Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Sep 14.
Published in final edited form as: Nat Cardiovasc Res. 2022 Mar 14;1:246–252. doi: 10.1038/s44161-022-00035-7

Developmental venous anomalies are a genetic primer for cerebral cavernous malformations

Daniel A Snellings 1, Romuald Girard 2, Rhonda Lightle 2, Abhinav Srinath 2, Sharbel Romanos 2, Ying Li 2, Chang Chen 2, Aileen A Ren 3, Mark L Kahn 3, Issam A Awad 2,*,‡, Douglas A Marchuk 1,*,‡
PMCID: PMC8958845  NIHMSID: NIHMS1777926  PMID: 35355835

Introductory Paragraph

Cerebral cavernous malformations (CCM) are a neurovascular anomaly that may occur sporadically, or be inherited due to autosomal dominant mutations in KRIT1, CCM2, or PDCD10. Individual lesions are caused by somatic mutations which have been identified in KRIT1, CCM2, PDCD10, MAP3K3, and PIK3CA. However, the interactions between mutations, and their relative contributions to sporadic versus familial cases remain unclear. We show that mutations in KRIT1, CCM2, PDCD10, and MAP3K3 are mutually exclusive, but may co-occur with mutations in PIK3CA. We also find that MAP3K3 mutations may cause sporadic, but not familial CCM. Furthermore, we find identical PIK3CA mutations in CCMs and adjacent developmental venous anomalies (DVA), a common vascular malformation frequently found in the vicinity of sporadic CCMs. However, somatic mutations in MAP3K3 are found only in the CCM. This suggests that sporadic CCMs are derived from cells of the DVA which have acquired an additional mutation in MAP3K3.

Main Text

CCMs are prone to hemorrhage often leading to stroke, seizures and disability. The inherited form of CCM disease is characterized by numerous lesions throughout the brain and spinal cord and is caused by an autosomal dominant loss of function (LOF) mutation in the genes encoding components of the CCM signaling complex: KRIT11,2, CCM23, or PDCD104. In contrast, sporadic CCMs typically occur as solitary lesions and form in the absence of an inherited germline mutation. Previous studies have established that somatic mutations in genes of the CCM complex cause biallelic LOF5–8; however, it is unclear how the recently identified mutations in MAP3K3 and PIK3CA fit into this existing model of pathogenesis.

The presence of multiple somatic mutations in CCMs also raises the question of how these mutations arise, especially in sporadic cases where none of the mutations are inherited. It has long been appreciated that sporadic CCMs often form in the vicinity of DVA, but the underlying cause has remained a long-standing mystery. DVA are the most common vascular malformation present in 6–14% of the adult population15–17 with the majority developing prior to the age of 2015. When assessed by magnetic resonance imaging, an adjacent DVA is identified in 24–32% of sporadic CCM cases12–14, and an even greater fraction of sporadic CCMs are found to be associated with a DVA at surgery12,14. One study focused on DVA reported an adjacent sporadic CCM in 6.9% of all DVAs in a general population (116 of 1689)15. These studies highlight the association between DVA and sporadic CCM. By contrast, familial CCM lesions have not been associated with DVA18. These combined data suggest that a DVA is not required for CCM formation but may be a predisposing factor in sporadic cases.

Results

To evaluate whether sporadic and familial CCMs have distinct somatic mutation spectra we identified somatic mutations present in 71 CCMs (20 familial CCMs and 51 sporadic/presumed sporadic CCMs). Familial and Sporadic CCM were identified by clinical and genetic characteristics (see Methods), whereas cases lacking information concerning family history (e.g., deidentified biobank samples) were classified as unknown. Mutations in KRIT1, CCM2, PDCD10, and PIK3CA were detected by targeted sequencing and/or droplet digital PCR (ddPCR) as previously described9. The common gain of function mutation in MAP3K3 (hg38 chr17:63691212, NM_002401.3, c.1323C>G; NP_002392, p.I441M) was detected by ddPCR (Supplemental Table 1).

The p.I441M mutation in MAP3K3 was identified in 15/51 sporadic CCMs and 0/20 familial CCMs (Figure 1A). We also screened for MAP3K3 p.I441M in 8 blood samples for which we were previously unable to identify a germline mutation in KRIT1, CCM2, PDCD10. None of the 8 blood samples harbored MAP3K3 p.I441M. Notably 11/51 sporadic CCMs harbored at least 1 somatic mutation in KRIT1, CCM2, or PDCD10, however none of these CCMs also had a mutation in MAP3K3 indicating that a mutual loss of the CCM complex and gain of function in MEKK3 (the protein product of MAP3K3) are not both required for CCM formation. As the CCM complex is known to be a direct inhibitor of MEKK3 activity19, these data strongly suggest identical functional consequences of these mutations.

