noise:
387 388
Bullmore, E., Sporns, O., 2009. Complex brain networks: graph theoretical analysis of structural and
389 390
Bush, G., Luu, P., Posner, M.I., 2000. Cognitive and emotional influences in anterior cingulate cortex.
391 392 393
Cauda, F., Torta, D.M., Sacco, K., D’Agata, F., Geda, E., Duca, S., Geminiani, G., Vercelli, A., 2013.
394 395
Chan, J., Wang, Y., Yan, J.H., Chen, H., 2016. Developmental implications of children’s brain networks
396 397
Chang, L.J., Yarkoni, T., Khaw, M.W., Sanfey, A.G., 2013. Decoding the role of the insula in human
398 399 400 401
Chen, A.C., Oathes, D.J., Chang, C., Bradley, T., Zhou, Z.-W., Williams, L.M., Glover, G.H.,
SC
Fractionation of the Brain's Default Network. Neuron 65, 550-562.
M AN U
brain BOLD oscillations. The Journal of Neuroscience 31, 7910-7919.
Kennedy, H., Fries, P., 2015. Visual areas exert feedforward and feedback influences through distinct frequency channels. Neuron 85, 390-401.
using temporal ICA to extract meaningful information from high-frequency fMRI signal fluctuations
TE D
during rest. Frontiers in human neuroscience 7, 1-12.
functional systems. Nature Reviews Neuroscience 10, 186-198.
EP
Trends in Cognitive Sciences 4, 215-222.
AC C
Functional anatomy of cortical areas characterized by Von Economo neurons. Brain Structure and Function 218, 1-20.
and learning. Reviews in the Neurosciences 27, 713-727.
cognition: functional parcellation and large-scale reverse inference. Cerebral Cortex 23, 739-749.
Deisseroth, K., Etkin, A., 2013. Causal interactions between fronto-parietal central executive and default-mode networks in humans. Proceedings of the National Academy of Sciences 110, 19944-19949. 19
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 402 403
Cole, M.W., Reynolds, J.R., Power, J.D., Repovs, G., Anticevic, A., Braver, T.S., 2013. Multi-task
404 405
Craig, A.D., 2009. How do you feel—now? The anterior insula and human awareness. Nature Reviews
406 407
Deen, B., Pitskel, N.B., Pelphrey, K.A., 2011. Three systems of insular functional connectivity
408 409
Evrard, H.C., Forro, T., Logothetis, N.K., 2012. Von Economo Neurons in the Anterior Insula of the
410 411 412
Finn, E.S., Shen, X., Scheinost, D., Rosenberg, M.D., Huang, J., Chun, M.M., Papademetris, X.,
413 414 415
Fox, M.D., Snyder, A.Z., Vincent, J.L., Corbetta, M., Van Essen, D.C., Raichle, M.E., 2005. The human
416 417 418
Glasser, M.F., Sotiropoulos, S.N., Wilson, J.A., Coalson, T.S., Fischl, B., Andersson, J.L., Xu, J.,
419 420
Gohel, S.R., Biswal, B.B., 2015. Functional integration between brain regions at rest occurs in
421 422 423
Kurth, F., Zilles, K., Fox, P.T., Laird, A.R., Eickhoff, S.B., 2010. A link between the systems: functional
424 425 426
Lakatos, P., Barczak, A., Neymotin, S.A., Mcginnis, T., Ross, D., Javitt, D.C., O'Connell, M.N., 2016.
427 428
Li, J.M., Bentley, W.J., Snyder, A.Z., Raichle, M.E., Snyder, L.H., 2015. Functional connectivity arises
429 430 431
Lin, F.-H., Chu, Y.-H., Hsu, Y.-C., Lin, J.-F.L., Tsai, K.W.-K., Tsai, S.-Y., Kuo, W.-J., 2015. Significant
432 433
Lindquist, K.A., Wager, T.D., Kober, H., Bliss-Moreau, E., Barrett, L.F., 2012. The brain basis of
434 435
Liu, F., Zhu, C., Wang, Y., Guo, W., Li, M., Wang, W., Long, Z., Meng, Y., Cui, Q., Zeng, L., Chen, H.,
connectivity reveals flexible hubs for adaptive task control. Nature Neuroscience 16, 1348-1355.
Neuroscience 10, 59-70.
RI PT
identified with cluster analysis. Cerebral Cortex 21, 1498-1506.
Macaque Monkey. Neuron 74, 482-489.
of brain connectivity. Nature Neuroscience 18, 1664-1671.
SC
Constable, R.T., 2015. Functional connectome fingerprinting: Identifying individuals based on patterns
M AN U
brain is intrinsically organized into dynamic, anticorrelated functional networks. Proceedings of the National Academy of Sciences of the United States of America 102, 9673-9678.
Jbabdi, S., Webster, M., Polimeni, J.R., 2013. The minimal preprocessing pipelines for the Human Connectome Project. NeuroImage 80, 105-124.
TE D
multiple-frequency bands. Brain Connectivity 5, 23-34.
differentiation and integration within the human insula revealed by meta-analysis. Brain Structure and
EP
Function 214, 519-534.
Global dynamics of selective attention and its lapses in primary auditory cortex. Nature Neuroscience
AC C
19, 1707-1717.
from a slow rhythmic mechanism. Proceedings of the National Academy of Sciences 112, 2527-2535.
feed-forward connectivity revealed by high frequency components of BOLD fMRI signals. NeuroImage 121, 69-77.
emotion: A meta- analytic review. Behavioral and Brain Sciences 35, 121-143.
2015. Disrupted cortical hubs in functional brain networks in social anxiety disorder. Clinical 20
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 436
Neurophysiology 22, 102-115.
437 438
Maris, E., Fries, P., van Ede, F., 2016. Diverse Phase Relations among Neuronal Rhythms and Their
439 440 441 442
Martino, M., Magioncalda, P., Huang, Z., Conio, B., Piaggio, N., Duncan, N.W., Rocchi, G., Escelsior,
443 444
Mason, M.F., Norton, M.I., Van Horn, J.D., Wegner, D.M., Grafton, S.T., Macrae, C.N., 2007.
445 446
Menon, V., Uddin, L.Q., 2010. Saliency, switching, attention and control: a network model of insula
447 448
Murphy, K., Fox, M.D., 2017. Towards a consensus regarding global signal regression for resting state
449 450 451
Murray, J.D., Bernacchia, A., Freedman, D.J., Romo, R., Wallis, J.D., Cai, X., Padoa-Schioppa, C.,
452 453
Namkung, H., Kim, S.H., Sawa, A., 2017. The Insula: An Underestimated Brain Area in Clinical
454 455 456
Nomi, J.S., Farrant, K., Damaraju, E., Rachakonda, S., Calhoun, V.D., Uddin, L.Q., 2016. Dynamic
457 458 459
Palva,
460 461
Penttonen, M., Buzsáki, G., 2003. Natural logarithmic relationship between brain oscillators. Thalamus
462 463
Pessoa, L., 2008. On the relationship between emotion and cognition. Nature Reviews Neuroscience 9,
464
Raichle, M.E., 2015. The brain's default mode network. Annual Review of Neuroscience 38, 433-447.
