1 mecp2 duplication causes aberrant gaba pathways ...€¦ · 3 recapitulate typical phenotypes in...

1

MECP2 duplication causes aberrant GABA pathways, circuits and 1

behaviors in transgenic monkeys: neural mappings to patients with autism 2

Dan-Chao Cai1#, Zhiwei Wang1,2#, Tingting Bo1,2#, Shengyao Yan1,2#, Yilin Liu1,2, Zhaowen 3

Liu3,4, Kristina Zeljic1,2, Xiaoyu Chen1,2, Yafeng Zhan1,2, Xiu Xu5, Yasong Du6, Yingwei 4

Wang7, Jing Cang8, Guang-Zhong Wang9, Jie Zhang4, Qiang Sun1, Zilong Qiu1, Shengjin 5

Ge8§, Zheng Ye1,2§, Zheng Wang1,2,10§ 6

7

1 Institute of Neuroscience, CAS Center for Excellence in Brain Science and Intelligence Technology, 8

State Key Laboratory of Neuroscience, CAS Key Laboratory of Primate Neurobiology, Chinese 9

Academy of Sciences, Shanghai, China 10 2 University of Chinese Academy of Sciences, China 11 3 Life Science Research Center, Engineering Research Center of Molecular and Neuro Imaging of 12

Ministry of Education, School of Life Science and Technology, Xidian University, Xi’an, China 13 4 Institute of Science and Technology for Brain Inspired Intelligence, Key Laboratory of 14

Computational Neuroscience and Brain Inspired Intelligence, Fudan University, Shanghai, China 15 5 Department of Child Healthcare, Children’s Hospital of Fudan University, Shanghai, China 16 6 Shanghai Mental Health Center, Shanghai Jiao Tong University School of Medicine, Shanghai, 17

China 18 7 Department of Anesthesiology, Huashan Hospital of Fudan University, Shanghai, China 19 8 Department of Anesthesiology, Zhongshan Hospital of Fudan University, Shanghai, China 20 9 Key Laboratory of Computational Biology, Chinese Academy of Sciences-Max Planck Partner 21

Institute for Computational Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of 22

Sciences, Shanghai, China 23 10 Kunming Institute of Zoology, Chinese Academy of Sciences, Kunming, China 24

25 # These authors contributed equally to this work 26 § Authors for correspondence: [email protected] ; [email protected] ; ge.shengjin@zs-27

hospital.sh.cn 28

Tel: +86-(0)21-5492-1713 29

Fax: +86-(0)21-5492-1735 30

31

certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was notthis version posted August 7, 2019. . https://doi.org/10.1101/728113doi: bioRxiv preprint

https://doi.org/10.1101/728113

2

One sentence summary: We identify shared circuit-level abnormalities between MECP2 1

transgenic monkeys and a stratified subgroup of human autism, and demonstrate the 2

translational need of a multimodal approach to capture multifaceted effects triggered by a 3

single genetic event in a genetically-engineered primate model. 4

5


https://doi.org/10.1101/728113

3

Abstract 1

MECP2 gain- and loss-of-function in genetically-engineered monkeys demonstrably 2

recapitulate typical phenotypes in patients, yet where MECP2 mutation affects the monkey 3

brain and whether/how it relates to autism pathology remains unknown. Using expression 4

profiles of 13,888 genes in 182 macaque neocortical samples, we first show that MECP2 5

coexpressed genes are enriched in GABA-related signaling pathways. We then perform 6

analyses on multiple phenotypic levels including locomotive and cognitive behavior, resting-7

state electroencephalography and fMRI in MECP2 overexpressed and wild-type macaque 8

monkeys. Behaviorally, transgenic monkeys exhibit hyperactive and repetitive locomotion, 9

greater separation anxiety response, and less flexibility in rule switching. Moreover, 10

decreased neural synchronization at beta frequency (12-30 Hz) is associated with greater 11

locomotion after peer separation. Further analysis of fMRI-derived connectomics reveals 12

widespread hyper- and hypo-connectivity, where hyper-connectivity prominently involving 13

prefrontal and cingulate networks accounts for deficits in cognitive flexibility. To map 14

MECP2-related aberrant circuits of monkeys to the pathological circuits of autistic patients, 15

individuals in a large public neuroimaging database of autism were clustered using 16

community detection on functional connectivity patterns. In a stratified cohort of 49 autisms 17

and 72 controls, the dysfunctional connectivity profile particularly in prefrontal and temporal 18

networks is highly correlated with that of transgenic monkeys, as is further responsible for 19

the severity of social communicative deficits in patients. Through establishing a circuit-based 20

construct link between transgenic animal models and stratified clinical patients, the present 21

findings with explicable biological causes are potentially amenable to translation for accurate 22

diagnosis and evaluation of future treatments in autism-related disorders. 23

24


https://doi.org/10.1101/728113

4

Keywords: autism spectrum disorder, MECP2 overexpression, transgenic monkeys, beta 1

oscillation, neural synchronization, resting-state fMRI, brain connectome, repetitive restricted 2

behavior, cognitive flexibility 3


https://doi.org/10.1101/728113

5

Introduction 1

Psychiatric disorders including autism spectrum disorder (ASD) are increasingly prevalent 2

(1), and substantially heterogeneous in genetic bases and phenotypic architecture, consisting 3

of subtypes with distinct biological mechanisms (2, 3). To achieve precise diagnosis and 4

treatment, patient stratification based on biological measures has been encouraged in recent 5

research initiatives (3, 4). Using non-invasive electrophysiological and neuroimaging 6

measures, neural subtypes or biotypes of mental illnesses have been identified based on 7

distinct patterns of brain network dysfunction, which are disease-relevant and predictive of 8

treatment responsiveness (5). Recent efforts have also been made to bring genetic disorders 9

with a high penetrance of ASD to the forefront of translational efforts to find treatments for 10

subpopulations of mechanism-based classification of ASD (6). Although this strategy holds 11

significant potential in filtering the complex etiology of autism, how specific genes with rare 12

copy number variants contribute to functional alterations in brain circuitry, and what neural 13

circuit basis may underlie particular pathological behaviors in ASD remains elusive (6). 14

An alternative route to dissect clinical heterogeneity and pursue the underlying 15

neuropsychiatric mechanisms is genetic animal modeling, where certain causative genes in 16

these biologically homogeneous samples are purposely manipulated to recapitulate some 17

dimensions of core symptoms in psychiatric patients (7). One could expect that such animal 18

models with a clear genetic basis would (at least partially) share circuit constructs with 19

certain psychiatric neural subtypes, thereby providing valuable opportunities for probing the 20

gene-circuit-behavior casual chain of events underlying complex brain disorders (7, 8). Here 21

we focus on a transgenic macaque model of ASD with mutations in Methyl-CpG binding 22

protein 2 (MECP2) (9). As extensively demonstrated in both humans (10) and rodent models 23

(11), MECP2 is one of few exceptional genes causing autistic features including intellectual 24

disability, motor dysfunction, anxiety and social behavior deficits. In a previous study, we 25


https://doi.org/10.1101/728113

6

reported the successful application of lentiviral-mediated methods to produce genetically 1

engineered macaque monkeys carrying extra copies of MECP2 that manifested less active 2

social contact and increased stereotypical behaviors (9). However, the relevant brain circuits 3

that mediate the causal effects of atypical MECP2 levels on phenotypic symptoms in 4

transgenic monkeys remain essentially unexplored. In addition, it is unknown whether the 5

neural circuit observed in the monkey model will map neatly onto homologs in specific 6

neural subtypes of autism patients. To address these questions, we employ a combination of 7

MECP2-coexpression network analysis, locomotive and cognitive behavioral tests (12-14), 8

resting-state functional magnetic resonance imaging (rsfMRI) (7, 15) and 9

electroencephalography (EEG) recordings (16), all of which are commonly administered in 10

primate species and potentially amenable to cross-species translation into clinical diagnosis 11

and future development of therapeutic interventions for autism-related disorders. We further 12

conduct a cross-species circuit mapping of dysfunctional connectivity profiles between 13

transgenic monkeys and subgroups of ASD patients who are stratified based on whole-brain 14

resting-state connectivity patterns. 15

16

Results 17

Association of MECP2 coexpression network with GABA function 18

To gain system-level insight into how MECP2 mutations in transgenic (TG) monkeys affect 19

the underlying biological process, we applied a weighted gene coexpression network analysis 20

(WGCNA) to the gene expression data from the neocortex of adult macaque monkeys 21

(Macaca mulatta). Expression profiles of 13,888 genes from 182 public samples were 22

utilized to construct the gene coexpression network (see Supplementary Materials and 23

Methods and Fig. 1A). We found that MECP2 belonged to a module consisting of 159 24

coregulated genes and was positively correlated with the Module Eigengene (ME) (r = 0.66, 25


https://doi.org/10.1101/728113

7

P < 2.2×10-16, Fig. 1, B and C). Further functional enrichment analysis revealed that this 1

module was highly associated (P < 0.01, FDR corrected) with four human diseases including 2

unipolar depression (P = 0.002), four gene pathways including GABA (gamma-aminobutyric 3

acid) function (P = 0.002), and three cellular components including clathrin-sculpted GABA 4

vesicular transport (P = 0.002) (Fig. 1D). Next we explored the transcriptome of cortical 5

samples from three deceased TG and four wild-type (WT) monkeys [PMID: 26808898] to 6

further examine the role of this GABA-related module in MECP2 transgenic monkeys. By 7

comparing the expression abundance of MECP2 top-linked genes between TG and WT 8

monkeys, we found that most of the genes that exhibit the strongest association with MECP2 9

expression were dramatically regulated in TG monkeys. Among those genes, 5 of them were 10

significantly up-regulated (pink color in Fig. 1C) while 18 genes were significantly down-11

regulated (purple color in Fig. 1C). More details of these dysregulated genes are provided in 12

fig. S1. Together the results in both TG and WT monkeys indicate a direct effect of MECP2 13

alteration on the expression of those network adjacent genes. 14

15

Abnormal locomotive behavior in transgenic monkeys 16

At 12~18 months of age, these TG monkeys exhibited an increased frequency of repetitive 17

circular locomotion (9). When they reached ~55 months of age, we repeated the analysis on 18

spontaneous locomotive behaviors in five TG and sixteen age-matched WT monkeys 19

(Macaca fascicularis) (see Supplementary Materials and Methods). Repetitive locomotion 20

including circular routing (movie S1), tumbling (movie S2), and cyclic routing in other more 21

complex paths (movie S3) were observed. General locomotion time and repetitive index in 22

home cage activity were analyzed by two independent observers to indicate the performance 23

in locomotive and repetitive behavior, respectively. The inter-rater reliability of the two 24

measures was 96.3% and 98.1%, respectively. The two measures were not correlated in either 25


https://doi.org/10.1101/728113

8

TG (r = -0.04, P = 0.952) or WT (r = -0.35, P = 0.182) monkeys. The total time spent in 1

general locomotion for TG monkeys was significantly greater than for WT monkeys (P = 2

0.005, Hedge’s g = 1.55, two-sample Student’s t-test). The repetitive index of locomotion 3

was also significantly higher in TG monkeys (P = 0.001, Hedge’s g = 1.94), as presented in 4

Fig. 2B and table S1. 5

TG monkeys also exhibited increased anxiety at an early age in addition to excessive 6

stereotyped locomotion (9). To further examine the relationship between abnormal 7

locomotion and trait anxiety, we designed a novel test (Fig. 2A) based on a classical peer 8

separation paradigm (17) (see Supplementary Materials and Methods). Absolute change of 9

locomotive measures after separation with familiar peers was estimated to indicate the 10

locomotive response to separation anxiety in five TG and seven WT monkeys. Both TG and 11

WT monkeys showed increased locomotion time after peer separation compared to the 12

adaptation period (P = 0.014; P = 0.040, respectively; Fig. 2C, upper panel). No group 13

difference was found in the amount of change in locomotion time (P = 0.138, Hedge’s g = 14

0.87; Fig. 2D, upper panel). TG monkeys also showed a marginally significant increase in 15

repetitive index (P = 0.056; Fig. 2C, lower panel) whereas WT monkeys showed no such 16

change (P = 0.547). There was a significant group difference in the amount of change in 17

repetitive index (P = 0.032, Hedge’s g = 1.34; Fig. 2D, lower panel). As a control study, the 18

same test monkeys were further paired with unfamiliar peer monkeys to mediate the degree 19

of separation anxiety. No group difference in induced behavioral changes was found in the 20

case of unfamiliar peer separation in either general locomotion time (P = 0.643) or repetitive 21

index (P = 0.953). More details of locomotion activity in the familiar and unfamiliar peer 22

separation test are provided in table S2 and table S3, respectively. 23

24

Increased regressive errors in reversal learning task 25


https://doi.org/10.1101/728113

9

To probe potential cognitive inflexibility induced by MECP2 dysfunction (18), we trained 1

five TG and four WT monkeys to perform a discrimination and reversal learning task on a 2

touch screen (12-14) (see Supplementary Materials and Methods). After the animals learned 3

stimulus-directed touching (fig. S2A and movie S4), they were trained to discriminate 4

between green and blue color, and to select the correct color for food reward when presented 5

with two colored buttons (Fig. 3A and movie S5). The rewarded color was initially set to 6

green, then reversed to blue after acquisition of the previous rule of color-reward association, 7

and then randomly switched across sessions (Fig. 3, A and B). Animals had to discover the 8

currently rewarded color through trial and error, and apply that rule in a session of 80 9

consecutive trials (Fig. 3B and movies S6-7). 10

Two types of error after rule switches were assessed. Perseverative errors were trials in 11

which monkeys continued to select the previously rewarded color following negative 12

feedback but prior to the first correct response, suggesting a failure to switch to a new rule. 13

By contrast, regressive errors were trials in which monkeys continued to select the previously 14

rewarded color after the first correct response, thus representing a preference for the old rule 15

or an inability to maintain a new rule (Fig. 3B). Notably, TG monkeys exhibited a 16

significantly greater regressive error rate for the first twenty performed trials than WT 17

monkeys (Fig. 3C and D, P = 0.038, Hedge’s g = 1.47, two-sample Student’s t-test), while 18

they did not demonstrate an increased rate of perseverative errors (P = 0.443, Hedges’ g = 19

0.46, Fig. 3E). Meanwhile, TG and WT monkeys omitted a similar number of trials (P = 20

0.710, Hedges’ g = 0.26, Fig. 3F), and performed equivalently on acquiring both stimulus-21

directed touching (fig. S2B) and stimulus-reward association rules (fig. S2, E to G). 22

Individual performance at all training stages is presented in fig. S2 (B to D, H to L). 23

24

Decreased beta synchronization in transgenic monkeys 25


https://doi.org/10.1101/728113

10

We conducted multi-channel scalp EEG recordings in five TG and sixteen WT monkeys (see 1

table S4 for more details). Light anesthesia state was maintained during data collection by 2

experienced anesthesiologists. Details on EEG data acquisition and analysis, and anesthesia 3

maintenance are provided in Supplementary Materials and Methods. Fig. 4A and Fig. 4B 4

show the spatial location of electrodes (see table S5 for full description) in a three-5

dimensional reconstruction of individual anatomical MRI scan and a standard macaque brain 6

template (19), respectively. Spectral power at each recording site, as well as phase 7

synchronization strength between each pair of recording sites were estimated and compared 8

across two groups in six canonical frequency bands (20), namely delta (1–4 Hz), theta (4–8 9

Hz), alpha (8–12 Hz), beta (12–30 Hz), low gamma (30–60 Hz), and high gamma (60–100 10

Hz). No significant group difference between TG and WT monkeys was found in the relative 11

power of either frequency band after correction for multiple comparisons (fig. S3). 12

Inter-site phase synchronization was estimated via de-biased weighted phase-lag index 13

(dwPLI) (21). Significant group differences were found specifically in the beta band at the 14

network level. Group-averaged beta synchronization networks of TG and WT monkeys with 15

age (linear and quadratic) and gender effect regressed out are presented in Fig. 4C, together 16

with the corresponding effect size (Hedges’ g value) of individual neural synchronization 17

network connections (Fig. 4D). We identified a fronto-parieto-occipital network that showed 18

a statistically significant decrease in beta synchronization in TG monkeys (Fig. 4E, edge-wise 19

P < 0.005, cluster-level corrected P = 0.010). This network consisted of two connections 20

within the frontal lobe (OF1-PF1, PF1-Fz) and four fronto-parietal connections (Fz-FP1, 21

OF1-P1, FT1- P1, FT2-P1) in the left hemisphere, and five parieto-occipital connections (P1-22

O2, P1-O4, P2-O2, P2-O4, P4-O4) in the right hemisphere (more details are specified in fig. 23

S4 and table S6). The averaged neural synchronization matrices of all frequency bands are 24


https://doi.org/10.1101/728113

11

shown in fig. S5. In contrast, no significant group difference in neural synchronization of 1

other frequency bands was observed. 2

Given the strong evidence of copy-number-dependent symptom severity observed in 3

both animal models and patients (10, 11), we attempted to explore potential associations 4

among genetic variability, circuit abnormality and behavioral phenotypes in the current 5

experimental setting. We adopted Manhattan distance to quantitatively describe the overall 6

abnormal extent of the connectivity fingerprint of individual TG relative to WT controls (15, 7

22). Such a ‘distance measure’ can determine whether different fingerprints are ‘close’ or 8

‘far’ from each other. Manhattan distances based on connections with abnormal beta 9

synchronization were calculated for each TG monkey. Spearman’s correlation analysis was 10

performed in a pairwise manner to probe associations among gene (MECP2 copy number), 11

circuit (Manhattan distance based on abnormal beta synchronization), and behavior (general 12

locomotion time and repetitive index in home cage, locomotion change induced by familiar 13

peer separation, and error rate in reversal learning task) in five TG monkeys. The overall 14

abnormality in beta synchronization was significantly correlated with locomotion time 15

change (r = 0.975, P = 0.033, Fig. 4F) after familiar peer separation, but not with the MECP2 16

copy number (r = 0.667, P = 0.267, Fig. 4G). 17

18

Aberrant functional connectivity in transgenic monkeys 19

We applied a between-group comparison of functional connectome derived from intrinsic 20

resting-state fMRI signals, which are highly sensitive and stable, and correlate well with 21

brain gene expression (23, 24). Details on MRI data acquisition and analysis are provided in 22

Supplementary Materials and Methods. We acquired whole-brain fMRI data and constructed 23

connectivity networks with 94 parcellated brain regions (also termed nodes, see table S7 for a 24

complete list of anatomical labels) in sixteen anesthetized macaques including five TG and 25


https://doi.org/10.1101/728113

12

eleven WT monkeys (table S8). Group-averaged connectivity networks of TG and WT 1

monkeys with age and gender as covariates are subjected to two-sample Student’s t-tests, as 2

shown in Fig. 5A. We identified a total of 50 functional connections that showed significant 3

differences between TG and WT groups (edge-wise P < 0.001, cluster-level corrected P = 4

0.019, top-right triangle in Fig. 5B), and then quantified the extent of these group effects by 5

using corresponding effect size (ES) (bottom-left triangle in Fig. 5B). Connectivity strength 6

of these connections in all monkeys is presented in fig. S6. Out of 94 nodes, 40 widespread 7

throughout the entire brain were found containing abnormal connections with remarkably 8

large effect sizes (table S9). The degree of abnormality of each node, i.e., the sum of effect 9

sizes of all abnormal edges connecting to this node (ESn), is presented in the interlayer of 10

Fig. 5C. The right primary somatosensory cortex (S1) showed decreased connectivity with 11

large ESn (top 25%, blue bars in Fig. 5C). In contrast, increased connectivity regions with the 12

top 25% largest ESn (red bars) covered a wide range of frontal cortex including left polar and 13

right centrolateral prefrontal cortex (PFCpol, PFCcl), right lateral orbitofrontal cortex 14

(PFCol) and frontal eye field (FEF), cingulate areas including bilateral retrosplenial and left 15

anterior cingulate cortices (CCr and CCa), and other areas including right medial parietal 16

(PCm) and left parahippocampus (PHC). The distribution of altered functional connections in 17

brain lobes are summarized in Fig. 5D (Z scores in bottom-left triangle, see Supplementary 18

Materials and Methods), after the non-uniform parcellation of the brain template was 19

accounted for in a standardized residual analysis (25). The distribution was preferentially 20

biased to prefrontal and cingulate cortices (P < 0.05, Bonferroni correction, top-right triangle 21

in Fig. 5D). 22

The unique functional connectivity profile of individual TG monkeys was calculated in a 23

similar way to their neural connectivity profiles. We plotted the connectivity fingerprint of 24

each TG, in which topological alterations in individual nodes are clearly illustrated (fig. S7). 25


https://doi.org/10.1101/728113

13

We found that Manhattan distances based on abnormal functional connections of monkeys 1

TG04, TG06, TG08, TG10 and TG11 are 8.06, 9.64, 9.17, 6.98 and 9.17, respectively, all of 2

which except TG10 were statistically different from WT monkeys (P = 0.025, 0.000, 0.000, 3

0.155 and 0.000, permutation test, fig. S7). The Manhattan distance was marginally 4

significantly associated with regressive error rate in cognitive flexibility test (r = 0.872, P = 5

0.067). To further dissect the neural circuit that may underlie a dimensional phenotype in the 6

identified network abnormalities, we quantitatively evaluated the relation between each 7

abnormal edge and behavioral measures, and then adjusted the statistical significance at the 8

cluster-level (see Supplementary Materials and Methods). Fourteen functional connections 9

that mostly stemmed from left PFCpol and left PHC were correlated with the regressive error 10

rate in the reversal learning task (edge-wise P < 0.05, cluster-level corrected P = 0.006, Fig. 11

5E). Strikingly, MECP2 copy number in TG monkeys was significantly associated with the 12

Manhattan distance of these dysfunctional connections (r = 1.000, P = 0.017, Fig. 5F) and 13

with the regressive error rate (r = 1.000, P = 0.017, Fig.5G). Seven other functional 14

connections that mostly emanated from right S1 were correlated with general locomotion 15

time in home cage (edge-wise P < 0.01, cluster-level corrected P = 0.005 and P = 0.037, 16

respectively). However, the overall abnormality in this subnetwork was not associated with 17

MECP2 copy number (r = 0.500, P = 0.450). More details on behavior-related edges are 18

provided in table S10. 19

20

Pathological circuits and symptoms in patients with autism 21

We proceeded to analyze the functional connectivity networks of human subjects using the 22

same strategy as in the monkey experiments (Fig. 6, A and B). Given the substantial 23

heterogeneity and large spectrum of clinical symptoms among ASD patients (26), individuals 24

were first clustered into subgroups with relatively homogenous brain network organization 25


https://doi.org/10.1101/728113

14

(Fig. 6C). Pathological circuits regulating behavioral phenotypes in different autism subtypes 1

were then evaluated via case-control comparison (Fig. 6D) and symptom correlation analysis 2

(Fig. 6E). 3

We used public data from the Autism Brain Imaging Data Exchange (ABIDE), which 4

shares MRI scans and phenotypic information from 539 patients with ASD and 573 typically 5

developing controls (TDCs) (27). After basic demographic and diagnostic screening, 90 6

adolescents with clinical diagnoses of autism and 140 matched TDCs were enrolled (table 7

S11). Covariate-free (including age, gender, full-scale IQ, and scanning sites) connectivity 8

networks were then reconstructed for these subjects based on the same brain parcellation 9

scheme used for monkeys (table S7). The enrolled subjects were then clustered into 10

subgroups based on their rsfMRI connectivity network using community detection on the 11

inter-participant spatial correlation matrix (see Supplementary Materials and Methods). Two 12

clusters were derived: subgroup 1 consisted of 49 autism and 72 TDCs, and subgroup 2 13

consisted of 41 autism and 68 TDCs (fig. S8A, tables S12 and S13). Brain network 14

homogeneity within each subgroup was significantly improved for both autism and TDCs (P 15

< 0.05, fig. S8B). No discernable differences in phenotypic and demographic characteristic 16

were found between these two subgroups of autism (fig. S8C). 17

As in the monkey dataset, we repeated the same statistical analysis for both subgroups in 18

parallel (Fig. 7 and fig. S9). In subgroup 1, 32 significant hypo-connections were observed in 19

27 brain nodes (edge-wise P < 0.001, cluster-level corrected P = 0.026, Fig. 7, B and C, and 20

table S14), most of which were located between temporal cortex (TC) and PFC, between TC 21

and orbitofrontal cortex (OFC), between TC and occipital cortex, and within TC (P < 0.05, 22

Bonferroni correction, Fig. 7D). Regions with the top 25% largest ESn were right PFCpol, 23

right PFCol, bilateral ventral TC (TCv), right TC polar (TCpol), right superior TC (TCs) and 24


https://doi.org/10.1101/728113

15

right primary auditory cortex (A1) (blue bars of interlayer in Fig. 7C). In contrast, no 1

abnormal connections in subgroup 2 withstood correction for multiple comparisons (fig. S9). 2

To map different dimensions/domains of autistic symptoms (as assessed by the Autism 3

Diagnostic Observation Schedule, ADOS) onto the connectivity fingerprint in subgroup 1, we 4

applied Manhattan distance to quantitatively describe the overall extent of circuit abnormality 5

in individual autism relative to an averaged TDC (22, 26) (see Supplementary Materials and 6

Methods). In 29 of 49 participants with research-reliable ADOS scores, we observed a 7

significant correlation between communication scores and their Manhattan distances 8

(Spearman’s r = 0.416, P = 0.025), but not with social interaction or restricted, repetitive 9

behavior (Fig. 7E and table S15). No subnetworks underlying specific symptoms were further 10

derived from the disrupted network. 11

12

Cross-species comparison of connectivity fingerprints 13

After independently evaluating functional abnormalities in monkey and human networks, we 14

next asked whether any relationship existed between the two (Fig. 6F). Although no 15

abnormal connections were shared between transgenic monkeys and subgroup 1 of patients 16

with autism (Figs. 5C and 7C), there were overlaps in abnormal brain nodes (Fig. 8A), which 17

include lateral prefrontal areas (left PFCcl, right PFCvl), lateral orbitofrontal areas (bilateral 18

PFCol), left temporal areas (TCi, TCv and TCpol), left cingulate (CCp and CCr), left PHC, 19

left amygdala, right inferior parietal area, and right primary visual area (V1). Comparing to 20

this subset of patients, monkeys exhibited further abnormality in other subdivisions of PFC 21

(bilateral PFCdl, left PFCpol, and right PFCcl, PFCdm and PFCm) and OFC (bilateral PFCoi 22

and PFCom), FEF, central TC (TCc), PCm, S1, and hippocampus. Abnormal brain areas 23

specific to the patient subgroup include bilateral superior and right medial TC (TCs and 24

TCm), bilateral insula, bilateral A1, and left V1 and bilateral secondary visual areas (V2). 25


https://doi.org/10.1101/728113

16

We further conducted cross-species comparison at network level by calculating pairwise 1

spatial correlations between group-different monkey networks and group-different human 2

networks (lower triangle of effect size matrices in Figs. 5B and 7B, see Supplementary 3

Materials and Methods). The connectivity fingerprint of monkeys substantially correlated 4

only with human subgroup 1 (r = -0.137, P = 8.75×10-20, Fig. 8B), but not with subgroup 2 (r 5

= 0.013, P = 0.408, fig. S10). To further test whether such cross-species similarity was 6

uniform across different parts of the brain, we divided the whole brain circuitry into eight 7

lobes/regions and computed their corresponding spatial correlations. As plotted in Fig. 8C, 8

correspondence of connectivity fingerprints between PFC and temporal, parietal, cingulate, 9

insula and subcortical areas, and between temporal and parietal, cingulate, occipital and 10

subcortical areas were remarkably high between the two primates (P < 0.05, Bonferroni 11

correction, table S16). Taking all between-lobe/regions spatial correlations into account, we 12

found that homologous networks between TG monkeys and autistic patients largely center on 13

PFC and temporal lobes, both of which show strong cross-species correlation with five other 14

lobes/regions (Fig. 8D). 15

16

Discussion 17

MECP2-related GABA dysfunction and beta desynchronization 18

The frequency-dependent finding in electrophysiological activity likely reflects MECP2-19

induced dysfunction in specific molecular pathways. Evidence from transgenic rodents has 20

demonstrated that MECP2 dysfunction alters synchrony and the overall excitation/inhibition 21

balance in brain circuits (28), with greater influence on GABAergic neurons (29). The 22

present transcriptome and functional enrichment analysis of macaque neocortex reveals a 23

strong association between MECP2 coexpressed genes and GABA function, implying altered 24

GABAergic action in the mutant monkeys. Moreover, results from both intracellular 25


https://doi.org/10.1101/728113

17

recording and genetic association studies suggest that generation of cortical beta oscillation 1

depends on GABAergic neurotransmission (30, 31). As such, we propose that the disrupted 2

beta synchronization observed in transgenic monkeys is very likely a consequence of the 3

MECP2-induced malfunction of GABAergic neurons. Similar dysfunction of GABAergic 4

signaling (32) and reduction in beta synchronization during resting state (33) have been 5

reported in ASD patients. Although no associations between spontaneous beta coherence and 6

behavioral symptoms have been reported in human patients thus far, this may be a potential 7

diagnostic biomarker for autism which merits future validation. 8

9

Circuit and behavior disruptions in transgenic monkeys and autistic humans 10

The prevailing hypothesis of disrupted cortical connectivity, in which deficiencies in the way 11

the brain coordinates and synchronizes activity among different regions account for clinical 12

symptoms of ASD, is central to the etiology and accurate diagnosis of ASD (26, 34-36). 13

However, human heterogeneity in terms of complex genetic backgrounds, diverse clinical 14

comorbidities, different genders, various developmental trajectories and medication status is 15

a general issue and formidable challenge, causing diverse and contradictory EEG (35, 37, 38) 16

and imaging findings of ASD (26, 27, 34). Therefore, in this study, we first stratified all 17

enrolled subjects through a data-driven approach to improve the inter-participant 18

homogeneity in whole-brain connectivity patterns. Disease-related disruptions in brain 19

networks were then assessed via case-control comparison in subgroups (39). As a result, a 20

hypo-connected network mainly located in prefrontal and temporal areas was identified in 21

one subgroup. Several brain areas underlying language processing and verbal communication 22

were found disrupted in this autism subgroup, including primary auditory and visual areas, 23

and superior and medial temporal areas (40). The biological significance of the identified 24

circuit was supported by the association between its overall disruption and communication 25


https://doi.org/10.1101/728113

18

deficits in the autism sample. 1

Cognitive flexibility deficits are proposed as potential psychological constructs 2

underpinning major autistic domains, including restricted and repetitive behavior, atypical 3

social interaction and abnormal communication (18, 41). In remarkable consistency with our 4

observation in monkeys, ASD patients are impaired at maintaining newly-rewarded responses 5

and inclined to revert back to previously-reinforced choices, which partially explains the 6

increased severity of restricted and repetitive behaviors (42). Interestingly, MECP2 copy 7

number positively correlates with the increased rate of regressive errors, suggesting a gene 8

dosage-dependent severity of the phenotype in these monkeys similar to that observed in 9

human autism (10). Our circuit-behavior analysis further pins down a pertinent distributed 10

network involving medial, orbital and lateral PFC regions, premotor, anterior and rostral 11

cingulate, inferior parietal, hippocampal and parahippocampal areas, responsible for their 12

performance in the reversal task. This finding is compatible with recent fMRI results in ASD 13

patients performing a modified reversal learning task, where reduced activation was found in 14

frontal and parietal networks that support flexible choices (43). Future studies are needed to 15

investigate the role of cognitive flexibility in autistic etiology, especially its link to social 16

communication and repetitive behavior. 17

18

Cross-species neural mappings 19

We proposed a new cross-species comparison strategy (as illustrated in Fig. 6) involving 20

human stratification and neural mappings from monkeys to human subgroups. Patient 21

stratification before cross-species mapping is an important prerequisite because within the 22

umbrella term “ASD” there exist very different subtypes of patients in this ABIDE large 23

repository, meanwhile the ‘animal model’ is not expected to resemble the human disorder in 24

every respect (7). As described above, data-driven stratification whereby multiple subgroups 25


https://doi.org/10.1101/728113

19

are biologically defined, should reveal multiple networks in parallel that may underlie 1

different symptomatic domains in autism (39). Fortunately, we found that one of two derived 2

subgroups demonstrated substantial overlapping abnormal brain regions with the monkey 3

model (Fig. 8A), among which the lateral orbitofrontal cortex plays a vital role in disrupted 4

circuits in both species (Fig. 5C and Fig. 7C). 5

It is somewhat surprising to observe a lack of shared connections between transgenic 6

monkeys and autistic patients, even though the dysfunctional connectivity profiles of the two 7

species were significantly correlated. As a matter of fact, it remains essentially unknown how 8

exactly the brain circuits of primate species correspond to each other, due to evolutionary 9

effects (7, 22). Given the substantial variations in functional connections subserving species-10

specific behavioral and cognitive adaptations (7, 44), topological characteristics of brain 11

regions are more likely to be evolutionarily conserved. This assumption is supported as cross-12

species similarity in dysconnectivity profiles was more prominent in prefrontal and temporal 13

cortices, thus implying etiological convergence between the monkey model and autism 14

patients. Nevertheless, it is rather intriguing that shared abnormalities in disconnected circuits 15

are tightly associated with cognitive flexibility deficits in monkeys whereas communication 16

deficits in humans, respectively. This “bent” circuit-behavioral mapping at two ends may 17

indicate a novel mechanism underlying social communication deficits in autism, where a 18

specific neurocognitive defect – preference for old cognitive rules or failure to maintain new 19

cognitive rules– disturbs executive function (41). 20

21

Conclusions 22

We present the first neurophysiological and neuroimaging evidence in genetically-engineered 23

macaque monkeys that genetic variants in MECP2 may predispose individuals to abnormal 24

locomotive and cognitive behaviors through dysregulation of neural and functional 25


https://doi.org/10.1101/728113

20

connectivity with prefrontal circuits. MECP2 duplication-induced effect on neural 1

connectivity of the primate model is both temporally (beta frequency range, 12-30 Hz) and 2

spatially (fronto-parieto-occipital network, prefrontal and cingulate network) dependent. 3

More importantly, dysfunctional connectivity profiles induced by MECP2 duplication are 4

mapped onto a subgroup of autistic individuals sharing similar profiles of brain circuits. 5

Taken together, all present examinations in nonhuman primates that are conveniently adapted 6

to human subjects not only hold crucial implications for accurate diagnosis of autism-related 7

disorders, but also offer new insights into the development of targeted behavioral 8

interventions that can improve atypical locomotion and social communication in autism-9

related disorders. 10

11

Supplementary Materials 12

Materials and Methods 13

Supplementary References 14

Fig. S1. Heatmap for the expression of MECP2 top correlated genes in the cortical regions of 15

MECP2 transgenic monkeys. 16

Fig. S2. Behavior results of color discrimination and reversal learning task. 17

Fig. S3. Topography of relative power differences between TG and WT monkeys for six 18

frequency bands. 19

Fig. S4. Strength of connections with significantly decreased beta synchronization in TG 20

monkeys. 21

Fig. S5. Averaged neural synchronization matrices for both TG and WT monkeys for six 22

frequency bands. 23

Fig. S6. Scatter plot of abnormal connections in transgenic monkeys. 24

Fig. S7. Connectivity fingerprint of individual transgenic monkeys. 25


https://doi.org/10.1101/728113

21

Fig. S8. Parsing heterogeneity in clinical cohorts. 1

Fig. S9. Abnormal functional connections of human autism in subgroup 2. 2

Fig. S10. Spatial correlation between entire effect size (ES) matrices of TG versus WT 3

monkeys and human autism versus TDC in subgroup 2. 4

Table S1. MECP2 copy number and locomotion of 5 TG and 16 WT monkeys in home cage. 5

Table S2. Locomotion of 5 TG and 7 WT monkeys in peer separation test with familiar peers. 6

Table S3. Locomotion of 5 TG and 7 WT monkeys in peer separation test with unfamiliar 7

peers. 8

Table S4. EEG data collection from 5 TG and 16 WT monkeys. 9

Table S5. Spatial location and anatomical label of all recording electrodes. 10

Table S6. Connections with decreased beta synchronization in TG monkeys. 11

Table S7. Cortical and subcortical parcellation and abbreviations. 12

Table S8. MRI data collection from 5 TG and 11 WT monkeys. 13

Table S9. Abnormal functional connections in transgenic monkeys. 14

Table S10. Group different edges correlating with behavioral measures. 15

Table S11. Characteristics of all human participants. 16

Table S12. Characteristics of human participants in subgroup 1. 17

Table S13. Characteristics of human participants in subgroup 2. 18

Table S14. Disrupted functional connections in human autism of subgroup 1. 19

Table S15. Correlations between connectivity fingerprint of human autism in subgroup 1 and 20

symptom severity. 21

Table S16. Correlations of group differences in connectivity networks between monkeys and 22

humans (subgroup 1). 23

Movie S1. Example of repetitive circular routing behaviors. 24

Movie S2. Example of repetitive tumbling behaviors. 25


https://doi.org/10.1101/728113

22

Movie S3. Example of repetitive cyclic routing in the path of a figure eight. 1

Movie S4. An example of monkey (TG04) performing stimulus-directed touching task. 2

Movie S5. An example of monkey (TG04) performing discrimination task. 3

Movie S6. An example of monkey (TG08) performing reversal learning task. 4

Movie S7. An example of monkey (WT032) performing reversal learning task. 5

6

References and Notes 7

1. G. Xu, L. Strathearn, B. Liu, W. Bao, Prevalence of autism spectrum disorder among 8

US children and adolescents, 2014-2016. JAMA 319, 81-82 (2018). 9

2. A. F. Marquand, T. Wolfers, M. Mennes, J. Buitelaar, C. F. Beckmann, Beyond 10

lumping and splitting: A review of computational approaches for stratifying 11

psychiatric disorders. Biol. Psychiatry Cogn. Neurosci. Neuroimaging 1, 433-447 12

(2016). 13

3. M. V. Lombardo, M.-C. Lai, S. Baron-Cohen, Big data approaches to decomposing 14

heterogeneity across the autism spectrum. Mol. Psychiatry advance online 15

publication, (2019). 16

4. T. Insel, B. Cuthbert, M. Garvey, R. Heinssen, D. S. Pine, K. Quinn, C. Sanislow, P. 17

Wang, Research Domain Criteria (RDoC): Toward a new classification framework 18

for research on mental disorders. Am J Psychiatr 167, 748-751 (2010). 19

5. A. T. Drysdale, L. Grosenick, J. Downar, K. Dunlop, F. Mansouri, Y. Meng, R. N. 20

Fetcho, B. Zebley, D. J. Oathes, A. Etkin, A. F. Schatzberg, K. Sudheimer, J. Keller, 21

H. S. Mayberg, F. M. Gunning, G. S. Alexopoulos, M. D. Fox, A. Pascual-Leone, H. 22

U. Voss, B. J. Casey, M. J. Dubin, C. Liston, Resting-state connectivity biomarkers 23

define neurophysiological subtypes of depression. Nat. Med. 23, 28-38 (2017). 24


https://doi.org/10.1101/728113

23

6. M. Sahin, M. Sur, Genes, circuits, and precision therapies for autism and related 1

neurodevelopmental disorders. Science 350, aab3897 (2015). 2

7. M. P. van den Heuvel, E. T. Bullmore, O. Sporns, Comparative connectomics. Trends 3

Cogn. Sci. 20, 345-361 (2016). 4

8. A. Bertero, A. Liska, M. Pagani, R. Parolisi, M. E. Masferrer, M. Gritti, M. 5

Pedrazzoli, A. Galbusera, A. Sarica, A. Cerasa, M. Buffelli, R. Tonini, A. Buffo, C. 6

Gross, M. Pasqualetti, A. Gozzi, Autism-associated 16p11.2 microdeletion impairs 7

prefrontal functional connectivity in mouse and human. Brain 141, 2055-2065 (2018). 8

9. Z. Liu, X. Li, J. T. Zhang, Y. J. Cai, T. L. Cheng, C. Cheng, Y. Wang, C. C. Zhang, 9

Y. H. Nie, Z. F. Chen, W. J. Bian, L. Zhang, J. Xiao, B. Lu, Y. F. Zhang, X. D. 10

Zhang, X. Sang, J. J. Wu, X. Xu, Z. Q. Xiong, F. Zhang, X. Yu, N. Gong, W. H. 11

Zhou, Q. Sun, Z. Qiu, Autism-like behaviours and germline transmission in 12

transgenic monkeys overexpressing MeCP2. Nature 530, 98-102 (2016). 13

10. M. B. Ramocki, Y. J. Tavyev, S. U. Peters, The MECP2 duplication syndrome. Am. J. 14

Med. Genet. A. 152A, 1079-1088 (2010). 15

11. R. C. Samaco, C. Mandel-Brehm, C. M. McGraw, C. A. Shaw, B. E. McGill, H. Y. 16

Zoghbi, Crh and Oprm1 mediate anxiety-related behavior and social approach in a 17

mouse model of MECP2 duplication syndrome. Nat. Genet. 44, 206-211 (2012). 18

12. P. G. Judge, D. W. Evans, K. K. Schroepfer, A. C. Gross, Perseveration on a reversal-19

learning task correlates with rates of self-directed behavior in nonhuman primates. 20

Behav. Brain Res. 222, 57-65 (2011). 21

13. M. J. Buckley, F. A. Mansouri, H. Hoda, M. Mahboubi, P. G. Browning, S. C. Kwok, 22

A. Phillips, K. Tanaka, Dissociable components of rule-guided behavior depend on 23

distinct medial and prefrontal regions. Science 325, 52-58 (2009). 24


https://doi.org/10.1101/728113

24

14. B. Jones, M. Mishkin, Limbic lesions and the problem of stimulus--reinforcement 1

associations. Exp. Neurol. 36, 362-377 (1972). 2

15. F.-X. Neubert, Rogier B. Mars, Adam G. Thomas, J. Sallet, Matthew F. S. 3

Rushworth, Comparison of human ventral frontal cortex areas for cognitive control 4

and language with areas in monkey frontal cortex. Neuron 81, 700-713 (2014). 5

16. R. Gil-da-Costa, G. R. Stoner, R. Fung, T. D. Albright, Nonhuman primate model of 6

schizophrenia using a noninvasive EEG method. Proc. Natl. Acad. Sci. U. S. A. 110, 7

15425-15430 (2013). 8

17. W. T. McKinney, Jr, S. J. Suomi, H. F. Harlow, Repetitive peer separations of 9

juvenile-age rhesus monkeys. Arch. Gen. Psychiatry 27, 200-203 (1972). 10

18. H. M. Geurts, B. Corbett, M. Solomon, The paradox of cognitive flexibility in autism. 11

Trends Cogn. Sci. 13, 74-82 (2009). 12

19. D. C. Van Essen, Surface-based approaches to spatial localization and registration in 13

primate cerebral cortex. Neuroimage 23, S97-S107 (2004). 14

20. A. C. Snyder, M. J. Morais, C. M. Willis, M. A. Smith, Global network influences on 15

local functional connectivity. Nat. Neurosci. 18, 736-743 (2015). 16

21. M. Vinck, R. Oostenveld, M. van Wingerden, F. Battaglia, C. M. Pennartz, An 17

improved index of phase-synchronization for electrophysiological data in the presence 18

of volume-conduction, noise and sample-size bias. Neuroimage 55, 1548-1565 19

(2011). 20

22. R. B. Mars, L. Verhagen, T. E. Gladwin, F. X. Neubert, J. Sallet, M. F. Rushworth, 21

Comparing brains by matching connectivity profiles. Neurosci. Biobehav. Rev. 60, 22

90-97 (2016). 23

23. J. Richiardi, A. Altmann, A. C. Milazzo, C. Chang, M. M. Chakravarty, T. 24

Banaschewski, G. J. Barker, A. L. Bokde, U. Bromberg, C. Buchel, P. Conrod, M. 25


https://doi.org/10.1101/728113

25

Fauth-Buhler, H. Flor, V. Frouin, J. Gallinat, H. Garavan, P. Gowland, A. Heinz, H. 1

Lemaitre, K. F. Mann, J. L. Martinot, F. Nees, T. Paus, Z. Pausova, M. Rietschel, T. 2

W. Robbins, M. N. Smolka, R. Spanagel, A. Strohle, G. Schumann, M. Hawrylycz, J. 3

B. Poline, M. D. Greicius, I. consortium, BRAIN NETWORKS. Correlated gene 4

expression supports synchronous activity in brain networks. Science 348, 1241-1244 5

(2015). 6

24. G. Z. Wang, T. G. Belgard, D. Mao, L. Chen, S. Berto, T. M. Preuss, H. Lu, D. H. 7

Geschwind, G. Konopka, Correspondence between resting-state activity and brain 8

gene expression. Neuron 88, 659-666 (2015). 9

25. D. J. Sheskin, Handbook of Parametric and Nonparametric Statistical Procedures: 10

Third Edition. (CRC Press, 2003). 11

26. A. Hahamy, M. Behrmann, R. Malach, The idiosyncratic brain: Distortion of 12

spontaneous connectivity patterns in autism spectrum disorder. Nat. Neurosci. 18, 13

302-309 (2015). 14

27. A. Di Martino, C. G. Yan, Q. Li, E. Denio, F. X. Castellanos, K. Alaerts, J. S. 15

Anderson, M. Assaf, S. Y. Bookheimer, M. Dapretto, B. Deen, S. Delmonte, I. 16

Dinstein, B. Ertl-Wagner, D. A. Fair, L. Gallagher, D. P. Kennedy, C. L. Keown, C. 17

Keysers, J. E. Lainhart, C. Lord, B. Luna, V. Menon, N. J. Minshew, C. S. Monk, S. 18

Mueller, R. A. Muller, M. B. Nebel, J. T. Nigg, K. O'Hearn, K. A. Pelphrey, S. J. 19

Peltier, J. D. Rudie, S. Sunaert, M. Thioux, J. M. Tyszka, L. Q. Uddin, J. S. 20

Verhoeven, N. Wenderoth, J. L. Wiggins, S. H. Mostofsky, M. P. Milham, The autism 21

brain imaging data exchange: Towards a large-scale evaluation of the intrinsic brain 22

architecture in autism. Mol. Psychiatry 19, 659-667 (2014). 23

28. H. Lu, R. T. Ash, L. He, S. E. Kee, W. Wang, D. Yu, S. Hao, X. Meng, K. Ure, A. 24

Ito-Ishida, B. Tang, Y. Sun, D. Ji, J. Tang, B. R. Arenkiel, S. M. Smirnakis, H. Y. 25


https://doi.org/10.1101/728113

26

Zoghbi, Loss and gain of MeCP2 cause similar hippocampal circuit dysfunction that 1

is rescued by deep brain stimulation in a Rett syndrome mouse model. Neuron 91, 2

739-747 (2016). 3

29. H. T. Chao, H. Chen, R. C. Samaco, M. Xue, M. Chahrour, J. Yoo, J. L. Neul, S. 4

Gong, H. C. Lu, N. Heintz, M. Ekker, J. L. Rubenstein, J. L. Noebels, C. Rosenmund, 5

H. Y. Zoghbi, Dysfunction in GABA signalling mediates autism-like stereotypies and 6

Rett syndrome phenotypes. Nature 468, 263-269 (2010). 7

30. M. A. Whittington, H. J. Faulkner, H. C. Doheny, R. D. Traub, Neuronal fast 8

oscillations as a target site for psychoactive drugs. Pharmacol. Ther. 86, 171-190 9

(2000). 10

31. B. Porjesz, L. Almasy, H. J. Edenberg, K. Wang, D. B. Chorlian, T. Foroud, A. 11

Goate, J. P. Rice, S. J. O'Connor, J. Rohrbaugh, S. Kuperman, L. O. Bauer, R. R. 12

Crowe, M. A. Schuckit, V. Hesselbrock, P. M. Conneally, J. A. Tischfield, T.-K. Li, 13

T. Reich, H. Begleiter, Linkage disequilibrium between the beta frequency of the 14

human EEG and a GABAA receptor gene locus. Proc. Natl. Acad. Sci. U. S. A. 99, 15

3729-3733 (2002). 16

32. C. E. Robertson, E. M. Ratai, N. Kanwisher, Reduced GABAergic action in the 17

autistic brain. Curr. Biol. 26, 80-85 (2016). 18

33. G. Shou, M. W. Mosconi, J. Wang, L. E. Ethridge, J. A. Sweeney, L. Ding, 19

Electrophysiological signatures of atypical intrinsic brain connectivity networks in 20

autism. J. Neural Eng. 14, 046010 (2017). 21

34. C. Ecker, S. Y. Bookheimer, D. G. Murphy, Neuroimaging in autism spectrum 22

disorder: Brain structure and function across the lifespan. Lancet. Neurol. 14, 1121-23

1134 (2015). 24


https://doi.org/10.1101/728113

27

35. K. Kessler, R. A. Seymour, G. Rippon, Brain oscillations and connectivity in autism 1

spectrum disorders (ASD): New approaches to methodology, measurement and 2

modelling. Neurosci. Biobehav. Rev. 71, 601-620 (2016). 3

36. P. J. Uhlhaas, W. Singer, Neuronal dynamics and neuropsychiatric disorders: Toward 4

a translational paradigm for dysfunctional large-scale networks. Neuron 75, 963-980 5

(2012). 6

37. M. E. Vissers, M. X. Cohen, H. M. Geurts, Brain connectivity and high functioning 7

autism: A promising path of research that needs refined models, methodological 8

convergence, and stronger behavioral links. Neurosci. Biobehav. Rev. 36, 604-625 9

(2012). 10

38. N. David, T. R. Schneider, I. Peiker, R. Al-Jawahiri, A. K. Engel, E. Milne, 11

Variability of cortical oscillation patterns: A possible endophenotype in autism 12

spectrum disorders? Neurosci. Biobehav. Rev. 71, 590-600 (2016). 13

39. E. Loth, W. Spooren, L. M. Ham, M. B. Isaac, C. Auriche-Benichou, T. 14

Banaschewski, S. Baron-Cohen, K. Broich, S. Bölte, T. Bourgeron, T. Charman, D. 15

Collier, F. de Andres-Trelles, S. Durston, C. Ecker, A. Elferink, M. Haberkamp, R. 16

Hemmings, M. H. Johnson, E. J. H. Jones, O. S. Khwaja, S. Lenton, L. Mason, V. 17

Mantua, A. Meyer-Lindenberg, M. V. Lombardo, L. O'Dwyer, K. Okamoto, G. J. 18

Pandina, L. Pani, A. M. Persico, E. Simonoff, S. Tauscher-Wisniewski, J. Llinares-19

Garcia, S. Vamvakas, S. Williams, J. K. Buitelaar, D. G. M. Murphy, Identification 20

and validation of biomarkers for autism spectrum disorders. Nat. Rev. Drug. Discov. 21

15, 70 (2015). 22

40. D. G. Amaral, C. M. Schumann, C. W. Nordahl, Neuroanatomy of autism. Trends 23

Neurosci. 31, 137-145 (2008). 24

41. M. C. Lai, M. V. Lombardo, S. Baron-Cohen, Autism. Lancet 383, 896-910 (2013). 25


https://doi.org/10.1101/728113

28

42. H. L. Miller, M. E. Ragozzino, E. H. Cook, J. A. Sweeney, M. W. Mosconi, 1

Cognitive set shifting deficits and their relationship to repetitive behaviors in autism 2

spectrum disorder. J. Autism Dev. Disord. 45, 805-815 (2015). 3

43. A. M. D'Cruz, M. W. Mosconi, M. E. Ragozzino, E. H. Cook, J. A. Sweeney, 4

Alterations in the functional neural circuitry supporting flexible choice behavior in 5

autism spectrum disorders. Transl. Psychiatry 6, e916 (2016). 6

44. R. B. Mars, S. N. Sotiropoulos, R. E. Passingham, J. Sallet, L. Verhagen, A. A. 7

Khrapitchev, N. Sibson, S. Jbabdi, Whole brain comparative anatomy using 8

connectivity blueprints. Elife 7, (2018). 9

10


https://doi.org/10.1101/728113

29

Acknowledgments 1

We would like to thank Hu Zhang, Ganxian Wang, Qinying Jiang and Wenwen Yu for their 2

assistance to monkey training and data acquisition, and also thank Drs. Trevor Robbins, 3

Freund Tamas, Charles Schroeder, Anna Roe, Ravi Menon, Stefan Everling and Muming Poo 4

for their stimulating discussions and suggestions during the preparation of this study. 5

Funding: This work was supported by the National Key R&D Program of China (No. 6

2017YFC1310400), the Strategic Priority Research Program of Chinese Academy of Science 7

(No. XDB32000000), grants from National Natural Science Foundation (81571300, 8

81527901, 31771174), Natural Science Foundation and Major Basic Research Program of 9

Shanghai (No. 16JC1420100), and Shanghai Municipal Science and Technology Major 10

Project (No. 2018SHZDZX05). Author contribution: D.-C.C., Z.-W.W., S.-Y.Y., and X.C., 11

with the help of Y.W., J.C. and S.G., conducted the EEG experiments. D.-C.C., with the help 12

of Z.W., analyzed the EEG data. Z.-W.W., T.-T.B., D.-C.C., and S.-Y.Y., with the help of 13

J.C. and Z.W., conducted animal MRI experiments. Z.-W.W., with the help of Y.-F.Z. and 14

Z.W., analyzed animal MRI data. T.-T.B., Y.-L.L. and S.-Y.Y., with the help of Z.Q., K.Z., 15

Z.Y. and Z.W., designed animal behavioral experiments. T.-T.B. and Y.-L.L. conducted the 16

reversal learning task and analyzed the behavioral data. S.-Y.Y. conducted the home cage 17

observation and peer separation test. S.-Y.Y. and D.-C.C. analyzed the locomotive behaviors. 18

Z.-W.W., with the help of D.-C.C., Y.-F.Z., X.X., G.-Z.W., Z.Y., Y.D. and Z.W., analyzed 19

the human data and cross-species comparison. Z.L., with the help of J.Z., conducted the gene 20

coexpression and enrichment analysis. Z.Q. and Q.S. generated the transgenic monkeys. 21

X.X., Y.D., and G.-Z.W. contributed to data interpretation. Z.W., Z.Y., and S.G. conceived 22

and supervised the project. Z.W. wrote the manuscript with the help of Z.Y., S.G., D.-C.C., 23

Z.-W.W., and K.Z. All authors read and approved the manuscript. 24

25


https://doi.org/10.1101/728113

30

Figure Legends 1

Fig. 1. WGCNA and functional enrichment analysis. (A) Overview of weighted gene 2

coexpression network analysis using public transcriptional data of 182 macaque cortical 3

samples, and functional enrichment analysis on identified gene module coexpressed with 4

MECP2. (B) Network analysis dendrograms representing assignment of 13,888 genes to 17 5

modules. (C) MECP2 coexpression network. Directed neighbors of MECP2 (n = 5) are 6

represented with filled circles; genes enriched in GABA-related pathways (n = 5) are 7

represented with filled squares; the top 20 genes with larger node degree are represented with 8

empty circles; and the other second-order neighbors of MECP2 are represented with smaller 9

gray circles. Pink and purple colors indicate upregulation and downregulation in MECP2 10

transgenic monkeys, respectively. Dark gray lines (edges) indicate the path from MECP2 to 11

GABA associated genes. (D) Four human diseases, four gene pathways, and three cellular 12

components are significantly enriched in constrained genes (FDR-corrected P < 0.01). The 13

enrichment P-values were calculated and corrected using ToppGene Suite. 14

15

Fig. 2. Excessive locomotive behaviors during home cage observation and peer 16

separation test in TG monkeys. (A) Peer separation paradigm. The test monkey was first 17

transferred from the home cage to a two-compartment test cage and stayed alone for 20 min. 18

A peer monkey was then introduced into the other compartment for 40 min, during which the 19

two monkeys were able to have visual, auditory and olfactory interactions. The test monkey 20

stayed alone for another 20 min after the peer monkey was removed. Home cage observation 21

and peer separation test were carried out on separate days. (B) Locomotion in home cage in 22

terms of general locomotion time (upper panel) and repetitive index (lower panel). (C) 23

Locomotion in the last 10 min of the adaptation and the first 10 min of the separation period 24


https://doi.org/10.1101/728113

31

in peer separation test. (D) Induced change in locomotive behaviors after peer separation. *** 1

P < 0.0001, ** P < 0.001, * P < 0.05, # P < 0.1. 2

3

Fig. 3. Cognitive flexibility tests in MECP2 mutants and WT monkeys. (A) Temporal 4

sequence of events in the color discrimination and reversal learning task. After subjects 5

maintained a 500 ms hand fixation, two choice buttons (blue and green colors) were 6

presented at randomized positions on the screen across trials. Subjects had to select the 7

correct color within 2000 ms to get a 500 ms reward. RT, reaction time; MT, movement time; 8

ITI, inter-trial interval. (B) An example of task sequence in the reversal learning test. The test 9

included 10 sessions per day, 80 trials per session. The color assigned for reward was pseudo-10

randomly switched between sessions. The inter-session interval (ISI) was about 1 min. Red 11

and black arrows indicate perseverative errors and regressive errors, respectively. G, green-12

reward session; B, blue-reward session. (C) The cumulative regressive error rate of 13

performed trials after rule switch was analyzed (n = 5, TG; n = 4, WT; *P < 0.05, two-sample 14

Student’s t-test). Error bars denote standard error of mean. (D-F) The percentage of 15

regressive errors (D), perseverative errors (E) and number of omitted trials (F) in TG and WT 16

groups in first 20 performed trials after rule switch (Student’s t-test). 17

18

Fig. 4. Decreased beta (12-30 Hz) synchronization in fronto-parieto-occipital networks 19

in TG monkeys. (A) Three-dimensional reconstruction of monkey TG04 wearing the EEG 20

cap based on T2-weighted MRI images. Bright circles indicate electrode rings. Bright ellipses 21

above circles indicate marked electrodes, including the reference (REF), most posterior (Oz) 22

and most lateral (18/24) electrodes. The ground electrode (not shown here) was located far 23

from the cortex beneath the left eye. Six recording channels far from the cortex (including 24

18/24) were disabled for subsequent recording. (B) Spatial organization of 21 active 25


https://doi.org/10.1101/728113

32

electrodes after mapping to the cortical surfaces of the F99 template brain. Brain lobes were 1

displayed in different colors. The electrodes were named after the spatial location. See table 2

S5 for more details. (C) Averaged covariates-free neural synchronization networks of TG 3

(bottom-left) and WT (top-right) monkeys. The numbers indicate the electrode ID as in table 4

S5. The color bar denotes the logarithm of dwPLI. (D) Effect sizes (ESs, Hedge’s g value, 5

bottom-left) and statistical P values (top-right) of statistical comparison between TG and WT 6

groups. Eleven connections widely distributed in fronto-parieto-occipital networks were 7

identified significantly decreased in TG monkeys (edge-wise P < 0.005, cluster-level 8

corrected P = 0.010, more details in fig. S3 and table S6). (E) Topographic distribution of 9

abnormal connections was illustrated in the F99 template brain. (F-G) Association of the 10

overall abnormality in beta synchronization (Manhattan distance) with general locomotion 11

time change after peer separation (F) and MECP2 copy number (G). Associations were 12

evaluated via Spearman’s rank correlation for TG monkeys only. 13

14

Fig. 5. Disrupted neural circuits in transgenic monkeys. (A) Covariate-free connectivity 15

network matrices for TG (bottom-left) and WT (top-right) monkeys. (B) Effect sizes (ESs) of 16

TG versus WT (bottom-left) are shown with corresponding P values (top-right, cluster-level 17

corrected P < 0.05, and edge-wise P < 0.001). Brain nodes are sorted and organized 18

according to the regions/lobes as listed in table S7. (C) Disrupted functional connections and 19

corresponding brain nodes in TG monkeys compared with WT. The red and blue bars in the 20

interlayer indicate positive and negative ES of each brain node (ESn), respectively, which 21

was defined by the sum of the ESs of all abnormal connections to this node. The dashed line 22

labels the top 25% of absolute ESn. The abbreviations and parcellation of brain nodes are 23

listed in table S7. (D) Spatial distribution of disrupted connections across the brain is plotted 24

(bottom-left) and the regions with statistical significance are highlighted (top-right, P < 0.05, 25


https://doi.org/10.1101/728113

33

Bonferroni correction). Inf, infinite; OC, occipital cortex; PC, parietal cortex; TC, temporal 1

cortex; PFC, prefrontal cortex; OFC, orbitofrontal cortex; CC, cingulate cortex; Ins, insula; 2

Sub, subcortical areas. (E) Subnetwork associated with regressive error rate in reversal 3

learning task at the significance level of P < 0.05 after correction for multiple comparisons. 4

(F) Gene-circuit association between MECP2 copy number and circuit abnormality 5

underlying regressive error rate. MD, Manhattan distance; RE, regressive error. (G) Gene-6

behavior association between MECP2 copy number and regressive error rate. Associations 7

were evaluated via Spearman’s rank correlation for TG monkeys only. 8

9

Fig. 6. Schematic of comparative analysis strategy between monkeys and humans. (A) 10

Group-level differences of functional connectivity networks between transgenic (TG) and 11

wild-type (WT) monkeys were statistically compared and their effect sizes were evaluated. 12

(B) The connectivity fingerprint of each TG monkey was identified and evaluated by 13

Manhattan distance (MD) relative to an averaged WT monkey. The relationships between 14

genetic, circuitry and behavioral aberrations were explored. (C) Stratification of human 15

participants was done through data-driven clustering based on the resting-state fMRI 16

connectivity network. (D) Group-level differences of connectivity networks between autism 17

and typically developing controls (TDC) were analyzed for each subgroup using the same 18

strategy as in the monkey experiments. (E) Connectivity fingerprint of individual autism was 19

identified and evaluated by MD. Their relations to the severity of dimensional symptoms 20

assessed by clinical scores were explored. (F) Cross-species similarity of altered functional 21

homologs in TG monkeys and autistic patients was estimated based on (A) and (D). 22

23

Fig. 7. Disrupted neural circuits in a subgroup of patients with autism. (A) Covariate-24

free connectivity network matrices for autism (bottom-left) and TDC (top-right) in subgroup 25


https://doi.org/10.1101/728113

34

1. (B) Effect sizes (ESs) of autism versus TDC (bottom-left) are shown with corresponding P 1

values (top-right, cluster-level P < 0.05, edge-wise P < 0.001). The brain nodes are sorted 2

and organized according to the regions/lobes as listed in table S7. (C) Disrupted functional 3

connections and corresponding brain nodes in patients with autism compared with TDC. The 4

blue bar in the interlayer indicates negative ESn, which was defined as the sum of the ESs of 5

all abnormal connections to this node. The dashed line labels the top 25% of absolute ESn. 6

The abbreviations and parcellation of brain nodes are listed in table S7. (D) Spatial 7

distribution of disrupted connections across the brain is plotted (bottom-left) and highlighted 8

with statistical significance (top-right, P < 0.05, Bonferroni correction). Inf, infinite; OC, 9

occipital cortex; PC, parietal cortex; TC, temporal cortex; PFC, prefrontal cortex; OFC, 10

orbitofrontal cortex; CC, cingulate cortex; Ins, insula; Sub, subcortical areas. (E) 11

Associations between Manhattan distances of each autism patient and the severity of 12

dimensional symptoms in communication (left), social interaction (middle), and stereotyped 13

behaviors (right), respectively. 14

15

Fig. 8. Cross-species comparison of connectivity fingerprints in monkeys and humans. 16

(A) Abnormal brain regions shared between transgenic monkeys and human patients. The 17

size of nodes indicates the total number of abnormal edges connected to the node in both 18

species. OFC, orbitofrontal cortex; TC, temporal cortex; CC, cingulate cortex; Sub, 19

subcortical areas; PFC, prefrontal cortex; OC, occipital cortex; PC, parietal cortex. (B) 20

Spatial correlation between the entire effect size (ES) matrices of TG versus WT monkeys 21

(shown in Fig. 5B) and patients with autism versus TDC in subgroup 1 (shown in Fig. 7B). 22

(C) Spatial correlation of ES matrices in different brain lobes between monkeys and humans. 23

Each scatter plot displays statistically significant correlations in different lobes between the 24

two primates (P < 0.05, Bonferroni correction), as summarized in (D). The size of brain lobes 25


https://doi.org/10.1101/728113

35

in the spider plot indicates the sum of all significant correlation coefficients of ESs between 1

the lobe and all other lobes of the two primates. 2

3


https://doi.org/10.1101/728113


https://doi.org/10.1101/728113

1 mecp2 duplication causes aberrant gaba pathways ...€¦ · 3 recapitulate typical phenotypes in...

Documents