Whole-Genome Omics Elucidates the Role of CCM1 and Progesterone in Cerebral Cavernous Malformations within CmPn Networks

Abstract

Cerebral cavernous malformations (CCMs) are abnormal expansions of brain capillaries that increase the risk of hemorrhagic strokes, with CCM1 mutations responsible for about 50% of familial cases. The disorder can cause irreversible brain damage by compromising the blood-brain barrier (BBB), leading to fatal brain hemorrhages. Studies show that progesterone and its derivatives significantly impact BBB integrity. The three CCM proteins (CCM1, CCM2, and CCM3) form the CCM signaling complex (CSC), linking classic and non-classic progesterone signaling within the CmPn network, which is crucial for maintaining BBB integrity. This study aimed to explore the relationship between CCM1 and key pathways of the CmPn signaling network using three mouse embryonic fibroblast lines (MEFs) with distinct CCM1 expressions. Omics and systems biology analysis investigated CCM1-mediated signaling within the CmPn network. Our findings reveal that CCM1 is essential for regulating cellular processes within progesterone-mediated CmPn/CmP signaling, playing a crucial role in maintaining microvessel integrity. This regulation occurs partly through gene transcription control. The critical role of CCM1 in these processes suggests it could be a promising therapeutic target for CCMs.

Keywords: CCM signaling complex (CSC); CSC-mPRs-PRG-nPRs (CmPn) signaling network; blood–brain barrier (BBB); cerebral cavernous malformations (CCMs); non-classic membrane receptor (mPRs); nuclear progesterone receptor (nPRs); progesterone (PRG) signaling; whole-genome omics.

Figure 1

Differentially expressed protein (DEP) profiles in the MEFs with three different genotypes of CCM1, including depletion (knockout, KO), endogenously low (WT), and ectopically excessive expression (knockin, KI/96) of CCM1. The protein profiles of mouse embryonic fibroblasts (MEFs) were examined for differential expression in relation to CCM1 depletion (knockout, KO), endogenous low levels (WT), and excess expression through knockin (KI/96). These DEPs were analyzed to investigate the impact of genetic backgrounds on protein expression profiles in MEFs. (A) To compare the differentially expressed protein (DEP) profiles of three pairs of mouse embryonic fibroblasts (MEFs) with total depletion (CCM1KO), endogenous low expression levels (CCM1—WT), or excess expression through knockin (CCM1KI/96) of CCM1, a Venn diagram was utilized. This diagram illustrates the number of DEPs with significant expression changes between the paired comparisons of CCM1 genotypes, as well as the number of specifically expressed proteins between two different CCM1 genotypes. (B) To depict the transcriptome profiles of differentially expressed proteins (DEPs) in mouse embryonic fibroblasts (MEFs) across three comparative pairs of CCM1 genotypes, a bar diagram was employed. The diagram showcases the number of upregulated (represented by red bars) and downregulated (represented by blue bars) DEPs between two different CCM1 genotypes. (C) A heatmap was used to visualize the differential expression levels and profiles of DEPs in three comparative pairs of CCM1 genotypes in mouse embryonic fibroblasts (MEFs) in response to PRG actions. The heatmap shows the upregulation (red lines) and downregulation (blue lines) of DEPs for each comparative genotype pair across the three different CCM1 genotypes under PRG treatment. The DEPs heatmap was created using t-test statistical analysis and visualized with clustering software, as detailed in the Methods section. (D) The ORA results highlighting core-enriched differentially expressed proteins (DEPs) in GO pathways were presented for both progesterone (PRG) treatment and vehicle control across three CCM1 genotypes. These genotypes are total depletion of CCM1 (CCM1-KO, left), endogenous low levels of CCM1 (CCM1-WT, middle), and ectopic overexpression of CCM1 through CCM1-knockin (CCM1-KI/96, right). p-values were adjusted across all three CCM1 genotypes. (E) Similarly, in the comparative ROA plots, red circles represent DEPs with lower significance in adjusted p-value, while blue circles indicate DEPs with higher significance in adjusted p-value. The size of the circles corresponds to the GeneRatio value, which measures whether genes from predefined sets (e.g., genes in the untreated group within a specific GO term or KEGG pathway) are more present than expected (over-represented) in another subset of the data (e.g., treated).