465 466
Rosanova, M., Casali, A., Bellina, V., Resta, F., Mariotti, M., Massimini, M., 2009. Natural frequencies
467
Schroeder, C.E., Lakatos, P., 2012. The signs of silence. Neuron 74, 770-772.
468
Siegel, M., Donner, T.H., Engel, A.K., 2012. Spectral fingerprints of large-scale neuronal interactions.
Potential Function. Trends in Neurosciences 39, 86-99.
A., Marozzi, V., Wolff, A., 2016. Contrasting variability patterns in the default mode and sensorimotor
RI PT
networks balance in bipolar depression and mania. Proceedings of the National Academy of Sciences 113, 4824-4829.
functional connectivity MRI. NeuroImage.
M AN U
function. Brain Structure and Function 214, 655-667.
SC
Wandering minds: the default network and stimulus-independent thought. Science 315, 393-395.
Pasternak, T., Seo, H., Lee, D., 2014. A hierarchy of intrinsic timescales across primate cortex. Nature Neuroscience 17, 1661-1663.
TE D
Neuroscience, Psychiatry, and Neurology. Trends in Neurosciences 40, 200-207.
functional network connectivity reveals unique and overlapping profiles of insula subdivisions. Human Brain Mapping 37, 1770-1787. Palva,
S.,
2012.
Infra-slow
fluctuations
in
electrophysiological
recordings,
EP
J.M.,
blood-oxygenation-level-dependent signals, and psychophysical time series. NeuroImage 62,
AC C
2201-2211.
& Related Systems 2, 145-152.
148-158.
of human corticothalamic circuits. The Journal of Neuroscience 29, 7679-7685.
21
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 469
Nature Reviews Neuroscience 13, 121-134.
470 471
Song, X., Zhang, Y., Liu, Y., 2014. Frequency specificity of regional homogeneity in the resting-state
472 473 474
Sridharan, D., Levitin, D.J., Menon, V., 2008. A critical role for the right fronto-insular cortex in
475 476
Thompson, W.H., Fransson, P., 2015. The frequency dimension of fMRI dynamic connectivity:
477 478 479
Touroutoglou, A., Hollenbeck, M., Dickerson, B.C., Barrett, L.F., 2012. Dissociable large-scale
480 481
Uddin, L.Q., 2015. Salience processing and insular cortical function and dysfunction. Nature Reviews
482 483 484
Uddin,
485 486
Van Essen, D.C., Smith, S.M., Barch, D.M., Behrens, T.E., Yacoub, E., Ugurbil, K., Consortium,
487 488 489
Wang, Y.-F., Dai, G.-S., Liu, F., Long, Z.-L., Yan, J.H., Chen, H.-F., 2015. Steady-state BOLD response
490 491
Wang, Y.-F., Liu, F., Long, Z.-L., Duan, X.-J., Cui, Q., Yan, J.H., Chen, H.-F., 2014. Steady-state
492 493 494
Wang, Y.F., Long, Z., Cui, Q., Liu, F., Jing, X.J., Chen, H., Guo, X.N., Yan, J.H., Chen, H.F., 2016.
495 496
Wen, X., Liu, Y., Yao, L., Ding, M., 2013. Top-down regulation of default mode activity in spatial
497 498
Wen, X., Yao, L., Liu, Y., Ding, M., 2012. Causal interactions in attention networks predict behavioral
499 500 501
Worsley, K.J., Marrett, S., Neelin, P., Vandal, A.C., Friston, K.J., Evans, A.C., 1996. A unified
502
Yuan, B.-K., Wang, J., Zang, Y.-F., Liu, D.-Q., 2014. Amplitude differences in high-frequency fMRI
human brain. PLoS ONE 9, e86818.
switching between central-executive and default-mode networks. Proceedings of the National Academy
RI PT
of Sciences 105, 12569-12574.
network connectivity, functional hubs and integration in the resting brain. NeuroImage 121, 227-242.
SC
networks anchored in the right anterior insula subserve affective experience and attention. NeuroImage
Neuroscience 16, 55-61. L.Q.,
Kinnison,
J.,
Pessoa,
M AN U
60.
L.,
Anderson,
M.L.,
2014.
Beyond
the
tripartite
cognition–emotion–interoception model of the human insular cortex. Journal of Cognitive Neuroscience 26, 16-27.
TE D
W.-M.H., 2013. The WU-Minn human connectome project: an overview. NeuroImage 80, 62-79.
to higher-order cognition modulates low frequency neural oscillations. Journal of Cognitive
EP
Neuroscience 27, 2406-2415.
BOLD response modulates low frequency neural oscillations. Scientific reports 4, 7376.
AC C
Low frequency steady‐state brain responses modulate large scale functional networks in a frequency ‐specific means. Human Brain Mapping 37, 381-394.
visual attention. The Journal of Neuroscience 33, 6444-6453.
performance. Journal of Neuroscience 32, 1284-1292.
statistical approach for determining significant signals in images of cerebral activation. Human Brain Mapping 4, 58-73.
22
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 503
signals between eyes open and eyes closed resting states. Frontiers in human neuroscience 8, 1-13.
504 505
Zabelina,
506 507
Zuo, X.-N., Di Martino, A., Kelly, C., Shehzad, Z.E., Gee, D.G., Klein, D.F., Castellanos, F.X., Biswal,
D.L.,
Andrews-Hanna,
J.R.,
2016.
Dynamic
network
interactions
supporting
internally-oriented cognition. Current Opinion in Neurobiology 40, 86-93.
B.B., Milham, M.P., 2010. The oscillating brain: complex and reliable. NeuroImage 49, 1432-1445.
AC C
EP
TE D
M AN U
SC
RI PT
508
23
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
Figure legends
510
Figure 1. The functional parcellation of the right anterior insula. Based
511
on seeds in the task-positive network and default mode network, the
512
functional connectivity patterns are separated in the dorsal and ventral
513
parts of the right anterior insula. The functional parcellation is remarkable
514
in lower frequency range from slow 6 to slow 3.
SC
515
RI PT
509
Figure 2. The effects of seed and frequency and their interaction. The
517
functional connectivity patterns of the right dorsal and ventral anterior
518
insula are interlaced in widespread regions (left). The frequency effect is
519
primarily located in the sensory, motor, and limbic systems (middle). The
520
seed by frequency interaction is mainly lies in the anterior insula, default
521
mode network, fronto-parietal network, and limbic system (right). The
522
results are corrected with FWE method (p < 0.05).
TE D
EP
523
M AN U
516
Figure 3. Functional connectivity patterns of the rdAI and rvAI in
525
frequency bands from slow 6 to slow 3. The rdAI is positively connected
526
to the goal-directed system and negatively connected to the default mode
527
network. The rvAI is positively linked to the default mode network and
528
limbic system and negatively linked to the sensory system. The results are
529
corrected with FWE method (p < 0.05). rdAI: right dorsal anterior insula;
530
rvAI: right ventral anterior insula.
