Integrated Gene and Isoform-Level Transcriptomic Analysis of Adverse Childhood Experiences in the Human Prefrontal Cortex

Abstract

Adverse childhood experiences (ACE) can lead to diverse outcomes, ranging from resilience to an increased risk of psychiatric disorders such as anxiety, depression, and posttraumatic stress disorder (PTSD). In mammals, most multiexon genes encode an average of 3.9 protein-coding isoforms, which amplify transcriptomic diversity and potentially exhibit distinct functional characteristics. Recent research has shown long-lasting transcriptomic changes associated with ACE, particularly in immune-related genes. However, differential isoform usage may not be captured when analyses are confined to gene-level expression. To date, no studies have explored isoform-level dysregulation in postmortem brains of individuals exposed to ACEs. Our study investigated transcriptomic dynamics across four prefrontal regions—the dorsolateral (dlPFC), dorsal Anterior Cingulate (dACC), orbitofrontal (OFC), and subgenual prefrontal (sgPFC) cortices—in a cohort of 22 donors with PTSD, comprising 11 with and 11 without ACE history. The OFC exhibited the highest number of differentially expressed genes (DEGs), followed by the sgPFC. Correspondingly, these regions also showed the most pronounced differential isoform usage, or “isoform switching”. Notably, our integrated transcriptomic analysis revealed that while PAQR6 was downregulated in the sgPFC among ACE-exposed individuals, its principal isoform (PAQR6-201) showed increased usage. Several genes exhibiting significant isoform switching did not display substantial differential gene expression. Functional pathway analysis revealed that genes with altered expression or isoform usage converged on neurogenesis regulation, with isoform-switching genes specifically enriched in gliogenesis. This study demonstrates that examining differential isoform usage unveils previously unrecognized genes potentially implicated in ACE. Future research should focus on characterizing the functional consequences of isoform-specific up- or downregulation to comprehensively understand transcriptomic dysregulation in complex psychiatric disorders.

Introduction

Adverse childhood experiences (ACE) represent a critical environmental factor that can profoundly induce complex molecular alterations, alter neurodevelopmental trajectories and increase susceptibility to psychiatric disorders such as posttraumatic stress disorder (PTSD), anxiety, and depression [1-4]. A recent study examining transcriptomic changes associated to childhood trauma within major depressive disorder (MDD) individuals reported MED22 as a differentially expressed gene (DEG) associated with neglect [5]. Immune dysregulation was also associated with emotional abuse during childhood via principal components of expression analysis [5]. Other transcriptomic analysis of ACE-related traits also highlights substantial immune system dysregulation [6,7]. Yet, these studies have primarily focused on gene-level changes, potentially overlooking crucial molecular nuances.

In mammals, most multiexon genes encode an average of 3.9 protein-coding isoforms [8], each potentially exhibiting unique functional characteristics [9-11]. Isoform usage varies extensively during development, resulting in ubiquitous, tissue-specific, or cell-type specific expression patterns [11-13]. Recent studies in complex neuropsychiatric disorders have demonstrated that isoform-specific changes in human postmortem brain can diverge significantly from gene-level expression patterns. For instance, in schizophrenia and autism spectrum disorders (ASD), isoform usage has been shown to be more disorder-specific than overall gene expression [14,15]. Although these findings underscore the importance of comprehensive, isoform-resolved transcriptomic profiling in neuropsychiatric conditions, the specific role of isoform-level dysregulation in ACE remains uncharacterized.

Another critical yet underexplored dimension of transcriptomic regulation corresponds to alternative splicing (AS), a mechanism that generates multiple protein isoforms from a single gene [16]. In AS, the splicing machinery binds to precursor mRNA, removing non-coding regions and splicing together coding exons through five well-characterized processes: skipping exon (SE), intron retention (IR), mutually exclusive exons (MXEs), alternative acceptor/donor splice sites (A3SS and A5SS). These mechanisms significantly amplify isoform diversity and can be linked to isoform-level dysregulation.