Figure 1

Differentially expressed protein (DEP) profiles in the MEFs with three different genotypes of CCM1, including depletion (knockout, KO), endogenously low (WT), and ectopically excessive expression (knockin, KI/96) of CCM1. The protein profiles of mouse embryonic fibroblasts (MEFs) were examined for differential expression in relation to CCM1 depletion (knockout, KO), endogenous low levels (WT), and excess expression through knockin (KI/96). These DEPs were analyzed to investigate the impact of genetic backgrounds on protein expression profiles in MEFs. (A) To compare the differentially expressed protein (DEP) profiles of three pairs of mouse embryonic fibroblasts (MEFs) with total depletion (CCM1KO), endogenous low expression levels (CCM1—WT), or excess expression through knockin (CCM1KI/96) of CCM1, a Venn diagram was utilized. This diagram illustrates the number of DEPs with significant expression changes between the paired comparisons of CCM1 genotypes, as well as the number of specifically expressed proteins between two different CCM1 genotypes. (B) To depict the transcriptome profiles of differentially expressed proteins (DEPs) in mouse embryonic fibroblasts (MEFs) across three comparative pairs of CCM1 genotypes, a bar diagram was employed. The diagram showcases the number of upregulated (represented by red bars) and downregulated (represented by blue bars) DEPs between two different CCM1 genotypes. (C) A heatmap was used to visualize the differential expression levels and profiles of DEPs in three comparative pairs of CCM1 genotypes in mouse embryonic fibroblasts (MEFs) in response to PRG actions. The heatmap shows the upregulation (red lines) and downregulation (blue lines) of DEPs for each comparative genotype pair across the three different CCM1 genotypes under PRG treatment. The DEPs heatmap was created using t-test statistical analysis and visualized with clustering software, as detailed in the Methods section. (D) The ORA results highlighting core-enriched differentially expressed proteins (DEPs) in GO pathways were presented for both progesterone (PRG) treatment and vehicle control across three CCM1 genotypes. These genotypes are total depletion of CCM1 (CCM1-KO, left), endogenous low levels of CCM1 (CCM1-WT, middle), and ectopic overexpression of CCM1 through CCM1-knockin (CCM1-KI/96, right). p-values were adjusted across all three CCM1 genotypes. (E) Similarly, in the comparative ROA plots, red circles represent DEPs with lower significance in adjusted p-value, while blue circles indicate DEPs with higher significance in adjusted p-value. The size of the circles corresponds to the GeneRatio value, which measures whether genes from predefined sets (e.g., genes in the untreated group within a specific GO term or KEGG pathway) are more present than expected (over-represented) in another subset of the data (e.g., treated).

