Your privacy, your choice

We use essential cookies to make sure the site can function. We also use optional cookies for advertising, personalisation of content, usage analysis, and social media.

By accepting optional cookies, you consent to the processing of your personal data - including transfers to third parties. Some third parties are outside of the European Economic Area, with varying standards of data protection.

See our privacy policy for more information on the use of your personal data.

for further information and to change your choices.
Skip to main content
Download PDF

Cortical morphological changes and associated transcriptional signatures in post-traumatic stress disorder and psychological resilience

BMC Medicine volume 22, Article number: 431 (2024) Cite this article

Abstract

Background

Individuals who have experienced severe traumatic events are estimated to have a post-traumatic stress disorder (PTSD) prevalence rate ranging from 10 to 50%, while those not affected by trauma exposure are often considered to possess psychological resilience. However, the neural mechanisms underlying the development of PTSD, especially resilience after trauma, remain unclear. This study aims to investigate changes of cortical morphometric similarity network (MSN) in PTSD and trauma-exposed healthy individuals (TEHI), as well as the associated molecular alterations in gene expression, providing potential targets for the prevention and intervention of PTSD.

Methods

We recruited PTSD patients and TEHI who had experienced severe earthquakes, and healthy controls who had not experienced earthquakes. We identified alterations in the whole-brain MSN changes in PTSD and TEHI, and established associations between these changes and brain-wide gene expression patterns from the Allen Human Brain Atlas microarray dataset using partial least squares regression.

Results

At the neuroimaging level, we found not only trauma-susceptible changes in TEHI same as those in PTSD, but also unique neurobiological alterations to counteract the deleterious impact of severe trauma. We identified 1444 and 2214 genes transcriptionally related to MSN changes in PTSD and TEHI, respectively. Functional enrichment analysis of weighted gene expression for PTSD and TEHI revealed distinct enrichments in Gene Ontology biological processes and Kyoto Encyclopedia of Genes and Genomes pathways. Furthermore, gene expression profiles of astrocytes, excitatory neurons, and microglial cells are highly related to MSN abnormalities in PTSD.

Conclusions

The formation of resilience may be by an active compensatory process of the brain. The combination of macroscopic neuroimaging changes and microscopic human brain transcriptomics could offer a more direct and in-depth understanding of the pathogenesis of PTSD and psychological resilience, shedding light on new targets for the prevention and treatment of PTSD.

Peer Review reports

Background

Post-traumatic stress disorder (PTSD) is a chronic psychiatric disorder that develops after experiencing or witnessing a traumatic event [1]. The lifetime prevalence of PTSD ranges from 10 to 50% after exposure to natural disasters, accidents and injuries, assaults, etc. [2, 3]. PTSD patients are characterized by intrusive and distressing symptoms, accompanied by major depression, substance abuse, suicide, and severely impaired social functioning [4]. Individuals who were exposed to trauma but did not develop PTSD are recognized as resilient. A critical unanswered question is why some trauma-exposed individuals develop PTSD whereas others remain resilient. Knowing the neuropathophysiological alterations of both PTSD patients and trauma-exposed healthy individuals (TEHI) is beneficial, as reducing stress-related risk factors and boosting resilience both contribute to the prevention and treatment of PTSD [5].

Neuroimaging techniques have been used to investigate the brain structural changes of PTSD. The consistent findings in PTSD are increased cortical thickness in the superior frontal gyrus [6], decreased gray matter volumes of lateral orbitofrontal gyri, hippocampus, lingual and superior parietal gyri, left superior temporal gyrus, and right insular [7, 8]. These abnormalities may serve as the structural foundation of the compromised functional brain network in PTSD, such as the fear circuits and the default mode network [9, 10]. In contrast, the coverage of TEHI-related brain morphometric changes is limited, as resilient individuals generally are not captured by the healthcare system and thus remain invisible in most studies. The limited research evidence shows that relevant areas of TEHI are smaller frontal and occipital gyrus, smaller supratentorial cranial vaults, hippocampus, and larger amygdala compared with non-trauma-exposed healthy controls (HC) [11,12,13]. In addition to overlapped changes with PTSD, some of these observations seem to be unique in TEHI, indicating a need to comprehensively understand the TEHI-related brain changes and the driving alterations at the microscale architecture.

Transcriptomic analyses contribute to understanding the molecular changes associated with PTSD. One study reported lower expression of the 18-kDa translocator protein gene and microglia-associated genes in postmortem brain tissue of female PTSD relative to non-trauma-exposed HC [14]. Another study analyzed gene expression and network differences of four prefrontal cortex subregions from postmortem tissue of PTSD patients, revealing the downregulation of an extensively connected group of interneuron transcripts. Integration with PTSD genome-wide association study (GWAS) data identified the interneuron synaptic gene ELFN1 as crucial to the genetic susceptibility for PTSD [15].

In contrast, the range of molecular mechanisms that lead to resilience is far from being fully determined. Several preclinical studies build up animal models (i.e., the chronic social defeat stress model) to investigate the genes that are responsible for susceptibility and resilience to extreme stress [16, 17]. It is worth noting that the numbers of regulated genes in most brain regions of resilient mice were much larger than in susceptible mice exposed to chronic social defeat stress [17, 18], suggesting that the gene expression changes in resilient mice are not necessarily indicative of pathology or the development of susceptibility, and may be part of the brain's adaptive response to stress, or compensatory alterations that enable resilience. For human studies, one study conducted GWAS of self-reported resilience among soldiers and revealed a genome-wide significant locus on chromosome 4 on an intergenic region upstream from a gene associated with the survival and regeneration promotion of damaged neurons [19]. However, there has not been a sequencing-based study to investigate the relevant molecular adaptations directly in the brain of resilient human beings.

Morphometric similarity network (MSN) is a recently developed indicator to measure individual-level morphological networks, which has been used in the research of normal brain development, aging, and neuropsychiatric disorders [20, 21]. Methodologically, MSN is based on the extraction of multiple morphometric features and the calculation of the inter-regional similarity of these features, which generates an association matrix representing the strength of morphometric similarity (MS) between each pair of cortical areas. Regional MSN is calculated as the average MS between a given cortical region and all others, representing the nodal similarity. Biologically, the morphological similarity potentially indicates a synchronized development or coordinated maturation among different brain regions [22]. The greater inter-regional morphometric similarity was more probably to belong to the same cytoarchitectonic classification and was related to stronger gene co-expression between regions [23]. One recent study linked MSN changes in patients with major depressive disorder (MDD) to human brain gene expression data and found that MDD-specific MSN differences were associated with microglia and neuronal-specific transcriptional changes [24].

To better understand the brain structural network changes and the driving molecular mechanisms of PTSD and resilience after trauma, we recruited a group of earthquake survivors, either with (N = 63) or without PTSD (N = 66), as well as trauma-unexposed HC (N = 52). First, we describe the regional MSN changes of both PTSD and TEHI compared with HC. Second, we investigate the relationship between MSN changes and anatomically patterned gene expressions from the Allen Human Brain Atlas (AHBA) microarray dataset, to obtain PTSD- and TEHI-related genes. Third, we conduct functional enrichment analyses to infer the ontological pathways of PTSD- and TEHI-related genes and compare these biological pathways. As we advocated above, we hypothesize firstly that both PTSD and TEHI show abnormal regional MSN in areas such as the prefrontal cortex related to pathology, and that TEHI will present additional changes that are unique. Secondly, functional enrichment analyses of PTSD- and TEHI-related genes may also indicate the psychopathological pathways for PTSD susceptibility and active mechanisms of resilience after severe trauma.

Methods

Participants

We recruited participants from one of the most devastated areas affected by the 2008 Wenchuan earthquake [25]. The inclusion criteria include experience or witness of the earthquake (e.g., homes destroyed, relatives killed, and being buried), being right-handed, and being aged between 18 and 60. The exclusion criteria were described in detail in our previous study [8], which mainly include the exclusion of any brain injury or physical disability, antipsychotic medication or psychotherapy, neurological diseases, psychiatric disorders other than PTSD such as schizophrenia, bipolar disorder, or substance-related and addictive disorders (comorbid depression and anxiety disorders were not excluded as these conditions usually co-occur with PTSD after trauma).