AC C
524
24
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 531
Figure 4. The relationship between rdAI and rvAI could predict the
533
coordinated ability of the rdAI in slow 6 and slow 5 frequency bands. The
534
more negative the correlation between the rdAI and rvAI, the larger
535
difference between the rdAI based anti-correlated systems.
RI PT
532
536
Figure 5. The model of coordinated roles of the rdAI and rvAI. The rdAI
538
and rvAI modulate the externally- and internally-oriented systems,
539
respectively. Being different from the previous model, we show the two
540
systems in details that the externally-oriented systems associated with the
541
rdAI and rvAI are relevant to output and input information, respectively.
542
Furthermore, the rvAI is particularly related to the limbic system.
AC C
EP
TE D
M AN U
SC
537
25
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
S1053-8119(17)30817-0
DOI:
10.1016/j.neuroimage.2017.10.004
Reference:
YNIMG 14382
To appear in:
NeuroImage
Received Date: 29 June 2017 Accepted Date: 2 October 2017
Please cite this article as: Wang, Y., Zhu, L., Zou, Q., Cui, Q., Liao, W., Duan, X., Biswal, B., Chen, H., Frequency dependent hub role of the dorsal and ventral right anterior insula, NeuroImage (2017), doi: 10.1016/j.neuroimage.2017.10.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
Running title: Sub-hubs of the rAI
Frequency dependent hub role of the dorsal and ventral right
RI PT
anterior insula
M AN U
Duan1, Bharat Biswal3,1, Huafu Chen1*.
SC
Yifeng Wang1, Lixia Zhu1, Qijun Zou1, Qian Cui2, Wei Liao1, Xujun
1 Key laboratory for NeuroInformation of Ministry of Education, School of Life Science and Technology and Center for Information in BioMedicine, University of Electronic Science and Technology of China,
TE D
Chengdu, 610054, China.
2 School of Political Science and Public Administration, University of Electronic Science and Technology of China, Chengdu 610054, China.
EP
3 Department of Biomedical Engineering, New Jersey Institute of
AC C
Technology, 607 Fenster Hall, University Height, Newark, NJ 07102, USA
* Correspondence: Huafu Chen Tele: +86-138-0800-3171; Fax: +86-28-83208238 Email: [email protected]. Or Yifeng Wang Email: [email protected] 1
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
Abstract
2
The right anterior insula (rAI) plays a crucial role in generating adaptive
3
behavior by orchestrating multiple brain networks. Based on functional
4
separation findings of the insula and spectral fingerprints theory of
5
cognitive functions, we hypothesize that the hub role of the rAI is region
6
and frequency dependent. Using the Human Connectome Project dataset
7
and backtracking approach, we segregate the rAI into dorsal and ventral
8
parts at frequency bands from slow 6 to slow 3, indicating the frequency
9
dependent functional separation of the rAI. Functional connectivity
10
analysis shows that, within lower than 0.198 Hz frequency range, the
11
dorsal and ventral parts of rAI form a complementary system to
12
synchronize with externally and internally-oriented networks. Moreover,
13
the relationship between the dorsal and ventral rAIs predicts the
14
relationship between anti-correlated networks associated with the dorsal
15
rAI at slow 6 and slow 5, suggesting a frequency dependent regulation of
16
the rAI to brain networks. These findings could improve our
17
understanding of the rAI by supporting the region and frequency
18
dependent function of rAI and its essential role in coordinating brain
19
systems relevant to internal and external environments.
20
Key words: anterior insula; frequency-specificity; functional separation;
21
hub; Human Connectome Project.
AC C
EP
TE D
M AN U
SC
RI PT
1
22 2
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 23
Introduction The human brain operates as a highly organized multi-scale system
25
consisting of multiple structural and functional networks (Bullmore and
26
Sporns, 2009; Chan et al., 2016). The execution of normal brain function
27
requires orchestration among networks such as the balance between
28
task-positive and task-negative networks (Fox et al., 2005). Inter-network
29
cooperation depends on the modulation of particular brain regions which
30
usually serve as functional hubs (Cole et al., 2013; Liu et al., 2015). The
31
right anterior insula (rAI) has been thought to be an important hub in
32
coordinating brain networks to produce adaptive behavior in various
33
cognitive processes (Chen et al., 2013; Sridharan et al., 2008).
34
Specifically, it has been confirmed that the rAI modulates the default
35
mode
36
attention-demanding tasks (Sridharan et al., 2008; Wen et al., 2013).
(DMN)
and
cognitive
control
network
in
EP
network
TE D
M AN U
SC
RI PT
24
Recent studies have suggested that the rAI could be further divided
38
into dorsal and ventral functional sub-parts (Chang et al., 2013; Deen et
39
al., 2011). The right dorsal AI (rdAI) is a core region of the salience
40
network (SN) and is connected to cognitive control networks. Typically, it
41
is activated in almost all tasks involving goal-directed cognition (Chang
42
et al., 2013; Nomi et al., 2016). By contrast, the right ventral AI (rvAI) is
43
rich in Von Economo neurons and has been demonstrated to be involved
44
in interoception, emotion, and social awareness via connections with the
AC C
37
3
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
limbic system (Evrard et al., 2012; Uddin, 2015). Similar to the limbic
46
system, the DMN is primarily involved in internal thoughts, such as
47
self-reference processing, episodic memory, planning and so on
48
(Andrews-Hanna et al., 2010; Mason et al., 2007; Raichle, 2015),
49
indicating that the DMN is more likely to be connected with the rvAI.
50
Therefore, we surmised that the rdAI and rvAI may be prone to connect
51
with externally- and internally-oriented systems, respectively (Menon and
52
Uddin, 2010; Zabelina and Andrews-Hanna, 2016).
SC
Besides
M AN U
53
RI PT
45
the
functional
separation
hypothesis,
the
spectral
fingerprints hypothesis (Maris et al., 2016; Siegel et al., 2012) suggests
55
that disparate cognitive processes are implemented by different brain
56
networks embedded in specific frequency bands. For instance, neural
57
oscillations in beta and gamma bands have been found to play distinctive
58
roles in top-down and bottom-up information streams (Bastos et al.,
59
2015). The externally-oriented and internally-oriented systems have also
60
been reported to be dominated by delta-theta and alpha-beta bands,
61
respectively (Lakatos et al., 2016). The spectral fingerprints of these two
62
systems in the low frequency blood oxygen level dependent (BOLD)
63
signal fluctuations is undetermined yet, although frequency-specific
64
functions have been widely observed in this frequency range in cognitive
65
processes and clinical cases (Martino et al., 2016; Palva and Palva, 2012;
66
Wang et al., 2016).
AC C
EP
TE D
54
4
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
In the current study, we investigated whether the hub role of the rAI
68
is region and frequency dependent. Specifically, we used the Human
69
Connectome Project (HCP) 100 unrelated dataset to explore whether the
70
externally- and internally-oriented systems are connected to different
71
parts of the rAI and whether these connections are frequency dependent.
72
In reverse, whether the sub-parts of rAI are connected to other systems
73
other than the conventional externally- and internally-oriented systems
74
and whether these connections are frequency dependent.