Figure 2

A multi-omics approach to analyzing proteomic and RNAseq data under various levels of CCM1 expression in response to progesterone (PRG) actions. Both protein and RNA expression profiles of MEFs across three different levels of CCM1 expression were investigated. The proteomic data included three distinct genotypes: complete depletion of CCM1 via CCM1-knockout (CCM1-KO), natural low expression levels of CCM1 (CCM1-WT), and ectopic overexpression of CCM1 via CCM1-knockin (CCM1-96). Meanwhile, the RNAseq data covered two levels of expression profiles: complete depletion (knockout) and ectopic overexpression levels of CCM1 via CCM1-knockin (CCM1-96). (A) The expression profiles of DEPs in response to PRG actions were investigated in MEFs with three different CCM1 genotypes. A Venn diagram was used to illustrate the specific DEPs affected by progesterone (PRG) treatment compared to vehicle controls across the three CCM1 genotypes in mouse embryonic fibroblasts (MEFs). (B) A bar diagram was used to depict the profiles of DEPs in response to PRG actions across three CCM1 genotypes. The diagram shows the distribution of upregulated DEPs (represented by a red bar) and downregulated DEPs (represented by a blue bar). (C) A heatmap was employed to visualize the profiles of DEPs in response to PRG actions across three CCM1 genotypes. The heatmap illustrates the distribution of upregulated DEPs (indicated by red lines) and downregulated DEPs (indicated by blue lines). Comparative analysis of DEPs was performed using pairwise t-test statistical analysis. (D) Similarly, for RNA expression profiling, Venn diagrams were used to compare expression patterns between untreated MEF KO controls and MEF-KO treated with progesterone (PRG) in the CCM1-depletion genotype (MEF KO). The analysis identified 751 unique differentially expressed genes (DEGs) upon exposure to PRG. (E) Likewise, for the ectopic overexpression of the CCM1 (MEF 96) genotype, we compared expression patterns between untreated MEF 96 controls and MEF 96 treated with progesterone (PRG). This analysis revealed 573 differentially expressed genes (DEGs) upon exposure to PRG. (F) A heatmap was generated to depict the profiles of differentially expressed genes (DEGs) in response to PRG actions across two CCM1 genotypes. The comprehensive analysis involved examining 1900 shared RNAseq profiles between the two CCM1 genotypes to investigate how exposure to PRG influences the regulation of common sequences across different levels of CCM1, both with and without PRG treatment. (G) The ORA results, highlighting core-enriched DEPs in both GO and KEGG pathways, were presented to show the differential responses to PRG actions across three CCM1 genotypes (CCM1-KO, left; CCM1-WT, middle; and CCM1-KI/96, right). The analysis plots depict data specifically filtered with CCM1-associated pathways from enriched DEPs in both GO and KEGG approaches (Figure 1D,E), with the original dataset provided in the Supplementary Materials. It is important to highlight that all samples underwent triplicate analysis, and statistical significance was evaluated using a t-test, with p-values less than 0.05 considered significant. (H) Additionally, the RNAseq data underwent ORA similar to the previous approach used for proteomic data. Importantly, we emphasized shared pathways between the RNAseq profiles and the proteomic data by highlighting them with bold borders, specifically focusing on Diabetic cardiomyopathy and Aminoacyl-tRNA biosynthesis. In the comparative ROA plots, the depiction and scale of circles remain consistent with the previous ones and uniform throughout this manuscript.

Figure 2

A multi-omics approach to analyzing proteomic and RNAseq data under various levels of CCM1 expression in response to progesterone (PRG) actions. Both protein and RNA expression profiles of MEFs across three different levels of CCM1 expression were investigated. The proteomic data included three distinct genotypes: complete depletion of CCM1 via CCM1-knockout (CCM1-KO), natural low expression levels of CCM1 (CCM1-WT), and ectopic overexpression of CCM1 via CCM1-knockin (CCM1-96). Meanwhile, the RNAseq data covered two levels of expression profiles: complete depletion (knockout) and ectopic overexpression levels of CCM1 via CCM1-knockin (CCM1-96). (A) The expression profiles of DEPs in response to PRG actions were investigated in MEFs with three different CCM1 genotypes. A Venn diagram was used to illustrate the specific DEPs affected by progesterone (PRG) treatment compared to vehicle controls across the three CCM1 genotypes in mouse embryonic fibroblasts (MEFs). (B) A bar diagram was used to depict the profiles of DEPs in response to PRG actions across three CCM1 genotypes. The diagram shows the distribution of upregulated DEPs (represented by a red bar) and downregulated DEPs (represented by a blue bar). (C) A heatmap was employed to visualize the profiles of DEPs in response to PRG actions across three CCM1 genotypes. The heatmap illustrates the distribution of upregulated DEPs (indicated by red lines) and downregulated DEPs (indicated by blue lines). Comparative analysis of DEPs was performed using pairwise t-test statistical analysis. (D) Similarly, for RNA expression profiling, Venn diagrams were used to compare expression patterns between untreated MEF KO controls and MEF-KO treated with progesterone (PRG) in the CCM1-depletion genotype (MEF KO). The analysis identified 751 unique differentially expressed genes (DEGs) upon exposure to PRG. (E) Likewise, for the ectopic overexpression of the CCM1 (MEF 96) genotype, we compared expression patterns between untreated MEF 96 controls and MEF 96 treated with progesterone (PRG). This analysis revealed 573 differentially expressed genes (DEGs) upon exposure to PRG. (F) A heatmap was generated to depict the profiles of differentially expressed genes (DEGs) in response to PRG actions across two CCM1 genotypes. The comprehensive analysis involved examining 1900 shared RNAseq profiles between the two CCM1 genotypes to investigate how exposure to PRG influences the regulation of common sequences across different levels of CCM1, both with and without PRG treatment. (G) The ORA results, highlighting core-enriched DEPs in both GO and KEGG pathways, were presented to show the differential responses to PRG actions across three CCM1 genotypes (CCM1-KO, left; CCM1-WT, middle; and CCM1-KI/96, right). The analysis plots depict data specifically filtered with CCM1-associated pathways from enriched DEPs in both GO and KEGG approaches (Figure 1D,E), with the original dataset provided in the Supplementary Materials. It is important to highlight that all samples underwent triplicate analysis, and statistical significance was evaluated using a t-test, with p-values less than 0.05 considered significant. (H) Additionally, the RNAseq data underwent ORA similar to the previous approach used for proteomic data. Importantly, we emphasized shared pathways between the RNAseq profiles and the proteomic data by highlighting them with bold borders, specifically focusing on Diabetic cardiomyopathy and Aminoacyl-tRNA biosynthesis. In the comparative ROA plots, the depiction and scale of circles remain consistent with the previous ones and uniform throughout this manuscript.

Figure 2

A multi-omics approach to analyzing proteomic and RNAseq data under various levels of CCM1 expression in response to progesterone (PRG) actions. Both protein and RNA expression profiles of MEFs across three different levels of CCM1 expression were investigated. The proteomic data included three distinct genotypes: complete depletion of CCM1 via CCM1-knockout (CCM1-KO), natural low expression levels of CCM1 (CCM1-WT), and ectopic overexpression of CCM1 via CCM1-knockin (CCM1-96). Meanwhile, the RNAseq data covered two levels of expression profiles: complete depletion (knockout) and ectopic overexpression levels of CCM1 via CCM1-knockin (CCM1-96). (A) The expression profiles of DEPs in response to PRG actions were investigated in MEFs with three different CCM1 genotypes. A Venn diagram was used to illustrate the specific DEPs affected by progesterone (PRG) treatment compared to vehicle controls across the three CCM1 genotypes in mouse embryonic fibroblasts (MEFs). (B) A bar diagram was used to depict the profiles of DEPs in response to PRG actions across three CCM1 genotypes. The diagram shows the distribution of upregulated DEPs (represented by a red bar) and downregulated DEPs (represented by a blue bar). (C) A heatmap was employed to visualize the profiles of DEPs in response to PRG actions across three CCM1 genotypes. The heatmap illustrates the distribution of upregulated DEPs (indicated by red lines) and downregulated DEPs (indicated by blue lines). Comparative analysis of DEPs was performed using pairwise t-test statistical analysis. (D) Similarly, for RNA expression profiling, Venn diagrams were used to compare expression patterns between untreated MEF KO controls and MEF-KO treated with progesterone (PRG) in the CCM1-depletion genotype (MEF KO). The analysis identified 751 unique differentially expressed genes (DEGs) upon exposure to PRG. (E) Likewise, for the ectopic overexpression of the CCM1 (MEF 96) genotype, we compared expression patterns between untreated MEF 96 controls and MEF 96 treated with progesterone (PRG). This analysis revealed 573 differentially expressed genes (DEGs) upon exposure to PRG. (F) A heatmap was generated to depict the profiles of differentially expressed genes (DEGs) in response to PRG actions across two CCM1 genotypes. The comprehensive analysis involved examining 1900 shared RNAseq profiles between the two CCM1 genotypes to investigate how exposure to PRG influences the regulation of common sequences across different levels of CCM1, both with and without PRG treatment. (G) The ORA results, highlighting core-enriched DEPs in both GO and KEGG pathways, were presented to show the differential responses to PRG actions across three CCM1 genotypes (CCM1-KO, left; CCM1-WT, middle; and CCM1-KI/96, right). The analysis plots depict data specifically filtered with CCM1-associated pathways from enriched DEPs in both GO and KEGG approaches (Figure 1D,E), with the original dataset provided in the Supplementary Materials. It is important to highlight that all samples underwent triplicate analysis, and statistical significance was evaluated using a t-test, with p-values less than 0.05 considered significant. (H) Additionally, the RNAseq data underwent ORA similar to the previous approach used for proteomic data. Importantly, we emphasized shared pathways between the RNAseq profiles and the proteomic data by highlighting them with bold borders, specifically focusing on Diabetic cardiomyopathy and Aminoacyl-tRNA biosynthesis. In the comparative ROA plots, the depiction and scale of circles remain consistent with the previous ones and uniform throughout this manuscript.