Figure 1. Mutations in MAP3K3 are mutually exclusive with CCM gene mutations and occur in the same cells as PIK3CA mutations.

Figure 1.

Open in a new tab

A. Mutations present in 71 cerebral cavernous malformations (CCMs). Disease type denotes whether the sample was familial (F), sporadic (S), or unknown (blank). The presence of somatic mutations in PIK3CA and MAP3K3 are denoted by black and purple bars respectively. Germline and somatic mutations (green and blue respectively) in KRIT1, CCM2, or PDCD10, are shown in CCM Mut 1 with the second-hit mutation shown in CCM Mut 2 if present. B-D. Nuclei genotypes determined by snDNA-seq. The left and right circles in each Venn diagram shows the number of nuclei with the PIK3CA or MAP3K3 mutations where the overlap shows nuclei harboring both mutations. *** P < 0.0001, P = 4.3E-27 (B), 9.4E-33 (C), 1.3E-05 (D). E. Summary of data presented in B-D including P values determined by χ2 of the observed number of double mutant nuclei to the expected value derived from a Poisson distribution as done previously 9.

The majority of CCM and verrucous venous malformations with a mutation in MAP3K3 harbor the p.I441M variant 10,11,20, however an alternative variant p.Y544H has also been identified in a venous malformation21. While ddPCR provides superior sensitivity and specificity compared to targeted sequencing, it is restricted to detecting a single mutation per assay. To determine whether other mutations that contribute to CCM pathogenesis—either MAP3K3 mutations besides p.I441M, or mutations in yet undiscovered genes—we performed whole-exome sequencing (mean depth 133×) on 8 sporadic CCMs for which no somatic mutations in KRIT1, CCM2, PDCD10, or MAP3K3 were found. No additional mutations in MAP3K3 were identified and no candidate variants in other genes passed QC filters (see Methods).

While somatic mutations in KRIT1, CCM2, PDCD10, and MAP3K3 are mutually exclusive, somatic gain of function mutations in PIK3CA may co-occur with any other mutation (Figure 1A). We have previously shown that co-occurring mutations in KRIT1/CCM2 and PIK3CA occur in the same clonal population of cells9. To determine whether MAP3K3 and PIK3CA mutations co-exist in the same cells we performed single-nucleus DNA-sequencing (snDNA-seq) on frozen tissue from three surgically resected CCMs determined to harbor both mutations (Figure 1B–D).

In CCMs 5002 and 5030, the vast majority of mutant nuclei harbor both mutations in MAP3K3 and in PIK3CA indicating that these mutations co-exist in the same cells. In CCM 5032, 37% (19/51) of mutant nuclei harbor both mutations. While this is a far lower fraction compared to other samples, it is significantly higher than may be expected by chance when sampling from 1405 total nuclei (P = 1.3E-05, Figure 1E). In bulk genetic analysis, the allele frequencies of PIK3CA and MAP3K3 mutations detected in CCM 5002 were 19% and 13% respectively. In snDNA-seq the allele frequencies of these mutations increased to 28.7% and 30.9% respectively. This difference likely reflects the mosaic nature of CCMs. As snDNA-seq requires nuclei harvested from frozen tissue, we must sample a new area of the frozen lesion than was sampled for bulk sequencing. Sampling from different sites of the same lesion often results in minor changes in allele frequency, however the drastic change in allele frequency we find in CCM 5002 suggests either that our initial sample of the lesion for bulk sequencing contained largely non-lesion tissue, or an uneven distribution of mutant cells in the lesion.

All three samples support the coexistence of MAP3K3 and PIK3CA somatic mutations in single cells, however it is worth noting that in each sample we also observe singly-mutated nuclei representing each possible genotype. This arrangement of mutations is biologically unlikely as it would require a somatic mutation in one gene, followed reversion of the mutation in the other gene. Instead, the observed singly-mutated genotypes are likely the result of “allelic dropout”, a common technical artifact in single-cell DNA sequencing methods22. As each allele is present in a single copy per cell, the inability to consistently amplify both alleles (e.g., due to incomplete nuclear lysis) leads to occasional, random dropout of an allele and misrepresentation of genotypes. Allelic dropout prevents us from accurately identifying the small populations of cells that acquired the first mutation prior to acquisition of the second mutation.