A total of 129 earthquake survivors were included in the present study, who were assessed by the DSM-IV Structured Clinical Interview (SCID) [26], the Clinical Administered PTSD Scale (CAPS) [27], Hamilton Depression Rating Scale-24 item (HAMD-24) and Hamilton Rating Scale for Anxiety (HAMA-14). According to SCID, 63 participants were diagnosed with PTSD (24 comorbid with depressive disorders and 2 comorbid with agoraphobia), and 66 TEHI did not meet the PTSD diagnostic criteria (all without any other axis I diagnoses). In addition, 52 trauma-unexposed healthy controls were recruited from areas unaffected by the earthquake through advertisement. This study was approved by the Medical Ethics Council of West China Hospital, Sichuan University. All the participants signed a written informed consent form for this study.

Imaging acquisition and processing

All participants underwent MRI examinations on the same 3.0 Tesla (T) MRI system (Siemens Trio, Erlangen, Germany) with a 12-channel phased-array head coil. For each participant, T1-weighted images were acquired using a magnetization prepared rapid acquisition gradient echo imaging sequence (slice thickness = 1 mm, TR/TE = 1900/2.26 ms, flip angle = 9°, FOV = 240 × 240 mm, number of slices = 176, data matrix = 256 × 256).

The cortical surface reconstruction was performed with recon-all (FreeSurfer 6.0.1, http:// surfer. nmr. mgh. harvard. edu/). The outcomes of this segmentation process were visually examined, and if necessary, manually corrected by adding control points. A cortical atlas with optimal resolution enables a more precise comprehension of the functional roles of specific brain regions. Finer regional parcellation can enhance the sensitivity and specificity of statistical analyses, thereby offering a more detailed characterization of connectivity patterns between brain regions. The cerebral cortex was parcellated into 308 regions [23, 28] derived from the 68 cortical regions in the D-K atlas [29]. Using a backtracking technique, this parcellation created roughly equal sizes (~ 500 mm2) for each region, reducing the impact of the diversity in parcel sizes. To obtain an individual surface parcellation for each participant, the parcellated D-K atlas was transformed to their surface [23].

For each region, seven features from the T1-weighted images were extracted, including surface area, gray matter volume, average cortical thickness, mean curvature, Gaussian curvature, fold index, and curvature index. To reduce the dimensional differences between different features, the morphological feature vectors were z-normalized across all brain regions in each participant. Therefore, for each participant, we obtained a feature matrix of brain regions (308 × 7).

Next, MSN construction and regional MSN calculation were performed from feature matrix of brain regions. The Pearson correlation coefficients were calculated from the seven normalized features of each paired brain region, thus constructing a morphological similarity network for each participant, denoted as a matrix (308 × 308) MSN. The MSN matrix is a symmetric matrix where each value MSN (i,j) represents the Pearson correlation between brain region i and brain region j. Larger absolute values indicate greater structural similarity or dissimilarity between brain regions i and brain j, while values closer to zero indicate less structural association between the two regions. To obtain regional MSN, we summed the rows or columns of the MSN matrix, resulting in a one-dimensional vector of length 308. Each value in this vector represents the sum of the corresponding brain region's total correlation effects with all other brain regions. Biologically, regional MSN reflects the overall structural connectivity strength of specific brain regions with others. These values illuminate the structural and functional interactions across different brain regions, crucial for understanding cognitive processing, information integration, and involvement in physiological and pathological processes. Brain regions with higher regional MSN values typically exhibit greater connectivity density and centrality. The stronger their coordination with other regions, the more indispensable they become in fundamental functional networks. This suggests that these regions play crucial roles in the brain network, effectively integrating and transmitting information. Their high connectivity and centrality enable rapid response and coordination with other regions, demonstrating strong synchrony [30, 31]. Conversely, brain regions with lower regional MSN values may exhibit lower connectivity density and centrality, indicating a relatively independent role within the brain network. This independence suggests that these regions rely more on their intrinsic functional and structural properties during development, rather than on close collaboration with other regions. Thus, regions with lower MSN values may demonstrate greater autonomy during development [32, 33]. Therefore, the seven aforementioned features were replaced by a comprehensive regional MSN index for each brain region.

The group average regional MSN was calculated for the three groups respectively, and the average regional MSN of the control group was basically consistent with a previous study [24]. The nuance in the rostral lateral cingulate cortex and superior frontal cortex of the right hemisphere may be attributed to the different features we included. For example, the gray matter feature Gaussian curvature, fold index were not utilized in the study of Li et al. [24]; however, two additional white matter features, fractional anisotropy and mean diffusivity, were included.

Analysis of regional MSN changes for PTSD and TEHI

To examine the differences in brain regions between groups (i.e., PTSD vs. HC and TEHI vs. HC, respectively), general linear models were used to extract a two-sided t-value and p-value. Regional MSN values as the dependent variable, group information as an independent variable, and age, sex, and education level were added as covariates. The following model was used: region MSNi = intercept + β0 × (group) + β1 × (age) + β2 × (sex) + β3 × (education). T-value and p-value corresponding to group information of each brain region were extracted. Significance was set at p < 0.05 with FDR correction for multiple comparisons across 308 regions to control the false positives rate. The remaining brain regions were believed to have significant differences between groups. Due to the importance of total intracranial volume (TIV) in brain imaging analysis, we included TIV as a covariate in the linear regression model, and we observed minimal changes in the t-maps of PTSD vs. HC and TEHI vs. HC (see Additional file 1: Fig. S1).

Regional gene expression estimation

We used the AHBA database to construct the relationship between regional MSN changes and transcriptomes. AHBA database is a publicly available database of brain-wide gene expressions. Regional microarray expression data were obtained from 6 healthy post-mortem brains (age = 42.50 ± 13.38 years; male/female = 5/1) provided by the Allen Human Brain Atlas [34]. Data were processed using the abagen toolbox(version 0.1.3; https://github.com/rmarkello/abagen), which mainly consists of the following seven steps: (1) matching microarray probes to gene annotation [35]; (2) filtering of probes (intensity-based filtering) that do not exceed 50% background noise [36]; (3) selecting probes that show the most consistent pattern of regional expression to RNA-seq data when multiple probes index the expression of the same gene; (4) samples of gene annotation assignment to brain regions in the D-K 308 atlas within 2 mm Euclidean distance of a parcel; (5) gene expression values were normalized across all genes and all tissue samples by a robust sigmoid function [37] to control for inter-subject variation; (6) strategy of differential stability was used to filter gene set in six brains; (7) samples assigned to the same brain region were averaged separately for each donor and then across donors.

These steps yielded a regional expression matrix with 308 rows, corresponding to brain regions, and 13,562 columns, corresponding to the retained genes. Since only the left half of the brain data set is available for all 6 individuals, we only used 152 brain regions of the left half of the brain for the analysis. Thus, we get a gene expression matrix of brain regions (152 regions × 13,562 gene expression levels).

PLS regression analysis between gene expression and regional MSN changes

The relationship between gene expression and regional MSN changes was linked by partial least squares (PLS) regression, which is an excellent dimension reduction decomposition method [38]. The gene expression data of 152 left brain regions were used as an independent variable, and the t-value of the corresponding brain regions was used as the dependent variable. The aim was to identify genes that might contribute to structural abnormalities in brain regions. PLS functions as a regression model where the weight assigned to each gene in the independent variable signifies its impact on the dependent variable (t-value of regional MSN). A threshold of ± 3 was applied; genes exceeding this threshold were deemed significantly correlated with brain region differences in PTSD/TEHI, identifying them as potential key genes influencing these variations. During decomposition, PLS constructs latent variables by maximizing covariance between independent and dependent variables, ensuring each set of potential variables represents a minimal correlation across groups. The first set of extracted latent variables, denoted as PLS1, captures the highest covariance and most accurately represents the relationship between the independent and dependent variables, which is an optimal linear combination of weighted gene expression scores that had the strongest correlation with the regional MSN changes. The first principal component of PLS (PLS1) explains the maximum variance between the predictor variable and the response variable. To assess the variability of each gene in PLS1, we did bootstrapping for 1000 times. Z values were calculated by the ratio of the weight of each gene to its bootstrap standard error. FDR correction showed that all genes with pFDR < 0.005 have Z values greater than 3 (positive PLS1 +) or less than − 3 (negative PLS1 −). These genes were considered to have significant contributions to MSN changes (PLS1 gene set). The same procedure was used for the TEHI vs. HC group.

Enrichment analysis

To capture the major functions of PLS1 + and PLS1 − gene sets, they were placed into the Metascape tool (https://metascape.org/gp/index.html#/main/step1) [39] for GO enrichments of biological processes and KEGG pathways enrichment analysis. All genes in the genome have been used as the enrichment background. Terms with a p-value < 0.01, a minimum count of 3, and an enrichment factor > 1.5 are collected and grouped into clusters based on their membership similarities.

Analysis of PTSD-related genes from the national center for biotechnology information

Genes associated with PTSD were selected from all databases of the National Center for Biotechnology Information (NCBI, https://www.ncbi.nlm.nih.gov), and 62 genes that were only significantly expressed in the brain were screened. Few genes related to psychological resilience using the brain tissue for assay could be found in the databases; therefore, we did not collect genes related to resilience for further analyses. We focused on 54 genes associated with PTSD that overlapped with 13,562 background genes. To verify the reliability of significant genes screened by PLS, the proportion of PLS1 + , and PLS1 − gene sets that overlapped with PTSD-related genes was calculated. Pearson correlation coefficients between the expression of these overlapping genes and the t-map were calculated separately to explore the contribution of these genes to PLS analysis. In order to avoid random effects, 1000 times spin permutation testing was carried out, all p-values were determined based on a one-sided test, and significance was set at pspin < 0.05 with FDR correction for multiple comparisons.

Analysis of other histological measures of differential gene expression

In order to test the association between differential gene expression (DGE) from postmortem tissue and the PLS1-weighted gene expression of PTSD-related changes in MSN, we obtained the DGE from four prefrontal regions of PTSD postmortem tissue compared with matched healthy individuals in the study of Girgenti et al. [15]. This study used RNA sequencing to examine the transcriptomic organization differences of PTSD. We only focused on the four prefrontal regions on the left hemisphere as only the left hemisphere of the AHBA database was used in the previous analysis. The correspondence between the four prefrontal regions and the brain regions in the D-K 308 atlas can be found in Additional file 1: Table S1. The p-value of gene log2 (fold change, FC) in each brain region was corrected by FDR, and genes with pFDR < 0.05 were screened, which were considered to have abnormal expression in the corresponding brain region. Genes with log2 FC > 0 were considered up-regulated genes; on the contrary, they were considered to be down-regulated. The abnormally expressed genes in all brain regions were collected and pooled, and 588 genes were obtained, among which 469 genes overlapped with the 13,562 background genes. Of these, 40 genes overlapped with the PLS1 + gene set. Spearman’s correlation analysis was performed between the Z value and DGE values of these overlapped genes. In addition, DGE from historical measures of other diseases was also obtained from Gandal et al. to perform the same analyses [40].

Assigning PTSD-related genes to cell types

To investigate the specific cell types in which PTSD-associated genes are predominantly expressed, we obtained a set of susceptible genes associated with neurodevelopmental disorders organized into seven cell classes: astrocytes, endothelial cells, microglia, excitatory neurons, inhibitory neurons, oligodendrocytes, oligodendrocyte precursors [41]. These genes were intersected with the PLS + gene set. The average gene expression of each cell type of the intersection genes in each of the left 152 brain regions was calculated. Then enrichment analysis was performed to explore the enrichment pathways of each cell type gene set (p < 0.05, FDR corrected). This analysis allows us to observe the expression patterns of PTSD-related genes assigned to specific cell types in different brain regions.

Results

Study design and clinical characteristics

This study combines structural magnetic resonance brain imaging data and human brain transcriptome information to establish the relationship between MSN changes and gene expression in both PTSD and TEHI (Fig. 1). The clinical characteristics of the PTSD, TEHI, and HC groups are shown in Table 1. Both the PTSD and TEHI groups differed in education, age, HAMD-24, and HAMA score compared with the HC group. In addition, there were group differences in sex and TIV between PTSD and HC.

Fig. 1
figure 1

Study overview. Structural magnetic resonance brain imaging data was collected from three groups: PTSD, TEHI, and HC. The brain was segmented and parcellated into D-K 308 atlas. Seven features from each brain region were extracted and the MSN network for each individual was calculated. T-maps for PTSD vs. HC and TEHI vs. HC were constructed. The gene expression data from the Allen Brain Atlas was assigned to 308 regions according to the D-K308 atlas. The brain gene expression data and t-maps were combined using the PLS method to identify significant genes for PLS1. Enrichment analyses were performed for these genes

Full size image
Table 1 The clinical characteristics of the PTSD, TEHI, and HC groups
Full size table

Regional MSN changes in PTSD and TEHI

The average regional MSN of the PTSD group, HC group, and TEHI group are shown respectively in Fig. 2a. The minimum value of average regional MSN was − 0.06 in the TEHI group and − 0.05 in the other two groups. The maximum value of average regional MSN was 0.04, the same in all three groups. To find out which brain regions were significantly different between groups, we constructed PTSD vs. HC t-map of regional MSN, and TEHI vs. HC t-map of regional MSN. As shown in Fig. 2b, the t-value represents the degree of the differences in brain regions. The TEHI group showed a more significant difference compared to HC, with a larger absolute t-value, and more brain regions of difference. We also compared the regional MSN between PTSD and TEHI, and no significant differences were found.

Fig. 2
figure 2

Regional MSN changes in PTSD and TEHI. a Average regional MSN of three groups. b T-maps of PTSD vs HC and TEHI vs HC. In PTSD, 2 cortical regions showed statistically significant differences (pFDR < 0.05). In TEHI, 24 cortical regions showed statistically significant differences (pFDR < 0.05). c Scatterplots showing the relationship between the average regional MSN (x-axis) and t-map (y-axis). The upper subplot represents the t-map for PTSD vs. HC (Pearson’s r(306) = 0.503, pspin < 0.001) and the lower subplot represents the t-map for TEHI vs. HC (Pearson’s r(306) = 0.624, pspin < 0.001). P value was determined based on a one-sided test

Full size image

We found that the maximum t-value in the PTSD group was 4.5, and the minimum value was − 4.01 compared with HC. The maximum t-value in the TEHI group was 5.3, and the minimum value was − 4.2 compared with HC. Both PTSD and TEHI groups showed increased regional MSN in the precentral gyrus of the left hemisphere and decreased regional MSN in the rostral middle frontal gyrus of the right hemisphere. In addition, in the TEHI group alone, increased regional MSN was found in the inferior parietal, parahippocampal, paracentral, and superior frontal gyrus in the left hemisphere, and precentral, superior frontal, supramarginal area in the right hemisphere; decreased regional MSN was found in lateral occipital, medial orbitofrontal, rostral middle frontal gyrus in the left hemisphere, and cuneus, lateral occipital, lateral orbitofrontal, lingual, medial orbitofrontal, peri calcarine area in the right hemisphere (Additional file 1: Table S2). To understand the trends of TEHI vs. PTSD in these identified differential brain regions, we extracted regional MSN values of these brain regions and performed a pairwise two-sided t-test among the PTSD, TEHI, and HC groups. The results showed that the left medial orbitofrontal, paracentral, rostral middle frontal area, and right cuneus were different for TEHI vs. PTSD, and TEHI vs. HC comparison (Additional file 1: Fig. S2). We did partial correlation analyses between values of regional MSN changes and clinical symptoms severity, and we found correlations in the TEHI cohort but not in the PTSD cohort (Additional file 1: Fig. S3).

Each dot in Fig. 2c represents the t-value and the regional MSN value in the HC group of the same brain region. Two-sided t-test were conducted on regional MSN values of 308 brain regions. After FDR correction for multiple comparisons, brain regions with a significance level of p < 0.05 were identified as showing significant differences across 308 brain regions. These regions were represented by red or blue dots, where red dots indicated brain regions with t-values > 0, indicating significantly increased regional MSN values compared to HC, and blue dots indicated brain regions with t-values < 0, indicating significantly decreased regional MSN values compared to HC. These findings are consistent with the statistically significant regions depicted in Fig. 2a. After 1000 times spin permutation testing, Pearson correlation coefficients of the PTSD group and TEHI group were calculated to be 0.50 and 0.62, respectively, showing a significant positive linear correlation (pspin < 0.0001), which means that the larger the absolute t-value, the greater the difference in regional MSN between the PTSD/TEHI group and the HC group. This indicates that the brain region shows higher overall similarity compared to HC (when the t-value is less than 0, indicating a negative correlation, suggesting opposite changes compared to the overall brain). It also suggests stronger connectivity of this brain region with others. As the t-value approaches 0, the differences between the PTSD and HC groups decrease. Therefore, this statement suggests that higher t-values indicate stronger connectivity between brain regions or greater overall similarity of the brain region. In the lower left quadrant, the number of brain regions in the PTSD group and TEHI group was roughly the same, 30.2% and 30.5%, respectively. In the upper right quadrant, the proportions of TEHI group and PTSD groups were 42.2% and 37.3%, respectively. These findings are consistent with a previous PTSD study [42], but contrary to major depressive disorder [24] and schizophrenia [28].

Weighted gene expression associated with regional MSN changes in PTSD and TEHI

The first principal component, the PLS1 score, was a weighted linear combination of all gene expressions, which showed a significant linear correlation with t-map (Fig. 3a–b). In the PTSD vs HC group, PLS1 accounts for a 27.34% variance in regional MSN changes after 1000 times spin permutation, pspin = 0.001. In TEHI vs. HC, PLS1 explained a 32.6% variance in regional MSN changes after 1000 times spin permutation, pspin < 0.001. Pearson’s correlation values were r(150) = 0.52, pspin < 0.001, and r(150) = 0.57, pspin < 0.001 in the PTSD group and TEHI group, respectively. After multiple comparisons corrected by FDR following the 1000 times spin permutation testing, the differences were statistically significant. The distribution of PLS1 score maps reflected the regional changes of gene expression in the cortical structure of the PTSD group or TEHI group, which was consistent with the regional MSN changes in the cortical region. Specifically, genes with positive PLS1 weights are overexpressed in areas where regional MSN was higher in the PTSD or TEHI group, whereas genes with negative PLS1 weights are under-expressed in areas where regional MSN was higher in the PTSD or TEHI group.

Fig. 3
figure 3

Weighted gene expression associated with regional MSN changes in PTSD and TEHI. a The weighted gene expression maps of regional PLS1 scores and the t-maps in the left hemisphere, for PTSD and TEHHI, respectively. b Scatterplots of regional PLS1 scores and t-map of regional MSN. The left plot represents the PTSD group (Pearson’s r(150) = 0.52, pspin < 0.001), and the right plot represents the TEHI group (Pearson’s r(150) = 0.57, pspin < 0.001). P value was determined based on a one-sided test

Full size image

Enrichment analysis of genes transcriptionally related to MSN changes

According to the PLS algorithm, the number of PLS1 + gene sets in the PTSD group was 1444. In the TEHI group, the number of PLS1 + gene sets was 2214. The Jaccard similarity coefficient is 0.64, with 35.50% of genes being unique to the TEHI group and only 1.11% being unique to the PTSD group (Fig. 4a–b). The corresponding analysis of the PLS1 − gene set was provided in the Additional file 1: Fig. S4.

Fig. 4
figure 4

Enrichment analysis of genes transcriptionally related to MSN changes. a A circos plot of genes overlapped between the PTSD group and the TEHI group for the PLS1 + gene set. b The enrichment network with its nodes displayed as pie sections for the genes shared between PTSD and TEHI. Each pie sector is proportional to the number of hits originating from a gene list. c–d Ontology terms for PLS1 + genes (Z > 3, pFDR < 0.05) for TEHI and PTSD. The horizontal axis represents the rich factor of the ontology terms. The size of the circle represents the number of genes involved in a given term. The color of the circles represents the degree of significant enrichment

Full size image

We tested the significant gene ontology (GO) biological processes and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways for the PLS1 − or PLS1 + gene set (Fig. 4c–d). The PLS1 + gene set in the PTSD group was enriched in pyruvate metabolism (KEGG), cellular ketone metabolic process (GO), carbon metabolism (KEGG), phagosome (KEGG), and so on. In the TEHI group, it was enriched in the phagosome (KEGG), salmonella infection (KEGG), pathways of neurodegeneration-multiple diseases (KEGG), generation of precursor metabolites and energy (GO), and so on.

PLS1-weighted expression of previously identified PTSD genes correlated with MSN changes

In order to further verify the relationship between the previously identified PTSD-related genes and regional MSN changes, we searched for genes associated with PTSD from NCBI and identified 62 PTSD-related genes by limiting the related tissues to the brain. A total of 54 genes were screened, which overlapped with 13,562 background genes. Considering only the PTSD group, 5 genes overlapped with the PLS1 + gene set, among which 2 genes (TAC1, HSP90AA1) were significantly positively correlated with t-map (TAC1, HSP90AA1). Ten genes overlapped with PLS1 − gene set, and 6 of them (UST, POGK, CACNA1C, NR3C1, RORA, NLGN1) were negatively correlated with t-map. The gene most significantly positively correlated with the t-map is the TACI gene, while the gene most significantly negatively correlated with the t-map is the POGK gene (Fig. 5a–b).

Fig. 5
figure 5

PLS1-weighted expression of previously identified PTSD genes correlated with MSN changes. a–b The PTSD-related genes expressed in the brain were screened from the NCBI database and the overlap with the PLS1 gene set. Positively correlated genes include TAC1: Pearson’s r(150) = 0.33, adjusted pspin = 0.017; WWC1: Pearson’s r(150) = 0.27, adjusted pspin = 0.088; HSP90AA1: Pearson’s r(150) = 0.28, adjusted pspin = 0.034; CRHBP: Pearson’s r(150) = 0.17, adjusted pspin = 0.210; NPY: Pearson’s r(150) = 0.16, adjusted pspin = 0.230. Negatively correlated genes include OPRL1: Pearson’s r(150) =  − 0.26, adjusted pspin = 0.054; ATP6AP1L: Pearson’s r(150) =  − 0.19, adjusted pspin = 0.191; NLGN1: Pearson’s r(150) =  − 0.39, adjusted pspin = 0; UST: Pearson’s r(150) =  − 0.34, adjusted pspin = 0.017; NR3C1: Pearson’s r(150) =  − 0.30, adjusted pspin = 0.082; POGK: Pearson’s r(150) =  − 0.43, adjusted pspin = 0; SNRNP35: Pearson’s r(150) =  − 0.28, adjusted pspin = 0.068; NR3C2: Pearson’s r(150) =  − 0.30, adjusted pspin = 0.042; CACNA1C: Pearson’s r(150) =  − 0.37, adjusted pspin = 0.008; RORA: Pearson’s r(150) =  − 0.32, adjusted pspin = 0.032). All p-values were derived from the spin test, determined based on a one-sided test, and adjusted using the FDR method. The asterisk (*) indicates p-values that remain significant after FDR correction with p < 0.05. c Weighted gene expression of PTSD-related MSN changes associated with transcriptional changes of dorsal lateral prefrontal cortex from PTSD postmortem brain tissue

Full size image

Weighted gene expression of PTSD-related MSN changes associated with transcriptional changes from PTSD postmortem brain tissue but not with differential gene expression of other disorders

We obtained the differentially expressed genes from four prefrontal brain regions of PTSD postmortem tissue RNA sequencing. For the brain region of the dorsal lateral prefrontal cortex (dlPFC), there were 19 up-regulated genes overlapped with the PLS1 + gene set, including ANXA2, BCL2L13, BHMT2, C15orf40, CALD1, CSPG4, IGDCC4, IL4R, KCNE4, LMCD1, LPP, LRRC57, NT5DC3, PDPK1, PLCE1, RESP18, TEAD3, TK2, and VIM. The Spearman correlation coefficient between Z scores and log2 FC of these genes was − 0.43 (p = 0.036). Similarly, for the medial orbital frontal cortex (medial OFC), there are 4 up-regulated genes overlapped with the PLS1 + gene set, including genes CLU, NT5DC3, RESP18, and TNFRSF21. The Spearman correlation coefficient between their Z scores and log2FC is 0.40 (p = 0.214). When looking at the down-regulated genes, 7 genes in medial OFC and 3 genes in the dorsal anterior cingulate cortex (dACC) overlapped with the PLS1 + gene set (Fig. 5c). The overlap with the PLS1 − gene set was described in the Additional file 1. We also examined the intersection genes between the PLS1 + gene set of PTSD and differentially expressed genes of schizophrenia, inflammatory bowel disease, autism spectrum disorder, alcohol abuse disorder, major depressive disorder, and bipolar disorder using histological microarray methods [40]. We did not find any association between PLS1 + weight and DGE of these disorders (Additional file 1: Fig. S5).

Expression of genes in specific cell types of PTSD

Each cell type has a different number of genes that overlap with the PLS1 + gene set in PTSD, as shown in Fig. 6a. It can be observed that genes positively associated with MSN changes in PTSD are mainly concentrated in astrocytes and excitatory neurons. The average gene expression of different cell types in the left hemisphere is depicted in Fig. 6b. For the cell type-specific genes, enrichment analysis showed that the genes related to MSN changes in PTSD are enriched for “leukocyte activation,” “TYROBP causal network,” and “innate immune response” in microglia. Changes in MSN identified in excitatory neurons were enriched with genes primarily involved in “vesicle-mediated transport,” “neuronal system,” and “neuronal projection development” (Fig. 6c). The overlaps between cell-type-specific genes and the PLS1-gene set were presented in Additional file 1: Fig. S6.

Fig. 6
figure 6

Expression of genes in specific cell types of PTSD. a The number of overlapping genes for each cell type and the corresponding p-values from the spatial permutation testing it were determined based on a one-sided test. b Gene expression maps of each cell type from overlapping genes between the PLS1 + gene and each cell type-specific gene. c Gene ontology terms enriched for overlapping genes between PLS1 + genes and each cell type-specific genes

Full size image

Discussion

This study first identified changes in regional MSN in PTSD and TEHI cohorts compared with HC. It was found that TEHI exhibited not only common regional MSN changes in brain regions shared with PTSD but also displayed a series of additional regional MSN changes in other brain regions. These MSN changes in TEHI may suggest potential factors for the development of resilience in TEHI individuals following trauma exposures. Furthermore, this study integrated a human genome-wide expression database to delineate genes associated with regional MSN alterations across the whole brain in both PTSD and TEHI separately. The results revealed a substantial overlap in genes related to MSN changes between the two conditions, with TEHI exhibiting a greater number of unique genes compared to PTSD. Gene enrichment analysis demonstrated that genes related to MSN changes in both conditions enriched in different functions and pathways, which potentially offered insights into the mechanisms underlying the onset of PTSD and the formation of resilience after trauma. Finally, by testing the genes with the greatest PLS1 weight (PLS1 + or PLS1 − gene set) enriched for genes previously reported to be implicated in the pathogenesis of PTSD, these findings provide an integrative comprehension of PTSD.

We found increased regional MSN in the left precentral gyrus and decreased regional MSN in the right rostral middle frontal gyrus. MSN represents the consistency between multiple morphological features, such as cortical thickness, across two cortical regions. Regional MSN encapsulates the cumulative degree of morphological similarity between a given brain region and all other brain regions in the entire brain. Higher values in the MSN indicate a greater likelihood of shared cellular architectural types between regions, accompanied by an increase in axonal connections [23]. The increased regional MSN that we observe in the left precentral gyrus suggests an increased architectonic similarity between it and the rest of the cortex, and the decreased regional MSN in the right rostral middle frontal gyrus may indicate greater morphometric differentiation with other cortical areas. The structural abnormalities of precentral gyrus [43] and rostral middle frontal gyrus [44] have been implicated in the pathogenesis of PTSD. The precentral gyrus and rostral middle frontal gyrus participate in the fear-related neurocircuitry [45, 46], the dysfunction of which is thought to be a crucial pathologic mechanism in PTSD [47]. Our findings of increased MSN in the precentral gyrus and decreased MSN in the rostral middle frontal gyrus provide insights into the structural basis of the network dysfunction.

The TEHI not only exhibits regional MSN abnormalities similar to PTSD but also demonstrates a range of regional MSN alterations in various other brain regions, including the parietal lobe, superior frontal gyrus, occipital lobe, middle frontal gyrus, orbital frontal gyrus, lingual gyrus, and parahippocampal gyrus. Among these, the MSN values of the left parahippocampal gyrus in TEHI show a negative correlation with both CAPS_total scores and CAPS_avoidance scores, while the right supramarginal area MSN values also exhibit a negative correlation with CAPS_avoidance scores. We did not find any associations between MSN changes and clinical symptoms in PTSD. These associations suggest that higher MSN values in the left parahippocampal gyrus and the right supramarginal area may be mechanisms underlying the development of psychological resilience. These findings indicate that there are not only trauma-susceptible changes in TEHI same as those in PTSD, but they also present compensatory neurobiological alterations to counteract the deleterious impact of severe trauma, thus maintaining mental well-being.

The neuroanatomical phenotypic alterations associated with PTSD and TEHI may be influenced by various factors, such as genetics, epigenetic modifications, and molecular signaling pathway changes [48]. Human imaging genetics has recently become a potent method for deciphering the molecular underpinnings of brain morphometric organization [49, 50]. In our study, there is a substantial overlap among genes associated with MSN changes in PTSD and TEHI, potentially reflecting an inherent anterior-to-posterior gradient of gene expression in the human cortical brain. Notably, 35% and 1% of genes were found to be specific to the respective MSN changes in TEHI and PTSD, possibly representing key molecular players in the formation of psychological resilience and PTSD. Additionally, functional enrichment analysis of weighted gene expression for PTSD and TEHI revealed distinct enrichments in GO biological processes and KEGG pathways for both conditions.

In PTSD, the identified KEGG pathway includes pyruvate metabolism, carbon metabolism, and so on. The recurring detection of anomalies in pyruvate metabolism in PTSD [51, 52] suggests impairments in mitochondrial tricarboxylic acid cycle and energy metabolism dysfunction [53], a phenomenon increasingly recognized as a contributing factor to the etiology of various psychiatric pathologies in recent years [54]. The carbon metabolism includes key aspects such as glycolysis, citric acid cycle, fatty acid and lipid metabolism, etc., which underpins various physiological processes, energy production, and the flow of carbon through ecosystems. Dysregulation of carbon metabolism can have profound implications for the development of PTSD [55]. In TEHI, the phagosome, salmonella infection neurodegeneration pathways, etc. were discovered. One recent study reported that in rats resilient to chronic social isolation stress (CSIS), the phagosome is the most activated pathway compared to CSIS-sensitive rats by hippocampal synaptoproteomic changes [56]. Phagosomes play a key role in the immune response by eliminating pathogens and foreign materials [57]. The immune activity associated with phagosomes may be a protective factor for resilience in TEHI [58].

We searched for PTSD-related genes from previous studies and found that 8 out of 15 genes were associated with MSN changes. The correlation between TAC1 gene expression and changes in regional MSN is the strongest positive, and POGK gene expression is the strongest negative. The TAC1 gene is responsible for producing four members of the tachykinin peptide hormone family, including substance P and neurokinin A, along with their related peptides, neuropeptide K and neuropeptide gamma [59]. These hormones are believed to act as neurotransmitters, engaging with nerve receptors and influencing smooth muscle cells [60]. One previous genome-wide association study reported that the POGK gene reached nominal significance for a main effect on PTSD diagnosis [61]. Although the exact function of the protein encoded by POGK is not known, this finding provides potential clues to explore the pathogenic molecules of PTSD.

We found PLS1-weighted expression from the PLS1 + gene set was associated with DGE in postmortem brain tissue of PTSD subjects, but not with DGE in the historical measures of other disorders (i.e., schizophrenia, inflammatory bowel disease, autism disorder, alcohol disorder, major depressive disorder, bipolar disorder). This suggests the specificity of the gene ranks related to MSN changes obtained by PLS in PTSD. The study by Girgenti et al. [15] examined the DGE of four prefrontal cortex subregions from PTSD postmortem tissue, and we only found genes significantly up-regulated in dlPFC were negatively associated with the overlapped genes from the PLS1 + gene set. This indicates that genes with increased dlPFC postmortem transcription in PTSD were overexpressed in areas where regional MSN was higher in PTSD. The observed MSN alterations in dlPFC appear to exhibit the closest relationship with molecular changes at the transcriptome level, suggesting its potential significance as a focal brain region for molecular pharmacological interventions.

We assigned the genes obtained from PLS analysis into seven cell types according to the study procedure of Seidlitz et al. [41]. We observed that the expression of marker genes for astrocytes and excitatory neurons constitutes the predominant proportion. Astrocytes, in addition to providing structural support to neurons, also play pivotal roles in metabolic support, synaptic transmission modulation, and inflammatory responses [62, 63]. A recent study has revealed that in mice subjected to early-life trauma, the activation of astrocytes in the cerebral cortex is heightened. These activated astrocytes exhibit an abnormal intensification in their phagocytic activity towards excitatory neurons, ultimately leading to aberrant neuronal activity, as well as anxiety- and depression-like behaviors [64]. Functional enrichment analysis further indicated that the most robustly enriched functions and pathways were predominantly centered around microglial cells, including the TYROP causal network in microglia, leukocyte activation, innate immune response, etc. The activation of microglial cells triggers the production of proinflammatory cytokines, thereby mediating the development of mood disorders [65, 66]. Furthermore, an additional study found that in a mouse model of PTSD, the elimination or inhibition of microglial cells through genetic or pharmacological methods mitigated PTSD-like symptoms [67].

The study has several limitations. Firstly, we employed the publicly available AHBA atlas, which contains genome-wide expression data from post-mortem brain tissue of six neuro-psychiatrically healthy individuals (with an average age of 43), thus does not facilitate comparisons between inter-group transcriptome-neuroimaging association. Additionally, there is substantial inter-individual variability in the demographic characteristics of the AHBA dataset contributors, which may induce bias. The brain imaging data collected in this study were sourced from participants in China, while the transcriptomes data were derived from sources African-American, Caucasian, and Hispanic. This discrepancy in data origins was not accounted for in the analysis of brain morphological differences and gene expression, which may have introduced bias into the results. Regrettably, due to the lack of postmortem brain collections in China, we hope that future researchers will dedicate their efforts to this field, making verification and comparison with our findings. Secondly, previous research lacks a dedicated genomic or transcriptomic investigation into the molecular underpinnings of psychological resilience. Consequently, we were unable to analyze cellular characterization and other functional validations of the genes obtained through the PLS method in this study. However, as researchers increasingly focus on the molecular mechanisms of resilience, future efforts may involve cross-referencing and validating the molecular findings from various studies. Finally, the sample size in this study is relatively small. In the future, it would be worthwhile to investigate whether there are similar and specific transcriptome-neuroimaging associations of PTSD and TEHI related to other trauma types (e.g., warfare and childhood trauma).

Conclusions

This study linked alterations in MSN changes related to PTSD and TEHI with gene expression levels, revealing the macroscopic structural network foundations and micro-molecular mechanisms underlying the development of PTSD and psychological resilience following severe traumatic events. TEHI individuals exhibit distinct MSN changes and associated gene functional enrichments, suggesting the formation of resilience may be by an active compensatory process of the brain. Future studies might consider promoting these protective factors to enhance resilience and reduce the incidence of PTSD. Furthermore, gene expression profiles of astrocytes, excitatory neurons, and microglial cells are highly correlated with MSN abnormalities in PTSD, indicating how previously reported genes, along with some unreported ones, drive changes in the PTSD structural brain network and mediate the risk of PTSD.

Availability of data and materials

The data that support the findings of this study are available upon reasonable requests directed to the corresponding author.

Abbreviations

HMRRC:

Huaxi MR Research Center

PTSD:

Post-traumatic stress disorder

MSN:

Morphometric similarity network

TEHI:

Trauma-exposed healthy individuals

HC:

Healthy controls

GWAS:

Genome-wide association study

DCLK2:

Doublecortin-like kinase 2

MDD:

Depressive disorder

AHBA:

Allen Human Brain Atlas

SCID:

DSM-IV Structured Clinical Interview

CAPS:

Clinical Administered PTSD Scale

HAMD-24:

Hamilton Depression Rating Scale-24 item

HAMA-14:

Hamilton Rating Scale for Anxiety

TIV:

Total intracranial volume

PLS:

Partial least squares

PLS1:

First principal component of PLS

NCBI:

National Center for Biotechnology Information

DGE:

Differential gene expression

FC:

Fold change

GO:

Gene ontology

KEGG:

Kyoto Encyclopedia of Genes and Genomes

dlPFC:

Dorsal lateral prefrontal cortex

medial OFC:

Medial orbital frontal cortex

dACC:

Dorsal anterior cingulate cortex

References

  1. American Psychiatric Association D, Association AP. Diagnostic and statistical manual of mental disorders: DSM-5, vol. 5. Washington, DC: American Psychiatric Association; 2013.

    Book  Google Scholar 

  2. Benjet C, Bromet E, Karam EG, Kessler RC, McLaughlin KA, Ruscio AM, Shahly V, Stein DJ, Petukhova M, Hill E, et al. The epidemiology of traumatic event exposure worldwide: results from the world mental health survey consortium. Psychol Med. 2016;46(2):327–43.

    Article  CAS  PubMed  Google Scholar 

  3. Beaglehole B, Mulder RT, Frampton CM, Boden JM, Newton-Howes G, Bell CJ. Psychological distress and psychiatric disorder after natural disasters: systematic review and meta-analysis. Br J Psychiatry. 2018;213(6):716–22.

    Article  PubMed  Google Scholar 

  4. Shalev A, Liberzon I, Marmar C. Post-traumatic stress disorder. N Engl J Med. 2017;376(25):2459–69.

    Article  PubMed  Google Scholar 

  5. Nestler EJ, Russo SJ. Neurobiological basis of stress resilience. Neuron. 2024;112(12):1911–29.

  6. Li L, Zhang Y, Zhao Y, Li Z, Kemp GJ, Wu M, Gong Q. Cortical thickness abnormalities in patients with post-traumatic stress disorder: a vertex-based meta-analysis. Neurosci Biobehav Rev. 2022;134: 104519.

    Article  PubMed  Google Scholar 

  7. Wang X, Xie H, Chen T, Cotton AS, Salminen LE, Logue MW, Clarke-Rubright EK, Wall J, Dennis EL, O’Leary BM, et al. Cortical volume abnormalities in posttraumatic stress disorder: an ENIGMA-psychiatric genomics consortium PTSD workgroup mega-analysis. Mol Psychiatry. 2021;26(8):4331–43.

    Article  PubMed  Google Scholar 

  8. Yuan M, Zhu H, Li Y, Ge F, Lui S, Gong Q, Qiu C, Song H, Zhang W. The DRD2 Taq1A polymorphism moderates the effect of PTSD symptom severity on the left hippocampal CA3 volume: a pilot study. Psychopharmacology. 2022;239(11):3431–8.

    Article  CAS  PubMed  Google Scholar 

  9. Ressler KJ, Berretta S, Bolshakov VY, Rosso IM, Meloni EG, Rauch SL, Carlezon WA, Jr. Post-traumatic stress disorder: clinical and translational neuroscience from cells to circuits. Nat Rev Neurol. 2022;18(5):273–88.

  10. Reuveni I, Bonne O, Giesser R, Shragai T, Lazarovits G, Isserles M, Schreiber S, Bick AS, Levin N. Anatomical and functional connectivity in the default mode network of post-traumatic stress disorder patients after civilian and military-related trauma. Hum Brain Mapp. 2016;37(2):589–99.

    Article  PubMed  Google Scholar 

  11. Fennema-Notestine C, Stein MB, Kennedy CM, Archibald SL, Jernigan TL. Brain morphometry in female victims of intimate partner violence with and without posttraumatic stress disorder. Biol Psychiatry. 2002;52(11):1089–101.

    Article  PubMed  Google Scholar 

  12. Vythilingam M, Luckenbaugh DA, Lam T, Morgan CA 3rd, Lipschitz D, Charney DS, Bremner JD, Southwick SM. Smaller head of the hippocampus in Gulf War-related posttraumatic stress disorder. Psychiatry Res. 2005;139(2):89–99.

    Article  PubMed  Google Scholar 

  13. Morey RA, Haswell CC, Hooper SR, De Bellis MD. Amygdala, hippocampus, and ventral medial prefrontal cortex volumes differ in maltreated youth with and without chronic posttraumatic stress disorder. Neuropsychopharmacology. 2016;41(3):791–801.

    Article  PubMed  Google Scholar 

  14. Bhatt S, Hillmer AT, Girgenti MJ, Rusowicz A, Kapinos M, Nabulsi N, Huang Y, Matuskey D, Angarita GA, Esterlis I, et al. PTSD is associated with neuroimmune suppression: evidence from PET imaging and postmortem transcriptomic studies. Nat Commun. 2020;11(1):2360.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Girgenti MJ, Wang J, Ji D, Cruz DA, Stein MB, Gelernter J, Young KA, Huber BR, Williamson DE, Traumatic Stress Brain Research G, et al. Transcriptomic organization of the human brain in post-traumatic stress disorder. Nat Neurosci. 2021;24(1):24–33.

    Article  CAS  PubMed  Google Scholar 

  16. Dion-Albert L, Cadoret A, Doney E, Kaufmann FN, Dudek KA, Daigle B, Parise LF, Cathomas F, Samba N, Hudson N, et al. Vascular and blood-brain barrier-related changes underlie stress responses and resilience in female mice and depression in human tissue. Nat Commun. 2022;13(1):164.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Krishnan V, Han MH, Graham DL, Berton O, Renthal W, Russo SJ, Laplant Q, Graham A, Lutter M, Lagace DC, et al. Molecular adaptations underlying susceptibility and resistance to social defeat in brain reward regions. Cell. 2007;131(2):391–404.

    Article  CAS  PubMed  Google Scholar 

  18. Bagot RC, Cates HM, Purushothaman I, Lorsch ZS, Walker DM, Wang J, Huang X, Schlüter OM, Maze I, Peña CJ, et al. Circuit-wide transcriptional profiling reveals brain region-specific gene networks regulating depression susceptibility. Neuron. 2016;90(5):969–83.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Stein MB, Choi KW, Jain S, Campbell-Sills L, Chen CY, Gelernter J, He F, Heeringa SG, Maihofer AX, Nievergelt C, et al. Genome-wide analyses of psychological resilience in U.S. Army soldiers. Am J Med Genet B Neuropsychiatr Genet. 2019;180(5):310–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Galdi P, Blesa M, Stoye DQ, Sullivan G, Lamb GJ, Quigley AJ, Thrippleton MJ, Bastin ME, Boardman JP. Neonatal morphometric similarity mapping for predicting brain age and characterizing neuroanatomic variation associated with preterm birth. Neuroimage Clin. 2020;25: 102195.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Homan P, Argyelan M, DeRosse P, Szeszko PR, Gallego JA, Hanna L, Robinson DG, Kane JM, Lencz T, Malhotra AK. Structural similarity networks predict clinical outcome in early-phase psychosis. Neuropsychopharmacology. 2019;44(5):915–22.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Alexander-Bloch A, Raznahan A, Bullmore E, Giedd J. The convergence of maturational change and structural covariance in human cortical networks. J Neurosci. 2013;33(7):2889–99.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Seidlitz J, Vasa F, Shinn M, Romero-Garcia R, Whitaker KJ, Vertes PE, Wagstyl K, Kirkpatrick Reardon P, Clasen L, Liu S, et al. Morphometric similarity networks detect microscale cortical organization and predict inter-individual cognitive variation. Neuron. 2018;97(1):231–247 e237.

    Article  CAS  PubMed  Google Scholar 

  24. Li J, Seidlitz J, Suckling J, Fan F, Ji GJ, Meng Y, Yang S, Wang K, Qiu J, Chen H, et al. Cortical structural differences in major depressive disorder correlate with cell type-specific transcriptional signatures. Nat Commun. 2021;12(1):1647.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  25. Stone R. Wenchuan earthquake. A deeply scarred land. Science. 2009;324(5928):713–4.

    Article  CAS  PubMed  Google Scholar 

  26. First M, Spitzer RL, Gibbon M. Structured clinical interview for DSM-IV axis I disorders. Washington: American Psychiatric Press; 1997.

    Google Scholar 

  27. Blake DD, Weathers FW, Nagy LM, Kaloupek DG, Gusman FD, Charney DS, Keane TM. The development of a Clinician-Administered PTSD Scale. J Trauma Stress. 1995;8(1):75–90.

    CAS  PubMed  Google Scholar 

  28. Morgan SE, Seidlitz J, Whitaker KJ, Romero-Garcia R, Clifton NE, Scarpazza C, van Amelsvoort T, Marcelis M, van Os J, Donohoe G, et al. Cortical patterning of abnormal morphometric similarity in psychosis is associated with brain expression of schizophrenia-related genes. Proc Natl Acad Sci U S A. 2019;116(19):9604–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  29. Desikan RS, Ségonne F, Fischl B, Quinn BT, Dickerson BC, Blacker D, Buckner RL, Dale AM, Maguire RP, Hyman BT, et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage. 2006;31(3):968–80.

    Article  PubMed  Google Scholar 

  30. van den Heuvel MP, Sporns O. Rich-club organization of the human connectome. J Neurosci. 2011;31(44):15775–86.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Sarli G, De Marco M, Hallikainen M, Soininen H, Bruno G, Venneri A. Regional strength of large-scale functional brain networks is associated with regional volumes in older adults and in Alzheimer’s Disease. Brain Connect. 2021;11(3):201–12.

    Article  PubMed  Google Scholar 

  32. Langs G, Wang D, Golland P, Mueller S, Pan R, Sabuncu MR, Sun W, Li K, Liu H. Identifying shared brain networks in individuals by decoupling functional and anatomical variability. Cereb Cortex. 2016;26(10):4004–14.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Zhang L, Ni H, Yu Z, Wang J, Qin J, Hou F, Yang A. Investigation on the alteration of brain functional network and its role in the identification of mild cognitive impairment. Front Neurosci. 2020;14: 558434.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Hawrylycz MJ, Lein ES, Guillozet-Bongaarts AL, Shen EH, Ng L, Miller JA, van de Lagemaat LN, Smith KA, Ebbert A, Riley ZL, et al. An anatomically comprehensive atlas of the adult human brain transcriptome. Nature. 2012;489(7416):391–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Arnatkeviciute A, Fulcher BD, Fornito A. A practical guide to linking brain-wide gene expression and neuroimaging data. Neuroimage. 2019;189:353–67.

    Article  PubMed  Google Scholar 

  36. Quackenbush J. Microarray data normalization and transformation. Nat Genet. 2002;32(Suppl):496–501.

    Article  CAS  PubMed  Google Scholar 

  37. Fulcher BD, Little MA, Jones NS. Highly comparative time-series analysis: the empirical structure of time series and their methods. J R Soc Interface. 2013;10(83):20130048.

    Article  PubMed  PubMed Central  Google Scholar 

  38. De Jong S. SIMPLS: an alternative approach to partial least squares regression. Chemom Intell Lab Syst. 1993;18(3):251–63.

    Article  Google Scholar 

  39. Zhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, Benner C, Chanda SK. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10(1):1523.

    Article  PubMed  PubMed Central  Google Scholar 

  40. Gandal MJ, Haney JR, Parikshak NN, Leppa V, Ramaswami G, Hartl C, Schork AJ, Appadurai V, Buil A, Werge TM, et al. Shared molecular neuropathology across major psychiatric disorders parallels polygenic overlap. Science. 2018;359(6376):693–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Seidlitz J, Nadig A, Liu S, Bethlehem RAI, Vértes PE, Morgan SE, Váša F, Romero-Garcia R, Lalonde FM, Clasen LS, et al. Transcriptomic and cellular decoding of regional brain vulnerability to neurogenetic disorders. Nat Commun. 2020;11(1):3358.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Xiao Y, Chen F, Lei W, Ke J, Dai Y, Qi R, Lu G, Zhong Y. Transcriptional signal and cell specificity of genes related to cortical structural differences of post-traumatic stress disorder. J Psychiatr Res. 2023;160:28–37.

    Article  CAS  PubMed  Google Scholar 

  43. Daniels JK, Frewen P, Theberge J, Lanius RA. Structural brain aberrations associated with the dissociative subtype of post-traumatic stress disorder. Acta Psychiatr Scand. 2016;133(3):232–40.

    Article  CAS  PubMed  Google Scholar 

  44. Qi S, Mu YF, Cui LB, Zhang J, Guo F, Tan QR, Shi M, Liu K, Xi YB, Zhang NY, et al. Anomalous gray matter structural networks in recent onset post-traumatic stress disorder. Brain Imaging Behav. 2018;12(2):390–401.

    Article  PubMed  Google Scholar 

  45. Lane RD, Reiman EM, Bradley MM, Lang PJ, Ahern GL, Davidson RJ, Schwartz GE. Neuroanatomical correlates of pleasant and unpleasant emotion. Neuropsychologia. 1997;35(11):1437–44.

    Article  CAS  PubMed  Google Scholar 

  46. Liu T, Ke J, Qi R, Zhang L, Zhang Z, Xu Q, Zhong Y, Lu G, Chen F. Altered functional connectivity of the amygdala and its subregions in typhoon-related post-traumatic stress disorder. Brain Behav. 2021;11(1): e01952.

    Article  PubMed  Google Scholar 

  47. Parsons RG, Ressler KJ. Implications of memory modulation for post-traumatic stress and fear disorders. Nat Neurosci. 2013;16(2):146–53.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  48. Yuan M, Yang B, Rothschild G, Mann JJ, Sanford LD, Tang X, Huang C, Wang C, Zhang W. Epigenetic regulation in major depression and other stress-related disorders: molecular mechanisms, clinical relevance and therapeutic potential. Signal Transduct Target Ther. 2023;8(1):309.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Li J, Long Z, Sheng W, Du L, Qiu J, Chen H, Liao W: Transcriptomic Similarity Informs Neuromorphic Deviations in Depression Biotypes. Biol Psychiatry 2023.

  50. Liu N, Zhang L, Tian T, Cheng J, Zhang B, Qiu S, Geng Z, Cui G, Zhang Q, Liao W et al: Cross-ancestry genome-wide association meta-analyses of hippocampal and subfield volumes. Nature genetics 2023.

  51. Muhie S, Gautam A, Yang R, Misganaw B, Daigle BJ Jr, Mellon SH, Flory JD, Abu-Amara D, Lee I, Wang K, et al. Molecular signatures of post-traumatic stress disorder in war-zone-exposed veteran and active-duty soldiers. Cell Rep Med. 2023;4(5):101045.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  52. Preston G, Emmerzaal T, Radenkovic S, Lanza IR, Oglesbee D, Morava E, Kozicz T. Cerebellar and multi-system metabolic reprogramming associated with trauma exposure and post-traumatic stress disorder (PTSD)-like behavior in mice. Neurobiol Stress. 2021;14: 100300.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  53. Somvanshi PR, Mellon SH, Flory JD, Abu-Amara D, Wolkowitz OM, Yehuda R, Jett M, Hood L, Marmar C, Doyle FJ 3rd. Mechanistic inferences on metabolic dysfunction in posttraumatic stress disorder from an integrated model and multiomic analysis: role of glucocorticoid receptor sensitivity. Am J Physiol Endocrinol Metab. 2019;317(5):E879–e898.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Iwata K. Mitochondrial involvement in mental disorders; energy metabolism, genetic, and environmental factors. Methods Mol Biol. 2019;1916:41–8.

    Article  CAS  PubMed  Google Scholar 

  55. de Vries GJ, Lok A, Mocking R, Assies J, Schene A, Olff M. Altered one-carbon metabolism in posttraumatic stress disorder. J Affect Disord. 2015;184:277–85.

    Article  PubMed  Google Scholar 

  56. Perić I, Costina V, Gass P, Findeisen P, Filipović D. Hippocampal synaptoproteomic changes of susceptibility and resilience of male rats to chronic social isolation. Brain Res Bull. 2021;166:128–41.

    Article  PubMed  Google Scholar 

  57. Flannagan RS, Jaumouillé V, Grinstein S. The cell biology of phagocytosis. Annu Rev Pathol. 2012;7:61–98.

    Article  CAS  PubMed  Google Scholar 

  58. Walker FR, Pfingst K, Carnevali L, Sgoifo A, Nalivaiko E. In the search for integrative biomarker of resilience to psychological stress. Neurosci Biobehav Rev. 2017;74(Pt B):310–20.

    Article  CAS  PubMed  Google Scholar 

  59. Chiwakata C, Brackmann B, Hunt N, Davidoff M, Schulze W, Ivell R. Tachykinin (substance-P) gene expression in Leydig cells of the human and mouse testis. Endocrinology. 1991;128(5):2441–8.

    Article  CAS  PubMed  Google Scholar 

  60. Holzer P. Local effector functions of capsaicin-sensitive sensory nerve endings: involvement of tachykinins, calcitonin gene-related peptide and other neuropeptides. Neuroscience. 1988;24(3):739–68.

    Article  CAS  PubMed  Google Scholar 

  61. Nievergelt CM, Maihofer AX, Mustapic M, Yurgil KA, Schork NJ, Miller MW, Logue MW, Geyer MA, Risbrough VB, O’Connor DT, et al. Genomic predictors of combat stress vulnerability and resilience in U.S. Marines: a genome-wide association study across multiple ancestries implicates PRTFDC1 as a potential PTSD gene. Psychoneuroendocrinology. 2015;51:459–71.

    Article  CAS  PubMed  Google Scholar 

  62. Endo F, Kasai A, Soto JS, Yu X, Qu Z, Hashimoto H, Gradinaru V, Kawaguchi R, Khakh BS. Molecular basis of astrocyte diversity and morphology across the CNS in health and disease. Science. 2022;378(6619):eadc9020.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  63. Tsai SF, Hsu PL, Chen YW, Hossain MS, Chen PC, Tzeng SF, Chen PS, Kuo YM. High-fat diet induces depression-like phenotype via astrocyte-mediated hyperactivation of ventral hippocampal glutamatergic afferents to the nucleus accumbens. Mol Psychiatry. 2022;27(11):4372–84.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  64. Byun YG, Kim NS, Kim G, Jeon YS, Choi JB, Park CW, Kim K, Jang H, Kim J, Kim E, et al. Stress induces behavioral abnormalities by increasing expression of phagocytic receptor MERTK in astrocytes to promote synapse phagocytosis. Immunity. 2023;56(9):2105–2120 e2113.

    Article  CAS  PubMed  Google Scholar 

  65. Zhang J, Rong P, Zhang L, He H, Zhou T, Fan Y, Mo L, Zhao Q, Han Y, Li S. IL4-driven microglia modulate stress resilience through BDNF-dependent neurogenesis. Science. Advances. 2021;7(12):eabb9888.

    CAS  Google Scholar 

  66. Du Preez A, Onorato D, Eiben I, Musaelyan K, Egeland M, Zunszain PA, Fernandes C, Thuret S, Pariante CM. Chronic stress followed by social isolation promotes depressive-like behaviour, alters microglial and astrocyte biology and reduces hippocampal neurogenesis in male mice. Brain Behav Immun. 2021;91:24–47.

    Article  PubMed  Google Scholar 

  67. Li S, Liao Y, Dong Y, Li X, Li J, Cheng Y, Cheng J, Yuan Z. Microglial deletion and inhibition alleviate behavior of post-traumatic stress disorder in mice. J Neuroinflammation. 2021;18(1):7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We thank all patients and volunteers for their contribution in this study.

Funding

This study was funded by 1·3·5 project for disciplines of excellence, West China Hospital, Sichuan University (Grant numbers ZYJC21004 to WZ), National Natural Science Foundation of China (Grant numbers 81871061 to WZ, 82171513 to HRZ, 82401796 to MLY), and Department of Science and Technology of Sichuan Province (Grant number 2023YFS0417 to MY).

Author information

Author notes

  1. Minlan Yuan and Lun Li contributed equally to this work.

Authors and Affiliations

  1. Mental Health Center and Psychiatric Laboratory, the State Key Laboratory of Biotherapy, West China Hospital of Sichuan University, 610041, Chengdu, People’s Republic of China

    Minlan Yuan, Lun Li, Hongru Zhu & Wei Zhang

  2. Huaxi Brain Research Center, West China Hospital of Sichuan University, 610041, Chengdu, People’s Republic of China

    Minlan Yuan, Hongru Zhu & Wei Zhang

  3. Sichuan Institute of Computer Sciences, 610041, Chengdu, People’s Republic of China

    Lun Li

  4. Med-X Center for Informatics, Sichuan University, 610041, Chengdu, People’s Republic of China

    Hongru Zhu

  5. Department of Interventional Medicine, Sichuan Science City Hospital, 621000, Mianyang, People’s Republic of China

    Bo Zheng

  6. Huaxi MR Research Center (HMRRC), Department of Radiology, West China Hospital of Sichuan University, 610041, Chengdu, People’s Republic of China

    Su Lui

  7. West China Biomedical Big Data Center, West China Hospital, Sichuan University, Chengdu, 610041, People’s Republic of China

    Wei Zhang

  8. Medical Big Data Center, Sichuan University, 610041, Chengdu, People’s Republic of China

    Wei Zhang

Authors
  1. Minlan Yuan
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Lun Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Hongru Zhu
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Bo Zheng
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Su Lui
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Wei Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

MLY and LL collected data, did the statistical analyses and wrote this paper. HRZ, BZ and SL provided expertise on protocol development and analyses. WZ was the primary designer of this study. All authors provided revision of the manuscript.

Corresponding author

Correspondence to Wei Zhang.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the ethics committee of the Medical Ethics Council of West China Hospital, Sichuan University (1054/2020). All the participants provided a written informed consent form for this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

12916_2024_3657_MOESM1_ESM.docx

Additional file 1: Figure S1. Pearson’s correlation for 308 t-statistical regional values for PTSD vs HC and TEHI vs HC differences with and without controlling for TIV. Table S1. The correspondence between the four prefrontal regions and the brain regions in the D–K 308 atlas. Table S2. significantly different brain regions in TEHI and PTSD respectively. Figure S2. Boxplot of regional MSN values between PTSD, TEHI and HC. Figure S3. Scatter plots between region MSN of significant brain regions and clinical scales in the TEHI group. Figure S4. Enrichment analysis of genes transcriptionally related to MSN changes for the PLS1- gene set. Figure S5. Associations between PLS1+ weighted gene expression and differentially gene expression. Figure S6. Expression of genes in specific cell types of PTSD for the PLS1- gene set.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Yuan, M., Li, L., Zhu, H. et al. Cortical morphological changes and associated transcriptional signatures in post-traumatic stress disorder and psychological resilience. BMC Med 22, 431 (2024). https://doi.org/10.1186/s12916-024-03657-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12916-024-03657-9

Keywords

BMC Medicine

ISSN: 1741-7015

Contact us