In this study, with the aim of identifying transcriptomic changes associated with ACE rather than to PTSD, we conducted a detailed transcriptomic analysis of postmortem prefrontal cortex (PFC) samples from donors with PTSD, comparing those with and without ACE. By integrating genome-wide differential expression, isoform usage, and alternative splicing analyses, we unravel genes and molecular mechanisms underlying ACE in the human PFC (Figure 1).

Figure 1.

Study design. We examined four different transcriptomic approaches in four prefrontal cortex subregions of PTSD donors, comparing those with and without exposure to ACE. dlPFC: dorsolateral prefrontal cortex, OFC: orbitofrontal cortex, dACC: dorsal anterior cingulate cortex, sgPFC: subgenual prefrontal cortex.

Methods

Brain Tissue Samples

Brain samples were obtained from the National Center for PTSD Brain Bank [17]. Freshly frozen samples from four distinct PFC subregions were collected: the dorsolateral PFC (dlPFC), orbitofrontal cortex (OFC), dorsal anterior cingulate cortex (dACC), and subgenual PFC (sgPFC). The study cohort comprised 22 donors diagnosed with PTSD, stratified into two groups with and without a history of ACE (n = 11/each) (Table 1).

Table 1. Demographic and Clinical Characteristics of Study Cohort.

	ACE (n=11)	Non-ACE (n=11)
Males	5	7
Females	6	4
Age	39.98 ± 12.45	38.65 ± 7.33
PMI	23.55 ± 5.23	23.00 ± 5.09
Smoking status	7	8
PTSD	11	11
MDD	10	9
European Ancestry	10	8
RIN
dACC	7.55 ± 0.65	6.76 ± 0.86
dlPFC	7.73 ± 0.66	7.40 ± 0.94
OFC	7.78 ± 0.73	8.00 ± 0.81
sgPFC	7.72 ± 0.41	7.76 ± 0.55

ACE history and psychiatric diagnoses were established through antemortem and postmortem assessment including psychological autopsy and review of medical records. In accordance with ethical guidelines, sample collection and diagnostic assessments were consented to by the participants’ next of kin and reviewed by the Institutional Review Board Committees of the Department of Veterans Affairs and Yale School of Medicine. After identifying the 11 donors with PTSD who reported exposure to ACE, we used the MatchIt R package [18] to select a comparison group of PTSD donors without ACE exposure matched by sex, age, smoking status, RNA integrity number (RIN), and psychiatric comorbidities.

RNA Extraction, Sequencing, and Data Processing

RNA was isolated from each PFC subregion using RNeasy Mini kit. Libraries were prepared using Stranded RNA-seq kit according to the manufacturer’s protocol. Sequencing was performed on an Illumina HiSeq4000 with 75-bp paired-end reads at a depth of 50 million reads [19]. FASTQ files were mapped to the human genome (ENSEMBL release 79, GRCh38) using STAR (v.2.5.3a).

Gene and Isoform Quantification

Transcript-level quantification (Figure 1) was performed using Salmon 1.10.2 [20]. BAM files were converted to FASTQ format using the bam2fq tool and analyzed against GENCODE reference transcripts (hg38, release 36) with the --validateMappings flag enabled.

Differential Expression Analysis

Differential expression at gene and transcript levels (Figure 1) were assessed using DESeq2 R package [21]. Estimated counts and abundance were imported into R using tximport [22] with GENCODE annotation files (hg38, release 36) [22] DESeqDataSetFromTximport function was used to import either gene or transcript counts into a DESeq2 matrix. Genes and transcripts with at least six counts in a minimum of eleven samples (equivalent to each experimental group size) were retained for analysis. Sex, postmortem interval (PMI), RNA integrity number (RIN), age, smoking status, and number of comorbid psychiatric diagnosis were added as covariates.

Transcript Usage Analysis

Isoform usage (Figure 1) was evaluated using IsoformSwitchAnalyzeR R package [23] via DEXSeq, which models transcript expression heterogeneity between ACE-exposed and unexposed donors. Analysis was performed with quality filters (geneExpressionCutoff = 1, isoformExpressionCutoff = 0.1 and removeSingleIsoformGenes = TRUE) and included the same covariates as the differential expression analysis. Differentially used isoforms were cross-referenced with APPRIS [24] and CanIsoNet [25] databases to identify principal isoforms.