To determine whether DVA and CCMs originate from a shared mutation, we collected three sporadic CCMs and sampled a portion of the associated DVA obtained during surgery (Figure 2A–C). Assays for mutations via ddPCR revealed that all three CCMs have a somatic activating mutation in PIK3CA and that the same mutation is present within the paired DVA samples at lower frequency (Figure 2D). Furthermore, ddPCR revealed that two of the CCMs harbored a mutation in MAP3K3 in addition to the previously noted mutation in PIK3CA. However, unlike the PIK3CA mutation, the MAP3K3 mutation was entirely absent from both DVA samples (Figure 2E). The presence of the PIK3CA, but not the MAP3K3, mutation in the DVA confirms that the PIK3CA mutation in the DVA did not arise via cross-sample contamination. The presence of multiple somatic mutations in these CCMs allows us to infer the developmental history of the lesion. The cancer field commonly uses the presence or absence of somatic mutations in clonal populations to track the evolutionary history of a tumor. Recent studies have expanded on this approach to use somatic mutations as endogenous barcodes to track embryonic development23. Using this same approach, we infer that the DVA was the first lesion to develop and that the associated CCM is derived from cells of the DVA following a somatic mutation in MAP3K3. The one CCM sample in which we found a mutation in PIK3CA but not MAP3K3 or CCM complex genes supports the role of PIK3CA in DVA development, but cannot be used to infer the temporal sequence of mutations. Notably, the lack of a MAP3K3/CCM complex mutation in 1 of 3 samples (33.3%) is consistent with our observations from bulk sequencing data where we did not identify MAP3K3/CCM complex mutations in 25 of 71 samples (35.2%).

Figure 2. Associated CCM and DVA harbor identical somatic mutations in PIK3CA.

Figure 2.

Open in a new tab

A, B, C. Axial magnetic resonance (MR) susceptibility weighted images acquired at 3 Tesla showing CCM (blue circle) and associated DVA branches sampled during surgery (red arrow) in individuals with cerebral cavernous malformation (CCM) 5080 (A) or 5081 (B) or 5082(C) (scale bars 15mm). The inset red box in C shows the region expanded to the right with the CCM and developmental venous anomaly (DVA) marked (scale bar 5mm). D, E. Somatic mutations in PIK3CA (D) and MAP3K3 (E) in CCM (top panels) and the associated DVA (bottom panels) from samples 5081, 5082, and 5083. Mutations were detected by droplet digital PCR and shown as the fluorescence of the reference probe on the x-axis, and the mutant probe on the y-axis. Droplets containing the reference allele, mutant allele, both, or neither, are colored in green, blue, orange, and black respectively. Percentage inset into each graph shows the variant allele frequency for the displayed mutation. If the mutation was determined to be present, the percentage is blue, else the percentage is red.

In addition to assaying the presence of PIK3CA mutations in DVA associated with CCM, we would ideally also assay DVA that are not associated with CCM. Unfortunately, DVA are benign malformations and are not resected unless associated with an additional pathology. This has precluded the direct assessment of PIK3CA mutations in DVA without a CCM. To address this limitation, we sought another source of tissue that could be non-invasively assayed for indirect evidence of PIK3CA activation. Thus, we collected plasma from individuals with DVA without a CCM and measured circulating miRNAs that might serve as biomarkers reflecting PIK3CA activity24.

We sequenced the plasma miRNomes of 12 individuals with a sporadic CCM associated with a DVA (CCM + DVA), 6 individuals with DVA only, and 7 healthy controls. Three plasma miRNAs were DE in the DVA only group when compared to healthy controls (P < 0.05; false discovery rate [FDR] corrected). One of the DE miRNAs, miR-134–5p (log2(FC)=−3.30), was downregulated and has been shown to inhibit PI3K/AKT signaling25 (Supplemental Table 2).

In addition, 18 plasma miRNAs were DE in patients with DVA only when compared to CCM + DVA (P < 0.05; FDR corrected). One of these 18 DE miRNAs, let-7c-5p (log2(FC)=−3.66) was downregulated and is known to target PIK3CA26,27 (Supplemental Table 2). Of interest, let-7c-5p also targets COL1A128, a DEG within the transcriptome of human sporadic CCM lesions (see Supplemental Information).

Additionally, 28 DE plasma miRNAs were identified between CCM + DVA and healthy controls (P < 0.05; FDR corrected). Four of these miRNAs putatively target PIK3CA: miR-148a-3p (log2(FC)=1.71), miR-148b-3p (log2(FC)=1.4), miR-128–3p (log2(FC)=1.35) and let-7c-5p (log2(FC)=2.07) (Supplemental Table 2)26,27,29–31.

Downregulation of a miRNA may lead to an upregulation of the targeted gene32. Even though these associations cannot be validated by somatic mutation analysis due to the lack of surgical tissue for these patients, the results of the circulating miRNome may reflect biomarkers of PIK3CA activation in patients harboring a DVA.

Discussion

In this study we have further interrogated the relationship between somatic mutations in KRIT1, CCM2, PDCD10, MAP3K3, and PIK3CA which contribute to the pathogenesis of CCM. We find that somatic mutations in MAP3K3 are not present in CCMs from individuals with familial CCM, consistent with a recent study10. We find that sporadic CCMs may harbor mutations in MAP3K3, KRIT1, CCM2, or PDCD10, but that the lesion will only have mutations in one of these genes. This implies that mutations in any of MAP3K3, KRIT1, CCM2, or PDCD10 are sufficient for CCM formation, without the need for mutations in a second gene. As the CCM complex is a direct inhibitor of MAP3K3 activity19, this pathway may be activated by either CCM complex LOF or by MAP3K3 GOF, but the mutual exclusivity of mutations in these genes suggests that only one of these events is necessary for lesion formation.