75
Methods
76
Data acquisition
M AN U
SC
RI PT
67
The resting-state functional magnetic resonance imaging (rfMRI)
78
data of 82 subjects were used in this study. The sample was selected from
79
the HCP 100 unrelated dataset (https://db.humanconnectome.org) with
80
the criteria that respiratory, cardiac, and head movement data are
81
available
82
rfMRI_REST2_LR, and rfMRI_REST2_RL). REST1 and REST2 were
83
acquired on two different days. The HCP scanning protocol was approved
84
by the local Institutional Review Board at Washington University in St.
85
Louis. Informed consent was obtained from all subjects. All participants
86
were scanned on a customized Siemens 3-T connectome-Skyra scanner.
87
The imaging parameters used to collect the rfMRI data were as follows:
88
TR = 720 ms; TE = 33.1 ms; flip angle = 52°; 2 mm isotropic voxels
EP
TE D
77
all
runs
(rfMRI_REST1_LR,
rfMRI_REST1_RL,
AC C
in
5
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
(FOV = 208 × 180 mm; 72 slices); multiband factor = 8; echo spacing =
90
0.58 ms; bandwidth (BW) = 2290 Hz/px; volumes = 1200. Full details on
91
the HCP dataset could be seen in Van Essen et al.’s article (Van Essen et
92
al., 2013).
93
Data preprocessing
RI PT
89
The data with minimal preprocessing pipeline (Glasser et al., 2013)
95
was used. This pipeline includes artifact removal, motion correction and
96
registration to standard space. Standard preprocessing procedure was
97
further applied to the data according to Finn et al.’s study (Finn et al.,
98
2015) which included removal of the linear trend, removal of linear
99
components related to the six motion parameters and their first
100
derivatives, regression of respiratory and cardiac noises, the mean time
101
courses of white matter and cerebrospinal fluid as well as the global
102
signal, smoothing with a full-width half-maximum (FWHM) of 6 mm.
103
The time series was band-pass filtered with the ideal filter (including six
104
frequency bands: slow 6: 0.001-0.01 Hz, slow 5: 0.01-0.027 Hz, slow 4:
105
0.027-0.073 Hz, slow 3: 0.073-0.198 Hz, show 2: 0.198-0.5 Hz, and slow
106
1: 0.5-0.694 Hz). The higher frequency boundary was determined by the
107
Nyquist theorem ((1 / 0.72 s) / 2 ≈ 0.694 Hz), whereas the lower
108
boundary was determined by the length of data (1 / (1200 * 0.72 s) ≈
109
0.001 Hz). The filtered time series of each run was normalized by
110
subtracting the mean and dividing by the standard deviation. The
AC C
EP
TE D
M AN U
SC
94
6
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 111
normalized time courses were linked together, forming a series of 4800
112
time points.
113
The segmentation of the rAI with backtracking approach To investigate whether the task-positive and task-negative networks
115
are connected to different parts of the rAI, we first computed the
116
functional connectivity (FC) between regions of interest (ROIs) in these
117
two networks and the rAI. The FC analysis within predefined six
118
frequency bands was performed based on ROIs proposed by Fox et al.
119
(Fox et al., 2005). Task-positive regions were centered in the intraparietal
120
sulcus (-25, -57, 46), the frontal eye field (25, -13, 50), and the middle
121
temporal region (-45, -69, -2). Task-negative regions were centered in the
122
medial prefrontal cortex (-1, 47, -4), posterior cingulate cortex (-5, -49,
123
40), and angular gyrus (-45, -67, 36). The radius of each ROI was 6 mm.
124
The target region of FC was determined with the rAI template (Deen et
125
al., 2011). The voxel-wise temporal correlation between each ROI and the
126
target region was computed and transformed to the Fisher’s z value in
127
each frequency band. Of note, we regressed out the global signal to
128
improve the specificity of correlations for different ROIs as suggested by
129
Murphy and Fox (Murphy and Fox, 2017). Because the global signal
130
regression could induce negative correlation (Murphy and Fox, 2017), we
131
replicated the FC analysis using data without global signal regression to
132
check the influence of this bias.
AC C
EP
TE D
M AN U
SC
RI PT
114
7
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
In order to obtain the functional parcellation of the rAI, conjunction
134
analysis was performed in each frequency band. Following Fox et al.’s
135
study (Fox et al., 2005), the effective region of rAI was determined by
136
one sample t-test (p < 0.05, z > 3, cluster size > 17 voxels) at each
137
frequency band. If the voxels in the rAI have significant positive
138
correlation with seeds in the task-positive network or negative correlation
139
with seeds in the DMN, they were labeled as 1; while voxels have
140
significant negative correlation with seeds in the task-positive network or
141
positive correlation with seeds in the DMN were labeled as -1. The
142
voxels without significant correlation with any of the seeds were equals
143
to 0. The labeled voxels were then overlaid, resulting in regions with
144
labels from 0 to ±6. The subparts of rAI were defined as regions with
145
label ≥ 5 or ≤ -5 (Fox et al., 2005).
146
The FC pattern of dorsal and ventral parts of the rAI
TE D
M AN U
SC
RI PT
133
To investigate the functional connectivity patterns (FCPs) of rdAI
148
and rvAI and their frequency effect, the voxel-wise FC was conducted
149
based on two seeds in the rAI in each frequency band. The region and
150
frequency effects were evaluated with the repeated measures analysis of
151
variance (ANOVA) using SPM12 (www.fil.ion.ucl.ac.uk/spm). The
152
frequency bands and seeds were served as within subject factors. Post hoc
153
analyses were conducted using paired-samples t-tests. All resulting maps
154
were corrected using family-wise error (FWE) method (p < 0.05) for
AC C
EP
147
8
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 155
multiple comparisons (Worsley et al., 1996).
156
The relationship between rdAI-rvAI correlation and the role of network
157
reconciliation We further asked whether the relationship between the rdAI and
159
rvAI influences their role of network reconciliation. To answer this
160
question, we conducted the Pearson’s correlation analysis (1) between the
161
rdAI-rvAI relationship and the anti-correlated systems associated with the
162
rdAI, and (2) between the rdAI-rvAI relationship and the anti-correlated
163
systems associated with the rvAI. The relationship between the rdAI and
164
rvAI and that between the anti-correlated systems were assessed with the
165
mean signal in each ROI. The ROIs of anti-correlated systems were
166
determined by regions with positive or negative effects in paired-samples
167
t-tests.
168
Results
169
The task-positive network and DMN are connected to different parts of
170
the rAI
SC
M AN U
TE D EP
AC C
171
RI PT
158
Because there is no prior evidence about the distinctive roles of
172
subparts of the rAI in modulating the task-positive network and DMN, we
173
cannot use predefined templates of rdAI and rvAI. Instead, we used the
174
backtracking approach to determine whether the rAI could be separated
175
into two different parts based on the FC with seeds in the task-positive 9
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
network and DMN. The conjunction analysis shows that seeds in the
177
task-positive network tend to be positively connected to the rdAI (Fig. 1,
178
yellow part) whereas those in the DMN tend to be positively connected
179
with the rvAI (Fig. 1, cyan part). The negative connections are opposite
180
to the positive connections. The dorsal-ventral segmentation can be seen
181
in frequency bands of slow 6, 5, 4, and 3, indicating the band-limited
182
segmentation of rAI. The dorsal-vental segmentation is independent of
183
global signal regression, as shown by almost the same distribution of
184
rdAI and rvAI with data without global signal regression (see Fig. S1).