Alternative Splicing Analysis

Alternative splicing events related to isoform switching were investigated using IsoformSwitchAnalyzeR [23]. Additionally, genome-wide quantification of five splicing mechanisms (Figure 1) (SE, RI, MXEs, A3SS, and A5SS) was performed using rMATS 4.2.1 [26]. Percent Spliced-In (PSI) values, segments of RNA (exon or splice junction) that are inserted into a final mRNA transcript after splicing, were calculated. PSI values range from 0 to 1, where a PSI of 1 represents the presence of an exon or splice junction across all transcripts, while PSI values below 1 indicate the percentage of transcripts containing the spliced junction [27], suggesting an alternative splicing event. PSI values were imported using the maser R package, filtering for an average of 10 reads per junction. Targeted analysis of PSI values was conducted to identify splicing events explaining significant changes in isoform usage, with sex, PMI, RIN, age, smoking, and comorbid psychiatric diagnoses as covariates.

Functional Enrichment Analysis

Enrichment analysis was performed using enrichr [28,29] and Metascape [30] online resources that integrate databases such as NCATS BioPlanet, Panther, Gene Ontology Consortium, and Kyoto Encyclopedia of Genes and Genomes (KEGG) [31]. Genes significantly (q value < 0.05) associated with ACE that showed differential expression or differential isoform usage were used as input for this analysis.

Results

Genome Wide Differential Gene Expression

Differential gene expression analysis identified 2, 55, 5, and 19 genes significantly (q value < 0.05) associated with ACE in the dlPFC, OFC, dACC, and sgPFC regions, respectively (Figure 2a-d and Supplementary Table 1). Notable differentially expressed genes associated with ACE included OLIG1 and S100B, markers of oligodendrocytes, cells involved in the brain immune response. Functional enrichment analysis with nominal p value < 0.05 revealed enrichment in GTP hydrolysis and joining of the 60S ribosomal subunit (R-HSA-72706), EGF EGFR signaling (WP437), and regulation of axon extension (GO:0030516). The latter pathway included genes such as L1CAM, SEMA4F, ISLR2, PTK2B, SPTB, S100B, NCDN, LLPH, OLIG1, MBD1, HEYL, and BCAR1. No pathways survived multiple test correction (Figure 3 and Supplementary Table 2).

Figure 2.

Differentially expressed genes. Volcanos plot shows differential expression genes associated with ACE across all prefrontal cortex subregions. a. dlPFC, b. OFC, c. dACC, d. sgPFC. Nominal (p value < 0.05) DEG are represented with red dots.

Figure 3.

Functional enrichment analysis. Heatmap represents convergent functional enrichment analysis of differentially expressed genes (Gene) and isoforms (IsoSwitch) significantly associated with ACE.

Genome Wide Differential Isoforms Expression

Significant differential isoform expression associated with ACE was detected only in the dACC and OFC regions. In the OFC, GMFB-201, LY6H-201, MALAT1-201, MRPL12-201, SULT1A1-201, TPR-201 were significant differential isoforms. CDH13-202, ESF1-201, GLUL-202, GSTM3-201, MT-RNR2-201, NPIPB11-201, SOCS7-201, TIAM2-201 were significant isoforms in the sgPFC (Figure 4 and Supplementary Table 3).

Figure 4.

Differentially expressed isoforms. Volcanos plot shows isoforms with significant differential usage that were associated with ACE across all prefrontal cortex subregions a. dlPFC, b. OFC, c. dACC, d. sgPFC. Nominal (p value < 0.05) isoforms exhibiting DTU are represented with red dots.