CCMs often develop as the result of multiple somatic mutations that co-exist within the same cells as we show with snDNA-seq. Although several somatic mutations occur in every cell division, the specificity of the mutations in CCM translates to a very low chance of acquiring these mutations within a single cell. This is especially true of somatic mutations in MAP3K3 and PIK3CA, both of which have very narrow spectra of activating mutations. Despite this improbability, the accumulation of these mutations in CCM seems to occur frequently. We propose that after an initial somatic mutation, the singly-mutated cell undergoes clonal expansion to form an intermediate lesion. In this study we identify 7 CCMs with either biallelic LOF in a CCM complex gene or MAP3K3 GOF in the absence a PIK3CA mutation, suggesting that PIK3CA activation is not required for CCM formation. Furthermore, previous work in mouse models has shown that loss of a CCM complex gene (with WT Pik3ca) leads to clonal expansion of the mutant cells33,34. As a result of this clonal expansion, the probability of creating a double-mutant cell increases by a factor of the clonal population size as there are more cells in which the second mutation may occur. The data presented in this study suggest that DVA function similarly; developing from a PIK3CA mutation that clonally expands, increasing the number of cells in which a second mutation may occur.

Plasma miRNA analysis of individuals with DVA-associated CCM and DVA without CCM revealed both groups exhibit differentially expressed miRNAs that putatively target PI3K/AKT signaling. Notably, it remains unclear if the circulating DE plasma miRNAs identified herein affect their predicted gene targets and associated biological pathways within the lesions35. While DVA only vs healthy controls revealed one DE miRNA that putatively targets PI3K/AKT signaling, DVA + CCM vs healthy controls revealed three DE miRNA targeting PI3K/AKT. This may reflect the synergistic effects of the CCM signaling pathway with PIK3CA mutation to drive PI3K/AKT signaling as previously reported9. One significant limitation of this exploratory miRNA study is the limited sample sizes of the cohorts. While further studies will be required to understand the effects of DVA and CCM on the circulating miRNome and identify biomarkers of PIK3CA activation, these data are thus far consistent with our observation of PIK3CA gain of function mutations in DVA associated with CCM. Furthermore, these data motivate further studies to identify circulating plasma miRNAs that may be a valuable clinical tool to non-invasively assay PIK3CA activation.

The presence of PIK3CA mutations in DVA suggests that DVA act as a genetic precursor to CCM, which would account for the strong association between sporadic CCM and DVA (Figure 3). Likewise, DVA are not associated with familial CCM because the presence of an inherited germline mutation in a CCM gene biases probability towards a CCM gene somatic mutation occurring first, as there exist many different mutations that may cause LOF, but far fewer that would cause GOF in PIK3CA.

Figure 3. Genetic model of CCM pathogenesis.

Figure 3.

Open in a new tab

The genetic trajectories that underly familial and sporadic cerebral cavernous malformation (CCM) pathogenesis. Familial CCMs already harbor a predisposing germline mutation in the CCM complex (KRIT1, CCM2, PDCD10) and are therefore most likely to develop without requiring association with a developmental venous anomaly (DVA) (top). In contrast, individuals without familial CCM—but who have a PIK3CA-mutant DVA—are predisposed to sporadic CCM formation adjacent to the DVA as one genetic ‘hit’ is already present (bottom). However, sporadic CCMs but could also develop in the absence of the DVA (top), depending on the temporal sequence of acquisition of somatic mutations. GOF, gain of function; LOF, loss of function.

Collecting tissue from CCM-associated DVA is challenging; however, collecting tissue from DVA not associated with CCM is yet more challenging as DVA are considered benign and are therefore not resected. We have attempted to address this limitation by studying biomarkers of PI3K activity which can be assayed noninvasively in blood plasma. Assaying the presence of PIK3CA mutations in DVA not associated with CCM will be the domain of future studies, but the data we present here demonstrate a clear link between DVA and PIK3CA, and suggest a model that explains the long recognized—but poorly understood—association between CCM and DVA.

While we are unable to address the presence of PIK3CA mutations in DVA not associated with CCM, it is worth noting that DVAs have been associated with other PI3K-related disorders36–39,40 including some cancers and neurological malformations, suggesting that DVA may have a role, possibly even as a genetic primer, in these other phenotypes.