185
Insert Fig. 1 about here
186
Effects of seed, frequency, and their interactions
M AN U
SC
RI PT
176
The ANOVA shows that the seed effect spreads over almost the
188
whole brain with interlaced distribution, except for part of the primary
189
visual and sensorimotor cortices (Fig. 2), indicating that the FCPs of rdAI
190
and rvAI cover almost the whole brain in the interphase way. The
191
frequency effect is mainly located in the sensory region (the visual,
192
auditory, olfactory, and somatosensory cortices) as well as the limbic
193
system (Fig. 2), indicating that communications between the rAI and
194
external- and internal-worlds are influenced by frequency. The interaction
195
of seed by frequency is primarily located in the AI, DMN, fronto-parietal
196
network and limbic system (Fig. 2), including both externally- and
AC C
EP
TE D
187
10
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
internally-oriented systems.
198
Insert Fig. 2 about here
199
Complementary FCPs of rdAI and rvAI
RI PT
197
We examined the resting-state whole brain functional connectivity
201
associate with rdAI and rvAI, respectively. Similar to previous findings
202
(Chang et al., 2013; Deen et al., 2011; Nomi et al., 2016; Touroutoglou et
203
al., 2012), the rdAI is positively related to the dorsal and ventral attention
204
networks and sensorimotor network (SMN), while the rvAI is positively
205
linked to the DMN and limbic system (Fig. 3). By contrast, the rdAI is
206
negatively connected to the DMN, while the rvAI is negatively associated
207
with the dorsal and ventral visual streams as well as the SMN. The
208
complementary pattern appears at frequency bands from slow 6 to slow 3.
209
In other words, the positive connection of the rdAI is to an
210
externally-oriented system involving in cognitive and motor control,
211
whereas that of the rvAI is to an internally-oriented system involving in
212
interoceptive, emotion, and self-related processes. Conversely, the
213
negative connection of the rdAI is to an internally self-related system
214
whereas that of the rvAI is to an externally visual-motor system. The
215
complementary connective pattern of rdAI and rvAI is independent of
216
global signal regression (see Fig. S2).
217
Insert Fig. 3 about here
AC C
EP
TE D
M AN U
SC
200
11
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 218
The correlation between the rdAI and rvAI predicts the anti-correlated
219
systems associated with the rdAI Significant negative correlations are found between the rdAI-rvAI
221
relationship and the anti-correlated systems associated with the rdAI at
222
slow 6 and slow 5 (Fig. 4). There are no significant correlations between
223
the rdAI-rvAI relationship and other systems (-0.182 < r < 0.152, p >
224
0.101). In other words, the effective modulation of rdAI to the
225
anti-correlated systems dependents on the anti-correlation between the
226
rdAI and rvAI in a frequency dependent way.
SC
Discussion
M AN U
228
Insert Fig. 4 about here
TE D
227
RI PT
220
In this study, we demonstrated that the rAI could be divided into two
230
functional parts which are frequency dependent. Specifically, their
231
functional connections are focused on from the slow 6 to slow 3
232
frequency bands. The dorsal and ventral parts of rAI are not simply
233
connected to the cognitive- and emotional-related systems, or to the
234
externally- and internally-oriented systems. Instead, the rvAI is specially
235
connected to the limbic internal system whereas the rdAI and rvAI are
236
linked to output- and input-related external systems, respectively. These
237
findings may improve our understanding of the low frequency spectral
238
fingerprints of cognition and the two coordinated systems in our brain.
AC C
EP
229
12
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 239
The externally- and internally-oriented systems associated with the rAI The role of rAI has not been fully understood. Initially, the
241
cognition-emotion dichotomy has been used to describe the functional
242
separation of the dorsal and ventral anterior cingulate cortex and insula
243
(Bush et al., 2000; Chang et al., 2013). In line with this, Touroutoglou et
244
al. (Touroutoglou et al., 2012) suggested two SNs in the human brain
245
based on the FC of rAI. They found that the dorsal and ventral SNs are
246
related to attentional and affective functions, respectively. However, the
247
cognition-emotion dichotomy of rdAI and rvAI has become increasingly
248
challenged in recent years (Lindquist et al., 2012; Pessoa, 2008; Uddin et
249
al., 2014). On the contrary, the internal-external dichotomy has garnered
250
more support (Menon and Uddin, 2010; Nomi et al., 2016; Zabelina and
251
Andrews-Hanna, 2016). This perspective agrees with most findings about
252
the role of rAI (Uddin, 2015; Uddin et al., 2014). Accordingly, we
253
hypothesized that the task-positive network and DMN should be
254
connected to the rdAI and rvAI, respectively.
SC
M AN U
TE D
EP
AC C
255
RI PT
240
In fact, the results show two somewhat different externally- and
256
internally-oriented systems associated with the rAI. Specifically, the rdAI
257
is positively associated with the goal-directed system and negatively
258
related to the DMN. This result is merited because the rdAI is activated in
259
almost all task domains accompanied by the deactivation of the DMN
260
(Binder, 2012; Kurth et al., 2010). Similar to the evidence that the rdAI 13
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
coordinates the task-positive network and DMN (Chen et al., 2013; Wen
262
et al., 2012), we suggest that the rvAI devotes itself to internal processing
263
while inhibiting the input of external information. This role is supported
264
by plenty of Von Economo neurons in the vAI which have a central role
265
in interoception, emotional experience and social awareness (Craig, 2009).
266
Based on these evidence, we expand the external-internal dichotomy to a
267
more detailed model and show it in Fig. 5. In this model, the rdAI is
268
engaged in the externally-oriented system. It coordinates the task-positive
269
and task-negative networks to produce adaptive behavior (Sridharan et al.,
270
2008; Wen et al., 2012). More importantly, we find the internally-oriented
271
role of the rvAI. The rvAI may be engaged in internal self-referential
272
processing (e.g., the somato-self associated with the limbic system and
273
the social-self associated with the DMN) while restraining the distraction
274
of external information. These roles are identical to the function of vAI in
275
homeostasis and the function of Von Economo neurons in self-awareness
276
(Cauda et al., 2013; Namkung et al., 2017).
SC
M AN U
TE D
EP
AC C
277
RI PT
261
Furthermore, the positive correlation between the rdAI and rvAI
278
could disturb the modulation of the rdAI to the anti-correlation between
279
the task-positive network and DMN, suggesting that the anti-correlated
280
systems are not solely determined by the rdAI. The relationship between
281
the rdAI and rvAI is of importance for the coordinating function of rdAI,
282
indicating the balance between functional integration and functional 14
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 283
separation.