Figure 3

Generation of Combined Non-mPR PRG Action Filters Using a Multi-Omics Approach. This experiment illustrates our approach to create a secondary filter associated with non-mPR-specific PRG actions. This filter aimed to identify all pass-through pathways induced by all PRG actions. Given the available data, our initially identified targets were all DEPs and DEGs in response to PRG actions through CCM1, some of which are not mediated through mPR-specific PRG actions. In this experiment, we defined all mPR-specific DEPs and DEGs by employing the filter extracted from an established dataset of DEPs/DEGs in response to mifepristone (MIF, as an antagonist to nPRs/GRs, but agonist to mPRs) treatment, to identify pathways shared by the glucocorticoid receptor (GR) and the classic nuclear progesterone receptor (nPR). It is crucial to reemphasize that MIF functions only as an antagonist to the pathways shared by the GRs and nPRs. However, it acts solely as an agonist and works synergistically with PRG on mPR-mediated signaling (Supplement Figure S3), forming the foundation for this experiment. (A) Combined GO/KEGG ORA enrichment of DEPs from proteomic data in response to PRG actions through CCM1. In the panel, pathways framed in yellow represent cell proliferation processes, such as DNA replication and DNA repair pathways, including single-stranded DNA helicase activity, and catalytic activity acting on DNA. In contrast, pathways framed in green indicate cell–cell adherent junctions. (B) GO/KEGG ORA pathway enrichment of DEGs from RNAseq transcriptional expression data in response to PRG modulation via CCM1. In the panel, pathways highlighted in yellow represent cell proliferation processes, while green frames indicate cell–cell adherent junctions. Additionally, red frames highlight cellular signal transduction factors, purple frames denote inflammatory factors, and black frames indicate angiogenic factors. Statistical significance was assessed using a Student’s t-test, with a significance threshold set at p < 0.05.

Figure 3

Generation of Combined Non-mPR PRG Action Filters Using a Multi-Omics Approach. This experiment illustrates our approach to create a secondary filter associated with non-mPR-specific PRG actions. This filter aimed to identify all pass-through pathways induced by all PRG actions. Given the available data, our initially identified targets were all DEPs and DEGs in response to PRG actions through CCM1, some of which are not mediated through mPR-specific PRG actions. In this experiment, we defined all mPR-specific DEPs and DEGs by employing the filter extracted from an established dataset of DEPs/DEGs in response to mifepristone (MIF, as an antagonist to nPRs/GRs, but agonist to mPRs) treatment, to identify pathways shared by the glucocorticoid receptor (GR) and the classic nuclear progesterone receptor (nPR). It is crucial to reemphasize that MIF functions only as an antagonist to the pathways shared by the GRs and nPRs. However, it acts solely as an agonist and works synergistically with PRG on mPR-mediated signaling (Supplement Figure S3), forming the foundation for this experiment. (A) Combined GO/KEGG ORA enrichment of DEPs from proteomic data in response to PRG actions through CCM1. In the panel, pathways framed in yellow represent cell proliferation processes, such as DNA replication and DNA repair pathways, including single-stranded DNA helicase activity, and catalytic activity acting on DNA. In contrast, pathways framed in green indicate cell–cell adherent junctions. (B) GO/KEGG ORA pathway enrichment of DEGs from RNAseq transcriptional expression data in response to PRG modulation via CCM1. In the panel, pathways highlighted in yellow represent cell proliferation processes, while green frames indicate cell–cell adherent junctions. Additionally, red frames highlight cellular signal transduction factors, purple frames denote inflammatory factors, and black frames indicate angiogenic factors. Statistical significance was assessed using a Student’s t-test, with a significance threshold set at p < 0.05.