Methods

Sample Collection

Surgically resected CCMs were obtained from consenting participants at the University of Chicago, the Barrow Neurological Institute, and the Angioma Alliance biobank. Additional DVA tissue was discretely dissected from the lesion during surgical resection of the associated CCM at the University of Chicago. This study was approved by each institution’s respective Institutional Review Board.

Familial and Sporadic Diagnosis

Familial-CCM patients harbor multiple lesions throughout the brain on MR susceptibility weighted imaging (SWI), a documented CCM1, CCM2, or CCM3 germline mutation, and/or first-degree relative with a history of CCM. Sporadic/solitary patients typically harbor a single lesion on SWI, or a cluster of CAs associated with a developmental venous anomaly41. Cases without clear information about family history—e.g., deidentified samples acquired from tissue biobanks—were classified as unknown.

DNA Extraction

DNA from CCM and DVA samples was extracted using the DNeasy blood and tissue kit (QIAGEN, catalog number 69504) per the manufacturers protocol. DNA purity was determined by Nanodrop and concentration was determined using the Qubit dsDNA BR assay kit (Invitrogen, catalog number Q32850) per the manufacturers protocol.

Droplet Digital PCR

Detection of MAP3K3 p.I441M was performed via ddPCR using a previously published probe set20 detailed and synthesized by Integrated DNA Technologies.

Forward Primer: 5′-TGCAGTACTATGGCTGTCTG-3′

Reverse Primer: 5′-GTCTCACATGCATTCAAGG-3′

Reference Allele Probe: 5′-HEX-CCTGACCATcTTCATGGAGTACA-IBlk-3′

Alternate Allele Probe: 5′-FAM-CCTGACCATgTTCATGGAGTACA-IBlk-3′

Assays were performed using 30–100ng of DNA with the QX200 AutoDG system (BioRad) and quantified with the QX200 droplet reader (BioRad). Analysis was performed with the QuantaSoft software (BioRad).

Sequencing

A total of 8 sporadic CCMs with no identified mutation in KRIT1, CCM2, PDCD10, or MAP3K3 (5001, 5005, 5006, 5022, 5024, 5036, 5078, and 5081) were used for whole-exome sequencing prepared using the SureSelect Human All Exon V7 probe set (Agilent, Design ID S31285117) per the manufacturers protocol. Prepared libraries were sequenced on one lane of a NovaSeq 6000 S4 flow cell for a mean depth of 133×.

Sequence Analysis

Sequencing data was processed using the Gene Analysis Toolkit (GATK, Broad Institute) while following the GATK best practices for somatic short variant discovery using Mutect2. Secondary variant detection was performed using gonomics (https://github.com/vertgenlab/gonomics) and bcftools mpileup to manually examine KRIT1, CCM2, PDCD10, and MAP3K3 for somatic variants. Putative variants were annotated using Funcotator (GATK), the catalog of somatic mutation in cancer (COSMIC), and the genome aggregation database (gnomAD). Putative variants were filtered according to the following criteria: greater than 50× total coverage, less than 90% strand specificity, greater than 5 reads supporting the alternate allele, greater than 1% alternate allele frequency, less than 1% population allele frequency, and predicted protein/splicing change.

Single-Nucleus DNA Sequencing

Nuclei isolates were prepared via Dounce homogenization of frozen tissue in Nuclei EZ Lysis Buffer (Sigma-Aldrich) and sorted to a single-nucleus suspension with a FACSAriaII (BD) (70um nozzle, 70psi, 4-Way Purity, chiller). Sequencing libraries from individual nuclei were prepared using the Tapestri platform (MissionBio) using a custom panel targeting KRIT1, CCM2, PDCD10, MAP3K3, and PIK3CA. Libraries were pooled and sequenced with a NextSeq Mid-Output 2×150bp kit (Illumina). Data processing and QC was performed with the MissionBio cloud analysis pipeline (v1.10.0). P values for mutation co-occurance was determined by χ2 test of observed and expected genotype counts as determined by a Poisson distribution9.

miRNA Extraction and Sequencing

Total plasma RNA was extracted from the plasma of 12 individuals with a sporadic CCM and an associated DVA (CCM + DVA), 6 individuals with DVA and without a CCM (DVA only), and 7 healthy controls using the miRNeasy Serum/Plasma Kit (Qiagen, Hilden, Germany) following the manufacturer isolation protocol. Diagnosis of CCM with an associated DVA, as well as DVA without a CCM lesion was confirmed on susceptibility weighted MR imaging. Illumina small RNA-Seq kits (Clontech, Mountain View, CA, USA) were then used to generate cDNA libraries, and sequencing was completed with the Illumina HiSeq 4000 platform (Illumina, San Diego, CA, USA), with single-end 50bp reads, at the University of Chicago Genomics Core. Differential miRNA analyses were completed between (1) CCM + DVA to DVA only and then (2) DVA only to healthy controls. The differentially expressed miRNAs were identified having P < 0.05, FDR-corrected. All analyses were completed using the sRNAToobox and DESeq2 R packages42,43.

Identification of Putative Targets

miRWalk 3.0 was queried to identify the putative gene targets of each of the DE miRNAs, using a random forest tree algorithm with a bonding prediction probability higher than 95% on the 3 different gene locations (3′ UTR, 5′UTR, and CDS)44. Putative gene targets of the DE miRNAs were identified in at least 2 of the 3 databases. DE miRNAs between (1) CCM + DVA and DVA only as well as (2) DVA only and healthy controls were then analyzed for potential targeting of the PI3K signalling pathway.

Data Availability:

Data not included in this paper can be accessed through NCBI (DNA sequencing, BioProject Accession: PRJNA802805) or GEO (RNA sequencing, Accession: GSE195732). Public datasets used here are available at COSMIC (cancer.sanger.ac.uk/cosmic), dbSNP (ncbi.nlm.nih.gov/snp), 1000 Genomes Project (internationalgenome.org), ExAC (gnomad.broadinstitute.org), miRWalk3.0 (mirwalk.umm.uni-heidelberg.de) and the DAVID database (david.ncifcrf.gov).

Code availability:

Variant calling software was implemented as part of Gonomics, an ongoing effort to develop an open-source genomics platform in the Go programming language. Gonomics can be accessed at github.com/vertgenlab/gonomics.

Supplementary Material

Supplemental Information
1777926_Sup_tables

Acknowledgements:

We thank the patients who donated tissue for this study. We thank Angioma Alliance, the Barrow Neurological Institute, and the University of Chicago for patient enrollment and sample collection. Nucleus sorting was performed in the Duke Human Vaccine Institute Research Flow Cytometry Shared Resource Facility. We thank Duke University School of Medicine for use of the Sequencing and Genomic Technologies Shared Resource for library preparation and sequencing. These studies were supported by National Institute of Health grants, P01NS092521 (DM, IA, MK), F31HL152738 (DS).

Footnotes

Competing interests: The authors declare no competing financial interests. IAA is Chairman of the Scientific Advisory Board for Angioma Alliance and provides expert opinions related to clinical care of cerebral cavernous malformations.

References

  • 1.Laberge-le Couteulx S et al. Truncating mutations in CCM1, encoding KRIT1, cause hereditary cavernous angiomas. Nat Genet 23, 189–193, doi: 10.1038/13815 (1999). [DOI] [PubMed] [Google Scholar]
  • 2.Sahoo T et al. Mutations in the gene encoding KRIT1, a Krev-1/rap1a binding protein, cause cerebral cavernous malformations (CCM1). Hum Mol Genet 8, 2325–2333, doi: 10.1093/hmg/8.12.2325 (1999). [DOI] [PubMed] [Google Scholar]
  • 3.Liquori CL et al. Mutations in a gene encoding a novel protein containing a phosphotyrosine-binding domain cause type 2 cerebral cavernous malformations. Am J Hum Genet 73, 1459–1464, doi: 10.1086/380314 (2003). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Bergametti F et al. Mutations within the programmed cell death 10 gene cause cerebral cavernous malformations. Am J Hum Genet 76, 42–51, doi: 10.1086/426952 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Gault J, Shenkar R, Recksiek P & Awad IA Biallelic somatic and germ line CCM1 truncating mutations in a cerebral cavernous malformation lesion. Stroke 36, 872–874, doi: 10.1161/01.STR.0000157586.20479.fd (2005). [DOI] [PubMed] [Google Scholar]
  • 6.Gault J et al. Cerebral cavernous malformations: somatic mutations in vascular endothelial cells. Neurosurgery 65, 138–144; discussion 144–135, doi: 10.1227/01.NEU.0000348049.81121.C1 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Akers AL, Johnson E, Steinberg GK, Zabramski JM & Marchuk DA Biallelic somatic and germline mutations in cerebral cavernous malformations (CCMs): evidence for a two-hit mechanism of CCM pathogenesis. Hum Mol Genet 18, 919–930, doi: 10.1093/hmg/ddn430 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.McDonald DA et al. Lesions from patients with sporadic cerebral cavernous malformations harbor somatic mutations in the CCM genes: evidence for a common biochemical pathway for CCM pathogenesis. Hum Mol Genet 23, 4357–4370, doi: 10.1093/hmg/ddu153 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Ren AA et al. PIK3CA and CCM mutations fuel cavernomas through a cancer-like mechanism. Nature 594, 271–276, doi: 10.1038/s41586-021-03562-8 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Weng J et al. Somatic MAP3K3 mutation defines a subclass of cerebral cavernous malformation. Am J Hum Genet 108, 942–950, doi: 10.1016/j.ajhg.2021.04.005 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Hong T et al. Somatic MAP3K3 and PIK3CA mutations in sporadic cerebral and spinal cord cavernous malformations. Brain, doi: 10.1093/brain/awab117 (2021). [DOI] [PubMed] [Google Scholar]
  • 12.Abdulrauf SI, Kaynar MY & Awad IA A comparison of the clinical profile of cavernous malformations with and without associated venous malformations. Neurosurgery 44, 41–46; discussion 46–47, doi: 10.1097/00006123-199901000-00020 (1999). [DOI] [PubMed] [Google Scholar]
  • 13.Wurm G, Schnizer M & Fellner FA Cerebral cavernous malformations associated with venous anomalies: surgical considerations. Neurosurgery 57, 42–58; discussion 42–58, doi: 10.1227/01.neu.0000163482.15158.5a (2005). [DOI] [PubMed] [Google Scholar]
  • 14.Porter PJ, Willinsky RA, Harper W & Wallace MC Cerebral cavernous malformations: natural history and prognosis after clinical deterioration with or without hemorrhage. J Neurosurg 87, 190–197, doi: 10.3171/jns.1997.87.2.0190 (1997). [DOI] [PubMed] [Google Scholar]
  • 15.Brinjikji W, El-Rida El-Masri A, Wald JT & Lanzino G Prevalence of Developmental Venous Anomalies Increases With Age. Stroke 48, 1997–1999, doi: 10.1161/STROKEAHA.116.016145 (2017). [DOI] [PubMed] [Google Scholar]
  • 16.Linscott LL, Leach JL, Jones BV & Abruzzo TA Developmental venous anomalies of the brain in children -- imaging spectrum and update. Pediatr Radiol 46, 394–406; quiz 391–393, doi: 10.1007/s00247-015-3525-3 (2016). [DOI] [PubMed] [Google Scholar]
  • 17.Gokce E, Acu B, Beyhan M, Celikyay F & Celikyay R Magnetic resonance imaging findings of developmental venous anomalies. Clin Neuroradiol 24, 135–143, doi: 10.1007/s00062-013-0235-9 (2014). [DOI] [PubMed] [Google Scholar]
  • 18.Dammann P et al. Correlation of the venous angioarchitecture of multiple cerebral cavernous malformations with familial or sporadic disease: a susceptibility-weighted imaging study with 7-Tesla MRI. J Neurosurg 126, 570–577, doi: 10.3171/2016.2.JNS152322 (2017). [DOI] [PubMed] [Google Scholar]
  • 19.Zhou Z et al. The cerebral cavernous malformation pathway controls cardiac development via regulation of endocardial MEKK3 signaling and KLF expression. Dev Cell 32, 168–180, doi: 10.1016/j.devcel.2014.12.009 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Couto JA et al. A somatic MAP3K3 mutation is associated with verrucous venous malformation. Am J Hum Genet 96, 480–486, doi: 10.1016/j.ajhg.2015.01.007 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Al-Qattan MM et al. Late-onset multiple venous malformations confined to the upper limb: link to somatic MAP3K3 mutations. J Hand Surg Eur Vol 45, 1023–1027, doi: 10.1177/1753193420922459 (2020). [DOI] [PubMed] [Google Scholar]
  • 22.Xu L et al. Clonal Evolution and Changes in Two AML Patients Detected with A Novel Single-Cell DNA Sequencing Platform. Sci Rep 9, 11119, doi: 10.1038/s41598-019-47297-z (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Bizzotto S et al. Landmarks of human embryonic development inscribed in somatic mutations. Science 371, 1249–1253, doi: 10.1126/science.abe1544 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Mori MA, Ludwig RG, Garcia-Martin R, Brandao BB & Kahn CR Extracellular miRNAs: From Biomarkers to Mediators of Physiology and Disease. Cell Metab 30, 656–673, doi: 10.1016/j.cmet.2019.07.011 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Yang C et al. Exosome miR-134–5p restrains breast cancer progression via regulating PI3K/AKT pathway by targeting ARHGAP1. J Obstet Gynaecol Res 47, 4037–4048, doi: 10.1111/jog.14983 (2021). [DOI] [PubMed] [Google Scholar]
  • 26.Yuan H et al. MicroRNA let-7c-5p promotes osteogenic differentiation of dental pulp stem cells by inhibiting lipopolysaccharide-induced inflammation via HMGA2/PI3K/Akt signal blockade. Clin Exp Pharmacol Physiol 46, 389–397, doi: 10.1111/1440-1681.13059 (2019). [DOI] [PubMed] [Google Scholar]
  • 27.Li Y, Li P & Wang N Effect of let-7c on the PI3K/Akt/FoxO signaling pathway in hepatocellular carcinoma. Oncol Lett 21, 96, doi: 10.3892/ol.2020.12357 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Srivastava SP, Hedayat AF, Kanasaki K & Goodwin JE microRNA Crosstalk Influences Epithelial-to-Mesenchymal, Endothelial-to-Mesenchymal, and Macrophage-to-Mesenchymal Transitions in the Kidney. Front Pharmacol 10, 904, doi: 10.3389/fphar.2019.00904 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Yin H et al. MiR-148a-3p Regulates Skeletal Muscle Satellite Cell Differentiation and Apoptosis via the PI3K/AKT Signaling Pathway by Targeting Meox2. Front Genet 11, 512, doi: 10.3389/fgene.2020.00512 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Cimino D et al. miR148b is a major coordinator of breast cancer progression in a relapse-associated microRNA signature by targeting ITGA5, ROCK1, PIK3CA, NRAS, and CSF1. FASEB J 27, 1223–1235, doi: 10.1096/fj.12-214692 (2013). [DOI] [PubMed] [Google Scholar]
  • 31.Huo L et al. miR-128–3p inhibits glioma cell proliferation and differentiation by targeting NPTX1 through IRS-1/PI3K/AKT signaling pathway. Exp Ther Med 17, 2921–2930, doi: 10.3892/etm.2019.7284 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Correia de Sousa M, Gjorgjieva M, Dolicka D, Sobolewski C & Foti M Deciphering miRNAs’ Action through miRNA Editing. Int J Mol Sci 20, doi: 10.3390/ijms20246249 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Detter MR, Snellings DA & Marchuk DA Cerebral Cavernous Malformations Develop Through Clonal Expansion of Mutant Endothelial Cells. Circ Res 123, 1143–1151, doi: 10.1161/CIRCRESAHA.118.313970 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Malinverno M et al. Endothelial cell clonal expansion in the development of cerebral cavernous malformations. Nat Commun 10, 2761, doi: 10.1038/s41467-019-10707-x (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Turchinovich A, Samatov TR, Tonevitsky AG & Burwinkel B Circulating miRNAs: cell-cell communication function? Front Genet 4, 119, doi: 10.3389/fgene.2013.00119 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Roux A et al. High Prevalence of Developmental Venous Anomaly in Diffuse Intrinsic Pontine Gliomas: A Pediatric Control Study. Neurosurgery 86, 517–523, doi: 10.1093/neuros/nyz298 (2020). [DOI] [PubMed] [Google Scholar]
  • 37.Brinjikji W et al. Facial Venous Malformations Are Associated with Cerebral Developmental Venous Anomalies. AJNR Am J Neuroradiol 39, 2103–2107, doi: 10.3174/ajnr.A5811 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Tan WH et al. The spectrum of vascular anomalies in patients with PTEN mutations: implications for diagnosis and management. J Med Genet 44, 594–602, doi: 10.1136/jmg.2007.048934 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Dhamija R et al. Neuroimaging abnormalities in patients with Cowden syndrome: Retrospective single-center study. Neurol Clin Pract 8, 207–213, doi: 10.1212/CPJ.0000000000000463 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Rasalkar DD & Paunipagar BK Developmental venous anomaly associated with cortical dysplasia. Pediatr Radiol 40 Suppl 1, S165, doi: 10.1007/s00247-010-1654-2 (2010). [DOI] [PubMed] [Google Scholar]
  • 41.Akers A et al. Synopsis of Guidelines for the Clinical Management of Cerebral Cavernous Malformations: Consensus Recommendations Based on Systematic Literature Review by the Angioma Alliance Scientific Advisory Board Clinical Experts Panel. Neurosurgery 80, 665–680, doi: 10.1093/neuros/nyx091 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Rueda A et al. sRNAtoolbox: an integrated collection of small RNA research tools. Nucleic Acids Res 43, W467–473, doi: 10.1093/nar/gkv555 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Love MI, Huber W & Anders S Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550, doi: 10.1186/s13059-014-0550-8 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Sticht C, De La Torre C, Parveen A & Gretz N miRWalk: An online resource for prediction of microRNA binding sites. PLoS One 13, e0206239, doi: 10.1371/journal.pone.0206239 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Information
NIHMS1777926-supplement-Supplemental_Information.pdf (26KB, pdf)
1777926_Sup_tables
NIHMS1777926-supplement-1777926_Sup_tables.xlsx (22.3KB, xlsx)

Data Availability Statement

Data not included in this paper can be accessed through NCBI (DNA sequencing, BioProject Accession: PRJNA802805) or GEO (RNA sequencing, Accession: GSE195732). Public datasets used here are available at COSMIC (cancer.sanger.ac.uk/cosmic), dbSNP (ncbi.nlm.nih.gov/snp), 1000 Genomes Project (internationalgenome.org), ExAC (gnomad.broadinstitute.org), miRWalk3.0 (mirwalk.umm.uni-heidelberg.de) and the DAVID database (david.ncifcrf.gov).

RESOURCES