284
Frequency specific functional systems of the rAI The frequency effect of the functional systems associated with the
286
rAI could be understood from three facets: (1) the functional connection
287
of the rAI is focused on the very low frequency range from slow 6 to
288
slow 3; (2) the frequency specific local connections within the low
289
frequency range; and (3) the frequency dependent modulation of the
290
rdAI-rvAI relationship to inter-network correlation.
M AN U
SC
RI PT
285
First, the FC of rdAI and rvAI appears only at frequency range from
292
slow 6 to slow 3. The lack of FC at slow 2 and slow 1 may be due to
293
decreased FC strength and increased noise as frequency increases,
294
although we have regressed out head motion parameters, respiration and
295
cardiac noises (Lin et al., 2015; Thompson and Fransson, 2015).
296
Alternatively, some studies have suggested that the higher frequency
297
range really has particular physiological meanings (Lin et al., 2015;
298
Thompson and Fransson, 2015). Compared with BOLD signal
299
fluctuations in the infra-slow frequency range (0.01-0.1 Hz), those in the
300
slow frequency range (0.1-1 Hz) has got much less attention, resulting in
301
inadequate understanding of its psychophysiological meanings which
302
warrants future investigations (Boubela et al., 2013; Gohel and Biswal,
303
2015; Yuan et al., 2014). The current results add new evidence to the
304
functional significance of low frequency BOLD signal fluctuations,
AC C
EP
TE D
291
15
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 305
indicating that the BOLD signal fluctuations in higher frequency range
306
deserve further research. Second, the frequency effect within the low frequency range is
308
primarily located in the sensory, motor, and limbic systems. These
309
regions are primarily involved in exogenous and endogenous information
310
processing. The lack of frequency effect in the association cortex may be
311
due to differentiated cytoarchitectures or functions of different brain
312
regions (Song et al., 2014). The natural or dominant frequency of the
313
brain is found to be region-specific in higher frequency range (Murray et
314
al., 2014; Rosanova et al., 2009), leaving its frequency characteristics in
315
the lower frequency range (< 1 Hz) largely unknown. Recently, the
316
frequency specificity of different brain regions has been constantly
317
reported (Baria et al., 2011; Song et al., 2014; Zuo et al., 2010),
318
indicating that frequency specificity is pervasive even in the very narrow
319
low frequency band (0.01-1 Hz). The results that frequency effect appears
320
in regions involved in information processing rather than in information
321
integration suggest that these two kinds of regions possess different
322
frequency characteristics. In previous studies, we observed responses in
323
the supplementary motor area in a wider frequency range than those in
324
the task control network (Wang et al., 2015; Wang et al., 2014). It,
325
therefore, seems that different brain regions can communicate with other
326
regions in distinctive frequency bands depending on spatial, time, or
AC C
EP
TE D
M AN U
SC
RI PT
307
16
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 327
cognitive factors (Siegel et al., 2012). These factors are warranted in
328
future
329
organizations of brain networks in the low frequency range.
studies
to
elucidate
mechanisms
of
frequency-specific
Third, the relationship between the rdAI and rvAI influences the
331
coordinated capability of the rdAI to the anti-correlated systems in a
332
frequency specific means. Of note, this is the first study to uncover the
333
frequency characteristics of the coordinated role of the subparts of rAI. It
334
seems that the coordinated role the rdAI dependents on the relationship
335
between the rdAI and rvAI in slow 6 and slow 5 frequency bands,
336
whereas in the higher frequency range, the rdAI and rvAI play their roles
337
independently. This result is in line with the perspective that more
338
complex and integrated functions are executed in lower frequency range
339
(Penttonen and Buzsáki, 2003). Although the dominant frequency of each
340
brain function is thought to be established by evolution and experience
341
(Schroeder and Lakatos, 2012), each brain region has been suggested to
342
have specific intrinsic frequency determined by the characteristics of
343
neural system (Li et al., 2015; Murray et al., 2014; Rosanova et al., 2009).
344
We show that the functional segregation and functional integration from
345
the point of frequency dependence, shedding new light on the functional
346
coordination and frequency characteristics of brain systems.
AC C
EP
TE D
M AN U
SC
RI PT
330
347
Overall, the frequency specific functional relationships are
348
consistent with recent viewpoint that different brain regions have 17
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
distinctive intrinsic frequency (Murray et al., 2014) and coherent with
350
other regions in different frequencies to fulfill various psychological
351
activities (Siegel et al., 2012). This underlying hypothesis may provide us
352
a frequency route to decoding complicated brain functions.
353
Conclusions
RI PT
349
In summary, we demonstrated for the first time that the externally-
355
and internally-oriented systems have different connection patterns with
356
the rdAI and rvAI. The frequency specificity is pervasive at the system
357
level. It influences not only the strength of inter-regional FC but also the
358
functional segregation and functional integration.
359
Acknowledgements
TE D
360
M AN U
SC
354
The work was supported by the 863 project (2015AA020505), the
362
Natural Science Foundation of China (61533006, 31600930, 31400901,
363
and 61673089), the Science Foundation of Ministry of Education of
364
China (14XJC190003), and the Fundamental Research Funds for the
365
Central Universities (ZYGX2016KYQD120 and ZYGX2015J141).
AC C
366
EP
361
367
Author contributions
368
Y.W., L.Z., and H.C. designed the experiment; Y.W., L.Z., and Q.Z.
369
analyzed the data; Y. F., L.Z., Q.Z., Q.C., W.L., X.D., B.B., and H.C.
370
wrote the paper. 18
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 371 372
Competing financial interests
373
The authors declare no competing financial interests.
RI PT
374 375
References
376 377
Andrews-Hanna, J.R., Reidler, J.S., Sepulcre, J., Poulin, R., Buckner, R.L., 2010. Functional-Anatomic
378 379
Baria, A.T., Baliki, M.N., Parrish, T., Apkarian, A.V., 2011. Anatomical and functional assemblies of
380 381 382
Bastos, A.M., Vezoli, J., Bosman, C.A., Schoffelen, J.-M., Oostenveld, R., Dowdall, J.R., De Weerd, P.,
383
Binder, J.R., 2012. Task-induced deactivation and the "resting" state. NeuroImage 62, 1086-1091.
384 385 386
Boubela, R.N., Kalcher, K., Huf, W., Kronnerwetter, C., Filzmoser, P., Moser, E., 2013. Beyond noise:
387 388
Bullmore, E., Sporns, O., 2009. Complex brain networks: graph theoretical analysis of structural and
389 390
Bush, G., Luu, P., Posner, M.I., 2000. Cognitive and emotional influences in anterior cingulate cortex.
391 392 393
Cauda, F., Torta, D.M., Sacco, K., D’Agata, F., Geda, E., Duca, S., Geminiani, G., Vercelli, A., 2013.
394 395
Chan, J., Wang, Y., Yan, J.H., Chen, H., 2016. Developmental implications of children’s brain networks
396 397
Chang, L.J., Yarkoni, T., Khaw, M.W., Sanfey, A.G., 2013. Decoding the role of the insula in human
398 399 400 401
Chen, A.C., Oathes, D.J., Chang, C., Bradley, T., Zhou, Z.-W., Williams, L.M., Glover, G.H.,
SC
Fractionation of the Brain's Default Network. Neuron 65, 550-562.