Figure 4

Identifying mPR-Specific pathways Using combined Non-mPR Progesterone Action Filters. The objective of this experiment was to identify signaling pathways specifically associated with the membrane progesterone receptor (mPR) by excluding the PRG actions that are not mPR-specific across the distinct CCM1 genotypes defined in the previous step (Supplementary Figure S3A). This approach aimed to generate mPR-specific signaling pathways in response to PRG actions via CCM1, as represented below. (A) To visualize the ORA results of core-enriched DEPs from GO and KEGG pathway enrichments, an integrative dot plot was used. The plot displayed data from three CCM1 genotypes that passed the filter for non-mPR-specific PRG actions. Triplicate analysis was conducted, and statistical significance was determined using a Student’s t-test with a p-value cut-off of less than 0.05. The integrative dot plot highlighted the enriched pathways associated with mPR-specific PRG action, indicated by red-framed, blue-framed, and black pathways. (B) Similarly, an integrative dot plot was employed to visualize the ORA signaling pathways for mPR-specific PRG actions of enriched RNAseq data from GO and KEGG pathway enrichments. These data, which underwent triplicate analysis, passed through the filter for non-mPR-specific PRG actions among the two CCM1 genotypes used in the RNAseq portion of this study. The statistical significance of the results was determined using a Student’s t-test, with a p-value cut-off of less than 0.05. (C) Finally, a summarized dot plot was created to visualize the combined ORA signaling pathways for mPR-specific PRG actions, highlighting the core-enriched DEPs and DEGs from both proteomic and RNAseq data.

Figure 4

Identifying mPR-Specific pathways Using combined Non-mPR Progesterone Action Filters. The objective of this experiment was to identify signaling pathways specifically associated with the membrane progesterone receptor (mPR) by excluding the PRG actions that are not mPR-specific across the distinct CCM1 genotypes defined in the previous step (Supplementary Figure S3A). This approach aimed to generate mPR-specific signaling pathways in response to PRG actions via CCM1, as represented below. (A) To visualize the ORA results of core-enriched DEPs from GO and KEGG pathway enrichments, an integrative dot plot was used. The plot displayed data from three CCM1 genotypes that passed the filter for non-mPR-specific PRG actions. Triplicate analysis was conducted, and statistical significance was determined using a Student’s t-test with a p-value cut-off of less than 0.05. The integrative dot plot highlighted the enriched pathways associated with mPR-specific PRG action, indicated by red-framed, blue-framed, and black pathways. (B) Similarly, an integrative dot plot was employed to visualize the ORA signaling pathways for mPR-specific PRG actions of enriched RNAseq data from GO and KEGG pathway enrichments. These data, which underwent triplicate analysis, passed through the filter for non-mPR-specific PRG actions among the two CCM1 genotypes used in the RNAseq portion of this study. The statistical significance of the results was determined using a Student’s t-test, with a p-value cut-off of less than 0.05. (C) Finally, a summarized dot plot was created to visualize the combined ORA signaling pathways for mPR-specific PRG actions, highlighting the core-enriched DEPs and DEGs from both proteomic and RNAseq data.

Figure 5

Identifying transcription factors (TFs) in mPR-specific regulatory pathways using ML/DL techniques with enrichment of mPR-specific DEPs and DEGs. After machine learning/deep learning (ML/DL)-based prediction, we identified 12 potential TFs and explored their functional roles using Entrez ID identifiers.