Differential Transcript Usage and Alternative Splicing

Differential transcript usage analysis identified 9, 17, 5, and 18 genes exhibiting differential isoform usage (q value < 0.05) in the dlPFC, OFC, dACC, and sgPFC regions, respectively (Supplementary Table 4). Functional enrichment analysis with nominal p value < 0.05 revealed enrichment in gliogenesis (GO:0042063), ceramide biosynthetic process (GO:0046513), positive regulation of axonogenesis (GO:0050772) and neural tube closure (GO:0001843). No pathways survived multiple test correction (Figure 3 and Supplementary Table 5).

Analysis using the APPRIS database identified 40 principal isoforms with differential usage between ACE-exposed and unexposed donors (Table 2). Alternative splicing events associated with differential transcriptional usage included SE, IR, MXE, A3SS, A5SS, alternative start points, and alternative end points (Supplementary Figure 1).

Table 2. ACE-associated Isoforms with Differential Usage.

	Decreased	Increased
Principal	dlPFC: NCOR2-202, PXK-202, YME1L1-202, ZMYND11-203,	dlPFC: BHLHB9-201, EIF2B4-201, ZMYND11-202,
	OFC: AASDH-201, CABIN1-201, DAAM2-201, GAPVD1-201, PER3-201, PTGR1-201, RNF14-202, SPON2-201, SRP54-201, TBC1D3L-202,	OFC: CECR2-201, DAAM2-202, SPON2-202, TBC1D3L-201, UBAP2L-202,
	dACC: R3HCC1L-202,	dACC: ASXL1-202, R3HCC1L-201,
	sgPFC: AASDH-201, BLOC1S2-202, CITED2-201, GBA-201, ISLR2-201, PLXNB2-201, SGMS1-202, SMAP2-202, TRO-201, UBE2E2-202	sgPFC: GBA-202, ISLR2-202, PAQR6-201, PPIP5K2-202, SNX16-201
Alternative		dlPFC: NCOR2-201,
	OFC: UBAP2L-203,	OFC: PER3-202,
		dACC: GRIA4-201, TRO-202,
	sgPFC: PPIP5K2-201, TCF3-201	sgPFC: TRO-202
Minor	dlPFC: LEPR-201, ST3GAL3-204,	dlPFC: EWSR1-202, LEPR-203, PXK-201, ST3GAL3-202,
	OFC: CECR2-202, MAP4K4-205, NTRK2-202, PFKFB3-203,	OFC: AASDH-202, CABIN1-202, RNF14-203, TRAPPC3-202,
	dACC: GRIA4-202, NCEH1-202,	dACC: NCEH1-201,
	sgPFC: MRS2-202, PAQR6-203, PUF60-203	sgPFC: AASDH-202, BLOC1S2-201, CLCC1-202, MRS2-201, PLXNB2-203, SGMS1-203, SLC35B2-204, SMAP2-201, TCF3-202, UBE2E2-201
Unknown	-	OFC: BTBD6-201, PTGR1-202, SRP54-203

Color meaning: Blue: dlPFC, Green + Bold: OFC, Purple + Italic: dACC, Orange + Underline: sgPFC

When comparing findings across examined transcriptomic approaches, only one gene (PAQR6) exhibiting differential isoform usage was found to be a differentially expressed gene in the sgPFC. Other genes such as NCOR2 did not show differential expression at gene-level but showed differential usage of its major isoform. These findings exhibit that isoform-level changes can be masked when only gene-level is examined (Figure 5).

Figure 5.

Isoform-level changes may be masked when evaluating only gene-level changes. Main changes observed in our study are summarized in this conceptual diagram. Across all four examined subregions, only one gene (PAQR6) in the sgPFC exhibited differential gene expression and differential isoform usage. Nevertheless, examining each approach separately, we observed that although PAQR6 was downregulated, individuals exposed to ACE showed higher usage of the major isoform PAQR6-201 and the gene level results were probably triggered by expression of minor isoforms. NCOR2 is another example of gene-level hiding isoform-level changes, due to a NCOR2 major isoform showing high usage in ACE-exposed individuals did not necessarily results in gene level changes. Examining alternative splicing mechanisms may also help to complementary understand isoform level changes in the brain.