M AN U
brain BOLD oscillations. The Journal of Neuroscience 31, 7910-7919.
Kennedy, H., Fries, P., 2015. Visual areas exert feedforward and feedback influences through distinct frequency channels. Neuron 85, 390-401.
using temporal ICA to extract meaningful information from high-frequency fMRI signal fluctuations
TE D
during rest. Frontiers in human neuroscience 7, 1-12.
functional systems. Nature Reviews Neuroscience 10, 186-198.
EP
Trends in Cognitive Sciences 4, 215-222.
AC C
Functional anatomy of cortical areas characterized by Von Economo neurons. Brain Structure and Function 218, 1-20.
and learning. Reviews in the Neurosciences 27, 713-727.
cognition: functional parcellation and large-scale reverse inference. Cerebral Cortex 23, 739-749.
Deisseroth, K., Etkin, A., 2013. Causal interactions between fronto-parietal central executive and default-mode networks in humans. Proceedings of the National Academy of Sciences 110, 19944-19949. 19
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 402 403
Cole, M.W., Reynolds, J.R., Power, J.D., Repovs, G., Anticevic, A., Braver, T.S., 2013. Multi-task
404 405
Craig, A.D., 2009. How do you feel—now? The anterior insula and human awareness. Nature Reviews
406 407
Deen, B., Pitskel, N.B., Pelphrey, K.A., 2011. Three systems of insular functional connectivity
408 409
Evrard, H.C., Forro, T., Logothetis, N.K., 2012. Von Economo Neurons in the Anterior Insula of the
410 411 412
Finn, E.S., Shen, X., Scheinost, D., Rosenberg, M.D., Huang, J., Chun, M.M., Papademetris, X.,
413 414 415
Fox, M.D., Snyder, A.Z., Vincent, J.L., Corbetta, M., Van Essen, D.C., Raichle, M.E., 2005. The human
416 417 418
Glasser, M.F., Sotiropoulos, S.N., Wilson, J.A., Coalson, T.S., Fischl, B., Andersson, J.L., Xu, J.,
419 420
Gohel, S.R., Biswal, B.B., 2015. Functional integration between brain regions at rest occurs in
421 422 423
Kurth, F., Zilles, K., Fox, P.T., Laird, A.R., Eickhoff, S.B., 2010. A link between the systems: functional
424 425 426
Lakatos, P., Barczak, A., Neymotin, S.A., Mcginnis, T., Ross, D., Javitt, D.C., O'Connell, M.N., 2016.
427 428
Li, J.M., Bentley, W.J., Snyder, A.Z., Raichle, M.E., Snyder, L.H., 2015. Functional connectivity arises
429 430 431
Lin, F.-H., Chu, Y.-H., Hsu, Y.-C., Lin, J.-F.L., Tsai, K.W.-K., Tsai, S.-Y., Kuo, W.-J., 2015. Significant
432 433
Lindquist, K.A., Wager, T.D., Kober, H., Bliss-Moreau, E., Barrett, L.F., 2012. The brain basis of
434 435
Liu, F., Zhu, C., Wang, Y., Guo, W., Li, M., Wang, W., Long, Z., Meng, Y., Cui, Q., Zeng, L., Chen, H.,
connectivity reveals flexible hubs for adaptive task control. Nature Neuroscience 16, 1348-1355.
Neuroscience 10, 59-70.
RI PT
identified with cluster analysis. Cerebral Cortex 21, 1498-1506.
Macaque Monkey. Neuron 74, 482-489.
of brain connectivity. Nature Neuroscience 18, 1664-1671.
SC
Constable, R.T., 2015. Functional connectome fingerprinting: Identifying individuals based on patterns
M AN U
brain is intrinsically organized into dynamic, anticorrelated functional networks. Proceedings of the National Academy of Sciences of the United States of America 102, 9673-9678.
Jbabdi, S., Webster, M., Polimeni, J.R., 2013. The minimal preprocessing pipelines for the Human Connectome Project. NeuroImage 80, 105-124.
TE D
multiple-frequency bands. Brain Connectivity 5, 23-34.
differentiation and integration within the human insula revealed by meta-analysis. Brain Structure and
EP
Function 214, 519-534.
Global dynamics of selective attention and its lapses in primary auditory cortex. Nature Neuroscience
AC C
19, 1707-1717.
from a slow rhythmic mechanism. Proceedings of the National Academy of Sciences 112, 2527-2535.
feed-forward connectivity revealed by high frequency components of BOLD fMRI signals. NeuroImage 121, 69-77.
emotion: A meta- analytic review. Behavioral and Brain Sciences 35, 121-143.
2015. Disrupted cortical hubs in functional brain networks in social anxiety disorder. Clinical 20
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 436
Neurophysiology 22, 102-115.
437 438
Maris, E., Fries, P., van Ede, F., 2016. Diverse Phase Relations among Neuronal Rhythms and Their
439 440 441 442
Martino, M., Magioncalda, P., Huang, Z., Conio, B., Piaggio, N., Duncan, N.W., Rocchi, G., Escelsior,
443 444
Mason, M.F., Norton, M.I., Van Horn, J.D., Wegner, D.M., Grafton, S.T., Macrae, C.N., 2007.
445 446
Menon, V., Uddin, L.Q., 2010. Saliency, switching, attention and control: a network model of insula
447 448
Murphy, K., Fox, M.D., 2017. Towards a consensus regarding global signal regression for resting state
449 450 451
Murray, J.D., Bernacchia, A., Freedman, D.J., Romo, R., Wallis, J.D., Cai, X., Padoa-Schioppa, C.,
452 453
Namkung, H., Kim, S.H., Sawa, A., 2017. The Insula: An Underestimated Brain Area in Clinical
454 455 456
Nomi, J.S., Farrant, K., Damaraju, E., Rachakonda, S., Calhoun, V.D., Uddin, L.Q., 2016. Dynamic
457 458 459
Palva,
460 461
Penttonen, M., Buzsáki, G., 2003. Natural logarithmic relationship between brain oscillators. Thalamus
462 463
Pessoa, L., 2008. On the relationship between emotion and cognition. Nature Reviews Neuroscience 9,
464
Raichle, M.E., 2015. The brain's default mode network. Annual Review of Neuroscience 38, 433-447.
465 466
Rosanova, M., Casali, A., Bellina, V., Resta, F., Mariotti, M., Massimini, M., 2009. Natural frequencies
467
Schroeder, C.E., Lakatos, P., 2012. The signs of silence. Neuron 74, 770-772.
468
Siegel, M., Donner, T.H., Engel, A.K., 2012. Spectral fingerprints of large-scale neuronal interactions.
Potential Function. Trends in Neurosciences 39, 86-99.
A., Marozzi, V., Wolff, A., 2016. Contrasting variability patterns in the default mode and sensorimotor
RI PT
networks balance in bipolar depression and mania. Proceedings of the National Academy of Sciences 113, 4824-4829.
functional connectivity MRI. NeuroImage.
