Altered Forebrain Functional Connectivity and Neurotransmission in a Kinase-Inactive Met Mouse Model of Autism

Abstract

MET, the gene encoding the tyrosine kinase receptor for hepatocyte growth factor, is a susceptibility gene for autism spectrum disorder (ASD). Genetically altered mice with a kinase-inactive Met offer a potential model for understanding neural circuit organization changes in autism. Here, we focus on the somatosensory thalamocortical circuitry because distinct somatosensory sensitivity phenotypes accompany ASD, and this system plays a major role in sensorimotor and social behaviors in mice. We employed resting-state functional magnetic resonance imaging and in vivo high-resolution proton MR spectroscopy to examine neuronal connectivity and neurotransmission of wild-type, heterozygous Met–Emx1, and fully inactive homozygous Met–Emx1 mice. Met–Emx1 brains showed impaired maturation of large-scale somatosensory network connectivity when compared with wild-type controls. Significant sex × genotype interaction in both network features and glutamate/gamma-aminobutyric acid (GABA) balance was observed. Female Met–Emx1 brains showed significant connectivity and glutamate/GABA balance changes in the somatosensory thalamocortical system when compared with wild-type brains. The glutamate/GABA ratio in the thalamus was correlated with the connectivity between the somatosensory cortex and the thalamus in heterozygous Met–Emx1 female brains. The findings support the hypothesis that aberrant functioning of the somatosensory thalamocortical system is at the core of the conspicuous somatosensory behavioral phenotypes observed in Met–Emx1 mice.

Keywords

autism resting-state functional magnetic resonance imaging magnetic resonance spectroscopy brain network somatosensory thalamocortical system met null mice GABA

Introduction

MET (also known as protein product of the c-MET proto-oncogene) is a receptor tyrosine kinase, which binds the ligand hepatocyte growth factor (HGF).¹ Decreased MET function is associated with an increased risk of autism spectrum disorder (ASD).^2

–5 MET transcript expression is prominent in the cerebral cortex, hippocampus, and amygdala.⁶ The changes in Met signaling during development could affect neuronal number as well as the complexity of the neuropil,^7
–9 consequently altering the structure or connectivity in the brain and ultimately impacting its function. In a recent neuroimaging study,¹⁰ when children and adolescents with autism-associated promoter variant in MET were clustered into homozygous and heterozygous groups, the MET risk allele had strong impact across individuals within the heterozygous group. The results suggested that MET risk genotype should be taken into account to stratify individuals with ASD. Human neuroimaging studies likely reflect the effects of multiple different changes at the genetic level, as reflected in the considerable heterogeneity observed between studies.

Impaired somatosensory processing and subsequent associated deficits in motor skills, interpersonal relations, social skills, and communication are well-documented in neurodevelopmental disorders, including ASD.¹¹ The term “tactile defensiveness” has emerged to depict the tactile hypo- or hypersensitivity of children with ASD. Mouse models offer a particularly useful paradigm to study the tactile sensitivity deficits. Mice rely heavily on tactile discrimination with their whiskers. A large portion of the primary somatosensory cortex is devoted to representation of whiskers by neural modules in a one-to-one and patterned fashion.¹² These neural modules known as “barrels” in the cortex also exist in subcortical structures, such as the sensory thalamus and the trigeminal brainstem, which carry the whisker inputs to the cortex. Thus, the mouse whisker-barrel system is uniquely advantageous to study the development, plasticity, adult organization, and function of the somatosensory system.

Previously, we have used a mouse line in which Met signaling is inactivated specifically in the cerebral cortex and hippocampus of Met–Emx1 mice. This mouse line was generated using a Cre-loxP recombination strategy. Unlike global Met null mutants,¹³ Met–Emx1 mice live to adulthood,¹⁴ allowing examination of brain structure during the postnatal and adult periods. As reported in an earlier anatomical magnetic resonance imaging (MRI) study,¹⁵ the rostral cortex, caudal hippocampus, dorsal striatum, thalamus, and corpus callosum were all larger in adult, but not in juvenile, Met–Emx1 mice relative to wild-type mice. The specificity of the changes suggests that aberrant expansion of the forebrain is consistent with continued axonal and dendritic growth, potentially leading to improper circuit formation and maintenance. In a separate electrophysiological study, we observed an excitatory/inhibitory (E/I) imbalance in somatosensory thalamocortical transmission in an in vitro slice preparation in the same mouse model,¹⁶ consistent with the E/I imbalance theory in autism.¹⁷ Behavioral deficits in Met–Emx1 mice have been reported as well.^18,19 Given the tactile impairments associated with neurodevelopmental disorders, genetic associations of MET with ASD, and our previous data of anatomical differences in structures and electrophysiology, the Met–Emx1 mouse line was chosen to study altered somatosensory responses at the circuit level in the intact live animal.

We employed in vivo high-resolution proton magnetic resonance spectroscopy (¹H-MRS) and resting-state functional magnetic resonance imaging (rsfMRI) to investigate the neurotransmitters (glutamate and gamma-aminobutyric acid [GABA]) and functional connectivity alterations, respectively, within the somatosensory thalamocortical circuitry. The neuroimaging method rsfMRI is used to investigate regional brain interactions when a subject is not performing an explicit task. Resting-state functional MRI signals are thought to arise from spontaneous low-frequency fluctuations in blood oxygen level-dependent (BOLD) signal. The coherence of BOLD signal—also named “functional connectivity”—can be examined using hypothesis-driven seed-based analysis, which is based on the correlation of the resting-state time series of a selected brain region (seed) with the time series of other regions.^20
–22 Functional connectivity detects tightly coupled regions in brain baseline functional systems.^23

–26 Therefore, rsfMRI can display basic brain functional organization, which reflects the formation and maintenance of neural networks.²⁷

In vivo ¹H-MRS is a noninvasive spectroscopic technique capable of measuring neurochemical concentrations in specific regions of the central nervous system.^28
–30 At high field strength (≥7 T) with a short echo time (≤25 ms), the resolution of coupled peaks, such as the CH₂ from glutamate and glutamine and GABA, can be clearly resolved, suggesting the potential of measuring GABA at 2.30 ppm without the need of spectral J-editing.^29,30 The application of in vivo ¹H-MRS provides biochemical information on glutamatergic and GABAergic functions without bias introduced from sample preparation. Alterations of correlation between GABA or glutamate and functional connectivity have been reported in mental disorders or following experimental modulation^31

–34 but not yet in ASD. Here, we provide correlations between neurotransmitter levels and functional connectivity within the somatosensory thalamocortical circuitry in Met–Emx1 mice.

Sets of discrete neural elements linked by connections can be considered a network. An analytical technique—graph theory analysis—is helpful in assessing the brain network properties. A network can be defined in graph theory as a set of nodes or vertices and the edges or lines between them can be quantitatively described by various measures (eg, clustering coefficient, local efficiency, small-worldness [SW]) to assess the network features such as integration, segregation, and resilience.³⁵ This method is used to characterize organization of networks in neuroimaging studies in humans, nonhuman primates, and rats,^36

–39 and it has paved the way for increasingly sophisticated investigations of brain connectivity in the study of ASD.^40,41 Recently, modular structure and hub community were detected in the mouse brain using rsfMRI,^42,43 indicating the presence of a well-organized brain network that shares similar features as primates and rats. In addition to seed-based functional connectivity analysis, we performed a graph theory analysis to compare the transgenic mice with homozygous and heterozygous genotype with wild-type mice in topological organization across brain regions.

Methods

Animals

Founder mice, Met-fx (#016974; Jackson Laboratory, Bar Harbor, Maine⁴⁴) and Emx1-Cre (#005628; Jackson Laboratory), were crossed and maintained on a C57BL/6J background for >30 generations. Mice with a single inactive Met allele, Met-fx/Emx1-Cre, are denoted as Met–Emx1 or heterozygous Met–Emx1, and those with 2 inactive alleles are denoted as Met-Met-Emx1 or homozygous Met–Emx1. Mice used in these experiments were adult male and female littermates from mating nonsibling heterozygotes. Wild-type mice included the Emx1-Cre allele alone, the Met-fx allele alone, and mice lacking any transgenes. In our previous studies, we observed no differences between these groups,^15,16 and therefore, we combined the data from these groups. We imaged 6 groups: wild-type male (n = 12), wild-type female (n = 12), heterozygous Met–Emx1 male (n = 12), heterozygous Met–Emx1 female (n = 12), homozygous Met-Met-Emx1 male (n = 12), and homozygous Met-Met-Emx1 female (n = 12) mice. All mice were adults, more than 90 days of age. All procedures conformed to the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee, University of Maryland, Baltimore.

Image Acquisition

We performed all experiments on a Bruker Biospec 7 T 30-cm horizontal bore scanner (Bruker Biospin MRI GmbH, Ettlingen, Germany) equipped with a BGA12S gradient system and interfaced to a Bruker Paravision 5.1 console. A Bruker 72-mm linear-volume coil served as the transmitter and a Bruker ¹H 4-element surface coil array served as the receiver. Anesthesia was induced using isoflurane (∼1%) with enriched O₂, which is considered appropriate based on previous evidence in rodent models.^45
–47 An MR compatible small-animal monitoring and gating system (SA Instruments, Inc., Stony Brook, New York) monitored the respiration rate and body temperature, which was maintained at 36°C to 37.5°C using a warm water bath circulation.

A 3-slice (axial, mid-sagittal, and coronal) scout image using rapid acquisition with fast low-angle shot localized the mouse brain. A fast shimming procedure (FASTMAP) improved the B₀ homogeneity covering the brain. Anatomic images (repetition time/effective echo time [TR/TE_eff = 5000/18.42 ms]) were obtained using a 2-dimensional rapid acquisition with relaxation enhancement sequence covering the entire brain. Imaging was performed over a 1.75-cm field of view (FOV) in the coronal plane with an in-plane resolution of 146 μm using 18 slices at 1-mm thickness.

Resting-state functional MRI was acquired matching the anatomic images using a single shot, spin echo planar imaging sequence (TR/TE = 1000/27.6 ms) with a 1.75-cm FOV and an in-plane resolution of 273 μm² using 18 slices at 1-mm thickness. Despite a weaker BOLD signal, spin echo provides much less imaging distortion than gradient echo when imaging mouse brains. Therefore, spin echo imaging was used in this study. Figure 1A shows the raw acquisition EPI images from a female mouse. Six hundred repetitions were taken, resulting in a total scanning time of around 10 minutes for each data set. Two rsfMRI sessions were acquired for each animal. During the rsfMRI experiment, we adjusted the isoflurane anesthesia to 0.75% to 1% and the respiratory rate of the mouse was ∼50 breaths/minute without any detectable motion.

Figure 1.

Brain imaging and spectroscopy. Demonstration of the raw EPI images of a female control mouse (A), designation of ROIs for rsfMRI analysis (B), connectivity map of control animals with right-side somatosensory cortex as the seed ROI (C), and representative ¹HMRS spectra from somatosensory cortex (top) and thalamus (bottom) (D). Amy indicates amygdala; Au, auditory cortex; Cg, cingulate cortex; CIC, inferior colliculus; GABA, gamma-aminobutyric acid; Gln, glutamine; Glu, glutamate; Glx, glutamate and glutamine; GSH, glutathione; Hip, hippocampus; hypotha, hypothalamus; Ins, myo-inositol; lCPu/mCPu, lateral/medial caudate putamen; M1/M2, primary/secondary motor cortex; MM, macromolecules; NAA, N-acetyl-aspartate; NAc, nucleus accumbens; Pir, piriform cortex; Pn, pontine nuclei; PnO, rostral pontine reticular nucleus; PrL/IL, prelimbic/infralimbic cortex; RSC, retrosplenial cortex; rsfMRI, resting-state functional magnetic resonance imaging; S1/S2, primary/secondary somatosensory cortex; Tau, taurine; tCho, glycerophosphocholine and phosphocholine; tCr, creatine and phosphocreatine; Tha, thalamus; V, visual cortex; ¹HMRS, high-resolution proton magnetic resonance spectroscopy.

High-resolution proton magnetic resonance spectroscopy data were obtained from the right somatosensory cortex (2.5 × 1 × 1.5 mm³, number of averages = 1000) and the right thalamus (2.0 × 2.5 × 1.5 mm³, number of averages = 600). Prior to acquiring the spectra, the FASTMAP procedure aided the adjustments of all first- and second-order shims over the voxel of interest. After shimming, a typical water line-width was 10 to 11 Hz (0.033–0.037 ppm). This allowed for good separation of the glutamate (γCH₃ 2.35 ppm, αCH 3.75 ppm) and GABA (αCH₂ 2.28 ppm, βCH₂ 1.89 ppm). A short echo time Point-RESolved Spectroscopy (PRESS) pulse sequence (TR/TE = 2500/10 ms) was used for MRS data acquisition.⁴⁸ The unsuppressed water signal from the prescribed voxel was collected as a biochemical concentration reference for each scan. The total in vivo imaging acquisition time was 2.5 hours.

Resting-State Functional MRI Processing

We conducted all rsfMRI image preprocessing and processing using SPM12 (http://www.fil.ion.ucl.ac.uk/spm/) and Analysis of Functional NeuroImages (AFNI) (http://afni.nimh.nih.gov/afni). The processing pipeline included slice-timing correction, motion correction, alignment to a study-specific mouse anatomic template, band-pass filtering, orthogonalization of motion-derived parameters, and smoothing. Specifically, the first 10 and last 50 volumes were excluded from each data set according to the stability of animals and the equipment. Slice-timing correction and motion correction were performed in SPM12 with the ninth slice as the reference slice and second-degree B-spline interpolation as the estimation method. The study-specific mouse anatomical template was created in SPM12 with 2 steps. First, the anatomical image of each mouse was aligned to the anatomical image of a representative mouse with fourth-degree B-spline interpolation. Then, the aligned anatomical images were averaged, yielding the study-specific template. Resting-state functional MRI of each mouse was aligned to the study-specific template with the original anatomic image as the source image in SPM12. Next, the motion-derived parameters were regressed and the data were band-pass filtered (0.01–0.1 Hz) and smoothed (Full width at half maximum [FWHM] = 0.6 mm) with 3dBandpass function in AFNI.

Thirty-seven regions of interest (ROIs) were manually defined based on the standard Paxinos and Watson mouse brain atlas⁴⁹ (Figure 1B). For each animal, the regionally averaged BOLD time series was extracted from each ROI, and then these extracted time courses of ROIs were correlated using the Pearson correlation coefficient, producing a 37 × 37 correlation matrix. A separate correlation matrix was generated each session for each mouse. A Fisher transformation yielded the z-score association matrices. Two sessions of z-score matrices were averaged for graph theory analysis.

Graph Theory Analysis

Graph theory analysis was performed in MATLAB (Ver. R2014a; MathWorks, Inc. Natick, Massachusetts, USA)-based program graph analysis toolbox (https://www.nitrc.org/projects/gat/). Binary adjacency matrices were derived by thresholding the 37 × 37 association matrices at a range of densities (0.01–0.6). Six global network measures (global efficiency [GEff], local efficiency, SW, Louvian modularity, normalized clustering coefficient, and normalized characteristic path lengths) were calculated at the selected range of densities. To avoid the bias of density selections, a range of densities (0.33–0.4) were used as the target range. The lower boundary was selected at the density when all nodes in the network had at least 1 connection based on the data from the wild-type animals in this study. The upper boundary was set to 0.4 when the averaged SW index was larger than 1.15, based on the range of the SW (eg, 1.1–1.2) used in previous publications.^50
–52 Values of the 6 global network measures were calculated for densities from 0.33 to 0.4 with the step of 0.01. The averages were calculated across the target density range and were used for estimating the effect of genotype, sex, and the genotype × sex interaction with 2-way analysis of variance (ANOVA). False discovery rate correction was performed to maintain a 5% type 1 error rate.⁵³ Corrected P values were demonstrated as Q values. One-way ANOVA was further performed in males and females on measurements that showed significant genotype × sex interactions. Post hoc analysis was performed with Tukey multiple comparison test. All of the statistical analyses in this study were performed in SPSS Statistics (Version 23.0; SPSS Inc., Chicago, Illinois).

Seed-Based Analysis

For each rsfMRI session, the regionally averaged time course of somatosensory cortex (S1 and S2 combined) was extracted and correlated with the time courses of other voxels in the whole brain to create a correlation map with somatosensory cortex as the seed ROI. For each animal, the correlation map was then transformed to a z-score connectivity map with the Fisher transformation, averaged across sessions, and subjected to a 1-way ANOVA (3dANOVA) analysis with males and females separated. 3dFWHMx determined the spatial smoothness of error variance, from which 3dClustSim estimated the required minimum cluster size, maintaining a 5% type 1 error rate (indicated as α in the Figure 3).⁵⁴ The averaged connectivity z scores (from the significant clusters in thalamus) were subjected to 2-sample t test for quantitative analysis with respect to genotype. Values are reported as mean ± standard error of the mean (SEM). Averaged z scores were used in the correlation analysis with neurotransmitter level. The connectivity profile of wild-type animals with right somatosensory cortex as the seed ROI is demonstrated in Figure 1C.

High-resolution proton Magnetic Resonance Spectroscopy Data Analysis

High-resolution proton magnetic resonance spectroscopy data were fitted using the LCModel package.⁵⁵ A simulated basis set of model metabolites appropriate for our acquisition parameters was obtained from Stephen Provencher (PhD; LCMODEL Inc., Oakville, ON, Canada; personal communication, November 2012). We based the criteria for selection of the reliable metabolite concentrations on the Cramer-Rao lower bounds (CRLB) with CRLB ≤20% for the rodent brain.^56,57 All concentrations were expressed as mean ± SEM. In the current study, the CRLB values for both GABA (14.23% ± 0.56%) and glutamate (4.89% ± 0.11%) fulfilled the criteria. Figure 1D illustrates representative in vivo high-resolution ¹HMRS spectrum from the right somatosensory cortex and thalamus of 1 male wild-type mouse. We analyzed the levels of GABA, glutamate, and the ratio of glutamate to GABA (glutamate/GABA) with 2-way ANOVA to assess the effects of genotype, sex, and genotype × sex interaction. One-way ANOVA was further performed in males and females when significant genotype × sex interaction was observed. Post hoc analysis was performed with Tukey test. The significance level was set at P < .05. Since only females demonstrated significant genotype effect in the seed-based analysis, we performed the correlation between functional connectivity (averaged connectivity z scores from the significant cluster in thalamus) and neurotransmitter levels only in females with Pearson correlation test within each group.

Data Availability

The data sets generated and analyzed during the current study are available from the corresponding author upon request.

Results

Network Topology Alteration

The average of global network measures in the target density range were significantly different with respect to the genotype effect in 2-way ANOVA (Figure 2A), that is, SW (genotype: F[2, 66] = 5.046, P = .009, Q = 0.024), normalized clustering coefficient (C/C_rand; genotype: F[2, 66] = 4.745, P = .012, Q = 0.024; sex: F[1, 66] = 4.143, P = .046, Q = 0.069), and modularity (ModL; genotype: F[2, 66] = 5.004, P = .009, Q = 0.024; sex: F[1, 66] = 5.353, P = .024, Q = 0.048). Post hoc analysis of the genotype effect showed significantly reduced values in heterozygous Met–Emx1 mice when compared with wild-type animals (SW: P = .006; C/C_rand: P = .008; ModL: P = .007) but not in homozygous Met–Emx1 mice. Significant effect of genotype, sex, and genotype × sex interaction was observed in local efficiency (MLocEff; genotype: F[2, 66] = 3.828, P = .027, Q = 0.04; sex: F[1, 66] = 6.523, P = .013, Q = 0.039; genotype × sex: F[2, 66] = 7.137, P = .002, Q = 0.012). Therefore, 1-way ANOVA was performed to further investigate the effect of genotype on MLocEff in males and females separately. A main effect of genotype was observed in females with F[2, 33] = 14.678, P < .001 (Figure 2B). MLocEff was significantly increased in female homozygous Met–Emx1 mice (P = .016) and decreased in heterozygous Met–Emx1 mice (P = .047) when compared with wild-type mice. A significant difference between heterozygous and homozygous Met–Emx1 mice was also observed (P < .001) that MLocEff in homozygous Met–Emx1 mice was significantly higher than in heterozygous Met–Emx1 mice. Only a main effect of sex was observed in the area under the curve of normalized characteristic path lengths (L/L_rand; F[1, 65] = 8.276, P = .005, Q = 0.03). No difference in the effect of genotype, sex, or genotype × sex interaction was observed in the area under the curve of GEff.

Figure 2.

Loss of Met allele changed global measurements. Alterations were observed within the densities between 0.33 and 0.4. Two-way ANOVA showed significant effects of genotype or sex and a genotype × sex interaction (A). One-way ANOVA was further performed on MLocEff in both males and females. Only females showed significant genotype effect (B). Values are means ± SEM. ^# P < .05, ^## P < .01, ^### P < 0.001 main genotype effect; ^† P < 0.05, ^†† P < 0.01 main sex effect; ^$$ P < .01 genotype × sex interaction; *P < .05, **P < .01, ***P < .001 between group difference. ANOVA indicates analysis of variance; C/C_rand, normalized clustering coefficient; GEff, global efficiency; HET, heterozygous; HOM, homozygous.; L/L_rand, normalized characteristic path lengths; MLocEff, local efficiency; ModL, modularity; SEM, standard error of the mean; SW, small-worldness; WT, wild-type.

Alteration of Somatosensory Thalamocortical Functional Connectivity

The functional connectivity maps of somatosensory cortex (S1 + S2) were subjected to 1-way ANOVA with males and females analyzed separately. Clusters in thalamus with main effect of genotype were only observed in females, P < .02 (type 1 error rate, α < 0.05), as shown in Figure 3A. In addition, between-group comparison showed increased somatosensory thalamocortical connectivity in female homozygous Met–Emx1 mice when compared with wild-type female mice, P < .01 (type 1 error rate, α < 0.05) (Figure 3B). The averaged connectivity z score of clusters with a main effect of genotype significantly increased 189% in female homozygous Met–Emx1 mice (0.17 ± 0.02) when compared with wild-type females (0.06 ± 0.01), P < .001. The z scores were subsequently used for investigation on the correlation between somatosensory thalamocortical functional connectivity and neurotransmitter levels. Heterozygous mice did not differ significantly from the wild-type mice.

Figure 3.

Impaired Met signaling leads to sex-specific differences in functional connectivity. One-way ANOVA shows significant functional connectivity alterations in thalamus with somatosensory cortex (S1 + S2) as the seed ROI in females but not in males (A). Additional post hoc testing demonstrated clusters of voxels in homozygous female mice that were significantly different to control animals (B). α indicates type 1 error rate; ANOVA, analysis of variance; HET: heterozygous, HOM, homozygous; ROI, region of interest; WT, wild-type.

Excitatory/Inhibitory Alteration in Met–Emx1 mice

Our previous in vitro electrophysiology study implicated impaired Met signaling in decreased GABAergic inhibition and increased E/I ratio.¹⁶ Magnetic resonance spectroscopy tested the hypothesis that Met–Emx1 mice have an altered E/I balance, as measured by in vivo glutamate and GABA levels. In the right somatosensory cortex (S1 + S2), no significant alteration in glutamate, GABA, or glutamate/GABA ratio was detected (Figure 4A). In the right thalamus, significant genotype × sex interaction was observed in glutamate/GABA ratio (sex: F[1, 66] = 5.468, P = .022; genotype × sex: F[2, 66] = 3.990, P = .023; Figure 4B). A 1-way ANOVA was performed with males and females separately to assess the differences among genotypes. Only female mice showed a main effect of genotype, with F[2, 33] = 6.483, P = .004. Post hoc analysis revealed higher glutamate/GABA ratio in female heterozygous Met–Emx1 mice when compared with female homozygous Met–Emx1 mice (P = .003). There was a trend toward increased glutamate/GABA ratio in female heterozygous Met–Emx1 mice when compared with female wild-type mice (P = .093; Figure 4B).

Figure 4.

Comparison of glutamate and GABA levels. The concentrations of glutamate and GABA and the glutamate/GABA ratio were measured in somatosensory cortex (A) and thalamus (B). Values are means ± SEM. Two-way ANOVA showed significant effect of sex and genotype × sex interaction of Glu/GABA ratio in thalamus. One-way ANOVA was further performed on measurements with significant genotype × sex interaction in both males and females. Only females showed significant genotype effect on Glu/GABA ratio in thalamus. ^## P < .01 main genotype effect; ^† P < .05 main sex effect; ^$ P < .05 genotype × sex interaction; **P < .01 between group difference. ⁁.05 < P < .1 trend of between-group difference. ANOVA indicates analysis of variance; GABA, gamma-aminobutyric acid; Glu, glutamate; HET, heterozygous; HOM, homozygous; SEM, standard error of the mean; WT, wild-type.

Correlation Between Glutamate/GABA Ratio and Functional Connectivity

The averaged connectivity z scores from the cluster in thalamus showing significant effect of genotype were used in the correlation analysis with neurotransmitter levels.

Correlations between thalamic glutamate/GABA ratio and somatosensory thalamocortical functional connectivity were estimated in females, with the 3 genotypic groups analyzed separately (Figure 5). Only a significant negative correlation between somatosensory thalamocortical functional connectivity and thalamic glutamate/GABA ratio was observed in heterozygous Met–Emx1 mice (r = −0.664, P = .019), while this correlation was absent in wild-type and homozygous Met–Emx1 mice.

Figure 5.

Correlation between thalamic glutamate/GABA ratio and somatosensory thalamocortical functional connectivity. Pearson correlation test revealed significant negative correlation between thalamic glutamate/GABA ratio and functional connectivity between somatosensory cortex and thalamus. GABA indicates gamma-aminobutyric acid.

Discussion

Our findings reveal several features of altered neural network communication in mice with targeted Met receptor inactivation. Decreased SW, normalized clustering coefficient, and modularity in heterozygous Met–Emx1 mice were observed when compared with wild-type ones. In addition, local efficiency in female mice was significantly increased in homozygous Met–Emx1 mice and decreased in heterozygous Met–Emx1 mice when compared with wild-type mice. In female Met–Emx1 mice, we found (1) a significantly increased functional connectivity between somatosensory cortex and thalamus in homozygous Met–Emx1 mice when compared with wild-type mice, (2) glutamate/GABA ratio in thalamus is negatively correlated with the somatosensory thalamocortical connectivity in heterozygous Met–Emx1 mice, and (3) a marginally increased glutamate/GABA ratio in thalamus of heterozygous Met–Emx1 mice and significantly higher glutamate/GABA ratio in female heterozygous Met–Emx1 mice when compared with female homozygous Met–Emx1 mice.

The reduced network modularity, SW, and normalized clustering coefficient in the current study agree with previous studies involving young autism patients.^58,59 Altered clustering coefficient and local efficiency indicated altered brain functional segregation.³⁵ The features of SW of brain functional networks were formed during typical brain maturational processes when functional connectivity tends to weaken locally and strengthen over longer distances.^60
–62 The present results demonstrated that cortical and hippocampal disruptions of HGF-Met signaling during the brain maturational process may impair the maturation of large-scale network community. Hyperconnectivity has been observed in thalamocortical system of individuals with ASD.^63,64 Enhanced somatosensory thalamocortical connectivity in the present study provides evidence that the disruption of HGF-Met signaling in the neocortex during development can lead to functional deficit in the somatosensory inhibitory system and contribute to the autistic phenotype.

Previous studies noted altered E/I balance in ASD, involving both glutamate and GABA (reviewed by Nelson and Valakh).¹⁷ Our previous in vitro electrophysiology study¹⁶ indicated that the E/I ratio increases due to a dramatic reduction in postsynaptic inhibition, as reflected in decreased amplitude of the spontaneous inhibitory postsynaptic currents in the thalamocortical pathway in Met–Emx1 mice.¹⁶ Our current study showed that although the absolute concentrations of glutamate and GABA were not altered in Met–Emx1 mice, the glutamate/GABA ratio showed an increase which provides in vivo evidence to support the E/I imbalance theory in kinase-inactive Met mice. Loss of Met activity, specifically in interneurons, has been reported to impair cognitive and procedural behaviors.⁸ The negative correlation between the glutamate/GABA ratio in the thalamus with the somatosensory thalamocortical connectivity in Met–Emx1, but not in wild-type mice, provides a link between the neurotransmission imbalance and thalamocortical system overconnectivity in this model. Future studies will design appropriate behavior tests to investigate whether Met–Emx1 mice show impaired thalamocortical system behavior, correlated to neurotransmission imbalance and brain functional overconnectivity.

Other novel findings of the current study are the difference between heterozygous and homozygous Met–Emx1 mice and between males and females. Sex differences, specifically a “male bias” or “female protective effect,” have been noticed in previous studies that males are affected more frequently than females.^65,66 Recently, there is an emerging awareness that the sex differences may be due to different phenotypic presentation in females and the failure of detecting autistic females by current diagnostic procedures.^65,67,68 Autism may manifest itself differently, and in some ways more subtly, in females.⁶⁹ Recent studies have identified sex-related differences in behavioral characteristics.^70,71 But research to date is yet to define or provide a systematic understanding of the female presentation.^69,72 In this study, male heterozygous Met–Emx1 mice presented more alterations than male homozygous Met–Emx1 mice when compared with wild-type mice in network topology, while both female heterozygous and homozygous Met–Emx1 mice showed network topological alterations and thalamic E/I imbalance. These findings indicate that fully inactive Met in the cerebral cortex and hippocampus does not necessarily result in more alterations than single allele inactivation of Met. The lack of differences in homozygous Met–Emx1 mice may be due to early compensation for the complete lack of Met Signaling, due to the inactivated allele. Also, the sex bias that males are more prone to autism-associated impairment may not hold true when focusing on a specific genetic basis.

Our results are consistent with some of the observations from individuals with the Met susceptibility alleles¹⁰ but not all. The genetic changes are different in each case. In the Met–Emx1 mice, the Met receptor is inactive, whereas in individuals, the rs1858830 variant is located in a noncoding region and appears to alter MET expression. Although the MET rs1858830 variant has a strong impact in both heterozygous and homozygous cases, the Met–Emx1 mice did not show a strong phenotype in the heterozygous group. A possible explanation is that in the Met–Emx1 mice, Met was altered only in the cerebral cortex and hippocampal regions, and the full phenotype may involve subcortical areas and inactivation of Met in broader brain areas may generate phenotypes more similar to the human results.

The present study should be viewed with the caveat that anesthetic isoflurane, also a vasodilator, can alter cerebrovascular activity and can have dose-dependent effects on both task-based and resting-state BOLD responses.^46,73 We used the minimum dose of isoflurane (1%) to maintain the immobility of the animals. Imaging with this dose or higher doses has shown preserved brain network activity and network organization in rodents.^42,73,74 Therefore, the functional connectivity observed in this study should represent the intrinsic property of functional brain organization. However, the possibility exists that mice with different genotypes respond differently to isoflurane. To resolve this question, future studies testing cerebral blood flow will be necessary.

In conclusion, our study revealed altered thalamocortical functional connectivity and correlations between neural circuits and neurotransmission in mice with the inactive autism-associated Met tyrosine kinase. The present results provide valuable in vivo evidence to support our hypothesis that aberrant functioning of the somatosensory thalamocortical system is at the core of the conspicuous somatosensory behavioral phenotypes observed in Met–Emx1 mice and in individuals with ASD. The relative level of glutamate and GABA can play an important role in the functioning of the somatosensory thalamocortical system.

Footnotes

Authors’ Note

S.T. acquired, analyzed, and interpreted the data; wrote the manuscript; and assisted with final editorial changes. W.Z. assisted with MR data acquisition. F.L. conducted the animal preparations. R.S.E. and E.M.P. designed the MRI experiment and contributed to the writing of the manuscript and interpretation of the data. S.X. assisted significantly with experimental design, MR data acquisition, data analysis, data interpretation, writing of the manuscript, and assisted with final editorial changes.

Acknowledgments

The authors are grateful to Dr. Rao Gullapalli for contributing to data interpretation. The authors thank Michelle Monroe and Shuxin Zhao for genotyping and breeding the mice.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Institutes of Health/ National Institutes of Neurological Disorders and Stroke (NS092216).

References

Naldini

Weidner

Vigna

. Scatter factor and hepatocyte growth factor are indistinguishable ligands for the MET receptor. EMBO J. 1991;10(10):2867–2878.

Campbell

Sutcliffe

Ebert

. A genetic variant that disrupts MET transcription is associated with autism. Proc Natl Acad Sci U S A. 2006;103(45):16834–16839.

Sousa

Clark

Toma

. MET and autism susceptibility: family and case-control studies. Eur J Hum Genet. 2009;17(6):749–758.

Jackson

Boccuto

Skinner

. Further evidence that the rs1858830 C variant in the promoter region of the MET gene is associated with autistic disorder. Autism Res. 2009;2(4):232–236.

Thanseem

Nakamura

Miyachi

. Further evidence for the role of MET in autism susceptibility. Neurosci Res. 2010;68(2):137–141.

Judson

Eagleson

Wang

Levitt

. Evidence of cell-nonautonomous changes in dendrite and dendritic spine morphology in the met-signaling-deficient mouse forebrain. J Comp Neurol. 2010;518(21):4463–4478.

Martins

Plachez

Powell

. Loss of embryonic MET signaling alters profiles of hippocampal interneurons. Dev Neurosci. 2007;29(1-2):143–158.

Martins

Shahrokh

Powell

. Genetic disruption of Met signaling impairs GABAergic striatal development and cognition. Neuroscience. 2011;176:199–209.

Powell

Campbell

Stanwood

Davis

Noebels

Levitt

. Genetic disruption of cortical interneuron development causes region- and GABA cell type-specific deficits, epilepsy, and behavioral dysfunction. J Neurosci. 2003;23(2):622–631.

10.

Rudie

Hernandez

Brown

. Autism-associated promoter variant in MET impacts functional and structural brain networks. Neuron. 2012;75(5):904–915.

11.

Cascio

. Somatosensory processing in neurodevelopmental disorders. J Neurodev Disord. 2010;2(2):62–69.

12.

Erzurumlu

Gaspar

. Development and critical period plasticity of the barrel cortex. Eur J Neurosci. 2012;35(10):1540–1553.

13.

Bladt

Riethmacher

Isenmann

Aguzzi

Birchmeier

. Essential role for the c-met receptor in the migration of myogenic precursor cells into the limb bud. Nature. 1995;376(6543):768–771.

14.

Judson

Bergman

Campbell

Eagleson

Levitt

. Dynamic gene and protein expression patterns of the autism-associated met receptor tyrosine kinase in the developing mouse forebrain. J Comp Neurol. 2009;513(5):511–531.

15.

Smith

Powell

. Age dependent forebrain structural changes in mice deficient in the autism associated gene Met tyrosine kinase. Neuroimage Clin. 2012;1(1):66–74.

16.

Erzurumlu

Powell

. Insulin-Independent GABAA receptor-mediated response in the barrel cortex of mice with impaired met activity. J Neurosci. 2016;36(13):3691–3697.

17.

Nelson

Valakh

. Excitatory/inhibitory balance and circuit homeostasis in autism spectrum disorders. Neuron. 2015;87(4):684–698.

18.

Thompson

Levitt

. Complete or partial reduction of the Met receptor tyrosine kinase in distinct circuits differentially impacts mouse behavior. J Neurodev Disord. 2015;7:35.

19.

Smith

Powell

. Mice with impaired met tyrosine kinase signaling demonstrate characteristics relevant to autism. Autism. 2012;S1:002.

20.

Biswal

Van Kylen

Hyde

. Simultaneous assessment of flow and BOLD signals in resting-state functional connectivity maps. NMR Biomed. 1997;10(4-5):165–170.

21.

Cordes

Haughton

Arfanakis

. Mapping functionally related regions of brain with functional connectivity MR imaging. AJNR Am J Neuroradiol. 2000;21(9):1636–1644.

22.

Jiang

Zang

Weng

. Modulation of functional connectivity during the resting state and the motor task. Hum Brain Mapp. 2004;22(1):63–71.

23.

Deco

Ponce-Alvarez

Mantini

Romani

Hagmann

Corbetta

. Resting-state functional connectivity emerges from structurally and dynamically shaped slow linear fluctuations. J Neurosci. 2013;33(27):11239–11252.

24.

Damoiseaux

Rombouts

Barkhof

. Consistent resting-state networks across healthy subjects. Proc Natl Acad Sci U S A. 2006;103(37):13848–13853.

25.

Fox

Raichle

. Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nat Rev Neurosci. 2007;8(9):700–711.

26.

Raichle

. Two views of brain function. Trends Cogn Sci. 2010;14(4):180–190.

27.

Guerra-Carrillo

Mackey

Bunge

. Resting-state fMRI: a window into human brain plasticity. Neuroscientist. 2014;20(5):522–533.

28.

Zhuo

Racz

. Early microstructural and metabolic changes following controlled cortical impact injury in rat: a magnetic resonance imaging and spectroscopy study. J Neurotrauma. 2011;28(10):2091–2102.

29.

Gullapalli

Frost

. Olanzapine antipsychotic treatment of adolescent rats causes long term changes in glutamate and GABA levels in the nucleus accumbens. Schizophr Res. 2015;161(2-3):452–457.

30.

Waddell

Zhu

. In vivo longitudinal proton magnetic resonance spectroscopy on neonatal hypoxic-ischemic rat brain injury: neuroprotective effects of acetyl-L-carnitine. Magn Reson Med. 2015;74(6):1530–1542.

31.

Horn

Steiner

. Glutamatergic and resting-state functional connectivity correlates of severity in major depression – the role of pregenual anterior cingulate cortex and anterior insula. Front Syst Neurosci. 2010;4:pii: 33.

32.

Chen

Yang

. Resting-state glutamate and GABA concentrations predict task-induced deactivation in the default mode network. J Neurosci. 2013;33(47):18566–18573.

33.

Kwon

Scheinost

Lacadie

. GABA, resting-state connectivity and the developing brain. Neonatology. 2014;106(2):149–155.

34.

Sampaio-Baptista

Filippini

Stagg

Near

Scholz

Johansen-Berg

. Changes in functional connectivity and GABA levels with long-term motor learning. Neuroimage. 2015;106:15–20.

35.

Rubinov

Sporns

. Complex network measures of brain connectivity: uses and interpretations. Neuroimage. 2010;52(3):1059–1069.

36.

Bressler

Menon

. Large-scale brain networks in cognition: emerging methods and principles. Trends Cogn Sci. 2010;14(6):277–290.

37.

Sporns

. From simple graphs to the connectome: networks in neuroimaging. Neuroimage. 2012;62(2):881–886.

38.

Telesford

Laurienti

Friedman

Kraft

Daunais

. The effects of alcohol on the nonhuman primate brain: a network science approach to neuroimaging. Alcohol Clin Exp Res. 2013;37(11):1891–1900.

39.

van den Heuvel

Scholtens

de Reus

. Topological organization of connectivity strength in the rat connectome. Brain Struct Funct. 2016;221(3):1719–1736.

40.

Verhoeven

De Cock

Lagae

Sunaert

. Neuroimaging of autism. Neuroradiology. 2010;52(1):3–14.

41.

Kennedy

Adolphs

. The social brain in psychiatric and neurological disorders. Trends Cogn Sci. 2012;16(11):559–572.

42.

Bardella

Bifone

Gabrielli

Gozzi

Squartini

. Hierarchical organization of functional connectivity in the mouse brain: a complex network approach. Sci Rep. 2016;6:32060.

43.

Liska

Galbusera

Schwarz

Gozzi

. Functional connectivity hubs of the mouse brain. Neuroimage. 2015;115:281–291.

44.

Huh

Factor

Sanchez

Uchida

Conner

Thorgeirsson

. Hepatocyte growth factor/c-met signaling pathway is required for efficient liver regeneration and repair. Proc Natl Acad Sci U S A. 2004;101(13):4477–4482.

45.

Ferron

Kroeger

Chever

Amzica

. Cortical inhibition during burst suppression induced with isoflurane anesthesia. J Neurosci. 2009;29(31):9850–9860.

46.

Liu

Zhu

Zhang

Chen

. The change of functional connectivity specificity in rats under various anesthesia levels and its neural origin. Brain Topogr. 2013;26(3):363–377.

47.

DeSimone

Pappas

Febo

. Forebrain knock-out of torsinA reduces striatal free-water and impairs whole-brain functional connectivity in a symptomatic mouse model of DYT1 dystonia. Neurobiol Dis. 2017;106:124–132.

48.

Chen

Yang

Gullapalli

Masri

. In vivo high-resolution localized (1) H MR spectroscopy in the awake rat brain at 7 T. Magn Reson Med. 2013;69(4):937–943.

49.

Paxinos

Franklin

. The Mouse Brain in Stereotaxic Coordinates. 4th ed. Cambridge, MA: Academic Press; 2012.

50.

Bernhardt

Chen

Evans

Bernasconi

. Graph-theoretical analysis reveals disrupted small-world organization of cortical thickness correlation networks in temporal lobe epilepsy. Cereb Cortex. 2011;21(9):2147–2157.

51.

Zhou

Zhang

Wang

Yuan

. Disrupted small-world brain network topology in pure conduct disorder. Oncotarget. 2017;8(39):65506–65524.

52.

Smit

Stam

Posthuma

Boomsma

de Geus

. Heritability of “small-world” networks in the brain: a graph theoretical analysis of resting-state EEG functional connectivity. Hum Brain Mapp. 2008;29(12):1368–1378.

53.

Benjamini

Hochberg

. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc. 1995;57(1):289–300.

54.

Bennett

Wolford

Miller

. The principled control of false positives in neuroimaging. Soc Cogn Affect Neurosci. 2009;4(4):4174–4122.

55.

Provencher

. Automatic quantitation of localized in vivo 1 H spectra with LCModel. NMR Biomed. 2001;14(4):260–264.

56.

Morgan

Kleven

Tulbert

Olson

Horita

Ronca

. Longitudinal 1 H MRS of rat forebrain from infancy to adulthood reveals adolescence as a distinctive phase of neurometabolite development. NMR Biomed. 2013;26(6):683–691.

57.

Harris

Yeh

Choi

. Altered neurochemical profile after traumatic brain injury: (1)H-MRS biomarkers of pathological mechanisms. J Cereb Blood Flow Metab. 2012;32(12):2122–2134.

58.

Shi

Wang

Peng

Wee

Shen

. Altered modular organization of structural cortical networks in children with autism. PLoS One. 2013;8(5):e63131.

59.

Barttfeld

Amoruso

Ais

. Organization of brain networks governed by long-range connections index autistic traits in the general population. J Neurodev Disord. 2013;5(1):16.

60.

Fair

Cohen

Power

. Functional brain networks develop from a “local to distributed” organization. PLoS Comput Biol. 2009;5(5):e1000381.

61.

Yap

Fan

Chen

Gilmore

Lin

Shen

. Development trends of white matter connectivity in the first years of life. PLoS One. 2011;6(9):e24678.

62.

Hagmann

Sporns

Madan

. White matter maturation reshapes structural connectivity in the late developing human brain. Proc Natl Acad Sci U S A. 2010;107(44):19067–19072.

63.

Nair

Carper

Abbott

. Regional specificity of aberrant thalamocortical connectivity in autism. Hum Brain Mapp. 2015;36(11):4497–4511.

64.

Woodward

Giraldo-Chica

Rogers

Cascio

. Thalamocortical dysconnectivity in autism spectrum disorder: an analysis of the Autism brain imaging data exchange. Biol Psychiatry Cogn Neurosci Neuroimaging. 2017;2(1):76–84.

65.

Werling

Geschwind

. Sex differences in autism spectrum disorders. Curr Opin Neurol. 2013;26(2):146–153.

66.

Werling

Geschwind

. Understanding sex bias in autism spectrum disorder. Proc Natl Acad Sci U S A. 2013;110(13):4868–4869.

67.

Ratto

Kenworthy

Yerys

. What about the girls? Sex-based differences in Autistic traits and adaptive skills. J Autism Dev Disord. 2018;48(5): 1698–1711.

68.

Ormond

Brownlow

Garnett

Rynkiewicz

Attwood

. Profiling Autism symptomatology: an exploration of the Q-ASC parental report scale in capturing sex differences in Autism. J Autism Dev Disord. 2018;48(2): 389–403.

69.

Dworzynski

Ronald

Bolton

Happe

. How different are girls and boys above and below the diagnostic threshold for autism spectrum disorders? J Am Acad Child Adolesc Psychiatry. 2012;51(8):788–797.

70.

Kopp

Gillberg

. The Autism Spectrum Screening Questionnaire (ASSQ)-Revised Extended Version (ASSQ-REV): an instrument for better capturing the autism phenotype in girls? A preliminary study involving 191 clinical cases and community controls. Res Dev Disabil. 2011;32(6):2875–2888.

71.

White

Wallace

Bascom

. Sex differences in parent-reported executive functioning and adaptive behavior in children and young adults with autism spectrum disorder. Autism Res. 2017;10(10):1653–1662.

72.

Lai

Lombardo

Auyeung

Chakrabarti

Baron-Cohen

. Sex/gender differences and autism: setting the scene for future research. J Am Acad Child Adolesc Psychiatry. 2015;54(1):11–24.

73.

Masamoto

Fukuda

Vazquez

Kim

. Dose-dependent effect of isoflurane on neurovascular coupling in rat cerebral cortex. Eur J Neurosci. 2009;30(2):242–250.

74.

Hutchison

Mirsattari

Jones

Gati

Leung

. Functional networks in the anesthetized rat brain revealed by independent component analysis of resting-state FMRI. J Neurophysiol. 2010;103(6):3398–3406.