PAQR6, which displayed differential isoform usage, was also identified as a differentially expressed gene, albeit with contrasting directional effects. In our genome-wide differential expression (where counts from all isoforms are integrated), PAQR6 was downregulated in ACE-exposed donors after multiple test correction. However, isoform-specific analysis revealed that PAQR6-201, the principal isoform of PAQR6, showed higher usage in ACE-exposed donors (Figure 6a-b). Further examination of PAQR6-related isoforms (PAQR6-202, PAQR6-203, PAQR6-204 and PAQR6-205) revealed higher expression of PAQR6-202, PAQR6-203, and PAQR6-205 in unexposed donors (Supplementary Figure 2b), potentially explaining the gene-level expression differences. Other genes with differential isoform usage did not show significant changes in overall gene expression.

Figure 6.

Differential isoform usage of PAQR6 and associated alternative splicing mechanisms involved. a. The PAQR6 isoforms exhibiting differential usage. b. Expression levels of PAQR6 isoforms between exposed and unexposed individuals to ACE. c. Intron retention (IR) event associated with PAQR6 isoforms. d. Donor splice sites (A5SS) event associated with PAQR6 isoforms.

When analyzing PSI values of splicing events quantified via rMATS, we encountered a limitation where PSI values reflected changes across all gene-related isoforms. This means that for genes where differential usage occurs between lower-expressed isoforms, PSI values may be predominantly influenced by isoforms with higher expression (Supplementary Table 6). For instance, splicing analysis of PAQR6 identified two events with significantly different PSI values between groups. An IR event causing retention of intron 3 in PAQR6-202, showed lower PSI values in ACE-exposed versus unexposed donors (Figure 6c), consistent with the expression pattern of this isoform (Supplementary Figure 2a). We also identified an A5SS event responsible for generating a larger exon 1 in PAQR6-202 and PAQR6-203 isoforms (Figure 6d). This event showed higher PSI values in ACE-exposed than unexposed donors, which, while inconsistent with PAQR6-202 and PAQR6-203 expression levels, reflects the higher expression of other isoforms in unexposed individuals (Supplementary Figure 2a).

NCOR2, a gene with 24 related isoforms, showed differential isoform usage between NCOR2-201 and NCOR2-202 (Figure 7a-b), despite no significant changes at the gene expression level. When examining counts across all isoforms, NCOR2-201 was the principal isoform with the highest expression, showing a 1.8-fold increase in ACE-exposed versus unexposed donors (Supplementary Figure 2b). Three splicing events that contribute to NCOR2-201 formation displayed nominal differences in PSI values between groups (p value < 0.1), consistent with the observed differential usage (Figure 7c-e). Specifically, an SE event responsible for inclusion of exon 21 in NCOR2-201 exhibited higher PSI values in ACE-exposed compared to unexposed donors (Figure 7d). Conversely, A3SS and A5SS events that generate larger exons in NCOR2-202 showed lower PSI values in ACE-exposed compared to unexposed donors (Figure 7c and Figure 7e).

Figure 7.

Differential isoform usage of NCOR2 and associated alternative splicing mechanisms. a. The NCOR2 isoforms exhibiting differential usage. b. Expression levels of NCOR2 isoforms between exposed and unexposed individuals to ACE. c. Acceptor splice sites (A3SS) event associated with NCOR2 isoforms. d. skipping exon (SE) event associated with NCOR2 isoforms. e. Donor splice sites (A5SS) event associated with NCOR2 isoforms.

Discussion

In this study, we examined the transcriptomic landscape of four prefrontal cortex subregions associated with ACE. We found that the OFC and sgPFC exhibited the largest transcriptomic variations between groups. ACE-associated differential expression was found at both gene and isoform levels, although identifying different sets of genes. The only overlapping gene was PAQR6, significantly downregulated at the gene level, yet whose principal isoform exhibited higher usage in ACE-exposed donors. The remaining isoforms exhibiting differential usage did not show changes at gene level and vice versa. Genes exhibiting significant changes at either gene or isoform level participate in critical neural developmental pathways, particularly regulation of neurogenesis and gliogenesis. Our study shed light on how ACE-associated transcriptomic alterations are frequently detectable only at the isoform level, revealing molecular signals that would remain completely masked when analysis is restricted to gene-level expression changes alone.

We found that most genes with differential isoform usage did not result in differential gene expression. By examining DTU, we report additional genes and biological pathways associated with ACE exposure, that would not be discovered if only analysis at gene level had been done. Similarly, a previous study examining differential gene expression and transcript usage in Alzheimer’s disease revealed that associated genes with differential transcriptional usage did not result in differential gene expression [32]. Furthermore, authors emphasized that examining DTU contributed to the discovery of key signals associated with Alzheimer including regulation of synapse transmission and immune response [32]. Similar discordance between genes showing differential expression or differential isoform usage was also reported in schizophrenia, ASD and bipolar disorder (BD) [15]. Although these studies showed that genes exhibiting differential gene expression and isoform usage converge in pathways such as inflammatory response, isoforms tend to be more disease-specific [15].

The advantage of examining transcriptomic changes at the isoform level was particularly evident with the PAQR6 gene. At the gene level, which includes counts across all related isoforms, PAQR6 was downregulated in the ACE-exposed OFC. However, when evaluating isoform usage, the principal isoform of this gene, PAQR6-201, showed higher usage in donors exposed to ACE. These findings suggest that despite the principal isoform of PAQR6 showing higher usage associated with ACE exposure, the higher expression of minor isoforms in unexposed donors masked the PAQR6-201 pattern. PAQR6 encodes progestin and adipoQ receptor family member 6, which exhibits higher expression in several brain regions including forebrain and amygdala compared to other PAQR family members. Activation of PAQR6 through neurosteroids is involved in inhibition of apoptosis for hippocampal neuronal cells [33]. Future analysis examining the role of PAQR6 isoforms across different brain regions will elucidate the role of PAQR6-201 in ACE exposure. Furthermore, additional work is needed to understand the functional differences when minor isoforms are overexpressed under certain conditions.

Our NCOR2 findings represent another example of the importance of examining multi-layer transcriptomic diversity. When NCOR2 gene expression was examined, no differential gene expression was found between exposed and unexposed individuals to ACE. However, when we extended our analysis to evaluate isoform usage, we found that the principal isoform NCOR2-201 was upregulated in donors exposed to ACE compared to those unexposed. NCOR2 is a member of the nuclear receptor corepressors (NCORs) that regulate gene expression by mediating the activation of histone deacetylase 3 (HDAC3). Studies conducted in animal models demonstrated that alterations of NCOR2 are linked to memory deficits and reduced expression of the GABAA receptor subunit α2 (GABRA2) in lateral hypothalamus GABAergic neurons [34]. In humans, NCOR2 dysfunction has been associated with neurocognitive disorders including ASD [34,35]. Future analysis examining how a principal isoform of NCOR2 impact gene regulation of targeted genes will shed light on the downstream effects of the transcriptomic alterations associated with ACE.

At the functional level, while no pathways survived multiple test correction, genes with differential expression and isoform usage associated with ACE showed nominal enrichment in neurogenesis and gliogenesis pathways. These biological processes promote synaptic plasticity through neuronal generation and myelinization [36], and their functional alteration has been consistently linked to stress response and psychiatric disorders [37]. Previous research has demonstrated that early life stress in rodent models result in impaired neurogenesis during adulthood [38]. In our study, several genes exhibiting transcriptomic dysregulation associated with ACE exposure, including L1CAM, MBD1, NTRK2, SEMA4F, DAAM2, PLXNB2, TIAM2, HEYL, ISLR2, GPRASP3, PTK2B, GBA1, LEPR, PFKFB3, EIF2B4, SOCS7, and OLIG1 were involved in neuro/gliogenesis processes. Although preliminary, these findings suggest promising new avenues to investigate whether gene-and particularly isoform-changes in the PFC may disrupt the neuro/gliogenesis processes in the brains of individuals exposed to ACE.

In the hippocampus, the disruption of neurogenesis processes in response to stress is mediated by immune-related factors [38]. Microglia, the immune cells in the central nervous system, play a critical role in neurogenesis and synaptic pruning during development [39,40]. Furthermore, altered microglial dynamics in the brain have been observed in rodents exposed to ACE [39,40]. Our study, examining human PFC subregions, reveals that oligodendrocyte markers such as S100B and OLIG1 were differentially expressed genes associated with ACE exposure. S100B has been implicated in neuroinflammation by mediating microglial and neuronal responses to neurotoxins, with high expression related to neuronal apoptosis [41,42]. Additionally, genes such as CITED2 and SMAP2, which correspond to microglial markers, exhibited principal isoforms with differential usage across groups. CITED2 has been postulated as a repressor of proinflammatory activation mediated by macrophages [43]. Alternative splicing, highly dynamic in immune-related processes, has been linked to maintenance of the balance between immunity and tolerance [44]. In the context of Alzheimer’s disease, microglia undergo unique transcriptional alterations with responses potentially varying in an isoform-specific manner [45,46]. Splicing isoform of genes such as GPR56 has been related to microglia‐mediated synaptic pruning [47]. Our study, reporting differential isoform usage of glial markers, aligns with previous reports linking ACE to immune system dysregulation [5,48]. Future studies are warranted to elucidate how specific isoforms mediate the interplay among brain immune response, neuro/gliogenesis, and ACE exposure.

Regarding the examined PFC subregions, most transcriptomic changes associated with ACE were found in the OFC and sgPFC. Studies indicate that medial PFC may modulate emotional response by suppressing the amygdala’s function, which is implicated in memory of emotional events [49]. Both OFC and sgPFC have extensive connections with amygdala. Neuroimaging research focusing on the OFC indicates that the volume of the OFC-amygdala circuit is affected in individuals exposed to ACE, implying emotional dysregulation and long-lasting consequences into adulthood [50]. Reduced cortical thickness in the OFC has also been reported in individuals exposed to chronic stress during infancy [51]. Further, supporting these findings, a neuroimaging study of 57 adolescents with and without ACE history demonstrated that exposed individuals exhibit significant negative functional connectivity between PFC and amygdala [52]. This aligns with extensive research in non-human primates and rodent models that has established the critical role of sgPFC (also referred to as Brodmann Area 25 or BA25) in regulating emotional processing, visceral functions, and stress responses through its direct neural projections to key limbic structures including the amygdala, hypothalamus and ventral striatum [53,54]. In humans, sgPFC alterations have been related to abnormal responses to emotional experiences [55]. Considering the strong connection of the OFC and sgPFC with the amygdala, a disruption of neuro/gliogenesis processes in these two areas may play a key role in response to ACE exposure. More work is needed to confirm the isoform-specific functions in PFC subregions and determine whether these can mediate the connection with the amygdala and the response to ACE exposure.

One of the regulatory mechanisms that contributes to isoform diversity across conditions is alternative splicing [9]. In our study, we report alternative splicing events that can trigger the increased usage of certain isoforms in the OFC of donors exposed to ACE. When PSI values were examined, we found low concordance between isoform usage and PSI values of related AS events, which is expected due to technical differences across employed approaches. While IsoformSwitchAnalyzeR reports the splicing event involved in the increased/decreased usage of two particular isoforms, rMATS quantifies PSI values considering all transcripts/isoforms that map the examined gene. That is, when comparing differential usage of two isoforms with gene PSI values, it sheds light on the relative expression of the differentially used isoforms with respect to the rest of the gene-related isoforms.

Our study highlights the critical role of isoforms-level changes in ACE, demonstrating how these molecular signals remain undetected when analysis is restricted to gene-level changes. By examining the transcriptomic landscape of complex traits across gene, isoform, and alternative splicing levels, we provide a more comprehensive framework for discovering novel molecular markers. The development of integrated bioinformatic tools analyzing these multiple transcriptomic mechanisms will significantly enhance the identification of biological markers in complex disorders, while providing insights into their potential functional implications. Furthermore, studies investigating functional differences across isoforms will clarify how differential expression of principal versus minor isoforms mediates environmental response. Future research with larger sample sizes across additional brain regions (amygdala and hippocampus) will identify novel genes and/or isoforms disrupted by ACE exposure. Additionally, extending these analyses to single-cell transcriptomics will provide finer resolution into molecular underpinnings that may be obscured in bulk-tissue evaluation. Finally, comparative studies examining ACE in resilient individuals or those with psychiatric disorders other than PTSD, are essential to fully understand the long-term consequences of ACE-related transcriptomic dysregulation and its impact on the development of psychiatric disorders.

Supplementary Material

Supplementary Figures

Supplementary Figures

Supplementary Table 1

Differentially expressed genes associated with ACE in four prefrontal cortex subregions.

Supplementary Table 2

Functional enrichment analysis of ACE-associated differentially expressed genes.

Supplementary Table 3

Differentially expressed isoforms associated with ACE in four prefrontal cortex subregions.

Supplementary Table 4

Differential isoform usage associated with ACE in four prefrontal cortex subregions.

Supplementary Table 5

Functional enrichment analysis of differential isoform usage.

Supplementary Table 6

PSI values of genes exhibiting isoform switching

Glossary

ACE

adverse childhood experiences

PTSD

posttraumatic stress disorder

PFC

prefrontal cortex

dlPFC

dorsolateral

dACC

dorsal Anterior Cingulate

OFC

orbitofrontal

sgPFC

subgenual prefrontal

DEGs

differentially expressed genes

ASD

autism spectrum disorders

AS

alternative splicing

SE

exon skipping

IR

intron retention

MXEs

mutually exclusive exons

A3SS and A5SS

alternative acceptor/donor splice sites

¹Members of the Traumatic Stress Brain Research Group include

Victor E. Alvarez, MD¹, David Benedek, MD², Alicia Che, PhD³, Dianne A. Cruz, MS⁴, David A. Davis, PhD⁵, Matthew J. Girgenti, PhD^3,6, Ellen Hoffman, MD, PhD³, Paul E. Holtzheimer, MD^6,7, Bertrand R. Huber, MD, PhD⁸, Alfred Kaye, MD, PhD³, John H. Krystal, MD^3,6, Adam T. Labadorf, PhD⁸, Terence M. Keane, PhD^6,8, Mark W. Logue, PhD^6,8, Ann McKee, MD⁸, Brian Marx, PhD^6,8, Mark W. Miller, PhD^6,8, Crystal Noller, PhD^6,7, Janitza Montalvo-Ortiz, PhD³, William K. Scott, PhD⁵, Paula Schnurr, PhD^6,7, Thor Stein, MD, PhD⁸, Robert Ursano, MD⁷, Douglas E. Williamson, PhD⁴, Erika J. Wolf, PhD^6,8, Keith A. Young, PhD⁹. Affiliations: 1. Boston University, Boston, MA. 2. Uniformed Services University of the Health Sciences, Bethesda, MD. 3. Yale University, New Haven, CT. 4. Duke University School of Medicine, Durham, NC. 5. University of Miami, Miami, FL. 6. National Center for PTSD. 7. Geisel School of Medicine at Dartmouth, Hanover, NH. 8. Boston University, Boston, MA. 9. Texas A&M University, College Station, TX.

Ethical Statement

The Institutional Review Board Committees of the Department of Veterans Affairs and Yale School of Medicine reviewed this study.

Funding Statement

This work was supported by the National Institute on Drug Abuse R21DA050160 and DP1DA058737 (JLMO); the U.S. Department of Veterans Affairs via the National Center for Posttraumatic Stress Disorder – Veterans Affairs Connecticut (JLMO), 1IK2CX002095-01A1 (JLMO), and the Kavli Institute for Neuroscience at Yale University Kavli Postdoctoral Award for Academic Diversity (JJMM). This publication was made possible in part by the Yale Center for Brain and Mind Health, which is sponsored by the Yale School of Medicine (DLNR).