M AN U
function. Brain Structure and Function 214, 655-667.
SC
Wandering minds: the default network and stimulus-independent thought. Science 315, 393-395.
Pasternak, T., Seo, H., Lee, D., 2014. A hierarchy of intrinsic timescales across primate cortex. Nature Neuroscience 17, 1661-1663.
TE D
Neuroscience, Psychiatry, and Neurology. Trends in Neurosciences 40, 200-207.
functional network connectivity reveals unique and overlapping profiles of insula subdivisions. Human Brain Mapping 37, 1770-1787. Palva,
S.,
2012.
Infra-slow
fluctuations
in
electrophysiological
recordings,
EP
J.M.,
blood-oxygenation-level-dependent signals, and psychophysical time series. NeuroImage 62,
AC C
2201-2211.
& Related Systems 2, 145-152.
148-158.
of human corticothalamic circuits. The Journal of Neuroscience 29, 7679-7685.
21
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 469
Nature Reviews Neuroscience 13, 121-134.
470 471
Song, X., Zhang, Y., Liu, Y., 2014. Frequency specificity of regional homogeneity in the resting-state
472 473 474
Sridharan, D., Levitin, D.J., Menon, V., 2008. A critical role for the right fronto-insular cortex in
475 476
Thompson, W.H., Fransson, P., 2015. The frequency dimension of fMRI dynamic connectivity:
477 478 479
Touroutoglou, A., Hollenbeck, M., Dickerson, B.C., Barrett, L.F., 2012. Dissociable large-scale
480 481
Uddin, L.Q., 2015. Salience processing and insular cortical function and dysfunction. Nature Reviews
482 483 484
Uddin,
485 486
Van Essen, D.C., Smith, S.M., Barch, D.M., Behrens, T.E., Yacoub, E., Ugurbil, K., Consortium,
487 488 489
Wang, Y.-F., Dai, G.-S., Liu, F., Long, Z.-L., Yan, J.H., Chen, H.-F., 2015. Steady-state BOLD response
490 491
Wang, Y.-F., Liu, F., Long, Z.-L., Duan, X.-J., Cui, Q., Yan, J.H., Chen, H.-F., 2014. Steady-state
492 493 494
Wang, Y.F., Long, Z., Cui, Q., Liu, F., Jing, X.J., Chen, H., Guo, X.N., Yan, J.H., Chen, H.F., 2016.
495 496
Wen, X., Liu, Y., Yao, L., Ding, M., 2013. Top-down regulation of default mode activity in spatial
497 498
Wen, X., Yao, L., Liu, Y., Ding, M., 2012. Causal interactions in attention networks predict behavioral
499 500 501
Worsley, K.J., Marrett, S., Neelin, P., Vandal, A.C., Friston, K.J., Evans, A.C., 1996. A unified
502
Yuan, B.-K., Wang, J., Zang, Y.-F., Liu, D.-Q., 2014. Amplitude differences in high-frequency fMRI
human brain. PLoS ONE 9, e86818.
switching between central-executive and default-mode networks. Proceedings of the National Academy
RI PT
of Sciences 105, 12569-12574.
network connectivity, functional hubs and integration in the resting brain. NeuroImage 121, 227-242.
SC
networks anchored in the right anterior insula subserve affective experience and attention. NeuroImage
Neuroscience 16, 55-61. L.Q.,
Kinnison,
J.,
Pessoa,
M AN U
60.
L.,
Anderson,
M.L.,
2014.
Beyond
the
tripartite
cognition–emotion–interoception model of the human insular cortex. Journal of Cognitive Neuroscience 26, 16-27.
TE D
W.-M.H., 2013. The WU-Minn human connectome project: an overview. NeuroImage 80, 62-79.
to higher-order cognition modulates low frequency neural oscillations. Journal of Cognitive
EP
Neuroscience 27, 2406-2415.
BOLD response modulates low frequency neural oscillations. Scientific reports 4, 7376.
AC C
Low frequency steady‐state brain responses modulate large scale functional networks in a frequency ‐specific means. Human Brain Mapping 37, 381-394.
visual attention. The Journal of Neuroscience 33, 6444-6453.
performance. Journal of Neuroscience 32, 1284-1292.
statistical approach for determining significant signals in images of cerebral activation. Human Brain Mapping 4, 58-73.
22
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 503
signals between eyes open and eyes closed resting states. Frontiers in human neuroscience 8, 1-13.
504 505
Zabelina,
506 507
Zuo, X.-N., Di Martino, A., Kelly, C., Shehzad, Z.E., Gee, D.G., Klein, D.F., Castellanos, F.X., Biswal,
D.L.,
Andrews-Hanna,
J.R.,
2016.
Dynamic
network
interactions
supporting
internally-oriented cognition. Current Opinion in Neurobiology 40, 86-93.
B.B., Milham, M.P., 2010. The oscillating brain: complex and reliable. NeuroImage 49, 1432-1445.
AC C
EP
TE D
M AN U
SC
RI PT
508
23
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT
Figure legends
510
Figure 1. The functional parcellation of the right anterior insula. Based
511
on seeds in the task-positive network and default mode network, the
512
functional connectivity patterns are separated in the dorsal and ventral
513
parts of the right anterior insula. The functional parcellation is remarkable
514
in lower frequency range from slow 6 to slow 3.
SC
515
RI PT
509
Figure 2. The effects of seed and frequency and their interaction. The
517
functional connectivity patterns of the right dorsal and ventral anterior
518
insula are interlaced in widespread regions (left). The frequency effect is
519
primarily located in the sensory, motor, and limbic systems (middle). The
520
seed by frequency interaction is mainly lies in the anterior insula, default
521
mode network, fronto-parietal network, and limbic system (right). The
522
results are corrected with FWE method (p < 0.05).
TE D
EP
523
M AN U
516
Figure 3. Functional connectivity patterns of the rdAI and rvAI in
525
frequency bands from slow 6 to slow 3. The rdAI is positively connected
526
to the goal-directed system and negatively connected to the default mode
527
network. The rvAI is positively linked to the default mode network and
528
limbic system and negatively linked to the sensory system. The results are
529
corrected with FWE method (p < 0.05). rdAI: right dorsal anterior insula;
530
rvAI: right ventral anterior insula.
AC C
524
24
Sub-hubs of the rAI
ACCEPTED MANUSCRIPT 531
Figure 4. The relationship between rdAI and rvAI could predict the
533
coordinated ability of the rdAI in slow 6 and slow 5 frequency bands. The
534
more negative the correlation between the rdAI and rvAI, the larger
535
difference between the rdAI based anti-correlated systems.
RI PT
532
536
Figure 5. The model of coordinated roles of the rdAI and rvAI. The rdAI
538
and rvAI modulate the externally- and internally-oriented systems,
539
respectively. Being different from the previous model, we show the two
540
systems in details that the externally-oriented systems associated with the
541
rdAI and rvAI are relevant to output and input information, respectively.
542
Furthermore, the rvAI is particularly related to the limbic system.
AC C
EP
TE D
M AN U
SC
537
25
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT